Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Rainfall and nest site competition delay Mountain Bluebird and Tree Swallow breeding but do not impact productivity

Rainfall and nest site competition delay Mountain Bluebird and Tree Swallow breeding but do not... Abstract Optimizing breeding phenology, an important aspect of fitness, is complex for migratory species as they must make key timing decisions early, and remotely, from breeding sites. We examined the role of weather (locally and cross-seasonally), cavity availability, and competitive exclusion in determining among-year variation in breeding phenology over 17 yr for 2 migratory, cavity-nesting birds: Mountain Bluebirds (Sialia currucoides; n = 462 nests) and Tree Swallows (Tachycineta bicolor; n = 572) using natural tree cavities in British Columbia, Canada. We assessed weather effects within the winter and migratory range and at our study sites. We quantified competition as the proportion of cavities occupied by European Starlings (Sturnus vulgaris) (for both species) and Mountain Bluebirds (for Tree Swallow only) in each year. For 229 bluebird and 177 swallow nests with known fates, we tested whether late years resulted in reduced productivity. Although the effects were small, heavy rainfall and strong diurnal westerly winds during migration were associated with breeding delays for Mountain Bluebirds. However, cavity availability (earlier breeding with increases) had a 5–8 × greater effect on timing than migratory conditions. There was no evidence that starling competition delayed bluebirds. In Tree Swallows, greater local daily rainfall was associated with delayed breeding, as was starling abundance (the effect of starlings was 1.4 × smaller than that of rainfall). Neither bluebird abundance nor cavity availability changed swallow phenology. Neither species showed reduced productivity in late breeding years. In both species, individuals that bred late relative to conspecifics within-year had smaller clutches and greater probability of nest failure. We conclude that breeding ground conditions, particularly cavity limitation and local rainfall (for swallows), are important drivers of breeding phenology for our focal species, but that the productivity cost of late years, at least for Tree Swallows, is minimal. RÉSUMÉ Optimiser la phénologie de la reproduction, un aspect important de la condition physique, est complexe pour les espèces migratrices car elles doivent prendre tôt des décisions clés concernant le moment de la reproduction, et ce loin des sites de reproduction. Nous avons examiné le rôle des conditions météorologiques (localement et de façon saisonnière), de la disponibilité des cavités et de l’exclusion compétitive dans la détermination de la variation interannuelle de la phénologie de la reproduction pendant 17 ans chez deux oiseaux migrateurs nichant dans des cavités: Sialia currucoides (n = 462 nids) et Tachycineta bicolor (n = 572), lesquels utilisent des cavités d’arbres naturelles en Colombie-Britannique, au Canada. Nous avons évalué les effets des conditions météorologiques dans l’aire de répartition hivernale et migratoire, ainsi qu’à nos sites d’étude. Nous avons quantifié la compétition comme étant la proportion de cavités occupées par Sturnus vulgaris (pour les deux espèces) et S. currucoides (pour T. bicolor seulement) à chaque année. Pour 229 nids de S. currucoides et 177 nids de T. bicolor dont le sort était connu, nous avons testé si les années tardives causaient une réduction de la productivité. Bien que les effets aient été faibles, de fortes précipitations et de forts vents diurnes provenant de l’ouest pendant la migration étaient associés aux retards dans la reproduction pour S. currucoides. Toutefois, la disponibilité des cavités (reproduction plus hâtive si augmentations) avait de 5 à 8 fois plus d’effet sur le moment de la reproduction que les conditions migratoires. Aucune preuve n’appuyait le fait que la compétition par S. vulgaris retardait la reproduction de S. currucoides. Chez T. bicolor, une plus grande quantité quotidienne de précipitations au niveau local était associée à un retard dans la reproduction, tout comme l’abondance de S. vulgaris (l’effet de S. vulgaris était 1,4 fois plus faible que celui des précipitations). Ni l’abondance de S. currucoides ni la disponibilité des cavités n’ont changé la phénologie de T. bicolor. Aucune de ces espèces ne présentait une réduction de la productivité lors des années où la reproduction était plus tardive. Chez les deux espèces, les individus qui se reproduisaient tardivement comparativement à leurs congénères au cours de l’année avaient des couvées plus petites et une plus grande probabilité d’échec de la reproduction. Nous concluons que les conditions sur les aires de reproduction, particulièrement la limitation des cavités et les précipitations locales (pour T. bicolor), sont des facteurs importants de la phénologie de la reproduction pour nos espèces cibles, mais que les coûts de la productivité lors des années tardives sont minimaux, du moins pour T. bicolor. INTRODUCTION Birds are under selective pressure to match breeding with food abundance, such that they are provisioning young when resources are at their peak (Perrins 1970, Siikamäki 1998, Visser et al. 2006, Both 2010). In seasonal environments, initiating breeding too early or late can additionally expose nesting adults and developing young to unfavorable conditions. This, in turn, can lead to adult or offspring mortality, lower quality eggs, and nest abandonment (Nager and Van Noordwijk 1992, Nilsson 1994, Drake et al. 2014a, Johnson and Dawson 2019, de Zwaan et al. 2019). Late breeding by individuals is associated with smaller clutches, fewer fledglings, fewer recruits, and reduced parental survival over winter, independent of individual qualities that may influence such relationships (Nilsson and Svensson 1996, Brown and Roth 2002, Harriman et al. 2017). Because the timing of peak food and optimal weather conditions can be variable among years, bird species often exhibit behavioral plasticity and adjust clutch initiation dates in response to environmental cues (e.g., rainfall [Hau 2001, Illera and Díaz 2006, Cavalcanti et al. 2016], temperature [Cresswell and Mccleery 2003, Wesolowski and Cholewa 2009, Drake and Martin 2018], and food abundance [Hau et al. 2000]). Achieving optimal timing is complicated among migratory species because these species must make other timing decisions (winter ground departure date and rate of movement during migration) that impact when they can breed (Marra et al. 2005, Åkesson et al. 2017). These decisions are made early and remotely from breeding sites, potentially increasing the probability of mistiming (Åkesson et al. 2017, Franks et al. 2018). Poor conditions (e.g., drought, storm events) on the wintering grounds and on the migratory route may also delay breeding, for example, by slowing fattening or reducing flight range when food is scarce (Studds and Marra 2007, Rockwell et al. 2012, Gómez et al. 2017) or by slowing movement or keeping individuals grounded when conditions aloft are hostile (Drake et al. 2014b, Vansteelant et al. 2015, Mitchell et al. 2015). In cavity-nesting species, achieving optimal timing is further complicated by nest site availability. Cavity limitation and intra- and inter-specific competition for cavities should delay clutch initiation as pairs seek to acquire a suitable nest site (Koch et al. 2012). Thus, while relationships between timing and some weather cues may represent tracking of conditions and resources, relationships between timing and cavity availability or competition should represent constraints and be exclusively costly, by producing a mismatch with optimal timing. In this paper, we examine the timing of breeding in 2 migratory, secondary cavity-nesting species: Mountain Bluebird (Sialia currucoides) and Tree Swallow (Tachycineta bicolor). These species co-occur in breeding habitat throughout western North America; at our study site they are the most numerous of 4 migratory, secondary cavity-nesting species recorded (the other 2 species being non-native European Starling [Sturnus vulgaris] and American Kestrel [Falco sparverius]). Mountain Bluebirds winter in the southwestern United States with some range extension into the northern interior of Mexico (Johnson and Dawson 2019). Tree Swallows winter in western Mexico, throughout the Gulf of Mexico, and along the east coast of Mexico and Central America (Winkler et al. 2011). The spring migratory period in Mountain Bluebird begins earlier (February vs. March) but lasts longer than that of Tree Swallow (~3 vs. 1.5 mo) (Winkler et al. 2011, Johnson and Dawson 2019). Both species use previously excavated and naturally occurring tree holes as nesting sites as well as nest boxes, when provided. Based on previous work (Koch et al. 2012, Wiebe 2016), we consider European Starlings (for both species) and bluebirds (for swallow) to be strong competitors for nesting space. All 3 species overlap in terms of tree hole preferences and clutch initiation dates (although species’ mean initiation dates differ) (Koch et al. 2012). We examine variation in the timing of breeding in Mountain Bluebirds and Tree Swallows in natural tree cavities in south-central British Columbia over 17 yr. We assess whether these species showed shifts in clutch initiation dates among years in response to relevant weather variables across their annual cycle (winter and spring temperatures and rainfall, and spring wind speed). We also test whether ecological constraints—specifically, annual cavity availability and the prevalence of nest site competitors—were associated with shifts in clutch initiation dates within each population. We then assess the relative importance of our predictor variables in determining when, on average, Tree Swallows and Mountain Bluebirds initiated breeding in a given year. Finally, we test if population-level shifts in timing among years impacted population productivity at our study site. We do this by separating the productivity impact of breeding late relative to conspecifics within a year from the contribution of a late breeding year, on average, for the population as a whole. METHODS Study Site and Breeding Activity Breeding data for Mountain Bluebird and Tree Swallow were collected between 1995 and 2011 at 2 study sites ~38 km apart, Riske Creek (52.0025°N, 122.4116°W) and Knife Creek (52.0068°N, 121.8619°W), near Williams Lake in south-central British Columbia, Canada. This region is part of the warm and dry Interior Douglas-fir biogeoclimatic zone. In this study, the Riske Creek site consisted of 16 mixed conifer stands (Douglas-fir [Pseudotsuga menziesii var. menziesii], lodgepole pine [Pinus contorta var. latifolia], and white and Englemann spruce hybrids [Picea glauca × engelmannii]) with trembling aspen (Populus tremuloides) within a grassland-wetland matrix. Knife Creek consisted of 11 mixed conifer stands with some deciduous riparian zone. Stands ranged from 7 to 32 ha in size. No nest boxes were present at either site and all nesting was done in natural tree cavities. Mountain Bluebird and Tree Swallow were 2 of a total of 32 cavity-nesting bird species found within the study sites over the monitoring period (Wesolowski and Martin 2018). During the 1995 to 2011 period, systematic searches were conducted between May 1 and July 31. These surveys were conducted for ~6–7 hr per stand per week by walking the entirety of each stand and examining previously identified nest sites and following birds (Aitken and Martin 2007). The number of stands monitored increased between 1995 and 1998, but thereafter survey effort was equivalent. The majority of nests were found in the laying or early incubation stage (Koch et al. 2012). Active nesting cavities were identified based on adult behavior (carrying nesting material or food or entering or exiting cavities) or the vocalizations of young (Martin et al. 2004). All cavities at the study site were given unique identifiers and their persistence and use was recorded in each year of the study. Between 1995 and 2004, we accessed active cavities up to 5.2 m above the ground using ladders and mirrors to assess the stage of breeding. One hundred and twelve nests (10% of the total dataset) were inaccessible during this 10-yr period. These inaccessible cavities were recorded as active based on adult behavior or begging chicks but could not be assigned a clutch initiation date or hatch date in the field. After 2004, all nests were accessible using a video camera mounted on a pole (TreeTop Peeper; Sandpiper Technologies, Manteca, California, USA) to identify the stage of breeding in cavities up to 15 m above the ground (Edworthy et al. 2012). Nests were checked, on average, every 5 days during their active phase and, when possible, clutch initiation date was determined using observed clutch size (if nests were found during laying) or observed final clutch size and hatch date (if nests were found during incubation), combined with a 1 egg day–1 laying interval and a 14- or 15-day mean incubation period for Mountain Bluebird and Tree Swallow, respectively (Koch et al. 2012). Nesting attempts found after young had hatched were not assigned a clutch initiation date in the field. Nest activity periods were recorded as the first and last days that each cavity was observed as actively being used (containing fresh nesting material, eggs, or nestlings). The number of fledglings produced by each nest was recorded when fledging was observed or when chicks were old enough at the last nest check to survive out of the nest and where there was no evidence of predation within the cavity or around the nesting site when the nest was found empty at the penultimate check. The presence of adults feeding chicks in the immediate nest area was also used as an indicator of success where fledging was not directly observed. Phenology Our initial dataset consisted of 590 Tree Swallow and 519 Mountain Bluebird breeding attempts, of which, 218 and 203 had direct information on clutch initiation date recorded in the field. The remaining records had phenology data (such as dates observed active, lay stage when found, and fledge date), but no clutch initiation date. We used this phenology data to infer clutch initiation dates and an associated prediction error for these records, as described below. We used known dates to inform a mixed-effects model predicting the timing of clutch initiation in the remaining records, and incorporated the prediction error for these records into our analyses by using multiple imputation (MI) from a conditional distribution (Schafer and Graham 2002, Schomaker and Heumann 2014; application to this system Drake and Martin 2018). This involved 3 steps. First, we conservatively backdated 86 nests found at the nestling stage to the egg stage using either reported clutch size (nestlings plus unhatched eggs, n = 17) or mean clutch size at the study site (5 eggs for both species, n = 69), a lay rate of 1 egg day–1, and the mean incubation period for each species (Winkler et al. 2011, Johnson and Dawson 2019). Second, we used known clutch initiation dates for both species (n = 417, 5 records lacked end dates so were dropped from the model) to model clutch initiation date as a function of (1) the date of the earliest known/backdated nest activity, (2) the stage of the nest at that date (pre-lay, egg, or unknown), (3) the latest date the nest was observed active, and (4) nest fate (fledged, failed pre-hatch, failed pre-fledge, or unknown). Species identity was included as a random factor (intercept). These models were run in R 3.6.0 (R Core Team 2019) using the package lme4 (Bates et al. 2015). We ran this model 5 times with a unique 20% set aside to test its predictive capability and to calculate a root mean-squared prediction error (RMSPE) for our modeled clutch initiation dates. The average performance of the model was high (5-run mean: both species: r = 0.98, n = 83.4; Tree Swallow alone, r = 0.94; Mountain Bluebird alone, r = 0.98) and the RMSPE was ± 2.86 days. Finally, we used the complete dataset (n = 417) to train the final model and predict clutch initiation dates for those nests lacking direct data. Error within these predicted dates was incorporated into all subsequent analyses by multiply imputing these dates using a Monte Carlo approach; predicted dates were adjusted in each imputation using values obtained from random draws of a normal distribution with a mean of zero and a standard deviation of the RMSPE calculated above (2.86 days). To restrict our analyses to initial nesting attempts, known second nesting attempts by the same breeding pair were removed and records were additionally limited to the first occupancy of each cavity in each year by each species. This removed possible re-nests in the same cavity by the same breeding pair, as well as late nesting attempts by pairs of the same species following abandonment or displacement of the first nesting pair. We could not, however, identify re-nesting if it occurred in cavities that had not been previously used within the year. We therefore cannot exclude the possibility that some re-nests made it into the final dataset. Our final dataset consisted of 462 Mountain Bluebird, and 572 Tree Swallow records (see Table 1 for annual sample sizes). We note that a subset of these records (112 bluebird and 89 swallow records) were used by Koch et al. (2012). TABLE 1. Annual number of breeding records for Mountain Bluebird, Tree Swallow, and European Starling as well as total suitable cavity availability (see Methods) at our monitoring sites in Central British Columbia, Canada. Percent values (bracketed) indicate cavity occupancy by availability for each species and were used as a metric of annual competition for nest sites. Italicized records at Knife Creek were dropped from final AIC analyses due to their small within-year sample size and the inclusion of “Site” as a random effect in these analyses (see Methods). Year Riske Creek site Knife Creek site Mountain Bluebird Tree Swallow European Starling Suitable cavity # Mountain Bluebird Tree Swallow European Starling Suitable cavity # 1995 17 (17%) 10 (10%) 9 (9%) 98 – – – – 1996 12 (9%) 10 (8%) 9 (7%) 132 0 1 (8%) 0 12 1997 14 (7%) 12 (6%) 29 (14%) 212 2 (5%) 9 (22%) 1 (2%) 40 1998 14 (5%) 17 (6%) 39 (15%) 267 3 (4%) 8 (10%) 1 (1%) 80 1999 8 (3%) 15 (5%) 37 (12%) 317 3 (3%) 7 (6%) 6 (5%) 117 2000 18 (5%) 21 (5%) 47 (12%) 387 0 9 (6%) 2 (1%) 150 2001 16 (4%) 22 (5%) 50 (11%) 435 0 11 (6%) 2 (1%) 183 2002 23 (5%) 22 (5%) 24 (5%) 476 3 (1%) 13 (6%) 4 (2%) 208 2003 24 (5%) 30 (6%) 32 (6%) 533 4 (2%) 13 (6%) 1 (0%) 221 2004 27 (5%) 40 (7%) 18 (3%) 597 1 (0%) 14 (5%) 0 273 2005 41 (6%) 34 (5%) 20 (3%) 631 3 (1%) 11 (4%) 1 (0%) 294 2006 24 (6%) 18 (4%) 14 (3%) 428 6 (3%) 9 (4%) 0 233 2007 30 (7%) 23 (5%) 10 (2%) 423 0 3 (1%) 1 (0%) 249 2008 33 (7%) 36 (8%) 12 (3%) 474 1 (0%) 7 (3%) 1 (0%) 263 2009 42 (10%) 44 (10%) 13 (3%) 436 5 (2%) 8 (3%) 0 285 2010 33 (7%) 50 (10%) 18 (4%) 481 4 (1%) 8 (3%) 0 282 2011 47 (11%) 33 (8%) 9 (2%) 420 4 (1%) 4 (1%) 0 283 Year Riske Creek site Knife Creek site Mountain Bluebird Tree Swallow European Starling Suitable cavity # Mountain Bluebird Tree Swallow European Starling Suitable cavity # 1995 17 (17%) 10 (10%) 9 (9%) 98 – – – – 1996 12 (9%) 10 (8%) 9 (7%) 132 0 1 (8%) 0 12 1997 14 (7%) 12 (6%) 29 (14%) 212 2 (5%) 9 (22%) 1 (2%) 40 1998 14 (5%) 17 (6%) 39 (15%) 267 3 (4%) 8 (10%) 1 (1%) 80 1999 8 (3%) 15 (5%) 37 (12%) 317 3 (3%) 7 (6%) 6 (5%) 117 2000 18 (5%) 21 (5%) 47 (12%) 387 0 9 (6%) 2 (1%) 150 2001 16 (4%) 22 (5%) 50 (11%) 435 0 11 (6%) 2 (1%) 183 2002 23 (5%) 22 (5%) 24 (5%) 476 3 (1%) 13 (6%) 4 (2%) 208 2003 24 (5%) 30 (6%) 32 (6%) 533 4 (2%) 13 (6%) 1 (0%) 221 2004 27 (5%) 40 (7%) 18 (3%) 597 1 (0%) 14 (5%) 0 273 2005 41 (6%) 34 (5%) 20 (3%) 631 3 (1%) 11 (4%) 1 (0%) 294 2006 24 (6%) 18 (4%) 14 (3%) 428 6 (3%) 9 (4%) 0 233 2007 30 (7%) 23 (5%) 10 (2%) 423 0 3 (1%) 1 (0%) 249 2008 33 (7%) 36 (8%) 12 (3%) 474 1 (0%) 7 (3%) 1 (0%) 263 2009 42 (10%) 44 (10%) 13 (3%) 436 5 (2%) 8 (3%) 0 285 2010 33 (7%) 50 (10%) 18 (4%) 481 4 (1%) 8 (3%) 0 282 2011 47 (11%) 33 (8%) 9 (2%) 420 4 (1%) 4 (1%) 0 283 Open in new tab TABLE 1. Annual number of breeding records for Mountain Bluebird, Tree Swallow, and European Starling as well as total suitable cavity availability (see Methods) at our monitoring sites in Central British Columbia, Canada. Percent values (bracketed) indicate cavity occupancy by availability for each species and were used as a metric of annual competition for nest sites. Italicized records at Knife Creek were dropped from final AIC analyses due to their small within-year sample size and the inclusion of “Site” as a random effect in these analyses (see Methods). Year Riske Creek site Knife Creek site Mountain Bluebird Tree Swallow European Starling Suitable cavity # Mountain Bluebird Tree Swallow European Starling Suitable cavity # 1995 17 (17%) 10 (10%) 9 (9%) 98 – – – – 1996 12 (9%) 10 (8%) 9 (7%) 132 0 1 (8%) 0 12 1997 14 (7%) 12 (6%) 29 (14%) 212 2 (5%) 9 (22%) 1 (2%) 40 1998 14 (5%) 17 (6%) 39 (15%) 267 3 (4%) 8 (10%) 1 (1%) 80 1999 8 (3%) 15 (5%) 37 (12%) 317 3 (3%) 7 (6%) 6 (5%) 117 2000 18 (5%) 21 (5%) 47 (12%) 387 0 9 (6%) 2 (1%) 150 2001 16 (4%) 22 (5%) 50 (11%) 435 0 11 (6%) 2 (1%) 183 2002 23 (5%) 22 (5%) 24 (5%) 476 3 (1%) 13 (6%) 4 (2%) 208 2003 24 (5%) 30 (6%) 32 (6%) 533 4 (2%) 13 (6%) 1 (0%) 221 2004 27 (5%) 40 (7%) 18 (3%) 597 1 (0%) 14 (5%) 0 273 2005 41 (6%) 34 (5%) 20 (3%) 631 3 (1%) 11 (4%) 1 (0%) 294 2006 24 (6%) 18 (4%) 14 (3%) 428 6 (3%) 9 (4%) 0 233 2007 30 (7%) 23 (5%) 10 (2%) 423 0 3 (1%) 1 (0%) 249 2008 33 (7%) 36 (8%) 12 (3%) 474 1 (0%) 7 (3%) 1 (0%) 263 2009 42 (10%) 44 (10%) 13 (3%) 436 5 (2%) 8 (3%) 0 285 2010 33 (7%) 50 (10%) 18 (4%) 481 4 (1%) 8 (3%) 0 282 2011 47 (11%) 33 (8%) 9 (2%) 420 4 (1%) 4 (1%) 0 283 Year Riske Creek site Knife Creek site Mountain Bluebird Tree Swallow European Starling Suitable cavity # Mountain Bluebird Tree Swallow European Starling Suitable cavity # 1995 17 (17%) 10 (10%) 9 (9%) 98 – – – – 1996 12 (9%) 10 (8%) 9 (7%) 132 0 1 (8%) 0 12 1997 14 (7%) 12 (6%) 29 (14%) 212 2 (5%) 9 (22%) 1 (2%) 40 1998 14 (5%) 17 (6%) 39 (15%) 267 3 (4%) 8 (10%) 1 (1%) 80 1999 8 (3%) 15 (5%) 37 (12%) 317 3 (3%) 7 (6%) 6 (5%) 117 2000 18 (5%) 21 (5%) 47 (12%) 387 0 9 (6%) 2 (1%) 150 2001 16 (4%) 22 (5%) 50 (11%) 435 0 11 (6%) 2 (1%) 183 2002 23 (5%) 22 (5%) 24 (5%) 476 3 (1%) 13 (6%) 4 (2%) 208 2003 24 (5%) 30 (6%) 32 (6%) 533 4 (2%) 13 (6%) 1 (0%) 221 2004 27 (5%) 40 (7%) 18 (3%) 597 1 (0%) 14 (5%) 0 273 2005 41 (6%) 34 (5%) 20 (3%) 631 3 (1%) 11 (4%) 1 (0%) 294 2006 24 (6%) 18 (4%) 14 (3%) 428 6 (3%) 9 (4%) 0 233 2007 30 (7%) 23 (5%) 10 (2%) 423 0 3 (1%) 1 (0%) 249 2008 33 (7%) 36 (8%) 12 (3%) 474 1 (0%) 7 (3%) 1 (0%) 263 2009 42 (10%) 44 (10%) 13 (3%) 436 5 (2%) 8 (3%) 0 285 2010 33 (7%) 50 (10%) 18 (4%) 481 4 (1%) 8 (3%) 0 282 2011 47 (11%) 33 (8%) 9 (2%) 420 4 (1%) 4 (1%) 0 283 Open in new tab To calculate the mean (±95% confidence interval [CI]) annual breeding dates reported in Figure 1, we generated 500 datasets using MI (see above) and calculated mean annual initiation dates and standard deviations for each run. Reported values are the averages of these 500 runs. We report the mean annual clutch initiation date for each species as “early” or “late” relative to the multi-year average for that species (Figure 1, green line). FIGURE 1. Open in new tabDownload slide Mean annual clutch initiation dates (days after April 1 ± 95% CI) for (A) Mountain Bluebirds and (B) Tree Swallows in south-central British Columbia. The green horizontal line represents the inter-annual average clutch initiation date for each species, with Tree Swallows initiating breeding, on average, 10 days later than Mountain Bluebird. Annual sample sizes can be found in Table 1. FIGURE 1. Open in new tabDownload slide Mean annual clutch initiation dates (days after April 1 ± 95% CI) for (A) Mountain Bluebirds and (B) Tree Swallows in south-central British Columbia. The green horizontal line represents the inter-annual average clutch initiation date for each species, with Tree Swallows initiating breeding, on average, 10 days later than Mountain Bluebird. Annual sample sizes can be found in Table 1. Weather Variables We obtained local rainfall data (mm) and daily minimum, maximum, and mean temperatures (°C) for the pre-breeding period from the Environment and Climate Change Canada weather station Williams Lake A (World Meteorological Organization [WMO] ID 71104; 52.1800°N, 122.0500°W; elevation 939.7 m; http://climate.weather.gc.ca) within 40 km of our field sites (see also Drake and Martin 2018). These values corresponded with incomplete data from a British Columbia Wildfire Service station at the Riske Creek study site (51.9603°N, 122.5000°W; elevation 929 m; Station 210) (Pearson’s r  =  0.80 for precipitation; Pearson’s r  =  0.95 and 0.97 for minimum and mean temperature, respectively). We obtained winter and migratory weather variables by specifying winter and migratory regions for Mountain Bluebird and Tree Swallow using NatureServe range maps (Ridgely et al. 2005), band return data (Environment and Climate Change Canada 2016), and geolocator data (Knight et al. 2018) for Tree Swallows (for maps, see Appendix). For Mountain Bluebird, we used the entire winter and migratory regions defined within this species’ NatureServe range map and also included breeding areas south of our study site within our defined migratory region. For the more broadly distributed Tree Swallow, geolocator (Knight et al. 2018) and band return data (Environment and Climate Change Canada 2016) supported the assumption that the individuals in British Columbia are a part of populations that remain in western North America throughout their annual cycle. We therefore restricted winter and migration ranges to the western United States and Mexico. We then collected spatially coded temperature, precipitation (surface), and wind vector data (averaged over the 850 and 925 millibar (mb) level [~1,500 and 700 m above mean sea level; Alerstam et al. 2011, Drake et al. 2014b, Huang et al. 2017]) from 1994 to 2011 from the National Center of Environmental Prediction (NCEP) Reanalysis 1 data archives at the National Oceanic and Atmospheric Administration-Cooperative Institute for Research in Environmental Sciences (NOAA-CIRES) Climate Diagnostics Center at Boulder, Colorado, USA (Kalnay et al. 1996) using R (package RNCEP; Kemp et al. 2012). In QGIS 2.10 (QGIS Development Team 2015), we restricted these data to spatial regions that overlapped with our defined wintering and migratory regions. For each species, we then calculated mean daily temperatures and precipitation over the entire winter region. We divided migratory regions into latitudinal blocks: 40–50°N, 30–40°N, and (for Tree Swallow only) 20–30°N based on the strong correlation of our weather variables within these blocks (that justified averaging), but not between them. Within each block of the migratory region we calculated mean daily temperature, precipitation, and (because both species are daytime migrants; Evans and O’Brien 2002, Winkler et al. 2011) diurnal (0600 to 1800 hours) east-west (U-) and north-south (V-) wind vectors. Nest Site Availability and Competition The availability of nest cavities changed dramatically over the 17-yr study period, with the fewest cavities available in 1996, and the most in 2005 (Cockle and Martin 2015). This change was largely due to a mountain pine beetle outbreak that peaked in 2005 (Cockle and Martin 2015). This outbreak killed pine trees and resulted in an increased density in primary excavators at the study site (Edworthy et al. 2011). Cavity availability subsequently declined with lower excavation rates and natural loss, post-outbreak. We quantified cavity availability for each breeding period (year) at our Riske and Knife Creek monitoring sites as the number of suitable cavities pre-existing, plus those identified in the current year, minus those lost over the previous winter (Table 1). The suitability of these cavities for our study species was based on observed use: we excluded all cavities made by excavators for which our focal species were rarely secondary occupants (<5% use by availability; e.g., Pileated Woodpecker [Dryocopus pileatus]) and, where excavator was unknown or the cavity naturally formed, we excluded cavities that had been initially occupied by secondary nesting species for which our focal species were rarely secondary occupants (<5% use by availability; e.g., northern flying squirrel [Glaucomys sabrinus]). We quantified direct competition between European Starlings and Mountain Bluebirds and Tree Swallows (Koenig 2003, Wiebe 2003, Koch et al. 2012) as the proportion of suitable cavities occupied by nesting starlings in a given year at each of our monitoring sites (Table 1). We also tested whether the abundance of nesting Mountain Bluebirds might delay Tree Swallow nesting given the overlap in cavity preferences between these 2 species and the fact that Mountain Bluebirds initiate earlier than Tree Swallows (Koch et al. 2012, Wiebe 2016). As with European Starlings, we quantified Mountain Bluebird competition as the proportion of suitable cavities occupied by active Mountain Bluebird nests in a given year at each of our monitoring sites (Table 1). Analysis All analyses were run in R 3.6.0 (R Core Team 2019). We used a linear mixed modeling approach to describe individual clutch initiation date as a function of our explanatory variables: “Year” was included in all models as a random effect (intercept) to account for the non-independence of data collected within the same year (package lme4; Bates et al. 2015). Models were competed using Akaike information criterion (AIC). To limit our final model candidate set, we used a 2-step hierarchical approach to model testing. In the first step, we identified weather variables that corresponded with timing; in the second step, we created the final model candidate set that tested these weather variables together with cavity availability and nest site competition metrics. In step 1, the performance of local, wintering, and migratory weather variables (temperature, rainfall, and migratory region wind speed) was tested using a sliding window approach (package climwin; van de Pol et al. 2016) covering the appropriate time window of the annual cycle for each species (Winkler et al. 2011, Johnson and Dawson 2019) but allowing the period assessed for both winter and breeding to encompass the spring migratory period (Appendix Table 4). Minimum window size was restricted to 10 days (providing time for a physiological reproductive response [i.e. ova development] to conditions) and the maximum window encompassed the entire time period (Williams 2012, Williams et al. 2015, Drake and Martin 2018). The best time window for each variable was determined using AIC. Sliding windows, by definition, represent multiple comparisons and result in an increased probability of false positives. We therefore calculated the probability that the AIC scores we obtained were the product of chance (i.e. Type I error) by using a response-data randomization program included for this purpose within climwin. Specifically, we considered our results to be spurious if our model AIC values did not differ significantly (P > 0.05) from those generated from sliding window analyses of 100 randomizations of the response data (for details, see van de Pol et al. 2016). Only weather variables that passed the Type I error test and that fell within ΔAIC ≤ 2 of the top performing weather variable (Burnham and Anderson 2004) were included in our final candidate set of models. Our final candidate model set included: standardized (z-score transformed), top-performing weather variables, cavity availability, and competition metrics on their own as well as biologically plausible additive models that included weather, cavity availability, and competition together. All models included “Site” as a random effect (intercept), in addition to “Year” (intercept), to account for the non-independence of breeding records within each study site. These were run against a null (the random effect of “Year” and “Site” only) model and a model that assumed a linear trend in phenology over the monitoring period (i.e. “Year” as a continuous variable). The final candidate set contained 13 models for Mountain Bluebird and 17 models for Tree Swallow (Tables 2 and 3, respectively). We excluded 11 years (1996–2002, 2004–2005, and 2007–2008; 16 breeding records) at Knife Creek for Mountain Bluebird and 2 years (1996 and 2007; 4 breeding records) at Knife Creek for Tree Swallow due to small sample sizes (<4 records) at these sites in those years (Table 1). For model testing, we imputed 100 datasets (see above) and bootstrapped each of these (n  =  20 per dataset), resulting in 2,000 runs of the candidate model set (100 MI × 20 bootstraps) (Schomaker and Heumann 2014). The influence (β value) of each candidate variable was then model-averaged over the entire candidate set within each run using AIC weights (wi) (Burnham and Anderson 2002, Burnham et al. 2011). The resulting value distributions were used to obtain our coefficients (Schomaker and Heumann 2014): reported beta values (β) are the median, model-averaged, β for each predictor variable, and the reported 85% and 95% CIs for these values are the 0.075–0.925 and 0.025–0.975 quantiles of their empirical distribution from the 2,000 MI × bootstrap runs (Figure 2). Variables with beta estimates whose 85% CI do not cross zero contribute to model fit under AIC selection criteria (Arnold 2010). We report 95% CIs for comparison with traditional P-value testing. To support our missing data approach, we re-ran our candidate models using only clutch initiation dates that were recorded in the field. These results are reported in the Appendix and do not differ substantively from the full analysis (Appendix Tables 5 and 6 and Appendix Figure 5). TABLE 2. Mountain Bluebird final candidate model ranking (n = 446 nests). Values under model parameters are standardized beta values. All values represent the median of 2,000 MI-bootstrap runs (see Methods). Null and annual trend models are italicized. Top models (≤ 4 ΔAIC) are in bold and represent 95% of total model support. k = number of parameters and wi = AIC weight. Model Intercept Year (trend) Cavity availability Starling occupancy Rainfall 40–50°N Westerly wind 40–50°N k loglik ΔAIC a wi MOBL7 147.70 – –3.18 – 1.73 – 6 –1,765.56 0.00 0.28 MOBL11 147.74 – –3.33 – – 1.46 6 –1,765.97 0.81 0.19 MOBL9 147.55 – –3.09 0.26 1.70 – 7 –1,765.16 1.26 0.15 MOBL3 148.01 – –4.21 – – – 5 –1,767.24 1.30 0.15 MOBL13 147.48 – –3.11 0.45 – 1.47 7 –1,765.46 1.86 0.11 MOBL5 147.81 – –4.02 0.38 – – 6 –1,766.89 2.65 0.07 MOBL6 145.85 – – – 3.49 – 5 –1,769.51 5.84 0.02 MOBL8 145.42 – – 0.99 3.15 – 6 –1,768.91 6.70 0.01 MOBL10 145.86 – – – – 3.37 5 –1,770.04 6.89 0.01 MOBL12 145.31 – – 1.32 – 3.06 6 –1,769.07 7.02 0.01 MOBL2 146.11 –3.24 – – – – 5 –1,770.49 7.80 0.01 MOBL4 144.91 – – 1.82 – – 5 –1,772.93 12.67 0.00 MOBL1 145.76 – – – – – 4 –1,774.10 12.98 0.00 Model Intercept Year (trend) Cavity availability Starling occupancy Rainfall 40–50°N Westerly wind 40–50°N k loglik ΔAIC a wi MOBL7 147.70 – –3.18 – 1.73 – 6 –1,765.56 0.00 0.28 MOBL11 147.74 – –3.33 – – 1.46 6 –1,765.97 0.81 0.19 MOBL9 147.55 – –3.09 0.26 1.70 – 7 –1,765.16 1.26 0.15 MOBL3 148.01 – –4.21 – – – 5 –1,767.24 1.30 0.15 MOBL13 147.48 – –3.11 0.45 – 1.47 7 –1,765.46 1.86 0.11 MOBL5 147.81 – –4.02 0.38 – – 6 –1,766.89 2.65 0.07 MOBL6 145.85 – – – 3.49 – 5 –1,769.51 5.84 0.02 MOBL8 145.42 – – 0.99 3.15 – 6 –1,768.91 6.70 0.01 MOBL10 145.86 – – – – 3.37 5 –1,770.04 6.89 0.01 MOBL12 145.31 – – 1.32 – 3.06 6 –1,769.07 7.02 0.01 MOBL2 146.11 –3.24 – – – – 5 –1,770.49 7.80 0.01 MOBL4 144.91 – – 1.82 – – 5 –1,772.93 12.67 0.00 MOBL1 145.76 – – – – – 4 –1,774.10 12.98 0.00 a Lowest median AIC value was 3,543.32. Open in new tab TABLE 2. Mountain Bluebird final candidate model ranking (n = 446 nests). Values under model parameters are standardized beta values. All values represent the median of 2,000 MI-bootstrap runs (see Methods). Null and annual trend models are italicized. Top models (≤ 4 ΔAIC) are in bold and represent 95% of total model support. k = number of parameters and wi = AIC weight. Model Intercept Year (trend) Cavity availability Starling occupancy Rainfall 40–50°N Westerly wind 40–50°N k loglik ΔAIC a wi MOBL7 147.70 – –3.18 – 1.73 – 6 –1,765.56 0.00 0.28 MOBL11 147.74 – –3.33 – – 1.46 6 –1,765.97 0.81 0.19 MOBL9 147.55 – –3.09 0.26 1.70 – 7 –1,765.16 1.26 0.15 MOBL3 148.01 – –4.21 – – – 5 –1,767.24 1.30 0.15 MOBL13 147.48 – –3.11 0.45 – 1.47 7 –1,765.46 1.86 0.11 MOBL5 147.81 – –4.02 0.38 – – 6 –1,766.89 2.65 0.07 MOBL6 145.85 – – – 3.49 – 5 –1,769.51 5.84 0.02 MOBL8 145.42 – – 0.99 3.15 – 6 –1,768.91 6.70 0.01 MOBL10 145.86 – – – – 3.37 5 –1,770.04 6.89 0.01 MOBL12 145.31 – – 1.32 – 3.06 6 –1,769.07 7.02 0.01 MOBL2 146.11 –3.24 – – – – 5 –1,770.49 7.80 0.01 MOBL4 144.91 – – 1.82 – – 5 –1,772.93 12.67 0.00 MOBL1 145.76 – – – – – 4 –1,774.10 12.98 0.00 Model Intercept Year (trend) Cavity availability Starling occupancy Rainfall 40–50°N Westerly wind 40–50°N k loglik ΔAIC a wi MOBL7 147.70 – –3.18 – 1.73 – 6 –1,765.56 0.00 0.28 MOBL11 147.74 – –3.33 – – 1.46 6 –1,765.97 0.81 0.19 MOBL9 147.55 – –3.09 0.26 1.70 – 7 –1,765.16 1.26 0.15 MOBL3 148.01 – –4.21 – – – 5 –1,767.24 1.30 0.15 MOBL13 147.48 – –3.11 0.45 – 1.47 7 –1,765.46 1.86 0.11 MOBL5 147.81 – –4.02 0.38 – – 6 –1,766.89 2.65 0.07 MOBL6 145.85 – – – 3.49 – 5 –1,769.51 5.84 0.02 MOBL8 145.42 – – 0.99 3.15 – 6 –1,768.91 6.70 0.01 MOBL10 145.86 – – – – 3.37 5 –1,770.04 6.89 0.01 MOBL12 145.31 – – 1.32 – 3.06 6 –1,769.07 7.02 0.01 MOBL2 146.11 –3.24 – – – – 5 –1,770.49 7.80 0.01 MOBL4 144.91 – – 1.82 – – 5 –1,772.93 12.67 0.00 MOBL1 145.76 – – – – – 4 –1,774.10 12.98 0.00 a Lowest median AIC value was 3,543.32. Open in new tab TABLE 3. Tree Swallow final candidate model ranking (n = 568 nests). Values under model parameters are standardized beta values. All values represent the median of 2,000 MI-bootstrap runs (see Methods). Null and annual trend models are italicized. Top models (≤ 4 ΔAIC) are in bold and represent 97% of total model support. k = number of parameters and wi = AIC weight. Model Intercept Year (trend) Local rainfall Cavity availability Starling occupancy Bluebird occupancy k loglik ΔAIC a wi TRES8 156.18 – 2.55 – 1.80 – 6 –2,114.50 0.00 0.32 TRES14 156.36 – 2.46 –0.61 1.78 – 7 –2,113.59 0.24 0.28 TRES15 156.28 – 2.57 – 1.80 –0.40 7 –2,113.89 0.83 0.21 TRES17 156.40 – 2.48 –0.53 1.80 –0.23 8 –2,113.09 1.28 0.17 TRES3 156.37 – 2.80 – – – 5 –2,119.27 7.49 0.01 TRES7 156.44 – 2.66 –0.72 – – 6 –2,118.57 8.13 0.01 TRES13 156.45 – 2.64 –1.04 – –0.49 7 –2,117.73 8.52 0.00 TRES10 156.38 – 2.81 – – –0.11 6 –2,118.89 8.78 0.00 TRES9 156.32 – – –1.11 2.22 – 6 –2,119.10 9.20 0.00 TRES5 156.13 – – – 2.05 – 5 –2,120.76 10.47 0.00 TRES16 156.32 – – –1.11 2.22 0.00 7 –2,118.78 10.60 0.00 TRES12 156.20 – – – 2.06 –0.42 6 –2,120.29 11.57 0.00 TRES2 156.55 –2.18 – – – – 5 –2,122.60 14.16 0.00 TRES4 156.28 – – –1.61 – – 5 –2,124.02 17.00 0.00 TRES1 156.34 – – – – – 4 –2,125.14 17.19 0.00 TRES11 156.30 – – –2.32 – –1.08 6 –2,123.22 17.44 0.00 TRES6 156.32 – – – – 0.11 5 –2,124.86 18.68 0.00 Model Intercept Year (trend) Local rainfall Cavity availability Starling occupancy Bluebird occupancy k loglik ΔAIC a wi TRES8 156.18 – 2.55 – 1.80 – 6 –2,114.50 0.00 0.32 TRES14 156.36 – 2.46 –0.61 1.78 – 7 –2,113.59 0.24 0.28 TRES15 156.28 – 2.57 – 1.80 –0.40 7 –2,113.89 0.83 0.21 TRES17 156.40 – 2.48 –0.53 1.80 –0.23 8 –2,113.09 1.28 0.17 TRES3 156.37 – 2.80 – – – 5 –2,119.27 7.49 0.01 TRES7 156.44 – 2.66 –0.72 – – 6 –2,118.57 8.13 0.01 TRES13 156.45 – 2.64 –1.04 – –0.49 7 –2,117.73 8.52 0.00 TRES10 156.38 – 2.81 – – –0.11 6 –2,118.89 8.78 0.00 TRES9 156.32 – – –1.11 2.22 – 6 –2,119.10 9.20 0.00 TRES5 156.13 – – – 2.05 – 5 –2,120.76 10.47 0.00 TRES16 156.32 – – –1.11 2.22 0.00 7 –2,118.78 10.60 0.00 TRES12 156.20 – – – 2.06 –0.42 6 –2,120.29 11.57 0.00 TRES2 156.55 –2.18 – – – – 5 –2,122.60 14.16 0.00 TRES4 156.28 – – –1.61 – – 5 –2,124.02 17.00 0.00 TRES1 156.34 – – – – – 4 –2,125.14 17.19 0.00 TRES11 156.30 – – –2.32 – –1.08 6 –2,123.22 17.44 0.00 TRES6 156.32 – – – – 0.11 5 –2,124.86 18.68 0.00 a Lowest median AIC value was 4,241.15. Open in new tab TABLE 3. Tree Swallow final candidate model ranking (n = 568 nests). Values under model parameters are standardized beta values. All values represent the median of 2,000 MI-bootstrap runs (see Methods). Null and annual trend models are italicized. Top models (≤ 4 ΔAIC) are in bold and represent 97% of total model support. k = number of parameters and wi = AIC weight. Model Intercept Year (trend) Local rainfall Cavity availability Starling occupancy Bluebird occupancy k loglik ΔAIC a wi TRES8 156.18 – 2.55 – 1.80 – 6 –2,114.50 0.00 0.32 TRES14 156.36 – 2.46 –0.61 1.78 – 7 –2,113.59 0.24 0.28 TRES15 156.28 – 2.57 – 1.80 –0.40 7 –2,113.89 0.83 0.21 TRES17 156.40 – 2.48 –0.53 1.80 –0.23 8 –2,113.09 1.28 0.17 TRES3 156.37 – 2.80 – – – 5 –2,119.27 7.49 0.01 TRES7 156.44 – 2.66 –0.72 – – 6 –2,118.57 8.13 0.01 TRES13 156.45 – 2.64 –1.04 – –0.49 7 –2,117.73 8.52 0.00 TRES10 156.38 – 2.81 – – –0.11 6 –2,118.89 8.78 0.00 TRES9 156.32 – – –1.11 2.22 – 6 –2,119.10 9.20 0.00 TRES5 156.13 – – – 2.05 – 5 –2,120.76 10.47 0.00 TRES16 156.32 – – –1.11 2.22 0.00 7 –2,118.78 10.60 0.00 TRES12 156.20 – – – 2.06 –0.42 6 –2,120.29 11.57 0.00 TRES2 156.55 –2.18 – – – – 5 –2,122.60 14.16 0.00 TRES4 156.28 – – –1.61 – – 5 –2,124.02 17.00 0.00 TRES1 156.34 – – – – – 4 –2,125.14 17.19 0.00 TRES11 156.30 – – –2.32 – –1.08 6 –2,123.22 17.44 0.00 TRES6 156.32 – – – – 0.11 5 –2,124.86 18.68 0.00 Model Intercept Year (trend) Local rainfall Cavity availability Starling occupancy Bluebird occupancy k loglik ΔAIC a wi TRES8 156.18 – 2.55 – 1.80 – 6 –2,114.50 0.00 0.32 TRES14 156.36 – 2.46 –0.61 1.78 – 7 –2,113.59 0.24 0.28 TRES15 156.28 – 2.57 – 1.80 –0.40 7 –2,113.89 0.83 0.21 TRES17 156.40 – 2.48 –0.53 1.80 –0.23 8 –2,113.09 1.28 0.17 TRES3 156.37 – 2.80 – – – 5 –2,119.27 7.49 0.01 TRES7 156.44 – 2.66 –0.72 – – 6 –2,118.57 8.13 0.01 TRES13 156.45 – 2.64 –1.04 – –0.49 7 –2,117.73 8.52 0.00 TRES10 156.38 – 2.81 – – –0.11 6 –2,118.89 8.78 0.00 TRES9 156.32 – – –1.11 2.22 – 6 –2,119.10 9.20 0.00 TRES5 156.13 – – – 2.05 – 5 –2,120.76 10.47 0.00 TRES16 156.32 – – –1.11 2.22 0.00 7 –2,118.78 10.60 0.00 TRES12 156.20 – – – 2.06 –0.42 6 –2,120.29 11.57 0.00 TRES2 156.55 –2.18 – – – – 5 –2,122.60 14.16 0.00 TRES4 156.28 – – –1.61 – – 5 –2,124.02 17.00 0.00 TRES1 156.34 – – – – – 4 –2,125.14 17.19 0.00 TRES11 156.30 – – –2.32 – –1.08 6 –2,123.22 17.44 0.00 TRES6 156.32 – – – – 0.11 5 –2,124.86 18.68 0.00 a Lowest median AIC value was 4,241.15. Open in new tab FIGURE 2. Open in new tabDownload slide Model-averaged standardized beta (β) values and their 85% and 95% confidence intervals (CIs) for weather and ecological variables expected to impact annual average clutch initiation dates as well as a linear change in phenology over the monitoring period for (A) Mountain Bluebird and (B) Tree Swallows. Positive beta values indicate breeding delays associated with increasing values of a given variable, negative beta values indicate breeding advancement. CIs are the empirical distribution of beta values for 2,000 MI-bootstrap runs (see Methods). Non-symmetrical CIs are due to model averaging: for example, beta values for a linear change in breeding dates over the monitoring period were consistently negative, but this model received little support within either species’ candidate set (wi < 0.01; Tables 2 and 3, Model 2). Model averaging therefore drove beta values towards zero in many MI-bootstrap runs. FIGURE 2. Open in new tabDownload slide Model-averaged standardized beta (β) values and their 85% and 95% confidence intervals (CIs) for weather and ecological variables expected to impact annual average clutch initiation dates as well as a linear change in phenology over the monitoring period for (A) Mountain Bluebird and (B) Tree Swallows. Positive beta values indicate breeding delays associated with increasing values of a given variable, negative beta values indicate breeding advancement. CIs are the empirical distribution of beta values for 2,000 MI-bootstrap runs (see Methods). Non-symmetrical CIs are due to model averaging: for example, beta values for a linear change in breeding dates over the monitoring period were consistently negative, but this model received little support within either species’ candidate set (wi < 0.01; Tables 2 and 3, Model 2). Model averaging therefore drove beta values towards zero in many MI-bootstrap runs. Clutch Initiation Date and Breeding Productivity We tested the impact of clutch initiation date on productivity for a subset of records for which we had data on clutch size and nest outcome (n = 229 for Mountain Bluebird and n = 177 for Tree Swallow). All of these records were from 2000 to 2011 (see Appendix Table 7 for annual sample sizes and productivity). We assumed timing could impact productivity at 2 points in a pathway between lay and fledging: (1) by influencing clutch size (nest investment) and then (2) by influencing nestling survival to fledging. We split timing into 2 components. First, we used mean clutch initiation dates within each year to assess the impact of “late” vs. “early” years for the population as a whole. We then calculated individual deviation from these annual means (clutch initiation date-annual average clutch initiation date) to assess the individual impact of being late or early relative to conspecifics, within year. For our productivity pathway, we tested (1) whether clutch size varied as a function of mean annual clutch initiation date and individual deviation from that mean. Clutch size was fit using a Conway-Maxwell-Poisson distribution to account for under-dispersion (package spaMM; Rousset and Ferdy 2014). We then tested (2) whether the number of fledglings produced by each nest varied as a function of clutch size, mean annual clutch initiation date, and individual deviation from that mean. Fledgling number best fit a Poisson distribution with zero inflation. We therefore tested both zero-inflated Poisson and Poisson hurdle models: hurdle models performed best for both species. Preliminary analysis additionally indicated that clutch size did not contribute to the binomial component of the hurdle model in Tree Swallows (i.e. nest failure) and we therefore dropped this term. Year was included in all models as a random effect (intercept only) to account for the non-independence of records within-year. Reported coefficients and standard error (SE) values (from which we calculated CIs) are the medians of 500 MI of each dataset. We additionally ran a simpler test for population-level effects of breeding delay by regressing mean fledge number within-year against mean annual clutch initiation date (n = 14, 1997–2011) for both species and weighting the response values by annual sample size. RESULTS Over our 17-yr monitoring period, average annual clutch initiation dates for Mountain Bluebird ranged from May 19 to June 8 (20 days) with an inter-annual average of May 26, while annual average dates for Tree Swallow ranged from May 30 to June 12 (13 days) with an inter-annual average date of June 6 (Figure 1). Among Mountain Bluebird, 80% of individual lay dates were within a 32-day window around this species’ annual average lay date (14 days prior to 18 days after). For Tree Swallows, 80% of lay dates were within a 25-day window (11 days prior to 14 days after the annual average lay date). Thus, phenology differences among individuals within-year were much larger than the observed differences in timing among years for these populations as a whole. The abundance of nesting Mountain Bluebirds and Tree Swallows fluctuated 5- and 3-fold, respectively, between 1998 and 2011 (where survey effort was constant). At Riske Creek, both species showed an increasing trend that occurred concurrently with increasing cavity abundance, and local lows in 2006–2007 that corresponded with decreased cavity availability. As a result, cavity occupancy by availability remained relatively stable over time (Table 1). The abundance of nesting European Starlings fluctuated 6-fold between 1998 and 2011, reaching a peak in 2000–2001, before collapsing and remaining at low numbers up to 2011 (Table 1). Factors Influencing Clutch Initiation Date Mountain Bluebird. Sliding window analyses of the correlation between annual timing and weather identified conditions in the northern portion (40–50°N) of the migratory range as possible predictors of Mountain Bluebird phenology. Greater rainfall and higher wind speeds within this region were associated with later lay dates for this species (best temporal windows: February 4 to March 24 and February 16–26, respectively). These 2 variables were included in the final candidate model set; other weather variables were either not competitive or failed to pass Type I error tests (Appendix Table 4A). Mountain Bluebird phenology was most strongly correlated with cavity number, with earlier breeding occurring in years with greater cavity availability (β = –3.05, Figures 2A and 3A). Weaker effects of migratory conditions between 40 and 50°N in the western US states (Washington State, Oregon, and Idaho; Appendix Figure 4) were indicated, with later breeding occurring in higher rainfall and stronger westerly wind speed years (rainfall β = 0.64; westerly wind speed β = 0.37 [only the 85% CI did not cross zero]; Figure 2A). Both the 85% and 95% CIs of European Starling competition crossed zero and we conclude that this factor did not predict timing (Figure 2A). There was a non-zero effect of year, indicating advancement in breeding dates over the monitoring period; however, this model performed poorly relative to our other candidate models and its model-averaged beta value was close to zero (β = –0.02 or a 0.05 day advancement over the 17-yr period; Figure 2A and Table 2). In real terms (and holding all other variables constant), our results indicated that Mountain Bluebirds advanced nesting by 3.8 days for every 200 additional cavities available, and shifted breeding by 10.2 and 5.4 days across the cavity extremes at Riske and Knife Creek, respectively (Figure 3A). Breeding delays of 1.5 days were associated with every additional 1 mm of daily rainfall in the northern migratory region and breeding dates shifted by 2.1 days across the precipitation extremes in our dataset. Delays of 1.1 days occurred when westerly winds increased by 5 m/s within the northern migratory region and breeding dates shifted by 1.3 days across the extremes in our dataset. Tree Swallow. Sliding window analyses of the correlation between annual timing and weather identified local spring rainfall at the study site (best temporal window: March 13 to May 22) as a possible predictor of phenology in Tree Swallow. Greater spring rainfall was associated with later lay dates; no other weather variables were competitive and only this variable was included in the final candidate model set (Appendix Table 4B). Tree Swallow phenology was most strongly correlated with spring rainfall at the study site, with delays occurring in higher rainfall years (β = 2.48; Figures 2B and 3B). European Starling prevalence was also correlated with breeding delays (β = 1.74; Figure 2B). Changes in cavity availability and nesting Mountain Bluebird abundance were not correlated with Tree Swallow timing (Figure 2B). As in our Mountain Bluebird models, there was a non-zero relationship with “Year”, indicating an advancement in breeding dates over the monitoring period; however, as with Mountain Bluebirds, this temporal model performed poorly relative to the other candidate models and its model-averaged beta value was negligible (β = –0.0006 or a 0.002 day advancement over the 17-yr period; Figure 2B and Table 3). In real terms (and holding all other variables constant), our results indicated that Tree Swallow nesting was delayed by 8.6 days when average daily local rainfall between March 13 and May 22 increased by 1 mm and shifted by 8.2 days across the extremes of rainfall in our 17-yr monitoring period (Figure 3B). Tree Swallows additionally showed a 4.0-day delay in clutch initiation with each 0.1 increase in the proportion of nest sites occupied by starlings; breeding dates shifted by 4.9 and 2.0 days across the extremes of starling occupancy at Riske and Knife Creek, respectively. Productivity and Clutch Initiation Dates Within our subset of records with known outcomes, Mountain Bluebirds (n = 229 over 12 yr) laid an average of 5.2 eggs per clutch (range: 2–7) and fledged an average of 2.4 young (range: 0–7); 61% of nests were successful (fledged at least 1 young). Tree Swallows (n = 177 over 12 yr) laid an average of 5.4 eggs per clutch (range: 2–8) and fledged an average of 2.1 young (range: 0–7); 50% of nests were successful (see Appendix Table 7 for annual values). Variation in annual average clutch initiation date was not associated with shifts in clutch size for either of our focal species (Mountain Bluebird β [95% CI]: –0.15 [0.25, –0.56]; Tree Swallow: –0.09 [0.31, –0.50]). Individual deviation from annual average clutch initiation dates was associated with egg number, such that those individuals breeding late, relative to their conspecifics within-year, laid smaller clutches (Mountain Bluebird: –0.34 [–0.17, –0.51]; Tree Swallow: –0.25 [–0.07, –0.42]). For Mountain Bluebird, individual delay within-year (β [95% CI]: 0.42 [0.13, 0.71]) and possibly larger clutch sizes (0.29 [–0.02, 0.60]) were associated with an increased probability of nest failure (the binomial portion of the fledgling hurdle model), while annual average initiation date was not (–0.05 [–0.73, 0.62]). After accounting for nest failure, clutch size alone predicted ultimate fledge number (0.14 [0.05, 0.23]). As a result, being late within-year reduced individual Mountain Bluebird productivity first via decreased clutch size, and second via a direct relationship between timing and nest failure. For Tree Swallow, individual delay within-year was possibly associated with nest failure (β [95% CI]: 0.26 [–0.07, 0.59]; 85% CI does not cross zero) while annual average initiation was not (0.06 [–0.56, 0.69]). After accounting for complete nest failure, clutch size alone predicted ultimate fledge number (0.15 [0.04, 0.27]). Thus, although the relationships were weaker, being late within-year impacted individual Tree Swallow productivity in the same manner as Mountain Bluebird: first via decreased clutch size, and second via a direct relationship between timing and nest failure. Our simpler analysis supported the findings above, with no relationship between mean annual fledge number and mean annual clutch initiation date for either species (Mountain Bluebird: r2 = 0.01, P = 0.31; Tree Swallow: r2 = –0.04, P = 0.52; n = 14 yr). DISCUSSION Our analyses examined the relative importance of weather cues (both breeding and nonbreeding) and ecological constraints in determining breeding phenology in 2 migratory, cavity-nesting species. Despite their migratory status, the timing of breeding for both Mountain Bluebirds and Tree Swallows in our study showed the strongest relationships with breeding ground factors: cavity availability in the case of Mountain Bluebirds, and spring rainfall and European Starling competition in the case of Tree Swallows. An effect of migratory conditions on timing—in the form of heavier rainfall and stronger westerly winds aloft, in the region immediately south of our breeding site—was also indicated for Mountain Bluebirds. The temporal window for these effects (February and March) are consistent with when our birds were most likely moving through this region, based on arrival dates for this species within the Williams Lake area (Johnson and Dawson 2019); standardized beta values put these migratory impacts at 5 × and 8 × weaker than that of cavity availability. The shift to earlier breeding by Mountain Bluebirds when cavities were more numerous may result from an increased opportunity to find suitable nest sites, and possibly nest sites away from conspecifics (Holt and Martin 1997). Corresponding changes in the number of breeding pairs at Riske Creek as cavities became more or less numerous would support the idea that our study sites were saturated and that more potential breeders were able to obtain a nesting site upon arrival from migration when cavity numbers permitted this. Alternatively, increased bluebird abundance might indicate that our sites were more desirable at the landscape level in some years, due to cavity number increases, or other changes associated with the mountain pine beetle outbreak (such as forest structure or food availability; Martin et al. 2006). Earlier breeding in these years might then be the product of preferential settlement. Tree Swallows did not show breeding advancement as cavities became more numerous. This is consistent with previous work where Tree Swallows did not show a numerical response to manipulations of cavity availability (Aitken and Martin 2008). Tree Swallows are more likely than Mountain Bluebirds to use cavities excavated by Red-naped Sapsuckers (Sphyrapicus nuchalis), which were the most abundant cavity-type at our field sites during the monitoring period (52% of all cavities; Robles and Martin 2013); and, unlike Mountain Bluebirds, Tree Swallows are not adverse to nesting in close proximity to each other (Holt and Martin 1997); thus, it is possible there was a sufficient number of suitable cavities for this species throughout the monitoring period. Tree Swallows did show delays with increasing European Starling but not Mountain Bluebird abundance at our study sites. Starling nest site preferences overlap with both of our study species, but starlings initiate clutches, on average, 1 week earlier than Mountain Bluebirds and 2 weeks earlier than Tree Swallows (Koch et al. 2012). This difference in phenology may explain the differences in the responses of our focal species to European Starling abundance: more starling pairs would have the advantage of prior ownership when Tree Swallows were seeking to obtain nest sites than when Mountain Bluebirds were acquiring nest sites. At the patch level, within our study site, early nesting in the presence of high starling densities has been associated with reduced breeding success for Tree Swallows, particularly when using flicker cavities (Koch et al. 2012, Robles and Martin 2013), and it is possible that swallows delay clutch initiation to reduce competition with starlings. The absence of an effect of Mountain Bluebird abundance on Tree Swallow timing may reflect the higher resource holding potential of Tree Swallows, as indicated by nest-box studies (Wiebe 2016). Weather effects on timing were evident in Tree Swallows, with wetter springs associated with late breeding. Our best temporal window for this weather effect ranged from March 13 to May 22, encompassing the period after these migrants arrived at the study site and just prior to when the majority initiated breeding. We do not have data on insect abundance but it is likely that rainfall delayed clutch initiation by reducing aerial insect availability for foraging adults (Winkler and Allen 1996, Irons et al. 2017). Aerial prey were less active on wet days (Poulsen 1996, Fournier et al. 2005, Grüebler et al. 2008) and resultant calorie limitation would slow female mass gain, post-arrival. Female Tree Swallows in eastern Canada now initiate breeding at a lower body mass than they did 20 yr ago, an effect that has been attributed to an increase in spring rainfall and therefore reduced food availability (Cox et al. 2019). Our calorie limitation hypothesis is also supported by negative relationships between rainfall during nesting development and parental provisioning, nestling mass, and survival (McCarty 2009, Cox et al. 2019). In our study region, frequency of rainfall during the breeding period is at its greatest in June, with an average of 11 days with >1 mm (1994–2011) during that month. Rainfall frequency then declines to 10 and 8 days in July and August, respectively. Wet springs are not correlated with wet summers, indicating that poor weather in the pre-breeding period does not provide Tree Swallows at our study site with any information as to conditions later in the season. In contrast to a continent-wide analysis (Dunn and Winkler 1999, Hussell 2003), we found no support for an effect of spring breeding ground temperatures on timing within our population (Appendix Table 4B). Wind and rainfall, rather than temperature, had the greatest impact on Tree Swallow breeding phenology in Alaska as well (Irons et al. 2017). Such differences may reflect regional differences in which environmental variables most limit aerial insect abundance. We would expect breeding delays associated with competition, nest site availability, and hostile local and migratory weather conditions to be costly. However, neither Mountain Bluebird nor Tree Swallow showed population-level reductions in productivity in late breeding years. It is probable that inter-annual shifts in timing (a maximum of 20 and 13 days, respectively) were not large enough to reduce productivity. If population-level delays had an equivalent impact on productivity as individual within-year delays, we would predict a decline of 0.8 fledglings for Mountain Bluebirds with a 20-day shift in lay date and a decline of 0.5 fledglings for Tree Swallows with a 13-day shift in lay date. However, population-level effects of timing on productivity are likely smaller than the effect of within-year delays for individuals. This is because late-breeding individuals within-year are often younger and/or have poorer quality territories than early-breeding individuals: factors that independently reduce productivity (Svensson and Nilsson 1995, Winkler and Allen 1996, Johnson and Dawson 2019). Mountain Bluebirds will attempt second broods if they initiate first broods early in the breeding period (Johnson and Dawson 2019) and, therefore, it is possible that the frequency of repeat nesting—and therefore total annual productivity—was lower at our sites for this species in late years. Our data did not allow us to assess this. Tree Swallows are single brooded in the northern parts of their range (Winkler et al. 2011) and our results therefore reflect annual productivity. Mountain Bluebird and Tree Swallow exhibited different responses to conditions than resident cavity-nesting species in south-central British Columbia. Like these 2 migrants, the majority of resident cavity nesters—a group including 4 woodpecker species and 3 small insectivores—showed marked phenological responses to local conditions; however, among residents, temperature was the key driver of timing as opposed to rainfall or cavity availability (Drake and Martin 2018). Among resident non-obligate excavators, Mountain Chickadee (Poecile gambeli; a secondary cavity nester) showed no response to cavity availability, while Red-breasted Nuthatch (Sitta canadensis; a facultative excavator) bred later as cavities became more numerous (Drake and Martin 2018). Our methodology allowed us to compare the relative strength of nonbreeding and local factors in determining the timing of breeding for our 2 migratory species. Our results indicate that breeding ground factors most strongly influence the phenology of Tree Swallows and Mountain Bluebirds. Nonbreeding conditions, often cited as a possible contributor to mistiming in migratory birds (e.g., Franks et al. 2018), had relatively little impact on clutch initiation dates over our 17-yr monitoring period. The fact that cavity number and the abundance of nest site competitors were strongly correlated with breeding phenology reinforces work that shows suitable cavities can be limiting for our focal species and emphasizes the importance of this resource for population management (Holt and Martin 1997, Wesolowski and Martin 2018). The relationship between increased local precipitation and breeding delays for Tree Swallow contributes to a body of literature showing that rainfall during the breeding period is costly for this species (Dawson 2008, Irons et al. 2017, Cox et al. 2019, but see McCarty and Winkler 1999). Despite these relationships, population-level shifts in timing among years were dwarfed by individual variation in timing within-year, indicating that individual-level factors (such as body condition, age, or territory quality) were more powerful drivers of timing for individuals than the range of external conditions they experienced within our study. Although population-level breeding delays are often assumed to result in similar productivity costs as within-year breeding delays for individuals, our work illustrates that such delays are not necessarily costly. FIGURE 3. Open in new tabDownload slide (A) Observed advancement in annual average Mountain Bluebird clutch initiation dates (days after April 1 ± 95% CI) as a function of increasing cavity availability. Timing is presented by study site (filled points = “Riske Creek”, hollow points = “Knife Creek”) because cavity availability differed between sites within year. (B) Observed delay in annual average Tree Swallow clutch initiation dates (days after April 1 ± 95% CI) as a function of increasing spring rainfall (average daily mm, March 13 to May 22). Numbers above error bars indicate study years (1995–2011). FIGURE 3. Open in new tabDownload slide (A) Observed advancement in annual average Mountain Bluebird clutch initiation dates (days after April 1 ± 95% CI) as a function of increasing cavity availability. Timing is presented by study site (filled points = “Riske Creek”, hollow points = “Knife Creek”) because cavity availability differed between sites within year. (B) Observed delay in annual average Tree Swallow clutch initiation dates (days after April 1 ± 95% CI) as a function of increasing spring rainfall (average daily mm, March 13 to May 22). Numbers above error bars indicate study years (1995–2011). Funding statement: Research funding came from the Natural Sciences and Engineering Research Council of Canada (NSERC), Environment and Climate Change Canada, the Sustainable Forest Management Network, part of Networks of Centres of Excellence of Canada, and Forest Renewal BC grants to K.M. Support for A.D. came from post-doctoral fellowships from TerreWEB (NSERC, UBC). Ethics statement: All protocols complied with the guidelines for the ethical use of animals and research, set forth by the Canadian Council on Animal Care, and were approved by the Animal Care and Biosafety Committee of the University of British Columbia (permit numbers: A04-0101 and A07-130). Author contributions: A.D. and K.M. conceived the idea and methods. K.M. collected data. A.D. analyzed the data and wrote the manuscript, K.M. edited the work. Data deposits: Analyses reported in this article can be reproduced using the data provided by Drake and Martin (2020). ACKNOWLEDGMENTS We thank numerous field assistants and graduate students who contributed to data collection. LITERATURE CITED Aitken , K. E. H. , and K. Martin ( 2007 ). The importance of excavators in hole-nesting communities: Availability and use of natural tree holes in old mixed forests of western Canada . Journal of Ornithology 148 : 425 – 434 . Google Scholar Crossref Search ADS WorldCat Aitken , K. E. H. , and K. Martin ( 2008 ). Resource selection plasticity and community responses to experimental reduction of a critical resource . Ecology 89 : 971 – 980 . Google Scholar Crossref Search ADS PubMed WorldCat Åkesson , S. , M. Ilieva , J. Karagicheva , E. Rakhimberdiev , B. Tomotani , and B. Helm ( 2017 ). Timing avian long-distance migration: From internal clock mechanisms to global flights . Philosophical Transactions of the Royal Society B: Biological Sciences 372 : 20160252 . Google Scholar Crossref Search ADS WorldCat Alerstam , T. , J. W. Chapman , J. Bäckman , A. D. Smith , H. Karlsson , C. Nilsson , D. R. Reynolds , R. H. G. Klaassen , and J. K. Hill ( 2011 ). Convergent patterns of long-distance nocturnal migration in noctuid moths and passerine birds . Proceedings of the Royal Society B: Biological Sciences 278 : 3074 – 3080 . Google Scholar Crossref Search ADS WorldCat Arnold , T. W . ( 2010 ). Uninformative parameters and model selection using Akaike’s Information Criterion . The Journal of Wildlife Management 74 : 1175 – 1178 . Google Scholar Crossref Search ADS WorldCat Bates , D. , M. Maechler , B. Bolker , and S. Walker ( 2015 ). Fitting linear mixed-effects models using lme4 . Journal of Statistical Software 67 : 1 – 48 . Google Scholar Crossref Search ADS WorldCat Both , C . ( 2010 ). Food availability, mistiming, and climatic change . In Effects of Climate Change on Birds (A. P. Møller, W. Fiedler, and P. Berthold, Editors). Oxford University Press , Oxford, UK. WorldCat Brown , W. P. , and R. R. Roth ( 2002 ). Temporal patterns of fitness and survival in the Wood Thrush . Ecology 83 : 958 – 969 . Google Scholar Crossref Search ADS WorldCat Burnham , K. P. , and D. R. Anderson ( 2002 ). Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach , 2nd edition. Springer-Verlag, New York, NY, USA . Google Preview WorldCat COPAC Burnham , K. P. , and D. R. Anderson ( 2004 ). Multimodel inference: Understanding AIC and BIC in model selection . Sociological Methods & Research 33 : 261 – 304 . Google Scholar Crossref Search ADS WorldCat Burnham , K. P. , D. R. Anderson , and K. P. Huyvaert ( 2011 ). AIC model selection and multimodel inference in behavioral ecology: Some background, observations, and comparisons . Behavioral Ecology and Sociobiology 65 : 23 – 35 . Google Scholar Crossref Search ADS WorldCat Cavalcanti , L. M. P. , L. V. de Paiva , L. F. França , L. M. P. Cavalcanti , L. V. de Paiva , and L. F. França ( 2016 ). Effects of rainfall on bird reproduction in a semi-arid Neotropical region . Zoologia (Curitiba) 33 : e20160018 . Google Scholar Crossref Search ADS WorldCat Cockle , K. L. , and K. Martin ( 2015 ). Temporal dynamics of a commensal network of cavity-nesting vertebrates: increased diversity during an insect outbreak . Ecology 96 : 1093 – 1104 . Google Scholar Crossref Search ADS PubMed WorldCat Cox , A. R. , R. J. Robertson , Á. Z. Lendvai , K. Everitt , and F. Bonier ( 2019 ). Rainy springs linked to poor nestling growth in a declining avian aerial insectivore (Tachycineta bicolor) . Proceedings of the Royal Society B: Biological Sciences 286 : 20190018 . Google Scholar Crossref Search ADS WorldCat Cresswell , W. , and R. McCleery ( 2003 ). How Great Tits maintain synchronization of their hatch date with food supply in response to long-term variability in temperature . Journal of Animal Ecology 72 : 356 – 366 . Google Scholar Crossref Search ADS WorldCat Dawson , R. D . ( 2008 ). Timing of breeding and environmental factors as determinants of reproductive performance of Tree Swallows . Canadian Journal of Zoology 86 : 843 – 850 . Google Scholar Crossref Search ADS WorldCat de Zwaan , D. R. , A. F. Camfield , E. C. MacDonald , and K. Martin ( 2019 ). Variation in offspring development is driven more by weather and maternal condition than predation risk . Functional Ecology 33 : 447 – 456 . Google Scholar Crossref Search ADS WorldCat Drake , A. , and K. Martin ( 2018 ). Local temperatures predict breeding phenology but do not result in breeding synchrony among a community of resident cavity-nesting birds . Scientific Reports 8 : 2756 . Google Scholar Crossref Search ADS PubMed WorldCat Drake , A. , and K. Martin ( 2020 ). Rainfall and nest site competition delay Mountain Bluebird and Tree Swallow breeding but do not impact productivity . The Auk: Ornithological Advances 137 :000–000. https://doi.org/10.5061/dryad.280gb5mmp. WorldCat Drake , A. , M. Martin , and D. J. Green ( 2014a ). Winter habitat use does not influence spring arrival dates or the reproductive success of Yellow Warblers breeding in the arctic . Polar Biology 37 : 181 – 191 . Google Scholar Crossref Search ADS WorldCat Drake , A. , C. A. Rock , S. P. Quinlan , M. Martin , and D. J. Green ( 2014b ). Wind speed during migration influences the survival, timing of breeding, and productivity of a Neotropical migrant, Setophaga petechia . PLoS One 9 : e97152 . Google Scholar Crossref Search ADS WorldCat Dunn , P. , and D. Winkler ( 1999 ). Climate change has affected the breeding date of Tree Swallows throughout North America . Proceedings of the Royal Society B: Biological Sciences 266 : 2487 – 2490 . Google Scholar Crossref Search ADS WorldCat Edworthy , A. B. , M. C. Drever , and K. Martin ( 2011 ). Woodpeckers increase in abundance but maintain fecundity in response to an outbreak of mountain pine bark beetles . Forest Ecology and Management 261 : 203 – 210 . Google Scholar Crossref Search ADS WorldCat Edworthy , A. B. , K. L. Wiebe , and K. Martin ( 2012 ). Survival analysis of a critical resource for cavity-nesting communities: Patterns of tree cavity longevity . Ecological Applications 22 : 1733 – 1742 . Google Scholar Crossref Search ADS PubMed WorldCat Environment and Climate Change Canada ( 2016 ). North American Bird Banding Program . Canadian Bird Banding Office, National Wildlife Research Centre, Carleton University, Ottawa, Canada . Google Preview WorldCat COPAC Evans , W. R. , and M. O’Brien ( 2002 ). Flight Calls of Migratory Birds: Eastern North American Landbirds . Old Bird Incorporated, Ithaca, NY, USA . http://www.oldbird.org. Google Preview WorldCat COPAC Fournier , F. , D. Pelletier , C. Vigneault , B. Goyette , and G. Boivin ( 2005 ). Effect of barometric pressure on flight initiation by Trichogramma pretiosum and Trichogramma evanescens (Hymenoptera: Trichogrammatidae) . Environmental Entomology 34 : 1534 – 1540 . Google Scholar Crossref Search ADS WorldCat Franks , S. E. , J. W. Pearce‐Higgins , S. Atkinson , J. R. Bell , M. S. Botham , T. M. Brereton , R. Harrington , and D. I. Leech ( 2018 ). The sensitivity of breeding songbirds to changes in seasonal timing is linked to population change but cannot be directly attributed to the effects of trophic asynchrony on productivity . Global Change Biology 24 : 957 – 971 . Google Scholar Crossref Search ADS PubMed WorldCat Gómez , C. , N. J. Bayly , D. R. Norris , S. A. Mackenzie , K. V. Rosenberg , P. D. Taylor , K. A. Hobson , and C. D. Cadena ( 2017 ). Fuel loads acquired at a stopover site influence the pace of intercontinental migration in a boreal songbird . Scientific Reports 7 : 3405 . Google Scholar Crossref Search ADS PubMed WorldCat Grüebler , M. U. , M. Morand , and B. Naef-Daenzer ( 2008 ). A predictive model of the density of airborne insects in agricultural environments . Agriculture, Ecosystems & Environment 123 : 75 – 80 . Google Scholar Crossref Search ADS WorldCat Harriman , V. B. , R. D. Dawson , L. E. Bortolotti , and R. G. Clark ( 2017 ). Seasonal patterns in reproductive success of temperate‐breeding birds: Experimental tests of the date and quality hypotheses . Ecology and Evolution 7 : 2122 – 2132 . Google Scholar Crossref Search ADS PubMed WorldCat Hau , M . ( 2001 ). Timing of breeding in variable environments: Tropical birds as model systems . Hormones and Behavior 40 : 281 – 290 . Google Scholar Crossref Search ADS PubMed WorldCat Hau , M. , M. Wikelski , and J. C. Wingfield ( 2000 ). Visual and nutritional food cues fine-tune timing of reproduction in a Neotropical rainforest bird . The Journal of Experimental Zoology 286 : 494 – 504 . Google Scholar Crossref Search ADS PubMed WorldCat Holt , R. F. , and K. Martin ( 1997 ). Landscape modification and patch selection: The demography of two secondary cavity nesters colonizing clearcuts . The Auk 114 : 443 – 455 . Google Scholar Crossref Search ADS WorldCat Huang , A. C. , C. A. Bishop , R. McKibbin , A. Drake , and D. J. Green ( 2017 ). Wind conditions on migration influence the annual survival of a Neotropical migrant, the western Yellow-breasted Chat (Icteria virens auricollis) . BMC Ecology 17 : 29 . Google Scholar Crossref Search ADS PubMed WorldCat Hussell , D. J. T . ( 2003 ). Climate change, spring temperatures, and timing of breeding of Tree Swallows (Tachycineta bicolor) in Southern Ontario . The Auk 120 : 607 – 618 . Google Scholar Crossref Search ADS WorldCat Illera , J. C. , and M. Díaz ( 2006 ). Reproduction in an endemic bird of a semiarid island: A food-mediated process . Journal of Avian Biology 37 : 447 – 456 . Google Scholar Crossref Search ADS WorldCat Irons , R. D. , A. Harding Scurr , A. P. Rose , J. C. Hagelin , T. Blake , and D. F. Doak ( 2017 ). Wind and rain are the primary climate factors driving changing phenology of an aerial insectivore . Proceedings of the Royal Society B: Biological Sciences 284 : 20170412 . Google Scholar Crossref Search ADS WorldCat Johnson , L. S. , and R. D. Dawson ( 2019 ). Mountain Bluebird (Sialia currucoides) version 2.0 . In The Birds of North America ( P. G. Rodewald , Editor). Cornell Lab of Ornithology , Ithaca, NY, USA . https://doi.org/10.2173/bna.moublu.02. Google Preview WorldCat COPAC Kalnay , E. , M. Kanamitsu , R. Kistler , W. Collins , D. Deaven , L. Gandin , M. Iredell , S. Saha , G. White , J. Woollen , et al. . ( 1996 ). The NCEP/NCAR 40-Year Reanalysis Project . Bulletin of the American Meteorological Society 77 : 437 – 471 . Google Scholar Crossref Search ADS WorldCat Kemp , M. U. , E. Emiel van Loon , J. Shamoun-Baranes , and W. Bouten ( 2012 ). RNCEP: Global weather and climate data at your fingertips . Methods in Ecology and Evolution 3 : 65 – 70 . Google Scholar Crossref Search ADS WorldCat Knight , S. M. , D. W. Bradley , R. G. Clark , E. A. Gow , M. Bélisle , L. L. Berzins , T. Blake , E. S. Bridge , L. Burke , R. D. Dawson , et al. . ( 2018 ). Constructing and evaluating a continent-wide migratory songbird network across the annual cycle . Ecological Monographs 88 : 445 – 460 . Google Scholar Crossref Search ADS WorldCat Koch , A. J. , K. Martin , and K. E. H. Aitken ( 2012 ). The relationship between introduced European Starlings and the reproductive activities of Mountain Bluebirds and Tree Swallows in British Columbia, Canada . Ibis 154 : 590 – 600 . Google Scholar Crossref Search ADS WorldCat Koenig , W. D . ( 2003 ). European Starlings and their effect on native cavity-nesting birds . Conservation Biology 17 : 1134 – 1140 . Google Scholar Crossref Search ADS WorldCat Marra , P. P. , C. M. Francis , R. S. Mulvihill , and F. R. Moore ( 2005 ). The influence of climate on the timing and rate of spring bird migration . Oecologia 142 : 307 – 315 . Google Scholar Crossref Search ADS PubMed WorldCat Martin , K. , K. E. H. Aitken , and K. L. Wiebe ( 2004 ). Nest sites and nest webs for cavity-nesting communities in interior British Columbia, Canada: Nest characteristics and niche partitioning . The Condor 106 : 5 – 19 . Google Scholar Crossref Search ADS WorldCat Martin , K. , A. Norris , and M. Drever ( 2006 ). Effects of bark beetle outbreaks on avian biodiversity in the British Columbia interior: Implications for critical habitat management . BC Journal of Ecosystems and Management 7 : 10 – 24 . WorldCat McCarty , J . ( 2009 ). Variation in growth of nestling Tree Swallows across multiple temporal and spatial scales . The Auk 118 : 176 – 190 . Google Scholar Crossref Search ADS WorldCat McCarty , J. P. , and D. W. Winkler ( 1999 ). Relative importance of environmental variables in determining the growth off nestling Tree Swallows Tachycineta bicolor . Ibis 141 : 286 – 296 . Google Scholar Crossref Search ADS WorldCat Mitchell , G. W. , B. K. Woodworth , P. D. Taylor , and D. R. Norris ( 2015 ). Automated telemetry reveals age specific differences in flight duration and speed are driven by wind conditions in a migratory songbird . Movement Ecology 3 : 19 . Google Scholar Crossref Search ADS PubMed WorldCat Nager , R. G. , and A. J. Van Noordwijk ( 1992 ). Energetic limitation in the egg-laying period of Great Tits . Proceedings of the Royal Society of London. Series B: Biological Sciences 249 : 259 – 263 . Google Scholar Crossref Search ADS WorldCat Nilsson , J.-Å . ( 1994 ). Energetic bottle-necks during breeding and the reproductive cost of being too early . Journal of Animal Ecology 63 : 200 – 208 . Google Scholar Crossref Search ADS WorldCat Nilsson , J.-Å. , and E. Svensson ( 1996 ). The cost of reproduction: A new link between current reproductive effort and future reproductive success . Proceedings of the Royal Society of London. Series B: Biological Sciences 263 : 711 – 714 . Google Scholar Crossref Search ADS WorldCat Perrins , C. M . ( 1970 ). The timing of birds’ breeding seasons . Ibis 112 : 242 – 255 . Google Scholar Crossref Search ADS WorldCat van de Pol , M. , L. D. Bailey , N. McLean , L. Rijsdijk , C. R. Lawson , and L. Brouwer ( 2016 ). Identifying the best climatic predictors in ecology and evolution . Methods in Ecology and Evolution 7 : 1246 – 1257 . Google Scholar Crossref Search ADS WorldCat Poulsen , B. O . ( 1996 ). Relationships between frequency of mixed-species flocks, weather and insect activity in a montane cloud forest in Ecuador . Ibis 138 : 466 – 470 . Google Scholar Crossref Search ADS WorldCat QGIS Development Team ( 2015 ). QGIS Geographic Information System . Open Source Geospatial Foundation . http://qgis.osgeo.org. Google Preview WorldCat COPAC R Core Team ( 2019 ). R: A Language and Environment for Statistical Computing . R Foundation for Statistical Computing, Vienna, Austria . http://www.r-project.org/. Google Preview WorldCat COPAC Ridgely , R. S. , T. F. Allnutt , T. Brooks , D. K. McNicol , D. W. Mehlman , B. E. Young , and J. R. Zook ( 2005 ). Digital Distribution Maps of the Birds of the Western Hemisphere. Version 2.1 . NatureServe, Arlington, VA, USA . Google Preview WorldCat COPAC Robles , H. , and K. Martin ( 2013 ). Resource quantity and quality determine the inter-specific associations between ecosystem engineers and resource users in a cavity-nest web . PLoS One 8 : e74694 . Google Scholar Crossref Search ADS PubMed WorldCat Rockwell , S. M. , C. I. Bocetti , and P. P. Marra ( 2012 ). Carry-over effects of winter climate on spring arrival date and reproductive success in an endangered migratory bird, Kirtland’s Warbler (Setophaga kirtlandii) . The Auk 129 : 744 – 752 . Google Scholar Crossref Search ADS WorldCat Rousset , F. , and J.- B. Ferdy ( 2014 ). Testing environmental and genetic effects in the presence of spatial autocorrelation . Ecography 37 : 781 – 790 . Google Scholar Crossref Search ADS WorldCat Schafer , J. L. , and J. W. Graham ( 2002 ). Missing data: Our view of the state of the art . Psychological Methods 7 : 147 – 177 . Google Scholar Crossref Search ADS PubMed WorldCat Schomaker , M. , and C. Heumann ( 2014 ). Model selection and model averaging after multiple imputation . Computational Statistics & Data Analysis 71 : 758 – 770 . Google Scholar Crossref Search ADS WorldCat Siikamäki , P . ( 1998 ). Limitation of reproductive success by food availability and breeding time in Pied Flycatchers . Ecology 79 : 1789 – 1796 . Google Scholar Crossref Search ADS WorldCat Studds , C. E. , and P. P. Marra ( 2007 ). Linking fluctuations in rainfall to nonbreeding season performance in a long-distance migratory bird, Setophaga ruticilla . Climate Research 35 : 115 – 122 . Google Scholar Crossref Search ADS WorldCat Svensson , E. , and J.- A. Nilsson ( 1995 ). Food supply, territory quality, and reproductive timing in the Blue Tit (Parus Caeruleus) . Ecology 76 : 1804 – 1812 . Google Scholar Crossref Search ADS WorldCat Vansteelant , W. M. G. , W. Bouten , R. H. G. Klaassen , B. J. Koks , A. E. Schlaich , J. van Diermen , E. E. van Loon , and J. Shamoun‐Baranes ( 2015 ). Regional and seasonal flight speeds of soaring migrants and the role of weather conditions at hourly and daily scales . Journal of Avian Biology 46 : 25 – 39 . Google Scholar Crossref Search ADS WorldCat Visser , M. E. , L. J. M. Holleman , and P. Gienapp ( 2006 ). Shifts in caterpillar biomass phenology due to climate change and its impact on the breeding biology of an insectivorous bird . Oecologia 147 : 164 – 172 . Google Scholar Crossref Search ADS PubMed WorldCat Wesolowski , T. , and M. Cholewa ( 2009 ). Climate variation and bird breeding seasons in a primeval temperate forest . Climate Research 38 : 199 – 208 . Google Scholar Crossref Search ADS WorldCat Wesolowski , T. , and K. Martin ( 2018 ). Tree holes and hole-nesting birds in European and North American forests. In Ecology and Conservation of Forest Birds ( G. Mikusiński , J.-M. Roberge and R. J. Fuller , Editors). Cambridge University Press , Cambridge, MA, USA . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Wiebe , K. L . ( 2003 ). Delayed timing as a strategy to avoid nest-site competition: Testing a model using data from starlings and flickers . Oikos 100 : 291 – 298 . Google Scholar Crossref Search ADS WorldCat Wiebe , K. L . ( 2016 ). Interspecific competition for nests: Prior ownership trumps resource holding potential for Mountain Bluebird competing with Tree Swallow . The Auk: Ornithological Advances 133 : 512 – 519 . WorldCat Williams , T. D . ( 2012 ). Physiological adaptations for breeding in Birds . Princeton University Press, Princeton, NJ, USA . Google Preview WorldCat COPAC Williams , T. D. , S. Bourgeon , A. Cornell , L. Ferguson , M. Fowler , R. B. Fronstin , and O. P. Love ( 2015 ). Mid-winter temperatures, not spring temperatures, predict breeding phenology in the European Starling Sturnus vulgaris . Open Science 2 : 140301 . WorldCat Winkler , D. W. , and P. E. Allen ( 1996 ). The seasonal decline in Tree Swallow clutch size: Physiological constraint or strategic adjustment? Ecology 77 : 922 – 932 . Winkler , D. W. , K. K. Hallinger , D. R. Ardia , R. J. Robertson , B. J. Stutchbury , and R. R. Cohen ( 2011 ). Tree Swallow (Tachycineta bicolor), version 2.0 . In The Birds of North America ( A. F. Poole , Editor). Cornell Lab of Ornithology , Ithaca, NY, USA . https://doi.org/10.2173/bna.11. Google Preview WorldCat COPAC APPENDIX WINTER AND MIGRATORY REGION WEATHER DATA Within the NCEP Reanalysis 1 climate dataset (Kalnay et al. 1996), surface variables (temperature and precipitation) are projected on a global T62 Gaussian grid composed of 192 equally spaced longitudinal and 94 variably spaced latitudinal points while wind vector data (westerly and southerly) are projected on a 2.5 × 2.5° spatial grid. In QGIS, we added a 0.95° buffer around surface data points and a 1.25° buffer around wind data points and included those points whose buffer physically overlapped with our defined winter and migration ranges (Appendix Figure 4). We then obtained daily mean values for our variables of interest over these points for the years 1994–2011. METHODS AND RESULTS USING FIELD-RECORDED DATES ONLY To support and supplement our missing data approach, we re-ran our candidate models using only clutch initiation dates that were recorded in the field. Sample sizes for each species are smaller (Tree Swallow = 212; Mountain Bluebird = 173). We bootstrapped each of these datasets 100 times to obtain CIs for our standardized beta values. Reported beta values (β) (Appendix Figure 5) are the median, model-averaged, β for each predictor variable, and the reported 85% and 95% CIs for these values are the 0.075–0.925 and the 0.025–0.975 quantiles of the empirical distribution of these bootstrapped values. Individual model results are shown in Appendix Table 5 and Appendix Table 6. While CIs are larger, our results are consistent with the expanded dataset used in our main analysis. For Mountain Bluebird, northern migratory wind and rain are associated with delays (95% CI) and there is no European Starling effect (as in our main analysis). The weak advancement by year is no longer significant for this species but the sign of the relationship is consistent. The only notable change for Mountain Bluebird is the relationship between cavity number and phenology: the sign of this relationship is consistent with the larger dataset but the effect size is notably smaller and crosses zero at the 85% CI (Appendix Figure 5A). Tree Swallows continue to show delays with increasing rainfall (95% CI) and starling abundance (85% CI) and a weak advancement with year (85% CI). Cavity availability is associated with a significant advancement in breeding dates for Tree Swallows within the reduced dataset (95% CI): the sign of this relationship is consistent with the larger dataset, but the effect is not significant in our main analysis (Appendix Figure 5B). Additionally, correspondence between annual mean clutch initiation dates recorded in the field and means that incorporate modeled dates using MI were high, particularly when years were restricted to those where there were greater than 2 known records (2000–2011: Pearson’s r = 0.94, all years with known dates (1998–2011): Pearson’s r = 0.83). APPENDIX FIGURE 4. Open in new tabDownload slide National Center of Environmental Prediction (NCEP) Reanalysis 1 climate data points for our Mountain Bluebird (A) and Tree Swallow (B) nonbreeding periods (for details see Appendix). Winter range surface temperature and precipitation data are denoted by blue triangles; migration surface temperature and rainfall data are denoted by orange triangles. Migration wind speed data are indicated by orange circles. Wintering data are averaged over all points. Migration data are averaged within 10°-latitudinal bands (20–30°N [Tree Swallow only], 30–40°N, and 40–50°N; see Methods). The red circle indicates the location of our monitoring sites. APPENDIX FIGURE 4. Open in new tabDownload slide National Center of Environmental Prediction (NCEP) Reanalysis 1 climate data points for our Mountain Bluebird (A) and Tree Swallow (B) nonbreeding periods (for details see Appendix). Winter range surface temperature and precipitation data are denoted by blue triangles; migration surface temperature and rainfall data are denoted by orange triangles. Migration wind speed data are indicated by orange circles. Wintering data are averaged over all points. Migration data are averaged within 10°-latitudinal bands (20–30°N [Tree Swallow only], 30–40°N, and 40–50°N; see Methods). The red circle indicates the location of our monitoring sites. APPENDIX FIGURE 5. Open in new tabDownload slide Comparison of model-averaged standardized beta (β) values and their confidence intervals (CIs) for (A) Mountain Bluebird and (B) Tree Swallows, between the full dataset used in the main manuscript (red; identical to Figure 2) and a reduced dataset that used field-recorded clutch initiation dates only (blue; Mountain Bluebird, n = 173; Tree Swallow, n = 212). CIs for results that use the full dataset are the empirical distribution of beta values for 2,000 MI-bootstrap runs (see Methods); CIs for results that used field-recorded data only are the empirical distribution of beta values for 100 bootstrap runs (see Appendix). Non-symmetrical CIs are due to model averaging: when models receive little support within a given run, averaging drives beta values towards zero. APPENDIX FIGURE 5. Open in new tabDownload slide Comparison of model-averaged standardized beta (β) values and their confidence intervals (CIs) for (A) Mountain Bluebird and (B) Tree Swallows, between the full dataset used in the main manuscript (red; identical to Figure 2) and a reduced dataset that used field-recorded clutch initiation dates only (blue; Mountain Bluebird, n = 173; Tree Swallow, n = 212). CIs for results that use the full dataset are the empirical distribution of beta values for 2,000 MI-bootstrap runs (see Methods); CIs for results that used field-recorded data only are the empirical distribution of beta values for 100 bootstrap runs (see Appendix). Non-symmetrical CIs are due to model averaging: when models receive little support within a given run, averaging drives beta values towards zero. APPENDIX TABLE 4. Ranked climate window analyses (climwin) output for weather variables that were tested as possible drivers of breeding phenology for (A) Mountain Bluebird (n = 462 nests) and (B) Tree Swallow (n = 572). Weather variables from winter and migratory ranges and the monitoring site were considered. Tested time windows were based on the phenology of each species (Winkler et al. 2011, Johnson and Dawson 2019) but allowed for the winter and breeding period to encompass the spring migratory period. Within climwin, ΔAIC values represent the difference from the null model (random effect of “Year” only) and are therefore negative when models perform better than the null. Bolded variables are within ΔAIC ≤ 2 of the top model and passed Type I error tests. These models were included in the final candidate set. A positive (+) relationship with timing indicates increasing values of a given weather variable are associated with later clutch initiation dates; negative (–) relationships indicate increasing values of a given weather variable are associated with earlier clutch initiation dates. Period/region Weather variable Temporal window ΔAIC (from null) Relationship with timing Tested Best (A) Mountain Bluebird (n = 462 nests) Winter Mean temperature Sept 1–April 30 Nov 18–Dec 17 –11.10 + Migration Rainfall 40–50°N Feb 1–April 30 Feb 4–Mar 24 –10.96 + Migration Westerlies 40–50°N Feb 1–April 30 Feb 16–Feb 26 –9.08 + Breeding Rainfall Feb 1–June 1 April 20–May 15 –8.79 + Migration Westerlies 30–40°N Feb 1–April 30 Feb 7–Mar 25 –7.92 + Breeding Mean temperature Feb 1–June 1 Feb 27–Mar 9 –7.91 – Winter Rainfall Sept 1–April 30 Oct 4–Oct 14 –7.81 – Breeding Minimum temperature Feb 1–June 1 Feb 27–Mar 9 –7.62 – Breeding Maximum temperature Feb 1–June 1 Feb 27–Mar 9 –7.31 – Migration Rainfall 30–40°N Feb 1–April 30 Feb 22–Mar 4 –5.56 – Migration Mean temperature 30–50°N Feb 1–April 30 Feb 15–Feb 25 –5.25 + Migration Southerlies 30–40°N Feb 1–April 30 April 3–April 22 –5.20 – Migration Southerlies 40–50°N Feb 1–April 30 March 3–Mar 13 –3.54 + (B) Tree Swallow (n = 572 nests) Breeding Rainfall Mar 1– June 1 Mar 13–May 22 –15.97 + Winter Rainfall Sept 1–April 15 Dec 11–Dec 28 –5.14 + Breeding Maximum temperature Mar 1– June 1 May 26–May 15 –5.00 – Migration Southerlies 40–50°N Mar 1–April 15 Mar 25–Apr 8 –3.84 – Breeding Mean temperature Mar 1– June 1 Mar 2–Mar 12 –3.68 – Breeding Minimum temperature Mar 1– June 1 Mar 2–Mar 12 –3.49 – Winter Mean temperature Sept 1–April 15 Nov 4–Jan 15 –2.85 – Migration Westerlies 30–40°N Mar 1–April 15 Mar 26–Apr 6 –2.57 – Migration Southerlies 30–40°N Mar 1–April 15 Mar 28–Apr 9 –2.37 – Migration Southerlies 20–30°N Mar 1–April 15 Mar 6–Mar 17 –1.76 + Migration Rainfall 40–50°N Mar 1–April 15 Mar 25–Apr 4 –1.75 – Migration Westerlies 40–50°N Mar 1–April 15 Apr 5–Apr 15 –1.05 + Migration Rainfall 30–40°N Mar 1–April 15 Mar 28–Apr 7 –0.54 – Migration Westerlies 20–30°N Mar 1–April 15 Mar 9–Mar 22 –0.45 – Migration Mean temperature 30–50°N Mar 1–April 15 Mar 1–Mar 11 0.05 – Migration Mean temperature 20–30°N Mar 1–April 15 Apr 2–Apr 12 0.64 – Migration Rainfall 20–30°N Mar 1–April 15 Mar 17–Mar 27 1.11 – Period/region Weather variable Temporal window ΔAIC (from null) Relationship with timing Tested Best (A) Mountain Bluebird (n = 462 nests) Winter Mean temperature Sept 1–April 30 Nov 18–Dec 17 –11.10 + Migration Rainfall 40–50°N Feb 1–April 30 Feb 4–Mar 24 –10.96 + Migration Westerlies 40–50°N Feb 1–April 30 Feb 16–Feb 26 –9.08 + Breeding Rainfall Feb 1–June 1 April 20–May 15 –8.79 + Migration Westerlies 30–40°N Feb 1–April 30 Feb 7–Mar 25 –7.92 + Breeding Mean temperature Feb 1–June 1 Feb 27–Mar 9 –7.91 – Winter Rainfall Sept 1–April 30 Oct 4–Oct 14 –7.81 – Breeding Minimum temperature Feb 1–June 1 Feb 27–Mar 9 –7.62 – Breeding Maximum temperature Feb 1–June 1 Feb 27–Mar 9 –7.31 – Migration Rainfall 30–40°N Feb 1–April 30 Feb 22–Mar 4 –5.56 – Migration Mean temperature 30–50°N Feb 1–April 30 Feb 15–Feb 25 –5.25 + Migration Southerlies 30–40°N Feb 1–April 30 April 3–April 22 –5.20 – Migration Southerlies 40–50°N Feb 1–April 30 March 3–Mar 13 –3.54 + (B) Tree Swallow (n = 572 nests) Breeding Rainfall Mar 1– June 1 Mar 13–May 22 –15.97 + Winter Rainfall Sept 1–April 15 Dec 11–Dec 28 –5.14 + Breeding Maximum temperature Mar 1– June 1 May 26–May 15 –5.00 – Migration Southerlies 40–50°N Mar 1–April 15 Mar 25–Apr 8 –3.84 – Breeding Mean temperature Mar 1– June 1 Mar 2–Mar 12 –3.68 – Breeding Minimum temperature Mar 1– June 1 Mar 2–Mar 12 –3.49 – Winter Mean temperature Sept 1–April 15 Nov 4–Jan 15 –2.85 – Migration Westerlies 30–40°N Mar 1–April 15 Mar 26–Apr 6 –2.57 – Migration Southerlies 30–40°N Mar 1–April 15 Mar 28–Apr 9 –2.37 – Migration Southerlies 20–30°N Mar 1–April 15 Mar 6–Mar 17 –1.76 + Migration Rainfall 40–50°N Mar 1–April 15 Mar 25–Apr 4 –1.75 – Migration Westerlies 40–50°N Mar 1–April 15 Apr 5–Apr 15 –1.05 + Migration Rainfall 30–40°N Mar 1–April 15 Mar 28–Apr 7 –0.54 – Migration Westerlies 20–30°N Mar 1–April 15 Mar 9–Mar 22 –0.45 – Migration Mean temperature 30–50°N Mar 1–April 15 Mar 1–Mar 11 0.05 – Migration Mean temperature 20–30°N Mar 1–April 15 Apr 2–Apr 12 0.64 – Migration Rainfall 20–30°N Mar 1–April 15 Mar 17–Mar 27 1.11 – Open in new tab APPENDIX TABLE 4. Ranked climate window analyses (climwin) output for weather variables that were tested as possible drivers of breeding phenology for (A) Mountain Bluebird (n = 462 nests) and (B) Tree Swallow (n = 572). Weather variables from winter and migratory ranges and the monitoring site were considered. Tested time windows were based on the phenology of each species (Winkler et al. 2011, Johnson and Dawson 2019) but allowed for the winter and breeding period to encompass the spring migratory period. Within climwin, ΔAIC values represent the difference from the null model (random effect of “Year” only) and are therefore negative when models perform better than the null. Bolded variables are within ΔAIC ≤ 2 of the top model and passed Type I error tests. These models were included in the final candidate set. A positive (+) relationship with timing indicates increasing values of a given weather variable are associated with later clutch initiation dates; negative (–) relationships indicate increasing values of a given weather variable are associated with earlier clutch initiation dates. Period/region Weather variable Temporal window ΔAIC (from null) Relationship with timing Tested Best (A) Mountain Bluebird (n = 462 nests) Winter Mean temperature Sept 1–April 30 Nov 18–Dec 17 –11.10 + Migration Rainfall 40–50°N Feb 1–April 30 Feb 4–Mar 24 –10.96 + Migration Westerlies 40–50°N Feb 1–April 30 Feb 16–Feb 26 –9.08 + Breeding Rainfall Feb 1–June 1 April 20–May 15 –8.79 + Migration Westerlies 30–40°N Feb 1–April 30 Feb 7–Mar 25 –7.92 + Breeding Mean temperature Feb 1–June 1 Feb 27–Mar 9 –7.91 – Winter Rainfall Sept 1–April 30 Oct 4–Oct 14 –7.81 – Breeding Minimum temperature Feb 1–June 1 Feb 27–Mar 9 –7.62 – Breeding Maximum temperature Feb 1–June 1 Feb 27–Mar 9 –7.31 – Migration Rainfall 30–40°N Feb 1–April 30 Feb 22–Mar 4 –5.56 – Migration Mean temperature 30–50°N Feb 1–April 30 Feb 15–Feb 25 –5.25 + Migration Southerlies 30–40°N Feb 1–April 30 April 3–April 22 –5.20 – Migration Southerlies 40–50°N Feb 1–April 30 March 3–Mar 13 –3.54 + (B) Tree Swallow (n = 572 nests) Breeding Rainfall Mar 1– June 1 Mar 13–May 22 –15.97 + Winter Rainfall Sept 1–April 15 Dec 11–Dec 28 –5.14 + Breeding Maximum temperature Mar 1– June 1 May 26–May 15 –5.00 – Migration Southerlies 40–50°N Mar 1–April 15 Mar 25–Apr 8 –3.84 – Breeding Mean temperature Mar 1– June 1 Mar 2–Mar 12 –3.68 – Breeding Minimum temperature Mar 1– June 1 Mar 2–Mar 12 –3.49 – Winter Mean temperature Sept 1–April 15 Nov 4–Jan 15 –2.85 – Migration Westerlies 30–40°N Mar 1–April 15 Mar 26–Apr 6 –2.57 – Migration Southerlies 30–40°N Mar 1–April 15 Mar 28–Apr 9 –2.37 – Migration Southerlies 20–30°N Mar 1–April 15 Mar 6–Mar 17 –1.76 + Migration Rainfall 40–50°N Mar 1–April 15 Mar 25–Apr 4 –1.75 – Migration Westerlies 40–50°N Mar 1–April 15 Apr 5–Apr 15 –1.05 + Migration Rainfall 30–40°N Mar 1–April 15 Mar 28–Apr 7 –0.54 – Migration Westerlies 20–30°N Mar 1–April 15 Mar 9–Mar 22 –0.45 – Migration Mean temperature 30–50°N Mar 1–April 15 Mar 1–Mar 11 0.05 – Migration Mean temperature 20–30°N Mar 1–April 15 Apr 2–Apr 12 0.64 – Migration Rainfall 20–30°N Mar 1–April 15 Mar 17–Mar 27 1.11 – Period/region Weather variable Temporal window ΔAIC (from null) Relationship with timing Tested Best (A) Mountain Bluebird (n = 462 nests) Winter Mean temperature Sept 1–April 30 Nov 18–Dec 17 –11.10 + Migration Rainfall 40–50°N Feb 1–April 30 Feb 4–Mar 24 –10.96 + Migration Westerlies 40–50°N Feb 1–April 30 Feb 16–Feb 26 –9.08 + Breeding Rainfall Feb 1–June 1 April 20–May 15 –8.79 + Migration Westerlies 30–40°N Feb 1–April 30 Feb 7–Mar 25 –7.92 + Breeding Mean temperature Feb 1–June 1 Feb 27–Mar 9 –7.91 – Winter Rainfall Sept 1–April 30 Oct 4–Oct 14 –7.81 – Breeding Minimum temperature Feb 1–June 1 Feb 27–Mar 9 –7.62 – Breeding Maximum temperature Feb 1–June 1 Feb 27–Mar 9 –7.31 – Migration Rainfall 30–40°N Feb 1–April 30 Feb 22–Mar 4 –5.56 – Migration Mean temperature 30–50°N Feb 1–April 30 Feb 15–Feb 25 –5.25 + Migration Southerlies 30–40°N Feb 1–April 30 April 3–April 22 –5.20 – Migration Southerlies 40–50°N Feb 1–April 30 March 3–Mar 13 –3.54 + (B) Tree Swallow (n = 572 nests) Breeding Rainfall Mar 1– June 1 Mar 13–May 22 –15.97 + Winter Rainfall Sept 1–April 15 Dec 11–Dec 28 –5.14 + Breeding Maximum temperature Mar 1– June 1 May 26–May 15 –5.00 – Migration Southerlies 40–50°N Mar 1–April 15 Mar 25–Apr 8 –3.84 – Breeding Mean temperature Mar 1– June 1 Mar 2–Mar 12 –3.68 – Breeding Minimum temperature Mar 1– June 1 Mar 2–Mar 12 –3.49 – Winter Mean temperature Sept 1–April 15 Nov 4–Jan 15 –2.85 – Migration Westerlies 30–40°N Mar 1–April 15 Mar 26–Apr 6 –2.57 – Migration Southerlies 30–40°N Mar 1–April 15 Mar 28–Apr 9 –2.37 – Migration Southerlies 20–30°N Mar 1–April 15 Mar 6–Mar 17 –1.76 + Migration Rainfall 40–50°N Mar 1–April 15 Mar 25–Apr 4 –1.75 – Migration Westerlies 40–50°N Mar 1–April 15 Apr 5–Apr 15 –1.05 + Migration Rainfall 30–40°N Mar 1–April 15 Mar 28–Apr 7 –0.54 – Migration Westerlies 20–30°N Mar 1–April 15 Mar 9–Mar 22 –0.45 – Migration Mean temperature 30–50°N Mar 1–April 15 Mar 1–Mar 11 0.05 – Migration Mean temperature 20–30°N Mar 1–April 15 Apr 2–Apr 12 0.64 – Migration Rainfall 20–30°N Mar 1–April 15 Mar 17–Mar 27 1.11 – Open in new tab APPENDIX TABLE 7. Annual productivity data for records where clutch size and nest fate were known (Mountain Bluebird, n = 229 nests; Tree Swallow, n = 177). Nest success is the proportion of nests that fledged at least one young. Year Mountain Bluebird Tree Swallow Clutch size mean (range) Fledge # mean (range) Nest success n Clutch size mean (range) Fledge # mean (range) Nest success n 2000 5.3 (5–6) 2.0 (0–4) 0.50 6 6.0 (4–8) 1.0 (0–6) 0.17 6 2001 5.4 (4–6) 1.4 (0–5) 0.44 9 5.2 (2–7) 2.8 (0–6) 0.67 6 2002 4.8 (2–6) 1.9 (0–5) 0.61 18 4.7 (4–6) 1.7 (0–5) 0.44 9 2003 5.0 (2–6) 3.0 (0–6) 0.72 18 5.4 (3–7) 1.1 (0–5) 0.31 16 2004 5.1 (4–6) 3.0 (0–6) 0.71 17 5.3 (2–7) 0.8 (0–4) 0.32 19 2005 5.1 (4–6) 2.3 (0–5) 0.58 12 5.6 (5–7) 2.8 (0–5) 0.75 8 2006 5.2 (4–6) 1.4 (0–5) 0.31 16 4.5 (3–6) 0.0 0.00 6 2007 5.1 (4–6) 2.3 (0–6) 0.70 23 5.4 (3–7) 2.4 (0–6) 0.75 12 2008 5.5 (5–7) 2.9 (0–5) 0.83 18 5.8 (4–8) 3.7 (0–6) 0.85 13 2009 5.3 (4–7) 3.0 (0–7) 0.62 29 5.3 (3–7) 2.3 (0–6) 0.56 27 2010 5.2 (3–6) 2.8 (0–6) 0.66 29 5.7 (4–7) 2.3 (0–7) 0.47 36 2011 5.4 (4–7) 1.7 (0–5) 0.47 34 5.6 (4–7) 2.5 (0–6) 0.58 19 Year Mountain Bluebird Tree Swallow Clutch size mean (range) Fledge # mean (range) Nest success n Clutch size mean (range) Fledge # mean (range) Nest success n 2000 5.3 (5–6) 2.0 (0–4) 0.50 6 6.0 (4–8) 1.0 (0–6) 0.17 6 2001 5.4 (4–6) 1.4 (0–5) 0.44 9 5.2 (2–7) 2.8 (0–6) 0.67 6 2002 4.8 (2–6) 1.9 (0–5) 0.61 18 4.7 (4–6) 1.7 (0–5) 0.44 9 2003 5.0 (2–6) 3.0 (0–6) 0.72 18 5.4 (3–7) 1.1 (0–5) 0.31 16 2004 5.1 (4–6) 3.0 (0–6) 0.71 17 5.3 (2–7) 0.8 (0–4) 0.32 19 2005 5.1 (4–6) 2.3 (0–5) 0.58 12 5.6 (5–7) 2.8 (0–5) 0.75 8 2006 5.2 (4–6) 1.4 (0–5) 0.31 16 4.5 (3–6) 0.0 0.00 6 2007 5.1 (4–6) 2.3 (0–6) 0.70 23 5.4 (3–7) 2.4 (0–6) 0.75 12 2008 5.5 (5–7) 2.9 (0–5) 0.83 18 5.8 (4–8) 3.7 (0–6) 0.85 13 2009 5.3 (4–7) 3.0 (0–7) 0.62 29 5.3 (3–7) 2.3 (0–6) 0.56 27 2010 5.2 (3–6) 2.8 (0–6) 0.66 29 5.7 (4–7) 2.3 (0–7) 0.47 36 2011 5.4 (4–7) 1.7 (0–5) 0.47 34 5.6 (4–7) 2.5 (0–6) 0.58 19 Open in new tab APPENDIX TABLE 7. Annual productivity data for records where clutch size and nest fate were known (Mountain Bluebird, n = 229 nests; Tree Swallow, n = 177). Nest success is the proportion of nests that fledged at least one young. Year Mountain Bluebird Tree Swallow Clutch size mean (range) Fledge # mean (range) Nest success n Clutch size mean (range) Fledge # mean (range) Nest success n 2000 5.3 (5–6) 2.0 (0–4) 0.50 6 6.0 (4–8) 1.0 (0–6) 0.17 6 2001 5.4 (4–6) 1.4 (0–5) 0.44 9 5.2 (2–7) 2.8 (0–6) 0.67 6 2002 4.8 (2–6) 1.9 (0–5) 0.61 18 4.7 (4–6) 1.7 (0–5) 0.44 9 2003 5.0 (2–6) 3.0 (0–6) 0.72 18 5.4 (3–7) 1.1 (0–5) 0.31 16 2004 5.1 (4–6) 3.0 (0–6) 0.71 17 5.3 (2–7) 0.8 (0–4) 0.32 19 2005 5.1 (4–6) 2.3 (0–5) 0.58 12 5.6 (5–7) 2.8 (0–5) 0.75 8 2006 5.2 (4–6) 1.4 (0–5) 0.31 16 4.5 (3–6) 0.0 0.00 6 2007 5.1 (4–6) 2.3 (0–6) 0.70 23 5.4 (3–7) 2.4 (0–6) 0.75 12 2008 5.5 (5–7) 2.9 (0–5) 0.83 18 5.8 (4–8) 3.7 (0–6) 0.85 13 2009 5.3 (4–7) 3.0 (0–7) 0.62 29 5.3 (3–7) 2.3 (0–6) 0.56 27 2010 5.2 (3–6) 2.8 (0–6) 0.66 29 5.7 (4–7) 2.3 (0–7) 0.47 36 2011 5.4 (4–7) 1.7 (0–5) 0.47 34 5.6 (4–7) 2.5 (0–6) 0.58 19 Year Mountain Bluebird Tree Swallow Clutch size mean (range) Fledge # mean (range) Nest success n Clutch size mean (range) Fledge # mean (range) Nest success n 2000 5.3 (5–6) 2.0 (0–4) 0.50 6 6.0 (4–8) 1.0 (0–6) 0.17 6 2001 5.4 (4–6) 1.4 (0–5) 0.44 9 5.2 (2–7) 2.8 (0–6) 0.67 6 2002 4.8 (2–6) 1.9 (0–5) 0.61 18 4.7 (4–6) 1.7 (0–5) 0.44 9 2003 5.0 (2–6) 3.0 (0–6) 0.72 18 5.4 (3–7) 1.1 (0–5) 0.31 16 2004 5.1 (4–6) 3.0 (0–6) 0.71 17 5.3 (2–7) 0.8 (0–4) 0.32 19 2005 5.1 (4–6) 2.3 (0–5) 0.58 12 5.6 (5–7) 2.8 (0–5) 0.75 8 2006 5.2 (4–6) 1.4 (0–5) 0.31 16 4.5 (3–6) 0.0 0.00 6 2007 5.1 (4–6) 2.3 (0–6) 0.70 23 5.4 (3–7) 2.4 (0–6) 0.75 12 2008 5.5 (5–7) 2.9 (0–5) 0.83 18 5.8 (4–8) 3.7 (0–6) 0.85 13 2009 5.3 (4–7) 3.0 (0–7) 0.62 29 5.3 (3–7) 2.3 (0–6) 0.56 27 2010 5.2 (3–6) 2.8 (0–6) 0.66 29 5.7 (4–7) 2.3 (0–7) 0.47 36 2011 5.4 (4–7) 1.7 (0–5) 0.47 34 5.6 (4–7) 2.5 (0–6) 0.58 19 Open in new tab APPENDIX TABLE 5. Mountain Bluebird final candidate model ranking using only field-recorded clutch initiation dates (n = 173 nests). Values under model parameters are standardized beta values. All values represent the median of 100 bootstrap runs (see Appendix). Null and annual trend models are italicized. There was little difference in model performance (all models ≤ 4 ΔAIC). k = number of parameters and wi = AIC weight. Model Intercept Year (trend) Cavity availability Starling occupancy Rainfall 40–50N Westerly wind 40–50N k loglik ΔAIC a wi MOBL10 143.77 – – – – 2.50 5 –684.99 0.00 0.18 MOBL8 143.67 – – –1.71 2.41 – 6 –684.28 0.71 0.12 MOBL6 143.73 – – – 2.34 – 5 –685.41 0.84 0.12 MOBL1 143.14 – – – – – 4 –686.70 1.29 0.09 MOBL12 143.66 – – –1.23 – 2.47 6 –684.70 1.57 0.08 MOBL11 144.17 – –0.77 – – 2.27 6 –684.75 1.67 0.08 MOBL3 144.83 – –2.08 – – – 5 –686.08 2.17 0.06 MOBL9 144.29 – –0.86 –1.72 2.42 – 7 –684.06 2.46 0.05 MOBL7 144.11 – –0.85 – 2.19 – 6 –685.15 2.47 0.05 MOBL4 143.08 – – –1.32 – – 5 –686.26 2.54 0.05 MOBL2 143.27 0.04 – – – – 5 –686.43 2.87 0.04 MOBL13 144.24 – –0.89 –1.41 – 2.24 7 –684.39 3.11 0.04 MOBL5 144.82 – –2.08 –1.53 – – 6 –685.50 3.16 0.04 Model Intercept Year (trend) Cavity availability Starling occupancy Rainfall 40–50N Westerly wind 40–50N k loglik ΔAIC a wi MOBL10 143.77 – – – – 2.50 5 –684.99 0.00 0.18 MOBL8 143.67 – – –1.71 2.41 – 6 –684.28 0.71 0.12 MOBL6 143.73 – – – 2.34 – 5 –685.41 0.84 0.12 MOBL1 143.14 – – – – – 4 –686.70 1.29 0.09 MOBL12 143.66 – – –1.23 – 2.47 6 –684.70 1.57 0.08 MOBL11 144.17 – –0.77 – – 2.27 6 –684.75 1.67 0.08 MOBL3 144.83 – –2.08 – – – 5 –686.08 2.17 0.06 MOBL9 144.29 – –0.86 –1.72 2.42 – 7 –684.06 2.46 0.05 MOBL7 144.11 – –0.85 – 2.19 – 6 –685.15 2.47 0.05 MOBL4 143.08 – – –1.32 – – 5 –686.26 2.54 0.05 MOBL2 143.27 0.04 – – – – 5 –686.43 2.87 0.04 MOBL13 144.24 – –0.89 –1.41 – 2.24 7 –684.39 3.11 0.04 MOBL5 144.82 – –2.08 –1.53 – – 6 –685.50 3.16 0.04 a Lowest median AIC value was 1,380.35. Open in new tab APPENDIX TABLE 5. Mountain Bluebird final candidate model ranking using only field-recorded clutch initiation dates (n = 173 nests). Values under model parameters are standardized beta values. All values represent the median of 100 bootstrap runs (see Appendix). Null and annual trend models are italicized. There was little difference in model performance (all models ≤ 4 ΔAIC). k = number of parameters and wi = AIC weight. Model Intercept Year (trend) Cavity availability Starling occupancy Rainfall 40–50N Westerly wind 40–50N k loglik ΔAIC a wi MOBL10 143.77 – – – – 2.50 5 –684.99 0.00 0.18 MOBL8 143.67 – – –1.71 2.41 – 6 –684.28 0.71 0.12 MOBL6 143.73 – – – 2.34 – 5 –685.41 0.84 0.12 MOBL1 143.14 – – – – – 4 –686.70 1.29 0.09 MOBL12 143.66 – – –1.23 – 2.47 6 –684.70 1.57 0.08 MOBL11 144.17 – –0.77 – – 2.27 6 –684.75 1.67 0.08 MOBL3 144.83 – –2.08 – – – 5 –686.08 2.17 0.06 MOBL9 144.29 – –0.86 –1.72 2.42 – 7 –684.06 2.46 0.05 MOBL7 144.11 – –0.85 – 2.19 – 6 –685.15 2.47 0.05 MOBL4 143.08 – – –1.32 – – 5 –686.26 2.54 0.05 MOBL2 143.27 0.04 – – – – 5 –686.43 2.87 0.04 MOBL13 144.24 – –0.89 –1.41 – 2.24 7 –684.39 3.11 0.04 MOBL5 144.82 – –2.08 –1.53 – – 6 –685.50 3.16 0.04 Model Intercept Year (trend) Cavity availability Starling occupancy Rainfall 40–50N Westerly wind 40–50N k loglik ΔAIC a wi MOBL10 143.77 – – – – 2.50 5 –684.99 0.00 0.18 MOBL8 143.67 – – –1.71 2.41 – 6 –684.28 0.71 0.12 MOBL6 143.73 – – – 2.34 – 5 –685.41 0.84 0.12 MOBL1 143.14 – – – – – 4 –686.70 1.29 0.09 MOBL12 143.66 – – –1.23 – 2.47 6 –684.70 1.57 0.08 MOBL11 144.17 – –0.77 – – 2.27 6 –684.75 1.67 0.08 MOBL3 144.83 – –2.08 – – – 5 –686.08 2.17 0.06 MOBL9 144.29 – –0.86 –1.72 2.42 – 7 –684.06 2.46 0.05 MOBL7 144.11 – –0.85 – 2.19 – 6 –685.15 2.47 0.05 MOBL4 143.08 – – –1.32 – – 5 –686.26 2.54 0.05 MOBL2 143.27 0.04 – – – – 5 –686.43 2.87 0.04 MOBL13 144.24 – –0.89 –1.41 – 2.24 7 –684.39 3.11 0.04 MOBL5 144.82 – –2.08 –1.53 – – 6 –685.50 3.16 0.04 a Lowest median AIC value was 1,380.35. Open in new tab APPENDIX TABLE 6. Tree Swallow final candidate model ranking using only field-recorded clutch initiation dates (n = 212 nests). Values are the median of 100 bootstrap runs (see Appendix). Null and annual trend models are italicized. Top models (≤ 4 ΔAIC) are in bold and represent 90% of total model support. k = number of parameters and wi = AIC weight. Model Intercept Year (trend) Local rainfall Cavity availability Starling occupancy Bluebird occupancy k loglik ΔAICa wi TRES14 155.06 – 2.00 –1.89 2.59 – 7 –742.26 0.00 0.48 TRES17 155.02 – 2.01 –1.87 2.57 –0.01 8 –742.17 1.98 0.18 TRES8 153.66 – 2.18 – 2.14 – 6 –745.06 3.45 0.09 TRES15 153.89 – 2.22 – 2.15 –0.96 7 –744.01 3.49 0.08 TRES9 155.20 – – –2.19 3.08 – 6 –745.27 3.88 0.07 TRES16 155.24 – – –2.30 3.09 0.20 7 –745.01 5.50 0.03 TRES13 154.61 – 2.20 –1.29 – –0.56 7 –745.68 6.84 0.02 TRES5 153.77 – – – 2.18 – 5 –748.07 7.36 0.01 TRES3 153.61 – 2.42 – – – 5 –748.37 7.97 0.01 TRES2 154.94 –2.62 – – – – 5 –748.56 8.35 0.01 TRES10 153.86 – 2.40 – – –0.66 6 –747.61 8.56 0.01 TRES7 154.61 – 2.29 –1.27 – – 6 –747.67 8.69 0.01 TRES12 154.01 – – – 2.24 –1.12 6 –747.71 8.76 0.01 TRES1 153.71 – – – – – 4 –750.70 10.53 0.00 TRES4 154.69 – – –1.42 – – 5 –749.72 10.67 0.00 TRES6 153.96 – – – – –0.66 5 –749.76 10.75 0.00 TRES11 154.64 – – –1.35 – –0.32 6 –749.21 11.75 0.00 Model Intercept Year (trend) Local rainfall Cavity availability Starling occupancy Bluebird occupancy k loglik ΔAICa wi TRES14 155.06 – 2.00 –1.89 2.59 – 7 –742.26 0.00 0.48 TRES17 155.02 – 2.01 –1.87 2.57 –0.01 8 –742.17 1.98 0.18 TRES8 153.66 – 2.18 – 2.14 – 6 –745.06 3.45 0.09 TRES15 153.89 – 2.22 – 2.15 –0.96 7 –744.01 3.49 0.08 TRES9 155.20 – – –2.19 3.08 – 6 –745.27 3.88 0.07 TRES16 155.24 – – –2.30 3.09 0.20 7 –745.01 5.50 0.03 TRES13 154.61 – 2.20 –1.29 – –0.56 7 –745.68 6.84 0.02 TRES5 153.77 – – – 2.18 – 5 –748.07 7.36 0.01 TRES3 153.61 – 2.42 – – – 5 –748.37 7.97 0.01 TRES2 154.94 –2.62 – – – – 5 –748.56 8.35 0.01 TRES10 153.86 – 2.40 – – –0.66 6 –747.61 8.56 0.01 TRES7 154.61 – 2.29 –1.27 – – 6 –747.67 8.69 0.01 TRES12 154.01 – – – 2.24 –1.12 6 –747.71 8.76 0.01 TRES1 153.71 – – – – – 4 –750.70 10.53 0.00 TRES4 154.69 – – –1.42 – – 5 –749.72 10.67 0.00 TRES6 153.96 – – – – –0.66 5 –749.76 10.75 0.00 TRES11 154.64 – – –1.35 – –0.32 6 –749.21 11.75 0.00 a Lowest median AIC value was 1,499.07. Open in new tab APPENDIX TABLE 6. Tree Swallow final candidate model ranking using only field-recorded clutch initiation dates (n = 212 nests). Values are the median of 100 bootstrap runs (see Appendix). Null and annual trend models are italicized. Top models (≤ 4 ΔAIC) are in bold and represent 90% of total model support. k = number of parameters and wi = AIC weight. Model Intercept Year (trend) Local rainfall Cavity availability Starling occupancy Bluebird occupancy k loglik ΔAICa wi TRES14 155.06 – 2.00 –1.89 2.59 – 7 –742.26 0.00 0.48 TRES17 155.02 – 2.01 –1.87 2.57 –0.01 8 –742.17 1.98 0.18 TRES8 153.66 – 2.18 – 2.14 – 6 –745.06 3.45 0.09 TRES15 153.89 – 2.22 – 2.15 –0.96 7 –744.01 3.49 0.08 TRES9 155.20 – – –2.19 3.08 – 6 –745.27 3.88 0.07 TRES16 155.24 – – –2.30 3.09 0.20 7 –745.01 5.50 0.03 TRES13 154.61 – 2.20 –1.29 – –0.56 7 –745.68 6.84 0.02 TRES5 153.77 – – – 2.18 – 5 –748.07 7.36 0.01 TRES3 153.61 – 2.42 – – – 5 –748.37 7.97 0.01 TRES2 154.94 –2.62 – – – – 5 –748.56 8.35 0.01 TRES10 153.86 – 2.40 – – –0.66 6 –747.61 8.56 0.01 TRES7 154.61 – 2.29 –1.27 – – 6 –747.67 8.69 0.01 TRES12 154.01 – – – 2.24 –1.12 6 –747.71 8.76 0.01 TRES1 153.71 – – – – – 4 –750.70 10.53 0.00 TRES4 154.69 – – –1.42 – – 5 –749.72 10.67 0.00 TRES6 153.96 – – – – –0.66 5 –749.76 10.75 0.00 TRES11 154.64 – – –1.35 – –0.32 6 –749.21 11.75 0.00 Model Intercept Year (trend) Local rainfall Cavity availability Starling occupancy Bluebird occupancy k loglik ΔAICa wi TRES14 155.06 – 2.00 –1.89 2.59 – 7 –742.26 0.00 0.48 TRES17 155.02 – 2.01 –1.87 2.57 –0.01 8 –742.17 1.98 0.18 TRES8 153.66 – 2.18 – 2.14 – 6 –745.06 3.45 0.09 TRES15 153.89 – 2.22 – 2.15 –0.96 7 –744.01 3.49 0.08 TRES9 155.20 – – –2.19 3.08 – 6 –745.27 3.88 0.07 TRES16 155.24 – – –2.30 3.09 0.20 7 –745.01 5.50 0.03 TRES13 154.61 – 2.20 –1.29 – –0.56 7 –745.68 6.84 0.02 TRES5 153.77 – – – 2.18 – 5 –748.07 7.36 0.01 TRES3 153.61 – 2.42 – – – 5 –748.37 7.97 0.01 TRES2 154.94 –2.62 – – – – 5 –748.56 8.35 0.01 TRES10 153.86 – 2.40 – – –0.66 6 –747.61 8.56 0.01 TRES7 154.61 – 2.29 –1.27 – – 6 –747.67 8.69 0.01 TRES12 154.01 – – – 2.24 –1.12 6 –747.71 8.76 0.01 TRES1 153.71 – – – – – 4 –750.70 10.53 0.00 TRES4 154.69 – – –1.42 – – 5 –749.72 10.67 0.00 TRES6 153.96 – – – – –0.66 5 –749.76 10.75 0.00 TRES11 154.64 – – –1.35 – –0.32 6 –749.21 11.75 0.00 a Lowest median AIC value was 1,499.07. Open in new tab Copyright © American Ornithological Society 2020. All rights reserved. For permissions, e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Auk: Ornithological Advances Oxford University Press

Rainfall and nest site competition delay Mountain Bluebird and Tree Swallow breeding but do not impact productivity

Auk: Ornithological Advances , Volume Advance Article – Apr 5, 2020

Loading next page...
 
/lp/oxford-university-press/rainfall-and-nest-site-competition-delay-mountain-bluebird-and-tree-JsN7cXj9OZ

References (88)

Copyright
Copyright © American Ornithological Society 2020. All rights reserved. For permissions, e-mail: journals.permissions@oup.com.
ISSN
0004-8038
eISSN
1938-4254
DOI
10.1093/auk/ukaa006
Publisher site
See Article on Publisher Site

Abstract

Abstract Optimizing breeding phenology, an important aspect of fitness, is complex for migratory species as they must make key timing decisions early, and remotely, from breeding sites. We examined the role of weather (locally and cross-seasonally), cavity availability, and competitive exclusion in determining among-year variation in breeding phenology over 17 yr for 2 migratory, cavity-nesting birds: Mountain Bluebirds (Sialia currucoides; n = 462 nests) and Tree Swallows (Tachycineta bicolor; n = 572) using natural tree cavities in British Columbia, Canada. We assessed weather effects within the winter and migratory range and at our study sites. We quantified competition as the proportion of cavities occupied by European Starlings (Sturnus vulgaris) (for both species) and Mountain Bluebirds (for Tree Swallow only) in each year. For 229 bluebird and 177 swallow nests with known fates, we tested whether late years resulted in reduced productivity. Although the effects were small, heavy rainfall and strong diurnal westerly winds during migration were associated with breeding delays for Mountain Bluebirds. However, cavity availability (earlier breeding with increases) had a 5–8 × greater effect on timing than migratory conditions. There was no evidence that starling competition delayed bluebirds. In Tree Swallows, greater local daily rainfall was associated with delayed breeding, as was starling abundance (the effect of starlings was 1.4 × smaller than that of rainfall). Neither bluebird abundance nor cavity availability changed swallow phenology. Neither species showed reduced productivity in late breeding years. In both species, individuals that bred late relative to conspecifics within-year had smaller clutches and greater probability of nest failure. We conclude that breeding ground conditions, particularly cavity limitation and local rainfall (for swallows), are important drivers of breeding phenology for our focal species, but that the productivity cost of late years, at least for Tree Swallows, is minimal. RÉSUMÉ Optimiser la phénologie de la reproduction, un aspect important de la condition physique, est complexe pour les espèces migratrices car elles doivent prendre tôt des décisions clés concernant le moment de la reproduction, et ce loin des sites de reproduction. Nous avons examiné le rôle des conditions météorologiques (localement et de façon saisonnière), de la disponibilité des cavités et de l’exclusion compétitive dans la détermination de la variation interannuelle de la phénologie de la reproduction pendant 17 ans chez deux oiseaux migrateurs nichant dans des cavités: Sialia currucoides (n = 462 nids) et Tachycineta bicolor (n = 572), lesquels utilisent des cavités d’arbres naturelles en Colombie-Britannique, au Canada. Nous avons évalué les effets des conditions météorologiques dans l’aire de répartition hivernale et migratoire, ainsi qu’à nos sites d’étude. Nous avons quantifié la compétition comme étant la proportion de cavités occupées par Sturnus vulgaris (pour les deux espèces) et S. currucoides (pour T. bicolor seulement) à chaque année. Pour 229 nids de S. currucoides et 177 nids de T. bicolor dont le sort était connu, nous avons testé si les années tardives causaient une réduction de la productivité. Bien que les effets aient été faibles, de fortes précipitations et de forts vents diurnes provenant de l’ouest pendant la migration étaient associés aux retards dans la reproduction pour S. currucoides. Toutefois, la disponibilité des cavités (reproduction plus hâtive si augmentations) avait de 5 à 8 fois plus d’effet sur le moment de la reproduction que les conditions migratoires. Aucune preuve n’appuyait le fait que la compétition par S. vulgaris retardait la reproduction de S. currucoides. Chez T. bicolor, une plus grande quantité quotidienne de précipitations au niveau local était associée à un retard dans la reproduction, tout comme l’abondance de S. vulgaris (l’effet de S. vulgaris était 1,4 fois plus faible que celui des précipitations). Ni l’abondance de S. currucoides ni la disponibilité des cavités n’ont changé la phénologie de T. bicolor. Aucune de ces espèces ne présentait une réduction de la productivité lors des années où la reproduction était plus tardive. Chez les deux espèces, les individus qui se reproduisaient tardivement comparativement à leurs congénères au cours de l’année avaient des couvées plus petites et une plus grande probabilité d’échec de la reproduction. Nous concluons que les conditions sur les aires de reproduction, particulièrement la limitation des cavités et les précipitations locales (pour T. bicolor), sont des facteurs importants de la phénologie de la reproduction pour nos espèces cibles, mais que les coûts de la productivité lors des années tardives sont minimaux, du moins pour T. bicolor. INTRODUCTION Birds are under selective pressure to match breeding with food abundance, such that they are provisioning young when resources are at their peak (Perrins 1970, Siikamäki 1998, Visser et al. 2006, Both 2010). In seasonal environments, initiating breeding too early or late can additionally expose nesting adults and developing young to unfavorable conditions. This, in turn, can lead to adult or offspring mortality, lower quality eggs, and nest abandonment (Nager and Van Noordwijk 1992, Nilsson 1994, Drake et al. 2014a, Johnson and Dawson 2019, de Zwaan et al. 2019). Late breeding by individuals is associated with smaller clutches, fewer fledglings, fewer recruits, and reduced parental survival over winter, independent of individual qualities that may influence such relationships (Nilsson and Svensson 1996, Brown and Roth 2002, Harriman et al. 2017). Because the timing of peak food and optimal weather conditions can be variable among years, bird species often exhibit behavioral plasticity and adjust clutch initiation dates in response to environmental cues (e.g., rainfall [Hau 2001, Illera and Díaz 2006, Cavalcanti et al. 2016], temperature [Cresswell and Mccleery 2003, Wesolowski and Cholewa 2009, Drake and Martin 2018], and food abundance [Hau et al. 2000]). Achieving optimal timing is complicated among migratory species because these species must make other timing decisions (winter ground departure date and rate of movement during migration) that impact when they can breed (Marra et al. 2005, Åkesson et al. 2017). These decisions are made early and remotely from breeding sites, potentially increasing the probability of mistiming (Åkesson et al. 2017, Franks et al. 2018). Poor conditions (e.g., drought, storm events) on the wintering grounds and on the migratory route may also delay breeding, for example, by slowing fattening or reducing flight range when food is scarce (Studds and Marra 2007, Rockwell et al. 2012, Gómez et al. 2017) or by slowing movement or keeping individuals grounded when conditions aloft are hostile (Drake et al. 2014b, Vansteelant et al. 2015, Mitchell et al. 2015). In cavity-nesting species, achieving optimal timing is further complicated by nest site availability. Cavity limitation and intra- and inter-specific competition for cavities should delay clutch initiation as pairs seek to acquire a suitable nest site (Koch et al. 2012). Thus, while relationships between timing and some weather cues may represent tracking of conditions and resources, relationships between timing and cavity availability or competition should represent constraints and be exclusively costly, by producing a mismatch with optimal timing. In this paper, we examine the timing of breeding in 2 migratory, secondary cavity-nesting species: Mountain Bluebird (Sialia currucoides) and Tree Swallow (Tachycineta bicolor). These species co-occur in breeding habitat throughout western North America; at our study site they are the most numerous of 4 migratory, secondary cavity-nesting species recorded (the other 2 species being non-native European Starling [Sturnus vulgaris] and American Kestrel [Falco sparverius]). Mountain Bluebirds winter in the southwestern United States with some range extension into the northern interior of Mexico (Johnson and Dawson 2019). Tree Swallows winter in western Mexico, throughout the Gulf of Mexico, and along the east coast of Mexico and Central America (Winkler et al. 2011). The spring migratory period in Mountain Bluebird begins earlier (February vs. March) but lasts longer than that of Tree Swallow (~3 vs. 1.5 mo) (Winkler et al. 2011, Johnson and Dawson 2019). Both species use previously excavated and naturally occurring tree holes as nesting sites as well as nest boxes, when provided. Based on previous work (Koch et al. 2012, Wiebe 2016), we consider European Starlings (for both species) and bluebirds (for swallow) to be strong competitors for nesting space. All 3 species overlap in terms of tree hole preferences and clutch initiation dates (although species’ mean initiation dates differ) (Koch et al. 2012). We examine variation in the timing of breeding in Mountain Bluebirds and Tree Swallows in natural tree cavities in south-central British Columbia over 17 yr. We assess whether these species showed shifts in clutch initiation dates among years in response to relevant weather variables across their annual cycle (winter and spring temperatures and rainfall, and spring wind speed). We also test whether ecological constraints—specifically, annual cavity availability and the prevalence of nest site competitors—were associated with shifts in clutch initiation dates within each population. We then assess the relative importance of our predictor variables in determining when, on average, Tree Swallows and Mountain Bluebirds initiated breeding in a given year. Finally, we test if population-level shifts in timing among years impacted population productivity at our study site. We do this by separating the productivity impact of breeding late relative to conspecifics within a year from the contribution of a late breeding year, on average, for the population as a whole. METHODS Study Site and Breeding Activity Breeding data for Mountain Bluebird and Tree Swallow were collected between 1995 and 2011 at 2 study sites ~38 km apart, Riske Creek (52.0025°N, 122.4116°W) and Knife Creek (52.0068°N, 121.8619°W), near Williams Lake in south-central British Columbia, Canada. This region is part of the warm and dry Interior Douglas-fir biogeoclimatic zone. In this study, the Riske Creek site consisted of 16 mixed conifer stands (Douglas-fir [Pseudotsuga menziesii var. menziesii], lodgepole pine [Pinus contorta var. latifolia], and white and Englemann spruce hybrids [Picea glauca × engelmannii]) with trembling aspen (Populus tremuloides) within a grassland-wetland matrix. Knife Creek consisted of 11 mixed conifer stands with some deciduous riparian zone. Stands ranged from 7 to 32 ha in size. No nest boxes were present at either site and all nesting was done in natural tree cavities. Mountain Bluebird and Tree Swallow were 2 of a total of 32 cavity-nesting bird species found within the study sites over the monitoring period (Wesolowski and Martin 2018). During the 1995 to 2011 period, systematic searches were conducted between May 1 and July 31. These surveys were conducted for ~6–7 hr per stand per week by walking the entirety of each stand and examining previously identified nest sites and following birds (Aitken and Martin 2007). The number of stands monitored increased between 1995 and 1998, but thereafter survey effort was equivalent. The majority of nests were found in the laying or early incubation stage (Koch et al. 2012). Active nesting cavities were identified based on adult behavior (carrying nesting material or food or entering or exiting cavities) or the vocalizations of young (Martin et al. 2004). All cavities at the study site were given unique identifiers and their persistence and use was recorded in each year of the study. Between 1995 and 2004, we accessed active cavities up to 5.2 m above the ground using ladders and mirrors to assess the stage of breeding. One hundred and twelve nests (10% of the total dataset) were inaccessible during this 10-yr period. These inaccessible cavities were recorded as active based on adult behavior or begging chicks but could not be assigned a clutch initiation date or hatch date in the field. After 2004, all nests were accessible using a video camera mounted on a pole (TreeTop Peeper; Sandpiper Technologies, Manteca, California, USA) to identify the stage of breeding in cavities up to 15 m above the ground (Edworthy et al. 2012). Nests were checked, on average, every 5 days during their active phase and, when possible, clutch initiation date was determined using observed clutch size (if nests were found during laying) or observed final clutch size and hatch date (if nests were found during incubation), combined with a 1 egg day–1 laying interval and a 14- or 15-day mean incubation period for Mountain Bluebird and Tree Swallow, respectively (Koch et al. 2012). Nesting attempts found after young had hatched were not assigned a clutch initiation date in the field. Nest activity periods were recorded as the first and last days that each cavity was observed as actively being used (containing fresh nesting material, eggs, or nestlings). The number of fledglings produced by each nest was recorded when fledging was observed or when chicks were old enough at the last nest check to survive out of the nest and where there was no evidence of predation within the cavity or around the nesting site when the nest was found empty at the penultimate check. The presence of adults feeding chicks in the immediate nest area was also used as an indicator of success where fledging was not directly observed. Phenology Our initial dataset consisted of 590 Tree Swallow and 519 Mountain Bluebird breeding attempts, of which, 218 and 203 had direct information on clutch initiation date recorded in the field. The remaining records had phenology data (such as dates observed active, lay stage when found, and fledge date), but no clutch initiation date. We used this phenology data to infer clutch initiation dates and an associated prediction error for these records, as described below. We used known dates to inform a mixed-effects model predicting the timing of clutch initiation in the remaining records, and incorporated the prediction error for these records into our analyses by using multiple imputation (MI) from a conditional distribution (Schafer and Graham 2002, Schomaker and Heumann 2014; application to this system Drake and Martin 2018). This involved 3 steps. First, we conservatively backdated 86 nests found at the nestling stage to the egg stage using either reported clutch size (nestlings plus unhatched eggs, n = 17) or mean clutch size at the study site (5 eggs for both species, n = 69), a lay rate of 1 egg day–1, and the mean incubation period for each species (Winkler et al. 2011, Johnson and Dawson 2019). Second, we used known clutch initiation dates for both species (n = 417, 5 records lacked end dates so were dropped from the model) to model clutch initiation date as a function of (1) the date of the earliest known/backdated nest activity, (2) the stage of the nest at that date (pre-lay, egg, or unknown), (3) the latest date the nest was observed active, and (4) nest fate (fledged, failed pre-hatch, failed pre-fledge, or unknown). Species identity was included as a random factor (intercept). These models were run in R 3.6.0 (R Core Team 2019) using the package lme4 (Bates et al. 2015). We ran this model 5 times with a unique 20% set aside to test its predictive capability and to calculate a root mean-squared prediction error (RMSPE) for our modeled clutch initiation dates. The average performance of the model was high (5-run mean: both species: r = 0.98, n = 83.4; Tree Swallow alone, r = 0.94; Mountain Bluebird alone, r = 0.98) and the RMSPE was ± 2.86 days. Finally, we used the complete dataset (n = 417) to train the final model and predict clutch initiation dates for those nests lacking direct data. Error within these predicted dates was incorporated into all subsequent analyses by multiply imputing these dates using a Monte Carlo approach; predicted dates were adjusted in each imputation using values obtained from random draws of a normal distribution with a mean of zero and a standard deviation of the RMSPE calculated above (2.86 days). To restrict our analyses to initial nesting attempts, known second nesting attempts by the same breeding pair were removed and records were additionally limited to the first occupancy of each cavity in each year by each species. This removed possible re-nests in the same cavity by the same breeding pair, as well as late nesting attempts by pairs of the same species following abandonment or displacement of the first nesting pair. We could not, however, identify re-nesting if it occurred in cavities that had not been previously used within the year. We therefore cannot exclude the possibility that some re-nests made it into the final dataset. Our final dataset consisted of 462 Mountain Bluebird, and 572 Tree Swallow records (see Table 1 for annual sample sizes). We note that a subset of these records (112 bluebird and 89 swallow records) were used by Koch et al. (2012). TABLE 1. Annual number of breeding records for Mountain Bluebird, Tree Swallow, and European Starling as well as total suitable cavity availability (see Methods) at our monitoring sites in Central British Columbia, Canada. Percent values (bracketed) indicate cavity occupancy by availability for each species and were used as a metric of annual competition for nest sites. Italicized records at Knife Creek were dropped from final AIC analyses due to their small within-year sample size and the inclusion of “Site” as a random effect in these analyses (see Methods). Year Riske Creek site Knife Creek site Mountain Bluebird Tree Swallow European Starling Suitable cavity # Mountain Bluebird Tree Swallow European Starling Suitable cavity # 1995 17 (17%) 10 (10%) 9 (9%) 98 – – – – 1996 12 (9%) 10 (8%) 9 (7%) 132 0 1 (8%) 0 12 1997 14 (7%) 12 (6%) 29 (14%) 212 2 (5%) 9 (22%) 1 (2%) 40 1998 14 (5%) 17 (6%) 39 (15%) 267 3 (4%) 8 (10%) 1 (1%) 80 1999 8 (3%) 15 (5%) 37 (12%) 317 3 (3%) 7 (6%) 6 (5%) 117 2000 18 (5%) 21 (5%) 47 (12%) 387 0 9 (6%) 2 (1%) 150 2001 16 (4%) 22 (5%) 50 (11%) 435 0 11 (6%) 2 (1%) 183 2002 23 (5%) 22 (5%) 24 (5%) 476 3 (1%) 13 (6%) 4 (2%) 208 2003 24 (5%) 30 (6%) 32 (6%) 533 4 (2%) 13 (6%) 1 (0%) 221 2004 27 (5%) 40 (7%) 18 (3%) 597 1 (0%) 14 (5%) 0 273 2005 41 (6%) 34 (5%) 20 (3%) 631 3 (1%) 11 (4%) 1 (0%) 294 2006 24 (6%) 18 (4%) 14 (3%) 428 6 (3%) 9 (4%) 0 233 2007 30 (7%) 23 (5%) 10 (2%) 423 0 3 (1%) 1 (0%) 249 2008 33 (7%) 36 (8%) 12 (3%) 474 1 (0%) 7 (3%) 1 (0%) 263 2009 42 (10%) 44 (10%) 13 (3%) 436 5 (2%) 8 (3%) 0 285 2010 33 (7%) 50 (10%) 18 (4%) 481 4 (1%) 8 (3%) 0 282 2011 47 (11%) 33 (8%) 9 (2%) 420 4 (1%) 4 (1%) 0 283 Year Riske Creek site Knife Creek site Mountain Bluebird Tree Swallow European Starling Suitable cavity # Mountain Bluebird Tree Swallow European Starling Suitable cavity # 1995 17 (17%) 10 (10%) 9 (9%) 98 – – – – 1996 12 (9%) 10 (8%) 9 (7%) 132 0 1 (8%) 0 12 1997 14 (7%) 12 (6%) 29 (14%) 212 2 (5%) 9 (22%) 1 (2%) 40 1998 14 (5%) 17 (6%) 39 (15%) 267 3 (4%) 8 (10%) 1 (1%) 80 1999 8 (3%) 15 (5%) 37 (12%) 317 3 (3%) 7 (6%) 6 (5%) 117 2000 18 (5%) 21 (5%) 47 (12%) 387 0 9 (6%) 2 (1%) 150 2001 16 (4%) 22 (5%) 50 (11%) 435 0 11 (6%) 2 (1%) 183 2002 23 (5%) 22 (5%) 24 (5%) 476 3 (1%) 13 (6%) 4 (2%) 208 2003 24 (5%) 30 (6%) 32 (6%) 533 4 (2%) 13 (6%) 1 (0%) 221 2004 27 (5%) 40 (7%) 18 (3%) 597 1 (0%) 14 (5%) 0 273 2005 41 (6%) 34 (5%) 20 (3%) 631 3 (1%) 11 (4%) 1 (0%) 294 2006 24 (6%) 18 (4%) 14 (3%) 428 6 (3%) 9 (4%) 0 233 2007 30 (7%) 23 (5%) 10 (2%) 423 0 3 (1%) 1 (0%) 249 2008 33 (7%) 36 (8%) 12 (3%) 474 1 (0%) 7 (3%) 1 (0%) 263 2009 42 (10%) 44 (10%) 13 (3%) 436 5 (2%) 8 (3%) 0 285 2010 33 (7%) 50 (10%) 18 (4%) 481 4 (1%) 8 (3%) 0 282 2011 47 (11%) 33 (8%) 9 (2%) 420 4 (1%) 4 (1%) 0 283 Open in new tab TABLE 1. Annual number of breeding records for Mountain Bluebird, Tree Swallow, and European Starling as well as total suitable cavity availability (see Methods) at our monitoring sites in Central British Columbia, Canada. Percent values (bracketed) indicate cavity occupancy by availability for each species and were used as a metric of annual competition for nest sites. Italicized records at Knife Creek were dropped from final AIC analyses due to their small within-year sample size and the inclusion of “Site” as a random effect in these analyses (see Methods). Year Riske Creek site Knife Creek site Mountain Bluebird Tree Swallow European Starling Suitable cavity # Mountain Bluebird Tree Swallow European Starling Suitable cavity # 1995 17 (17%) 10 (10%) 9 (9%) 98 – – – – 1996 12 (9%) 10 (8%) 9 (7%) 132 0 1 (8%) 0 12 1997 14 (7%) 12 (6%) 29 (14%) 212 2 (5%) 9 (22%) 1 (2%) 40 1998 14 (5%) 17 (6%) 39 (15%) 267 3 (4%) 8 (10%) 1 (1%) 80 1999 8 (3%) 15 (5%) 37 (12%) 317 3 (3%) 7 (6%) 6 (5%) 117 2000 18 (5%) 21 (5%) 47 (12%) 387 0 9 (6%) 2 (1%) 150 2001 16 (4%) 22 (5%) 50 (11%) 435 0 11 (6%) 2 (1%) 183 2002 23 (5%) 22 (5%) 24 (5%) 476 3 (1%) 13 (6%) 4 (2%) 208 2003 24 (5%) 30 (6%) 32 (6%) 533 4 (2%) 13 (6%) 1 (0%) 221 2004 27 (5%) 40 (7%) 18 (3%) 597 1 (0%) 14 (5%) 0 273 2005 41 (6%) 34 (5%) 20 (3%) 631 3 (1%) 11 (4%) 1 (0%) 294 2006 24 (6%) 18 (4%) 14 (3%) 428 6 (3%) 9 (4%) 0 233 2007 30 (7%) 23 (5%) 10 (2%) 423 0 3 (1%) 1 (0%) 249 2008 33 (7%) 36 (8%) 12 (3%) 474 1 (0%) 7 (3%) 1 (0%) 263 2009 42 (10%) 44 (10%) 13 (3%) 436 5 (2%) 8 (3%) 0 285 2010 33 (7%) 50 (10%) 18 (4%) 481 4 (1%) 8 (3%) 0 282 2011 47 (11%) 33 (8%) 9 (2%) 420 4 (1%) 4 (1%) 0 283 Year Riske Creek site Knife Creek site Mountain Bluebird Tree Swallow European Starling Suitable cavity # Mountain Bluebird Tree Swallow European Starling Suitable cavity # 1995 17 (17%) 10 (10%) 9 (9%) 98 – – – – 1996 12 (9%) 10 (8%) 9 (7%) 132 0 1 (8%) 0 12 1997 14 (7%) 12 (6%) 29 (14%) 212 2 (5%) 9 (22%) 1 (2%) 40 1998 14 (5%) 17 (6%) 39 (15%) 267 3 (4%) 8 (10%) 1 (1%) 80 1999 8 (3%) 15 (5%) 37 (12%) 317 3 (3%) 7 (6%) 6 (5%) 117 2000 18 (5%) 21 (5%) 47 (12%) 387 0 9 (6%) 2 (1%) 150 2001 16 (4%) 22 (5%) 50 (11%) 435 0 11 (6%) 2 (1%) 183 2002 23 (5%) 22 (5%) 24 (5%) 476 3 (1%) 13 (6%) 4 (2%) 208 2003 24 (5%) 30 (6%) 32 (6%) 533 4 (2%) 13 (6%) 1 (0%) 221 2004 27 (5%) 40 (7%) 18 (3%) 597 1 (0%) 14 (5%) 0 273 2005 41 (6%) 34 (5%) 20 (3%) 631 3 (1%) 11 (4%) 1 (0%) 294 2006 24 (6%) 18 (4%) 14 (3%) 428 6 (3%) 9 (4%) 0 233 2007 30 (7%) 23 (5%) 10 (2%) 423 0 3 (1%) 1 (0%) 249 2008 33 (7%) 36 (8%) 12 (3%) 474 1 (0%) 7 (3%) 1 (0%) 263 2009 42 (10%) 44 (10%) 13 (3%) 436 5 (2%) 8 (3%) 0 285 2010 33 (7%) 50 (10%) 18 (4%) 481 4 (1%) 8 (3%) 0 282 2011 47 (11%) 33 (8%) 9 (2%) 420 4 (1%) 4 (1%) 0 283 Open in new tab To calculate the mean (±95% confidence interval [CI]) annual breeding dates reported in Figure 1, we generated 500 datasets using MI (see above) and calculated mean annual initiation dates and standard deviations for each run. Reported values are the averages of these 500 runs. We report the mean annual clutch initiation date for each species as “early” or “late” relative to the multi-year average for that species (Figure 1, green line). FIGURE 1. Open in new tabDownload slide Mean annual clutch initiation dates (days after April 1 ± 95% CI) for (A) Mountain Bluebirds and (B) Tree Swallows in south-central British Columbia. The green horizontal line represents the inter-annual average clutch initiation date for each species, with Tree Swallows initiating breeding, on average, 10 days later than Mountain Bluebird. Annual sample sizes can be found in Table 1. FIGURE 1. Open in new tabDownload slide Mean annual clutch initiation dates (days after April 1 ± 95% CI) for (A) Mountain Bluebirds and (B) Tree Swallows in south-central British Columbia. The green horizontal line represents the inter-annual average clutch initiation date for each species, with Tree Swallows initiating breeding, on average, 10 days later than Mountain Bluebird. Annual sample sizes can be found in Table 1. Weather Variables We obtained local rainfall data (mm) and daily minimum, maximum, and mean temperatures (°C) for the pre-breeding period from the Environment and Climate Change Canada weather station Williams Lake A (World Meteorological Organization [WMO] ID 71104; 52.1800°N, 122.0500°W; elevation 939.7 m; http://climate.weather.gc.ca) within 40 km of our field sites (see also Drake and Martin 2018). These values corresponded with incomplete data from a British Columbia Wildfire Service station at the Riske Creek study site (51.9603°N, 122.5000°W; elevation 929 m; Station 210) (Pearson’s r  =  0.80 for precipitation; Pearson’s r  =  0.95 and 0.97 for minimum and mean temperature, respectively). We obtained winter and migratory weather variables by specifying winter and migratory regions for Mountain Bluebird and Tree Swallow using NatureServe range maps (Ridgely et al. 2005), band return data (Environment and Climate Change Canada 2016), and geolocator data (Knight et al. 2018) for Tree Swallows (for maps, see Appendix). For Mountain Bluebird, we used the entire winter and migratory regions defined within this species’ NatureServe range map and also included breeding areas south of our study site within our defined migratory region. For the more broadly distributed Tree Swallow, geolocator (Knight et al. 2018) and band return data (Environment and Climate Change Canada 2016) supported the assumption that the individuals in British Columbia are a part of populations that remain in western North America throughout their annual cycle. We therefore restricted winter and migration ranges to the western United States and Mexico. We then collected spatially coded temperature, precipitation (surface), and wind vector data (averaged over the 850 and 925 millibar (mb) level [~1,500 and 700 m above mean sea level; Alerstam et al. 2011, Drake et al. 2014b, Huang et al. 2017]) from 1994 to 2011 from the National Center of Environmental Prediction (NCEP) Reanalysis 1 data archives at the National Oceanic and Atmospheric Administration-Cooperative Institute for Research in Environmental Sciences (NOAA-CIRES) Climate Diagnostics Center at Boulder, Colorado, USA (Kalnay et al. 1996) using R (package RNCEP; Kemp et al. 2012). In QGIS 2.10 (QGIS Development Team 2015), we restricted these data to spatial regions that overlapped with our defined wintering and migratory regions. For each species, we then calculated mean daily temperatures and precipitation over the entire winter region. We divided migratory regions into latitudinal blocks: 40–50°N, 30–40°N, and (for Tree Swallow only) 20–30°N based on the strong correlation of our weather variables within these blocks (that justified averaging), but not between them. Within each block of the migratory region we calculated mean daily temperature, precipitation, and (because both species are daytime migrants; Evans and O’Brien 2002, Winkler et al. 2011) diurnal (0600 to 1800 hours) east-west (U-) and north-south (V-) wind vectors. Nest Site Availability and Competition The availability of nest cavities changed dramatically over the 17-yr study period, with the fewest cavities available in 1996, and the most in 2005 (Cockle and Martin 2015). This change was largely due to a mountain pine beetle outbreak that peaked in 2005 (Cockle and Martin 2015). This outbreak killed pine trees and resulted in an increased density in primary excavators at the study site (Edworthy et al. 2011). Cavity availability subsequently declined with lower excavation rates and natural loss, post-outbreak. We quantified cavity availability for each breeding period (year) at our Riske and Knife Creek monitoring sites as the number of suitable cavities pre-existing, plus those identified in the current year, minus those lost over the previous winter (Table 1). The suitability of these cavities for our study species was based on observed use: we excluded all cavities made by excavators for which our focal species were rarely secondary occupants (<5% use by availability; e.g., Pileated Woodpecker [Dryocopus pileatus]) and, where excavator was unknown or the cavity naturally formed, we excluded cavities that had been initially occupied by secondary nesting species for which our focal species were rarely secondary occupants (<5% use by availability; e.g., northern flying squirrel [Glaucomys sabrinus]). We quantified direct competition between European Starlings and Mountain Bluebirds and Tree Swallows (Koenig 2003, Wiebe 2003, Koch et al. 2012) as the proportion of suitable cavities occupied by nesting starlings in a given year at each of our monitoring sites (Table 1). We also tested whether the abundance of nesting Mountain Bluebirds might delay Tree Swallow nesting given the overlap in cavity preferences between these 2 species and the fact that Mountain Bluebirds initiate earlier than Tree Swallows (Koch et al. 2012, Wiebe 2016). As with European Starlings, we quantified Mountain Bluebird competition as the proportion of suitable cavities occupied by active Mountain Bluebird nests in a given year at each of our monitoring sites (Table 1). Analysis All analyses were run in R 3.6.0 (R Core Team 2019). We used a linear mixed modeling approach to describe individual clutch initiation date as a function of our explanatory variables: “Year” was included in all models as a random effect (intercept) to account for the non-independence of data collected within the same year (package lme4; Bates et al. 2015). Models were competed using Akaike information criterion (AIC). To limit our final model candidate set, we used a 2-step hierarchical approach to model testing. In the first step, we identified weather variables that corresponded with timing; in the second step, we created the final model candidate set that tested these weather variables together with cavity availability and nest site competition metrics. In step 1, the performance of local, wintering, and migratory weather variables (temperature, rainfall, and migratory region wind speed) was tested using a sliding window approach (package climwin; van de Pol et al. 2016) covering the appropriate time window of the annual cycle for each species (Winkler et al. 2011, Johnson and Dawson 2019) but allowing the period assessed for both winter and breeding to encompass the spring migratory period (Appendix Table 4). Minimum window size was restricted to 10 days (providing time for a physiological reproductive response [i.e. ova development] to conditions) and the maximum window encompassed the entire time period (Williams 2012, Williams et al. 2015, Drake and Martin 2018). The best time window for each variable was determined using AIC. Sliding windows, by definition, represent multiple comparisons and result in an increased probability of false positives. We therefore calculated the probability that the AIC scores we obtained were the product of chance (i.e. Type I error) by using a response-data randomization program included for this purpose within climwin. Specifically, we considered our results to be spurious if our model AIC values did not differ significantly (P > 0.05) from those generated from sliding window analyses of 100 randomizations of the response data (for details, see van de Pol et al. 2016). Only weather variables that passed the Type I error test and that fell within ΔAIC ≤ 2 of the top performing weather variable (Burnham and Anderson 2004) were included in our final candidate set of models. Our final candidate model set included: standardized (z-score transformed), top-performing weather variables, cavity availability, and competition metrics on their own as well as biologically plausible additive models that included weather, cavity availability, and competition together. All models included “Site” as a random effect (intercept), in addition to “Year” (intercept), to account for the non-independence of breeding records within each study site. These were run against a null (the random effect of “Year” and “Site” only) model and a model that assumed a linear trend in phenology over the monitoring period (i.e. “Year” as a continuous variable). The final candidate set contained 13 models for Mountain Bluebird and 17 models for Tree Swallow (Tables 2 and 3, respectively). We excluded 11 years (1996–2002, 2004–2005, and 2007–2008; 16 breeding records) at Knife Creek for Mountain Bluebird and 2 years (1996 and 2007; 4 breeding records) at Knife Creek for Tree Swallow due to small sample sizes (<4 records) at these sites in those years (Table 1). For model testing, we imputed 100 datasets (see above) and bootstrapped each of these (n  =  20 per dataset), resulting in 2,000 runs of the candidate model set (100 MI × 20 bootstraps) (Schomaker and Heumann 2014). The influence (β value) of each candidate variable was then model-averaged over the entire candidate set within each run using AIC weights (wi) (Burnham and Anderson 2002, Burnham et al. 2011). The resulting value distributions were used to obtain our coefficients (Schomaker and Heumann 2014): reported beta values (β) are the median, model-averaged, β for each predictor variable, and the reported 85% and 95% CIs for these values are the 0.075–0.925 and 0.025–0.975 quantiles of their empirical distribution from the 2,000 MI × bootstrap runs (Figure 2). Variables with beta estimates whose 85% CI do not cross zero contribute to model fit under AIC selection criteria (Arnold 2010). We report 95% CIs for comparison with traditional P-value testing. To support our missing data approach, we re-ran our candidate models using only clutch initiation dates that were recorded in the field. These results are reported in the Appendix and do not differ substantively from the full analysis (Appendix Tables 5 and 6 and Appendix Figure 5). TABLE 2. Mountain Bluebird final candidate model ranking (n = 446 nests). Values under model parameters are standardized beta values. All values represent the median of 2,000 MI-bootstrap runs (see Methods). Null and annual trend models are italicized. Top models (≤ 4 ΔAIC) are in bold and represent 95% of total model support. k = number of parameters and wi = AIC weight. Model Intercept Year (trend) Cavity availability Starling occupancy Rainfall 40–50°N Westerly wind 40–50°N k loglik ΔAIC a wi MOBL7 147.70 – –3.18 – 1.73 – 6 –1,765.56 0.00 0.28 MOBL11 147.74 – –3.33 – – 1.46 6 –1,765.97 0.81 0.19 MOBL9 147.55 – –3.09 0.26 1.70 – 7 –1,765.16 1.26 0.15 MOBL3 148.01 – –4.21 – – – 5 –1,767.24 1.30 0.15 MOBL13 147.48 – –3.11 0.45 – 1.47 7 –1,765.46 1.86 0.11 MOBL5 147.81 – –4.02 0.38 – – 6 –1,766.89 2.65 0.07 MOBL6 145.85 – – – 3.49 – 5 –1,769.51 5.84 0.02 MOBL8 145.42 – – 0.99 3.15 – 6 –1,768.91 6.70 0.01 MOBL10 145.86 – – – – 3.37 5 –1,770.04 6.89 0.01 MOBL12 145.31 – – 1.32 – 3.06 6 –1,769.07 7.02 0.01 MOBL2 146.11 –3.24 – – – – 5 –1,770.49 7.80 0.01 MOBL4 144.91 – – 1.82 – – 5 –1,772.93 12.67 0.00 MOBL1 145.76 – – – – – 4 –1,774.10 12.98 0.00 Model Intercept Year (trend) Cavity availability Starling occupancy Rainfall 40–50°N Westerly wind 40–50°N k loglik ΔAIC a wi MOBL7 147.70 – –3.18 – 1.73 – 6 –1,765.56 0.00 0.28 MOBL11 147.74 – –3.33 – – 1.46 6 –1,765.97 0.81 0.19 MOBL9 147.55 – –3.09 0.26 1.70 – 7 –1,765.16 1.26 0.15 MOBL3 148.01 – –4.21 – – – 5 –1,767.24 1.30 0.15 MOBL13 147.48 – –3.11 0.45 – 1.47 7 –1,765.46 1.86 0.11 MOBL5 147.81 – –4.02 0.38 – – 6 –1,766.89 2.65 0.07 MOBL6 145.85 – – – 3.49 – 5 –1,769.51 5.84 0.02 MOBL8 145.42 – – 0.99 3.15 – 6 –1,768.91 6.70 0.01 MOBL10 145.86 – – – – 3.37 5 –1,770.04 6.89 0.01 MOBL12 145.31 – – 1.32 – 3.06 6 –1,769.07 7.02 0.01 MOBL2 146.11 –3.24 – – – – 5 –1,770.49 7.80 0.01 MOBL4 144.91 – – 1.82 – – 5 –1,772.93 12.67 0.00 MOBL1 145.76 – – – – – 4 –1,774.10 12.98 0.00 a Lowest median AIC value was 3,543.32. Open in new tab TABLE 2. Mountain Bluebird final candidate model ranking (n = 446 nests). Values under model parameters are standardized beta values. All values represent the median of 2,000 MI-bootstrap runs (see Methods). Null and annual trend models are italicized. Top models (≤ 4 ΔAIC) are in bold and represent 95% of total model support. k = number of parameters and wi = AIC weight. Model Intercept Year (trend) Cavity availability Starling occupancy Rainfall 40–50°N Westerly wind 40–50°N k loglik ΔAIC a wi MOBL7 147.70 – –3.18 – 1.73 – 6 –1,765.56 0.00 0.28 MOBL11 147.74 – –3.33 – – 1.46 6 –1,765.97 0.81 0.19 MOBL9 147.55 – –3.09 0.26 1.70 – 7 –1,765.16 1.26 0.15 MOBL3 148.01 – –4.21 – – – 5 –1,767.24 1.30 0.15 MOBL13 147.48 – –3.11 0.45 – 1.47 7 –1,765.46 1.86 0.11 MOBL5 147.81 – –4.02 0.38 – – 6 –1,766.89 2.65 0.07 MOBL6 145.85 – – – 3.49 – 5 –1,769.51 5.84 0.02 MOBL8 145.42 – – 0.99 3.15 – 6 –1,768.91 6.70 0.01 MOBL10 145.86 – – – – 3.37 5 –1,770.04 6.89 0.01 MOBL12 145.31 – – 1.32 – 3.06 6 –1,769.07 7.02 0.01 MOBL2 146.11 –3.24 – – – – 5 –1,770.49 7.80 0.01 MOBL4 144.91 – – 1.82 – – 5 –1,772.93 12.67 0.00 MOBL1 145.76 – – – – – 4 –1,774.10 12.98 0.00 Model Intercept Year (trend) Cavity availability Starling occupancy Rainfall 40–50°N Westerly wind 40–50°N k loglik ΔAIC a wi MOBL7 147.70 – –3.18 – 1.73 – 6 –1,765.56 0.00 0.28 MOBL11 147.74 – –3.33 – – 1.46 6 –1,765.97 0.81 0.19 MOBL9 147.55 – –3.09 0.26 1.70 – 7 –1,765.16 1.26 0.15 MOBL3 148.01 – –4.21 – – – 5 –1,767.24 1.30 0.15 MOBL13 147.48 – –3.11 0.45 – 1.47 7 –1,765.46 1.86 0.11 MOBL5 147.81 – –4.02 0.38 – – 6 –1,766.89 2.65 0.07 MOBL6 145.85 – – – 3.49 – 5 –1,769.51 5.84 0.02 MOBL8 145.42 – – 0.99 3.15 – 6 –1,768.91 6.70 0.01 MOBL10 145.86 – – – – 3.37 5 –1,770.04 6.89 0.01 MOBL12 145.31 – – 1.32 – 3.06 6 –1,769.07 7.02 0.01 MOBL2 146.11 –3.24 – – – – 5 –1,770.49 7.80 0.01 MOBL4 144.91 – – 1.82 – – 5 –1,772.93 12.67 0.00 MOBL1 145.76 – – – – – 4 –1,774.10 12.98 0.00 a Lowest median AIC value was 3,543.32. Open in new tab TABLE 3. Tree Swallow final candidate model ranking (n = 568 nests). Values under model parameters are standardized beta values. All values represent the median of 2,000 MI-bootstrap runs (see Methods). Null and annual trend models are italicized. Top models (≤ 4 ΔAIC) are in bold and represent 97% of total model support. k = number of parameters and wi = AIC weight. Model Intercept Year (trend) Local rainfall Cavity availability Starling occupancy Bluebird occupancy k loglik ΔAIC a wi TRES8 156.18 – 2.55 – 1.80 – 6 –2,114.50 0.00 0.32 TRES14 156.36 – 2.46 –0.61 1.78 – 7 –2,113.59 0.24 0.28 TRES15 156.28 – 2.57 – 1.80 –0.40 7 –2,113.89 0.83 0.21 TRES17 156.40 – 2.48 –0.53 1.80 –0.23 8 –2,113.09 1.28 0.17 TRES3 156.37 – 2.80 – – – 5 –2,119.27 7.49 0.01 TRES7 156.44 – 2.66 –0.72 – – 6 –2,118.57 8.13 0.01 TRES13 156.45 – 2.64 –1.04 – –0.49 7 –2,117.73 8.52 0.00 TRES10 156.38 – 2.81 – – –0.11 6 –2,118.89 8.78 0.00 TRES9 156.32 – – –1.11 2.22 – 6 –2,119.10 9.20 0.00 TRES5 156.13 – – – 2.05 – 5 –2,120.76 10.47 0.00 TRES16 156.32 – – –1.11 2.22 0.00 7 –2,118.78 10.60 0.00 TRES12 156.20 – – – 2.06 –0.42 6 –2,120.29 11.57 0.00 TRES2 156.55 –2.18 – – – – 5 –2,122.60 14.16 0.00 TRES4 156.28 – – –1.61 – – 5 –2,124.02 17.00 0.00 TRES1 156.34 – – – – – 4 –2,125.14 17.19 0.00 TRES11 156.30 – – –2.32 – –1.08 6 –2,123.22 17.44 0.00 TRES6 156.32 – – – – 0.11 5 –2,124.86 18.68 0.00 Model Intercept Year (trend) Local rainfall Cavity availability Starling occupancy Bluebird occupancy k loglik ΔAIC a wi TRES8 156.18 – 2.55 – 1.80 – 6 –2,114.50 0.00 0.32 TRES14 156.36 – 2.46 –0.61 1.78 – 7 –2,113.59 0.24 0.28 TRES15 156.28 – 2.57 – 1.80 –0.40 7 –2,113.89 0.83 0.21 TRES17 156.40 – 2.48 –0.53 1.80 –0.23 8 –2,113.09 1.28 0.17 TRES3 156.37 – 2.80 – – – 5 –2,119.27 7.49 0.01 TRES7 156.44 – 2.66 –0.72 – – 6 –2,118.57 8.13 0.01 TRES13 156.45 – 2.64 –1.04 – –0.49 7 –2,117.73 8.52 0.00 TRES10 156.38 – 2.81 – – –0.11 6 –2,118.89 8.78 0.00 TRES9 156.32 – – –1.11 2.22 – 6 –2,119.10 9.20 0.00 TRES5 156.13 – – – 2.05 – 5 –2,120.76 10.47 0.00 TRES16 156.32 – – –1.11 2.22 0.00 7 –2,118.78 10.60 0.00 TRES12 156.20 – – – 2.06 –0.42 6 –2,120.29 11.57 0.00 TRES2 156.55 –2.18 – – – – 5 –2,122.60 14.16 0.00 TRES4 156.28 – – –1.61 – – 5 –2,124.02 17.00 0.00 TRES1 156.34 – – – – – 4 –2,125.14 17.19 0.00 TRES11 156.30 – – –2.32 – –1.08 6 –2,123.22 17.44 0.00 TRES6 156.32 – – – – 0.11 5 –2,124.86 18.68 0.00 a Lowest median AIC value was 4,241.15. Open in new tab TABLE 3. Tree Swallow final candidate model ranking (n = 568 nests). Values under model parameters are standardized beta values. All values represent the median of 2,000 MI-bootstrap runs (see Methods). Null and annual trend models are italicized. Top models (≤ 4 ΔAIC) are in bold and represent 97% of total model support. k = number of parameters and wi = AIC weight. Model Intercept Year (trend) Local rainfall Cavity availability Starling occupancy Bluebird occupancy k loglik ΔAIC a wi TRES8 156.18 – 2.55 – 1.80 – 6 –2,114.50 0.00 0.32 TRES14 156.36 – 2.46 –0.61 1.78 – 7 –2,113.59 0.24 0.28 TRES15 156.28 – 2.57 – 1.80 –0.40 7 –2,113.89 0.83 0.21 TRES17 156.40 – 2.48 –0.53 1.80 –0.23 8 –2,113.09 1.28 0.17 TRES3 156.37 – 2.80 – – – 5 –2,119.27 7.49 0.01 TRES7 156.44 – 2.66 –0.72 – – 6 –2,118.57 8.13 0.01 TRES13 156.45 – 2.64 –1.04 – –0.49 7 –2,117.73 8.52 0.00 TRES10 156.38 – 2.81 – – –0.11 6 –2,118.89 8.78 0.00 TRES9 156.32 – – –1.11 2.22 – 6 –2,119.10 9.20 0.00 TRES5 156.13 – – – 2.05 – 5 –2,120.76 10.47 0.00 TRES16 156.32 – – –1.11 2.22 0.00 7 –2,118.78 10.60 0.00 TRES12 156.20 – – – 2.06 –0.42 6 –2,120.29 11.57 0.00 TRES2 156.55 –2.18 – – – – 5 –2,122.60 14.16 0.00 TRES4 156.28 – – –1.61 – – 5 –2,124.02 17.00 0.00 TRES1 156.34 – – – – – 4 –2,125.14 17.19 0.00 TRES11 156.30 – – –2.32 – –1.08 6 –2,123.22 17.44 0.00 TRES6 156.32 – – – – 0.11 5 –2,124.86 18.68 0.00 Model Intercept Year (trend) Local rainfall Cavity availability Starling occupancy Bluebird occupancy k loglik ΔAIC a wi TRES8 156.18 – 2.55 – 1.80 – 6 –2,114.50 0.00 0.32 TRES14 156.36 – 2.46 –0.61 1.78 – 7 –2,113.59 0.24 0.28 TRES15 156.28 – 2.57 – 1.80 –0.40 7 –2,113.89 0.83 0.21 TRES17 156.40 – 2.48 –0.53 1.80 –0.23 8 –2,113.09 1.28 0.17 TRES3 156.37 – 2.80 – – – 5 –2,119.27 7.49 0.01 TRES7 156.44 – 2.66 –0.72 – – 6 –2,118.57 8.13 0.01 TRES13 156.45 – 2.64 –1.04 – –0.49 7 –2,117.73 8.52 0.00 TRES10 156.38 – 2.81 – – –0.11 6 –2,118.89 8.78 0.00 TRES9 156.32 – – –1.11 2.22 – 6 –2,119.10 9.20 0.00 TRES5 156.13 – – – 2.05 – 5 –2,120.76 10.47 0.00 TRES16 156.32 – – –1.11 2.22 0.00 7 –2,118.78 10.60 0.00 TRES12 156.20 – – – 2.06 –0.42 6 –2,120.29 11.57 0.00 TRES2 156.55 –2.18 – – – – 5 –2,122.60 14.16 0.00 TRES4 156.28 – – –1.61 – – 5 –2,124.02 17.00 0.00 TRES1 156.34 – – – – – 4 –2,125.14 17.19 0.00 TRES11 156.30 – – –2.32 – –1.08 6 –2,123.22 17.44 0.00 TRES6 156.32 – – – – 0.11 5 –2,124.86 18.68 0.00 a Lowest median AIC value was 4,241.15. Open in new tab FIGURE 2. Open in new tabDownload slide Model-averaged standardized beta (β) values and their 85% and 95% confidence intervals (CIs) for weather and ecological variables expected to impact annual average clutch initiation dates as well as a linear change in phenology over the monitoring period for (A) Mountain Bluebird and (B) Tree Swallows. Positive beta values indicate breeding delays associated with increasing values of a given variable, negative beta values indicate breeding advancement. CIs are the empirical distribution of beta values for 2,000 MI-bootstrap runs (see Methods). Non-symmetrical CIs are due to model averaging: for example, beta values for a linear change in breeding dates over the monitoring period were consistently negative, but this model received little support within either species’ candidate set (wi < 0.01; Tables 2 and 3, Model 2). Model averaging therefore drove beta values towards zero in many MI-bootstrap runs. FIGURE 2. Open in new tabDownload slide Model-averaged standardized beta (β) values and their 85% and 95% confidence intervals (CIs) for weather and ecological variables expected to impact annual average clutch initiation dates as well as a linear change in phenology over the monitoring period for (A) Mountain Bluebird and (B) Tree Swallows. Positive beta values indicate breeding delays associated with increasing values of a given variable, negative beta values indicate breeding advancement. CIs are the empirical distribution of beta values for 2,000 MI-bootstrap runs (see Methods). Non-symmetrical CIs are due to model averaging: for example, beta values for a linear change in breeding dates over the monitoring period were consistently negative, but this model received little support within either species’ candidate set (wi < 0.01; Tables 2 and 3, Model 2). Model averaging therefore drove beta values towards zero in many MI-bootstrap runs. Clutch Initiation Date and Breeding Productivity We tested the impact of clutch initiation date on productivity for a subset of records for which we had data on clutch size and nest outcome (n = 229 for Mountain Bluebird and n = 177 for Tree Swallow). All of these records were from 2000 to 2011 (see Appendix Table 7 for annual sample sizes and productivity). We assumed timing could impact productivity at 2 points in a pathway between lay and fledging: (1) by influencing clutch size (nest investment) and then (2) by influencing nestling survival to fledging. We split timing into 2 components. First, we used mean clutch initiation dates within each year to assess the impact of “late” vs. “early” years for the population as a whole. We then calculated individual deviation from these annual means (clutch initiation date-annual average clutch initiation date) to assess the individual impact of being late or early relative to conspecifics, within year. For our productivity pathway, we tested (1) whether clutch size varied as a function of mean annual clutch initiation date and individual deviation from that mean. Clutch size was fit using a Conway-Maxwell-Poisson distribution to account for under-dispersion (package spaMM; Rousset and Ferdy 2014). We then tested (2) whether the number of fledglings produced by each nest varied as a function of clutch size, mean annual clutch initiation date, and individual deviation from that mean. Fledgling number best fit a Poisson distribution with zero inflation. We therefore tested both zero-inflated Poisson and Poisson hurdle models: hurdle models performed best for both species. Preliminary analysis additionally indicated that clutch size did not contribute to the binomial component of the hurdle model in Tree Swallows (i.e. nest failure) and we therefore dropped this term. Year was included in all models as a random effect (intercept only) to account for the non-independence of records within-year. Reported coefficients and standard error (SE) values (from which we calculated CIs) are the medians of 500 MI of each dataset. We additionally ran a simpler test for population-level effects of breeding delay by regressing mean fledge number within-year against mean annual clutch initiation date (n = 14, 1997–2011) for both species and weighting the response values by annual sample size. RESULTS Over our 17-yr monitoring period, average annual clutch initiation dates for Mountain Bluebird ranged from May 19 to June 8 (20 days) with an inter-annual average of May 26, while annual average dates for Tree Swallow ranged from May 30 to June 12 (13 days) with an inter-annual average date of June 6 (Figure 1). Among Mountain Bluebird, 80% of individual lay dates were within a 32-day window around this species’ annual average lay date (14 days prior to 18 days after). For Tree Swallows, 80% of lay dates were within a 25-day window (11 days prior to 14 days after the annual average lay date). Thus, phenology differences among individuals within-year were much larger than the observed differences in timing among years for these populations as a whole. The abundance of nesting Mountain Bluebirds and Tree Swallows fluctuated 5- and 3-fold, respectively, between 1998 and 2011 (where survey effort was constant). At Riske Creek, both species showed an increasing trend that occurred concurrently with increasing cavity abundance, and local lows in 2006–2007 that corresponded with decreased cavity availability. As a result, cavity occupancy by availability remained relatively stable over time (Table 1). The abundance of nesting European Starlings fluctuated 6-fold between 1998 and 2011, reaching a peak in 2000–2001, before collapsing and remaining at low numbers up to 2011 (Table 1). Factors Influencing Clutch Initiation Date Mountain Bluebird. Sliding window analyses of the correlation between annual timing and weather identified conditions in the northern portion (40–50°N) of the migratory range as possible predictors of Mountain Bluebird phenology. Greater rainfall and higher wind speeds within this region were associated with later lay dates for this species (best temporal windows: February 4 to March 24 and February 16–26, respectively). These 2 variables were included in the final candidate model set; other weather variables were either not competitive or failed to pass Type I error tests (Appendix Table 4A). Mountain Bluebird phenology was most strongly correlated with cavity number, with earlier breeding occurring in years with greater cavity availability (β = –3.05, Figures 2A and 3A). Weaker effects of migratory conditions between 40 and 50°N in the western US states (Washington State, Oregon, and Idaho; Appendix Figure 4) were indicated, with later breeding occurring in higher rainfall and stronger westerly wind speed years (rainfall β = 0.64; westerly wind speed β = 0.37 [only the 85% CI did not cross zero]; Figure 2A). Both the 85% and 95% CIs of European Starling competition crossed zero and we conclude that this factor did not predict timing (Figure 2A). There was a non-zero effect of year, indicating advancement in breeding dates over the monitoring period; however, this model performed poorly relative to our other candidate models and its model-averaged beta value was close to zero (β = –0.02 or a 0.05 day advancement over the 17-yr period; Figure 2A and Table 2). In real terms (and holding all other variables constant), our results indicated that Mountain Bluebirds advanced nesting by 3.8 days for every 200 additional cavities available, and shifted breeding by 10.2 and 5.4 days across the cavity extremes at Riske and Knife Creek, respectively (Figure 3A). Breeding delays of 1.5 days were associated with every additional 1 mm of daily rainfall in the northern migratory region and breeding dates shifted by 2.1 days across the precipitation extremes in our dataset. Delays of 1.1 days occurred when westerly winds increased by 5 m/s within the northern migratory region and breeding dates shifted by 1.3 days across the extremes in our dataset. Tree Swallow. Sliding window analyses of the correlation between annual timing and weather identified local spring rainfall at the study site (best temporal window: March 13 to May 22) as a possible predictor of phenology in Tree Swallow. Greater spring rainfall was associated with later lay dates; no other weather variables were competitive and only this variable was included in the final candidate model set (Appendix Table 4B). Tree Swallow phenology was most strongly correlated with spring rainfall at the study site, with delays occurring in higher rainfall years (β = 2.48; Figures 2B and 3B). European Starling prevalence was also correlated with breeding delays (β = 1.74; Figure 2B). Changes in cavity availability and nesting Mountain Bluebird abundance were not correlated with Tree Swallow timing (Figure 2B). As in our Mountain Bluebird models, there was a non-zero relationship with “Year”, indicating an advancement in breeding dates over the monitoring period; however, as with Mountain Bluebirds, this temporal model performed poorly relative to the other candidate models and its model-averaged beta value was negligible (β = –0.0006 or a 0.002 day advancement over the 17-yr period; Figure 2B and Table 3). In real terms (and holding all other variables constant), our results indicated that Tree Swallow nesting was delayed by 8.6 days when average daily local rainfall between March 13 and May 22 increased by 1 mm and shifted by 8.2 days across the extremes of rainfall in our 17-yr monitoring period (Figure 3B). Tree Swallows additionally showed a 4.0-day delay in clutch initiation with each 0.1 increase in the proportion of nest sites occupied by starlings; breeding dates shifted by 4.9 and 2.0 days across the extremes of starling occupancy at Riske and Knife Creek, respectively. Productivity and Clutch Initiation Dates Within our subset of records with known outcomes, Mountain Bluebirds (n = 229 over 12 yr) laid an average of 5.2 eggs per clutch (range: 2–7) and fledged an average of 2.4 young (range: 0–7); 61% of nests were successful (fledged at least 1 young). Tree Swallows (n = 177 over 12 yr) laid an average of 5.4 eggs per clutch (range: 2–8) and fledged an average of 2.1 young (range: 0–7); 50% of nests were successful (see Appendix Table 7 for annual values). Variation in annual average clutch initiation date was not associated with shifts in clutch size for either of our focal species (Mountain Bluebird β [95% CI]: –0.15 [0.25, –0.56]; Tree Swallow: –0.09 [0.31, –0.50]). Individual deviation from annual average clutch initiation dates was associated with egg number, such that those individuals breeding late, relative to their conspecifics within-year, laid smaller clutches (Mountain Bluebird: –0.34 [–0.17, –0.51]; Tree Swallow: –0.25 [–0.07, –0.42]). For Mountain Bluebird, individual delay within-year (β [95% CI]: 0.42 [0.13, 0.71]) and possibly larger clutch sizes (0.29 [–0.02, 0.60]) were associated with an increased probability of nest failure (the binomial portion of the fledgling hurdle model), while annual average initiation date was not (–0.05 [–0.73, 0.62]). After accounting for nest failure, clutch size alone predicted ultimate fledge number (0.14 [0.05, 0.23]). As a result, being late within-year reduced individual Mountain Bluebird productivity first via decreased clutch size, and second via a direct relationship between timing and nest failure. For Tree Swallow, individual delay within-year was possibly associated with nest failure (β [95% CI]: 0.26 [–0.07, 0.59]; 85% CI does not cross zero) while annual average initiation was not (0.06 [–0.56, 0.69]). After accounting for complete nest failure, clutch size alone predicted ultimate fledge number (0.15 [0.04, 0.27]). Thus, although the relationships were weaker, being late within-year impacted individual Tree Swallow productivity in the same manner as Mountain Bluebird: first via decreased clutch size, and second via a direct relationship between timing and nest failure. Our simpler analysis supported the findings above, with no relationship between mean annual fledge number and mean annual clutch initiation date for either species (Mountain Bluebird: r2 = 0.01, P = 0.31; Tree Swallow: r2 = –0.04, P = 0.52; n = 14 yr). DISCUSSION Our analyses examined the relative importance of weather cues (both breeding and nonbreeding) and ecological constraints in determining breeding phenology in 2 migratory, cavity-nesting species. Despite their migratory status, the timing of breeding for both Mountain Bluebirds and Tree Swallows in our study showed the strongest relationships with breeding ground factors: cavity availability in the case of Mountain Bluebirds, and spring rainfall and European Starling competition in the case of Tree Swallows. An effect of migratory conditions on timing—in the form of heavier rainfall and stronger westerly winds aloft, in the region immediately south of our breeding site—was also indicated for Mountain Bluebirds. The temporal window for these effects (February and March) are consistent with when our birds were most likely moving through this region, based on arrival dates for this species within the Williams Lake area (Johnson and Dawson 2019); standardized beta values put these migratory impacts at 5 × and 8 × weaker than that of cavity availability. The shift to earlier breeding by Mountain Bluebirds when cavities were more numerous may result from an increased opportunity to find suitable nest sites, and possibly nest sites away from conspecifics (Holt and Martin 1997). Corresponding changes in the number of breeding pairs at Riske Creek as cavities became more or less numerous would support the idea that our study sites were saturated and that more potential breeders were able to obtain a nesting site upon arrival from migration when cavity numbers permitted this. Alternatively, increased bluebird abundance might indicate that our sites were more desirable at the landscape level in some years, due to cavity number increases, or other changes associated with the mountain pine beetle outbreak (such as forest structure or food availability; Martin et al. 2006). Earlier breeding in these years might then be the product of preferential settlement. Tree Swallows did not show breeding advancement as cavities became more numerous. This is consistent with previous work where Tree Swallows did not show a numerical response to manipulations of cavity availability (Aitken and Martin 2008). Tree Swallows are more likely than Mountain Bluebirds to use cavities excavated by Red-naped Sapsuckers (Sphyrapicus nuchalis), which were the most abundant cavity-type at our field sites during the monitoring period (52% of all cavities; Robles and Martin 2013); and, unlike Mountain Bluebirds, Tree Swallows are not adverse to nesting in close proximity to each other (Holt and Martin 1997); thus, it is possible there was a sufficient number of suitable cavities for this species throughout the monitoring period. Tree Swallows did show delays with increasing European Starling but not Mountain Bluebird abundance at our study sites. Starling nest site preferences overlap with both of our study species, but starlings initiate clutches, on average, 1 week earlier than Mountain Bluebirds and 2 weeks earlier than Tree Swallows (Koch et al. 2012). This difference in phenology may explain the differences in the responses of our focal species to European Starling abundance: more starling pairs would have the advantage of prior ownership when Tree Swallows were seeking to obtain nest sites than when Mountain Bluebirds were acquiring nest sites. At the patch level, within our study site, early nesting in the presence of high starling densities has been associated with reduced breeding success for Tree Swallows, particularly when using flicker cavities (Koch et al. 2012, Robles and Martin 2013), and it is possible that swallows delay clutch initiation to reduce competition with starlings. The absence of an effect of Mountain Bluebird abundance on Tree Swallow timing may reflect the higher resource holding potential of Tree Swallows, as indicated by nest-box studies (Wiebe 2016). Weather effects on timing were evident in Tree Swallows, with wetter springs associated with late breeding. Our best temporal window for this weather effect ranged from March 13 to May 22, encompassing the period after these migrants arrived at the study site and just prior to when the majority initiated breeding. We do not have data on insect abundance but it is likely that rainfall delayed clutch initiation by reducing aerial insect availability for foraging adults (Winkler and Allen 1996, Irons et al. 2017). Aerial prey were less active on wet days (Poulsen 1996, Fournier et al. 2005, Grüebler et al. 2008) and resultant calorie limitation would slow female mass gain, post-arrival. Female Tree Swallows in eastern Canada now initiate breeding at a lower body mass than they did 20 yr ago, an effect that has been attributed to an increase in spring rainfall and therefore reduced food availability (Cox et al. 2019). Our calorie limitation hypothesis is also supported by negative relationships between rainfall during nesting development and parental provisioning, nestling mass, and survival (McCarty 2009, Cox et al. 2019). In our study region, frequency of rainfall during the breeding period is at its greatest in June, with an average of 11 days with >1 mm (1994–2011) during that month. Rainfall frequency then declines to 10 and 8 days in July and August, respectively. Wet springs are not correlated with wet summers, indicating that poor weather in the pre-breeding period does not provide Tree Swallows at our study site with any information as to conditions later in the season. In contrast to a continent-wide analysis (Dunn and Winkler 1999, Hussell 2003), we found no support for an effect of spring breeding ground temperatures on timing within our population (Appendix Table 4B). Wind and rainfall, rather than temperature, had the greatest impact on Tree Swallow breeding phenology in Alaska as well (Irons et al. 2017). Such differences may reflect regional differences in which environmental variables most limit aerial insect abundance. We would expect breeding delays associated with competition, nest site availability, and hostile local and migratory weather conditions to be costly. However, neither Mountain Bluebird nor Tree Swallow showed population-level reductions in productivity in late breeding years. It is probable that inter-annual shifts in timing (a maximum of 20 and 13 days, respectively) were not large enough to reduce productivity. If population-level delays had an equivalent impact on productivity as individual within-year delays, we would predict a decline of 0.8 fledglings for Mountain Bluebirds with a 20-day shift in lay date and a decline of 0.5 fledglings for Tree Swallows with a 13-day shift in lay date. However, population-level effects of timing on productivity are likely smaller than the effect of within-year delays for individuals. This is because late-breeding individuals within-year are often younger and/or have poorer quality territories than early-breeding individuals: factors that independently reduce productivity (Svensson and Nilsson 1995, Winkler and Allen 1996, Johnson and Dawson 2019). Mountain Bluebirds will attempt second broods if they initiate first broods early in the breeding period (Johnson and Dawson 2019) and, therefore, it is possible that the frequency of repeat nesting—and therefore total annual productivity—was lower at our sites for this species in late years. Our data did not allow us to assess this. Tree Swallows are single brooded in the northern parts of their range (Winkler et al. 2011) and our results therefore reflect annual productivity. Mountain Bluebird and Tree Swallow exhibited different responses to conditions than resident cavity-nesting species in south-central British Columbia. Like these 2 migrants, the majority of resident cavity nesters—a group including 4 woodpecker species and 3 small insectivores—showed marked phenological responses to local conditions; however, among residents, temperature was the key driver of timing as opposed to rainfall or cavity availability (Drake and Martin 2018). Among resident non-obligate excavators, Mountain Chickadee (Poecile gambeli; a secondary cavity nester) showed no response to cavity availability, while Red-breasted Nuthatch (Sitta canadensis; a facultative excavator) bred later as cavities became more numerous (Drake and Martin 2018). Our methodology allowed us to compare the relative strength of nonbreeding and local factors in determining the timing of breeding for our 2 migratory species. Our results indicate that breeding ground factors most strongly influence the phenology of Tree Swallows and Mountain Bluebirds. Nonbreeding conditions, often cited as a possible contributor to mistiming in migratory birds (e.g., Franks et al. 2018), had relatively little impact on clutch initiation dates over our 17-yr monitoring period. The fact that cavity number and the abundance of nest site competitors were strongly correlated with breeding phenology reinforces work that shows suitable cavities can be limiting for our focal species and emphasizes the importance of this resource for population management (Holt and Martin 1997, Wesolowski and Martin 2018). The relationship between increased local precipitation and breeding delays for Tree Swallow contributes to a body of literature showing that rainfall during the breeding period is costly for this species (Dawson 2008, Irons et al. 2017, Cox et al. 2019, but see McCarty and Winkler 1999). Despite these relationships, population-level shifts in timing among years were dwarfed by individual variation in timing within-year, indicating that individual-level factors (such as body condition, age, or territory quality) were more powerful drivers of timing for individuals than the range of external conditions they experienced within our study. Although population-level breeding delays are often assumed to result in similar productivity costs as within-year breeding delays for individuals, our work illustrates that such delays are not necessarily costly. FIGURE 3. Open in new tabDownload slide (A) Observed advancement in annual average Mountain Bluebird clutch initiation dates (days after April 1 ± 95% CI) as a function of increasing cavity availability. Timing is presented by study site (filled points = “Riske Creek”, hollow points = “Knife Creek”) because cavity availability differed between sites within year. (B) Observed delay in annual average Tree Swallow clutch initiation dates (days after April 1 ± 95% CI) as a function of increasing spring rainfall (average daily mm, March 13 to May 22). Numbers above error bars indicate study years (1995–2011). FIGURE 3. Open in new tabDownload slide (A) Observed advancement in annual average Mountain Bluebird clutch initiation dates (days after April 1 ± 95% CI) as a function of increasing cavity availability. Timing is presented by study site (filled points = “Riske Creek”, hollow points = “Knife Creek”) because cavity availability differed between sites within year. (B) Observed delay in annual average Tree Swallow clutch initiation dates (days after April 1 ± 95% CI) as a function of increasing spring rainfall (average daily mm, March 13 to May 22). Numbers above error bars indicate study years (1995–2011). Funding statement: Research funding came from the Natural Sciences and Engineering Research Council of Canada (NSERC), Environment and Climate Change Canada, the Sustainable Forest Management Network, part of Networks of Centres of Excellence of Canada, and Forest Renewal BC grants to K.M. Support for A.D. came from post-doctoral fellowships from TerreWEB (NSERC, UBC). Ethics statement: All protocols complied with the guidelines for the ethical use of animals and research, set forth by the Canadian Council on Animal Care, and were approved by the Animal Care and Biosafety Committee of the University of British Columbia (permit numbers: A04-0101 and A07-130). Author contributions: A.D. and K.M. conceived the idea and methods. K.M. collected data. A.D. analyzed the data and wrote the manuscript, K.M. edited the work. Data deposits: Analyses reported in this article can be reproduced using the data provided by Drake and Martin (2020). ACKNOWLEDGMENTS We thank numerous field assistants and graduate students who contributed to data collection. LITERATURE CITED Aitken , K. E. H. , and K. Martin ( 2007 ). The importance of excavators in hole-nesting communities: Availability and use of natural tree holes in old mixed forests of western Canada . Journal of Ornithology 148 : 425 – 434 . Google Scholar Crossref Search ADS WorldCat Aitken , K. E. H. , and K. Martin ( 2008 ). Resource selection plasticity and community responses to experimental reduction of a critical resource . Ecology 89 : 971 – 980 . Google Scholar Crossref Search ADS PubMed WorldCat Åkesson , S. , M. Ilieva , J. Karagicheva , E. Rakhimberdiev , B. Tomotani , and B. Helm ( 2017 ). Timing avian long-distance migration: From internal clock mechanisms to global flights . Philosophical Transactions of the Royal Society B: Biological Sciences 372 : 20160252 . Google Scholar Crossref Search ADS WorldCat Alerstam , T. , J. W. Chapman , J. Bäckman , A. D. Smith , H. Karlsson , C. Nilsson , D. R. Reynolds , R. H. G. Klaassen , and J. K. Hill ( 2011 ). Convergent patterns of long-distance nocturnal migration in noctuid moths and passerine birds . Proceedings of the Royal Society B: Biological Sciences 278 : 3074 – 3080 . Google Scholar Crossref Search ADS WorldCat Arnold , T. W . ( 2010 ). Uninformative parameters and model selection using Akaike’s Information Criterion . The Journal of Wildlife Management 74 : 1175 – 1178 . Google Scholar Crossref Search ADS WorldCat Bates , D. , M. Maechler , B. Bolker , and S. Walker ( 2015 ). Fitting linear mixed-effects models using lme4 . Journal of Statistical Software 67 : 1 – 48 . Google Scholar Crossref Search ADS WorldCat Both , C . ( 2010 ). Food availability, mistiming, and climatic change . In Effects of Climate Change on Birds (A. P. Møller, W. Fiedler, and P. Berthold, Editors). Oxford University Press , Oxford, UK. WorldCat Brown , W. P. , and R. R. Roth ( 2002 ). Temporal patterns of fitness and survival in the Wood Thrush . Ecology 83 : 958 – 969 . Google Scholar Crossref Search ADS WorldCat Burnham , K. P. , and D. R. Anderson ( 2002 ). Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach , 2nd edition. Springer-Verlag, New York, NY, USA . Google Preview WorldCat COPAC Burnham , K. P. , and D. R. Anderson ( 2004 ). Multimodel inference: Understanding AIC and BIC in model selection . Sociological Methods & Research 33 : 261 – 304 . Google Scholar Crossref Search ADS WorldCat Burnham , K. P. , D. R. Anderson , and K. P. Huyvaert ( 2011 ). AIC model selection and multimodel inference in behavioral ecology: Some background, observations, and comparisons . Behavioral Ecology and Sociobiology 65 : 23 – 35 . Google Scholar Crossref Search ADS WorldCat Cavalcanti , L. M. P. , L. V. de Paiva , L. F. França , L. M. P. Cavalcanti , L. V. de Paiva , and L. F. França ( 2016 ). Effects of rainfall on bird reproduction in a semi-arid Neotropical region . Zoologia (Curitiba) 33 : e20160018 . Google Scholar Crossref Search ADS WorldCat Cockle , K. L. , and K. Martin ( 2015 ). Temporal dynamics of a commensal network of cavity-nesting vertebrates: increased diversity during an insect outbreak . Ecology 96 : 1093 – 1104 . Google Scholar Crossref Search ADS PubMed WorldCat Cox , A. R. , R. J. Robertson , Á. Z. Lendvai , K. Everitt , and F. Bonier ( 2019 ). Rainy springs linked to poor nestling growth in a declining avian aerial insectivore (Tachycineta bicolor) . Proceedings of the Royal Society B: Biological Sciences 286 : 20190018 . Google Scholar Crossref Search ADS WorldCat Cresswell , W. , and R. McCleery ( 2003 ). How Great Tits maintain synchronization of their hatch date with food supply in response to long-term variability in temperature . Journal of Animal Ecology 72 : 356 – 366 . Google Scholar Crossref Search ADS WorldCat Dawson , R. D . ( 2008 ). Timing of breeding and environmental factors as determinants of reproductive performance of Tree Swallows . Canadian Journal of Zoology 86 : 843 – 850 . Google Scholar Crossref Search ADS WorldCat de Zwaan , D. R. , A. F. Camfield , E. C. MacDonald , and K. Martin ( 2019 ). Variation in offspring development is driven more by weather and maternal condition than predation risk . Functional Ecology 33 : 447 – 456 . Google Scholar Crossref Search ADS WorldCat Drake , A. , and K. Martin ( 2018 ). Local temperatures predict breeding phenology but do not result in breeding synchrony among a community of resident cavity-nesting birds . Scientific Reports 8 : 2756 . Google Scholar Crossref Search ADS PubMed WorldCat Drake , A. , and K. Martin ( 2020 ). Rainfall and nest site competition delay Mountain Bluebird and Tree Swallow breeding but do not impact productivity . The Auk: Ornithological Advances 137 :000–000. https://doi.org/10.5061/dryad.280gb5mmp. WorldCat Drake , A. , M. Martin , and D. J. Green ( 2014a ). Winter habitat use does not influence spring arrival dates or the reproductive success of Yellow Warblers breeding in the arctic . Polar Biology 37 : 181 – 191 . Google Scholar Crossref Search ADS WorldCat Drake , A. , C. A. Rock , S. P. Quinlan , M. Martin , and D. J. Green ( 2014b ). Wind speed during migration influences the survival, timing of breeding, and productivity of a Neotropical migrant, Setophaga petechia . PLoS One 9 : e97152 . Google Scholar Crossref Search ADS WorldCat Dunn , P. , and D. Winkler ( 1999 ). Climate change has affected the breeding date of Tree Swallows throughout North America . Proceedings of the Royal Society B: Biological Sciences 266 : 2487 – 2490 . Google Scholar Crossref Search ADS WorldCat Edworthy , A. B. , M. C. Drever , and K. Martin ( 2011 ). Woodpeckers increase in abundance but maintain fecundity in response to an outbreak of mountain pine bark beetles . Forest Ecology and Management 261 : 203 – 210 . Google Scholar Crossref Search ADS WorldCat Edworthy , A. B. , K. L. Wiebe , and K. Martin ( 2012 ). Survival analysis of a critical resource for cavity-nesting communities: Patterns of tree cavity longevity . Ecological Applications 22 : 1733 – 1742 . Google Scholar Crossref Search ADS PubMed WorldCat Environment and Climate Change Canada ( 2016 ). North American Bird Banding Program . Canadian Bird Banding Office, National Wildlife Research Centre, Carleton University, Ottawa, Canada . Google Preview WorldCat COPAC Evans , W. R. , and M. O’Brien ( 2002 ). Flight Calls of Migratory Birds: Eastern North American Landbirds . Old Bird Incorporated, Ithaca, NY, USA . http://www.oldbird.org. Google Preview WorldCat COPAC Fournier , F. , D. Pelletier , C. Vigneault , B. Goyette , and G. Boivin ( 2005 ). Effect of barometric pressure on flight initiation by Trichogramma pretiosum and Trichogramma evanescens (Hymenoptera: Trichogrammatidae) . Environmental Entomology 34 : 1534 – 1540 . Google Scholar Crossref Search ADS WorldCat Franks , S. E. , J. W. Pearce‐Higgins , S. Atkinson , J. R. Bell , M. S. Botham , T. M. Brereton , R. Harrington , and D. I. Leech ( 2018 ). The sensitivity of breeding songbirds to changes in seasonal timing is linked to population change but cannot be directly attributed to the effects of trophic asynchrony on productivity . Global Change Biology 24 : 957 – 971 . Google Scholar Crossref Search ADS PubMed WorldCat Gómez , C. , N. J. Bayly , D. R. Norris , S. A. Mackenzie , K. V. Rosenberg , P. D. Taylor , K. A. Hobson , and C. D. Cadena ( 2017 ). Fuel loads acquired at a stopover site influence the pace of intercontinental migration in a boreal songbird . Scientific Reports 7 : 3405 . Google Scholar Crossref Search ADS PubMed WorldCat Grüebler , M. U. , M. Morand , and B. Naef-Daenzer ( 2008 ). A predictive model of the density of airborne insects in agricultural environments . Agriculture, Ecosystems & Environment 123 : 75 – 80 . Google Scholar Crossref Search ADS WorldCat Harriman , V. B. , R. D. Dawson , L. E. Bortolotti , and R. G. Clark ( 2017 ). Seasonal patterns in reproductive success of temperate‐breeding birds: Experimental tests of the date and quality hypotheses . Ecology and Evolution 7 : 2122 – 2132 . Google Scholar Crossref Search ADS PubMed WorldCat Hau , M . ( 2001 ). Timing of breeding in variable environments: Tropical birds as model systems . Hormones and Behavior 40 : 281 – 290 . Google Scholar Crossref Search ADS PubMed WorldCat Hau , M. , M. Wikelski , and J. C. Wingfield ( 2000 ). Visual and nutritional food cues fine-tune timing of reproduction in a Neotropical rainforest bird . The Journal of Experimental Zoology 286 : 494 – 504 . Google Scholar Crossref Search ADS PubMed WorldCat Holt , R. F. , and K. Martin ( 1997 ). Landscape modification and patch selection: The demography of two secondary cavity nesters colonizing clearcuts . The Auk 114 : 443 – 455 . Google Scholar Crossref Search ADS WorldCat Huang , A. C. , C. A. Bishop , R. McKibbin , A. Drake , and D. J. Green ( 2017 ). Wind conditions on migration influence the annual survival of a Neotropical migrant, the western Yellow-breasted Chat (Icteria virens auricollis) . BMC Ecology 17 : 29 . Google Scholar Crossref Search ADS PubMed WorldCat Hussell , D. J. T . ( 2003 ). Climate change, spring temperatures, and timing of breeding of Tree Swallows (Tachycineta bicolor) in Southern Ontario . The Auk 120 : 607 – 618 . Google Scholar Crossref Search ADS WorldCat Illera , J. C. , and M. Díaz ( 2006 ). Reproduction in an endemic bird of a semiarid island: A food-mediated process . Journal of Avian Biology 37 : 447 – 456 . Google Scholar Crossref Search ADS WorldCat Irons , R. D. , A. Harding Scurr , A. P. Rose , J. C. Hagelin , T. Blake , and D. F. Doak ( 2017 ). Wind and rain are the primary climate factors driving changing phenology of an aerial insectivore . Proceedings of the Royal Society B: Biological Sciences 284 : 20170412 . Google Scholar Crossref Search ADS WorldCat Johnson , L. S. , and R. D. Dawson ( 2019 ). Mountain Bluebird (Sialia currucoides) version 2.0 . In The Birds of North America ( P. G. Rodewald , Editor). Cornell Lab of Ornithology , Ithaca, NY, USA . https://doi.org/10.2173/bna.moublu.02. Google Preview WorldCat COPAC Kalnay , E. , M. Kanamitsu , R. Kistler , W. Collins , D. Deaven , L. Gandin , M. Iredell , S. Saha , G. White , J. Woollen , et al. . ( 1996 ). The NCEP/NCAR 40-Year Reanalysis Project . Bulletin of the American Meteorological Society 77 : 437 – 471 . Google Scholar Crossref Search ADS WorldCat Kemp , M. U. , E. Emiel van Loon , J. Shamoun-Baranes , and W. Bouten ( 2012 ). RNCEP: Global weather and climate data at your fingertips . Methods in Ecology and Evolution 3 : 65 – 70 . Google Scholar Crossref Search ADS WorldCat Knight , S. M. , D. W. Bradley , R. G. Clark , E. A. Gow , M. Bélisle , L. L. Berzins , T. Blake , E. S. Bridge , L. Burke , R. D. Dawson , et al. . ( 2018 ). Constructing and evaluating a continent-wide migratory songbird network across the annual cycle . Ecological Monographs 88 : 445 – 460 . Google Scholar Crossref Search ADS WorldCat Koch , A. J. , K. Martin , and K. E. H. Aitken ( 2012 ). The relationship between introduced European Starlings and the reproductive activities of Mountain Bluebirds and Tree Swallows in British Columbia, Canada . Ibis 154 : 590 – 600 . Google Scholar Crossref Search ADS WorldCat Koenig , W. D . ( 2003 ). European Starlings and their effect on native cavity-nesting birds . Conservation Biology 17 : 1134 – 1140 . Google Scholar Crossref Search ADS WorldCat Marra , P. P. , C. M. Francis , R. S. Mulvihill , and F. R. Moore ( 2005 ). The influence of climate on the timing and rate of spring bird migration . Oecologia 142 : 307 – 315 . Google Scholar Crossref Search ADS PubMed WorldCat Martin , K. , K. E. H. Aitken , and K. L. Wiebe ( 2004 ). Nest sites and nest webs for cavity-nesting communities in interior British Columbia, Canada: Nest characteristics and niche partitioning . The Condor 106 : 5 – 19 . Google Scholar Crossref Search ADS WorldCat Martin , K. , A. Norris , and M. Drever ( 2006 ). Effects of bark beetle outbreaks on avian biodiversity in the British Columbia interior: Implications for critical habitat management . BC Journal of Ecosystems and Management 7 : 10 – 24 . WorldCat McCarty , J . ( 2009 ). Variation in growth of nestling Tree Swallows across multiple temporal and spatial scales . The Auk 118 : 176 – 190 . Google Scholar Crossref Search ADS WorldCat McCarty , J. P. , and D. W. Winkler ( 1999 ). Relative importance of environmental variables in determining the growth off nestling Tree Swallows Tachycineta bicolor . Ibis 141 : 286 – 296 . Google Scholar Crossref Search ADS WorldCat Mitchell , G. W. , B. K. Woodworth , P. D. Taylor , and D. R. Norris ( 2015 ). Automated telemetry reveals age specific differences in flight duration and speed are driven by wind conditions in a migratory songbird . Movement Ecology 3 : 19 . Google Scholar Crossref Search ADS PubMed WorldCat Nager , R. G. , and A. J. Van Noordwijk ( 1992 ). Energetic limitation in the egg-laying period of Great Tits . Proceedings of the Royal Society of London. Series B: Biological Sciences 249 : 259 – 263 . Google Scholar Crossref Search ADS WorldCat Nilsson , J.-Å . ( 1994 ). Energetic bottle-necks during breeding and the reproductive cost of being too early . Journal of Animal Ecology 63 : 200 – 208 . Google Scholar Crossref Search ADS WorldCat Nilsson , J.-Å. , and E. Svensson ( 1996 ). The cost of reproduction: A new link between current reproductive effort and future reproductive success . Proceedings of the Royal Society of London. Series B: Biological Sciences 263 : 711 – 714 . Google Scholar Crossref Search ADS WorldCat Perrins , C. M . ( 1970 ). The timing of birds’ breeding seasons . Ibis 112 : 242 – 255 . Google Scholar Crossref Search ADS WorldCat van de Pol , M. , L. D. Bailey , N. McLean , L. Rijsdijk , C. R. Lawson , and L. Brouwer ( 2016 ). Identifying the best climatic predictors in ecology and evolution . Methods in Ecology and Evolution 7 : 1246 – 1257 . Google Scholar Crossref Search ADS WorldCat Poulsen , B. O . ( 1996 ). Relationships between frequency of mixed-species flocks, weather and insect activity in a montane cloud forest in Ecuador . Ibis 138 : 466 – 470 . Google Scholar Crossref Search ADS WorldCat QGIS Development Team ( 2015 ). QGIS Geographic Information System . Open Source Geospatial Foundation . http://qgis.osgeo.org. Google Preview WorldCat COPAC R Core Team ( 2019 ). R: A Language and Environment for Statistical Computing . R Foundation for Statistical Computing, Vienna, Austria . http://www.r-project.org/. Google Preview WorldCat COPAC Ridgely , R. S. , T. F. Allnutt , T. Brooks , D. K. McNicol , D. W. Mehlman , B. E. Young , and J. R. Zook ( 2005 ). Digital Distribution Maps of the Birds of the Western Hemisphere. Version 2.1 . NatureServe, Arlington, VA, USA . Google Preview WorldCat COPAC Robles , H. , and K. Martin ( 2013 ). Resource quantity and quality determine the inter-specific associations between ecosystem engineers and resource users in a cavity-nest web . PLoS One 8 : e74694 . Google Scholar Crossref Search ADS PubMed WorldCat Rockwell , S. M. , C. I. Bocetti , and P. P. Marra ( 2012 ). Carry-over effects of winter climate on spring arrival date and reproductive success in an endangered migratory bird, Kirtland’s Warbler (Setophaga kirtlandii) . The Auk 129 : 744 – 752 . Google Scholar Crossref Search ADS WorldCat Rousset , F. , and J.- B. Ferdy ( 2014 ). Testing environmental and genetic effects in the presence of spatial autocorrelation . Ecography 37 : 781 – 790 . Google Scholar Crossref Search ADS WorldCat Schafer , J. L. , and J. W. Graham ( 2002 ). Missing data: Our view of the state of the art . Psychological Methods 7 : 147 – 177 . Google Scholar Crossref Search ADS PubMed WorldCat Schomaker , M. , and C. Heumann ( 2014 ). Model selection and model averaging after multiple imputation . Computational Statistics & Data Analysis 71 : 758 – 770 . Google Scholar Crossref Search ADS WorldCat Siikamäki , P . ( 1998 ). Limitation of reproductive success by food availability and breeding time in Pied Flycatchers . Ecology 79 : 1789 – 1796 . Google Scholar Crossref Search ADS WorldCat Studds , C. E. , and P. P. Marra ( 2007 ). Linking fluctuations in rainfall to nonbreeding season performance in a long-distance migratory bird, Setophaga ruticilla . Climate Research 35 : 115 – 122 . Google Scholar Crossref Search ADS WorldCat Svensson , E. , and J.- A. Nilsson ( 1995 ). Food supply, territory quality, and reproductive timing in the Blue Tit (Parus Caeruleus) . Ecology 76 : 1804 – 1812 . Google Scholar Crossref Search ADS WorldCat Vansteelant , W. M. G. , W. Bouten , R. H. G. Klaassen , B. J. Koks , A. E. Schlaich , J. van Diermen , E. E. van Loon , and J. Shamoun‐Baranes ( 2015 ). Regional and seasonal flight speeds of soaring migrants and the role of weather conditions at hourly and daily scales . Journal of Avian Biology 46 : 25 – 39 . Google Scholar Crossref Search ADS WorldCat Visser , M. E. , L. J. M. Holleman , and P. Gienapp ( 2006 ). Shifts in caterpillar biomass phenology due to climate change and its impact on the breeding biology of an insectivorous bird . Oecologia 147 : 164 – 172 . Google Scholar Crossref Search ADS PubMed WorldCat Wesolowski , T. , and M. Cholewa ( 2009 ). Climate variation and bird breeding seasons in a primeval temperate forest . Climate Research 38 : 199 – 208 . Google Scholar Crossref Search ADS WorldCat Wesolowski , T. , and K. Martin ( 2018 ). Tree holes and hole-nesting birds in European and North American forests. In Ecology and Conservation of Forest Birds ( G. Mikusiński , J.-M. Roberge and R. J. Fuller , Editors). Cambridge University Press , Cambridge, MA, USA . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Wiebe , K. L . ( 2003 ). Delayed timing as a strategy to avoid nest-site competition: Testing a model using data from starlings and flickers . Oikos 100 : 291 – 298 . Google Scholar Crossref Search ADS WorldCat Wiebe , K. L . ( 2016 ). Interspecific competition for nests: Prior ownership trumps resource holding potential for Mountain Bluebird competing with Tree Swallow . The Auk: Ornithological Advances 133 : 512 – 519 . WorldCat Williams , T. D . ( 2012 ). Physiological adaptations for breeding in Birds . Princeton University Press, Princeton, NJ, USA . Google Preview WorldCat COPAC Williams , T. D. , S. Bourgeon , A. Cornell , L. Ferguson , M. Fowler , R. B. Fronstin , and O. P. Love ( 2015 ). Mid-winter temperatures, not spring temperatures, predict breeding phenology in the European Starling Sturnus vulgaris . Open Science 2 : 140301 . WorldCat Winkler , D. W. , and P. E. Allen ( 1996 ). The seasonal decline in Tree Swallow clutch size: Physiological constraint or strategic adjustment? Ecology 77 : 922 – 932 . Winkler , D. W. , K. K. Hallinger , D. R. Ardia , R. J. Robertson , B. J. Stutchbury , and R. R. Cohen ( 2011 ). Tree Swallow (Tachycineta bicolor), version 2.0 . In The Birds of North America ( A. F. Poole , Editor). Cornell Lab of Ornithology , Ithaca, NY, USA . https://doi.org/10.2173/bna.11. Google Preview WorldCat COPAC APPENDIX WINTER AND MIGRATORY REGION WEATHER DATA Within the NCEP Reanalysis 1 climate dataset (Kalnay et al. 1996), surface variables (temperature and precipitation) are projected on a global T62 Gaussian grid composed of 192 equally spaced longitudinal and 94 variably spaced latitudinal points while wind vector data (westerly and southerly) are projected on a 2.5 × 2.5° spatial grid. In QGIS, we added a 0.95° buffer around surface data points and a 1.25° buffer around wind data points and included those points whose buffer physically overlapped with our defined winter and migration ranges (Appendix Figure 4). We then obtained daily mean values for our variables of interest over these points for the years 1994–2011. METHODS AND RESULTS USING FIELD-RECORDED DATES ONLY To support and supplement our missing data approach, we re-ran our candidate models using only clutch initiation dates that were recorded in the field. Sample sizes for each species are smaller (Tree Swallow = 212; Mountain Bluebird = 173). We bootstrapped each of these datasets 100 times to obtain CIs for our standardized beta values. Reported beta values (β) (Appendix Figure 5) are the median, model-averaged, β for each predictor variable, and the reported 85% and 95% CIs for these values are the 0.075–0.925 and the 0.025–0.975 quantiles of the empirical distribution of these bootstrapped values. Individual model results are shown in Appendix Table 5 and Appendix Table 6. While CIs are larger, our results are consistent with the expanded dataset used in our main analysis. For Mountain Bluebird, northern migratory wind and rain are associated with delays (95% CI) and there is no European Starling effect (as in our main analysis). The weak advancement by year is no longer significant for this species but the sign of the relationship is consistent. The only notable change for Mountain Bluebird is the relationship between cavity number and phenology: the sign of this relationship is consistent with the larger dataset but the effect size is notably smaller and crosses zero at the 85% CI (Appendix Figure 5A). Tree Swallows continue to show delays with increasing rainfall (95% CI) and starling abundance (85% CI) and a weak advancement with year (85% CI). Cavity availability is associated with a significant advancement in breeding dates for Tree Swallows within the reduced dataset (95% CI): the sign of this relationship is consistent with the larger dataset, but the effect is not significant in our main analysis (Appendix Figure 5B). Additionally, correspondence between annual mean clutch initiation dates recorded in the field and means that incorporate modeled dates using MI were high, particularly when years were restricted to those where there were greater than 2 known records (2000–2011: Pearson’s r = 0.94, all years with known dates (1998–2011): Pearson’s r = 0.83). APPENDIX FIGURE 4. Open in new tabDownload slide National Center of Environmental Prediction (NCEP) Reanalysis 1 climate data points for our Mountain Bluebird (A) and Tree Swallow (B) nonbreeding periods (for details see Appendix). Winter range surface temperature and precipitation data are denoted by blue triangles; migration surface temperature and rainfall data are denoted by orange triangles. Migration wind speed data are indicated by orange circles. Wintering data are averaged over all points. Migration data are averaged within 10°-latitudinal bands (20–30°N [Tree Swallow only], 30–40°N, and 40–50°N; see Methods). The red circle indicates the location of our monitoring sites. APPENDIX FIGURE 4. Open in new tabDownload slide National Center of Environmental Prediction (NCEP) Reanalysis 1 climate data points for our Mountain Bluebird (A) and Tree Swallow (B) nonbreeding periods (for details see Appendix). Winter range surface temperature and precipitation data are denoted by blue triangles; migration surface temperature and rainfall data are denoted by orange triangles. Migration wind speed data are indicated by orange circles. Wintering data are averaged over all points. Migration data are averaged within 10°-latitudinal bands (20–30°N [Tree Swallow only], 30–40°N, and 40–50°N; see Methods). The red circle indicates the location of our monitoring sites. APPENDIX FIGURE 5. Open in new tabDownload slide Comparison of model-averaged standardized beta (β) values and their confidence intervals (CIs) for (A) Mountain Bluebird and (B) Tree Swallows, between the full dataset used in the main manuscript (red; identical to Figure 2) and a reduced dataset that used field-recorded clutch initiation dates only (blue; Mountain Bluebird, n = 173; Tree Swallow, n = 212). CIs for results that use the full dataset are the empirical distribution of beta values for 2,000 MI-bootstrap runs (see Methods); CIs for results that used field-recorded data only are the empirical distribution of beta values for 100 bootstrap runs (see Appendix). Non-symmetrical CIs are due to model averaging: when models receive little support within a given run, averaging drives beta values towards zero. APPENDIX FIGURE 5. Open in new tabDownload slide Comparison of model-averaged standardized beta (β) values and their confidence intervals (CIs) for (A) Mountain Bluebird and (B) Tree Swallows, between the full dataset used in the main manuscript (red; identical to Figure 2) and a reduced dataset that used field-recorded clutch initiation dates only (blue; Mountain Bluebird, n = 173; Tree Swallow, n = 212). CIs for results that use the full dataset are the empirical distribution of beta values for 2,000 MI-bootstrap runs (see Methods); CIs for results that used field-recorded data only are the empirical distribution of beta values for 100 bootstrap runs (see Appendix). Non-symmetrical CIs are due to model averaging: when models receive little support within a given run, averaging drives beta values towards zero. APPENDIX TABLE 4. Ranked climate window analyses (climwin) output for weather variables that were tested as possible drivers of breeding phenology for (A) Mountain Bluebird (n = 462 nests) and (B) Tree Swallow (n = 572). Weather variables from winter and migratory ranges and the monitoring site were considered. Tested time windows were based on the phenology of each species (Winkler et al. 2011, Johnson and Dawson 2019) but allowed for the winter and breeding period to encompass the spring migratory period. Within climwin, ΔAIC values represent the difference from the null model (random effect of “Year” only) and are therefore negative when models perform better than the null. Bolded variables are within ΔAIC ≤ 2 of the top model and passed Type I error tests. These models were included in the final candidate set. A positive (+) relationship with timing indicates increasing values of a given weather variable are associated with later clutch initiation dates; negative (–) relationships indicate increasing values of a given weather variable are associated with earlier clutch initiation dates. Period/region Weather variable Temporal window ΔAIC (from null) Relationship with timing Tested Best (A) Mountain Bluebird (n = 462 nests) Winter Mean temperature Sept 1–April 30 Nov 18–Dec 17 –11.10 + Migration Rainfall 40–50°N Feb 1–April 30 Feb 4–Mar 24 –10.96 + Migration Westerlies 40–50°N Feb 1–April 30 Feb 16–Feb 26 –9.08 + Breeding Rainfall Feb 1–June 1 April 20–May 15 –8.79 + Migration Westerlies 30–40°N Feb 1–April 30 Feb 7–Mar 25 –7.92 + Breeding Mean temperature Feb 1–June 1 Feb 27–Mar 9 –7.91 – Winter Rainfall Sept 1–April 30 Oct 4–Oct 14 –7.81 – Breeding Minimum temperature Feb 1–June 1 Feb 27–Mar 9 –7.62 – Breeding Maximum temperature Feb 1–June 1 Feb 27–Mar 9 –7.31 – Migration Rainfall 30–40°N Feb 1–April 30 Feb 22–Mar 4 –5.56 – Migration Mean temperature 30–50°N Feb 1–April 30 Feb 15–Feb 25 –5.25 + Migration Southerlies 30–40°N Feb 1–April 30 April 3–April 22 –5.20 – Migration Southerlies 40–50°N Feb 1–April 30 March 3–Mar 13 –3.54 + (B) Tree Swallow (n = 572 nests) Breeding Rainfall Mar 1– June 1 Mar 13–May 22 –15.97 + Winter Rainfall Sept 1–April 15 Dec 11–Dec 28 –5.14 + Breeding Maximum temperature Mar 1– June 1 May 26–May 15 –5.00 – Migration Southerlies 40–50°N Mar 1–April 15 Mar 25–Apr 8 –3.84 – Breeding Mean temperature Mar 1– June 1 Mar 2–Mar 12 –3.68 – Breeding Minimum temperature Mar 1– June 1 Mar 2–Mar 12 –3.49 – Winter Mean temperature Sept 1–April 15 Nov 4–Jan 15 –2.85 – Migration Westerlies 30–40°N Mar 1–April 15 Mar 26–Apr 6 –2.57 – Migration Southerlies 30–40°N Mar 1–April 15 Mar 28–Apr 9 –2.37 – Migration Southerlies 20–30°N Mar 1–April 15 Mar 6–Mar 17 –1.76 + Migration Rainfall 40–50°N Mar 1–April 15 Mar 25–Apr 4 –1.75 – Migration Westerlies 40–50°N Mar 1–April 15 Apr 5–Apr 15 –1.05 + Migration Rainfall 30–40°N Mar 1–April 15 Mar 28–Apr 7 –0.54 – Migration Westerlies 20–30°N Mar 1–April 15 Mar 9–Mar 22 –0.45 – Migration Mean temperature 30–50°N Mar 1–April 15 Mar 1–Mar 11 0.05 – Migration Mean temperature 20–30°N Mar 1–April 15 Apr 2–Apr 12 0.64 – Migration Rainfall 20–30°N Mar 1–April 15 Mar 17–Mar 27 1.11 – Period/region Weather variable Temporal window ΔAIC (from null) Relationship with timing Tested Best (A) Mountain Bluebird (n = 462 nests) Winter Mean temperature Sept 1–April 30 Nov 18–Dec 17 –11.10 + Migration Rainfall 40–50°N Feb 1–April 30 Feb 4–Mar 24 –10.96 + Migration Westerlies 40–50°N Feb 1–April 30 Feb 16–Feb 26 –9.08 + Breeding Rainfall Feb 1–June 1 April 20–May 15 –8.79 + Migration Westerlies 30–40°N Feb 1–April 30 Feb 7–Mar 25 –7.92 + Breeding Mean temperature Feb 1–June 1 Feb 27–Mar 9 –7.91 – Winter Rainfall Sept 1–April 30 Oct 4–Oct 14 –7.81 – Breeding Minimum temperature Feb 1–June 1 Feb 27–Mar 9 –7.62 – Breeding Maximum temperature Feb 1–June 1 Feb 27–Mar 9 –7.31 – Migration Rainfall 30–40°N Feb 1–April 30 Feb 22–Mar 4 –5.56 – Migration Mean temperature 30–50°N Feb 1–April 30 Feb 15–Feb 25 –5.25 + Migration Southerlies 30–40°N Feb 1–April 30 April 3–April 22 –5.20 – Migration Southerlies 40–50°N Feb 1–April 30 March 3–Mar 13 –3.54 + (B) Tree Swallow (n = 572 nests) Breeding Rainfall Mar 1– June 1 Mar 13–May 22 –15.97 + Winter Rainfall Sept 1–April 15 Dec 11–Dec 28 –5.14 + Breeding Maximum temperature Mar 1– June 1 May 26–May 15 –5.00 – Migration Southerlies 40–50°N Mar 1–April 15 Mar 25–Apr 8 –3.84 – Breeding Mean temperature Mar 1– June 1 Mar 2–Mar 12 –3.68 – Breeding Minimum temperature Mar 1– June 1 Mar 2–Mar 12 –3.49 – Winter Mean temperature Sept 1–April 15 Nov 4–Jan 15 –2.85 – Migration Westerlies 30–40°N Mar 1–April 15 Mar 26–Apr 6 –2.57 – Migration Southerlies 30–40°N Mar 1–April 15 Mar 28–Apr 9 –2.37 – Migration Southerlies 20–30°N Mar 1–April 15 Mar 6–Mar 17 –1.76 + Migration Rainfall 40–50°N Mar 1–April 15 Mar 25–Apr 4 –1.75 – Migration Westerlies 40–50°N Mar 1–April 15 Apr 5–Apr 15 –1.05 + Migration Rainfall 30–40°N Mar 1–April 15 Mar 28–Apr 7 –0.54 – Migration Westerlies 20–30°N Mar 1–April 15 Mar 9–Mar 22 –0.45 – Migration Mean temperature 30–50°N Mar 1–April 15 Mar 1–Mar 11 0.05 – Migration Mean temperature 20–30°N Mar 1–April 15 Apr 2–Apr 12 0.64 – Migration Rainfall 20–30°N Mar 1–April 15 Mar 17–Mar 27 1.11 – Open in new tab APPENDIX TABLE 4. Ranked climate window analyses (climwin) output for weather variables that were tested as possible drivers of breeding phenology for (A) Mountain Bluebird (n = 462 nests) and (B) Tree Swallow (n = 572). Weather variables from winter and migratory ranges and the monitoring site were considered. Tested time windows were based on the phenology of each species (Winkler et al. 2011, Johnson and Dawson 2019) but allowed for the winter and breeding period to encompass the spring migratory period. Within climwin, ΔAIC values represent the difference from the null model (random effect of “Year” only) and are therefore negative when models perform better than the null. Bolded variables are within ΔAIC ≤ 2 of the top model and passed Type I error tests. These models were included in the final candidate set. A positive (+) relationship with timing indicates increasing values of a given weather variable are associated with later clutch initiation dates; negative (–) relationships indicate increasing values of a given weather variable are associated with earlier clutch initiation dates. Period/region Weather variable Temporal window ΔAIC (from null) Relationship with timing Tested Best (A) Mountain Bluebird (n = 462 nests) Winter Mean temperature Sept 1–April 30 Nov 18–Dec 17 –11.10 + Migration Rainfall 40–50°N Feb 1–April 30 Feb 4–Mar 24 –10.96 + Migration Westerlies 40–50°N Feb 1–April 30 Feb 16–Feb 26 –9.08 + Breeding Rainfall Feb 1–June 1 April 20–May 15 –8.79 + Migration Westerlies 30–40°N Feb 1–April 30 Feb 7–Mar 25 –7.92 + Breeding Mean temperature Feb 1–June 1 Feb 27–Mar 9 –7.91 – Winter Rainfall Sept 1–April 30 Oct 4–Oct 14 –7.81 – Breeding Minimum temperature Feb 1–June 1 Feb 27–Mar 9 –7.62 – Breeding Maximum temperature Feb 1–June 1 Feb 27–Mar 9 –7.31 – Migration Rainfall 30–40°N Feb 1–April 30 Feb 22–Mar 4 –5.56 – Migration Mean temperature 30–50°N Feb 1–April 30 Feb 15–Feb 25 –5.25 + Migration Southerlies 30–40°N Feb 1–April 30 April 3–April 22 –5.20 – Migration Southerlies 40–50°N Feb 1–April 30 March 3–Mar 13 –3.54 + (B) Tree Swallow (n = 572 nests) Breeding Rainfall Mar 1– June 1 Mar 13–May 22 –15.97 + Winter Rainfall Sept 1–April 15 Dec 11–Dec 28 –5.14 + Breeding Maximum temperature Mar 1– June 1 May 26–May 15 –5.00 – Migration Southerlies 40–50°N Mar 1–April 15 Mar 25–Apr 8 –3.84 – Breeding Mean temperature Mar 1– June 1 Mar 2–Mar 12 –3.68 – Breeding Minimum temperature Mar 1– June 1 Mar 2–Mar 12 –3.49 – Winter Mean temperature Sept 1–April 15 Nov 4–Jan 15 –2.85 – Migration Westerlies 30–40°N Mar 1–April 15 Mar 26–Apr 6 –2.57 – Migration Southerlies 30–40°N Mar 1–April 15 Mar 28–Apr 9 –2.37 – Migration Southerlies 20–30°N Mar 1–April 15 Mar 6–Mar 17 –1.76 + Migration Rainfall 40–50°N Mar 1–April 15 Mar 25–Apr 4 –1.75 – Migration Westerlies 40–50°N Mar 1–April 15 Apr 5–Apr 15 –1.05 + Migration Rainfall 30–40°N Mar 1–April 15 Mar 28–Apr 7 –0.54 – Migration Westerlies 20–30°N Mar 1–April 15 Mar 9–Mar 22 –0.45 – Migration Mean temperature 30–50°N Mar 1–April 15 Mar 1–Mar 11 0.05 – Migration Mean temperature 20–30°N Mar 1–April 15 Apr 2–Apr 12 0.64 – Migration Rainfall 20–30°N Mar 1–April 15 Mar 17–Mar 27 1.11 – Period/region Weather variable Temporal window ΔAIC (from null) Relationship with timing Tested Best (A) Mountain Bluebird (n = 462 nests) Winter Mean temperature Sept 1–April 30 Nov 18–Dec 17 –11.10 + Migration Rainfall 40–50°N Feb 1–April 30 Feb 4–Mar 24 –10.96 + Migration Westerlies 40–50°N Feb 1–April 30 Feb 16–Feb 26 –9.08 + Breeding Rainfall Feb 1–June 1 April 20–May 15 –8.79 + Migration Westerlies 30–40°N Feb 1–April 30 Feb 7–Mar 25 –7.92 + Breeding Mean temperature Feb 1–June 1 Feb 27–Mar 9 –7.91 – Winter Rainfall Sept 1–April 30 Oct 4–Oct 14 –7.81 – Breeding Minimum temperature Feb 1–June 1 Feb 27–Mar 9 –7.62 – Breeding Maximum temperature Feb 1–June 1 Feb 27–Mar 9 –7.31 – Migration Rainfall 30–40°N Feb 1–April 30 Feb 22–Mar 4 –5.56 – Migration Mean temperature 30–50°N Feb 1–April 30 Feb 15–Feb 25 –5.25 + Migration Southerlies 30–40°N Feb 1–April 30 April 3–April 22 –5.20 – Migration Southerlies 40–50°N Feb 1–April 30 March 3–Mar 13 –3.54 + (B) Tree Swallow (n = 572 nests) Breeding Rainfall Mar 1– June 1 Mar 13–May 22 –15.97 + Winter Rainfall Sept 1–April 15 Dec 11–Dec 28 –5.14 + Breeding Maximum temperature Mar 1– June 1 May 26–May 15 –5.00 – Migration Southerlies 40–50°N Mar 1–April 15 Mar 25–Apr 8 –3.84 – Breeding Mean temperature Mar 1– June 1 Mar 2–Mar 12 –3.68 – Breeding Minimum temperature Mar 1– June 1 Mar 2–Mar 12 –3.49 – Winter Mean temperature Sept 1–April 15 Nov 4–Jan 15 –2.85 – Migration Westerlies 30–40°N Mar 1–April 15 Mar 26–Apr 6 –2.57 – Migration Southerlies 30–40°N Mar 1–April 15 Mar 28–Apr 9 –2.37 – Migration Southerlies 20–30°N Mar 1–April 15 Mar 6–Mar 17 –1.76 + Migration Rainfall 40–50°N Mar 1–April 15 Mar 25–Apr 4 –1.75 – Migration Westerlies 40–50°N Mar 1–April 15 Apr 5–Apr 15 –1.05 + Migration Rainfall 30–40°N Mar 1–April 15 Mar 28–Apr 7 –0.54 – Migration Westerlies 20–30°N Mar 1–April 15 Mar 9–Mar 22 –0.45 – Migration Mean temperature 30–50°N Mar 1–April 15 Mar 1–Mar 11 0.05 – Migration Mean temperature 20–30°N Mar 1–April 15 Apr 2–Apr 12 0.64 – Migration Rainfall 20–30°N Mar 1–April 15 Mar 17–Mar 27 1.11 – Open in new tab APPENDIX TABLE 7. Annual productivity data for records where clutch size and nest fate were known (Mountain Bluebird, n = 229 nests; Tree Swallow, n = 177). Nest success is the proportion of nests that fledged at least one young. Year Mountain Bluebird Tree Swallow Clutch size mean (range) Fledge # mean (range) Nest success n Clutch size mean (range) Fledge # mean (range) Nest success n 2000 5.3 (5–6) 2.0 (0–4) 0.50 6 6.0 (4–8) 1.0 (0–6) 0.17 6 2001 5.4 (4–6) 1.4 (0–5) 0.44 9 5.2 (2–7) 2.8 (0–6) 0.67 6 2002 4.8 (2–6) 1.9 (0–5) 0.61 18 4.7 (4–6) 1.7 (0–5) 0.44 9 2003 5.0 (2–6) 3.0 (0–6) 0.72 18 5.4 (3–7) 1.1 (0–5) 0.31 16 2004 5.1 (4–6) 3.0 (0–6) 0.71 17 5.3 (2–7) 0.8 (0–4) 0.32 19 2005 5.1 (4–6) 2.3 (0–5) 0.58 12 5.6 (5–7) 2.8 (0–5) 0.75 8 2006 5.2 (4–6) 1.4 (0–5) 0.31 16 4.5 (3–6) 0.0 0.00 6 2007 5.1 (4–6) 2.3 (0–6) 0.70 23 5.4 (3–7) 2.4 (0–6) 0.75 12 2008 5.5 (5–7) 2.9 (0–5) 0.83 18 5.8 (4–8) 3.7 (0–6) 0.85 13 2009 5.3 (4–7) 3.0 (0–7) 0.62 29 5.3 (3–7) 2.3 (0–6) 0.56 27 2010 5.2 (3–6) 2.8 (0–6) 0.66 29 5.7 (4–7) 2.3 (0–7) 0.47 36 2011 5.4 (4–7) 1.7 (0–5) 0.47 34 5.6 (4–7) 2.5 (0–6) 0.58 19 Year Mountain Bluebird Tree Swallow Clutch size mean (range) Fledge # mean (range) Nest success n Clutch size mean (range) Fledge # mean (range) Nest success n 2000 5.3 (5–6) 2.0 (0–4) 0.50 6 6.0 (4–8) 1.0 (0–6) 0.17 6 2001 5.4 (4–6) 1.4 (0–5) 0.44 9 5.2 (2–7) 2.8 (0–6) 0.67 6 2002 4.8 (2–6) 1.9 (0–5) 0.61 18 4.7 (4–6) 1.7 (0–5) 0.44 9 2003 5.0 (2–6) 3.0 (0–6) 0.72 18 5.4 (3–7) 1.1 (0–5) 0.31 16 2004 5.1 (4–6) 3.0 (0–6) 0.71 17 5.3 (2–7) 0.8 (0–4) 0.32 19 2005 5.1 (4–6) 2.3 (0–5) 0.58 12 5.6 (5–7) 2.8 (0–5) 0.75 8 2006 5.2 (4–6) 1.4 (0–5) 0.31 16 4.5 (3–6) 0.0 0.00 6 2007 5.1 (4–6) 2.3 (0–6) 0.70 23 5.4 (3–7) 2.4 (0–6) 0.75 12 2008 5.5 (5–7) 2.9 (0–5) 0.83 18 5.8 (4–8) 3.7 (0–6) 0.85 13 2009 5.3 (4–7) 3.0 (0–7) 0.62 29 5.3 (3–7) 2.3 (0–6) 0.56 27 2010 5.2 (3–6) 2.8 (0–6) 0.66 29 5.7 (4–7) 2.3 (0–7) 0.47 36 2011 5.4 (4–7) 1.7 (0–5) 0.47 34 5.6 (4–7) 2.5 (0–6) 0.58 19 Open in new tab APPENDIX TABLE 7. Annual productivity data for records where clutch size and nest fate were known (Mountain Bluebird, n = 229 nests; Tree Swallow, n = 177). Nest success is the proportion of nests that fledged at least one young. Year Mountain Bluebird Tree Swallow Clutch size mean (range) Fledge # mean (range) Nest success n Clutch size mean (range) Fledge # mean (range) Nest success n 2000 5.3 (5–6) 2.0 (0–4) 0.50 6 6.0 (4–8) 1.0 (0–6) 0.17 6 2001 5.4 (4–6) 1.4 (0–5) 0.44 9 5.2 (2–7) 2.8 (0–6) 0.67 6 2002 4.8 (2–6) 1.9 (0–5) 0.61 18 4.7 (4–6) 1.7 (0–5) 0.44 9 2003 5.0 (2–6) 3.0 (0–6) 0.72 18 5.4 (3–7) 1.1 (0–5) 0.31 16 2004 5.1 (4–6) 3.0 (0–6) 0.71 17 5.3 (2–7) 0.8 (0–4) 0.32 19 2005 5.1 (4–6) 2.3 (0–5) 0.58 12 5.6 (5–7) 2.8 (0–5) 0.75 8 2006 5.2 (4–6) 1.4 (0–5) 0.31 16 4.5 (3–6) 0.0 0.00 6 2007 5.1 (4–6) 2.3 (0–6) 0.70 23 5.4 (3–7) 2.4 (0–6) 0.75 12 2008 5.5 (5–7) 2.9 (0–5) 0.83 18 5.8 (4–8) 3.7 (0–6) 0.85 13 2009 5.3 (4–7) 3.0 (0–7) 0.62 29 5.3 (3–7) 2.3 (0–6) 0.56 27 2010 5.2 (3–6) 2.8 (0–6) 0.66 29 5.7 (4–7) 2.3 (0–7) 0.47 36 2011 5.4 (4–7) 1.7 (0–5) 0.47 34 5.6 (4–7) 2.5 (0–6) 0.58 19 Year Mountain Bluebird Tree Swallow Clutch size mean (range) Fledge # mean (range) Nest success n Clutch size mean (range) Fledge # mean (range) Nest success n 2000 5.3 (5–6) 2.0 (0–4) 0.50 6 6.0 (4–8) 1.0 (0–6) 0.17 6 2001 5.4 (4–6) 1.4 (0–5) 0.44 9 5.2 (2–7) 2.8 (0–6) 0.67 6 2002 4.8 (2–6) 1.9 (0–5) 0.61 18 4.7 (4–6) 1.7 (0–5) 0.44 9 2003 5.0 (2–6) 3.0 (0–6) 0.72 18 5.4 (3–7) 1.1 (0–5) 0.31 16 2004 5.1 (4–6) 3.0 (0–6) 0.71 17 5.3 (2–7) 0.8 (0–4) 0.32 19 2005 5.1 (4–6) 2.3 (0–5) 0.58 12 5.6 (5–7) 2.8 (0–5) 0.75 8 2006 5.2 (4–6) 1.4 (0–5) 0.31 16 4.5 (3–6) 0.0 0.00 6 2007 5.1 (4–6) 2.3 (0–6) 0.70 23 5.4 (3–7) 2.4 (0–6) 0.75 12 2008 5.5 (5–7) 2.9 (0–5) 0.83 18 5.8 (4–8) 3.7 (0–6) 0.85 13 2009 5.3 (4–7) 3.0 (0–7) 0.62 29 5.3 (3–7) 2.3 (0–6) 0.56 27 2010 5.2 (3–6) 2.8 (0–6) 0.66 29 5.7 (4–7) 2.3 (0–7) 0.47 36 2011 5.4 (4–7) 1.7 (0–5) 0.47 34 5.6 (4–7) 2.5 (0–6) 0.58 19 Open in new tab APPENDIX TABLE 5. Mountain Bluebird final candidate model ranking using only field-recorded clutch initiation dates (n = 173 nests). Values under model parameters are standardized beta values. All values represent the median of 100 bootstrap runs (see Appendix). Null and annual trend models are italicized. There was little difference in model performance (all models ≤ 4 ΔAIC). k = number of parameters and wi = AIC weight. Model Intercept Year (trend) Cavity availability Starling occupancy Rainfall 40–50N Westerly wind 40–50N k loglik ΔAIC a wi MOBL10 143.77 – – – – 2.50 5 –684.99 0.00 0.18 MOBL8 143.67 – – –1.71 2.41 – 6 –684.28 0.71 0.12 MOBL6 143.73 – – – 2.34 – 5 –685.41 0.84 0.12 MOBL1 143.14 – – – – – 4 –686.70 1.29 0.09 MOBL12 143.66 – – –1.23 – 2.47 6 –684.70 1.57 0.08 MOBL11 144.17 – –0.77 – – 2.27 6 –684.75 1.67 0.08 MOBL3 144.83 – –2.08 – – – 5 –686.08 2.17 0.06 MOBL9 144.29 – –0.86 –1.72 2.42 – 7 –684.06 2.46 0.05 MOBL7 144.11 – –0.85 – 2.19 – 6 –685.15 2.47 0.05 MOBL4 143.08 – – –1.32 – – 5 –686.26 2.54 0.05 MOBL2 143.27 0.04 – – – – 5 –686.43 2.87 0.04 MOBL13 144.24 – –0.89 –1.41 – 2.24 7 –684.39 3.11 0.04 MOBL5 144.82 – –2.08 –1.53 – – 6 –685.50 3.16 0.04 Model Intercept Year (trend) Cavity availability Starling occupancy Rainfall 40–50N Westerly wind 40–50N k loglik ΔAIC a wi MOBL10 143.77 – – – – 2.50 5 –684.99 0.00 0.18 MOBL8 143.67 – – –1.71 2.41 – 6 –684.28 0.71 0.12 MOBL6 143.73 – – – 2.34 – 5 –685.41 0.84 0.12 MOBL1 143.14 – – – – – 4 –686.70 1.29 0.09 MOBL12 143.66 – – –1.23 – 2.47 6 –684.70 1.57 0.08 MOBL11 144.17 – –0.77 – – 2.27 6 –684.75 1.67 0.08 MOBL3 144.83 – –2.08 – – – 5 –686.08 2.17 0.06 MOBL9 144.29 – –0.86 –1.72 2.42 – 7 –684.06 2.46 0.05 MOBL7 144.11 – –0.85 – 2.19 – 6 –685.15 2.47 0.05 MOBL4 143.08 – – –1.32 – – 5 –686.26 2.54 0.05 MOBL2 143.27 0.04 – – – – 5 –686.43 2.87 0.04 MOBL13 144.24 – –0.89 –1.41 – 2.24 7 –684.39 3.11 0.04 MOBL5 144.82 – –2.08 –1.53 – – 6 –685.50 3.16 0.04 a Lowest median AIC value was 1,380.35. Open in new tab APPENDIX TABLE 5. Mountain Bluebird final candidate model ranking using only field-recorded clutch initiation dates (n = 173 nests). Values under model parameters are standardized beta values. All values represent the median of 100 bootstrap runs (see Appendix). Null and annual trend models are italicized. There was little difference in model performance (all models ≤ 4 ΔAIC). k = number of parameters and wi = AIC weight. Model Intercept Year (trend) Cavity availability Starling occupancy Rainfall 40–50N Westerly wind 40–50N k loglik ΔAIC a wi MOBL10 143.77 – – – – 2.50 5 –684.99 0.00 0.18 MOBL8 143.67 – – –1.71 2.41 – 6 –684.28 0.71 0.12 MOBL6 143.73 – – – 2.34 – 5 –685.41 0.84 0.12 MOBL1 143.14 – – – – – 4 –686.70 1.29 0.09 MOBL12 143.66 – – –1.23 – 2.47 6 –684.70 1.57 0.08 MOBL11 144.17 – –0.77 – – 2.27 6 –684.75 1.67 0.08 MOBL3 144.83 – –2.08 – – – 5 –686.08 2.17 0.06 MOBL9 144.29 – –0.86 –1.72 2.42 – 7 –684.06 2.46 0.05 MOBL7 144.11 – –0.85 – 2.19 – 6 –685.15 2.47 0.05 MOBL4 143.08 – – –1.32 – – 5 –686.26 2.54 0.05 MOBL2 143.27 0.04 – – – – 5 –686.43 2.87 0.04 MOBL13 144.24 – –0.89 –1.41 – 2.24 7 –684.39 3.11 0.04 MOBL5 144.82 – –2.08 –1.53 – – 6 –685.50 3.16 0.04 Model Intercept Year (trend) Cavity availability Starling occupancy Rainfall 40–50N Westerly wind 40–50N k loglik ΔAIC a wi MOBL10 143.77 – – – – 2.50 5 –684.99 0.00 0.18 MOBL8 143.67 – – –1.71 2.41 – 6 –684.28 0.71 0.12 MOBL6 143.73 – – – 2.34 – 5 –685.41 0.84 0.12 MOBL1 143.14 – – – – – 4 –686.70 1.29 0.09 MOBL12 143.66 – – –1.23 – 2.47 6 –684.70 1.57 0.08 MOBL11 144.17 – –0.77 – – 2.27 6 –684.75 1.67 0.08 MOBL3 144.83 – –2.08 – – – 5 –686.08 2.17 0.06 MOBL9 144.29 – –0.86 –1.72 2.42 – 7 –684.06 2.46 0.05 MOBL7 144.11 – –0.85 – 2.19 – 6 –685.15 2.47 0.05 MOBL4 143.08 – – –1.32 – – 5 –686.26 2.54 0.05 MOBL2 143.27 0.04 – – – – 5 –686.43 2.87 0.04 MOBL13 144.24 – –0.89 –1.41 – 2.24 7 –684.39 3.11 0.04 MOBL5 144.82 – –2.08 –1.53 – – 6 –685.50 3.16 0.04 a Lowest median AIC value was 1,380.35. Open in new tab APPENDIX TABLE 6. Tree Swallow final candidate model ranking using only field-recorded clutch initiation dates (n = 212 nests). Values are the median of 100 bootstrap runs (see Appendix). Null and annual trend models are italicized. Top models (≤ 4 ΔAIC) are in bold and represent 90% of total model support. k = number of parameters and wi = AIC weight. Model Intercept Year (trend) Local rainfall Cavity availability Starling occupancy Bluebird occupancy k loglik ΔAICa wi TRES14 155.06 – 2.00 –1.89 2.59 – 7 –742.26 0.00 0.48 TRES17 155.02 – 2.01 –1.87 2.57 –0.01 8 –742.17 1.98 0.18 TRES8 153.66 – 2.18 – 2.14 – 6 –745.06 3.45 0.09 TRES15 153.89 – 2.22 – 2.15 –0.96 7 –744.01 3.49 0.08 TRES9 155.20 – – –2.19 3.08 – 6 –745.27 3.88 0.07 TRES16 155.24 – – –2.30 3.09 0.20 7 –745.01 5.50 0.03 TRES13 154.61 – 2.20 –1.29 – –0.56 7 –745.68 6.84 0.02 TRES5 153.77 – – – 2.18 – 5 –748.07 7.36 0.01 TRES3 153.61 – 2.42 – – – 5 –748.37 7.97 0.01 TRES2 154.94 –2.62 – – – – 5 –748.56 8.35 0.01 TRES10 153.86 – 2.40 – – –0.66 6 –747.61 8.56 0.01 TRES7 154.61 – 2.29 –1.27 – – 6 –747.67 8.69 0.01 TRES12 154.01 – – – 2.24 –1.12 6 –747.71 8.76 0.01 TRES1 153.71 – – – – – 4 –750.70 10.53 0.00 TRES4 154.69 – – –1.42 – – 5 –749.72 10.67 0.00 TRES6 153.96 – – – – –0.66 5 –749.76 10.75 0.00 TRES11 154.64 – – –1.35 – –0.32 6 –749.21 11.75 0.00 Model Intercept Year (trend) Local rainfall Cavity availability Starling occupancy Bluebird occupancy k loglik ΔAICa wi TRES14 155.06 – 2.00 –1.89 2.59 – 7 –742.26 0.00 0.48 TRES17 155.02 – 2.01 –1.87 2.57 –0.01 8 –742.17 1.98 0.18 TRES8 153.66 – 2.18 – 2.14 – 6 –745.06 3.45 0.09 TRES15 153.89 – 2.22 – 2.15 –0.96 7 –744.01 3.49 0.08 TRES9 155.20 – – –2.19 3.08 – 6 –745.27 3.88 0.07 TRES16 155.24 – – –2.30 3.09 0.20 7 –745.01 5.50 0.03 TRES13 154.61 – 2.20 –1.29 – –0.56 7 –745.68 6.84 0.02 TRES5 153.77 – – – 2.18 – 5 –748.07 7.36 0.01 TRES3 153.61 – 2.42 – – – 5 –748.37 7.97 0.01 TRES2 154.94 –2.62 – – – – 5 –748.56 8.35 0.01 TRES10 153.86 – 2.40 – – –0.66 6 –747.61 8.56 0.01 TRES7 154.61 – 2.29 –1.27 – – 6 –747.67 8.69 0.01 TRES12 154.01 – – – 2.24 –1.12 6 –747.71 8.76 0.01 TRES1 153.71 – – – – – 4 –750.70 10.53 0.00 TRES4 154.69 – – –1.42 – – 5 –749.72 10.67 0.00 TRES6 153.96 – – – – –0.66 5 –749.76 10.75 0.00 TRES11 154.64 – – –1.35 – –0.32 6 –749.21 11.75 0.00 a Lowest median AIC value was 1,499.07. Open in new tab APPENDIX TABLE 6. Tree Swallow final candidate model ranking using only field-recorded clutch initiation dates (n = 212 nests). Values are the median of 100 bootstrap runs (see Appendix). Null and annual trend models are italicized. Top models (≤ 4 ΔAIC) are in bold and represent 90% of total model support. k = number of parameters and wi = AIC weight. Model Intercept Year (trend) Local rainfall Cavity availability Starling occupancy Bluebird occupancy k loglik ΔAICa wi TRES14 155.06 – 2.00 –1.89 2.59 – 7 –742.26 0.00 0.48 TRES17 155.02 – 2.01 –1.87 2.57 –0.01 8 –742.17 1.98 0.18 TRES8 153.66 – 2.18 – 2.14 – 6 –745.06 3.45 0.09 TRES15 153.89 – 2.22 – 2.15 –0.96 7 –744.01 3.49 0.08 TRES9 155.20 – – –2.19 3.08 – 6 –745.27 3.88 0.07 TRES16 155.24 – – –2.30 3.09 0.20 7 –745.01 5.50 0.03 TRES13 154.61 – 2.20 –1.29 – –0.56 7 –745.68 6.84 0.02 TRES5 153.77 – – – 2.18 – 5 –748.07 7.36 0.01 TRES3 153.61 – 2.42 – – – 5 –748.37 7.97 0.01 TRES2 154.94 –2.62 – – – – 5 –748.56 8.35 0.01 TRES10 153.86 – 2.40 – – –0.66 6 –747.61 8.56 0.01 TRES7 154.61 – 2.29 –1.27 – – 6 –747.67 8.69 0.01 TRES12 154.01 – – – 2.24 –1.12 6 –747.71 8.76 0.01 TRES1 153.71 – – – – – 4 –750.70 10.53 0.00 TRES4 154.69 – – –1.42 – – 5 –749.72 10.67 0.00 TRES6 153.96 – – – – –0.66 5 –749.76 10.75 0.00 TRES11 154.64 – – –1.35 – –0.32 6 –749.21 11.75 0.00 Model Intercept Year (trend) Local rainfall Cavity availability Starling occupancy Bluebird occupancy k loglik ΔAICa wi TRES14 155.06 – 2.00 –1.89 2.59 – 7 –742.26 0.00 0.48 TRES17 155.02 – 2.01 –1.87 2.57 –0.01 8 –742.17 1.98 0.18 TRES8 153.66 – 2.18 – 2.14 – 6 –745.06 3.45 0.09 TRES15 153.89 – 2.22 – 2.15 –0.96 7 –744.01 3.49 0.08 TRES9 155.20 – – –2.19 3.08 – 6 –745.27 3.88 0.07 TRES16 155.24 – – –2.30 3.09 0.20 7 –745.01 5.50 0.03 TRES13 154.61 – 2.20 –1.29 – –0.56 7 –745.68 6.84 0.02 TRES5 153.77 – – – 2.18 – 5 –748.07 7.36 0.01 TRES3 153.61 – 2.42 – – – 5 –748.37 7.97 0.01 TRES2 154.94 –2.62 – – – – 5 –748.56 8.35 0.01 TRES10 153.86 – 2.40 – – –0.66 6 –747.61 8.56 0.01 TRES7 154.61 – 2.29 –1.27 – – 6 –747.67 8.69 0.01 TRES12 154.01 – – – 2.24 –1.12 6 –747.71 8.76 0.01 TRES1 153.71 – – – – – 4 –750.70 10.53 0.00 TRES4 154.69 – – –1.42 – – 5 –749.72 10.67 0.00 TRES6 153.96 – – – – –0.66 5 –749.76 10.75 0.00 TRES11 154.64 – – –1.35 – –0.32 6 –749.21 11.75 0.00 a Lowest median AIC value was 1,499.07. Open in new tab Copyright © American Ornithological Society 2020. All rights reserved. For permissions, e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Journal

Auk: Ornithological AdvancesOxford University Press

Published: Apr 5, 2020

There are no references for this article.