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

Learn More →

Path and correlation analysis of perennial ryegrass (Lolium perenne L.) seed yield components

Path and correlation analysis of perennial ryegrass (Lolium perenne L.) seed yield components IntroductionPerennial ryegrass (Lolium perenne L.) is used extensively in temperate pastures as a production source of high‐quality grass for feeding animals or as turf for home lawns, golf courses and recreational areas. However, perennial ryegrass seed crops are biologically inefficient at producing seed (Elgersma, ).Seed production is a key trait for all forage and turf grass species (Boelt & Studer, ). Grassland is estimated to cover approximately 72 million hectares within the European Union (EU) with grass production estimated at a total value of 23 billion Euro (ESA, ). Denmark is the largest grass seed‐producing country within the EU. To ensure economic sustainability of new improved cultivars within today's market, detailed understanding of seed production and the interaction between dynamic yield components is required to improve the efficiencies of the biological system (Studer et al., ; Wilkins & Humphreys, ).Seed yield is highly dependent on various seed yield components starting with tiller production and reproductive induction. Reproductive tillers, inflorescences or seed heads are all terms used to define the same morphological unit: a modified panicle. After a period of cool weather and increasing day lengths, tillers will undergo secondary induction and develop to form a single inflorescence. As perennial ryegrass has the ability to asexually produce daughter tillers, Havstad, Aamlid, Heide, and Junttila () investigated the transmission of the reproductive stimuli from primary to daughter tillers and concluded that daughter tillers did not necessarily have to be directly exposed to inducive conditions. However, problematic to increasing seed yield potential is that these later forming tillers had significantly smaller inflorescence sizes (Havstad et al., ). Tillers formed in late spring are expected to remain vegetative.In Denmark, a large proportion of perennial ryegrass seed crops are sown in spring under a spring cereal crop, typically barley (Hordeum vulgare L.) (Boelt, ). The promotion of tillering in grass seed crops is particularly favoured by an open stand, whereby red light (620–750 nm) can be perceived by the crown (Deregibus, Sanchez, & Casal, ). Low plant density therefore responds positively by increased tiller number per unit area.Inflorescence number is highly influenced by agronomic management, for example nitrogen (N) rate and timings (Boelt, ; Hebblethwaite & Ivins, ), but is a low genetically heritable trait (Studer et al., ). Recent agronomic advances, that is the increased use of plant growth regulators, have not significantly altered inflorescence number when applied within the seed production year (Chastain, Young, Silberstein, & Garbacik, ). Inflorescence number is not believed to be a seed yield limiting factor (Hampton & Hebblethwaite, ; Rolston, McCloy, Trethewey, & Chynoweth, 2010). However, the inflorescence size is dependent on the time of formation with autumn‐initiated tillers being larger than spring‐initiated tillers (Havstad et al., ; Ryle, ; Yamada et al., ).Within inflorescences are spikelets the next seed yield‐determining factor. Inflorescences contain 15–30 spikelets bilaterally arranged in an alternating pattern along the rachis (Thorogood, ). Spikelet number per inflorescence is genetically controlled (Byrne et al., ) and is a highly heritable trait (h2 = .83) (Bugge, ). Spikelets contain 6–10 florets, arranged in an alternate bilateral position around the rachilla. Florets comprise the “flower” unit, which resides within the lemma and palea (Elgersma & Śnieżko, ; Langer, ). Florets per spikelet have been shown to be influenced by management, for example with N (Hampton, ) and by genetics (Bugge, ; Elgersma, ). However, of greater importance to floret number per spikelet is the date of tiller/inflorescence formation with a significantly higher number of florets per spikelet in earlier initiating tillers compared with later (Ryle, ). Further confounding floret number is spikelet position on the inflorescence (i.e., basal or distal positions) with fewer florets per spikelet observed at distal positions (Anslow, ; Ryle, ).Potential seed number per unit area can be determined by the multiplication of these three yield components: inflorescence number per unit area, spikelets per inflorescence and florets per spikelet. However, to produce a satisfactory yield of quality seed, seed set and thousand seed weight (TSW) are important. Factors influencing seed set are crop morphology, including lodging at flowering time (Rolston, Trethewey, Chynoweth, & McCloy, ), weather at time of pollination (Elgersma, Stephenson, & den Nijs, ) and self‐incompatibility (Walter, Studer, & Kölliker, ).The final seed yield component is TSW. Thousand seed weight is influenced by many parameters. Genetics (Smith, McFarlane, Croft, Trigg, & Kearney, ), date of pollination (Anslow, ; Warringa, Struik, Visser, & Kreuzer, ) and soil moisture status (Chastain, King, Garbacik, Young, & Wysocki, ; Chynoweth, Rolston, & McCloy, ) predominately influence TSW at seed harvest.Together, all five seed yield components can determine total theoretical seed yield potential. Total theoretical potential seed yield is defined by the following equation:TTPSY=((n1×n2×n3)/1,000)×((n4/100)×(n5)/1,000))where subscripts denote: TTPSY, Total Theoretical Potential Seed Yield (g seed/m2); n, maximum number of: 1, inflorescence per size category (n/m2); 2, spikelets per inflorescence (n/inflorescence); 3, floret per spikelet (n/spikelet); 4, seed set (%); 5, seed weight (g).Understanding any (seed) yield equation and knowing that these yield components form in a sequential manner, that is spikelets develop after an inflorescence has been established (as described in maize (Zea mays L.) by Fisher and Palmer ()), allows for unidirectional causal path analysis. Path analysis, described by Li (), has been applied in many agricultural crops, including wheat (Triticum aestivum L.) (Blue, Mason, & Sander, ) by associating partial correlations of yield components on yield assuming a bidirectional causation. The bidirectional causation assumes that later developing yield components can influence those earlier forming components. However, Wright () emphasised the importance of causal relationships of sequentially developing yield components; therefore, path analysis of sequentially developing yield components assuming unidirectional causation is an appropriate method of analysis (Rao, Rao, & Prasad, ) and has been applied in maize (Agrama, ). The unidirectional causation assumes that later developing yield components have no influence on earlier forming yield components (Dofing & Knight, ; Fisher & Palmer, ; Wright, ), which is an appropriate method of analysis.We aim to analyse and describe the yield components responsible for harvested seed yields utilising correlation and path analysis and their interaction in field‐based production systems. With this methodology, we aim at increasing our understanding of the interaction between seed yield components which will help explain how stress factors during field production may affect both the potential seed yield and the harvested yield. The following research will enable perennial ryegrass seed producers, seed production researchers and plant breeders to more effectively target higher seed yields.Materials and MethodsIn six commercial seed fields across Denmark, seed yield components were assessed for two growing seasons, 2013 and 2014. Each field was sown with the same intermediate flowering, diploid turf‐type perennial ryegrass cultivar, “Esquire.” Seed yield component determination involved randomly sampling four replicates of .25 m2 at growth stage (GS) 59 (Zadoks, Chang, & Konzak, ). Samples were collected throughout the field ensuring at least 100 m separation between each replicate. All inflorescences per replicate were size graded by length into 80–100, 100–120, 120–140, 140–160 and 160–200 mm categories and counted. Within each size grade, 20 representatively sampled inflorescences were randomly selected and had spikelet number counted, along with 30 spikelets being assessed for floret number.Prior to harvest, GS 92‐93 (Zadoks et al., ), four replicates of .25 m2 were sampled from each field. From these replicated samples, 20 seed heads were representatively sampled per size grade and hand threshed. Seed samples were used to determine seed set (%) and TSW (g). Seed set (%) is defined as being the proportion of seeds that have caryopsis that is greater than one third of the total seed length (as per ISTA ()). Seed samples were air blown with variable air rates in a “Seedburo General Seed Blower HMC67L/C” (Seedburo Equipment Co., Des Plaines, IL, USA). Seed and blowings (inert matter) fractions were confirmed as seed/inert matter with visual microscope inspection and the respective proportion calculated. On clean seed samples, TSW was calculated using a “Wintersteiger SeedCount R‐25+” (Wintersteiger AG, Austria). After field harvest, seed yields were obtained for each field from seed company databases. Field‐harvested seed yields (g seed/m2) were divided by the respective inflorescence count proportion to give seed yield per inflorescence size (g seed per inflorescence size grade) here referred to as seed yield (s) unless otherwise stated. Total theoretical potential seed yields (g seed/m2) were calculated for each inflorescence category utilising the highest yield component observed.Standardised partial regression coefficients (termed path coefficients) and correlation coefficients were determined on data that were log transformed to normalise data before being standardised (i.e., to give mean of 0 and standard deviation of 1). These transformations created linear relationships among seed yield components. Correlation coefficients (rxy) were determined between seed yield components where x and y denote different seed yield components (1–5 including s).Direct path coefficients, Pxs, assessed the direct correlation coefficient of the given seed yield component, x, on seed yield, s. These direct path coefficients between seed yield components and seed yield were arranged in sequential order of development (Fisher & Palmer, ). The sequential ordering of yield components allowed for the sum of direct and indirect causal‐admissible path correlation coefficients, here termed total path correlation coefficients (Txs), to be calculated using the following equations:T1s=P1s+r12P2s+r13P3s+r14P4s+r15P5sT2s=P2s+r23P3s+r24P4s+r25P5sT3s=P3s+r34P4s+r35P5sT4s=P4s+r45P5sT5s=P5swhere subscripts denote: s, seed yield (g inflorescence/size/m2); 1, inflorescence size (mm); 2, spikelets per inflorescence; 3, florets per spikelet; 4, seed set (%); 5, thousand seed weight (g).Here, the total path correlation coefficient between seed set and seed yield for example (T4s = P4s + r45P5s) is defined as the sum of the direct effect of seed set on seed yield (P4s) plus the indirect effect seed set has on seed yield via the correlation between seed set and thousand seed weight (r45) multiplied by the direct effect of pollination rate on seed yield (P5s). Indirect effects were simplified reductions in the models presented.A linear mixed model was utilised to produce path coefficients, Pxs, with random residual terms associated with field and replicate. Correlation coefficients (rxy) utilised Pearson's product moment regression analysis. Data analyses were performed using R version 3.1.3 (R Core Team, ) and the lme4 package (Bates, Maechler, Bolker, & Walker, ).ResultsSeed yield and seed yield componentsAll seed yield components were individually assessed to determine the yearly influence on these components utilising Welch two‐sided t tests. Harvested seed yield and inflorescence number were not significantly different between 2013 and 2014, with an average harvested seed yield of 196 ± 2.77 g seed/m2 (mean ± SE) with inflorescence numbers greater than 2,800 inflorescences/m2 (Table ). All seed yield components (2–5) were significantly different between assessment year (p < .001). The majority of seed yield components were significantly larger in 2014, with only florets per spikelet being greater in 2013 (Table ).Mean and standard errors (mean ± SE) of seed yield components (1–5) and seed yield (s) of six diploid turf perennial ryegrass (Lolium perenne L. cv. Esquire) fields grown in Denmark during the 2013 and 2014 harvest seasonsHarvested seed yield (g seed/m2)Inflorescence number (inflorescence/m2)Inflorescence size (mm)Spikelets per inflorescenceFlorets per spikeletSeed set (%)TSW (g)Seed yield (g seed inflorescence/size/m2)(1)(2)(3)(4)(5)(s)2013195 ± 1.052818 ± 109145 ± .1614.2 ± .286.56 ± .1851.5 ± .871.70 ± .0339.0 ± 3.312014197 ± 5.452913 ± 126148 ± .1617.0 ± .395.14 ± .1961.7 ± 1.091.95 ± .0339.4 ± 3.68Mean196 ± 2.772866 ± 37.6146 ± .1115.6 ± .275.85 ± .1556.6 ± .841.83 ± .0339.2 ± 2.47Significancensns–************nsnsNon‐significant; ***p < .001.Correlation coefficientsCorrelation coefficients were calculated between seed yield components. As these were defined as causal admissible, correlation coefficients were only analysed with seed yield components that formed sequentially. Inflorescence size had the greatest influence on the majority of seed yield components. Inflorescence size was correlated with spikelets per inflorescence and florets per spikelet with a correlation coefficient of .78 (Table ), indicating that larger inflorescences have greater spikelet and floret numbers (i.e., increased yield potential). Inflorescence size did not correlate with seed set (%). Florets per spikelet was negatively correlated with seed set (R = −.278) indicating that an increased number of florets per spikelet may influence harvested seed yields negatively.Correlation coefficients (rxy) between seed yield components (1–5) and seed yield (s) for six diploid turf perennial ryegrass (Lolium perenne L.—cv. Esquire) seed production fields grown throughout Denmark during 2013 and 2014Inflorescence size (mm)Spikelets per inflorescenceFlorets per spikeletSeed set (%)TSW (g)(1)(2)(3)(4)(5)Spikelets per inflorescence(2).782***––––Florets per spikelet(3).781***.462***–––Seed set (%)(4).008ns.206*−.278**––TSW (g)(5).665***.694***.326***.397***–Seed yield (g seed inflorescence/size/m2)(s).714***.665***.596***−.027ns.541***nsNon‐significant; *p < .05; **p < .01; ***p < .001.Path analysisSix different seed production fields grown over two harvest seasons were utilised for the path analysis which comprises direct (Pxs), indirect and total (Txs) path coefficients for each seed yield component with each beginning with the direct path coefficient. Direct path coefficients for inflorescence size (P1s), seed set (P4s) and TSW (P5s) showed no significant effect on seed yield (s) (Table ), while all other seed yield components have a significant direct influence on overall seed yield. The direct path coefficients of spikelet per inflorescence (P2s) and florets per spikelet (P5s) had the largest direct influence on seed yield with path coefficients of .342 and .237, respectively.Direct (Pxs), indirect and total path coefficients (Txs) for the influence that each seed yield component (1–5) has on seed yield of six diploid turf perennial ryegrass (Lolium perenne L.—cv. Esquire) seed production fields grown throughout Denmark during 2013 and 2014Inflorescence size (mm)Spikelets per inflorescenceFlorets per spikeletSeed set (%)TSW (g)(1)(2)(3)(4)(5)Direct effect (Pxs).154ns.342*.237*−.098ns.162nsIndirect effect via:Spikelets per inflorescence(2).268––––Florets per spikelet(3).185.110–––Seed set (%)(4)−.001−.020.027––TSW (g)(5).107.112.053.064–Total effect (Txs).714.544.317−.033.162nsNon‐significant; *p < .05.While inflorescence size had a low direct effect on seed yield (P1s), it had the greatest impact on seed yield when the total path correlation coefficient of .714 was calculated (T1s; Table ). The large effect of inflorescence size on overall seed yield is due to the seed yield‐determining factor being the first seed yield component used in this analysis and with high correlation coefficients (≥.665) with numerous seed yield components (r12, r13 and r15; Table ).Seed set had the lowest total path coefficient correlation (T4s = −.033). Contributing to the low total path correlation coefficient, seed set had a low direct path coefficient (P4s = −.098; Table ) and low correlations (<.206; Table ) from earlier seed yield‐determining factors (r14, r24, r34). As TSW is the last seed yield component determined, TSW had a low total path correlation coefficient (T5s = .162) (Table ).Seed yield per inflorescence sizeTo better understand the large total path coefficient that inflorescence size (T1s; Table ) has on seed yield, seed yield per respective inflorescence categories requires assessment. While there is no year effect on harvested seed yield (Table ), the contribution that each inflorescence size category adds to overall harvested yield varies (Table ). For the 2014 harvest year, inflorescence size 160–200 mm contributed 60% of the seed yield, whereas in 2013 the same size category contributed 42% a reduction of 38.2 g seed/m2 (Table ). The reduction of the largest size category in 2013 harvest year appears to be compensated by the significantly greater seed yield (p < .05; 12.8 g seed/m2) for the 120–140 mm inflorescence size category and marginally significant (p < .1) seed yield increase in the 140–160 mm inflorescence size category compared to 2014.Perennial ryegrass (Lolium perenne L.) seed yield for the respective inflorescence size categories (g seed/m2) contribute to total seed yield (g seed/m2) for the 2013 and 2014 harvest years, including standard errors (mean ± SE) when grown in DenmarkInflorescence size category (mm)80–100 mm (g seed/m2)100–120 mm (g seed/m2)120–140 mm (g seed/m2)140–160 mm (g seed/m2)160–200 mm (g seed/m2)20136.09 ± 1.1922.3 ± 5.7338.1 ± 5.1445.6 ± 4.8182.8 ± 12.120144.36 ± .4711.7 ± .9825.3 ± 1.3134.8 ± 2.22121 ± 13.2Mean5.23 ± .6617.0 ± 3.1031.7 ± 2.9640.2 ± 2.87102 ± 9.76Significancensns*††nsNon‐significant; †p < .1; *p < .05.Theoretical maximum seed yieldFor this given diploid turf cultivar, a theoretical maximum seed yield has been calculated at 974 g seed/m2 (Table ). Theoretical maximum seed yield has been obtained by multiplying maximum seed yield components per inflorescence size. When harvested seed yields are compared to that theoretical maximum yield, harvested yields represent only 20% of theoretical maximum potential seed yield (Table ). Significant yield increases should be observed when increasing the size of smaller inflorescences to that of intermediate or largest size categories thus allowing the potential yield of these categories to be realised.Total theoretical potential seed yield (TTPSY) (g seed/m2) of perennial ryegrass (Lolium perenne L.) for the respective inflorescence size category contribute to the total seed yield (g seed/m2) for the 2013 and 2014 harvest years, including standard errors (mean ± SE)Inflorescence size category (mm)Total seed yield (g seed/m2)80–100 mm (g seed/m2)100–120 mm (g seed/m2)120–140 mm (g seed/m2)140–160 mm (g seed/m2)160–200 mm (g seed/m2)201326.6 ± 4.90106 ± 24.7195 ± 22.6206 ± 19.2421 ± 75.0956 ± 29.3201430.8 ± 5.0366.4 ± 6.13142 ± 12.8175 ± 13.1577 ± 51.7992 ± 17.8Mean28.7 ± 3.5486.4 ± 13.4168 ± 14.1191 ± 12.0499 ± 48.3974 ± 18.3DiscussionSeed yield is a result of seed yield components, agronomic and environmental factors and their interaction. Both spikelets per inflorescence and florets per spikelet have a significant direct effect on harvested seed yield (Table ). Interestingly, florets per spikelet was negatively correlated with seed set (Table ) suggesting that an increased number of florets per spikelet reduced seed set. Inflorescence size has the largest influence on seed yield via indirect causal‐admissible path coefficients resulting in a large total path coefficient (Table ). Harvested seed yields did not significantly differ between harvest years (Table ), but the contribution each size category made towards final seed yield significantly varies from 4.4 to 121 g seed/m2 (Table ).Inflorescence number was not directly included in the path analysis above as it is not believed to be a yield‐limiting factor. Rolston, McCloy, et al. () stated that relatively low inflorescence numbers are required for adequate seed yield: between 1,800 and 2,400 inflorescences/m2. Hampton and Hebblethwaite () stated that seed yields plateaued with inflorescence number between 2,000 and 4,000/m2. While seed yield was not limited by inflorescence number, the findings by Hampton and Hebblethwaite () and Rolston, McCloy, et al. () did not include the inflorescence size, which as shown in the results here is a significant yield‐determining factor. Therefore, only using inflorescence number to indicate yield without considering the relative proportion of inflorescence sizes may lead to misinterpretation of seed yield components. Above these thresholds, seed yield is limited by other seed yield components which have been included in this analysis with particular reference to inflorescence size and its overall influence on the causal‐admissible paths (Table ).The majority of seed production crops are grown under controlled and highly intensive management systems. To maximise seed yield potential, a shift in inflorescence sizes, that is an increase in the mean inflorescence size, should see large advances in seed yield. The shift in overall inflorescence size distribution should result in an increase in harvested seed yields predominately via the total and indirect effect that inflorescence size has on sequentially developing seed yield components. The total path coefficient (T1s = .714; Table ) suggests that this is a feasible approach to increase harvested seed yields in Denmark. Correlation coefficients between inflorescence size and spikelets per inflorescence (r12 = .782; Table ), florets per spikelet (r13 = .781) and TSW (r15 = .665) reveal the importance of inflorescence size on sequentially developing seed yield components and therefore harvested seed yield.Tiller development in perennial ryegrass occurs over two growing seasons from autumn of the establishment year until floral initiation during the spring of the following year, or what can be termed the seed production year. After floral initiation, tillers develop to form a single inflorescence. Due to this extended period of tiller development and owing to the overall importance inflorescence size contributes to seed yield, autumn‐ or very late winter/early spring‐applied N should increase the proportions of the larger inflorescence categories. Brown () showed that crops sown with establishment fertiliser had a greater proportion of larger inflorescence sizes with Boelt () showing an increased number of inflorescences during the seed production year in response to autumn‐applied N. Moreover, numerous authors including Ryle (), Colvill and Marshall () and Havstad et al. () have all concluded that tillers with a larger apex, that is those formed in autumn or promoted via N, offered a greater seed yield potential. However, this N application should be restricted in order not to promote the production of additional vegetative tillers that can significantly compete for limited resources thus limiting harvested seed yields. Additionally, small sized inflorescences (<120 mm; Table ) that contribute little to harvested seed yield (<25 g seed/m2) and account for only 10% of total yield potential (Table ) need to be limited. The proposed management strategy to increase the proportion of larger inflorescence sizes should enable seed producers to effectively increase harvested seed yields.Plant breeding objectives are predominately focussed on vegetative rather than on seed reproductive traits (Rolston, Trethewey, McCloy, & Chynoweth, ; Wilkins & Humphreys, ). From the model results, breeding for seed production potential is possible via spikelet number. Spikelet number per inflorescence has a significant influence on seed yield with a direct path coefficient of .342 (Table ) and total path coefficient of .544. Fortunately for plant breeders, Byrne et al. () concluded that spikelet number is under the influence of genetic control with a high heritability (Bugge, ). Including spikelet number as a breeding objective should significantly increase realised seed production yields.While increasing spikelet number, plant breeders should also aim to reduce the potential number of seed positions available (via florets per spikelet) yet increasing the seed weights at distal floret positions within the spikelet. The importance of TSW to overall seed yield may be overlooked if only the total path coefficient is observed (T5s = .162; Table ); however, TSW still remains an important seed yield‐determining factor when looking at the respective correlation coefficient of .541 (Table ). Rolston et al. () showed that over 10 floret positions available, seed weights varied from >2.5 mg for basal florets to <1.0 mg in distal positions. If utilising 1.5 mg/seed as a minimum for contribution to yield, only seven florets would contribute to seed yield. The reduction in florets per spikelet will have further benefit on seed yield as florets per spikelet was negatively correlated with seed set (r34 = −.278; Table ). Accordingly, breeding for seven or less florets per spikelet, but increasing the average seed weight in distal spikelets, would be a valid method of breeding for increased seed yields, although due diligence is required as florets per spikelet had a significant direct effect on seed yield (Table ).The study was limited by the destructive sample approach taken during the first yield assessment; therefore, seed set and TSW assessments were conducted on an additional set of replications taken prior to seed shattering and harvest. Furthermore, harvested seed yields may differ between and within years, owing to different combine harvester parameters or weather conditions at the time of harvest of each location; however, due diligence was taken to align when samples were taken with further measures in the statistical analysis undertaken to ensure that an accurate representation of the data was observed.ConclusionsThis path and correlation analysis is the first of its kind in field perennial ryegrass seed production and the first to establish seed yield per inflorescence size categories. The results give insight and discussion as to that seed yield component(s) seed producers, seed production researchers and plant breeders can manipulate in order to increase harvested seed yields under field conditions. For seed producers, focus needs to be placed on increasing the proportion of larger inflorescences, which offer a 20‐fold increase in harvested seed yields over the smallest inflorescence size; plant breeders should focus on increasing spikelet number while placing less emphasis on floret number.ReferencesAgrama, H. (1996). Sequential path analysis of grain yield and its components in maize. Plant Breeding, 115, 343–346.Anslow, R. C. (1963). Seed formation in perennial ryegrass. I. Anther exsertion and seed set. Grass and Forage Science, 18, 90–96.Bates, D., Maechler, M., Bolker, B., & Walker, S. (2015). lme4: Linear mixed‐effects models using Eigen and S4. R package version 1.1‐9. Retrieved from https://CRAN.R-project.org/package=lme4Blue, E. N., Mason, S. C., & Sander, D. H. (1990). Influence of planting date, seeding rate, and phosphorus rate on wheat yield. Agronomy Journal, 82, 762–768.Boelt, B. (1997). Undersowing Poa pratensis L., Festuca rubra L., Festuca pratensis Huds., Dactylis glomerata L. and Lolium perenne L. for seed production in five cover crops III. The effect of the amount of autumn applied nitrogen on the seed yield of the undersown grasses. Journal of Applied Seed Production, 15, 41–47.Boelt, B., & Studer, B. (2010). Breeding for grass seed yield. In B. Boller, U. K. Posselt & F. Veronesi (Eds.), Fodder crops and amenity grasses (pp. 161–174). New York, NY: Springer.Brown, K. (1977). Effects of seeding rate and fertiliser on crop development and seed production of ‘Grasslands Tama’ Westerwolds ryegrass. New Zealand Journal of Experimental Agriculture, 5, 215–218.Bugge, G. (1987). Selection for seed yield in Lolium perenne L. Plant Breeding, 98, 149–155.Byrne, S., Guiney, E., Barth, S., Donnison, I. S., Mur, L. A. J., & Milbourne, D. (2009). Identification of coincident QTL for days to heading, spike length and spikelets per spike in Lolium perenne L. Euphytica, 166, 61–70.Chastain, T. G., King, C. M., Garbacik, C. J., Young, W. C., & Wysocki, D. (2015). Irrigation frequency and seasonal timing effects on perennial ryegrass (Lolium perenne L.) seed production. Field Crops Research, 180, 126–134.Chastain, T. G., Young, W. C., Silberstein, T. B., & Garbacik, C. J. (2014). Performance of trinexapac‐ethyl on Lolium perenne seed crops in diverse lodging environments. Field Crops Research, 157, 65–70.Chynoweth, R., Rolston, M. P., & McCloy, B. L. (2012). Irrigation management of perennial ryegrass (Lolium perenne) seed crops. Agronomy New Zealand, 42, 77–85.Colvill, K. E., & Marshall, C. (1984). Tiller dynamics and assimilate partitioning in Lolium perenne with particular reference to flowering. Annals of Applied Biology, 104, 543–557.Deregibus, V. A., Sanchez, R. A., & Casal, J. J. (1983). Effects of light quality on tiller production in Lolium spp. Plant Physiology, 72, 900–902.Dofing, S., & Knight, C. (1992). Alternative model for path analysis of small‐grain yield. Crop Science, 32, 487–489.Elgersma, A. (1990a). Heritability estimates of spaced‐plant traits in three perennial ryegrass (Lolium perenne L.) cultivars. Euphytica, 51, 163–171.Elgersma, A. (1990b). Seed yield related to crop development and to yield components in nine cultivars of perennial ryegrass (Lolium perenne L.). Euphytica, 49, 141–154.Elgersma, A., & Śnieżko, R. (1988). Cytology of seed development related to floret position in perennial ryegrass (Lolium perenne L.). Euphytica, 39, 59–68.Elgersma, A., Stephenson, A. G., & den Nijs, A. P. M. (1989). Effects of genotype and temperature on pollen tube growth in perennial ryegrass (Lolium perenne L.). Sexual Plant Reproduction, 2, 225–230.ESA (2016). European Seed Association. Retrieved from https://www.euroseeds.eu. 13 December 2016.Fisher, K. S., & Palmer, A. F. E. (1983). Maize. In W. H. Smith (Ed.), Symposium on potential productivity of field crops under different environments (pp. 155–180). Los Banos, PA: International Rice Research Institute.Hampton, J. G. (1987). Effect of nitrogen rate and time of application on seed yield in perennial ryegrass cv. Grasslands Nui. New Zealand Journal of Experimental Agriculture, 15, 9–16.Hampton, J. G., & Hebblethwaite, P. D. (1983). Yield components of the perennial ryegrass (Lolium perenne L.) seed crop. Journal of Applied Seed Production, 1, 23–25.Havstad, L. T., Aamlid, T. S., Heide, O. M., & Junttila, O. (2004). Transfer of flower induction stimuli to non‐exposed tillers in a selection of temperate grasses. Acta Agriculturae Scandinavica, Section B‐Soil & Plant Science, 54, 23–30.Hebblethwaite, P. D., & Ivins, J. D. (1978). Nitrogen studies in Lolium perenne grown for seed. II. Timing of nitrogen application. Journal of the British Grassland Society, 33, 159–166.ISTA (2013). International rules for seed testing. Bassersdorf, Switzerland: International Seed Testing Association.Langer, R. H. M. (1990). Pasture plants. Auckland, New Zealand: Oxford University Press.Li, C. (1956). The concept of path coefficient and its impact on population genetics. Biometrics, 12, 190–210.R Core Team (2015). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from http://www.R-project.org/Rao, C., Rao, A., & Prasad, A. (1991). Effect of inadmissible paths in path‐analysis. Indian Journal of Agricultural Sciences, 61, 471–475.Rolston, M., McCloy, B., Trethewey, J., & Chynoweth, R. (2010a). Removing early spring emerged reproductive growing points enhances seed yield of Italian ryegrass. Agronomy New Zealand, 40, 133–139.Rolston, M. P., Trethewey, J. A. K., Chynoweth, R., & McCloy, B. L. (2010b). Trinexapac‐ethyl delays lodging and increases seed yield in perennial ryegrass seed crops. New Zealand Journal of Agricultural Research, 53, 403–406.Rolston, M. P., Trethewey, J. A. K., McCloy, B. L., & Chynoweth, R. (2007). Achieving forage ryegrass seed yields of 3000 kg/ha and limitations to higher yields. In T. S. Aamlid, L. T. Havstad & B. Boelt (Eds.), 6th International herbage seed conference (pp. 100–106). Gjennestad, Norway: Bioforsk.Ryle, G. (1964). The influence of date of origin of the shoot and level of nitrogen on ear size in three perennial grasses. Annals of Applied Biology, 53, 311–323.Smith, K. F., McFarlane, N. M., Croft, V. M., Trigg, P. J., & Kearney, G. A. (2003). The effects of ploidy and seed mass on the emergence and early vigour of perennial ryegrass (Lolium perenne L.) cultivars. Australian Journal of Experimental Agriculture, 43, 481–486.Studer, B., Jensen, L. B., Hentrup, S., Brazauskas, G., Kölliker, R., & Lübberstedt, T. (2008). Genetic characterisation of seed yield and fertility traits in perennial ryegrass (Lolium perenne L.). Theoretical and Applied Genetics, 117, 781–791.Thorogood, D. (2002). Perennial ryegrass (Lolium perenne L.). In M. D. Casler & R. R. Duncan (Eds.), Turfgrass biology, genetics, and breeding (pp. 75–105). Hoboken, NY: John Wiley & Sons, Inc.Walter, A., Studer, B., & Kölliker, R. (2012). Advanced phenotyping offers opportunities for improved breeding of forage and turf species. Annals of Botany, 110(6), 1271–1279.Warringa, J. W., Struik, P. C., Visser, R. D., & Kreuzer, A. D. H. (1998). The pattern of flowering, seed set, seed growth and ripening along the ear of Lolium perenne. Australian Journal of Plant Physiology, 25, 213–223.Wilkins, P. W., & Humphreys, M. O. (2003). Progress in breeding perennial forage grasses for temperate agriculture. The Journal of Agricultural Science, 140, 129–150.Wright, S. (1921). Correlation and causation. Journal of Agricultural Research, 20, 557–585.Yamada, T., Jones, E. S., Cogan, N. O. I., Vecchies, A. C., Nomura, T., Hisano, H., … Forster, J. W. (2004). QTL analysis of morphological, developmental, and winter hardiness‐associated traits in perennial ryegrass. Crop Science, 44, 925–935.Zadoks, J. C., Chang, T. T., & Konzak, C. F. (1974). A decimal code for growth stages of cereals. Weed Research, 14, 415–421. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Agronomy and Crop Science Wiley

Path and correlation analysis of perennial ryegrass (Lolium perenne L.) seed yield components

Download PDF
7 pages

Loading next page...
 
/lp/wiley/path-and-correlation-analysis-of-perennial-ryegrass-lolium-perenne-l-AXyg0eudAL

References (33)

Publisher
Wiley
Copyright
Copyright © 2017 Blackwell Verlag GmbH
ISSN
0931-2250
eISSN
1439-037X
DOI
10.1111/jac.12202
Publisher site
See Article on Publisher Site

Abstract

IntroductionPerennial ryegrass (Lolium perenne L.) is used extensively in temperate pastures as a production source of high‐quality grass for feeding animals or as turf for home lawns, golf courses and recreational areas. However, perennial ryegrass seed crops are biologically inefficient at producing seed (Elgersma, ).Seed production is a key trait for all forage and turf grass species (Boelt & Studer, ). Grassland is estimated to cover approximately 72 million hectares within the European Union (EU) with grass production estimated at a total value of 23 billion Euro (ESA, ). Denmark is the largest grass seed‐producing country within the EU. To ensure economic sustainability of new improved cultivars within today's market, detailed understanding of seed production and the interaction between dynamic yield components is required to improve the efficiencies of the biological system (Studer et al., ; Wilkins & Humphreys, ).Seed yield is highly dependent on various seed yield components starting with tiller production and reproductive induction. Reproductive tillers, inflorescences or seed heads are all terms used to define the same morphological unit: a modified panicle. After a period of cool weather and increasing day lengths, tillers will undergo secondary induction and develop to form a single inflorescence. As perennial ryegrass has the ability to asexually produce daughter tillers, Havstad, Aamlid, Heide, and Junttila () investigated the transmission of the reproductive stimuli from primary to daughter tillers and concluded that daughter tillers did not necessarily have to be directly exposed to inducive conditions. However, problematic to increasing seed yield potential is that these later forming tillers had significantly smaller inflorescence sizes (Havstad et al., ). Tillers formed in late spring are expected to remain vegetative.In Denmark, a large proportion of perennial ryegrass seed crops are sown in spring under a spring cereal crop, typically barley (Hordeum vulgare L.) (Boelt, ). The promotion of tillering in grass seed crops is particularly favoured by an open stand, whereby red light (620–750 nm) can be perceived by the crown (Deregibus, Sanchez, & Casal, ). Low plant density therefore responds positively by increased tiller number per unit area.Inflorescence number is highly influenced by agronomic management, for example nitrogen (N) rate and timings (Boelt, ; Hebblethwaite & Ivins, ), but is a low genetically heritable trait (Studer et al., ). Recent agronomic advances, that is the increased use of plant growth regulators, have not significantly altered inflorescence number when applied within the seed production year (Chastain, Young, Silberstein, & Garbacik, ). Inflorescence number is not believed to be a seed yield limiting factor (Hampton & Hebblethwaite, ; Rolston, McCloy, Trethewey, & Chynoweth, 2010). However, the inflorescence size is dependent on the time of formation with autumn‐initiated tillers being larger than spring‐initiated tillers (Havstad et al., ; Ryle, ; Yamada et al., ).Within inflorescences are spikelets the next seed yield‐determining factor. Inflorescences contain 15–30 spikelets bilaterally arranged in an alternating pattern along the rachis (Thorogood, ). Spikelet number per inflorescence is genetically controlled (Byrne et al., ) and is a highly heritable trait (h2 = .83) (Bugge, ). Spikelets contain 6–10 florets, arranged in an alternate bilateral position around the rachilla. Florets comprise the “flower” unit, which resides within the lemma and palea (Elgersma & Śnieżko, ; Langer, ). Florets per spikelet have been shown to be influenced by management, for example with N (Hampton, ) and by genetics (Bugge, ; Elgersma, ). However, of greater importance to floret number per spikelet is the date of tiller/inflorescence formation with a significantly higher number of florets per spikelet in earlier initiating tillers compared with later (Ryle, ). Further confounding floret number is spikelet position on the inflorescence (i.e., basal or distal positions) with fewer florets per spikelet observed at distal positions (Anslow, ; Ryle, ).Potential seed number per unit area can be determined by the multiplication of these three yield components: inflorescence number per unit area, spikelets per inflorescence and florets per spikelet. However, to produce a satisfactory yield of quality seed, seed set and thousand seed weight (TSW) are important. Factors influencing seed set are crop morphology, including lodging at flowering time (Rolston, Trethewey, Chynoweth, & McCloy, ), weather at time of pollination (Elgersma, Stephenson, & den Nijs, ) and self‐incompatibility (Walter, Studer, & Kölliker, ).The final seed yield component is TSW. Thousand seed weight is influenced by many parameters. Genetics (Smith, McFarlane, Croft, Trigg, & Kearney, ), date of pollination (Anslow, ; Warringa, Struik, Visser, & Kreuzer, ) and soil moisture status (Chastain, King, Garbacik, Young, & Wysocki, ; Chynoweth, Rolston, & McCloy, ) predominately influence TSW at seed harvest.Together, all five seed yield components can determine total theoretical seed yield potential. Total theoretical potential seed yield is defined by the following equation:TTPSY=((n1×n2×n3)/1,000)×((n4/100)×(n5)/1,000))where subscripts denote: TTPSY, Total Theoretical Potential Seed Yield (g seed/m2); n, maximum number of: 1, inflorescence per size category (n/m2); 2, spikelets per inflorescence (n/inflorescence); 3, floret per spikelet (n/spikelet); 4, seed set (%); 5, seed weight (g).Understanding any (seed) yield equation and knowing that these yield components form in a sequential manner, that is spikelets develop after an inflorescence has been established (as described in maize (Zea mays L.) by Fisher and Palmer ()), allows for unidirectional causal path analysis. Path analysis, described by Li (), has been applied in many agricultural crops, including wheat (Triticum aestivum L.) (Blue, Mason, & Sander, ) by associating partial correlations of yield components on yield assuming a bidirectional causation. The bidirectional causation assumes that later developing yield components can influence those earlier forming components. However, Wright () emphasised the importance of causal relationships of sequentially developing yield components; therefore, path analysis of sequentially developing yield components assuming unidirectional causation is an appropriate method of analysis (Rao, Rao, & Prasad, ) and has been applied in maize (Agrama, ). The unidirectional causation assumes that later developing yield components have no influence on earlier forming yield components (Dofing & Knight, ; Fisher & Palmer, ; Wright, ), which is an appropriate method of analysis.We aim to analyse and describe the yield components responsible for harvested seed yields utilising correlation and path analysis and their interaction in field‐based production systems. With this methodology, we aim at increasing our understanding of the interaction between seed yield components which will help explain how stress factors during field production may affect both the potential seed yield and the harvested yield. The following research will enable perennial ryegrass seed producers, seed production researchers and plant breeders to more effectively target higher seed yields.Materials and MethodsIn six commercial seed fields across Denmark, seed yield components were assessed for two growing seasons, 2013 and 2014. Each field was sown with the same intermediate flowering, diploid turf‐type perennial ryegrass cultivar, “Esquire.” Seed yield component determination involved randomly sampling four replicates of .25 m2 at growth stage (GS) 59 (Zadoks, Chang, & Konzak, ). Samples were collected throughout the field ensuring at least 100 m separation between each replicate. All inflorescences per replicate were size graded by length into 80–100, 100–120, 120–140, 140–160 and 160–200 mm categories and counted. Within each size grade, 20 representatively sampled inflorescences were randomly selected and had spikelet number counted, along with 30 spikelets being assessed for floret number.Prior to harvest, GS 92‐93 (Zadoks et al., ), four replicates of .25 m2 were sampled from each field. From these replicated samples, 20 seed heads were representatively sampled per size grade and hand threshed. Seed samples were used to determine seed set (%) and TSW (g). Seed set (%) is defined as being the proportion of seeds that have caryopsis that is greater than one third of the total seed length (as per ISTA ()). Seed samples were air blown with variable air rates in a “Seedburo General Seed Blower HMC67L/C” (Seedburo Equipment Co., Des Plaines, IL, USA). Seed and blowings (inert matter) fractions were confirmed as seed/inert matter with visual microscope inspection and the respective proportion calculated. On clean seed samples, TSW was calculated using a “Wintersteiger SeedCount R‐25+” (Wintersteiger AG, Austria). After field harvest, seed yields were obtained for each field from seed company databases. Field‐harvested seed yields (g seed/m2) were divided by the respective inflorescence count proportion to give seed yield per inflorescence size (g seed per inflorescence size grade) here referred to as seed yield (s) unless otherwise stated. Total theoretical potential seed yields (g seed/m2) were calculated for each inflorescence category utilising the highest yield component observed.Standardised partial regression coefficients (termed path coefficients) and correlation coefficients were determined on data that were log transformed to normalise data before being standardised (i.e., to give mean of 0 and standard deviation of 1). These transformations created linear relationships among seed yield components. Correlation coefficients (rxy) were determined between seed yield components where x and y denote different seed yield components (1–5 including s).Direct path coefficients, Pxs, assessed the direct correlation coefficient of the given seed yield component, x, on seed yield, s. These direct path coefficients between seed yield components and seed yield were arranged in sequential order of development (Fisher & Palmer, ). The sequential ordering of yield components allowed for the sum of direct and indirect causal‐admissible path correlation coefficients, here termed total path correlation coefficients (Txs), to be calculated using the following equations:T1s=P1s+r12P2s+r13P3s+r14P4s+r15P5sT2s=P2s+r23P3s+r24P4s+r25P5sT3s=P3s+r34P4s+r35P5sT4s=P4s+r45P5sT5s=P5swhere subscripts denote: s, seed yield (g inflorescence/size/m2); 1, inflorescence size (mm); 2, spikelets per inflorescence; 3, florets per spikelet; 4, seed set (%); 5, thousand seed weight (g).Here, the total path correlation coefficient between seed set and seed yield for example (T4s = P4s + r45P5s) is defined as the sum of the direct effect of seed set on seed yield (P4s) plus the indirect effect seed set has on seed yield via the correlation between seed set and thousand seed weight (r45) multiplied by the direct effect of pollination rate on seed yield (P5s). Indirect effects were simplified reductions in the models presented.A linear mixed model was utilised to produce path coefficients, Pxs, with random residual terms associated with field and replicate. Correlation coefficients (rxy) utilised Pearson's product moment regression analysis. Data analyses were performed using R version 3.1.3 (R Core Team, ) and the lme4 package (Bates, Maechler, Bolker, & Walker, ).ResultsSeed yield and seed yield componentsAll seed yield components were individually assessed to determine the yearly influence on these components utilising Welch two‐sided t tests. Harvested seed yield and inflorescence number were not significantly different between 2013 and 2014, with an average harvested seed yield of 196 ± 2.77 g seed/m2 (mean ± SE) with inflorescence numbers greater than 2,800 inflorescences/m2 (Table ). All seed yield components (2–5) were significantly different between assessment year (p < .001). The majority of seed yield components were significantly larger in 2014, with only florets per spikelet being greater in 2013 (Table ).Mean and standard errors (mean ± SE) of seed yield components (1–5) and seed yield (s) of six diploid turf perennial ryegrass (Lolium perenne L. cv. Esquire) fields grown in Denmark during the 2013 and 2014 harvest seasonsHarvested seed yield (g seed/m2)Inflorescence number (inflorescence/m2)Inflorescence size (mm)Spikelets per inflorescenceFlorets per spikeletSeed set (%)TSW (g)Seed yield (g seed inflorescence/size/m2)(1)(2)(3)(4)(5)(s)2013195 ± 1.052818 ± 109145 ± .1614.2 ± .286.56 ± .1851.5 ± .871.70 ± .0339.0 ± 3.312014197 ± 5.452913 ± 126148 ± .1617.0 ± .395.14 ± .1961.7 ± 1.091.95 ± .0339.4 ± 3.68Mean196 ± 2.772866 ± 37.6146 ± .1115.6 ± .275.85 ± .1556.6 ± .841.83 ± .0339.2 ± 2.47Significancensns–************nsnsNon‐significant; ***p < .001.Correlation coefficientsCorrelation coefficients were calculated between seed yield components. As these were defined as causal admissible, correlation coefficients were only analysed with seed yield components that formed sequentially. Inflorescence size had the greatest influence on the majority of seed yield components. Inflorescence size was correlated with spikelets per inflorescence and florets per spikelet with a correlation coefficient of .78 (Table ), indicating that larger inflorescences have greater spikelet and floret numbers (i.e., increased yield potential). Inflorescence size did not correlate with seed set (%). Florets per spikelet was negatively correlated with seed set (R = −.278) indicating that an increased number of florets per spikelet may influence harvested seed yields negatively.Correlation coefficients (rxy) between seed yield components (1–5) and seed yield (s) for six diploid turf perennial ryegrass (Lolium perenne L.—cv. Esquire) seed production fields grown throughout Denmark during 2013 and 2014Inflorescence size (mm)Spikelets per inflorescenceFlorets per spikeletSeed set (%)TSW (g)(1)(2)(3)(4)(5)Spikelets per inflorescence(2).782***––––Florets per spikelet(3).781***.462***–––Seed set (%)(4).008ns.206*−.278**––TSW (g)(5).665***.694***.326***.397***–Seed yield (g seed inflorescence/size/m2)(s).714***.665***.596***−.027ns.541***nsNon‐significant; *p < .05; **p < .01; ***p < .001.Path analysisSix different seed production fields grown over two harvest seasons were utilised for the path analysis which comprises direct (Pxs), indirect and total (Txs) path coefficients for each seed yield component with each beginning with the direct path coefficient. Direct path coefficients for inflorescence size (P1s), seed set (P4s) and TSW (P5s) showed no significant effect on seed yield (s) (Table ), while all other seed yield components have a significant direct influence on overall seed yield. The direct path coefficients of spikelet per inflorescence (P2s) and florets per spikelet (P5s) had the largest direct influence on seed yield with path coefficients of .342 and .237, respectively.Direct (Pxs), indirect and total path coefficients (Txs) for the influence that each seed yield component (1–5) has on seed yield of six diploid turf perennial ryegrass (Lolium perenne L.—cv. Esquire) seed production fields grown throughout Denmark during 2013 and 2014Inflorescence size (mm)Spikelets per inflorescenceFlorets per spikeletSeed set (%)TSW (g)(1)(2)(3)(4)(5)Direct effect (Pxs).154ns.342*.237*−.098ns.162nsIndirect effect via:Spikelets per inflorescence(2).268––––Florets per spikelet(3).185.110–––Seed set (%)(4)−.001−.020.027––TSW (g)(5).107.112.053.064–Total effect (Txs).714.544.317−.033.162nsNon‐significant; *p < .05.While inflorescence size had a low direct effect on seed yield (P1s), it had the greatest impact on seed yield when the total path correlation coefficient of .714 was calculated (T1s; Table ). The large effect of inflorescence size on overall seed yield is due to the seed yield‐determining factor being the first seed yield component used in this analysis and with high correlation coefficients (≥.665) with numerous seed yield components (r12, r13 and r15; Table ).Seed set had the lowest total path coefficient correlation (T4s = −.033). Contributing to the low total path correlation coefficient, seed set had a low direct path coefficient (P4s = −.098; Table ) and low correlations (<.206; Table ) from earlier seed yield‐determining factors (r14, r24, r34). As TSW is the last seed yield component determined, TSW had a low total path correlation coefficient (T5s = .162) (Table ).Seed yield per inflorescence sizeTo better understand the large total path coefficient that inflorescence size (T1s; Table ) has on seed yield, seed yield per respective inflorescence categories requires assessment. While there is no year effect on harvested seed yield (Table ), the contribution that each inflorescence size category adds to overall harvested yield varies (Table ). For the 2014 harvest year, inflorescence size 160–200 mm contributed 60% of the seed yield, whereas in 2013 the same size category contributed 42% a reduction of 38.2 g seed/m2 (Table ). The reduction of the largest size category in 2013 harvest year appears to be compensated by the significantly greater seed yield (p < .05; 12.8 g seed/m2) for the 120–140 mm inflorescence size category and marginally significant (p < .1) seed yield increase in the 140–160 mm inflorescence size category compared to 2014.Perennial ryegrass (Lolium perenne L.) seed yield for the respective inflorescence size categories (g seed/m2) contribute to total seed yield (g seed/m2) for the 2013 and 2014 harvest years, including standard errors (mean ± SE) when grown in DenmarkInflorescence size category (mm)80–100 mm (g seed/m2)100–120 mm (g seed/m2)120–140 mm (g seed/m2)140–160 mm (g seed/m2)160–200 mm (g seed/m2)20136.09 ± 1.1922.3 ± 5.7338.1 ± 5.1445.6 ± 4.8182.8 ± 12.120144.36 ± .4711.7 ± .9825.3 ± 1.3134.8 ± 2.22121 ± 13.2Mean5.23 ± .6617.0 ± 3.1031.7 ± 2.9640.2 ± 2.87102 ± 9.76Significancensns*††nsNon‐significant; †p < .1; *p < .05.Theoretical maximum seed yieldFor this given diploid turf cultivar, a theoretical maximum seed yield has been calculated at 974 g seed/m2 (Table ). Theoretical maximum seed yield has been obtained by multiplying maximum seed yield components per inflorescence size. When harvested seed yields are compared to that theoretical maximum yield, harvested yields represent only 20% of theoretical maximum potential seed yield (Table ). Significant yield increases should be observed when increasing the size of smaller inflorescences to that of intermediate or largest size categories thus allowing the potential yield of these categories to be realised.Total theoretical potential seed yield (TTPSY) (g seed/m2) of perennial ryegrass (Lolium perenne L.) for the respective inflorescence size category contribute to the total seed yield (g seed/m2) for the 2013 and 2014 harvest years, including standard errors (mean ± SE)Inflorescence size category (mm)Total seed yield (g seed/m2)80–100 mm (g seed/m2)100–120 mm (g seed/m2)120–140 mm (g seed/m2)140–160 mm (g seed/m2)160–200 mm (g seed/m2)201326.6 ± 4.90106 ± 24.7195 ± 22.6206 ± 19.2421 ± 75.0956 ± 29.3201430.8 ± 5.0366.4 ± 6.13142 ± 12.8175 ± 13.1577 ± 51.7992 ± 17.8Mean28.7 ± 3.5486.4 ± 13.4168 ± 14.1191 ± 12.0499 ± 48.3974 ± 18.3DiscussionSeed yield is a result of seed yield components, agronomic and environmental factors and their interaction. Both spikelets per inflorescence and florets per spikelet have a significant direct effect on harvested seed yield (Table ). Interestingly, florets per spikelet was negatively correlated with seed set (Table ) suggesting that an increased number of florets per spikelet reduced seed set. Inflorescence size has the largest influence on seed yield via indirect causal‐admissible path coefficients resulting in a large total path coefficient (Table ). Harvested seed yields did not significantly differ between harvest years (Table ), but the contribution each size category made towards final seed yield significantly varies from 4.4 to 121 g seed/m2 (Table ).Inflorescence number was not directly included in the path analysis above as it is not believed to be a yield‐limiting factor. Rolston, McCloy, et al. () stated that relatively low inflorescence numbers are required for adequate seed yield: between 1,800 and 2,400 inflorescences/m2. Hampton and Hebblethwaite () stated that seed yields plateaued with inflorescence number between 2,000 and 4,000/m2. While seed yield was not limited by inflorescence number, the findings by Hampton and Hebblethwaite () and Rolston, McCloy, et al. () did not include the inflorescence size, which as shown in the results here is a significant yield‐determining factor. Therefore, only using inflorescence number to indicate yield without considering the relative proportion of inflorescence sizes may lead to misinterpretation of seed yield components. Above these thresholds, seed yield is limited by other seed yield components which have been included in this analysis with particular reference to inflorescence size and its overall influence on the causal‐admissible paths (Table ).The majority of seed production crops are grown under controlled and highly intensive management systems. To maximise seed yield potential, a shift in inflorescence sizes, that is an increase in the mean inflorescence size, should see large advances in seed yield. The shift in overall inflorescence size distribution should result in an increase in harvested seed yields predominately via the total and indirect effect that inflorescence size has on sequentially developing seed yield components. The total path coefficient (T1s = .714; Table ) suggests that this is a feasible approach to increase harvested seed yields in Denmark. Correlation coefficients between inflorescence size and spikelets per inflorescence (r12 = .782; Table ), florets per spikelet (r13 = .781) and TSW (r15 = .665) reveal the importance of inflorescence size on sequentially developing seed yield components and therefore harvested seed yield.Tiller development in perennial ryegrass occurs over two growing seasons from autumn of the establishment year until floral initiation during the spring of the following year, or what can be termed the seed production year. After floral initiation, tillers develop to form a single inflorescence. Due to this extended period of tiller development and owing to the overall importance inflorescence size contributes to seed yield, autumn‐ or very late winter/early spring‐applied N should increase the proportions of the larger inflorescence categories. Brown () showed that crops sown with establishment fertiliser had a greater proportion of larger inflorescence sizes with Boelt () showing an increased number of inflorescences during the seed production year in response to autumn‐applied N. Moreover, numerous authors including Ryle (), Colvill and Marshall () and Havstad et al. () have all concluded that tillers with a larger apex, that is those formed in autumn or promoted via N, offered a greater seed yield potential. However, this N application should be restricted in order not to promote the production of additional vegetative tillers that can significantly compete for limited resources thus limiting harvested seed yields. Additionally, small sized inflorescences (<120 mm; Table ) that contribute little to harvested seed yield (<25 g seed/m2) and account for only 10% of total yield potential (Table ) need to be limited. The proposed management strategy to increase the proportion of larger inflorescence sizes should enable seed producers to effectively increase harvested seed yields.Plant breeding objectives are predominately focussed on vegetative rather than on seed reproductive traits (Rolston, Trethewey, McCloy, & Chynoweth, ; Wilkins & Humphreys, ). From the model results, breeding for seed production potential is possible via spikelet number. Spikelet number per inflorescence has a significant influence on seed yield with a direct path coefficient of .342 (Table ) and total path coefficient of .544. Fortunately for plant breeders, Byrne et al. () concluded that spikelet number is under the influence of genetic control with a high heritability (Bugge, ). Including spikelet number as a breeding objective should significantly increase realised seed production yields.While increasing spikelet number, plant breeders should also aim to reduce the potential number of seed positions available (via florets per spikelet) yet increasing the seed weights at distal floret positions within the spikelet. The importance of TSW to overall seed yield may be overlooked if only the total path coefficient is observed (T5s = .162; Table ); however, TSW still remains an important seed yield‐determining factor when looking at the respective correlation coefficient of .541 (Table ). Rolston et al. () showed that over 10 floret positions available, seed weights varied from >2.5 mg for basal florets to <1.0 mg in distal positions. If utilising 1.5 mg/seed as a minimum for contribution to yield, only seven florets would contribute to seed yield. The reduction in florets per spikelet will have further benefit on seed yield as florets per spikelet was negatively correlated with seed set (r34 = −.278; Table ). Accordingly, breeding for seven or less florets per spikelet, but increasing the average seed weight in distal spikelets, would be a valid method of breeding for increased seed yields, although due diligence is required as florets per spikelet had a significant direct effect on seed yield (Table ).The study was limited by the destructive sample approach taken during the first yield assessment; therefore, seed set and TSW assessments were conducted on an additional set of replications taken prior to seed shattering and harvest. Furthermore, harvested seed yields may differ between and within years, owing to different combine harvester parameters or weather conditions at the time of harvest of each location; however, due diligence was taken to align when samples were taken with further measures in the statistical analysis undertaken to ensure that an accurate representation of the data was observed.ConclusionsThis path and correlation analysis is the first of its kind in field perennial ryegrass seed production and the first to establish seed yield per inflorescence size categories. The results give insight and discussion as to that seed yield component(s) seed producers, seed production researchers and plant breeders can manipulate in order to increase harvested seed yields under field conditions. For seed producers, focus needs to be placed on increasing the proportion of larger inflorescences, which offer a 20‐fold increase in harvested seed yields over the smallest inflorescence size; plant breeders should focus on increasing spikelet number while placing less emphasis on floret number.ReferencesAgrama, H. (1996). Sequential path analysis of grain yield and its components in maize. Plant Breeding, 115, 343–346.Anslow, R. C. (1963). Seed formation in perennial ryegrass. I. Anther exsertion and seed set. Grass and Forage Science, 18, 90–96.Bates, D., Maechler, M., Bolker, B., & Walker, S. (2015). lme4: Linear mixed‐effects models using Eigen and S4. R package version 1.1‐9. Retrieved from https://CRAN.R-project.org/package=lme4Blue, E. N., Mason, S. C., & Sander, D. H. (1990). Influence of planting date, seeding rate, and phosphorus rate on wheat yield. Agronomy Journal, 82, 762–768.Boelt, B. (1997). Undersowing Poa pratensis L., Festuca rubra L., Festuca pratensis Huds., Dactylis glomerata L. and Lolium perenne L. for seed production in five cover crops III. The effect of the amount of autumn applied nitrogen on the seed yield of the undersown grasses. Journal of Applied Seed Production, 15, 41–47.Boelt, B., & Studer, B. (2010). Breeding for grass seed yield. In B. Boller, U. K. Posselt & F. Veronesi (Eds.), Fodder crops and amenity grasses (pp. 161–174). New York, NY: Springer.Brown, K. (1977). Effects of seeding rate and fertiliser on crop development and seed production of ‘Grasslands Tama’ Westerwolds ryegrass. New Zealand Journal of Experimental Agriculture, 5, 215–218.Bugge, G. (1987). Selection for seed yield in Lolium perenne L. Plant Breeding, 98, 149–155.Byrne, S., Guiney, E., Barth, S., Donnison, I. S., Mur, L. A. J., & Milbourne, D. (2009). Identification of coincident QTL for days to heading, spike length and spikelets per spike in Lolium perenne L. Euphytica, 166, 61–70.Chastain, T. G., King, C. M., Garbacik, C. J., Young, W. C., & Wysocki, D. (2015). Irrigation frequency and seasonal timing effects on perennial ryegrass (Lolium perenne L.) seed production. Field Crops Research, 180, 126–134.Chastain, T. G., Young, W. C., Silberstein, T. B., & Garbacik, C. J. (2014). Performance of trinexapac‐ethyl on Lolium perenne seed crops in diverse lodging environments. Field Crops Research, 157, 65–70.Chynoweth, R., Rolston, M. P., & McCloy, B. L. (2012). Irrigation management of perennial ryegrass (Lolium perenne) seed crops. Agronomy New Zealand, 42, 77–85.Colvill, K. E., & Marshall, C. (1984). Tiller dynamics and assimilate partitioning in Lolium perenne with particular reference to flowering. Annals of Applied Biology, 104, 543–557.Deregibus, V. A., Sanchez, R. A., & Casal, J. J. (1983). Effects of light quality on tiller production in Lolium spp. Plant Physiology, 72, 900–902.Dofing, S., & Knight, C. (1992). Alternative model for path analysis of small‐grain yield. Crop Science, 32, 487–489.Elgersma, A. (1990a). Heritability estimates of spaced‐plant traits in three perennial ryegrass (Lolium perenne L.) cultivars. Euphytica, 51, 163–171.Elgersma, A. (1990b). Seed yield related to crop development and to yield components in nine cultivars of perennial ryegrass (Lolium perenne L.). Euphytica, 49, 141–154.Elgersma, A., & Śnieżko, R. (1988). Cytology of seed development related to floret position in perennial ryegrass (Lolium perenne L.). Euphytica, 39, 59–68.Elgersma, A., Stephenson, A. G., & den Nijs, A. P. M. (1989). Effects of genotype and temperature on pollen tube growth in perennial ryegrass (Lolium perenne L.). Sexual Plant Reproduction, 2, 225–230.ESA (2016). European Seed Association. Retrieved from https://www.euroseeds.eu. 13 December 2016.Fisher, K. S., & Palmer, A. F. E. (1983). Maize. In W. H. Smith (Ed.), Symposium on potential productivity of field crops under different environments (pp. 155–180). Los Banos, PA: International Rice Research Institute.Hampton, J. G. (1987). Effect of nitrogen rate and time of application on seed yield in perennial ryegrass cv. Grasslands Nui. New Zealand Journal of Experimental Agriculture, 15, 9–16.Hampton, J. G., & Hebblethwaite, P. D. (1983). Yield components of the perennial ryegrass (Lolium perenne L.) seed crop. Journal of Applied Seed Production, 1, 23–25.Havstad, L. T., Aamlid, T. S., Heide, O. M., & Junttila, O. (2004). Transfer of flower induction stimuli to non‐exposed tillers in a selection of temperate grasses. Acta Agriculturae Scandinavica, Section B‐Soil & Plant Science, 54, 23–30.Hebblethwaite, P. D., & Ivins, J. D. (1978). Nitrogen studies in Lolium perenne grown for seed. II. Timing of nitrogen application. Journal of the British Grassland Society, 33, 159–166.ISTA (2013). International rules for seed testing. Bassersdorf, Switzerland: International Seed Testing Association.Langer, R. H. M. (1990). Pasture plants. Auckland, New Zealand: Oxford University Press.Li, C. (1956). The concept of path coefficient and its impact on population genetics. Biometrics, 12, 190–210.R Core Team (2015). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from http://www.R-project.org/Rao, C., Rao, A., & Prasad, A. (1991). Effect of inadmissible paths in path‐analysis. Indian Journal of Agricultural Sciences, 61, 471–475.Rolston, M., McCloy, B., Trethewey, J., & Chynoweth, R. (2010a). Removing early spring emerged reproductive growing points enhances seed yield of Italian ryegrass. Agronomy New Zealand, 40, 133–139.Rolston, M. P., Trethewey, J. A. K., Chynoweth, R., & McCloy, B. L. (2010b). Trinexapac‐ethyl delays lodging and increases seed yield in perennial ryegrass seed crops. New Zealand Journal of Agricultural Research, 53, 403–406.Rolston, M. P., Trethewey, J. A. K., McCloy, B. L., & Chynoweth, R. (2007). Achieving forage ryegrass seed yields of 3000 kg/ha and limitations to higher yields. In T. S. Aamlid, L. T. Havstad & B. Boelt (Eds.), 6th International herbage seed conference (pp. 100–106). Gjennestad, Norway: Bioforsk.Ryle, G. (1964). The influence of date of origin of the shoot and level of nitrogen on ear size in three perennial grasses. Annals of Applied Biology, 53, 311–323.Smith, K. F., McFarlane, N. M., Croft, V. M., Trigg, P. J., & Kearney, G. A. (2003). The effects of ploidy and seed mass on the emergence and early vigour of perennial ryegrass (Lolium perenne L.) cultivars. Australian Journal of Experimental Agriculture, 43, 481–486.Studer, B., Jensen, L. B., Hentrup, S., Brazauskas, G., Kölliker, R., & Lübberstedt, T. (2008). Genetic characterisation of seed yield and fertility traits in perennial ryegrass (Lolium perenne L.). Theoretical and Applied Genetics, 117, 781–791.Thorogood, D. (2002). Perennial ryegrass (Lolium perenne L.). In M. D. Casler & R. R. Duncan (Eds.), Turfgrass biology, genetics, and breeding (pp. 75–105). Hoboken, NY: John Wiley & Sons, Inc.Walter, A., Studer, B., & Kölliker, R. (2012). Advanced phenotyping offers opportunities for improved breeding of forage and turf species. Annals of Botany, 110(6), 1271–1279.Warringa, J. W., Struik, P. C., Visser, R. D., & Kreuzer, A. D. H. (1998). The pattern of flowering, seed set, seed growth and ripening along the ear of Lolium perenne. Australian Journal of Plant Physiology, 25, 213–223.Wilkins, P. W., & Humphreys, M. O. (2003). Progress in breeding perennial forage grasses for temperate agriculture. The Journal of Agricultural Science, 140, 129–150.Wright, S. (1921). Correlation and causation. Journal of Agricultural Research, 20, 557–585.Yamada, T., Jones, E. S., Cogan, N. O. I., Vecchies, A. C., Nomura, T., Hisano, H., … Forster, J. W. (2004). QTL analysis of morphological, developmental, and winter hardiness‐associated traits in perennial ryegrass. Crop Science, 44, 925–935.Zadoks, J. C., Chang, T. T., & Konzak, C. F. (1974). A decimal code for growth stages of cereals. Weed Research, 14, 415–421.

Journal

Journal of Agronomy and Crop Science – Wiley

Published: Aug 1, 2017

Keywords: ;

There are no references for this article.