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

Learn More →

The moss traits that rule cyanobacterial colonization

The moss traits that rule cyanobacterial colonization Annals of Botany 129: 147–159, 2022 https://doi.org/10.1093/aob/mcab127, available online at www.academic.oup.com/aob 1,2,3, , 2,3 Xin Liu * and Kathrin Rousk CAS Key Laboratory of Mountain Ecological Restoration and Bioresource Utilization & Ecological Restoration and Biodiversity Conservation Key Laboratory of Sichuan Province, Chengdu Institute of Biology, Chinese Academy of Sciences, Chengdu 610041, China, Department of Biology, Terrestrial Ecology Section, University of Copenhagen, Universitetsparken 15, 2100 Copenhagen, Denmark and Center for Permafrost (CENPERM), University of Copenhagen, Øster Voldgade 10, 1350 Copenhagen, Denmark *For correspondence. E-mail liuxin1@cib.ac.cn Received: 17 May 2021 Returned for revision: 5 September 21 Editorial decision: 1 October 2021 Accepted: 8 October 2021 Electronically published: 10 October 2021 • Background and Aims Cyanobacteria associated with mosses represent a main nitrogen (N) source in pristine, high-latitude and -altitude ecosystems due to their ability to fix N . However, despite progress made regarding moss–cyanobacteria associations, the factors driving the large interspecific variation in N fixation activity be- tween moss species remain elusive. The aim of the study was to identify the traits of mosses that determine cyano- bacterial colonization and thus N fixation activity. • Methods Four moss species varying in N fixation activity were used to assess cyanobacterial abundance and activity to correlate it with moss traits (morphological, chemical, water-balance traits) for each species. • Key Results Moss hydration rate was one of the pivotal traits, explaining 56 and 38 % of the variation in fix- N ation and cyanobacterial colonization, respectively, and was linked to morphological traits of the moss species. Higher abundance of cyanobacteria was found on shoots with smaller leaves, and with a high frequency of leaves. High phenol concentration inhibited N fixation but not colonization. These traits driving interspecific variation in cyanobacterial colonization, however, are also affected by the environment, and lead to intraspecific variation. Approximately 24 % of paraphyllia, filamentous appendages on Hylocomium splendens stems, were colonized by cyanobacteria. • Conclusions Our findings show that interspecific variations in moss traits drive differences in cyanobacterial colonization and thus, N fixation activity among moss species. The key traits identified here that control moss- associated N fixation and cyanobacterial colonization could lead to improved predictions of N fixation in dif- 2 2 ferent moss species as a function of their morphology. Key words: Bryophytes, moss colony, Cyanobacteria, nitrogen fixation, water retention, functional trait. N fixation activity between mosses growing in the same habitat INTRODUCTION occur. This lack of understanding of the seemingly random col- Nitrogen (N) fixation performed by moss-associated cyano- onization patterns among moss species hampers efforts to up- bacteria is a main N source in many pristine ecosystems, such scale N fixation across moss species at a larger scale. as boreal forests and subarctic tundra, where N deposition is Cyanobacteria are the dominant N fixer associated with low and plant growth is commonly limited by N availability mosses (Leppänen et  al., 2013), and N fixation activity is (LeBauer and Treseder, 2008; Wang et  al., 2010). Given the commonly correlated with the number of epiphytic cyano- high abundance of mosses in these ecosystems, the association bacterial cells on mosses (Rousk et  al., 2013a). Thus, differ- contributes significantly to ecosystems’ N cycle by accounting ences in the abundance of epiphytic cyanobacteria likely drive for up to 50 % of the total N input (e.g. Rousk and Michelsen, the interspecific variation of moss-associated N fixation rates. 2017). Wide-ranging taxa of mosses have been found to host Cyanobacterial abundance, on the other hand, is affected by dif- cyanobacteria, but N fixation activity of the association ferent factors, such as the capability to move or disperse, and varies greatly among moss species (Rousk et al., 2015; Stuart the environmental factors affecting these processes (Solheim et al., 2021). For instance, N fixation activity in Hylocomium and Zielke, 2002). Among these factors, traits of the moss host, splendens can be more than double the activity found in which provide microsites for epiphytic cyanobacteria (Dalton Pleurozium schreberi, although they are co-dominant in boreal and Chatfield, 1985), might play a key role as moss traits vary forests (Stuart et al., 2021), while the activity in H. splendens is greatly among species (Niinemets and Tobias, 2019), and cer - only one-sixth the activity found in Sphagnum in arctic tundra tain moss species seem to be especially colonized by cyano- (Rousk et  al., 2015). To date, although many factors, such as bacteria (Solheim et  al., 1996). However, to date, it remains nutrient availability (Gundale et al., 2011; Rousk et al., 2017) unknown whether moss traits affect the colonization of cyano- and moisture content (Rousk et al., 2015, 2018) affect N fix- bacteria and, if so, which suite of moss traits facilitates cyano- ation in mosses, we do not know why these large variations in bacterial colonization that leads to differences between moss © The Author(s) 2021. Published by Oxford University Press on behalf of the Annals of Botany Company. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/ by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 148 Liu & Rousk — Moss traits that rule cyanobacterial colonization hosts. The paucity of data on the effects of species-specific concentration of host mosses might lead to variation in cyano- moss traits on cyanobacterial colonization limits our under - bacterial colonization and activity. Similarly, pH is an influ- standing of the relationship between moss and cyanobacteria, ential factor structuring microbial communities (Rousk et  al., and thereby of the functioning of moss-dominated ecosystems 2009) and affecting N fixation rates in mosses (Alvarenga and such as boreal forests. Rousk, 2021). To date, it is unknown if differences in the chem- Studies have repeatedly found that moisture promotes N ical environment among moss species lead to variations in the fixation in moss–cyanobacteria associations (Smith, 1984; number of epiphytic cyanobacteria. Jackson et al., 2011; Gundale et al., 2012; Rousk et al., 2018). The purpose of the study reported herein was to identify the This could be the result of either a direct positive effect of water moss traits that affect cyanobacterial colonization, and thus N availability on cyanobacterial activity, or indirectly, promoting fixation activity, using four different moss species that have the colonization of cyanobacteria, or a combination of both. To been shown to vary in N fixation. To accomplish this, we meas- form an effective N-fixing association with plants, vegetative ured acetylene reduction rate as a measure of N fixation ac- 2 2 cyanobacterial cells differentiate into motile and short-lived tivity, and assessed cyanobacterial colonization and abundance hormogonia with a gliding activity for 48–72  h (Meeks and at shoot and colony (group of shoots) level and linked this to Elhai, 2002). The capacity to move would affect the number of water balance traits (maximum water content, water absorption cyanobacteria colonizing mosses (Santi et  al., 2013). A  moist rate and water loss rate of moss colonies), chemical traits (pH, surface or liquid is required for hormogonia to glide or swim total phenols), colony structural traits (frequency of shoots and (Brahamsha and Bhaya, 2014), and it is likely that passive height) and morphological traits (shoot length, frequency of water flow carries cyanobacteria from lower parts of a moss leaves, leaf area, etc., Table 1) at shoot as well as at leaf level of shoot to its upper segments (Broady, 1979). Studies concerning the four moss species collected in the subarctic region. We hy- the mobility of hormogonia are mainly focused on structures pothesized that (1) N fixation activity is correlated with cyano- of cyanobacteria (e.g. Adams and Duggan, 2008; Wilde and bacterial colonization in all investigated moss species, (2) the Mullineaux, 2015), while the characteristics of the moss host moss species that has the highest water absorption rate hosts the may affect movement capacity and hence the abundance of epi-most cyanobacteria, (3) moss traits (e.g. frequency of shoots) phytic cyanobacteria. If moss colonies absorb water at different that facilitate water absorption increase cyanobacterial colon- rates, the rate of water flow may result in differences in cyano- ization, (4) moss shoots with higher frequency of leaves host bacterial abundance between moss species and thereby dif-fer more cyanobacterial colonizers, and (5) high phenol concentra- ences in N fixation. tion and low pH inhibit cyanobacterial colonization. Water retention capacity and water absorption rate of mosses may be important for hosting cyanobacteria, and are affected by colony structure (e.g. density and height) as well as by traits MATERIALS AND METHODS of individual shoots (e.g. leaf frequency). Indeed, moss colony density and height have been reported to correlate positively Moss sampling with water retention capacity (Proctor, 1982; Elumeeva et  al., 2011), shoot morphology controls dehydration rate of mosses Moss samples were collected at two sites in Northern Sweden. (Cruz de Carvalho et  al., 2019), and leaf width is positively Aulacomnium turgidum (Wahlenb.) Schwägr., Hylocomium related to water retention in Sphagnum mosses (Bengtsson splendens (Hedw.) Schimp. and Tomentypnum nitens (Hedw.) et  al., 2020). However, to our knowledge, there is no empir - Loeske (Supplementary Data Figure S1) were collected in June ical evidence that these hydrology-related traits affect cyano2019 in a subarctic dry heath close to the - Abisko Scientific bacterial colonization of mosses. Moreover, moss traits might Research Station (68°19′02″N, 18°50′04″E). The mean an- affect colonization rates directly. Mosses are assumed to host nual air temperature in Abisko is 0.2  °C, and the mean annual and protect cyanobacteria between or on their leaves (Dalton precipitation is 337  mm (30-year mean 1986–2015, Abisko and Chatfield, 1985), which vary 54-fold in size and 28-fold Scientific Research Station 2016). The site was dominated in frequency among species (Niinemets and Tobias, 2019), by mosses and Vaccinium uliginosum, Andromeda polifolia and cyanobacterial filaments are found between moss stems and Rhododendron lapponicum (see Rousk and Michelsen, and leaves (Solheim and Zielke, 2002). Thus, moss species 2017). Pleurozium schreberi (Brid.) Mitt., which did not occur with larger leaves and higher leaf frequency should host more at this site, was collected in August 2019 in a boreal forest cyanobacteria due to higher availability of colonization sites. near Arvidsjaur (64°58′49″N 19°33′59″E). This extended the Yet again, the relationships between colony- (e.g. frequency of study to another species that has been shown previously to shoots and colony height), shoot- (e.g. frequency of leaves and differ in N fixation rates compared to H.  splendens despite shoot length) and leaf-level (e.g. leaf width and area) traits and their similar morphology and habitat preferences (Jean et  al., cyanobacteria colonization remain elusive. 2020). Mean annual temperature and precipitation are approxi- Another possible mechanism that regulates cyanobacterial mately 1  °C and 570 mm respectively. The forest was domin- colonization is the production of chemical compounds by the ated by Picea abies, V. vitis-idaea, V. myrtillus and Empetrum host. Mosses are known to produce and accumulate inhibitory hermaphroditum (see Rousk et al., 2013b). From this site, sam- compounds like phenols (Erickson and Miksche, 1974). These ples of H. splendens were also collected in order to identify dif- compounds can lead to inhibition of bacterial growth (Rousk ferences in the measured variables across ecosystems, as well et  al., 2013a) and contributes to the low decomposability as to ascertain if differences in N fixation is a species or an of moss litter (Lang et  al., 2009). Differences in phenol ecosystem effect. Liu & Rousk — Moss traits that rule cyanobacterial colonization 149 Table 1. Evaluated variables and their symbols, definitions and units Symbol Description Unit −1 −1 AR Acetylene reduction rate nmol g DW h F Frequency of cyanobacteria colonized leaves % L,Colonized −1 CC Cyanobacteria count on stem leaves Cells leaf L,St −1 CC Cyanobacteria count on branch leaves Cells leaf L,Br −2 CD Cyanobacteria density on stem leaves Cells mm leaf L,St −2 CD Cyanobacteria density on branch leaves Cells mm leaf L,Br WC Maximum water content % Max W Time needed for moss colony to absorb water from air dried status to 50 % of min Absorb maximum water content. Larger number means lower hydration rate. W Time needed for moss colony to lose water from 100 to 50 % of maximum h Lose water content. Larger number means lower desiccation rate. H Height of moss colony mm Colony −2 F Frequency of Shoot Shoots cm Sh pH pH - −1 Phenol Total phenol concentration mg GAE g [C] Carbon concentration % [N] Nitrogen concentration % L Length of shoot mm Sh −1 F Frequency of leaves – number of (stem and branch) leaves per unit shoot length Leaves mm shoot W Basal width of stem leaf mm L,Base,St W Maximum width of stem leaf mm L,Max,St L Length of stem leaf mm L,St A Area of stem leaf mm L,St W Basal width of branch leaf mm L,Base,Br W Maximum width of branch leaf mm L,Max,Br L Length of branch leaf mm L,Br A Area of branch leaf mm L,Br Three separate monospecific moss colonies, with at least and 8 mm for T. nitens, which represents 1 year’s length growth 5 m distance from each other, were selected for each species (Bauer et  al., 2007), and other sections. Brown segments that at the subarctic and boreal sites. Uniform moss colonies of were partly decomposed were not included in the measure- 15 cm × 15 cm were sampled and carefully transported to the ments. Fully hydrated leaves from these chosen branches and laboratory in Copenhagen. The samples were assessed for N stem sections (351 branches/stem sections in total: 18 for fixation activity, water balance (maximum water content, water A.  turgidum, 72 for P.  schreberi, 72 for T.  nitens and 189 for absorption and -loss rate), chemical (pH and total phenols), H. splendens) were measured. shoot and leaf morphological traits, and cyanobacterial counts. To be able to link morphological traits to cyanobacterial col- The H. splendens samples from the boreal forest site were as-onization, we destructively harvested all the leaves from 1- to sessed for N fixation activity, water balance, and chemical 3-mm lengths of stem segments or branches through scraping traits. to enable counting. The number of scraped-off leaves was counted using an Olympus SZX16 stereo microscope. In total, 4857 leaves (328 for A. turgidum, 1141 for P. schreberi, 1469 for T. nitens and 1919 for H. splendens) were scraped off. The Morphological traits and cyanobacterial colonization number of cyanobacteria-colonized leaves among scraped- off leaves was counted using an Olympus BX61 ultraviolet- Three shoots from each sample were randomly selected to fluorescence microscope with a green filter. The frequency of measure shoot and leaf morphological traits (n  =  9 per spe- colonized leaves for each stem segment or branch was calculated cies). We measured the length of each shoot after submer - according to the number of colonized leaves and the number of ging in double distilled (dd) H O to ensure full hydration. For all scraped-off leaves. Digital images of five randomly selected H.  splendens, shoots were divided into three segments ac- leaves were taken with a USB2.0 CMOS Camera (ToupTek, cording to innate growth markers. These segments were: top- Hangzhou, China) attached to the stereo microscope. From most current year segment which are younger than one-year these images, the maximum width, basal width, length and area (ca. 11 mm), the segment younger than 2 years but older than of individual leaves (Table 1) were measured with ImageJ 2.35 1  year, and the segment older than 2  years. The stem and six (Wayne Rasbund, National Institutes of Health, Bethesda, MD, random branches were chosen from each segment for leaf USA). The cyanobacterial cells on the five selected leaves per measurements and cyanobacterial counting. For P.  schreberi stem segment or branch were counted using the Olympus BX61 and T.  nitens, six random branches were chosen from each ultraviolet-fluorescence microscope. The leaf size and number shoot. For A.  turgidum, which does not have branches, only of cyanobacterial cells for 1755 leaves (90 for A. turgidum, 360 stems leaves were measured. Stems of A. turgidum P, . schreberi for P. schreberi, 360 for T. nitens and 945 for H. splendens) were and T.  nitens were divided into two sections: the top sections, measured. Microscopic counting, rather than using proxies of approximately 9 mm for A.  turgidum, 13 mm for P.  schreberi 150 Liu & Rousk — Moss traits that rule cyanobacterial colonization cyanobacterial quantity (Renaudin et  al., 2021), enabled us to assay (ARA). For this, 20-mL glass vials containing ten fully link leaf traits with cyanobacterial colonization. hydrated moss shoots (n = 3 for each species) were sealed and 10 % of the headspace was replaced with acetylene. The moss samples were incubated for 10 h at 12 °C, 10 h at 6 °C, then 4 h at 12 °C. Ethylene generated in the headspace by the cyano- Water balance bacterial nitrogenase enzyme was measured by gas chroma- Water absorption. To measure water absorption rates of tography with a flame ionization detector using an automatic moss colonies, the bottom of transparent polypropylene cups, headspace sampler (Agilent, 8890 GC System, Agilent, Santa 3.7 cm in diameter, were cut off and replaced with cotton mesh Clara, USA). The fresh moss shoots were oven-dried at 65 °C (Supplementary Data Fig. S2). We placed a round moss colonyfor 48  , h and ground into fine powder, which was subsequently which fits the area of the cup, into the cup. Moss patches had used for total carbon (TC), total nitrogen (TN) and phenol con- similar height to those in the field, and included both green and centration measurement. basal parts, while decomposed parts were excluded. Cups filled with moss colonies were air-dried in a venti- lated growth chamber for one week. The air temperature in the Nutrient concentrations, total phenols and pH measurements chamber was 6 °C in the night (1800–0600 h) and 12 °C during daytime (0600–1800 h). The relative humidity in the chamber Carbon and N concentrations in moss tissue were assessed varied between 51.2 and 97.7 %, with an average of 76.6 %. The with a Vario Macro Cube Elemental Analyzer (Elementar, cups were then weighed and placed into a plastic box, which Germany). Total phenols were measured in moss tissue that was filled with ddH O to 1 cm depth. The cups were weighed had been ground into fine powder and then suspended in 10 mL at time intervals of 2 min for the initial 20 min, then weighed at ethanol. Samples were shaken for 120 min and then centrifuged intervals of 5 min for the rest of the first hour, and then weighed at 3600 g for 10 min. The supernatant was analysed for phenols every hour until their mass became nearly constant. We added using the Folin–Ciocalteu reagent. The absorbance was meas- ddH O to the box during the experiment to maintain the same ured at 725 nm using a spectrophotometer. The pH of mosses water level throughout the measurements. was measured in 3 g fresh moss tissue that was submerged in Water loss. The same cups filled with moss colonies were 15 mL ddH O and then shaken for 60 min. The pH of the ex- used to determine the water loss rate of the moss colonies. tracts was determined with a pH electrode. Water (ddH O) was added to the box to immerse the moss col- onies and left for 12 h, allowing the colonies to reach full hy- dration. The cups were then placed on a tilted plastic surface Statistical analyses for 5 min to let the surplus water run down. The samples were weighed and kept in the same chamber as above and allowed to We first performed principal component analyses (PCAs) dry. The cups were weighed every hour during the initial 12 h using acetylene reduction rate and cyanobacterial colonization and every 6 h for a maximum of 120 h until their mass became variables, as well as the water balance, chemical and morpho- nearly constant. Then we counted the number of moss shoots in logical traits to obtain an overview of the multidimensional each cup, oven-dried the mosses at 65 °C for 48 h and recorded cyanobacterial colonization and moss traits spectrum of vari- their dry weight. ation. Because of close relationships of cyanobacterial count, Calculations. Maximum water content of moss colonies were density and morphological traits of branch leaves to those of calculated and expressed as percentage of dry weight (Table 1). stem leaves, only variables and traits of stem leaves were in- Exponential functions weight =  K/(1 + exp(a + b × time)) and cluded in the analyses. The relationships between acetylene weight = a × exp(−b × time) were fitted to the moss weight and reduction rate and frequency of colonized leaves were tested time data in the water absorption and loss experiment, respect- with linear regression analyses. We used linear regressions to ively. The parameters, K, a and b, were calculated using the identify the main traits driving the variation in cyanobacterial “nls()” function in R. The exponential functions closely imitate colonization and N fixation rate: (1) relationships of water the moss weight changes through time during water absorption balance traits (maximum water content, hydration and desic- and loss processes (Supplementary Data Fig. S2). The mean R cation rate) with cyanobacterial activity and colonization; (2) of the water absorption relationship was 0.87 (0.76–0.94), and the effects of colony trait (frequency of shoots), shoot and leaf the mean R of the water loss relationship was 0.96 (0.87–0.99). traits (shoot length and leaf width) on hydration rate and cyano- The time for 50 % water absorption and the time for 50 % water bacterial colonization; (3) the relationships between cyanobac- loss, which is referred to later to hydration rate and desiccation terial colonization and shoot as well as leaf traits (frequency rate, respectively, were calculated using “uniroot()” function in of leaves, leaf length maximum and basal leaf width, and leaf R (R Core Team, 2019). area); and (4) the effects of chemical traits (pH and phenols) on N fixation activity and cyanobacterial colonization. Differences in N fixation activity across moss species were compared with one-way ANOVA. Species-specific differences N fixation in cyanobacterial colonization were compared by linear mixed Fully hydrated mosses were kept in the above-mentioned models, in which species identity was included as a fixed effect chamber for 1  week before measurement to minimize poten- and colony and shoot were included as random effects. The tial variability of N fixation activity between sampling times. ANOVA and linear mixed models were followed by Tukey’s N fixation activity was assessed using the acetylene reduction HSD test. Variation in frequency of colonized leaves across 2 Liu & Rousk — Moss traits that rule cyanobacterial colonization 151 different sections of moss shoots as well as across moss species The PCA for moss traits revealed two major dimensions of was compared by two-way ANOVA. Differences in N fixation moss traits covariation at colony level (Fig. 1B). The first PCA activity, colony structure and chemical traits of H.  splendens axis accounted for 45  % of the variation for moss traits and between sites were tested with t-tests. To demonstrate the vari- was mainly driven by water balance traits (hydration and des- ance pattern of moss traits, intra- and inter-specific coefficients iccation rate) and shoot length. The second axis accounted for of variation (CVs) of water balance, colony, chemical, shoot 30 % of the overall variance and was primarily related to leaf and leaf morphological traits were calculated. The differences area and frequency of leaves, suggesting a leaf size versus leaf in CV between intra- and inter- species were compared by number trade-off (Supplementary Data Fig. S4). Krishnamoorthy and Lee’s modified signed-likelihood ratio test (Marwick and Krishnamoorthy, 2019). Acetylene reduction rate, leaf cyanobacterial count and cyanobacterial density were Cyanobacterial colonization and hydration rate log10-transformed before all analyses. All analyses were con- ducted with R v.3.6.1 (R Core Team, 2019), and all tests were Significant differences in frequency of colonized leaves considered significant when P < 0.05. (F , F  = 7.03, P < 0.001) were found among species. L,Colonized 3,24 Maximum water content (WC , F   =  16.17, P  <  0.001), Max 3,14 hydration rate (time for 50  % water absorption, W , Absorb RESULTS F   =  10.27, P < 0.001) and decicassion rate (time for 50 % 3,14 water loss, W , F  = 9.27, P = 0.001, Fig. 3) were signifi- Lose 3,14 Moss species differences in N fixation activity cantly different among species. Hylocomium splendens, with the fastest hydration rate, had the highest cyanobacterial colon- Large variations were found in N fixation activity and cyano- ization frequency and N fixation activity. bacterial colonization between moss species (F  =  12.10, 3,14 Regression analysis revealed a strong negative relationship P  <  0.001, Table 2). The highest N fixation activity was between time for 50  % water absorption (W ) and N fix- −1 −1 Absorb 2 found in H.  splendens (113.00  ±  67.18  nmol  g   DW h ), ation activity (i.e. positive relation between hydration rate and while the lowest N fixation was found in P.  schreberi N fixation activity), and the moss colony hydration rate ex- −1 −1 (0.69  ±  0.23  nmol  g  DW h ). The N fixation activ- plained 56 % of the variation in N fixation activity (R  = 0.56, ities of T. nitens and A. turgidum were 42.02  ±  30.12 and P  <  0.001, Fig. 4). A  similar relationship was found between −1 −1 12.19  ±  3.93  nmol  g   DW h , respectively. Accordingly, a hydration rate and cyanobacterial colonization, for W Absorb higher frequency of H. splendens leaves (54.19 ± 9.36 %) was explained 38  % variation in frequency of colonized leaves colonized compared with Tnitens . , A. turgidum and P. schreberi (F , R  = 0.38, P = 0.034, Fig. 4). L,Colonized leaves (19.14  ±  5.24, 34.54  ±  6.25 and 18.19  ±  2.76  %, There was no significant relationship between shoot fre- respectively). quency (F ) and hydration rate. But shoot length (L) and basal Sh Sh width of stem leaves (W ) explained 42 % (P = 0.024) and L,Base,St 43 % (P = 0.020) variation in hydration rate (W ), respect- Absorb Covariation of N fixation activity, cyanobacterial colonization ively (Fig. 5). and moss traits The PCA for N fixation activity and cyanobacterial colon- ization revealed one major principal component axis, which Links between cyanobacterial colonization and explained 81  % of the overall variance (Fig. 1A), suggesting morphological traits a strong covariation of N fixation activity and cyanobacterial colonization. Confirming the patterns in the PCA, we found The longer the moss shoot (L ), the fewer leaves were Sh significant and positive correlations among N fixation ac- colonized at shoot level (F ; R   =  0.13, P  =  0.033, L,Colonized tivities and frequency of colonized leaves (F , Fig. 2), Supplementary Data Fig. S5). Similarly, shoot length was L,Colonized cyanobacterial count and density (CC , CD and CD , negatively correlated with cyanobacterial count (branch leaves, L,Br L,St L,Br Supplementary Data Figure S3). R   =  0.21, P = 0.020) and density at shoot level (stem leaves, −1 −1 Table 2. Mean values and s.e. (n = 3) of acetylene reduction rates (AR, nmol g  DW h ), frequency of colonized leaves (F , %), L,Colozied cyanobacteria count on stem leaf (CC , cells per leaf) and branch leaf (CC , cells per leaf) and cyanobacteria density on stem leaf L,St L,Br −2 −2 (CD , cells mm leaf area) and branch leaf (CD , cells mm leaf area), and their differences among species. Differences in AR be- L,St L,Br tween species were tested with one-way ANOVA. Differences in cyanobacterial colonization were tested with a linear mixed model in which species was included as a fixed factor and colony and shoot were included as random effects Aulacomnium turgidum Hylocomium splendens Pleurozium schreberi Tomentypnum nitens F df P AR 12.19 (3.93) 113.00 (67.18) 0.69 (0.23) 42.02 (30.12) 12.10 3,14 <0.001 F 34.54 (6.25) 54.19 (9.36) 18.19 (2.76) 19.14 (5.24) 7.04 3,24 0.001 L,Colonized CC 23.20 (5.39) 28.95 (11.67) 7.50 (1.64) 12.97 (6.32) 1.91 3,23 0.156 L,St CC – 26.63 (8.65) 4.99 (0.77) 7.73 (2.61) 2.97 2,15 0.082 L,Br CD 14.72 (3.72) 85.51 (36.21) 7.12 (2.18) 10.05 (3.91) 6.49 3,23 0.002 L,St CD – 135.84 (45.03) 7.75 (1.40) 13.32 (4.54) 9.80 2,15 0.002 L,Br 152 Liu & Rousk — Moss traits that rule cyanobacterial colonization –0.6 –0.4 –0.2 0 0.2 0.4 0.6 1.5 R = 0.44, P = 0.02 AR 0.5 CD L,St L,Colonized –0.5 CC 10 L,St –1 A. turgidum P. schreberi A. turgidum H. splendens T. nitens –1.5 –2 H. splendens –4 –2 0 24 P. schreberi PC1 (81% explained) 0.1 T. nitens –0.4 –0.2 0 0.2 0.4 0.6 20 40 60 L,St F (%) L,Colonized L,St L, Base, St F Sh L, Max, St Fig. 2 Acetylene reduction rate (AR) in relation to the frequency of leaves Phenol 0.2 Lose colonized by cyanobacteria (F ). Each coloured dot represents one moss L,Colonized Absorb colony, which was grouped by species identity indicated with different colours. Sh [C] H Acetylene reduction rate was log10-transformed before analysis. Frequency of Colony colonized leaves is the mean value of three shoots (n = 3), and 10 (A. turgidum), WC –0.2 Max pH –2 40 (P.  schreberi and T.  nitens) and 105 leaves (H.  splendens) were examined [N] on each shoot. The grey shading around the regression lines represents 95 % confidence intervals of the fitted values. –4 –0.6 –6 –4 –2 0 246 Cyanobacterial colonization as affected by pH and PC1 (45% explained) phenol content Fig. 1. PCA of cyanobacterial colonization and acetylene reduction rate (A) There was no significant correlation between pH with either and PCA of chemical and water balance, colony structural and shoot as well N fixation activity or frequency of colonized leaves (F ). 2 L,Colonized as leaf morphological traits (B). Acetylene reduction rate (AR), cyanobacteria N fixation activity was negatively related to total phenol con- count (CC) and cyanobacteria density (CD) were log10-transformed before tent of mosses (R  = 0.57, P < 0.001, Fig. 7), while there was no analysis. Because of close relationships of cyanobacteria count, density and significant correlation between phenol content and colonization morphological traits of branch leaves to those of stem leaves, only variables and traits of stem leaves were included in the analyses. (F ). L,Colonized 2 2 R   =  0.16, P  =  0.018; branch leaves, R   =  0.24, P  =  0.011, Ecosystem differences in H. splendens colonies and colonization Supplementary Data Fig. S5). A higher frequency of leaves was rates of paraphyllia colonized in the lower sections (below 15 mm) than in the top section (first 11–15 mm) of the moss shoot in all four moss spe- No significant differences were found in N fixation activ- cies (Supplementary Data Fig. S6). ities in H. splendens between the two ecosystems (arctic tundra Fewer leaves were colonized (F ) with increasing versus boreal forest), although the mean N fixation activity of L,Colonized leaf basal width (W , R   =  0.56, P  =  0.005, Fig. 5). H.  splendens colonies from the tundra site was over 7 times L,Base,St Cyanobacterial count (CC) and density (CD) were negatively higher than that of H.  splendens colonies from the forest site related to all assessed leaf size traits (leaf length, ; leaf basal L (Fig. 8). However, H.  splendens colonies from the tundra site width, W ; leaf maximum width, W ; and leaf area, A ) were characterized by a higher shoot frequency (F, P = 0.016), L,Base L,Max L Sh at colony level (Fig. 6). Similar negative correlations were also lower colony height (H , P = 0.015) and lower desiccation Colony found at leaf level (Supplementary Data Fig. S7). rate (longer time for 50 % desiccation, W , P = 0.006, Fig. 8) Lose Cyanobacterial densities of branch leaves (CD) and stem than colonies from the forest site. L,Br leaves (CD ) showed positive correlations with frequency of Stems of H.  splendens are covered by filamentous append- L,St leaves (F; branch leaves, R   =  0.77, P  =  0.002; stem leaves, ages, paraphyllia. Cyanobacteria colonized 19.2 and 30.0 % of R   =  0.36, P  =  0.041, Fig. 6). Shoot-level analyses revealed paraphyllia on the youngest, top sections and older sections, similar positive relations between frequency of leaves ( ) F respectively (Fig. 9). The frequency of paraphyllia colonized by and cyanobacterial colonization (CC, CD CD and cyanobacteria was significantly lower than that of leaves on the L,Br L,St, L,Br F , Supplementary Data Fig. S5). respective shoot sections (F  = 10.47, P = 0.003). L,Colonized 1,31 PC2 (30% explained) PC2 (13% explained) –1 –1 AR (nmol g DW h ) Liu & Rousk — Moss traits that rule cyanobacterial colonization 153 a P < 0.01 P < 0.01 100.0 a 60 ab 10.0 1.0 0.1 0 a a P < 0.01 P < 0.01 30 P < 0.01 1500 b ab ab 10 b 0 0 0 At Hs Ps Tn At Hs Ps Tn At Hs Ps Tn Fig. 3 Differences in acetylene reduction (AR), frequency of colonized leaves (F ), maximum water content (WC ), time for 50  % water absorption L,Colonized Max (W , hydration rate) and time for 50 % water loss (W , desiccation rate) among moss species. Larger W and W indicate lower hydration rate and des- Absorb Lose Absorb Lose iccation rate, respectively. Different lower case letters above error bars indicate significant (P < 0.05) differences among species according to Tukey’s HSD test. Letters on x-axes are acronyms of studied species, i.e. AtA , ulacomnium turgidum; Hs, Hylocomium splendens; Ps, Pleurozium schreberi; Tn, Tomentypnum nitens. 2 2 2 0.1 R = 0.33, P = 0.01 R = 0.56, P < 0.01 R = 0.21, P < 0.06 A. turgidum H. splendens P. schreberi T. nitens Arctic tundra Boreal forest 2 2 2 R = 0.05, P = 0.47 R = 0.38, P = 0.03 R = 0.17, P = 0.19 10 20 30 40 10 20 30 1250 1500 1750 2000 W (Min) W (h) WC (%) Absorb Lose Max −1 −1 Fig. 4 Acetylene reduction rate (AR, nmol g dw h ) and frequency of colonized leaves (F ) in relation to water balance traits, maximum water content L,Colonized (WC ), time for 50 % water absorption (W , hydration rate) and time for 50 % water loss (W , desiccation rate). Each coloured dot represents one moss Max Absorb Lose colony grouped by species identity indicated with different colours and ecosystem types, arctic tundra (filled circles) and boreal forest (filled triangles). The grey shading around the regression lines represent 95 % confidence intervals of the fitted values. F (%) –1 –1 L,Colonized AR (nmol g DW h ) WC (%) –1 –1 Max AR (nmol g DW h ) F ,Colonized (%) W (Min) Absorb W (h) Lose 154 Liu & Rousk — Moss traits that rule cyanobacterial colonization 50 2 2 R = 0.23, P = 0.11 R = 0.42, P = 0.02 R = 0.43, P = 0.02 A. turgidum H. splendens P. schreberi T. nitens Arctic tundra Boreal forest 2 2 R = 0.56, P < 0.01 R < 0.01, P = 0.98 R = 0.23, P = 0.11 24 6 810 30 50 70 0.3 0.4 0.5 0.6 0.7 –2 F (shoots cm ) L (mm) W (mm) Sh Sh L,Base,St Fig. 5 Time for 50 % water absorption (W , hydration rate) and frequency of colonized leaves (F ) in relation to frequency of moss shoots (F ), shoot Absorb L,Colonized Sh length (L ), and basal width of stem leaves (W ). Each coloured dot represents one colony grouped by species identity indicated with different colours. W Sh L,Base,St Absorb and F were measured at colony level, and F , L , and W represent the mean value of three shoots. The grey shading around the regression lines Sh L,Colonized Sh L,Base,St represent 95 % confidence interval of the fitted values. DISCUSSION Bay et al. (2013)’s finding that Polytrichum commune induces hormogonia, the cyanobacterial infectious units, but does not This study presents the first demonstration of the linkage be- host cyanobacteria. This could be attributed to the endohydric tween cyanobacterial colonization and traits of the moss hosts, nature of the moss in which, compared with ectohydric mosses, while the moss could be simply used as a substrate by colon- water is conducted internally and the surface of the moss is izing cyanobacteria. Further, our data supports our first hypoth- water-repellent (Proctor, 2000), thereby preventing cyanobac- esis (H1), stating that N fixation activity in moss–cyanobacteria terial colonization via external water flow along the moss shoot. associations is closely related to the abundance of epiphytic Hence, our findings could explain moisture’s positive effect on cyanobacteria. A strong link between cyanobacterial abundance N fixation in moss–cyanobacteria associations found in earlier and N fixation activity has been found in previous studies (e.g. studies (e.g. Gundale et al., 2012; Rousk et al., 2018). Smith, 1984; Rousk et al., 2013a). This allows an interchange- Our results that moss colonies comprising longer shoots able use of activity and abundance as well as extrapolation from and with wider leaves take more time to reach 50 % hydration one to the other, easing large-scale experiments. (Fig. 5) confirmed the morphological control over hydration of moss colonies (Larson, 1981). We found morphological traits (shoot length and leaf width) related negatively to cyanobac- Linking hydration rate and cyanobacterial colonization terial colonization, and conversely, traits that facilitate hydra- Our results reveal that hydration rate of the moss host, not tion increase cyanobacterial colonization, supporting our third water content per se, controls cyanobacterial colonization (Fig. hypothesis (H3). Shoot length was negatively correlated with 4), corroborating our second hypothesis (H2), that the moss cyanobacterial count and density (Supplementary Data Fig. species with the highest hydration rate hosts the most cyano- S5). This could be the result of lower hydration rates of colonies bacteria. Cyanobacteria can only move in liquid or on moist comprising longer shoots, or the longer distances the cyanobac- surfaces (Brahamsha and Bhaya, 2014) and their mobility is teria need to move to colonize longer shoots, or a combination transient (Bay et al., 2013). Hence, higher hydration rates could of both. Confirming the shoot length effect, we found a lower promote cyanobacterial colonization of moss leaves via passivfrequenc e y of colonized leaves at the top 1- to 2-cm section transport along the moss shoot. This notion could also explain than that for other sections (Supplementary Data Fig. S6). W (min) F (%) Absorb L,Colonized Liu & Rousk — Moss traits that rule cyanobacterial colonization 155 2 2 2 2 R = 0.30, P = 0.01 R = 0.55, P < 0.01 R = 0.44, P < 0.01 R = 0.37, P < 0.01 2 2 2 R = 0.61, P < 0.01 R = 0.49, P < 0.01 R = 0.70, P < 0.01 R = 0.60, P < 0.01 12 3 0.2 0.3 0.4 0.5 0.6 0.7 0.4 0.6 0.8 1.0 0.5 1.0 1.5 L (mm) W (mm) W (mm) A (mm ) L L, Base L, Max L 2 2 Species Stem R = 0.03, P = 0.61 Stem R = 0.36, P = 0.04 2 2 Branch R = 0.61, P = 0.01 Branch R = 0.77, P < 0.01 A. turgidum 1000 H. splendens 100 P. schreberi T. nitens 10 Types Branch Stem 25 75 100 25 75 100 50 50 –1 –1 F (leaves mm shoot) F (leaves mm shoot) L L Fig. 6 Cyanobacterial cells per leaf (CC) and per leaf area (CD) in relation to leaf morphological traits, leaf ), length basal (L width (W ), maximum width L L,Base (W ) and leaf area (A ), as well as the cyanobacterial abundance (CC and CD) in relation to frequency of leav ). Each coloured es (F points represents mean L,Max L L value of one colony. The grey shading around the regression lines represent 95 % confidence interval on the fitted values. Leaf width was negatively correlated with hydration rate, and that moss shoots with higher frequency of leaves host more thus negatively correlated with colonization (Fig. 5). Although cyanobacteria. wider leaves were found to hold more water by enabling leaves The negative relationship between leaf size and colonization to curve in Sphagnum species (Såstad and Flatberg, 1993; is likely driven by a trade-off between leaf size and number. Bengtsson et  al., 2020), the increase in leaf width results in a Leaf size is negatively correlated to leaf number in both vascular lower number of leaves per unit shoot length (Niinemets and plants (Kleiman and Aarssen, 2007) and mosses (Niinemets Tobias, 2019), which reduces water-holding capacity and hy- and Tobias, 2019). We found a strong negative correlation be- dration rate at the shoot level. tween leaf frequency (F) and area (A) (Supplementary Data L L Fig. S4), confirming the leaf size versus number trade-off. The trade-off implies that the smaller the leaf size, the more leaves per unit stem length the plant can carry. Mosses with up Leaf size and number trade-off, and the effect on cyanobacterial −1 to 55 leaves  mm shoot length fall at the extreme end of the colonization size versus number trade-off (Niinemets and Tobias, 2019). In Surprisingly, all measured traits related to leaf size (area, H.  splendens collected from the tundra, we found a range of −1 width etc.) were negatively correlated with cyanobacterial 52–115 leaves mm shoot, which is much higher than leaf fre- colonization, probably related to desiccation rates of larger quencies , from the study above, and an area of 0.23 ± 0.01 mm broader leaves, while leaf frequency was positively related to for branch leaves and 0.48 ± 0.04 mm for stem leaves. In cor - colonization (Fig. 6), supporting our fourth hypothesis (H4), respondence to the extremely high leaf frequency, H. splendens –2 CD (cells mm leaf) CC (cells per leaf) CC (cells per leaf) –1 CD (cells mm leaf) 156 Liu & Rousk — Moss traits that rule cyanobacterial colonization 2 2 R = 0.14, P = 0.13 R = 0.57, P < 0.01 0.1 R = 0.24, P = 0.11 R = 0.01, P = 0.75 A. turgidum H. splendens P. schreberi T. nitens Arctic tundra Boreal forest 5.1 5.3 5.5 5.7 12345 –1 pH Phenol (mg GAE g DW) −1 −1 Fig. 7 Acetylene reduction rate (AR, nmol g dw h ) and frequency of cyanobacteria-colonized leaves (F ) in relation to pH and total phenol content of L,Colonized moss colonies. Each coloured dot represents one moss colony grouped by species identity indicated with different colours and site indicated with shapes. The grey shading around the regression line represent 95 % confidence interval of the fitted values. hosts the most cyanobacteria among the studied mosses (Fig. The cyanobacterial infective units, hormogonia, are of higher 6), and has the highest N fixation activity (Figs 2 and 3). On tolerance than other cyanobacterial cells and could be less af- the other end of the trade-off spectrum, P. schreberi has the lar - fected by phenols while colonizing (Damervet  al al., 1991), but gest leaves, with an area of 0.70 ± 0.05 mm for branch leaves once differentiated to N-fixing heterocytes, phenols can in- and 1.41 ± 0.07 mm for stem leaves, and a low leaf frequency hibit N fixation activity. However, a study of cycad–cyanobac- −1 (33.90 ± 3.78 leaves mm shoot) among the three pleurocarpous teria symbiosis suggested that phenols provide a mechanism mosses. Pleurozium schreberi was the least colonized by cyano- for excluding other microbes and permitting cyanobacteria to bacteria and had the lowest N fixation activity. Tomentypnum grow in cycad roots (Obukowicz et al., 1981). Further study is nitens has slightly smaller leaves (0.64 ± 0.02 mm for branch needed to disentangle the effects of phenolics and other sec- leaves and 1.26 ± 0.09 mm for stem leaves) than P. schreberi ondary compounds on moss and cyanobacteria associations. −1 and a similar leaf frequency of 33.35 ± 4.24 leaves mm shoot The studied mosses, with pH ranges from 5.0 to 6.0, might length. AccordinglyT , .  nitens has a slightly higher cyanobac- limit the activity of cyanobacteria, since the optimum pH for terial colonization rate than P.  schreberi. Although the non-N fixation in mosses is between 5.9 and 6.2 (Smith, 1984). branching acrocarp, A.  turgidum, has the largest leaf area However, neither N fixation nor colonization was signifi- (1.68 ± 0.07 mm ) and lowest leaf frequency (9.12 ± 0.98 leaves cantly affected by pH in our study. The pH of the moss with the −1 mm shoot length) among the studied mosses, colonization rate highest cyanobacterial colonization and activity H.  , splendens, and N fixation activity was still higher than that of P. schreberi, ranged between 5.28 and 5.72, which is similar to the pH of the probably because water balance of non-branching acrocarps deleast colonized moss, - P.  schreberi, whose pH ranged between pend primarily on shoot density (Niinemets and Tobias, 2019). 5.32 and 5.64. This suggests that pH is not the most important Species with an extremely high frequency of small leaves havdeterminant of c e yanobacterial colonization and activity given shoots with large surface area, which increase water absorption the moss pH does not drop below 5. (Larson, 1981), and provide more potential colonization sites for cyanobacteria than those mosses with fewer, larger leaves. The role of ecosystem types We note that a potential weakness of our trait analyses is that Phenol content and pH effects on cyanobacterial colonization they included both inter- and intraspecific variation. We argue that In our experiment, phenols affected cyanobacterial activity trait correlations were mainly driven by interspecific variations. negatively, but not cyanobacterial colonization. This fits with Samples of each species were harvested in the same habitat, and previous findings showing that moss phenol content did not cor- intraspecific trait variations, which are attributed to environmental relate with cyanobacterial colonization (Rousk et  al., 2013a). factors should be small (Roos et al., 2019). The analyses of CVs –1 –1 F (%) L,Colonized AR (nmol g DW h ) Liu & Rousk — Moss traits that rule cyanobacterial colonization 157 P = 0.15 P = 0.01 100 60 P = 0.02 1 0 0 P = 0.67 1.5 1.00 P = 0.16 P = 0.10 0.75 1.0 0.50 0.5 0.25 0 0 0.00 P = 0.69 P = 0.22 10 P < 0.01 0 0 0 Tundra Forest Tundra Forest Tundra Forest −1 −1 Fig. 8 Differences in acetylene reduction rate (AR, nmol g DW  h ), colony structure (F , frequency of shoots and H , colony height), pH, phenol content, Sh Colony nitrogen content ([N]) and water balance-related traits (WC , maximum water content, W , time for 50 % hydration and W , time for 50 % desiccation) of Max Absorb Lose H. splendens between the two sampled sites (arctic tundra and boreal forest). Leaf Paraphyllium 0.1 mm 0.1 mm Top section Others Fig. 9 Cyanobacteria colonization (%) on leaves and paraphyllia of H. splendens. The left figure compares the frequency of colonized leaves with the frequency of colonized paraphyllia. Each coloured dot represents a respective section from one H. splendens shoot; bars show the mean ± s.e. (n = 9). Frequency of cyano- bacterial colonization was significantly higher on leaves than on paraphyllium (P = 0.003, two-way ANOVA). The picture on the right shows cyanobacteria (bright orange dots) on H. splendens leaves and paraphyllia, and the picture in the top-right corner shows cyanobacteria attached to a paraphyllium. Photos were taken under an Olympus BX61. Frequency of colonization (%) WC (%) –1 –1 Max pH AR (nmol g DW h ) –1 Phenol (mg GAE g DW) W (min) Absorb –2 F (shoots cm ) sh [N] (%) H (mm) W (h) Colony Lose 158 Liu & Rousk — Moss traits that rule cyanobacterial colonization of traits confirmed that intraspecific variations of traits were gen- paraphyllia could promote N uptake by the moss host by cre- erally smaller than interspecific variations (Supplementary Data ating a link between epiph ytic cyanobacteria and the moss stem. Table S1). For instance, the intraspecific CV of P. schreberi leaf These open questions and unknowns call for further research on frequency is significantly smaller than interspecific CV (0.12 the paraphyllium’s role in the mosses’ N uptake, which could versus 0.82, P  =  0.04, modified signed-likelihood ratio test). act as a lens through which to resolve the link between traits Moreover, to demonstrate the intraspecific variation of traits and and ecological functions. N fixation rate, we compared several traits of H. splendens be- In conclusion, our findings emphasize that the hydration rate tween ecosystem types (subarctic tundra versus boreal forest). of a moss colony is a key trait regulating cyanobacterial colon- Hylocomium splendens colonies from the tundra were shorter ization. On the other hand, species-specific morphological traits and had more shoots per area than the colonies from the forest control hydration rate. Chemical traits of the moss host seem (Fig. 7). These differences between ecosystems are likely drivento be less important than morphological traits in re gulating by environmental factors, such as temperature, moisture (Bisbee cyanobacterial colonization. We could also demonstrate that et  al., 2001) and nutrient availability, as the tundra has a lower a considerable portion of c yanobacteria colonize paraphyllia, mean annual temperature (0.2 versus 1  °C) and precipitationa pre viously overlooked structure in terms of cyanobacterial (337 versus 570  mm) than the forest site (Rousk et  al., 2014; colonization. Variation in moss species-specific traits drives Rousk and Michelsen, 2017), impacting moss growth. Shoot cyanobacterial colonization, but intraspecific variation that af- size controls water fluxes in moss colonies (Elumeeva et  al., fects colonization is in turn driven by environmental factors. 2011) (see section: Linking hydration rate and cyanobacterialGi ven that mosses are a key source of N to ecosystems where colonization), and can thereby control cyanobacterial coloniza-they dominate the ground cover, uncovering the relation be- tion. Fine-scale variations in colony structure, such as changestween moss traits and c yanobacterial colonization will ultim- in shoot frequency, alter surface roughness and further interact ately result in a better estimation of the amount of N input – as with wind flow affecting boundary-layer properties and desic- dependent on moss traits – and can provide new perspectives cation rate (Rice and Schneider, 2004Michel ; et al., 2012). This and information for understanding the relationship between is confirmed by the distinct desiccation rate for H.  splendens mosses and cyanobacteria. colony between tundra and forest sites (Fig. 7). It is possible that trait variation, driven by ecosystem specific factors (e.g. N avail- ability, temperature), are responsible for intraspecific differences SUPPLEMENTARY DATA in cyanobacterial colonization and thus N fixation, while relative Supplementary data are available online at https://academic. interspecific variation should remain similar in ecosystems with oup.com/aob and consist of the following. Figure S1: pictures different abiotic conditions. of the studied moss species. Figure S2: change in moss colony We did not assess cyanobacterial colonization for H. splendens weight over time for four moss species during water absorp- from the forest site, but given that N fixation activity is closely tion and loss. Figure S3: acetylene reduction rate in relation to linked to cyanobacterial colonization, and activity was not dif- the cyanobacteria count and density on moss leaves. Figure S4: ferent between the sites, it is likely that colonization is similar relationship between leaf area and frequency of leaves for in- between the sites, too (Fig. 8). Moreover, in contrast with dividual moss shoots. Figure S5: relationships between cyano- H. splendens collected from the forest site, P. schreberi had still −1 −1 bacterial colonization and shoot traits. Figure S6: differences in lower N fixation activity (27.75 versus 0.68 nmol g  DW h , frequency of colonized leaves between top segments and lower Fig. 3). Hence, the variation in cyanobacterial colonization and, segments. Figure S7: relationships between cyanobacterial col- thus N fixation activity, should be largely controlled by species- onization and leaf size. Table S1: intra- and interspecific CVs of specific traits, but intraspecific variation that affects colonization water balance, colony, chemical and morphological traits. is also in turn driven by the environment. FUNDING Morphological peculiarity of H. splendens Xin Liu was sponsored by the Second Tibetan Plateau Scientific Our results present the first evidence that cyanobacterial col- Expedition and Research program (STEP) (2019QZKK0301), onization of paraphyllia can be substantial. About 20–30 % of the National Natural Science Foundation of China (No. paraphyllia were colonized by cyanobacteria (Fig. 9). Given 31600357) and a visiting scholarship from Chinese Academy the large number of paraphyllia along the moss shoot, asso- of Sciences. Funding was also provided by the Independent ciated cyanobacteria can account for a considerable portion Research Fund Denmark, Sapere Aude Grant to K.R. (grant id: of N fixation. Hence, focusing on moss leaves as coloniza- 7027-00011B). tion sites could vastly underestimate cyanobacterial coloniza- tion along moss shoots. Paraphyllia occur in taxa of only a few moss families, such as Hylocomiaceae, Thuidiaceae and ACKNOWLEDGEMENTS Brachytheciaceae (Spirina et al., 2020). If those taxa that have paraphyllia host more cyanobacteria than taxa without these We thank F. Ekelund for help with microscopic analyses, A.T. structures is an open question. Moreover, paraphyllia, which F. Permin for help with the ethylene analyses, and E. Nielsen are photosynthetic filaments or leaf-like structures composed and G.  Sylvester for assistance with the chemical analyses. of live cells, are located between leaf and stem, where man X.L.  y and K.R.  designed the experiment. K.R.  collected the N -fixing cyanobacteria occur. Although these thick-walled samples. X.L. performed the laboratory experiment. X.L. ana- cells may have low nutrient exchange rates, it is possible that lysed the data. Both authors wrote the paper. The authors Liu & Rousk — Moss traits that rule cyanobacterial colonization 159 Meeks JC, Elhai J. 2002. Regulation of cellular differentiation in filamentous declare that they have no conflicts of interest with the contents cyanobacteria in free-living and plant-associated symbiotic growth states. of this article. Microbiology and Molecular Biology Reviews 66: 94–121. Michel  P, Lee  WG, During  HJ, Cornelissen  JHC. 2012. Species traits and their non-additive interactions control the water economy of bryophyte cushions. Journal of Ecology 100: 222–231. LITERATURE CITED Niinemets Ü, Tobias M. 2019. Canopy leaf area index at its higher end: dissec- Adams DG, Duggan PS. 2008. Cyanobacteria–bryophyte symbioses. Journal tion of structural controls from leaf to canopy scales in bryophytes. New of Experimental Botany 59: 1047–1058. Phytologist 223: 118–133. Alvarenga DO, Rousk K. 2021. Indirect effects of climate change inhibit N Obukowicz  M, Schaller  M, Kennedy  GS. 1981. Ultrastructure and phen- fixation associated with the feathermoss Hylocomium splendens in sub- olic histochemistry of the Cycas revoluta-Anabaena symbiosis. New arctic tundra. Science of The Total Environment 795: 148676. Phytologist 87: 751–759. Bauer  IE, Tirlea  D, Bhatti  JS, Errington  RC. 2007. Environmental and Proctor MCF. 1982. Physiological ecology: water relations, light and tempera- biotic controls on bryophyte productivity along forest to peatland eco- ture responses, carbon balance. In: Smith  AJE, ed. Bryophyte ecology. tones. Canadian Journal of Botany-Revue Canadienne De Botanique 85: Dordrecht: Springer, 333–381. 463–475. Proctor  MCF. 2000. Physiological ecology. In: Shaw  AJ, Goffinet  B, eds. Bay G, Nahar N, Oubre M, et al. 2013. Boreal feather mosses secrete chem- Bryophyte biology. Cambridge: Cambridge University Press, 225–247. ical signals to gain nitrogen. New Phytologist 200: 54–60. R Core Team. 2019. R: A language and environment for statistical computing. Bengtsson F, Granath G, Cronberg N, Rydin H. 2020. Mechanisms behind Vienna, Austria: R Foundation for Statistical Computing. species-specific water economy responses to water level drawdown in peat Renaudin  M, Darnajoux  R, Bellenger  JP. 2021. Quantification of moss- mosses. Annals of Botany 126: 219–230. associated cyanobacteria using phycocyanin pigment extraction. Frontiers Bisbee K, Gower S, Norman J, Nordheim E. 2001. Environmental controls in Microbiology 11: 611792. on ground cover species composition and productivity in a boreal black Rice  SK, Schneider  N. 2004. Cushion size, surface roughness, and the con- spruce forest. Oecologia 129: 261–270. trol of water balance and carbon flux in the cushion moss Leucobryum Brahamsha  B, Bhaya  D. 2014. Motility in unicellular and filamentous glaucum (Leucobryaceae). American Journal of Botany 91: 1164–1172. Cyanobacteria. In: Flores E, Herrero A. eds. The cell biology of cyanobac- Roos  RE, van  Zuijlen  K, Birkemoe  T, et  al. 2019. Contrasting drivers of teria. Seville: Caister Academic Press, 233–262. community-level trait variation for vascular plants, lichens and bryophytes Broady PA. 1979. The Signy Island terrestrial reference sites: IX. The ecology across an elevational gradient. Functional Ecolo33 gy : 2430– 2446. of the algae of site 2, a moss carpet. British Antarctic Survey Bulletin 47 : Rousk  K, Michelsen  A. 2017. Ecosystem nitrogen fixation throughout the 13–29. snow-free period in subarctic tundra: effects of willow and birch litter add- Cruz de Carvalho R, Maurício A, Pereira MF, da Silva JM, Branquinho C. ition and warming. Global Change Biology 23: 1552–1563. 2019. All for one: The role of colony morphology in bryophyte desicca-Rousk J, Brookes PC, Bååth E. 2009. Contrasting soil pH effects on fungal tion tolerance. Frontiers in Plant Science 10: 1360. and bacterial growth suggest functional redundancy in carbon mineraliza- Dalton  DA, Chatfield  JM. 1985. A new nitrogen–fixing cyanophyte-hepatic tion. Applied and Environmental Microbiology 75: 1589–1596. association: Nostoc and Porella. American Journal of Botany 72: 781–784. Rousk  K, DeLuca  TH, Rousk  J. 2013a. The cyanobacterial role in the re- Damerval T, Guglielmi G, Houmard J, De Marsac NT. 1991. Hormogonium sistance of feather mosses to decomposition—Toward a new hypothesis. differentiation in the cyanobacterium Calothrix: A photoregulated devel- PLoS ONE 8: e62058. opmental process. The Plant Cell 3: 191–201. Rousk  K, Rousk  J, Jones  DL, Zackrisson  O, DeLuca  TH. 2013b. Feather Elumeeva  TG, Soudzilovskaia  NA, During  HJ, Cornelissen  JHC. 2011. moss nitrogen acquisition across natural fertility gradients in boreal for - The importance of colony structure versus shoot morphology for the water ests. Soil Biology & Biochemistry 61: 86–95. balance of 22 subarctic bryophyte species. Journal of Vegetation Science Rousk K, Jones DL, DeLuca TH. 2014. Moss-nitrogen input to boreal forest 22: 152–164. soils: Tracking N in a field experiment. Soil Biology & Biochemistry 72: Erickson M, Miksche GE. 1974. On the occurrence of lignin or polyphenols 100–104. in some mosses and liverworts. Phytochemistry 13: 2295–2299. Rousk  K, Sorensen  PL, Lett  S, Michelsen  A. 2015. Across-habitat com- Gundale  MJ, Deluca  TH, Nordin  A. 2011. Bryophytes attenuate anthropo- parison of diazotroph activity in the subarctic. Microbial Ecology, 69: genic nitrogen inputs in boreal forests. Global Change Biology17 : 778–787. 2743–2753. Rousk  K, Degboe  J, Michelsen  A, Bradley  R, Bellenger  JP. 2017. Gundale  MJ, Wardle  DA, Nilsson  MC. 2012. The effect of altered Molybdenum and phosphorus limitation of moss-associated nitrogen fix- macroclimate on N-fixation by boreal feather mosses. Biology Letter8 s : ation in boreal ecosystems. New Phytologist 214: 97–107. 805–808. Rousk K, Sorensen PL, Michelsen A. 2018. What drives biological nitrogen Jackson BG, Martin P, Nilsson MC, Wardle DA. 2011. Response of feather fixation in high arctic tundra: Moisture or temperature? Ecosphere 9: moss associated N2 fixation and litter decomposition to variations in simu- e02117. lated rainfall intensity and frequency. Oikos 120: 570–581. Santi  C, Bogusz  D, Franche  C. 2013. Biological nitrogen fixation in non- Jean  M, Holland-Moritz  H, Melvin  AM, Johnstone  JF, Mack  MC. 2020. legume plants. Annals of Botany 111: 743–767. Experimental assessment of tree canopy and leaf litter controls on Såstad  SM, Flatberg  KI. 1993. Leaf morphology of Sphagnum strictum in the microbiome and nitrogen fixation rates of two boreal mosses. New norway, related to habitat characteristics. Lindbergia 18: 71–77. Phytologist 227: 1335–1349. Smith VR. 1984. Effects of abiotic factors on acetylene reduction by cyanobac- Kleiman  D, Aarssen  LW. 2007. The leaf size/number trade-off in trees. teria epiphytic on moss at a subantarctic island. Applied and Environmental Journal of Ecology 95: 376–382. Microbiology 48: 594–600. Lang SI, Cornelissen JHC, Klahn T, et al. 2009. An experimental comparison Solheim B, Endal A, Vigstad H. 1996. Nitrogen fixation in Arctic vegetation of chemical traits and litter decomposition rates in a diverse range of sub- and soils from Svalbard, Norway. Polar Biology 16: 35–40. arctic bryophyte, lichen and vascular plant species. Journal of Ecology Solheim  B, Zielke  M. 2002. Associations between cyanobacteria and 97: 886–900. mosses. In: Rai AN, Bergman B, Rasmussen U, eds. Cyanobacteria in Larson DW. 1981. Differential wetting in some lichens and mosses: The role Symbiosis. Dordrecht: Springer, 137–152. of morphology. The Bryologist 84: 1–15. Spirina UN, Voronkova TV, Ignatov MS. 2020. Are all paraphyllia the same? LeBauer DS, Treseder KK. 2008. Nitrogen limitation of net primary product- Frontiers in Plant Science 11: 858. ivity in terrestrial ecosystems is globally distributed. Ecolo 89 gy : 371–379. Stuart JEM, Holland-Moritz H, Lewis LR, et al. 2021. Host identity as a driver Leppänen  SM, Salemaa  M, Smolander  A, Mäkipää  R, Tiirola  M. 2013. of moss-associated N2 fixation rates in Alaska. Ecosystems24 : 530–547. Nitrogen fixation and methanotrophy in forest mosses along a N depos- Wang  YP, Law  RM, Pak  B. 2010. A global model of carbon, nitrogen ition gradient. Environmental and Experimental Botany 90: 62–69. and phosphorus cycles for the terrestrial biosphere. Biogeosciences 7: Marwick  B, Krishnamoorthy  K. 2019. cvequality: tests for the equality 2261–2282. of coefficients of variation from multiple groups. R software package Wilde  A, Mullineaux  CW. 2015. Motility in cyanobacteria: polysaccharide version 0.1.3. tracks and Type IV pilus motors. Molecular Microbiology 98: 998–1001. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Annals of Botany Oxford University Press

The moss traits that rule cyanobacterial colonization

Annals of Botany , Volume 129 (2): 14 – Oct 10, 2021

Loading next page...
 
/lp/oxford-university-press/the-moss-traits-that-rule-cyanobacterial-colonization-UZ00bo9AdB

References (51)

Publisher
Oxford University Press
Copyright
© The Author(s) 2021. Published by Oxford University Press on behalf of the Annals of Botany Company.
ISSN
0305-7364
eISSN
1095-8290
DOI
10.1093/aob/mcab127
Publisher site
See Article on Publisher Site

Abstract

Annals of Botany 129: 147–159, 2022 https://doi.org/10.1093/aob/mcab127, available online at www.academic.oup.com/aob 1,2,3, , 2,3 Xin Liu * and Kathrin Rousk CAS Key Laboratory of Mountain Ecological Restoration and Bioresource Utilization & Ecological Restoration and Biodiversity Conservation Key Laboratory of Sichuan Province, Chengdu Institute of Biology, Chinese Academy of Sciences, Chengdu 610041, China, Department of Biology, Terrestrial Ecology Section, University of Copenhagen, Universitetsparken 15, 2100 Copenhagen, Denmark and Center for Permafrost (CENPERM), University of Copenhagen, Øster Voldgade 10, 1350 Copenhagen, Denmark *For correspondence. E-mail liuxin1@cib.ac.cn Received: 17 May 2021 Returned for revision: 5 September 21 Editorial decision: 1 October 2021 Accepted: 8 October 2021 Electronically published: 10 October 2021 • Background and Aims Cyanobacteria associated with mosses represent a main nitrogen (N) source in pristine, high-latitude and -altitude ecosystems due to their ability to fix N . However, despite progress made regarding moss–cyanobacteria associations, the factors driving the large interspecific variation in N fixation activity be- tween moss species remain elusive. The aim of the study was to identify the traits of mosses that determine cyano- bacterial colonization and thus N fixation activity. • Methods Four moss species varying in N fixation activity were used to assess cyanobacterial abundance and activity to correlate it with moss traits (morphological, chemical, water-balance traits) for each species. • Key Results Moss hydration rate was one of the pivotal traits, explaining 56 and 38 % of the variation in fix- N ation and cyanobacterial colonization, respectively, and was linked to morphological traits of the moss species. Higher abundance of cyanobacteria was found on shoots with smaller leaves, and with a high frequency of leaves. High phenol concentration inhibited N fixation but not colonization. These traits driving interspecific variation in cyanobacterial colonization, however, are also affected by the environment, and lead to intraspecific variation. Approximately 24 % of paraphyllia, filamentous appendages on Hylocomium splendens stems, were colonized by cyanobacteria. • Conclusions Our findings show that interspecific variations in moss traits drive differences in cyanobacterial colonization and thus, N fixation activity among moss species. The key traits identified here that control moss- associated N fixation and cyanobacterial colonization could lead to improved predictions of N fixation in dif- 2 2 ferent moss species as a function of their morphology. Key words: Bryophytes, moss colony, Cyanobacteria, nitrogen fixation, water retention, functional trait. N fixation activity between mosses growing in the same habitat INTRODUCTION occur. This lack of understanding of the seemingly random col- Nitrogen (N) fixation performed by moss-associated cyano- onization patterns among moss species hampers efforts to up- bacteria is a main N source in many pristine ecosystems, such scale N fixation across moss species at a larger scale. as boreal forests and subarctic tundra, where N deposition is Cyanobacteria are the dominant N fixer associated with low and plant growth is commonly limited by N availability mosses (Leppänen et  al., 2013), and N fixation activity is (LeBauer and Treseder, 2008; Wang et  al., 2010). Given the commonly correlated with the number of epiphytic cyano- high abundance of mosses in these ecosystems, the association bacterial cells on mosses (Rousk et  al., 2013a). Thus, differ- contributes significantly to ecosystems’ N cycle by accounting ences in the abundance of epiphytic cyanobacteria likely drive for up to 50 % of the total N input (e.g. Rousk and Michelsen, the interspecific variation of moss-associated N fixation rates. 2017). Wide-ranging taxa of mosses have been found to host Cyanobacterial abundance, on the other hand, is affected by dif- cyanobacteria, but N fixation activity of the association ferent factors, such as the capability to move or disperse, and varies greatly among moss species (Rousk et al., 2015; Stuart the environmental factors affecting these processes (Solheim et al., 2021). For instance, N fixation activity in Hylocomium and Zielke, 2002). Among these factors, traits of the moss host, splendens can be more than double the activity found in which provide microsites for epiphytic cyanobacteria (Dalton Pleurozium schreberi, although they are co-dominant in boreal and Chatfield, 1985), might play a key role as moss traits vary forests (Stuart et al., 2021), while the activity in H. splendens is greatly among species (Niinemets and Tobias, 2019), and cer - only one-sixth the activity found in Sphagnum in arctic tundra tain moss species seem to be especially colonized by cyano- (Rousk et  al., 2015). To date, although many factors, such as bacteria (Solheim et  al., 1996). However, to date, it remains nutrient availability (Gundale et al., 2011; Rousk et al., 2017) unknown whether moss traits affect the colonization of cyano- and moisture content (Rousk et al., 2015, 2018) affect N fix- bacteria and, if so, which suite of moss traits facilitates cyano- ation in mosses, we do not know why these large variations in bacterial colonization that leads to differences between moss © The Author(s) 2021. Published by Oxford University Press on behalf of the Annals of Botany Company. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/ by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 148 Liu & Rousk — Moss traits that rule cyanobacterial colonization hosts. The paucity of data on the effects of species-specific concentration of host mosses might lead to variation in cyano- moss traits on cyanobacterial colonization limits our under - bacterial colonization and activity. Similarly, pH is an influ- standing of the relationship between moss and cyanobacteria, ential factor structuring microbial communities (Rousk et  al., and thereby of the functioning of moss-dominated ecosystems 2009) and affecting N fixation rates in mosses (Alvarenga and such as boreal forests. Rousk, 2021). To date, it is unknown if differences in the chem- Studies have repeatedly found that moisture promotes N ical environment among moss species lead to variations in the fixation in moss–cyanobacteria associations (Smith, 1984; number of epiphytic cyanobacteria. Jackson et al., 2011; Gundale et al., 2012; Rousk et al., 2018). The purpose of the study reported herein was to identify the This could be the result of either a direct positive effect of water moss traits that affect cyanobacterial colonization, and thus N availability on cyanobacterial activity, or indirectly, promoting fixation activity, using four different moss species that have the colonization of cyanobacteria, or a combination of both. To been shown to vary in N fixation. To accomplish this, we meas- form an effective N-fixing association with plants, vegetative ured acetylene reduction rate as a measure of N fixation ac- 2 2 cyanobacterial cells differentiate into motile and short-lived tivity, and assessed cyanobacterial colonization and abundance hormogonia with a gliding activity for 48–72  h (Meeks and at shoot and colony (group of shoots) level and linked this to Elhai, 2002). The capacity to move would affect the number of water balance traits (maximum water content, water absorption cyanobacteria colonizing mosses (Santi et  al., 2013). A  moist rate and water loss rate of moss colonies), chemical traits (pH, surface or liquid is required for hormogonia to glide or swim total phenols), colony structural traits (frequency of shoots and (Brahamsha and Bhaya, 2014), and it is likely that passive height) and morphological traits (shoot length, frequency of water flow carries cyanobacteria from lower parts of a moss leaves, leaf area, etc., Table 1) at shoot as well as at leaf level of shoot to its upper segments (Broady, 1979). Studies concerning the four moss species collected in the subarctic region. We hy- the mobility of hormogonia are mainly focused on structures pothesized that (1) N fixation activity is correlated with cyano- of cyanobacteria (e.g. Adams and Duggan, 2008; Wilde and bacterial colonization in all investigated moss species, (2) the Mullineaux, 2015), while the characteristics of the moss host moss species that has the highest water absorption rate hosts the may affect movement capacity and hence the abundance of epi-most cyanobacteria, (3) moss traits (e.g. frequency of shoots) phytic cyanobacteria. If moss colonies absorb water at different that facilitate water absorption increase cyanobacterial colon- rates, the rate of water flow may result in differences in cyano- ization, (4) moss shoots with higher frequency of leaves host bacterial abundance between moss species and thereby dif-fer more cyanobacterial colonizers, and (5) high phenol concentra- ences in N fixation. tion and low pH inhibit cyanobacterial colonization. Water retention capacity and water absorption rate of mosses may be important for hosting cyanobacteria, and are affected by colony structure (e.g. density and height) as well as by traits MATERIALS AND METHODS of individual shoots (e.g. leaf frequency). Indeed, moss colony density and height have been reported to correlate positively Moss sampling with water retention capacity (Proctor, 1982; Elumeeva et  al., 2011), shoot morphology controls dehydration rate of mosses Moss samples were collected at two sites in Northern Sweden. (Cruz de Carvalho et  al., 2019), and leaf width is positively Aulacomnium turgidum (Wahlenb.) Schwägr., Hylocomium related to water retention in Sphagnum mosses (Bengtsson splendens (Hedw.) Schimp. and Tomentypnum nitens (Hedw.) et  al., 2020). However, to our knowledge, there is no empir - Loeske (Supplementary Data Figure S1) were collected in June ical evidence that these hydrology-related traits affect cyano2019 in a subarctic dry heath close to the - Abisko Scientific bacterial colonization of mosses. Moreover, moss traits might Research Station (68°19′02″N, 18°50′04″E). The mean an- affect colonization rates directly. Mosses are assumed to host nual air temperature in Abisko is 0.2  °C, and the mean annual and protect cyanobacteria between or on their leaves (Dalton precipitation is 337  mm (30-year mean 1986–2015, Abisko and Chatfield, 1985), which vary 54-fold in size and 28-fold Scientific Research Station 2016). The site was dominated in frequency among species (Niinemets and Tobias, 2019), by mosses and Vaccinium uliginosum, Andromeda polifolia and cyanobacterial filaments are found between moss stems and Rhododendron lapponicum (see Rousk and Michelsen, and leaves (Solheim and Zielke, 2002). Thus, moss species 2017). Pleurozium schreberi (Brid.) Mitt., which did not occur with larger leaves and higher leaf frequency should host more at this site, was collected in August 2019 in a boreal forest cyanobacteria due to higher availability of colonization sites. near Arvidsjaur (64°58′49″N 19°33′59″E). This extended the Yet again, the relationships between colony- (e.g. frequency of study to another species that has been shown previously to shoots and colony height), shoot- (e.g. frequency of leaves and differ in N fixation rates compared to H.  splendens despite shoot length) and leaf-level (e.g. leaf width and area) traits and their similar morphology and habitat preferences (Jean et  al., cyanobacteria colonization remain elusive. 2020). Mean annual temperature and precipitation are approxi- Another possible mechanism that regulates cyanobacterial mately 1  °C and 570 mm respectively. The forest was domin- colonization is the production of chemical compounds by the ated by Picea abies, V. vitis-idaea, V. myrtillus and Empetrum host. Mosses are known to produce and accumulate inhibitory hermaphroditum (see Rousk et al., 2013b). From this site, sam- compounds like phenols (Erickson and Miksche, 1974). These ples of H. splendens were also collected in order to identify dif- compounds can lead to inhibition of bacterial growth (Rousk ferences in the measured variables across ecosystems, as well et  al., 2013a) and contributes to the low decomposability as to ascertain if differences in N fixation is a species or an of moss litter (Lang et  al., 2009). Differences in phenol ecosystem effect. Liu & Rousk — Moss traits that rule cyanobacterial colonization 149 Table 1. Evaluated variables and their symbols, definitions and units Symbol Description Unit −1 −1 AR Acetylene reduction rate nmol g DW h F Frequency of cyanobacteria colonized leaves % L,Colonized −1 CC Cyanobacteria count on stem leaves Cells leaf L,St −1 CC Cyanobacteria count on branch leaves Cells leaf L,Br −2 CD Cyanobacteria density on stem leaves Cells mm leaf L,St −2 CD Cyanobacteria density on branch leaves Cells mm leaf L,Br WC Maximum water content % Max W Time needed for moss colony to absorb water from air dried status to 50 % of min Absorb maximum water content. Larger number means lower hydration rate. W Time needed for moss colony to lose water from 100 to 50 % of maximum h Lose water content. Larger number means lower desiccation rate. H Height of moss colony mm Colony −2 F Frequency of Shoot Shoots cm Sh pH pH - −1 Phenol Total phenol concentration mg GAE g [C] Carbon concentration % [N] Nitrogen concentration % L Length of shoot mm Sh −1 F Frequency of leaves – number of (stem and branch) leaves per unit shoot length Leaves mm shoot W Basal width of stem leaf mm L,Base,St W Maximum width of stem leaf mm L,Max,St L Length of stem leaf mm L,St A Area of stem leaf mm L,St W Basal width of branch leaf mm L,Base,Br W Maximum width of branch leaf mm L,Max,Br L Length of branch leaf mm L,Br A Area of branch leaf mm L,Br Three separate monospecific moss colonies, with at least and 8 mm for T. nitens, which represents 1 year’s length growth 5 m distance from each other, were selected for each species (Bauer et  al., 2007), and other sections. Brown segments that at the subarctic and boreal sites. Uniform moss colonies of were partly decomposed were not included in the measure- 15 cm × 15 cm were sampled and carefully transported to the ments. Fully hydrated leaves from these chosen branches and laboratory in Copenhagen. The samples were assessed for N stem sections (351 branches/stem sections in total: 18 for fixation activity, water balance (maximum water content, water A.  turgidum, 72 for P.  schreberi, 72 for T.  nitens and 189 for absorption and -loss rate), chemical (pH and total phenols), H. splendens) were measured. shoot and leaf morphological traits, and cyanobacterial counts. To be able to link morphological traits to cyanobacterial col- The H. splendens samples from the boreal forest site were as-onization, we destructively harvested all the leaves from 1- to sessed for N fixation activity, water balance, and chemical 3-mm lengths of stem segments or branches through scraping traits. to enable counting. The number of scraped-off leaves was counted using an Olympus SZX16 stereo microscope. In total, 4857 leaves (328 for A. turgidum, 1141 for P. schreberi, 1469 for T. nitens and 1919 for H. splendens) were scraped off. The Morphological traits and cyanobacterial colonization number of cyanobacteria-colonized leaves among scraped- off leaves was counted using an Olympus BX61 ultraviolet- Three shoots from each sample were randomly selected to fluorescence microscope with a green filter. The frequency of measure shoot and leaf morphological traits (n  =  9 per spe- colonized leaves for each stem segment or branch was calculated cies). We measured the length of each shoot after submer - according to the number of colonized leaves and the number of ging in double distilled (dd) H O to ensure full hydration. For all scraped-off leaves. Digital images of five randomly selected H.  splendens, shoots were divided into three segments ac- leaves were taken with a USB2.0 CMOS Camera (ToupTek, cording to innate growth markers. These segments were: top- Hangzhou, China) attached to the stereo microscope. From most current year segment which are younger than one-year these images, the maximum width, basal width, length and area (ca. 11 mm), the segment younger than 2 years but older than of individual leaves (Table 1) were measured with ImageJ 2.35 1  year, and the segment older than 2  years. The stem and six (Wayne Rasbund, National Institutes of Health, Bethesda, MD, random branches were chosen from each segment for leaf USA). The cyanobacterial cells on the five selected leaves per measurements and cyanobacterial counting. For P.  schreberi stem segment or branch were counted using the Olympus BX61 and T.  nitens, six random branches were chosen from each ultraviolet-fluorescence microscope. The leaf size and number shoot. For A.  turgidum, which does not have branches, only of cyanobacterial cells for 1755 leaves (90 for A. turgidum, 360 stems leaves were measured. Stems of A. turgidum P, . schreberi for P. schreberi, 360 for T. nitens and 945 for H. splendens) were and T.  nitens were divided into two sections: the top sections, measured. Microscopic counting, rather than using proxies of approximately 9 mm for A.  turgidum, 13 mm for P.  schreberi 150 Liu & Rousk — Moss traits that rule cyanobacterial colonization cyanobacterial quantity (Renaudin et  al., 2021), enabled us to assay (ARA). For this, 20-mL glass vials containing ten fully link leaf traits with cyanobacterial colonization. hydrated moss shoots (n = 3 for each species) were sealed and 10 % of the headspace was replaced with acetylene. The moss samples were incubated for 10 h at 12 °C, 10 h at 6 °C, then 4 h at 12 °C. Ethylene generated in the headspace by the cyano- Water balance bacterial nitrogenase enzyme was measured by gas chroma- Water absorption. To measure water absorption rates of tography with a flame ionization detector using an automatic moss colonies, the bottom of transparent polypropylene cups, headspace sampler (Agilent, 8890 GC System, Agilent, Santa 3.7 cm in diameter, were cut off and replaced with cotton mesh Clara, USA). The fresh moss shoots were oven-dried at 65 °C (Supplementary Data Fig. S2). We placed a round moss colonyfor 48  , h and ground into fine powder, which was subsequently which fits the area of the cup, into the cup. Moss patches had used for total carbon (TC), total nitrogen (TN) and phenol con- similar height to those in the field, and included both green and centration measurement. basal parts, while decomposed parts were excluded. Cups filled with moss colonies were air-dried in a venti- lated growth chamber for one week. The air temperature in the Nutrient concentrations, total phenols and pH measurements chamber was 6 °C in the night (1800–0600 h) and 12 °C during daytime (0600–1800 h). The relative humidity in the chamber Carbon and N concentrations in moss tissue were assessed varied between 51.2 and 97.7 %, with an average of 76.6 %. The with a Vario Macro Cube Elemental Analyzer (Elementar, cups were then weighed and placed into a plastic box, which Germany). Total phenols were measured in moss tissue that was filled with ddH O to 1 cm depth. The cups were weighed had been ground into fine powder and then suspended in 10 mL at time intervals of 2 min for the initial 20 min, then weighed at ethanol. Samples were shaken for 120 min and then centrifuged intervals of 5 min for the rest of the first hour, and then weighed at 3600 g for 10 min. The supernatant was analysed for phenols every hour until their mass became nearly constant. We added using the Folin–Ciocalteu reagent. The absorbance was meas- ddH O to the box during the experiment to maintain the same ured at 725 nm using a spectrophotometer. The pH of mosses water level throughout the measurements. was measured in 3 g fresh moss tissue that was submerged in Water loss. The same cups filled with moss colonies were 15 mL ddH O and then shaken for 60 min. The pH of the ex- used to determine the water loss rate of the moss colonies. tracts was determined with a pH electrode. Water (ddH O) was added to the box to immerse the moss col- onies and left for 12 h, allowing the colonies to reach full hy- dration. The cups were then placed on a tilted plastic surface Statistical analyses for 5 min to let the surplus water run down. The samples were weighed and kept in the same chamber as above and allowed to We first performed principal component analyses (PCAs) dry. The cups were weighed every hour during the initial 12 h using acetylene reduction rate and cyanobacterial colonization and every 6 h for a maximum of 120 h until their mass became variables, as well as the water balance, chemical and morpho- nearly constant. Then we counted the number of moss shoots in logical traits to obtain an overview of the multidimensional each cup, oven-dried the mosses at 65 °C for 48 h and recorded cyanobacterial colonization and moss traits spectrum of vari- their dry weight. ation. Because of close relationships of cyanobacterial count, Calculations. Maximum water content of moss colonies were density and morphological traits of branch leaves to those of calculated and expressed as percentage of dry weight (Table 1). stem leaves, only variables and traits of stem leaves were in- Exponential functions weight =  K/(1 + exp(a + b × time)) and cluded in the analyses. The relationships between acetylene weight = a × exp(−b × time) were fitted to the moss weight and reduction rate and frequency of colonized leaves were tested time data in the water absorption and loss experiment, respect- with linear regression analyses. We used linear regressions to ively. The parameters, K, a and b, were calculated using the identify the main traits driving the variation in cyanobacterial “nls()” function in R. The exponential functions closely imitate colonization and N fixation rate: (1) relationships of water the moss weight changes through time during water absorption balance traits (maximum water content, hydration and desic- and loss processes (Supplementary Data Fig. S2). The mean R cation rate) with cyanobacterial activity and colonization; (2) of the water absorption relationship was 0.87 (0.76–0.94), and the effects of colony trait (frequency of shoots), shoot and leaf the mean R of the water loss relationship was 0.96 (0.87–0.99). traits (shoot length and leaf width) on hydration rate and cyano- The time for 50 % water absorption and the time for 50 % water bacterial colonization; (3) the relationships between cyanobac- loss, which is referred to later to hydration rate and desiccation terial colonization and shoot as well as leaf traits (frequency rate, respectively, were calculated using “uniroot()” function in of leaves, leaf length maximum and basal leaf width, and leaf R (R Core Team, 2019). area); and (4) the effects of chemical traits (pH and phenols) on N fixation activity and cyanobacterial colonization. Differences in N fixation activity across moss species were compared with one-way ANOVA. Species-specific differences N fixation in cyanobacterial colonization were compared by linear mixed Fully hydrated mosses were kept in the above-mentioned models, in which species identity was included as a fixed effect chamber for 1  week before measurement to minimize poten- and colony and shoot were included as random effects. The tial variability of N fixation activity between sampling times. ANOVA and linear mixed models were followed by Tukey’s N fixation activity was assessed using the acetylene reduction HSD test. Variation in frequency of colonized leaves across 2 Liu & Rousk — Moss traits that rule cyanobacterial colonization 151 different sections of moss shoots as well as across moss species The PCA for moss traits revealed two major dimensions of was compared by two-way ANOVA. Differences in N fixation moss traits covariation at colony level (Fig. 1B). The first PCA activity, colony structure and chemical traits of H.  splendens axis accounted for 45  % of the variation for moss traits and between sites were tested with t-tests. To demonstrate the vari- was mainly driven by water balance traits (hydration and des- ance pattern of moss traits, intra- and inter-specific coefficients iccation rate) and shoot length. The second axis accounted for of variation (CVs) of water balance, colony, chemical, shoot 30 % of the overall variance and was primarily related to leaf and leaf morphological traits were calculated. The differences area and frequency of leaves, suggesting a leaf size versus leaf in CV between intra- and inter- species were compared by number trade-off (Supplementary Data Fig. S4). Krishnamoorthy and Lee’s modified signed-likelihood ratio test (Marwick and Krishnamoorthy, 2019). Acetylene reduction rate, leaf cyanobacterial count and cyanobacterial density were Cyanobacterial colonization and hydration rate log10-transformed before all analyses. All analyses were con- ducted with R v.3.6.1 (R Core Team, 2019), and all tests were Significant differences in frequency of colonized leaves considered significant when P < 0.05. (F , F  = 7.03, P < 0.001) were found among species. L,Colonized 3,24 Maximum water content (WC , F   =  16.17, P  <  0.001), Max 3,14 hydration rate (time for 50  % water absorption, W , Absorb RESULTS F   =  10.27, P < 0.001) and decicassion rate (time for 50 % 3,14 water loss, W , F  = 9.27, P = 0.001, Fig. 3) were signifi- Lose 3,14 Moss species differences in N fixation activity cantly different among species. Hylocomium splendens, with the fastest hydration rate, had the highest cyanobacterial colon- Large variations were found in N fixation activity and cyano- ization frequency and N fixation activity. bacterial colonization between moss species (F  =  12.10, 3,14 Regression analysis revealed a strong negative relationship P  <  0.001, Table 2). The highest N fixation activity was between time for 50  % water absorption (W ) and N fix- −1 −1 Absorb 2 found in H.  splendens (113.00  ±  67.18  nmol  g   DW h ), ation activity (i.e. positive relation between hydration rate and while the lowest N fixation was found in P.  schreberi N fixation activity), and the moss colony hydration rate ex- −1 −1 (0.69  ±  0.23  nmol  g  DW h ). The N fixation activ- plained 56 % of the variation in N fixation activity (R  = 0.56, ities of T. nitens and A. turgidum were 42.02  ±  30.12 and P  <  0.001, Fig. 4). A  similar relationship was found between −1 −1 12.19  ±  3.93  nmol  g   DW h , respectively. Accordingly, a hydration rate and cyanobacterial colonization, for W Absorb higher frequency of H. splendens leaves (54.19 ± 9.36 %) was explained 38  % variation in frequency of colonized leaves colonized compared with Tnitens . , A. turgidum and P. schreberi (F , R  = 0.38, P = 0.034, Fig. 4). L,Colonized leaves (19.14  ±  5.24, 34.54  ±  6.25 and 18.19  ±  2.76  %, There was no significant relationship between shoot fre- respectively). quency (F ) and hydration rate. But shoot length (L) and basal Sh Sh width of stem leaves (W ) explained 42 % (P = 0.024) and L,Base,St 43 % (P = 0.020) variation in hydration rate (W ), respect- Absorb Covariation of N fixation activity, cyanobacterial colonization ively (Fig. 5). and moss traits The PCA for N fixation activity and cyanobacterial colon- ization revealed one major principal component axis, which Links between cyanobacterial colonization and explained 81  % of the overall variance (Fig. 1A), suggesting morphological traits a strong covariation of N fixation activity and cyanobacterial colonization. Confirming the patterns in the PCA, we found The longer the moss shoot (L ), the fewer leaves were Sh significant and positive correlations among N fixation ac- colonized at shoot level (F ; R   =  0.13, P  =  0.033, L,Colonized tivities and frequency of colonized leaves (F , Fig. 2), Supplementary Data Fig. S5). Similarly, shoot length was L,Colonized cyanobacterial count and density (CC , CD and CD , negatively correlated with cyanobacterial count (branch leaves, L,Br L,St L,Br Supplementary Data Figure S3). R   =  0.21, P = 0.020) and density at shoot level (stem leaves, −1 −1 Table 2. Mean values and s.e. (n = 3) of acetylene reduction rates (AR, nmol g  DW h ), frequency of colonized leaves (F , %), L,Colozied cyanobacteria count on stem leaf (CC , cells per leaf) and branch leaf (CC , cells per leaf) and cyanobacteria density on stem leaf L,St L,Br −2 −2 (CD , cells mm leaf area) and branch leaf (CD , cells mm leaf area), and their differences among species. Differences in AR be- L,St L,Br tween species were tested with one-way ANOVA. Differences in cyanobacterial colonization were tested with a linear mixed model in which species was included as a fixed factor and colony and shoot were included as random effects Aulacomnium turgidum Hylocomium splendens Pleurozium schreberi Tomentypnum nitens F df P AR 12.19 (3.93) 113.00 (67.18) 0.69 (0.23) 42.02 (30.12) 12.10 3,14 <0.001 F 34.54 (6.25) 54.19 (9.36) 18.19 (2.76) 19.14 (5.24) 7.04 3,24 0.001 L,Colonized CC 23.20 (5.39) 28.95 (11.67) 7.50 (1.64) 12.97 (6.32) 1.91 3,23 0.156 L,St CC – 26.63 (8.65) 4.99 (0.77) 7.73 (2.61) 2.97 2,15 0.082 L,Br CD 14.72 (3.72) 85.51 (36.21) 7.12 (2.18) 10.05 (3.91) 6.49 3,23 0.002 L,St CD – 135.84 (45.03) 7.75 (1.40) 13.32 (4.54) 9.80 2,15 0.002 L,Br 152 Liu & Rousk — Moss traits that rule cyanobacterial colonization –0.6 –0.4 –0.2 0 0.2 0.4 0.6 1.5 R = 0.44, P = 0.02 AR 0.5 CD L,St L,Colonized –0.5 CC 10 L,St –1 A. turgidum P. schreberi A. turgidum H. splendens T. nitens –1.5 –2 H. splendens –4 –2 0 24 P. schreberi PC1 (81% explained) 0.1 T. nitens –0.4 –0.2 0 0.2 0.4 0.6 20 40 60 L,St F (%) L,Colonized L,St L, Base, St F Sh L, Max, St Fig. 2 Acetylene reduction rate (AR) in relation to the frequency of leaves Phenol 0.2 Lose colonized by cyanobacteria (F ). Each coloured dot represents one moss L,Colonized Absorb colony, which was grouped by species identity indicated with different colours. Sh [C] H Acetylene reduction rate was log10-transformed before analysis. Frequency of Colony colonized leaves is the mean value of three shoots (n = 3), and 10 (A. turgidum), WC –0.2 Max pH –2 40 (P.  schreberi and T.  nitens) and 105 leaves (H.  splendens) were examined [N] on each shoot. The grey shading around the regression lines represents 95 % confidence intervals of the fitted values. –4 –0.6 –6 –4 –2 0 246 Cyanobacterial colonization as affected by pH and PC1 (45% explained) phenol content Fig. 1. PCA of cyanobacterial colonization and acetylene reduction rate (A) There was no significant correlation between pH with either and PCA of chemical and water balance, colony structural and shoot as well N fixation activity or frequency of colonized leaves (F ). 2 L,Colonized as leaf morphological traits (B). Acetylene reduction rate (AR), cyanobacteria N fixation activity was negatively related to total phenol con- count (CC) and cyanobacteria density (CD) were log10-transformed before tent of mosses (R  = 0.57, P < 0.001, Fig. 7), while there was no analysis. Because of close relationships of cyanobacteria count, density and significant correlation between phenol content and colonization morphological traits of branch leaves to those of stem leaves, only variables and traits of stem leaves were included in the analyses. (F ). L,Colonized 2 2 R   =  0.16, P  =  0.018; branch leaves, R   =  0.24, P  =  0.011, Ecosystem differences in H. splendens colonies and colonization Supplementary Data Fig. S5). A higher frequency of leaves was rates of paraphyllia colonized in the lower sections (below 15 mm) than in the top section (first 11–15 mm) of the moss shoot in all four moss spe- No significant differences were found in N fixation activ- cies (Supplementary Data Fig. S6). ities in H. splendens between the two ecosystems (arctic tundra Fewer leaves were colonized (F ) with increasing versus boreal forest), although the mean N fixation activity of L,Colonized leaf basal width (W , R   =  0.56, P  =  0.005, Fig. 5). H.  splendens colonies from the tundra site was over 7 times L,Base,St Cyanobacterial count (CC) and density (CD) were negatively higher than that of H.  splendens colonies from the forest site related to all assessed leaf size traits (leaf length, ; leaf basal L (Fig. 8). However, H.  splendens colonies from the tundra site width, W ; leaf maximum width, W ; and leaf area, A ) were characterized by a higher shoot frequency (F, P = 0.016), L,Base L,Max L Sh at colony level (Fig. 6). Similar negative correlations were also lower colony height (H , P = 0.015) and lower desiccation Colony found at leaf level (Supplementary Data Fig. S7). rate (longer time for 50 % desiccation, W , P = 0.006, Fig. 8) Lose Cyanobacterial densities of branch leaves (CD) and stem than colonies from the forest site. L,Br leaves (CD ) showed positive correlations with frequency of Stems of H.  splendens are covered by filamentous append- L,St leaves (F; branch leaves, R   =  0.77, P  =  0.002; stem leaves, ages, paraphyllia. Cyanobacteria colonized 19.2 and 30.0 % of R   =  0.36, P  =  0.041, Fig. 6). Shoot-level analyses revealed paraphyllia on the youngest, top sections and older sections, similar positive relations between frequency of leaves ( ) F respectively (Fig. 9). The frequency of paraphyllia colonized by and cyanobacterial colonization (CC, CD CD and cyanobacteria was significantly lower than that of leaves on the L,Br L,St, L,Br F , Supplementary Data Fig. S5). respective shoot sections (F  = 10.47, P = 0.003). L,Colonized 1,31 PC2 (30% explained) PC2 (13% explained) –1 –1 AR (nmol g DW h ) Liu & Rousk — Moss traits that rule cyanobacterial colonization 153 a P < 0.01 P < 0.01 100.0 a 60 ab 10.0 1.0 0.1 0 a a P < 0.01 P < 0.01 30 P < 0.01 1500 b ab ab 10 b 0 0 0 At Hs Ps Tn At Hs Ps Tn At Hs Ps Tn Fig. 3 Differences in acetylene reduction (AR), frequency of colonized leaves (F ), maximum water content (WC ), time for 50  % water absorption L,Colonized Max (W , hydration rate) and time for 50 % water loss (W , desiccation rate) among moss species. Larger W and W indicate lower hydration rate and des- Absorb Lose Absorb Lose iccation rate, respectively. Different lower case letters above error bars indicate significant (P < 0.05) differences among species according to Tukey’s HSD test. Letters on x-axes are acronyms of studied species, i.e. AtA , ulacomnium turgidum; Hs, Hylocomium splendens; Ps, Pleurozium schreberi; Tn, Tomentypnum nitens. 2 2 2 0.1 R = 0.33, P = 0.01 R = 0.56, P < 0.01 R = 0.21, P < 0.06 A. turgidum H. splendens P. schreberi T. nitens Arctic tundra Boreal forest 2 2 2 R = 0.05, P = 0.47 R = 0.38, P = 0.03 R = 0.17, P = 0.19 10 20 30 40 10 20 30 1250 1500 1750 2000 W (Min) W (h) WC (%) Absorb Lose Max −1 −1 Fig. 4 Acetylene reduction rate (AR, nmol g dw h ) and frequency of colonized leaves (F ) in relation to water balance traits, maximum water content L,Colonized (WC ), time for 50 % water absorption (W , hydration rate) and time for 50 % water loss (W , desiccation rate). Each coloured dot represents one moss Max Absorb Lose colony grouped by species identity indicated with different colours and ecosystem types, arctic tundra (filled circles) and boreal forest (filled triangles). The grey shading around the regression lines represent 95 % confidence intervals of the fitted values. F (%) –1 –1 L,Colonized AR (nmol g DW h ) WC (%) –1 –1 Max AR (nmol g DW h ) F ,Colonized (%) W (Min) Absorb W (h) Lose 154 Liu & Rousk — Moss traits that rule cyanobacterial colonization 50 2 2 R = 0.23, P = 0.11 R = 0.42, P = 0.02 R = 0.43, P = 0.02 A. turgidum H. splendens P. schreberi T. nitens Arctic tundra Boreal forest 2 2 R = 0.56, P < 0.01 R < 0.01, P = 0.98 R = 0.23, P = 0.11 24 6 810 30 50 70 0.3 0.4 0.5 0.6 0.7 –2 F (shoots cm ) L (mm) W (mm) Sh Sh L,Base,St Fig. 5 Time for 50 % water absorption (W , hydration rate) and frequency of colonized leaves (F ) in relation to frequency of moss shoots (F ), shoot Absorb L,Colonized Sh length (L ), and basal width of stem leaves (W ). Each coloured dot represents one colony grouped by species identity indicated with different colours. W Sh L,Base,St Absorb and F were measured at colony level, and F , L , and W represent the mean value of three shoots. The grey shading around the regression lines Sh L,Colonized Sh L,Base,St represent 95 % confidence interval of the fitted values. DISCUSSION Bay et al. (2013)’s finding that Polytrichum commune induces hormogonia, the cyanobacterial infectious units, but does not This study presents the first demonstration of the linkage be- host cyanobacteria. This could be attributed to the endohydric tween cyanobacterial colonization and traits of the moss hosts, nature of the moss in which, compared with ectohydric mosses, while the moss could be simply used as a substrate by colon- water is conducted internally and the surface of the moss is izing cyanobacteria. Further, our data supports our first hypoth- water-repellent (Proctor, 2000), thereby preventing cyanobac- esis (H1), stating that N fixation activity in moss–cyanobacteria terial colonization via external water flow along the moss shoot. associations is closely related to the abundance of epiphytic Hence, our findings could explain moisture’s positive effect on cyanobacteria. A strong link between cyanobacterial abundance N fixation in moss–cyanobacteria associations found in earlier and N fixation activity has been found in previous studies (e.g. studies (e.g. Gundale et al., 2012; Rousk et al., 2018). Smith, 1984; Rousk et al., 2013a). This allows an interchange- Our results that moss colonies comprising longer shoots able use of activity and abundance as well as extrapolation from and with wider leaves take more time to reach 50 % hydration one to the other, easing large-scale experiments. (Fig. 5) confirmed the morphological control over hydration of moss colonies (Larson, 1981). We found morphological traits (shoot length and leaf width) related negatively to cyanobac- Linking hydration rate and cyanobacterial colonization terial colonization, and conversely, traits that facilitate hydra- Our results reveal that hydration rate of the moss host, not tion increase cyanobacterial colonization, supporting our third water content per se, controls cyanobacterial colonization (Fig. hypothesis (H3). Shoot length was negatively correlated with 4), corroborating our second hypothesis (H2), that the moss cyanobacterial count and density (Supplementary Data Fig. species with the highest hydration rate hosts the most cyano- S5). This could be the result of lower hydration rates of colonies bacteria. Cyanobacteria can only move in liquid or on moist comprising longer shoots, or the longer distances the cyanobac- surfaces (Brahamsha and Bhaya, 2014) and their mobility is teria need to move to colonize longer shoots, or a combination transient (Bay et al., 2013). Hence, higher hydration rates could of both. Confirming the shoot length effect, we found a lower promote cyanobacterial colonization of moss leaves via passivfrequenc e y of colonized leaves at the top 1- to 2-cm section transport along the moss shoot. This notion could also explain than that for other sections (Supplementary Data Fig. S6). W (min) F (%) Absorb L,Colonized Liu & Rousk — Moss traits that rule cyanobacterial colonization 155 2 2 2 2 R = 0.30, P = 0.01 R = 0.55, P < 0.01 R = 0.44, P < 0.01 R = 0.37, P < 0.01 2 2 2 R = 0.61, P < 0.01 R = 0.49, P < 0.01 R = 0.70, P < 0.01 R = 0.60, P < 0.01 12 3 0.2 0.3 0.4 0.5 0.6 0.7 0.4 0.6 0.8 1.0 0.5 1.0 1.5 L (mm) W (mm) W (mm) A (mm ) L L, Base L, Max L 2 2 Species Stem R = 0.03, P = 0.61 Stem R = 0.36, P = 0.04 2 2 Branch R = 0.61, P = 0.01 Branch R = 0.77, P < 0.01 A. turgidum 1000 H. splendens 100 P. schreberi T. nitens 10 Types Branch Stem 25 75 100 25 75 100 50 50 –1 –1 F (leaves mm shoot) F (leaves mm shoot) L L Fig. 6 Cyanobacterial cells per leaf (CC) and per leaf area (CD) in relation to leaf morphological traits, leaf ), length basal (L width (W ), maximum width L L,Base (W ) and leaf area (A ), as well as the cyanobacterial abundance (CC and CD) in relation to frequency of leav ). Each coloured es (F points represents mean L,Max L L value of one colony. The grey shading around the regression lines represent 95 % confidence interval on the fitted values. Leaf width was negatively correlated with hydration rate, and that moss shoots with higher frequency of leaves host more thus negatively correlated with colonization (Fig. 5). Although cyanobacteria. wider leaves were found to hold more water by enabling leaves The negative relationship between leaf size and colonization to curve in Sphagnum species (Såstad and Flatberg, 1993; is likely driven by a trade-off between leaf size and number. Bengtsson et  al., 2020), the increase in leaf width results in a Leaf size is negatively correlated to leaf number in both vascular lower number of leaves per unit shoot length (Niinemets and plants (Kleiman and Aarssen, 2007) and mosses (Niinemets Tobias, 2019), which reduces water-holding capacity and hy- and Tobias, 2019). We found a strong negative correlation be- dration rate at the shoot level. tween leaf frequency (F) and area (A) (Supplementary Data L L Fig. S4), confirming the leaf size versus number trade-off. The trade-off implies that the smaller the leaf size, the more leaves per unit stem length the plant can carry. Mosses with up Leaf size and number trade-off, and the effect on cyanobacterial −1 to 55 leaves  mm shoot length fall at the extreme end of the colonization size versus number trade-off (Niinemets and Tobias, 2019). In Surprisingly, all measured traits related to leaf size (area, H.  splendens collected from the tundra, we found a range of −1 width etc.) were negatively correlated with cyanobacterial 52–115 leaves mm shoot, which is much higher than leaf fre- colonization, probably related to desiccation rates of larger quencies , from the study above, and an area of 0.23 ± 0.01 mm broader leaves, while leaf frequency was positively related to for branch leaves and 0.48 ± 0.04 mm for stem leaves. In cor - colonization (Fig. 6), supporting our fourth hypothesis (H4), respondence to the extremely high leaf frequency, H. splendens –2 CD (cells mm leaf) CC (cells per leaf) CC (cells per leaf) –1 CD (cells mm leaf) 156 Liu & Rousk — Moss traits that rule cyanobacterial colonization 2 2 R = 0.14, P = 0.13 R = 0.57, P < 0.01 0.1 R = 0.24, P = 0.11 R = 0.01, P = 0.75 A. turgidum H. splendens P. schreberi T. nitens Arctic tundra Boreal forest 5.1 5.3 5.5 5.7 12345 –1 pH Phenol (mg GAE g DW) −1 −1 Fig. 7 Acetylene reduction rate (AR, nmol g dw h ) and frequency of cyanobacteria-colonized leaves (F ) in relation to pH and total phenol content of L,Colonized moss colonies. Each coloured dot represents one moss colony grouped by species identity indicated with different colours and site indicated with shapes. The grey shading around the regression line represent 95 % confidence interval of the fitted values. hosts the most cyanobacteria among the studied mosses (Fig. The cyanobacterial infective units, hormogonia, are of higher 6), and has the highest N fixation activity (Figs 2 and 3). On tolerance than other cyanobacterial cells and could be less af- the other end of the trade-off spectrum, P. schreberi has the lar - fected by phenols while colonizing (Damervet  al al., 1991), but gest leaves, with an area of 0.70 ± 0.05 mm for branch leaves once differentiated to N-fixing heterocytes, phenols can in- and 1.41 ± 0.07 mm for stem leaves, and a low leaf frequency hibit N fixation activity. However, a study of cycad–cyanobac- −1 (33.90 ± 3.78 leaves mm shoot) among the three pleurocarpous teria symbiosis suggested that phenols provide a mechanism mosses. Pleurozium schreberi was the least colonized by cyano- for excluding other microbes and permitting cyanobacteria to bacteria and had the lowest N fixation activity. Tomentypnum grow in cycad roots (Obukowicz et al., 1981). Further study is nitens has slightly smaller leaves (0.64 ± 0.02 mm for branch needed to disentangle the effects of phenolics and other sec- leaves and 1.26 ± 0.09 mm for stem leaves) than P. schreberi ondary compounds on moss and cyanobacteria associations. −1 and a similar leaf frequency of 33.35 ± 4.24 leaves mm shoot The studied mosses, with pH ranges from 5.0 to 6.0, might length. AccordinglyT , .  nitens has a slightly higher cyanobac- limit the activity of cyanobacteria, since the optimum pH for terial colonization rate than P.  schreberi. Although the non-N fixation in mosses is between 5.9 and 6.2 (Smith, 1984). branching acrocarp, A.  turgidum, has the largest leaf area However, neither N fixation nor colonization was signifi- (1.68 ± 0.07 mm ) and lowest leaf frequency (9.12 ± 0.98 leaves cantly affected by pH in our study. The pH of the moss with the −1 mm shoot length) among the studied mosses, colonization rate highest cyanobacterial colonization and activity H.  , splendens, and N fixation activity was still higher than that of P. schreberi, ranged between 5.28 and 5.72, which is similar to the pH of the probably because water balance of non-branching acrocarps deleast colonized moss, - P.  schreberi, whose pH ranged between pend primarily on shoot density (Niinemets and Tobias, 2019). 5.32 and 5.64. This suggests that pH is not the most important Species with an extremely high frequency of small leaves havdeterminant of c e yanobacterial colonization and activity given shoots with large surface area, which increase water absorption the moss pH does not drop below 5. (Larson, 1981), and provide more potential colonization sites for cyanobacteria than those mosses with fewer, larger leaves. The role of ecosystem types We note that a potential weakness of our trait analyses is that Phenol content and pH effects on cyanobacterial colonization they included both inter- and intraspecific variation. We argue that In our experiment, phenols affected cyanobacterial activity trait correlations were mainly driven by interspecific variations. negatively, but not cyanobacterial colonization. This fits with Samples of each species were harvested in the same habitat, and previous findings showing that moss phenol content did not cor- intraspecific trait variations, which are attributed to environmental relate with cyanobacterial colonization (Rousk et  al., 2013a). factors should be small (Roos et al., 2019). The analyses of CVs –1 –1 F (%) L,Colonized AR (nmol g DW h ) Liu & Rousk — Moss traits that rule cyanobacterial colonization 157 P = 0.15 P = 0.01 100 60 P = 0.02 1 0 0 P = 0.67 1.5 1.00 P = 0.16 P = 0.10 0.75 1.0 0.50 0.5 0.25 0 0 0.00 P = 0.69 P = 0.22 10 P < 0.01 0 0 0 Tundra Forest Tundra Forest Tundra Forest −1 −1 Fig. 8 Differences in acetylene reduction rate (AR, nmol g DW  h ), colony structure (F , frequency of shoots and H , colony height), pH, phenol content, Sh Colony nitrogen content ([N]) and water balance-related traits (WC , maximum water content, W , time for 50 % hydration and W , time for 50 % desiccation) of Max Absorb Lose H. splendens between the two sampled sites (arctic tundra and boreal forest). Leaf Paraphyllium 0.1 mm 0.1 mm Top section Others Fig. 9 Cyanobacteria colonization (%) on leaves and paraphyllia of H. splendens. The left figure compares the frequency of colonized leaves with the frequency of colonized paraphyllia. Each coloured dot represents a respective section from one H. splendens shoot; bars show the mean ± s.e. (n = 9). Frequency of cyano- bacterial colonization was significantly higher on leaves than on paraphyllium (P = 0.003, two-way ANOVA). The picture on the right shows cyanobacteria (bright orange dots) on H. splendens leaves and paraphyllia, and the picture in the top-right corner shows cyanobacteria attached to a paraphyllium. Photos were taken under an Olympus BX61. Frequency of colonization (%) WC (%) –1 –1 Max pH AR (nmol g DW h ) –1 Phenol (mg GAE g DW) W (min) Absorb –2 F (shoots cm ) sh [N] (%) H (mm) W (h) Colony Lose 158 Liu & Rousk — Moss traits that rule cyanobacterial colonization of traits confirmed that intraspecific variations of traits were gen- paraphyllia could promote N uptake by the moss host by cre- erally smaller than interspecific variations (Supplementary Data ating a link between epiph ytic cyanobacteria and the moss stem. Table S1). For instance, the intraspecific CV of P. schreberi leaf These open questions and unknowns call for further research on frequency is significantly smaller than interspecific CV (0.12 the paraphyllium’s role in the mosses’ N uptake, which could versus 0.82, P  =  0.04, modified signed-likelihood ratio test). act as a lens through which to resolve the link between traits Moreover, to demonstrate the intraspecific variation of traits and and ecological functions. N fixation rate, we compared several traits of H. splendens be- In conclusion, our findings emphasize that the hydration rate tween ecosystem types (subarctic tundra versus boreal forest). of a moss colony is a key trait regulating cyanobacterial colon- Hylocomium splendens colonies from the tundra were shorter ization. On the other hand, species-specific morphological traits and had more shoots per area than the colonies from the forest control hydration rate. Chemical traits of the moss host seem (Fig. 7). These differences between ecosystems are likely drivento be less important than morphological traits in re gulating by environmental factors, such as temperature, moisture (Bisbee cyanobacterial colonization. We could also demonstrate that et  al., 2001) and nutrient availability, as the tundra has a lower a considerable portion of c yanobacteria colonize paraphyllia, mean annual temperature (0.2 versus 1  °C) and precipitationa pre viously overlooked structure in terms of cyanobacterial (337 versus 570  mm) than the forest site (Rousk et  al., 2014; colonization. Variation in moss species-specific traits drives Rousk and Michelsen, 2017), impacting moss growth. Shoot cyanobacterial colonization, but intraspecific variation that af- size controls water fluxes in moss colonies (Elumeeva et  al., fects colonization is in turn driven by environmental factors. 2011) (see section: Linking hydration rate and cyanobacterialGi ven that mosses are a key source of N to ecosystems where colonization), and can thereby control cyanobacterial coloniza-they dominate the ground cover, uncovering the relation be- tion. Fine-scale variations in colony structure, such as changestween moss traits and c yanobacterial colonization will ultim- in shoot frequency, alter surface roughness and further interact ately result in a better estimation of the amount of N input – as with wind flow affecting boundary-layer properties and desic- dependent on moss traits – and can provide new perspectives cation rate (Rice and Schneider, 2004Michel ; et al., 2012). This and information for understanding the relationship between is confirmed by the distinct desiccation rate for H.  splendens mosses and cyanobacteria. colony between tundra and forest sites (Fig. 7). It is possible that trait variation, driven by ecosystem specific factors (e.g. N avail- ability, temperature), are responsible for intraspecific differences SUPPLEMENTARY DATA in cyanobacterial colonization and thus N fixation, while relative Supplementary data are available online at https://academic. interspecific variation should remain similar in ecosystems with oup.com/aob and consist of the following. Figure S1: pictures different abiotic conditions. of the studied moss species. Figure S2: change in moss colony We did not assess cyanobacterial colonization for H. splendens weight over time for four moss species during water absorp- from the forest site, but given that N fixation activity is closely tion and loss. Figure S3: acetylene reduction rate in relation to linked to cyanobacterial colonization, and activity was not dif- the cyanobacteria count and density on moss leaves. Figure S4: ferent between the sites, it is likely that colonization is similar relationship between leaf area and frequency of leaves for in- between the sites, too (Fig. 8). Moreover, in contrast with dividual moss shoots. Figure S5: relationships between cyano- H. splendens collected from the forest site, P. schreberi had still −1 −1 bacterial colonization and shoot traits. Figure S6: differences in lower N fixation activity (27.75 versus 0.68 nmol g  DW h , frequency of colonized leaves between top segments and lower Fig. 3). Hence, the variation in cyanobacterial colonization and, segments. Figure S7: relationships between cyanobacterial col- thus N fixation activity, should be largely controlled by species- onization and leaf size. Table S1: intra- and interspecific CVs of specific traits, but intraspecific variation that affects colonization water balance, colony, chemical and morphological traits. is also in turn driven by the environment. FUNDING Morphological peculiarity of H. splendens Xin Liu was sponsored by the Second Tibetan Plateau Scientific Our results present the first evidence that cyanobacterial col- Expedition and Research program (STEP) (2019QZKK0301), onization of paraphyllia can be substantial. About 20–30 % of the National Natural Science Foundation of China (No. paraphyllia were colonized by cyanobacteria (Fig. 9). Given 31600357) and a visiting scholarship from Chinese Academy the large number of paraphyllia along the moss shoot, asso- of Sciences. Funding was also provided by the Independent ciated cyanobacteria can account for a considerable portion Research Fund Denmark, Sapere Aude Grant to K.R. (grant id: of N fixation. Hence, focusing on moss leaves as coloniza- 7027-00011B). tion sites could vastly underestimate cyanobacterial coloniza- tion along moss shoots. Paraphyllia occur in taxa of only a few moss families, such as Hylocomiaceae, Thuidiaceae and ACKNOWLEDGEMENTS Brachytheciaceae (Spirina et al., 2020). If those taxa that have paraphyllia host more cyanobacteria than taxa without these We thank F. Ekelund for help with microscopic analyses, A.T. structures is an open question. Moreover, paraphyllia, which F. Permin for help with the ethylene analyses, and E. Nielsen are photosynthetic filaments or leaf-like structures composed and G.  Sylvester for assistance with the chemical analyses. of live cells, are located between leaf and stem, where man X.L.  y and K.R.  designed the experiment. K.R.  collected the N -fixing cyanobacteria occur. Although these thick-walled samples. X.L. performed the laboratory experiment. X.L. ana- cells may have low nutrient exchange rates, it is possible that lysed the data. Both authors wrote the paper. The authors Liu & Rousk — Moss traits that rule cyanobacterial colonization 159 Meeks JC, Elhai J. 2002. Regulation of cellular differentiation in filamentous declare that they have no conflicts of interest with the contents cyanobacteria in free-living and plant-associated symbiotic growth states. of this article. Microbiology and Molecular Biology Reviews 66: 94–121. Michel  P, Lee  WG, During  HJ, Cornelissen  JHC. 2012. Species traits and their non-additive interactions control the water economy of bryophyte cushions. Journal of Ecology 100: 222–231. LITERATURE CITED Niinemets Ü, Tobias M. 2019. Canopy leaf area index at its higher end: dissec- Adams DG, Duggan PS. 2008. Cyanobacteria–bryophyte symbioses. Journal tion of structural controls from leaf to canopy scales in bryophytes. New of Experimental Botany 59: 1047–1058. Phytologist 223: 118–133. Alvarenga DO, Rousk K. 2021. Indirect effects of climate change inhibit N Obukowicz  M, Schaller  M, Kennedy  GS. 1981. Ultrastructure and phen- fixation associated with the feathermoss Hylocomium splendens in sub- olic histochemistry of the Cycas revoluta-Anabaena symbiosis. New arctic tundra. Science of The Total Environment 795: 148676. Phytologist 87: 751–759. Bauer  IE, Tirlea  D, Bhatti  JS, Errington  RC. 2007. Environmental and Proctor MCF. 1982. Physiological ecology: water relations, light and tempera- biotic controls on bryophyte productivity along forest to peatland eco- ture responses, carbon balance. In: Smith  AJE, ed. Bryophyte ecology. tones. Canadian Journal of Botany-Revue Canadienne De Botanique 85: Dordrecht: Springer, 333–381. 463–475. Proctor  MCF. 2000. Physiological ecology. In: Shaw  AJ, Goffinet  B, eds. Bay G, Nahar N, Oubre M, et al. 2013. Boreal feather mosses secrete chem- Bryophyte biology. Cambridge: Cambridge University Press, 225–247. ical signals to gain nitrogen. New Phytologist 200: 54–60. R Core Team. 2019. R: A language and environment for statistical computing. Bengtsson F, Granath G, Cronberg N, Rydin H. 2020. Mechanisms behind Vienna, Austria: R Foundation for Statistical Computing. species-specific water economy responses to water level drawdown in peat Renaudin  M, Darnajoux  R, Bellenger  JP. 2021. Quantification of moss- mosses. Annals of Botany 126: 219–230. associated cyanobacteria using phycocyanin pigment extraction. Frontiers Bisbee K, Gower S, Norman J, Nordheim E. 2001. Environmental controls in Microbiology 11: 611792. on ground cover species composition and productivity in a boreal black Rice  SK, Schneider  N. 2004. Cushion size, surface roughness, and the con- spruce forest. Oecologia 129: 261–270. trol of water balance and carbon flux in the cushion moss Leucobryum Brahamsha  B, Bhaya  D. 2014. Motility in unicellular and filamentous glaucum (Leucobryaceae). American Journal of Botany 91: 1164–1172. Cyanobacteria. In: Flores E, Herrero A. eds. The cell biology of cyanobac- Roos  RE, van  Zuijlen  K, Birkemoe  T, et  al. 2019. Contrasting drivers of teria. Seville: Caister Academic Press, 233–262. community-level trait variation for vascular plants, lichens and bryophytes Broady PA. 1979. The Signy Island terrestrial reference sites: IX. The ecology across an elevational gradient. Functional Ecolo33 gy : 2430– 2446. of the algae of site 2, a moss carpet. British Antarctic Survey Bulletin 47 : Rousk  K, Michelsen  A. 2017. Ecosystem nitrogen fixation throughout the 13–29. snow-free period in subarctic tundra: effects of willow and birch litter add- Cruz de Carvalho R, Maurício A, Pereira MF, da Silva JM, Branquinho C. ition and warming. Global Change Biology 23: 1552–1563. 2019. All for one: The role of colony morphology in bryophyte desicca-Rousk J, Brookes PC, Bååth E. 2009. Contrasting soil pH effects on fungal tion tolerance. Frontiers in Plant Science 10: 1360. and bacterial growth suggest functional redundancy in carbon mineraliza- Dalton  DA, Chatfield  JM. 1985. A new nitrogen–fixing cyanophyte-hepatic tion. Applied and Environmental Microbiology 75: 1589–1596. association: Nostoc and Porella. American Journal of Botany 72: 781–784. Rousk  K, DeLuca  TH, Rousk  J. 2013a. The cyanobacterial role in the re- Damerval T, Guglielmi G, Houmard J, De Marsac NT. 1991. Hormogonium sistance of feather mosses to decomposition—Toward a new hypothesis. differentiation in the cyanobacterium Calothrix: A photoregulated devel- PLoS ONE 8: e62058. opmental process. The Plant Cell 3: 191–201. Rousk  K, Rousk  J, Jones  DL, Zackrisson  O, DeLuca  TH. 2013b. Feather Elumeeva  TG, Soudzilovskaia  NA, During  HJ, Cornelissen  JHC. 2011. moss nitrogen acquisition across natural fertility gradients in boreal for - The importance of colony structure versus shoot morphology for the water ests. Soil Biology & Biochemistry 61: 86–95. balance of 22 subarctic bryophyte species. Journal of Vegetation Science Rousk K, Jones DL, DeLuca TH. 2014. Moss-nitrogen input to boreal forest 22: 152–164. soils: Tracking N in a field experiment. Soil Biology & Biochemistry 72: Erickson M, Miksche GE. 1974. On the occurrence of lignin or polyphenols 100–104. in some mosses and liverworts. Phytochemistry 13: 2295–2299. Rousk  K, Sorensen  PL, Lett  S, Michelsen  A. 2015. Across-habitat com- Gundale  MJ, Deluca  TH, Nordin  A. 2011. Bryophytes attenuate anthropo- parison of diazotroph activity in the subarctic. Microbial Ecology, 69: genic nitrogen inputs in boreal forests. Global Change Biology17 : 778–787. 2743–2753. Rousk  K, Degboe  J, Michelsen  A, Bradley  R, Bellenger  JP. 2017. Gundale  MJ, Wardle  DA, Nilsson  MC. 2012. The effect of altered Molybdenum and phosphorus limitation of moss-associated nitrogen fix- macroclimate on N-fixation by boreal feather mosses. Biology Letter8 s : ation in boreal ecosystems. New Phytologist 214: 97–107. 805–808. Rousk K, Sorensen PL, Michelsen A. 2018. What drives biological nitrogen Jackson BG, Martin P, Nilsson MC, Wardle DA. 2011. Response of feather fixation in high arctic tundra: Moisture or temperature? Ecosphere 9: moss associated N2 fixation and litter decomposition to variations in simu- e02117. lated rainfall intensity and frequency. Oikos 120: 570–581. Santi  C, Bogusz  D, Franche  C. 2013. Biological nitrogen fixation in non- Jean  M, Holland-Moritz  H, Melvin  AM, Johnstone  JF, Mack  MC. 2020. legume plants. Annals of Botany 111: 743–767. Experimental assessment of tree canopy and leaf litter controls on Såstad  SM, Flatberg  KI. 1993. Leaf morphology of Sphagnum strictum in the microbiome and nitrogen fixation rates of two boreal mosses. New norway, related to habitat characteristics. Lindbergia 18: 71–77. Phytologist 227: 1335–1349. Smith VR. 1984. Effects of abiotic factors on acetylene reduction by cyanobac- Kleiman  D, Aarssen  LW. 2007. The leaf size/number trade-off in trees. teria epiphytic on moss at a subantarctic island. Applied and Environmental Journal of Ecology 95: 376–382. Microbiology 48: 594–600. Lang SI, Cornelissen JHC, Klahn T, et al. 2009. An experimental comparison Solheim B, Endal A, Vigstad H. 1996. Nitrogen fixation in Arctic vegetation of chemical traits and litter decomposition rates in a diverse range of sub- and soils from Svalbard, Norway. Polar Biology 16: 35–40. arctic bryophyte, lichen and vascular plant species. Journal of Ecology Solheim  B, Zielke  M. 2002. Associations between cyanobacteria and 97: 886–900. mosses. In: Rai AN, Bergman B, Rasmussen U, eds. Cyanobacteria in Larson DW. 1981. Differential wetting in some lichens and mosses: The role Symbiosis. Dordrecht: Springer, 137–152. of morphology. The Bryologist 84: 1–15. Spirina UN, Voronkova TV, Ignatov MS. 2020. Are all paraphyllia the same? LeBauer DS, Treseder KK. 2008. Nitrogen limitation of net primary product- Frontiers in Plant Science 11: 858. ivity in terrestrial ecosystems is globally distributed. Ecolo 89 gy : 371–379. Stuart JEM, Holland-Moritz H, Lewis LR, et al. 2021. Host identity as a driver Leppänen  SM, Salemaa  M, Smolander  A, Mäkipää  R, Tiirola  M. 2013. of moss-associated N2 fixation rates in Alaska. Ecosystems24 : 530–547. Nitrogen fixation and methanotrophy in forest mosses along a N depos- Wang  YP, Law  RM, Pak  B. 2010. A global model of carbon, nitrogen ition gradient. Environmental and Experimental Botany 90: 62–69. and phosphorus cycles for the terrestrial biosphere. Biogeosciences 7: Marwick  B, Krishnamoorthy  K. 2019. cvequality: tests for the equality 2261–2282. of coefficients of variation from multiple groups. R software package Wilde  A, Mullineaux  CW. 2015. Motility in cyanobacteria: polysaccharide version 0.1.3. tracks and Type IV pilus motors. Molecular Microbiology 98: 998–1001.

Journal

Annals of BotanyOxford University Press

Published: Oct 10, 2021

Keywords: Bryophytes; moss colony; Cyanobacteria; nitrogen fixation; water retention; functional trait

There are no references for this article.