Dispersal Ecology of Desert Mosses Along Gradients of Elevation, Wildfire Disturbance and Local Niche

Propagule soil-banks are important for colonization and for the maintenance of regional diversity among bryophytes of temperate regions, yet they have not previously been reported from hot desert ecosystems and little is known about the relationship between soil-bank diversity and elevation. I used emergence germination methods to explore patterns of species composition and richness in bryophyte soil-banks along a 1400 m elevational gradient spanning three major vegetation zones in the Mojave Desert of the southwestern United States. Climate variables related to water availability (mean annual precipitation, relative humidity and soil moisture) were significantly and positively correlated with elevation. A total of 17 bryophyte taxa were present in soils collected at twelve sites. Community compositions shifted with increasing elevation, suggesting that soil-banks along the gradient are an important “cryptic” component of the regional species pool. Bayesian linear regressions revealed that three measures of community diversity were positively associated with elevation: species richness, Shannon entropy and phylogenetically-controlled species richness. Positive diversity–elevation trends in soil-banks of the Mojave Desert are the likely product of increasing moisture availability and reproductive frequency at higher elevations in dry mountain ranges. Introduction In deserts of North America, bryophyte diversity reflects the ability of individuals to reproduce, disperse and establish in the face of challenging abiotic conditions. Successful colonization results from at least one of three processes: 1) dispersal over distance; 2) soil-bank dormancy; and 3) vegetative expansion of existing plants (During 1979). Soil-banks are of 7 particular importance, not only because they function as reservoirs of taxonomic and genetic diversity for bryophytes (Hock et al. 2008), but also because they allow plants to avoid risks associated with temporal fluctuations in stress and resource availability (During 1997). Organs and tissues from which a new plant can arise (hereafter referred to as “propagules”) can remain dormant and viable for nearly 50 years in dry soils (Whitehouse 1984), suggesting that soil-banks are likely to exist in deserts even in locations where surface plants may be absent. Assuming the persistence of propagules in soils of arid localities, soil-banks could be a key mechanism of “dispersal in time” which permits desert bryophytes to avoid harsh abiotic conditions and conditionally establish during favorable periods. Despite the significant role of soil crust bryophytes for nutrient dynamics, hydrological cycling and soil stabilization in arid ecosystems (Eldridge and Greene 1994), we know little of how soil-banks might contribute to bryophyte diversity either locally or across arid landscapes. Elsewhere, bryophyte soil-bank diversity has been described from a variety of mesic habitats (reviewed by During 1997). Only a single study has investigated soil-banks from a semi-arid region (a dry savanna in Zimbabwe), where soil communities exhibited surprisingly high diversity compared to surface communities (During 2007) and featured one moss genus that was new to science (Zander and During 1999). These findings suggest that bryophyte soil-banks in desert shrublands might similarly harbor cryptic taxonomic richness. The association of species richness with elevation is a fundamental concern of both plant ecology (Whittaker 1960) and biogeography (Lomolino 2008). In bryophyte communities surveyed along elevational gradients, species richness can vary in a positive (Austrheim 2002), negative (Geissler and Velluti 1996) or unimodal fashion (Wolf 1993), trends which depend on the length of the elevational gradient surveyed (Nogués-Bravo et al. 2008). Elevation–richness patterns of bryophytes are typically associated with moisture gradients that covary with elevation (Lee and La Roi 1979), reflecting the fact that bryophytes are highly dependent on external water for growth and reproduction. In deserts of the southwestern United States, persistent desiccation 8 and water limitation restrict reproductive frequency, particularly at low elevations (Stark and Castetter 1987; Benassi et al. 2011). In the absence of successful propagule production and dispersal, meager deposition rates are also expected to be reflected by minimal species richness in low elevation soil-banks. While many low elevations (<1000 m) are typified by low species richness of surface bryophyte communities (Nash et al. 1977), the mechanisms maintaining diversity in sub-surface soil-banks may operate differently than in surface communities, and it remains to be seen if elevation–richness patterns are consistent between each profile. The purposes of this study were to characterize the diversity of bryophyte soil-bank communities in the Mojave Desert and to quantify changes in sub-surface diversity along an elevational gradient of 1400 m. Because bryophyte reproduction in the region increases with elevation and increasing moisture availability, I hypothesized that three separate measures of soilbank diversity would be positively related to elevation. Although similar approaches have been applied to other plant communities, to my knowledge this study represents the first to investigate a soil-bank community of bryophytes in a truly arid temperate region, and is the first to explicitly test the elevation–richness relationship for any bryophyte soil-bank. Materials and Methods Study area and field collections The Spring Mountains are a dry mountain range in the Mojave Desert, approximately 40 km northwest of the city of Las Vegas in southern Nevada, U.S.A. (36.006°, –115.452°). On the eastern slopes of the Spring Mountains (Fig. 2.1), limestone outcrops and wooded foothills descend to sloping alluvial fans composed of shallow, well-drained, loamy-skeletal soils (Lato 2006). Much of the terrain is dissected by dry washes created during episodic rainfall events. Annual precipitation in the area is sparse and highly variable (mean = 283 mm; SD ± 127 mm; Fig. 2.2), mostly provided by winter rain-showers and occasional late-summer showers (National Climatic Data Center records for Spring Mountain State Park, 1152 m a.s.l., period 1977–2011). 9 At low elevations near 1000 m, local vegetation was dominated by the perennial evergreen shrub Larrea tridentata (DC.) Coville. Middle elevations (1200–1800 m) were characterized by the shrub Coleogyne ramosissima Torr., which intergraded at its upper limits with the small trees Pinus monophylla Torr. & Frém. and Juniperus osteosperma (Torr.) Little. Upper elevations (> 2000 m) were dominated by tall coniferous trees, mainly Pinus ponderosa Lawson & C. Lawson. In late September and early October of 2011, I surveyed twelve sites (Table 2.1) arrayed along two parallel elevational gradients ascending the eastern slopes of the Spring Mountains (Fig. 2.3). The first transect (Kyle Canyon) included five sites spanning 1664–2235 m, while the second transect (Lee Canyon) included seven sites between 1019–2442 m. Sites were crudely spaced at intervals of ~200 m. The presence of mature vegetation and intact substrates at all sites indicated the absence of any recent fires or mechanical disturbances. I collected 4 soil cores (10 cm diameter × 3 cm depth) at each site from patches that were free of any large rocks, litter or surface plants (including existing patches of bryophyte shoots). The 48 soil cores were sealed and immediately transported to the lab at room temperature in dry darkness before germination. Some cores were breached during transport and were discarded, yielding a total of 40 cores. For germination, I air-dried and sieved each soil sample (355 μm fine wire mesh) to remove small rocks, coarse organic matter and any remnant plant parts, then placed a 4 cm 3 subsample into sterile polystyrene culture dishes (35 mm diameter, Falcon item 351008, BectonDickinson, Franklin Lakes, NJ, USA). Cultures were moistened, loosely covered and maintained in a humid environment within a growth chamber (Percival model E30B, Boone, IA, USA) under a cycle of 12 hr light (20°C) and 12 hr dark (8°C). I also randomized two sterile soil samples among the culture dishes to detect unwanted aerial spore contamination. Soil cultures were monitored twice weekly for moisture and germination; any vascular plant germinants were immediately removed. Preliminary observations suggested that growth of microorganisms (e.g., fungi, cyanobacteria, algae) in non-sterile soil had only a nominal effect on moss germination rates compared to sterile soil. Soil cultures were grown for 5 months to allow all propagules to 10 germinate to identifiable stages, and at monthly intervals I recorded the presence and identity of bryophyte species in each sample. Bryophyte nomenclature follows Norris and Shevock (2004), with the exception of the family Bryaceae, which follows the recent revisions of Spence (2011). Statistical analyses All analyses were performed in the R computing environment (version 2.15.0; R Core Development Team 2012) using supplemental packages as noted. To determine how climatic variables were associated with elevation, I obtained interpolated values for mean annual precipitation, temperature, relative humidity and soil moisture for each of the twelve field sites from the FetchClimate web service (http://fetchclimate.cloudapp.net/), and then calculated pairwise Spearman correlation coefficients (significant if P ≤ 0.05) using the packages Hmisc and corrplot. The FetchClimate web service generates values based solely on climate data observations (National Oceanic and Atmospheric Administration records, 1948–2012, http://www.esrl.noaa.gov/psd/) without regard to elevation information. For the species data, I assessed species compositions by calculating ordination site scores with non-metric multidimensional scaling (NMS; Kruskal 1964) based on Bray-Curtis distances and implemented in the vegan package. A two-dimensional solution was sought a priori from a maximum of 999 random starting configurations. Composition–elevation trends were assessed visually by plotting site scores in the ordination space according to elevation. To assess the relationship between diversity and elevation, I first calculated three distinct measures of diversity for each of the 12 sites. The first diversity measure, species richness (SR), was the number of species in the pooled collection of soil cores at each site. The second measure was Shannon entropy (H′), calculated for each site according to Oksanen et al. (2011). This measure integrates the number of species at each site with the evenness in the abundances among species, where abundances were considered as the frequency of each species among each of the soil cores per site. The third measure, phylogenetic species richness (PSR) was essentially an estimate of the effective species richness of a site after accounting for taxonomic relatedness. 11 PSR was calculated in the picante package as the number of species at a given site multiplied by the phylogenetic distance (“branch length” in a phylogenetic tree) between those species (Helmus et al. 2007). Therefore, related species give lower values for a site, and distantly related species give higher values, ranging from a minimum of zero to a maximum equal to the number of species present at a given site. The tree upon which calculations were performed was manually constructed using taxonomic distinctions defined by the Bryophyte Names Authority List (http://www.mobot.org/MOBOT/tropicos/most/bryolist.shtml) of the Missouri Botanical Garden. To detect diversity–elevation trends, I conducted separate Bayesian linear regressions for each of the three diversity response variables using the package MHadaptive (Chivers 2012). A Bayesian approach allows one to make a probability statement about the parameter of interest (in this case, the slope of the diversity–elevation trendline) given only the data at hand, rather than a statement about how frequently those data would be expected under a given null hypothesis (Gelman et al. 2004). Therefore, small sample sizes (in this study, N=12) are not a constraint, provided that observations are considered independent and exchangeable. Bayesian methods are based on the joint probability of observing the data given a hypothesis (calculated using a likelihood function) and the prior probability distribution of the hypothesis (the “prior”); this joint probability is proportional to the conditional probability distribution (the “posterior”) for the parameter of interest, which is effectually the desired outcome. In this study, I used flat, uninformative priors, assumed normally-distributed errors, and sampled the posterior probability distribution with a Metropolis-Hastings algorithm (Chivers 2012) employing 10,000 iterations, of which the first 2,000 iterations (burn-in) were discarded. Results Among twelve sites on the eastern slope of the Spring Mountains, elevation was significantly and positively correlated with mean annual precipitation (Spearman correlation coefficient = 0.60), mean annual soil moisture (0.85) and mean annual relative air humidity 12 (0.85); elevation was negatively correlated with mean annual air temperature (–0.74). A total of 17 bryophyte species (and one fern) were present in the entire collection of soil samples (Table 2.2). Of all bryophyte species, 58% represented the taxonomic family Pottiaceae, Bryaceae 17%, Funariaceae 11%, Ditrichaceae 5% and Encalyptaceae 5%. There were few widespread species, and most species were present in only a few sites (Table 2.3). External contamination in the growth chamber was not detected by the sterile controls. The NMS ordination yielded two convergent solutions after ten random starts, with a final stress of 10.9%. The topology of NMS site scores plotted in species space indicated an association between species composition and elevation (Fig. 2.4). For example, communities at lower elevation sites were associated with xeric species including Syntrichia caninervis Mitt. and Crossidium aberrans Holz. & E.B. Bartram, while mesic species such as Ceratodon purpureus (Hedw.) Brid. were present at higher elevations (Table 2.3). The Bayesian regressions revealed that all three diversity measures (SR, H′, PSR) were positively associated with elevation (Table 2.1). Ninety-five percent Bayesian credible intervals for the regression slope parameter (β) did not include zero, consistent with a positive diversity–elevation hypothesis (Fig 2.5). Posterior means for each slope parameter were: SR = 2.47 species; H′ = 0.597; and PSR = 1.82 species per change in 1000 m. Discussion This study is the first to characterize species composition and diversity of bryophyte soilbanks in the driest desert of the southwestern United States, and the first to connect trends in soilbank diversity to elevation. The most noteworthy finding was that soil-bank diversity in the Mojave Desert was positively associated with elevation. Elevation was itself significantly correlated with climatic variables including available moisture and temperature, indicating that elevation can be a useful indicator of the stresses which bryophytes experience. Moisture and elevation are tightly linked in desert mountain ranges of the southwestern United States, providing strong gradients which influence plant growth and reproduction (Smith et al. 1997). 13 Studies in the region have shown fidelity between elevation and bryophyte species compositions of surface communities, which change along gradients corresponding to vegetation and climate. For example, Clark (2012) documented compositional changes across 1000 m of vertical relief spanning three vegetation zones at Grand Canyon National Park in northern Arizona, revealing that acrocarpic, xeric mosses of the Pottiaceae and Grimmiaceae families predominated at lower elevations and were superseded by mesic mosses (plus several liverworts) at higher elevations. I observed very similar patterns in soil-banks of the Spring Mountains, where the Pottiaceae family was well represented, although the saxicolous family Grimmiaceae was conspicuously absent from soil profiles, perhaps because its members require lithic substrates for germination (Keever 1957). Outside of the southwestern United States, Stehn et al. (2010) and Slack (1977) have observed similarly changing species compositions in montane surface communities of the eastern United States, supporting the generality of the concept of compositional turnover along montane elevational gradients. Like species compositions, area-wide species richness in bryophyte soil-banks of the Spring Mountains (17 taxa) corresponded well with the generally low species numbers found in the surface vegetation of other Mojave Desert sites. For example, Thompson et al. (2005) encountered 11 taxa (4 shared in common with the current study) in surface communities at other Spring Mountain locations between 1030 – 1440 m, Bowker et al. (2000) reported 6 taxa (5 shared) at 1494 m, and Nash et al. (1977) reported just 3 species (3 shared) from 1000 m, perhaps indicative of moisture limitations at lower elevations. The moisture limitation concept was supported by my finding strong positive correlations between elevation and indicators of moisture availability which included annual precipitation, soil moisture, and relative humidity. Curiously, there was no strong association between elevation and the seed-bank of vascular plants at similar locations in the Spring Mountains (Abella and Springer 2012), perhaps indicating that water availability has quite different effects depending on the extent to which different plant groups rely on external water. 14 The generally positive elevation–diversity trend observed among bryophyte soil-banks in this study appears to parallel the positive trends observed at moderate elevations both in the Mojave Desert and elsewhere in the world. For example, Bisang et al. (2003) suggested that soilbank richness in a tropical rainforest may increase with elevation. However, above approximately 2000 m, bryophyte species richness in surface communities apparently declines, probably as a result of increasing stresses (e.g., moisture and temperature stresses) in some alpine zones (Geissler and Velluti 1996). Studies of very broad elevational gradients in excess of 3000 m (e.g., Wolf 1993; Grau et al. 2007) often reveal a hump-shaped, unimodal relationship for bryophyte species richness in surface vegetation, with low-elevation increases eventually peaking and tapering in high montane and alpine areas. Had the current study been extended through the conifer forest and above upper treeline into the alpine zone, a similar unimodal peak and decline might have been expected for soil-bank species. Despite covering three major vegetation–climate zones, the areas surveyed herein covered only the lower portion of a mountain range that extends upward to 3633 m. Observing only a portion of a larger elevation gradient could influence the apparent shape of the elevation–richness pattern (McCain and Grytnes 2010), which may explain the monotonic positive relationship which I observed in soil-banks of the Spring Mountains. Bryophyte diversity at landscape scales is perhaps best viewed in a context where local communities are sustained by dispersal from the regional species pool (During and Lloret 2001; Leibold et al. 2004). In this regard, local richness and compositions are the product of immigration events which are then filtered by the capacity of bryophytes to tolerate local conditions. Like the “seed rain” of many vascular plants, the deposition of bryophyte spores (Miles and Longton 1992) and asexual propagules (Pohjamo et al. 2006) is expected to be greatest immediately around source plants, and declines rapidly with distance. This is true from scales of several dozens of meters (Lönnell et al. 2012) to several hundreds of kilometers (Sundberg 2012), suggesting that proximity to reproductive plants may be a strong determinant of the density and composition of spores that comprise soil-banks (but see Hylander 2009). I 15 hypothesize that the accrual of propagules in soil-banks is the result of proximity to propaguleproducing plants at higher elevations. Reproductive success increases with moisture and elevation in southwestern USA deserts (Stark and Castetter 1987), which suggests that soil-banks should be richest at higher elevations where productive propagule-producing plants occur more frequently. Conversely, low-elevation sites far from productive plants are expected to yield fewer soil-bank species, not necessarily because propagules cannot survive there, but because their arrival would require infrequent long-distance dispersal events originating from distant montane sources. Although this study did not directly measure distance-dependency, one way to verify these hypotheses would be to survey soil-bank richness (in tandem with spore deposition rates) at varying distances from known reproductive populations. To summarize, this study documented 17 species of bryophytes in sub-surface soil-banks along an elevational gradient covering 1400 m in a temperate arid mountain range. The positive relationships of species richness, Shannon entropy, and phylogenetically-controlled species richness with increasing elevation indicate that future work should attempt to connect soil-bank diversity with distance from reproductive surface communities along the elevation gradient. The positive diversity–elevation trend present in the Mojave Desert suggests that similar patterns may exist in bryophyte soil-banks of other temperate arid regions where moisture and temperature influences are closely associated with changes in elevation.