Variation in species‐level plant functional traits over wetland indicator status categories

Abstract Wetland indicator status (WIS) describes the habitat affinity of plant species and is used in wetland delineations and resource inventories. Understanding how species‐level functional traits vary across WIS categories may improve designations, elucidate mechanisms of adaptation, and explain habitat optima and niche. We investigated differences in species‐level traits of riparian flora across WIS categories, extending their application to indicate hydrologic habitat. We measured or compiled data on specific leaf area (SLA), stem specific gravity (SSG), seed mass, and mature height of 110 plant species that occur along the Colorado River in Grand Canyon, Arizona. Additionally, we measured leaf δ13C, δ15N, % carbon, % nitrogen, and C/N ratio of 56 species with C3 photosynthesis. We asked the following: (i) How do species‐level traits vary over WIS categories? (ii) Does the pattern differ between herbaceous and woody species? (iii) How well do multivariate traits define WIS categories? (iv) Which traits are correlated? The largest trait differences among WIS categories for herbaceous species occurred for SSG, seed mass, % leaf carbon and height, and for woody species occurred for height, SSG, and δ13C. SSG increased and height decreased with habitat aridity for both woody and herbaceous species. The δ13C and hence water use efficiency of woody species increased with habitat aridity. Water use efficiency of herbaceous species increased with habitat aridity via greater occurrence of C4 grasses. Multivariate trait assemblages differed among WIS categories. Over all species, SLA was correlated with height, δ13C, % leaf N, and C/N; height was correlated with SSG and % leaf C; SSG was correlated with % leaf C. Adaptations of both herbaceous and woody riparian species to wet, frequently inundated habitats include low‐density stem tissue. Adaptations to drier habitats in the riparian zone include short, high‐density cavitation‐resistant stem tissue, and high water use efficiency. The results enhance understanding about using traits to describe plant habitat in riparian systems.


| INTRODUCTION
The use of plant functional traits in ecology aims to elucidate links between species' characteristics, habitat, distribution, and community assembly (Grime, 1977;Poff, 1997;Ackerly, Knight, Weiss, Barton, & Starmer, 2002;Merritt, Scott, Poff, Auble, & Lytle, 2010). This use is founded on the theory that taxonomically unrelated species show convergent adaptations to similar environmental conditions and assumes that the distributions of plant traits over environmental gradients are predictable and explain the mechanisms defining the niche of a species (Keddy, 1992;Reich, Walters, & Ellsworth, 1997;Reich et al., 2003;Shipley et al., 2016). A growing number of studies report consistent trends in interspecific variation in traits across environmental gradients at regional to global scales (Chave et al., 2009;Kattge et al., 2011;Wright et al., 2004) and have provided support to plant strategy theory by elucidating trade-offs in traits for establishment and persistence in different environments (Grime, 1977;Lavorel & Garnier, 2002;Westoby & Wright, 2006). However, the extent to which relationships between functional traits and environmental gradients apply to specific regional and local contexts warrants further exploration (Shipley et al., 2016).
Riparian ecosystems in arid regions are ideal for investigating species-level trait variation across habitats to understand adaptation to inundation, flooding, and aridity. These ecosystems are highly species diverse relative to upland plant communities and include sharp spatial gradients of inundation, fluvial disturbance, and water availability across relatively short distances within a river valley (Horton, Kolb, & Hart, 2001;Naiman, Decamps, & Pollock, 1993;Shafroth, Stromberg, & Patten, 2002). Because hydrologic and geomorphic characteristics define the dominant disturbance regime of riparian ecosystems (i.e., flooding), vegetation communities adjacent to rivers are often arranged in elevational bands along gradients of disturbance and moisture availability (Nilsson & Svedmark, 2002;Sluis & Tandarich, 2004).
Riparian plants in these communities display a range of morphological and physiological adaptations that influence their growth, reproduction, water balance, and survival in conditions ranging from drought to inundation and prolonged submergence (Catford & Jansson, 2014;Lytle & Poff, 2004).
In the United States, a national wetland plant list was established in 1988 as part of a wetland classification system widely used in resource inventories, wetland delineations, and research (Cowardin, Carter, Golet, & LaRoe, 1979;Reed, 1988). The plant list is periodically updated and is publically available (Lichvar, Butterwick, Kirchner, & Melvin, 2016) and consists of categorical designations of the most typical hydrologic habitat in wetland and riparian areas for hundreds of plant species based on expert opinions. These designations are often used to identify wetland habitat.
Following Stromberg (2013), we extend the use of these designations to describe the hydrologic habitat in which each species is likely to occur. Investigation of quantitative functional traits of species on the plant list may elucidate the underpinnings of designations, facilitate classification of new species, and refine the overall designation scheme.
Our study aims to understand mechanisms of adaptation by comparing traits of species across wetland indicator status (WIS) categories (Lichvar et al., 2016;Reed, 1988). For example, Stromberg (2013) used species-level comparisons among WIS categories to show that obligate riparian species have shallower roots than facultative (FAC) riparian species and thus demonstrated a functional linkage between a plant morphological characteristic and niche. A small number of measurable traits have been postulated to represent the myriad of plant functional strategies and trait syndromes (Weiher et al., 1999;Westoby, 1998). For example, the leaf-height-seed scheme uses specific leaf area (SLA), plant maximum height, and seed mass to condense variation in a wider suite of traits and quantify plant strategy (Westoby, 1998). SLA represents the amount of leaf area per investment in leaf mass and is correlated with photosynthetic capacity, nitrogen content, leaf longevity, and leaf architecture (Ordoñez et al., 2009;Westoby, 1998). Low SLA has been reported in habitats that are arid (Wright et al., 2004), infertile (Hodgson et al., 2011), open (Ackerly, Knight, Weiss, Barton, & Starmer, 2002), and in more drought tolerant species (Mitchell, Veneklaas, Lambers, & Burgess, 2008). High SLA, in contrast, has been associated with wetter habitats (Cornwell & Ackerly, 2009) and may promote leaf gas exchange in plants coping with submergence (Mommer, Wolters-Arts, Andersen, Visser, & Pedersen, 2007;Voesenek, Colmer, Pierik, Millenaar, & Peeters, 2006). Plant height is related to competitive ability for light and soil resources through the expectation of larger root systems for taller plants (Keddy & Shipley, 1989;Westoby, 1998). Tall height also may be an adaptation to flooding by allowing less complete and shorter duration submergence of the stem (Voesenek et al., 2006). Seed mass is related to reproductive strategy, drought tolerance, and seedling establishment (Larios, Búrquez, Becerra, & Lawrence Venable, 2014;Leishman, Westoby, & Jurado, 1995;Westoby, 1998). Larger seed size may confer greater drought tolerance and survivorship by providing greater metabolic reserves for root elongation and construction of thick cell walls during seedling establishment and often occurs in species that occupy less disturbed sites and in areas where competition is more intense (Leishman & Westoby, 1994;Westoby, Jurado, & Leishman, 1992).
Other traits associated with water use and drought tolerance may also be informative in understanding adaptations of riparian plants.

Our study focuses on riparian vegetation of the Colorado River in
Grand Canyon in northern Arizona where completion of Glen Canyon Dam in 1964 altered flood frequency and magnitude as well as sediment regime and transport processes, severely altering the historic disturbance regime and leading to vegetation establishment, encroachment and narrowing of the active river channel (Gloss, Lovich, & Melis, 2005;Sankey, Ralston, Grams, Schmidt, & Cagney, 2015;Topping, Schmidt, & Vierra, 2003;Turner & Karpiscak, 1980). Additionally, river regulation created a permanent base flow that enabled the establishment and proliferation of plant species that were historically absent or rare (Stevens, Schmidt, Ayers, & Brown, 1995). Predictions of a warmer future climate and changes in water availability (e.g., reduced headwater snowpack and timing of flows) in this arid region may cause further alterations in river flows and the riparian vegetation community (Dominguez, Rivera, Lettenmaier, & Castro, 2012;Hayhoe et al., 2004;Seager & Vecchi, 2010;Seager et al., 2007). Better knowledge about distributions of species' functional traits across gradients of water availability and inundation is needed in order to inform predictions of plant community changes in response to further alterations to river flows and changing moisture availability and temperature and may help refine wetland plant classification.
We compared nine traits (SSG, seed mass, plant height, SLA, foliar δ 13 C and δ 15 N, foliar carbon and nitrogen concentration, foliar carbonto-nitrogen ratio) of 110 plant species that commonly occur along the Colorado River in Grand Canyon National Park, Arizona among five WIS categories (Lichvar et al., 2016;Reed, 1988). These categories describe the most typical habitat of a plant species and the affinity of species for conditions ranging between frequently flooded, inundated areas to arid habitats in uplands adjacent to the river. The species we investigated include a mixture of natives and non-natives, many of which occur in riparian ecosystems throughout the world. Thus, our study, while focused on riparian flora at a specific river, has broader application for understanding adaptations of riparian plants to environmental gradients. We addressed four questions: (i) How do specieslevel functional traits related to dispersal strategy, competitive ability, leaf physiology, water relations, and drought tolerance of Colorado River riparian flora vary over WIS categories? (ii) Does the pattern differ between herbaceous and woody species? (iii) How well do multivariate traits define WIS categories? (iv) Which traits are correlated?

| Site description
Our study area is 362 km of the Colorado River between Lees Ferry and Diamond Creek, AZ. This includes all of Marble Canyon and much of Grand Canyon (referred to collectively as Grand Canyon).
Vegetation communities vary along the river and are comprised of species found in the Sonoran, Great Basin, and Mojave Deserts (Huisinga, Makarick, & Watters, 2006;McLaughlin, 1986). The frequency and magnitude of flooding has been greatly reduced since the construction of Glen Canyon Dam, and variation in discharge is largely controlled by dam operations (Gloss et al., 2005). The river channel averages 100 meters wide and is largely confined by bedrock (Howard & Dolan, 1981). There is a steep moisture gradient from perpetually flooded areas in the active channel and high desert environments on colluvial side slopes and predam fluvial deposits furthest from the channel. Channel constrictions occurring at side canyons (debris fans) resulting from debris flows reduce the river's velocity and allow for deposition of sediments and formation of sandbars downstream from fans (Dolan, Howard, & Gallenson, 1974;Howard & Dolan, 1981;Schmidt, 1990). These sandbars form habitats within the channel and,

| Data collection
We used riparian vegetation surveys conducted by the US Geological Survey (USGS) at sandbars along the Colorado River within the Grand Canyon National Park to guide species selection. We measured or T A B L E 1 Traits used for this study, their units, source, and brief interpretation compiled data for 110 species (see Table 1 for summary of traits, see Appendices S1 and S2 for a list of species sampled and sampling locations), which included those that were most commonly documented in vegetation monitoring surveys in years 2012-2014, as well as many less common species. The species include six trees with the C3 photosynthetic pathway (C3), 18 C3 woody shrubs, 16 C3 grasses, 38 C3 forbs, three C3 herbaceous subshrubs, one woody shrub with the C4 photosynthetic pathway (C4), 22 C4 grasses, two C4 forbs, and three forbs and one subshrub of undetermined photosynthetic pathway.
Our field sampling approach emphasized estimation of mean trait values at the species level and did not formally consider intraspecific variation (e.g., among habitats for a given species) based on recent findings that trait rankings of species in large datasets typically are not sensitive to the exact location of sampling (Kazakou et al., 2014;Ordoñez, 2014 The SSG was calculated from these measurements as dry mass per volume (g/cm 3 ).
We obtained data on height and seed mass of the species from previously compiled databases and flora. Plant heights were compiled from floral descriptions found in A Utah Flora (Welsh, Atwood, Goodrich, & Higgins, 1993), the Jepson Desert Manual (Baldwin, 2002), and the Flora of North America (Flora of North America Editorial Committee 1993). Height was calculated as the average of the maximum and minimum height reported for mature plants in these floras. We found height data from this source to be strongly correlated with heights, we measured directly on species at field collection sites (Spearman r s = .83, p < .05). Seed mass data were compiled from the KEW Royal Botanic Garden Seed Information Database (http://data. kew.org/sid/) and consist of the average weight in grams per 1,000 seeds.
We measured additional foliar traits (δ 13 C, δ 15 N, percent carbon and nitrogen, and carbon/nitrogen ratio) on 56 species that have the C3 photosynthetic pathway. These additional measurements focused on species with the C3 photosynthetic pathway because differences in δ 13 C can be more directly interpreted as changes in leaf gas exchange and water use efficiency in C3 species than in C4 species (Cernusak et al., 2013). Leaves used in measurement of SLA were also used for

| Wetland indicator status
We used existing WIS categories to classify the typical habitat of each species. We used this approach because WIS is available for many species, it is based on many expert opinions, and it is often used in regulation.

| Data analysis
We compared species-level traits over WIS categories. Comparisons were carried out for all species pooled and for woody and herbaceous species separately. We tested for differences in traits among categories using ANOVA and Tukey's Honestly Significant Difference (HSD) for data that met normality and variance assumptions, and nonpara-  (Westoby, 1998), and due to their importance in plant drought tolerance (Brodribb, 2009;Hacke et al., 2001). Seed mass and height data were natural log-transformed to improve skewed distributions, and data for all traits were standardized so that values ranged between −1 and 1. The analysis was run for 999 permutations using Euclidian distance. Principal components analysis ordination with a biplot overlay of traits was used to display the results.

| Herbaceous species
For herbaceous species, statistically significant differences in trait values among WIS categories occurred for seed mass (x 2 = 15.12 p < .01) and SSG (x 2 = 20.73, p < .01), with marginally significant differences and trends for height (F = 2.20, p = .08), δ 13 C (F = 2.27, p = .10), δ 15 N (F = 2.32 p = .09), and % leaf carbon (F = 2.88, p = .05; Figure 2). Seed mass was lowest in the FACW category, but similar in other categories. SSG was lowest in the OBL category and increased in more upland categories. Height generally decreased from the OBL category to more upland categories. The δ 13 C was lowest in the FAC category and highest in the UPL category. The δ 15 N was lowest in the FACU category and highest in the FACW category. Percent leaf carbon was lowest in the FAC category, but similar in other categories. SLA was lowest in the OBL category, but overall differences in SLA were not significant (F = 1.65, p = .17). Percent leaf nitrogen and leaf C/N did not differ significantly among WIS categories and had no obvious trends ( Figure 2).

| Woody species
For woody species, statistically significant differences in trait values among WIS categories occurred for height (F = 6.61, p < .01), with marginally significant differences and trends for SSG (x 2 = 7.47, p = .06) and δ 13 C (x 2 = 6.81, p = .08; Figure 3). SSG was lowest in the FACW category, but similar in other categories. Height was lowest in the UPL category. The δ 13 C was lowest in the FACW category and increased in more upland categories. Seed mass, SLA, δ 15 N, % leaf nitrogen, % leaf carbon, and leaf C/N did not differ significantly among WIS categories and had no obvious trends (Figure 3).

| Multivariate analysis
The principal components analysis for data pooled over all species based on traits of height, seed mass, SLA, and SSG showed moderate separation of some WIS categories (Figure 4). PERMANOVA showed significant differences in trait suites among WIS categories (Pseudo

| CORRELATION AMONG TRAITS
Significant rank correlations were detected between SLA and height, δ 13 C, % leaf nitrogen, and C/N ratio (Table 2). Height was significantly correlated with SSG and % leaf carbon. SSG was significantly correlated with % leaf carbon.

| Trait variation among wetland indicator status categories
Our results for several functional traits of both woody and herbaceous species suggest convergent adaptation to extremely wet (OBL) and  (Chave et al., 2009;Hacke et al., 2001;Kyle & Leishman, 2009;Pockman & Sperry, 2000), yet adds new information for riparian herbaceous plants of the southwestern USA. Our finding of an increase in SSG of herbaceous species from wetland habitats to drier more upland habitats is consistent with previous reports that grass species with more lignified roots (Wahl & Ryer, 2000) and stems (Lens et al., 2016) are more resistant to drought-induced xylem cavitation and supports the hypothesis of similar xylem structure-function relationships in herbaceous and woody species (Lens et al., 2016;Tixier et al., 2013).
Our results suggest an adaptive role of seed mass to inundation and aridity in riparian ecosystems. Both herbaceous and woody species had the lowest seed mass in the FACW habitat category and greater seed mass in more arid, upland habitat categories. Seed mass of herbaceous species from the OBL category was higher than expected and similar to mass in more upland categories. Lower seed mass may be related to buoyancy and dispersal ability in plants, imparting a selective dispersal advantage for those plants growing near water (Merritt & Wohl, 2002).
This finding shows that some herbaceous species with large seeds can successfully occupy the wettest sites along rivers, likely because they also have other traits that facilitate tolerance of inundation and saturated soil, such as low-density aerenchymous stems, leaves, and roots.
Overall, our results suggest that species along the Colorado River in Grand Canyon occupy more arid, less inundated sites in part by having large seeds that promote large seedlings and deep roots. and Atriplex canescens. Decreasing height with increasing habitat aridity is consistent with other reports for riparian ecosystems (e.g., Kyle & Leishman, 2009) and may be explained by greater constraints on canopy growth by water stress in upland sites, and greater below-ground growth allocation of xerophytic vegetation (Canadell et al., 1996;Stromberg, 2013).
Results for leaf tissue δ 13 C of plants with the C3 photosynthetic pathway differed between herbaceous and woody plants. Differences in δ 13 C of herbaceous species having C3 photosynthesis were not consistently related to aridity of WIS categories, with the highest values occurring in both wet (FACW) and dry (UPL) categories. However, the proportion of herbaceous species with the C4 photosynthetic pathway increased from wet (0.0 OBL, 0.17 FACW) to dry WIS categories (0.33 FACU, 0.4 UPL). This pattern is consistent with an increase in water use efficiency of herbaceous species with habitat aridity in our study due to greater occurrence of C4 grasses in drier habitats. C4 grasses are well-known to have higher δ 13 C and water use efficiency (Cernusak et al., 2013) and typically occupy hotter sites than C3 grasses (Gurevitch, Scheiner, & Fox, 2006). Given this interpretation of our results for herbaceous species, results for water use efficiency of woody species were similar to herbaceous species. Leaf tissue δ 13 C and hence water use efficiency of woody species consistently increased with WIS category aridity. This finding for riparian vegetation within the Grand Canyon is consistent with earlier investigations of desert riparian vegetation in southwestern North America (Ehleringer & Cooper, 1988) and comparisons among woody species over wide habitat and climatic gradients (Diefendorf et al., 2010). Overall, we conclude that both woody and herbaceous species of riparian vegetation of Grand Canyon increase water use efficiency with habitat aridity through species replacement, with this change for herbaceous species primarily driven by a shift from species with C3 photosynthesis to C4 photosynthesis.
For example, SLA of herbaceous species in the OBL category with obvious aerenchyma tissue based on visual observations of cut stems (genera Scirpus, Juncus, and Typha) averaged 106 cm 2 /g and ranged from 53 to 197 cm 2 /g, whereas average SLA for the only species in the OBL category that did not have a vertical linear leaf with pronounced aerenchyma (Symphiotrichum divaricatum) was 302 cm 2 /g.
Our results for leaf δ 15 N suggest a trend of change over WIS categories in nitrogen sources for herbaceous species. δ 15 N tended to vary more over WIS categories for herbaceous species (p = .09) than woody species (p = .88). The δ 15 N of herbaceous species was greatest in the wettest WIS category (FACW) included in the analysis. This result suggests more variation in N sources over habitats for herbaceous species than woody species, perhaps due to shallower roots of herbaceous species. Frequency and magnitude of flooding and cycles of soil wetting and drying can alter riparian litter accumulation, nutrient dynamics, and discrimination against δ 15 N by soil microbes (Austin et al., 2004;Sponseller & Fisher, 2006). For example, Hall et al. (2015) reported δ 15 N enrichment in riparian species with submerged roots because of isotopic enrichment from benthic denitrification.
Another factor possibly contributing to variability in N sources available to herbaceous species in our study is the density of woody species across the riparian landscape. Transitional riparian habitats along the Colorado River in Grand Canyon between the river edge and xeric uplands have greater woody vegetation and thus higher production of litter, likely contributing to faster biochemical processing and nutrient dynamics than more xeric habitats (Kennedy & Ralston, 2012;Sankey et al., 2015). The trend in δ 15 N for herbaceous species over WIS categories in our study suggests spatial variation in nitrogen sources that should be explored further.

| Multivariate traits
Principal components analysis based on traits of height, SLA, SSG, and seed mass showed moderate separation of species among WIS categories ( Figure 4)  (r = −.66 and r = .42, respectively) and seed mass to PCO2 (r = −.35) suggest these traits explained substantial variation among categories.
Specifically, hydrophytic species in Grand Canyon have a combination of lower SLA, SSG, and seed mass than nonhydrophytic species.

| Trait correlations
SLA has been suggested as an easily measured surrogate of the leaf economics spectrum because it is correlated with other leaf traits, such as leaf nitrogen concentration, maximum photosynthetic rate, decomposition rate, and water use efficiency (Diefendorf et al., 2010;Pérez-Harguindeguy et al., 2013;Reich, 2014;Wright et al., 2004).
Our results are consistent with this suggestion for the 56 species with C3 photosynthesis we investigated. SLA was significantly correlated with δ 13 C, % leaf nitrogen, leaf C/N, and height. The weak but significant correlation (r s = −.28, p = .003) between SLA and height was unexpected based on Westoby's leaf-height-seed scheme (1998), which postulates that these traits are orthogonal, or independent axes of trait variation. Significant correlation between SSG and percent leaf carbon (r s = .51, p = .00005) suggests selection on stem and leaf traits collectively, consistent with the conclusions of other studies (Bucci et al., 2004;Ishida et al., 2008;Maherali, Moura, Caldeira, Willson, & Jackson, 2006). This collective selection is a likely driver for the positive correlation of percent leaf carbon and height (r s = .34, p = .01), and the positive correlation of SSG and height (r s = .20, p = .04) with greater height and canopy requiring greater support and overall carbon investment in tissue.
Our investigation of functional traits of vegetation that dominates riparian ecosystems along the Colorado River in Grand Canyon revealed five major findings. First, SSG, or what is also termed "stem density" or "wood density" (Chave et al., 2009)

ACKNOWLEDGMENTS
Thanks to the USGS for logistical support and funding through the WaterSMART program and to Grand Canyon National Park for allowing us to work and sample in the Park (Permit# GRCA-2015-SCI-0015). Patrick Shafroth and Scott Vanderkooi (USGS), and two anonymous reviewers provided helpful comments on the manuscript.
We devote this manuscript to our deceased colleague Daniel Sarr. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the US Government.

DATA ACCESSIBILITY
Trait data for the project have been archived at ScienceBase, https:// doi.org/10.5066/F7BV7DTQ. ScienceBase is maintained by the United States Geological Survey.