Functional structure of plant communities along salinity gradients in Iranian salt marshes

Abstract Salt marshes are unique habitats between sea or saline lakes and land that need to be conserved from the effects of global change. Understanding the variation in functional structure of plant community along environmental gradients is critical to predict the response of plant communities to ongoing environmental changes. We evaluated the changes in the functional structure of halophytic communities along soil gradients including salinity, in Iranian salt marshes; Lake Urmia, Lake Meyghan, Musa estuary, and Nayband Bay (Iran). We established 48 plots from 16 sites in four salt marshes and sampled 10 leaves per species to measure leaf functional traits. Five soil samples were sampled from each plot and 30 variables were analyzed. We examined the changes in the functional structure of plant communities (i.e., functional diversity [FD] and community weighted mean [CWM]) along local soil gradients using linear mixed effect models. Our results showed that FD and CWM of leaf thickness tended to increase with salinity, while those indices related to leaf shape decreased following soil potassium content. Our results suggest that the variations in functional structure of plant communities along local soil gradients reveal the effect of different ecological processes (e.g., niche differentiation related to the habitat heterogeneity) that drive the assembly of halophytic plant communities in SW Asian salt marshes.


| INTRODUC TI ON
Salinization is a global degradation process affecting not only soil quality and plant distribution but also the ecosystem services provided by healthy drylands (Flowers et al., 2010;Parida and Das, 2005;Wang et al., 2015). Paradoxically, natural habitats linked to saline soils located between sea or saline lakes and land such as "salt marshes" have been considered of significant importance for nature conservation (Isacch et al., 2006;Milotić et al., 2010;Tabot and Adams, 2013). They are extremely influenced by fluctuations of salinity, which are periodically caused by annual rainfalls, flooding, and inundation (Chapman, 1977;Clarke and Hannon, 1970;Gleason, 1926;Tug et al., 2012). Consequently, salt marshes are fragile and many of them face critical anthropogenic disturbance (Bouchard et al., 2003;Zahran and Willis, 1992). Indeed, direct and indirect anthropogenic factors such as overgrazing, agriculture and intensive irrigation may severely disturb salt marshes and so natural vegetation of these areas (Tug et al., 2012).
Salt marshes are frequently used as study systems to examine plant community structure (Wilson and Whittaker, 1995;Zedler, 1977;Orson and Howes, 1992), as they are colonized by relatively simple plant communities with few dominant species and very low plant diversity (Asri and Ghorbanli, 1997;Cutini et al., 2010;Ghorbanalizadeh et al., 2020;Tug et al., 2012). Generally plant communities in SW and Central Asian salt marshes are dominated by plants specialized to saline soils which their zonation is depending on local topography, existing macroclimate and hydrological conditions.
Usually eu-halophytes such as annual Salicornia spp. and C 3 annual where fresh or brackish water inflow from rivers, streams, aquifers, and wetlands. The C 4 transitional plant formations with species of Chenopodiaceae family are common in moderately saline soils or ruderal salt affected soils as a usually wide zone ending to xerophytic steppes largely dominated by Artemisia and Stipa species. (Akhani, 2004;Akhani, 2015;Akhani and Mucina, 2015;Djamali et al. 2011;Ghorbanalizadeh et al., 2020).
Several studies on the relationship between vegetation and soil showed that salt concentration in the groundwater, soil salinity, elevation, K, Na, Ca, and Mg content in the soil are strong determinants of soil-vegetation dynamics in salt marshes (Brewer and Grace, 1990;Cantero et al., 1998;He et al., 2011;Rogel et al., 2000). However, although there is ample information on the differences in plant community composition of salt marshes at the global (Adam, 2002;Simas et al., 2001) and regional scale (Asri, 1998;Ghorbanalizadeh et al., 2020;Niering and Scott Warren, 1980), we have no information on the effect of Na salinity and other soil variables on the functional structure of halophyte communities. Understanding the underlying factors that determine the functional structure of plant communities along local gradients is essential to predict the response of vegetation to different global change drivers and to define conservation programs to protect salt marshes and halophyte communities.
Here, we evaluated the functional structure of plant communities in salt marshes from Iran in relation to salinity and other relevant soil variables. In this regard, we aimed to answer two questions: (i) how does functional plant diversity change with increased soil salinity? (ii) What are the most dominant trait values for certain functional traits in the plant communities occurring along salinity gradients? 2 | MATERIAL S AND ME THODS

| Study area
This study was conducted in four salt marshes in Iran: Lake Urmia (NW Iran), Lake Meyghan (Central Iran), Musa estuary and Nayband Bay (South of Iran) (Figure 1; Appendix S1; See Matinzadeh et al. 2019 for further details). All studied salt marshes are dominated by halophytic and salt-tolerance plants and present local salinity gradients, which lead to a natural zonation of halophytic vegetation. The two first of these salt marshes are inland and both suffer the reduction of water income due to agriculture and unsustainable irrigation management. They are located in a semiarid region with similar bioclimate and so their precipitation is largely similar. The last two salt marshes are coastal with similar bioclimate. Their vegetation is influenced by inundation and tide, which is remarkably different to inland salt marshes (Akhani, 2004;Akhani, 2015;Ghorbanalizadeh et al., 2020).
Lake Urmia is known as the largest inland lake in Iran and second largest hypersaline lake in the World (Stone, 2015;ULRP, 2018).
This area is part of the tropical desertic (Trd) bioclimate (Djamali et al., 2011) (Figure 1c). Nayband Bay is located near the Asaluyeh industrial zone on the Persian Gulf coasts. Its tidal vegetation includes mostly mangrove forests of Avicennia marina followed by Arthrocaulon macrostachyum on the high saline shores, Sporobolus arabicus on saline sand shores, and end to xerophyte plants on coastal dunes and dry plains (Akhani, 2004). The inflow of seawater in the coastal area and lowlands, the presence of soil layers containing salts and the high transpiration rates caused by strong winds are the main factors causing soil salinity in this area (range of EC = 1.07-10.52 and pH = 8.1-9.8). The presence of sodium and chloride as the main ions along with other ions such as calcium, potassium, magnesium, and sulfate contribute to the high salinity of this area (Akhani, 2015). This area is typical of a tropical xeric (Trx) bioclimate (Djamali et al., 2011) ( Figure 1d).

| Functional traits measurement
Six continuous plant traits including plant height (PH), leaf thickness (LT), leaf shape (LS; leaf length (LL)/leaf width (LW)), leaf area (LA), leaf perimeter (LP), specific leaf area (SLA) and leaf dry matter content (LDMC) were measured, obtaining the average of each trait value per species and plot (Table 1). In addition, six categorical traits were included in Appendix S2 to provide more information about study species. These traits, including life history, growth form, and life form were determined according to Pérez-Harguindeguy et al. (2013), salt-tolerance category was determined based on Breckle (1990), eco-morphotypes according to Breckle (1986) and photosynthetic pathway by information available in the literature (Akhani et al., 1997;Akhani and Ziegler, 2002;Osborne et al., 2014;Rudov et al., 2020) The "water saturated-leaf mass" was measured with a precision balance (Sartorius, TE153S, d = 0.001 g) after rehydrating samples for 12 h (and succulent plants for 6 h), and LT was measured by a digital micrometer (Mitutoyo, MDC-25SB, d = 0.001 mm) (Pérez-Harguindeguy et al., 2013). In detail, we measured thickness of the young succulent stems of stem succulent plants (e.g., Salicornia) as LT trait. The scans of water-saturated leaves were used to calculate LA, LP, LL, and LW using image analysis software ImageJ 1.42q (National Institutes of Health, USA; http://rsb.info.nih.gov/ij). Leaves were subsequently oven dried at 70-75°C for 72 h, and weighed to obtain the "dry leaf mass". SLA was calculated as the ratio of LA to its dry leaf mass (mm 2 mg −1 ) (Minden et al., 2012), and LDMC as the dry leaf mass divided by the water-saturated leaf mass (Vendramini et al., 2002).

| Soil chemical analyses
Soil samples were air-dried, milled in a ball mill (Retsch Mixer MM400) and then dissolved in HCl-HNO 3 (9,3) using Microwave Acid Digestion (speedwave MWS-3 + , BERGHOF). The filtrated extract solution was used to determine Aluminium (Al), Arsenic (As), Calcium diluting with distilled water to 1:2.5 and 1:5 (g: ml), respectively. Soil EC was used to measure and as a surrogate for soil salinity. Also, high Na was used as indicative of high salinity in soil, since Na is a main cation in the studied sites (Akhani, 1989(Akhani, , 2015Asri and Ghorbanli, 1997). The percentage of gypsum was determined gravimetrically comparing the weight of samples dried at 50 and 105 °C (Porta et al., 1986). Soil carbonate was estimated with a Bernard calcimeter (Bolukbasi et al., 2016) and organic matter was measured through the wet oxidation method (Heanes, 1984). Soil texture was  Song et al., 2014;Valencia et al., 2015). For that, we used the mean pairwise distance (MPD): where n is the species number in the plots, δ the trait distance matrix, and δ i,j the Gower distance (Gower, 1971;Podani, 1999) of any given single trait between species i and j.
The correlation of the traits was analyzed by cor function in R package "stats" (R Core Team, 2018). LP was excluded before analysis to reduce correlation (>70%). After that, trait values were centered and scaled (Muscarella and Uriarte, 2016) using the scale function in Base R (R Core Team, 2018). To compute MPD and CWM independent to local species richness, the functional structure of 34 observed plots (only plots with more than two species were included in the analyses) were compared to 999 random communities derived from null models using an "independent swap" algorithm. We previously tested reshuffling effects on our results, through running three different algorithms of null models: (a) frequency, (b) richness where Obs is the observed value of MPD or CWM in the communities, and Mean null and SD null are the mean and standard deviation of MPD or CWM in the null communities, respectively.

SES of MPD (SES-MPD) was calculated for six continuous traits
in each plot, using the function ses.mpd in the R package "picante" (Kembel et al., 2010). SES of CWM (SES-CWM) was calculated within each community for each six continuous traits separately by the Equation 3.

| Statistical analyses
We examined the variation of plant community functional structure along soil gradients using linear mixed effect models (LMMs), such that soil elements were considered as predictors or fixed effects and sites as random effects. By this model, we analyzed the effect of soil variables on trait distribution patterns (SES-MPD or SES-CWM; as response factor). To reduce correlation among variables, we checked the correlation of soil variables using cor function and selected pH, OM, As, K, Mg, Na, P, Si, and N. After that, we performed a principal components analysis (PCA) using prcomp function in R package "stats" selecting the less orthogonal variables with high loading in the first two PCA axes (Appendix S6). As a result, the soil variables selected were K, Mg, Na, and N. Then, VIF scores were used to check for multicollinearity using the vif function in the package "car". Before that, soil parameters were log-transformed by log function in Base R (R Core Team, 2018). The statistical significance of each predictor in the model was determined using likelihood ratio tests (Winter, 2015). Finally, the best model was selected with Na, K, Mg, and N as factors showing the highest contribution to the statistical significance of the model.

| RE SULTS
The There were no significant patterns for SES-MPD along soil N gradients ( Figure 5a). However, we observed an increasing significant trend of SES-CWM for LA along the soil N gradients (Figure 5b). However, CWM results for LT showed that the most successful strategy in high salinity is that of succulent plants with thick leaves ( Figure 2b). So, the increasing pattern of MPD for LT along salinity gradients could be mainly related to habitat heterogeneity within plots, which provides different niches for succulent (e.g., Salicornia iranica) and salt-recreting halophytes (e.g., Atriplex tatarica). In detail, succulent C 3 halophytes of chenopods occur in the wettest and saltiest parts of saline habitats, such as margins of saline rivers, salty lakes, and sea, while salt-recreting C 4 halophytes of chenopods mostly occupy the drier parts of saline soil gradient in transition between hygro-halophytes and xerophytes (Akhani et al., 2003;Frey & Kürschner, 1983). Previous studies have reported a strong heterogeneity in space and time in saline and arid conditions, which can support species with different traits delimiting different niches (de Bello et al., 2006;Ricotta and Moretti, 2011;Scherrer et al., 2018).

| DISCUSS ION
On the contrary, MPD reduction and low CWM for LT indicated the dominance of thin-leaved plants (e.g., Alhagi maurorum, Artemisia spicigera) at low saline condition. Together with thinner leaves, these species may be suited with certain phenological and ecological achievements such as earlier germination, well-developed root system, vegetative growth, and delayed senescence (i.e., a longer growing season) that enable them to colonize more benign soils with low salinity by avoiding periods of increased salinity (Rozema and Schat, 2013) (Figure 2). In addition, salt-tolerance plants are the weaker competitor in this condition and so they were excluded.
Consequently, our results point at a shift in the dominance strategy along salinity gradients: from two salt-tolerance strategies including succulent halophytes (e.g., Suaeda altissima, Salicornia iranica, F I G U R E 3 SES-MPD (a) and -CWM (b) of traits along potassium (K) gradients. Log-transformed values of soil K are shown. P-values for every plots show the significance of this relation based on likelihood ratio test. Solid lines represent significant relation, while the dashed lines represent non-significant correlation in the regressions. The 95% confidence intervals for the regressions are shown. LT, leaf thickness; PH, plant height; LS, leaf shape; LA, leaf area; SLA, specific leaf area; LDMC, leaf dry matter content Halimocnemis rarifolium, Climacoptera crassa, and Halocnemum strobilaceum) and salt-recreting halophytes (e.g., Atriplex tatarica, Aeluropus littoralis, and Frankenia hirsuta) in the most stressful saline parts of the gradients, to salt-avoidance strategy characterized by thin leaves and the aforementioned avoidance mechanisms (e.g.,
CWM results for SLA in response to shifts in soil K (Figure 3b) indicated a positive relationship between K soil content and high SLA values. SLA reflects plant resource-use strategy in many environments and relates to plant relative growth rate, photosynthetic efficiency and nutrient conservation strategies (Valencia et al., 2015;Wang et al., 2015;Wilson et al., 1999). Allocation of more resources to photosynthesis and growth is typical in resource-rich habitats (Thuiller et al., 2010;Wang et al., 2015;Wellstein et al., 2013).
Therefore, the dominant plants growing on soils with high K content tend to have high SLA and faster relative growth rates, which could be annual succulent plants. These plants with high SLA can store water in thick tissues of their main photosynthetic organs.
Accordingly, our results indicate that these annual succulent plants (e.g., Suaeda altissima, Bienertia cycloptera, and Climacoptera lanata) a decreased production of fibrous tissue and thinner cell walls and a higher allocation of resources to growth and photosynthesis in these species (Grubb et al., 2015). Furthermore, the opposing pattern of CWM for SLA and LT (Figure 4b) in the soil Mg gradient could be related to more variation and plasticity of SLA than LT (Adler et al., 2014;Gross et al., 2013;Vendramini et al., 2002). Phragmites australis; Appendix S3), which is in line with previous findings supporting the view that taller plants might be significantly related to higher availability of some macronutrients such as Mg (Wellstein et al., 2013).
LA and LS can also affect leaf thermal and water conductance since small leaves can help to keep water and leaf temperatures lower in hot and dry conditions (Cornwell and Ackerly, 2009). Low CWM for LA in high soil Na content (Figure 2b) might be due to the adaptation of plants to higher physiological drought which expected under high soil salinity. When saline stress is high, stomata tend to be closed, leading to high leaf temperature and high damage to large leaves (Cornwell and Ackerly, 2009). Our results, similar to previous studies, showed that plants with high LA generally tend to grow in milder saline habitats, because their photosynthetic rates would be high in this condition (Cornwell and Ackerly, 2009;Gross et al., 2013). Low CWM for LA and LS or the dominance of small-leaved plants in low soil Mg and N, respectively, could be due to macronutrient (i.e., Mg and N) deficiencies (Figures 4b and 5b) (Trubat et al., 2006;Watson, 1947).
High MPD of LS in plots with low soil K may indicate the high niche differentiation in heterogeneous environments between functionally different plants with similar responses to low K and also high Na content in the soil (i.e., large-elongated non-succulent leaves, e.g., Bolboschoenus glaucus and small-scaly succulent leaves, e.g., Caroxylon imbricatum) (Figure 3a; Appendix S3). Furthermore, the high CWM for LA and LS in low soil K (Figure 3b) indicates the most successful strategy is that of large and long leaved plants such as Bolboschoenus glaucus, which are distributed in margins of saline and moderately saline lakes, salty and brackish swamps (Akhani and Ghorbanli, 1993). These results may be explained by the antagonism effect of K and Na in the soil, especially in sodic or saline-sodic soils, where that high Na content and saline stress is linked to low soil K (Matinzadeh et al., 2013;Wakeel, 2013).

| CON CLUS ION
Our results demonstrate that the functional structure of plant communities in Iranian salt marshes may change along sodium and associated nutrient content gradients. We found that the increase of MPD and CWM for LT along soil Na gradients would be related to niche differentiation in heterogeneous environments and dominance of F I G U R E 5 SES-MPD (a) and -CWM (b) of traits along nitrogen (N) gradients. Log-transformed values of soil N are shown. P-values for every plots show the significance of this relation based on likelihood ratio test. Solid lines represent significant relation, while the dashed lines represent non-significant correlation in the regressions. The 95% confidence intervals for the regressions are shown. LT, leaf thickness; PH, plant height; LS, leaf shape; LA, leaf area; SLA, specific leaf area; LDMC, leaf dry matter content succulent plants in high saline soil. In addition, MPD and CWM of LS decreased along soil K gradients and the plants with small-elongated leaves are dominant in high K content in the soil. Furthermore, the reduction of CWM for LA along soil Na gradients indicates the plants with small leaves are the most successful plants in high Na content in the soil, while the increase of CWM for LA towards soil N availability would be mainly related to success of plants with large leaves and high photosynthetic rates in high macronutrient availability. We conclude that the variations in functional structure of plant communities along environmental gradients can display some ecological processes such as niche differentiation related to habitat heterogeneity, which can influence the assembly of halophyte communities in Iranian salt marshes.

ACK N OWLED G M ENTS
The authors would like to thank Atefeh Ghorbanalizadeh for her helps during field studies and identification of some plant species, and Vahid Sedghipour and Moslem Doostmohammadi for joining field studies. We also thank Andreu Cera for his help in starting statistical analyses with R.

CO N FLI C T O F I NTE R E S T
The authors have no conflict of interests to declare.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that supports the findings of this study are available in the supplementary material of this article.