Seasonality drives herbaceous community beta diversity in lithologically different rocky outcrops in Brazil

1Centro Universitário Redentor, BR 356, Cidade Nova, Itaperuna, Rio de Janeiro, Brazil 2Universidade Federal de Viçosa, Department of Plant Biology, Campus Viçosa, 36570-900, Viçosa-MG, Brazil 3Universidade Federal de Viçosa, Department of Plant Biology, Laboratory of Ecology and Evolution of Plants, Campus Viçosa, 36570-900, Viçosa-MG, Brazil 4Universidade Federal de Viçosa, Department of Soil, Campus Universitário, 36570-900, Viçosa-MG, Brazil *Corresponding author: jaquelinabiologa@yahoo.com.br REGULAR PAPER


INTRODUCTION
Temporal environmental fluctuations, such as climatic seasonality, exert strong controlling forces on species diversity patterns from different ecological communities (Ferreira et al. 2017;Bao et al. 2019;Duong et al. 2019;Villa et al. 2019a). Therefore, understanding how seasonal environmental variation determines plant community diversity has been one of the central aspects of ecological studies (e.g., Bao et al. 2017;Poorter et al. 2019). Seasonal climatic conditions (i.e., precipitation) generally determine the temporal dynamics of germination and dormancy (Silveira et al. 2016;Tonkin et al. 2017), and the amount of water available for plants to grow and establish is an important limiting factor in several global ecosystems (e.g., Ulrich et al. 2014;Le Bagousse-Pinguet et al. 2017;Bao et al. 2018). Specifical-ly, the temporal variation in precipitation can substantially affect species composition and richness patterns in different plant communities (e.g., Klanderud et al. 2015;Le Bagousse-Pinguet et al. 2017;Bao et al. 2018), particularly for water-dependent herbaceous species with short life cycles (Royo & Ristau 2013;Bao et al. 2017). Conversely, seasonality provides the necessary conditions for the regeneration of plant species (Poorter et al. 2019;Rozendaal et al. 2019). Compared to stable systems, predictable seasonal precipitation can promote high species turnover (Tonkin et al. 2017), particularly beta diversity (Shimadzu et al. 2013;Bao et al. 2017Bao et al. , 2018. Beta diversity is a measure of the temporal or spatial variation of species composition (Tuomisto 2010;Anderson et al. 2011). The simplest definition of taxonomic beta diversity, and one of the most used, is that it is the percentage of dissimilarity in the species composition of two communities (Koleff et al. 2003). However, most studies on species composition variation have used spatial analysis along environmental gradients, and studies on temporal variations are scarce (Ruhí et al. 2017).
The rocky outcrops in Brazil, known as campos rupestres, are hotspots of terrestrial species diversity (Alves et al. 2014;Fernandes 2016). Although this region occupies less than 1% of the Brazilian territory, it hosts nearly 15% of the national flora (Silveira et al. 2016) and nearly 40% of the species are narrow endemics (ca. 2000 species), many of which are threatened with extinction (Martinelli & Moraes 2013). Campos rupestres comprises a mosaic of plant communities that have adapted to the landscape heterogeneity of mountain ranges with varying levels of tolerance to edaphoclimatic factors (Negreiros et al. 2014;Fernandes 2016). The vegetation faces specific environmental conditions of high stress due to shallow nutrient-poor soils with low water-holding capacity, which leads to low water availability and droughtinduced growth limitation (Benites et al. 2007;Jacobi et al. 2007;Fernandes 2016;Figueira et al. 2016;Schaefer et al. 2016;Silveira et al. 2016). The diversity of habitat types is also considerable and the region contains peculiar landscapes occurring mainly on quartzite and ironstone formations (Benites et al. 2007;Silveira et al. 2016). The rocky outcrops are also recognized as ecosystems with marked environmental seasonality (Silveira et al. 2016). Due to changes in climate, mountains are suitable systems for identifying changes in diversity patterns (Pecl et al. 2017). Tropical mountains have higher temperatures at low elevations and overall wetter conditions than temperate mountains (McCain & Grytnes 2010). However, most studies about changes in species composition and diversity in plant communities of Brazilian rocky outcrops have analysed environmental gradients (e.g., Alves et al. 2014;Nunes et al. 2015;Mota et al. 2018;Alves-Silva et al. 2019), while temporal differences in the composition of the herbaceous community remain poorly understood.
In this study, we investigated the effects of seasonal differences in precipitation and variability in soil fertility on species richness and beta diversity of two different herbaceous plant communities in the Brazilian campos rupestres. For that reason, we sampled herbaceous species on an ironstone outcrop in the Carajás National Forest (Pará state) and a quartzite outcrop in the Iron Quadrangle (Minas Gerais state). We established the following two research questions: (1) How do the species richness and beta diversity of each site change between the dry and the wet season? (2) What are the effects of seasonal differences in precipitation (i.e., between the dry and the wet season) and variability in soil fertility on species richness and beta diversity? We hypothesize that seasonality, determined by variation in precipitation, and soil fertility, determined by variability in nutrient-related soil properties, drives species diversity (i.e., richness and beta diversity) patterns of herbaceous communities at local scale.

Study sites
The study was carried out at two different Brazilian campos rupestres. (1) The ironstone outcrop formation, canga (lateritic capping), known as Serra Sul (738-762 m a.s.l.), located in the Carajás National Forest (6°24′37.7″S, 50°21′31.0″W) conservation unit in south-eastern Amazonia, Pará state, Brazil. The ironstone canga (ferruginous campo rupestre islands) covers approximately 9031 ha (2.28%) of the total FLONA area of 395,827 ha (Nunes et al. 2015). (2) A quartzite outcrop formation in a private nature reserve at Fazenda da Capanema (20°13′41.4″S, 43°36′48.7″W). This site is located on the eastern border of the Iron Quadrangle, Minas Gerais state, Brazil. We selected sampling areas at each site that were as homogeneous as possible. Additionally, this criterion allows for controlling for possible effects of fine spatial scale differences and analysing the direct effects of seasonality and soil fertility.

Climate and soil characteristics
The climate in Carajás National Forest is tropical, hot and humid (type "Aw" in the Köppen climate classification) (Alvares et al. 2014) with a dry season between May and October that receives 14.5% of the total annual rainfall and an average of 50 mm precipitation per month (INMET 2020; supplementary file 1A). The rainy season, between November and April, receives an average of 244 mm precipitation per month and 79% of the total annual rainfall (Lopes et al. 2013). The mean annual temperature ranges from 23.5°C at 835 m a.s.l. to 26.2°C at 203 m a.s.l. (Silva et al. 1996). The position of the Intertropical Convergence Zone and the intensity of convection in the Amazon are two major factors driving differences in regional and seasonal precipitation (Lopes et al. 2013). In addition, moist trade winds from the tropical Atlantic and evapotranspiration from the Amazon rainforest also contribute to the precipitation (Alizadeh et al. 2017).
The climate in the Iron Quadrangle is subtropical with a dry winter and a hot summer (type "Cwa" in the Köppen climate classification) (Alvares et al. 2014), and is strongly influenced by altitude. This region has a dry season between April and September with an average of 32 mm precipitation per month and a rainy season between October and March with an average of 200 mm precipitation per month (IN-MET 2020; supplementary file 1B). Throughout the year, the region experiences considerable temperature variability, reaching below 10°C in winter and above 30°C in summer (Vincent 2004;Ferrari et al. 2016). Seasonal moisture variation in the soil of the campo rupestre is indicative of severe water deficiency during the winter/dry season (Ferrari et al. 2016). Ferrari et al. (2016) also pointed out that the first 30 cm of topsoil has the highest water deficiency, reaching values close to zero.
The soil in Carajás rocky outcrop is acidic and has low phosphorus content (Nunes et al. 2015). The ironstone areas, comprises a friable soil that rarely exceeds 5 cm of horizon A. In some areas, the soil is formed from old abandoned termite mounds, forming micropockets of organic soil that fill fractures and dissolution depressions in the canga . The Iron Quadrangle quartzite formation, the soil is also very acid and extremely oligotrophic and high levels of exchangeable Al. In general, both areas represent a harsh condition with low levels of nutrients and low water availability due to the shallow soil .
These rocky outcrops have been recently proposed as one of the world's old climatically buffered infertile landscapes (OCBILs), which have reduced water and nutrient availability and contain shallow soils (Silveira et al. 2016).

Soil properties measurements
In order to measure the soil properties within each plot by season and site, a surface soil sample was collected (up to 10 cm depth). Soil properties of the samples were measured in the Soil Analysis Laboratory of the Federal University of Viçosa, following standard protocols (EMBRAPA 1997). The following parameters were assessed: available exchangeable P, K, Ca, Na, Mg, Fe, Cu, Mn, Zn, Al, exchangeable acidity (H + Al), pH (H 2 O), organic matter (OM), sum of exchangeable bases (SB), effective cation exchange capacity (t), cation exchange capacity acidity (T), potential index (m), bases saturation index (V), and Na saturation index (ISNa). Soil data was in part locally available or could otherwise be obtained from the harmonized world soil database (HWSD version 1.2; FAO et al. 2012).

Vegetation sampling
We carried out vegetation sampling in 20 plots (1 × 1 m) in Carajás and 40 plots (1 × 1 m) in the Iron Quadrangle. We sampled each plot twice, once during the dry season and once during the wet season. In the Carajás ironstone outcrop, the plants were sampled in July 2007 and January 2008, while in the Iron Quadrangle quartzite outcrop, the plant sampling was carried out in July 2011 and February 2012. In each plot, the number of individuals of each species was counted (Braun-Blanquet 1979). The species were also classified according to Raunkiaer's biological spectrum (e.g., Alves-Silva et al. 2019).

Data analyses
Differences in species richness between the two sampled areas were evaluated using individual-based data to estimate rarefaction and extrapolation curves using the first Hill number (Chao et al. 2014), and extrapolations were made based on abundance data for the 1 × 1 m plots (Colwell et al. 2012). Rarefaction and extrapolation curves present the lines that represent the mean values and the bands the standard deviation with 95% confidence intervals (i.e., Campos et al. 2018; Alves-Silva et al. 2019). These estimates were obtained using the 'iNEXT' package (Hsieh et al. 2016).
We performed principal coordinate analysis (PCoA) based on Bray-Curtis dissimilarity matrix among seasons to explore the effects of seasonality on the patterns of species composition. PCoA is run on distance matrix (Monte-Carlo, 999 permutations) and can directly represent the relationships among samples to determine differences in species composition. The PCoA was performed using the 'vegan' package in R (Oksanen et al. 2018). We used permutation multivariate analysis of variance (PERMANOVA, 9999 permutations) to determine differences in species composition by using the 'adonis' routine, available within the vegan package (Oksanen et al. 2018).
To assess beta diversity, we used the betadisper function in vegan, based on the applied multivariate analysis of dispersal (PERMDISP) method (Anderson 2006). We used distance to group centroids to assess variability in species composition, based on the Bray-Curtis metric, on the abundance data (Monte-Carlo, 9999 permutations). In this analysis, the greater the dispersal of the data points, the greater the heterogeneity of the groups. We tested for differences in beta diversity between seasons for each site using PERMANO-VAs (Anderson 2001). All data and statistical analyses were conducted in R version 3.2.2 (R Core Team 2017). Finally, we used the two-way cluster analyses using Bray-Curtis dissimilarity based on abundance data using the PC-ORD software (i.e., Villa et al. 2019b) to identify similarity patterns for the species between seasons.
We used principal components analysis (PCA) on the correlation matrix to reduce the number of redundant soil properties (e.g., Schmitz et al. 2020) in each site (fig 2), using the 'FactoMineR' package (Husson et al. 2017). Thus, we reduced the correlations among local soil variables using the first PCA axis for mainly the variability of nutrientrelated soil properties (PCA1) (e.g., Schmitz et al. 2020). To explain the main effects of variability of nutrient-related soil properties (PCA1) and seasons on species richness and beta diversity, we constructed linear mixed effect models (LMM) after having evaluated normality (Crawley 2013). The models were tested for each site to explain local effects of seasonality and soil variability. We observed that soil texture does not vary between seasons and was not an important predictor between tested models. Predictor variables were grouped into two categories, i.e., the variability of nutrient-related soil properties as a continuous variable and the season as a categorical variable that has two levels (i.e., dry and wet season). The univariate and interaction effect of the predictors on richness and beta diversity was evaluated, and the plots were considered as a random factor, and the PCA axis predictor and seasons as a fixed factor. All models were calculated using the packages 'lme4' (Bates et al. 2015(Bates et al. , 2017 in the platform R version 3.6.0 (R Core Team 2019).
We observed similar richness patterns between seasons without significant differences using both individual-based rarefaction and extrapolation curves ( fig. 2A-B). However,  there are notable differences in the number of species at both sites between the wet and the dry season, especially for the chamaephytes and therophytes ( fig. 3).
Species compositions showed significantly differences between dry and wet seasons at both sites, in Carajás (PER-MANOVA: F 1,38 = 6.70, p < 0.001; fig. 4A) and in Iron Quadrangle (PERMANOVA: F 1,77 = 14.54, p < 0.001; fig. 4C), forming two groups on the first axis. According to the first two PCoA axes, seasonality explains 36.13% and 38.12% of the variance in species for the Carajás and Iron Quadrangle site, respectively ( fig. 4). From this analysis, most of the sample points for the dry season clustered together in the multidimensional space, with less distance from the centroid, compared to a more dispersed pattern for the wet season. We did not observe changes in beta diversity between seasons at the Carajás site (PERMANOVA: F 1,38 = 1.41, p < 0.24; fig.  4B) or the Iron Quadrangle (PERMANOVA: F 1,77 = 5.78, p < 0.10; fig. 4D).
The two-way cluster analyses broadly divided the plant species into two groups based on season, which can be clearly seen by the two main branches of the dendrogram at both sites, in Carajás (fig. 5) and in Iron Quadrangle ( fig. 6).
Despite the lack of differences in beta diversity between the two sites, the models indicate that beta diversity is main-ly explained by seasonality within both sites, but without an effect on species richness, and there is no effect of variability of nutrient-related soil properties (table 1).

DISCUSSION
Our results indicate that seasonality (i.e., changes between dry and wet season) determines the temporal variation of beta diversity of the studied rocky outcrops. This indicates that the predictable seasonal precipitation is closely linked to the community composition on this type of rocky outcrop formation, where there typically is a marked seasonal pattern of water deficit (Silveira et al. 2016). Likewise, it was possible to observe how seasonality can promote similar

Table 1 -Candidate mixed effect models predicting the effect of variability of nutrient-related soil properties (PCA1) and seasons on beta diversity with Gaussian error distribution (linear mixed effects model -lmer).
The models were tested for each site to explain local effects of seasonality and soil variability. The p-value of each predictor is given as: * p < 0.001. beta diversity patterns between seasons at each site. In different rocky outcrops, there are plant communities that are well adapted to extreme environmental conditions, mainly to water and nutrient deficits, as well as shallow soils (Lüttge 1997;Gröger & Huber 2007). These species have different mechanisms and adaptive strategies to tolerate scarce substrates and water, as well as the high temperatures due to the bare rock surface (i.e., Lüttge 1997). Several studies in America and Africa show that the vegetation on rocky outcrops is dominated by succulents, geophytic and xerophytic species, and drought-tolerant species and drought-avoiding species such as therophytes (Lüttge 1997;Gröger & Huber 2007). Our results allow us to confirm the hypothesis that seasonality, based on changes between a dry and a wet season, determines species composition and beta diversity, but not species richness. However, soil fertility determined by variability of nutrient-related soil properties has no effect on species composition and beta diversity. Some plants that are not drought-resistant use drought avoidance as a mechanism to allocate more energy to survival during certain periods, to germinate and grow sufficient aerial biomass during a short period of time, and to conclude their entire life cycle before the end of wet season. For example, therophytes that germinate after the beginning of the wet season can develop in a short period of time while the humidity is relatively high (Lüttge 1997;Gröger & Huber 2007;Villa et al. 2018). Likewise, geophytes are another group of plants that exhibit this mechanism. They store resources in their underground organs (rhizomes, tubers, bulbs) in order to survive periods of drought and to protect them against excessive water loss (Lüttge 1997). Conversely, a classic example of a tolerance strategy in plants is desiccation tolerance, which allows plants to almost complete dehydrate their tissues and cellular protoplasm (between 85 and 95% of the total water content of the plant), and dehydrated and rehydrated between the dry and wet seasons (Dinakar et al. 2012). Reactive photosynthetic activity is usually reactivated within a few days after the onset of precipitation and rehydration of the plants (Lüttge 1997;Dinakar et al. 2012).

Response variable
In this study, we found that, despite the lack of differences in species richness between dry and wet season, there are notable differences in beta diversity. We assume that fine-scale variations in species composition change considerably due to the influence of water availability, resulting in the coexistence of a higher number of species (Bao et al. 2018). Therefore, seasonal environmental variation, such as precipitation, can also be a determining factor in the availability of resources (i.e., water) and the coexistence of herbaceous community species in the same space (e.g., Ulrich et al. 2014;Le Bagousse-Pinguet et al. 2017;Bao et al. 2018). Therefore, it is possible that the availability of water may be considered an important abiotic filter shaping species composition at a fine-scale in the rocky outcrops studied.
Species richness did not show differences between seasons, which is more likely due to species redundancies in the two studied sites, and this pattern may be a response to high environmental stress, mainly the reduced water and nutrient availability (Silveira et al. 2016). However, we found considerable differences in the community composition between seasons. In this sense, we attribute this change in community composition to the predictable effect of seasonally variable precipitation. Moreover, the low water storage capacity of the soil in the rocky outcrops is an additional environmental filter determining the temporal dynamics of the plant communities' diversity.
Our results provide evidence that seasonality shapes the temporal beta diversity in herbaceous communities. However, most of the studies on Brazilian rocky outcrops have analysed changes in woody community composition along environmental gradients, and among habitats and phytophysiognomies within the campo rupestre s. lat. (e.g., Alves et al. 2014;Nunes et al. 2015;Prata et al. 2018;Silva et al. 2019). On the other hand, the majority of studies on plant communities on rocky outcrops, using an environmental gradient approach, have analysed the direct effects of physico-chemical parameters of the soil on the community diversity and structure (e.g., Nunes et al. 2015;Neri et al. 2016). Overall, these studies concluded that the spatial variation of resources in these edaphic environments is one of the main factors affecting woody species diversity. However, few studies have evaluated these ecological patterns within the herbaceous communities that represent some of the most important phy- tophysiognomies of the campos rupestres. In fact, these communities are responsible for a large part of the diversity and endemism of these ecosystems (Alves et al. 2014;Fernandes 2016). In contrast, our study reveals that soil fertility does not explain the temporal variation in species richness and composition of herbaceous communities.
Therefore, we emphasize that, when studying the diversity of herbaceous communities, one should consider both spatial variations using the environmental gradients approach, as well as seasonal environmental variation. It has been reported that in rocky outcrops (i.e., campos rupestres, inselbergs), where there is a considerable water deficit, annual short cycle species (i.e., ephemeral plants) are common (Silveira et al. 2016;Villa et al. 2018). These plants have developed an adaptive strategy to avoid the dry season and we presume that they are responsible for the considerable species turnover during the dry season, when these ephemeral species are absent, thus determining the high beta diversity.
The high beta diversity may reflect a higher habitat heterogeneity, which can be induced by seasonal environmental changes (Heino et al. 2015;Tonkin et al. 2017;Bao et al. 2019), and species establish according to their degree of ecological tolerance (Pott & Silva 2015;Tonkin et al. 2017). We presume that despite the lack of difference in beta diversity between the seasons, and the effect of seasonality on beta diversity, it is observed that environmental heterogeneity promoted by seasonality may lead to an increase in the number of coexisting species. In this context, seasonality probably causes divergence of the community composition (high beta diversity) by increasing environmental filtering (Myers et al. 2015; Bao et al. 2019). For example, the seasonality of precipitation can regulate the water availability in the soil, which, in turn, regulates species richness and the abundance of seedlings from the seed bank (Bao et al. 2017).

CONCLUSION
The herbaceous species composition and beta diversity is considerably different between seasons on the two rocky outcrops studied. However, species richness did not differ between seasons at either site, which had similar species turnovers that maintained the differences in species composition. Our result corroborates the hypothesis that seasonality determined by changes between the dry and the wet season determines beta diversity, but not species richness. However, soil fertility determined by variability of nutrient-related soil properties has no effect on species composition and beta diversity. Finally, we suggest that future studies on herbaceous community assemblies in lithologically different rocky outcrops should evaluate the temporal effect of the environment, such as the seasonal variation of precipitation; as well as the species responses to the temporal resource variability.

SUPPLEMENTARY FILES
Five supplementary files are associated with this paper: