Niche and neutral-based processes differ in importance for common and rare species in a metacommunity of anurans in subtropical grasslands

Most species-rich communities consist of a limited number of common species and many rare species. This widespread ecological pattern can arise due to the predominance of either niche-based or neutral-based processes in community structuring. We tested two hypotheses related to this pattern for an anuran metacommunity composed of 35 species distributed in 187 ponds in the South Brazilian grasslands: (1) for common species, niche processes should prevail over neutral ones because common species evolved under a wider range of ecological filters than rare species; (2) for rare species, neutral processes should prevail because rare species are more prone to ecological drift and dispersion limitation. We found that the densities of both common (i.e., the 25% higher density ones) and rare (i.e., the 75% lower density ones) species depended mostly on pond structure and land use, indicating a similar response to niche-based processes in both groups of species. The importance of pond descriptors and type of land use, however, differed between the two groups, confirming that common and rare species represent ecologically distinct groups within the metacommunity. In addition, neutral processes—related to dispersion limitation or not—influenced both species groups, but the spatial filters selected explained a smaller portion of the variation in species densities compared to environmental descriptors. Our findings showcase the prevalence of niche processes in structuring communities of common and rare species of anurans in subtropical grasslands and indicate the variables of pond structure and land use that are relevant to guide conservation practices for the metacommunity of anurans in the South Brazilian grasslands. 

Methods
The study was conducted in the South Brazilian grasslands located in the state of Rio Grande do Sul, Brazil. Data were collected from 187 ponds using the 'survey at breeding sites' method.

For each pond, 36 environmental descriptors were measured. To accomplish this, we visually measured each of the 34 pond descriptors (e.g., type of pond edge, depth, area, type of vegetation, pH, turbidity) following the methodology outlined by Iop et al. (2020). Additionally, we recorded two variables related to the sampling date that could potentially impact the number of records: the length of the photoperiod and the percentage of lunar disk (moon phase).

Landscape descriptors were determined through digital land use and land cover (LULC) mapping for the Landscape Sampling Units (LSUs), considering the base years of 2002. The scale was based on two false-color compositions. These landscape descriptors correspond to the area of each LULC class around each LSU. We employed three buffer zones (1 km, 3 km, and 5 km) to calculate these descriptors, taking into account the reported anuran dispersion range in the literature, which varies between 400 m and 10 km. In total, nine landscape descriptors (Agriculture, Forest, Pasture, Rocky Outcrop, Sandy, Forestry, Urban, Water, and Wetland) were measured within the buffers generated by the three rays.

To assess whether geographic space influences the distribution of common and rare species in the metacommunity, we generated Moran’s eigenvector maps (MEM) using the geographic coordinates of each pond as a proxy for spatial patterns related to dispersion and ecological drift.

Redundancy analysis (RDA) was employed to evaluate the role of environmental characteristics (local and landscape descriptors) and spatial patterns in the organization of the common and rare anuran community. RDA combines multiple regression with principal component analysis to identify a series of linear combinations of explanatory variables (i.e., local environmental descriptors, landscape descriptors, and spatial filters) that best explain the variation in the response matrix (i.e., density of species in each survey site).