Biogeographical patterns of species richness in stream diatoms from southwestern South America

Abstract

The latitudinal diversity gradient (LDG) hypothesis has been validated for many taxon groups, but so far, stream diatoms have not conformed to this pattern. Research on diatoms that includes data from South America is lacking, and our study aims to address this knowledge gap. Previous studies have successfully explained stream diatom species richness by considering niche dimensionality of physicochemical variables. Moreover, in southwestern South America, the observed biogeographical pattern differs from LDG and has been shown to be determined by historical factors. We used a dataset comprising 373 records of stream diatom communities located between 35° S and 52° S latitude, southwestern South America. The dataset included physicochemical river water variables, climate data, and ice sheet cover from the Last Glacial Maximum. We explored geographical patterns of diatom species richness and evaluated 12 different causal mechanisms, including climate-related theories, physicochemical and climatical exploratory analyses, historical factors, and niche dimensionality. A metacommunity analysis was conducted to evaluate the possible nested structure due to historical factors. We observed an increase in diatom species richness from south to north. Models containing both physicochemical and climatic predictors explained the highest proportion of variation in the data. Silica, which was correlated with latitude, and flow velocity, which did not show any spatial pattern, were the most important predictors. Historical factors and nested structure did not play any role. Contrary to what has been reported in the literature, we found no support for climate-related explanations of species richness. Instead, theories related to niche dimensionality and local factors provided better explanations, consistent with previous related research. We suggest that the increase in diatom richness in the north of our study region is due to a higher nutrient supply in these rivers, rather than a due to larger species pool in the area.

Keywords: Clementsian structure; Glasonian structure; PATICE; flow velocity; latitudinal diversity gradient; number of limiting resources; silica.

FIGURE 1

Spatial distribution of the 373 study sites. Filled circles show the geographical distribution of the sampling locations in southern Chile. The inset map shows the location of the study area in South America.

FIGURE 2

Species richness and estimated species richness in relation to latitude (DD = decimal degree) for the Complete dataset (373 points). (a) Species richness per site. (b) Species richness per latitudinal band. (c) Estimated species richness and confidence interval for each latitudinal band. The smoothed line represents the GAM prediction for the “richness ~ latitude” model, and the gray area is its confidence interval. Solid line = significant model. Dashed line = non‐significant model.

FIGURE 3

Correlation matrix between predictor variables based on correlation coefficient r (Gradient color) and correlation significance following Pearson tests (***p < .001; **p < .01; *p < .05). Species richness = Rich. For spatial variables: Lat = latitude, Lon = longitude. For climatical variables: T_mean = Annual mean temperature, T_max = Maximum temperature during the warmest month, T_min = minimum temperature of coldest month, T_sd = temperature seasonality. For physicochemical variables: T_insitu = In situ temperature, EC = Electrical conductivity, Osat = Oxygen saturation, TP = total phosphorus, SiO₂ = Silica, Vel = Flow velocity, NLR = Number of limiting nutrient. For historical factors: Glac = Proportion of time under ice sheet the last 35 ka per site. Plot designed with R function corrplot::corrplot (Wei & Simko, 2021).

FIGURE 4

Single‐predictor relationships between species richness and the significant predictor variables from GAMs for the Complete dataset (373 points). The smoothed line represents the GAM prediction and the gray area its confident interval. Solid line = significant model. (a) Diatom richness in relation to cell density and smoothed line from Species‐energy theory model. (b) Species richness in relation to annual maximal temperature (T_max). (c) Species richness in relation to annual temperature seasonality (T_sd). (d) Species richness in relation to annual average temperature (T_mean). For b–d, smoothed splines are from the Climatical effect model. See text for more details.

FIGURE 5

Single‐predictor relationships between species richness and the significant predictor variables from GAMs for the Chemical dataset (182 points). The smoothed line represents the GAM prediction and the gray area its confident interval. Solid line = significant model. Species richness is shown in relation to (a) in situ temperature (T_insitu), (b) pH, (c) silica (SiO₂), (d) saturation oxygen (Osat), (e) annual mean temperature (T_mean), (f) flow velocity (Vel), and (g) number of limiting resources (NLR). a–f are smoothed splines from the total effect model and (g) is a linear model from the Niche dimensionality model.