Two Connectivity Metrics and Their Relation with Fish Species Richness and Composition in Southern Brazilian Coastal Lakes

Habitat area, productivity, heterogeneity and disturbance are known to influence species richness and composition in aquatic communities. However, connectivity may also have a considerable importance where colonization dynamics is an important process. For example, poorly-connected patches (e.g., lakes) may have different communities when compared to well-connected patches, even when environmental conditions are similar.  In spite of its importance, measuring connectivity in aquatic systems is still a work in progress. We introduce two connectivity metrics for use in lake systems, and test their potential to explain fish species richness and composition. The first proposed metric is Primary Connectivity (PC), which measures the connectivity between a lake and all lakes directly connected to it. PC is a function of cost-distance, number of connections and area of connected lakes. The other metric is the Estuarine Connectivity (EC), which is the cost-distance from the estuarine connection to the sea to each lake in the system. Cost-distances incorporate the idea that connectivity is a function of both the distance between lakes and the resistance that different types of connections impose on fish  dispersal. Cost-distances were calculated with a cost-grow algorithm in the GIS Idrisi Andes, and using 4 levels of connection resistance to dispersal. We calculated connectivity metrics for the 41 lakes draining to the Tramandaí estuary (southern Brazil), and extracted the values for the 22 lakes for which fish data were available. We used multiple regression with PC, EC, lake area (A) and euclidean distance to the sea (DSC) as predictors of species richness. The relation between species composition and connectivity was examined using Mantel correlation tests. PC, EC, and area were important for predicting species richness (best model, Akaike´s AIC). However, differences in species composition were correlated only with EC (r = 0.4, p = 0.01), with no influence from PC, lake area or DSC, indicating that the main differences in composition among lakes were related to the estuarine-freshwater gradient. We suggest that PC and EC may be useful as connectivity metrics in investigations of fish community patterns in lakes. PC and EC are not mutually correlated and measure different aspects of connectivity. PC, emphasizes neighboring lakes as sources of colonizers, while EC is a larger extent measure of connectivity. We also suggest that different connectivity indices may be developed according to the particular lake system (coastal lakes, floopplain lakes, mountain lakes) and ecological problem in question.