Multiscale Environmental Influences on Fish Assemblage Structure of South Atlantic Coastal Plain Streams

Abstract

South Atlantic coastal plain streams are unique and understudied freshwater environments that provide crucial habitats for a wide range of aquatic taxa. We investigated patterns in fish assemblages across South Carolina's coastal plain, and developed statistical models to identify the dominant multiscale abiotic environmental factors that influence assemblage structure. Cluster and indicator species analyses of fish assemblage data collected at 208 wadeable streams indicated the presence of four predominant species associations that commonly occur in the South Atlantic coastal plain, which we termed the (1) fluvial, (2) Eastern Mudminnow, (3) centrarchid, and (4) nonfluvial assemblages. A random forest analysis indicated that geographic gradients and instream habitat variables associated with velocity, channel form, stream size, depth, and large wood generally played a greater role in distinguishing fish assemblages than catchment land cover. A follow‐up principal component analysis showed that these instream habitat variables showed weak relationships with current anthropogenic land‐cover conditions. We suggest several possible explanations for our findings, including (1) a sustained dynamic equilibrium among catchment, riparian, and instream conditions over time may encourage assemblage partitioning among differential instream habitats, (2) past landscape conditions may have a greater influence on current instream habitat conditions and fish assemblage structure than current landscape conditions, (3) weak relationships between catchment land cover and instream habitats may be common in low‐elevation regions, and (4) seasonal variation in hydrologic patterns may dominantly influence instream habitat conditions and fish assemblage structure. Our study adds essential information towards a better understanding of how South Atlantic coastal plain fish assemblages respond to environmental factors across multiple spatial scales. Such knowledge will help improve management and conservation strategies, as well as assist in the development of appropriate indicators for standardized evaluations of ecological integrity.

State‐ and Regional‐Scale Patterns and Drivers of Freshwater Fish Functional Diversity in the Southeastern USA

“…; Marion et al. ). These environmental variables served as the predictor variables in our Random Forest models (see Data Analysis section below) to determine drivers of functional diversity across South Carolina streams.…”

“…; Marion et al. ). A standard set of quality assurance and quality control (QA–QC) procedures were followed for all field measurements and data entered.…”

“…; Marion et al. ). Due to our inability to procure environmental or land cover data to match the temporal scale associated with our SALCC fish sampling data, we did not consider environmental or land cover drivers at the regional scale.…”

“…; Marion et al. ). Random Forest is a machine learning extension of CART (i.e., Classification and Regression Trees) in which a “tree” is constructed to partition multivariate data into homogenous groups.…”

“…Environmental variable data.-At the state scale, environmental and land cover data were collected or quantified within stream reaches at all sampling locations following standard protocols developed by the SCDNR (Scott et al 2009;Marion et al 2015). These environmental variables served as the predictor variables in our Random Forest models (see Data Analysis section below) to determine drivers of functional diversity across South Carolina streams.…”