Can Species Distribution Models Aid Bioassessment when Reference Sites are Lacking? Tests Based on Freshwater Fishes

Abstract

Recent literature reviews of bioassessment methods raise questions about use of least-impacted reference sites to characterize natural conditions that no longer exist within contemporary landscapes. We explore an alternate approach for bioassessment that uses species site occupancy data from museum archives as input for species distribution models (SDMs) stacked to predict species assemblages of freshwater fishes in Texas. When data for estimating reference conditions are lacking, deviation between richness of contemporary versus modeled species assemblages could provide a means to infer relative biological integrity at appropriate spatial scales. We constructed SDMs for 100 freshwater fish species to compare predicted species assemblages to data on contemporary assemblages acquired by four independent surveys that sampled 269 sites. We then compared site-specific observed/predicted ratios of the number of species at sites to scores from a multimetric index of biotic integrity (IBI). Predicted numbers of species were moderately to strongly correlated with the numbers observed by the four surveys. We found significant, though weak, relationships between observed/predicted ratios and IBI scores. SDM-based assessments identified patterns of local assemblage change that were congruent with IBI inferences; however, modeling artifacts that likely contributed to over-prediction of species presence may restrict the stand-alone use of SDM-derived patterns for bioassessment and therefore warrant examination. Our results suggest that when extensive standardized survey data that include reference sites are lacking, as is commonly the case, SDMs derived from generally much more readily available species site occupancy data could be used to provide a complementary tool for bioassessment.

Rapid Assessment for Identifying Species of Greatest Conservation Need: Towards a Unified Approach

“…We retained relative abundances provided by the authors as calculated ([N of individuals of a species]/[total N of individuals within the fish community] × 100). In other examples, studies sampling a site once tend to underestimate species richness for an area more so than studies sampling a site multiple times over a year (Labay et al 2015), and false absences (failure to detect a species when present; Gu and Swihart 2004;Denes et al 2015) are common sampling problems even when sites are sampled with multiple gear types over extended sampling periods. Among the 24 data sets used in this study, source materials were from the published literature (57%), student theses (22%), unpublished data (17%), and agency reports (4%).…”

“…For example, seines do not capture fish that are small enough to pass through the mesh or fish that are large enough to outswim the net (Hayes et al 1996); therefore, seines have a tendency to underestimate species occurrence and relative abundance within the community (Bayley and Peterson 2001;Fischer and Paukert 2009). In other examples, studies sampling a site once tend to underestimate species richness for an area more so than studies sampling a site multiple times over a year (Labay et al 2015), and false absences (failure to detect a species when present; Gu and Swihart 2004;Denes et al 2015) are common sampling problems even when sites are sampled with multiple gear types over extended sampling periods. Similar to other published studies (e.g., Perkin and Bonner 2011;Kollaus et al 2015), we acknowledge potential limitations in comparing relative abundances and community data among studies.…”

Fisheries Volume 44 Number 10 October 2019

“…For example, seines do not capture fi sh that are small enough to pass through the mesh or fi sh that are large enough to outswim the net (Hayes et al 1996 ); therefore, seines have a tendency to underestimate species occurrence and relative abundance within the community (Bayley and Peterson 2001 ;Fischer and Paukert 2009 ) . In other examples, studies sampling a site once tend to underestimate species richness for an area more so than studies sampling a site multiple times over a year (Labay et al 2015 ) , and false absences (failure to detect a species when present; Gu and Swihart 2004 ;Denes et al 2015 ) are common sampling problems even when sites are sampled with multiple gear types over extended sampling periods. Similar to other published studies (e.g., Perkin and Bonner 2011 ;Kollaus et al 2015 ), we acknowledge potential limitations in comparing relative abundances and community data among studies.…”

A predictive index based on environmental filters for the bioassessment of river basins without reference areas in Atlantic Forest biome, Brazil

“…In such cases, where "minimally disturbed" or "historical condition" sites (sensu Stoddard et al 2006) are missing, some protocols relax the reference criteria and use "best attainable" or "least disturbed" sites as reference (Stoddard et al 2006). The misuse of this practice can generate a series of problems (Chessman 2006, Labay et al 2015, Elias et al 2015. First, these definitions are often arbitrary and inconsistent, thus impractical to apply them in broad spatial scales (Cao & Hawkins 2011).…”