Predicting Species Distributions from Samples Collected along Roadsides

Abstract

Predictive models of species distributions are typically developed with data collected along roads. Roadside sampling may provide a biased (nonrandom) sample; however, it is currently unknown whether roadside sampling limits the accuracy of predictions generated by species distribution models. We tested whether roadside sampling affects the accuracy of predictions generated by species distribution models by using a prospective sampling strategy designed specifically to address this issue. We built models from roadside data and validated model predictions at paired locations on unpaved roads and 200 m away from roads (off road), spatially and temporally independent from the data used for model building. We predicted species distributions of 15 bird species on the basis of point-count data from a landbird monitoring program in Montana and Idaho (U.S.A.). We used hierarchical occupancy models to account for imperfect detection. We expected predictions of species distributions derived from roadside-sampling data would be less accurate when validated with data from off-road sampling than when it was validated with data from roadside sampling and that model accuracy would be differentially affected by whether species were generalists, associated with edges, or associated with interior forest. Model performance measures (kappa, area under the curve of a receiver operating characteristic plot, and true skill statistic) did not differ between model predictions of roadside and off-road distributions of species. Furthermore, performance measures did not differ among edge, generalist, and interior species, despite a difference in vegetation structure along roadsides and off road and that 2 of the 15 species were more likely to occur along roadsides. If the range of environmental gradients is surveyed in roadside-sampling efforts, our results suggest that surveys along unpaved roads can be a valuable, unbiased source of information for species distribution models.

Coupling Traditional Monitoring and Citizen Science to Disentangle the Invasion of Halyomorpha halys

“…The occurrence data collected on BugMap platform exhibit a well-known geographical bias pattern termed travel-time bias (Table S2) [38,66]. Reports are especially concentrated in the urban area of Trento and in the neighboring villages, while records from open field are rare.…”

A Bayesian method for assessing multi-scale species-habitat relationships

“…During April-July of 2016, we conducted multi-species 500 m fixed-radius aural point count surveys (Robbins et al 1986) at roadside sites (Mccarthy et al 2012) located across the State of Nebraska. Spatially balanced survey sites were selected to reflect the background landcover composition such that predictors of interest were sampled along a gradient from low to high proportions in amounts found within the State.…”

Off-road sampling reveals a different grassland bird community than roadside sampling: implications for survey design and estimates to guide conservation

“…Bird distribution and abundance may be affected by roads through avoidance, increased mortality from vehicles, decreased mating success, increased nest predation, and habitat changes associated with roads (Lima and Valone 1991, Forman and Alexander 1998, Spellerberg 1998, Forman et al 2002, Fletcher and Koford 2003, Benítez-López et al 2010. Such effects may exist several hundred meters or even several kilometers from roads (Benítez-López et al 2010), perhaps depending on which of the above mechanisms is operating (e.g., Koper et al 2009, McCarthy et al 2012. Restricting surveys to roadsides may limit the representativeness of BBS data if habitat composition differs away from roads.…”

“…Restricting surveys to roadsides may limit the representativeness of BBS data if habitat composition differs away from roads. Such data may result in skewed abundance, distribution, and community composition data (Thogmartin et al 2006, Betts et al 2007, Niemuth et al 2007, McCarthy et al 2012, potentially reducing the reliability of associated population and trend estimates and distribution models that are developed to guide conservation-related programs (Bart et al 1995, 2004, Hutto et al 1995, Francis et al 2005, Sauer et al 2005. Therefore, quantifying and accounting for potential sources of bias from using roadside BBS data is one of the greatest priorities for expanding the utility of this valuable continental survey (Francis et al 2005, Sauer et al 2005, Betts et al 2007, McCarthy et al 2012.…”

“…These count data are increasingly used to achieve other goals, such as generating continental population estimates (e.g., Rich et al 2004, Thogmartin et al 2006) and building species distribution models (e.g., Scott et al 2002, Thogmartin et al 2004, Guisan and Thuiller 2005, Niemuth et al 2005, Murray et al 2008. When combined with Geographic Information Systems (GIS) and remote sensing data, spatial distribution models can predict species occurrences and can estimate species' distributions over large areas (Scott et al 2002, McCarthy et al 2012. However, such spatial extrapolations most often assume that sampling is representative of the habitat and bird community across the broader landscape.…”