Assessing coastal species distribution models through the integration of terrestrial, oceanic and atmospheric data

Abstract

Aim When considering multiple species in distribution models, environmental variables should be selected that describe the group of species' environmental requirements. This can, however, be challenging in locations such as coastal areas, where different species may respond to terrestrial, oceanic and/or atmospheric conditions. Here, we evaluate the use of remotely sensed (RS) terrestrial, oceanic and interpolated climate variables, as well as a more detailed shore‐zone data set, as a means of modelling the distributions of coastal bird species with diverse habitat requirements. Location Coastal British Columbia, Canada. Methods Boosted regression trees were used to model the distributions of 60 species of coastal birds using each environmental variable group, run individually and in combination. Models were assessed for their predictive ability and model fit, as well as for model overfitting. Results Models that incorporated terrestrial data were found to produce the highest model fit and predictive ability of the broad‐scale environmental groups. Incorporating the fine‐scale shore‐zone data offered little improvement, as did selecting the best model in terms of predictive ability by species. Model fit and predictive ability were also found to vary by functional feeding group, with insectivores and benthivores being the best‐modelled. Main conclusions Having access to both broad‐scale RS environmental data and more detailed coastal shore‐zone data produced modest improvements over employing RS environmental data alone. Models using only the terrestrial data set, however, performed similarly to the best single model type (terrestrial + shore‐zone), indicating that broad‐scale environmental data also offer an effective means of estimating coastal bird distributions. Testing multiple environmental variable groups in different combinations and selecting the best model allowed models to be optimized by species; conversely, the results of a ‘one model fits all’ approach were comparable to those of the best models, indicating that a single‐model approach is also valid.

The grazing impacts of four barren ground caribou herds (Rangifer tarandus groenlandicus) on their summer ranges: An application of archived remotely sensed vegetation productivity data

“…While originally applied as an index to describe plant communities (Coops et al, 2009a;Fitterer, Nelson, Coops, & Wulder, 2012) and animal diversity (Coops, Wulder, & Iwanicka, 2009b;Andrew, Wulder, Coops, & Baillargeon, 2012;Fitterer, Nelson, Coops, Wulder, & Mahony, 2013;Rickbeil, Coops, Drever, & Nelson, 2014b), DHI is also a useful predictor of individual coastal bird species distributions (Rickbeil et al, 2014a) and for describing forage conditions for moose (Alces alces) in Ontario, Canada (Michaud et al, 2014). The DHI estimates three components of landscape productivity -the yearly sum or overall productivity, the seasonality (the change between the maximum and minimum productivity throughout the year), and the minimum annual productivity (not considered here as all arctic vegetation goes to 0 in terms of fPAR values owing to the short growing season).…”

Predicting environmental mitigation requirements for hydropower projects through the integration of biophysical and socio-political geographies

“…Since one of our goals was spatial prediction beyond the spatial extent of our dataset, we did not implement methods for accounting for spatial autocorrelation because previously developed methods do not allow for prediction beyond the dataset (Dormann et al, 2007;Rickbeil et al, 2014). We recognize that our models did not use an independent validation dataset, but rather a split of our original dataset.…”

“…A full list of each of the mitigation types catalogued in the database and the percent of times each was required, including each of the 132 Tier 3 categories, is presented in Appendix A. Descriptions of each of the Tier 3 categories are provided in Appendix A of Schramm et al (2016). Predictive models were built only if a mitigation type was required for at least 5% (Rickbeil et al, 2014) of the plants in the mitigation database. Models were not built for the very broad Tier 1 categories.…”

“…Before running the models, all predictor variables were assessed for collinearity using Pearson's correlation coefficients (r). When r values exceeded 0.7 (Dormann et al, 2013), the variable deemed more functionally applicable to hydropower mitigation (Arganaraz et al, 2015;Rickbeil et al, 2014) or that was derived at a higher spatial resolution was retained ( Table 1). The data were split into training (80%) and validation (20%) data using the caret package in R, which creates random splits within each class so that the overall class distribution is preserved as well as possible (Kuhn, 2008).…”

“…Predictive models were built only if a mitigation type was required for at least 5% (Rickbeil et al, 2014) of the plants in the mitigation database, resulting in 57 Tier 3 mitigation types being modelled and all 20 of the Tier 2 mitigations being modelled (see Table 2 for modeling results). Eight of the 57 Tier 3 models were rejected due to either a CV ROC or validation ROC b0.7, leaving 49 Tier 3 models with at least an acceptable fit.…”