Predicting species distributions from museum and herbarium records using multiresponse models fitted with multivariate adaptive regression splines

Abstract

Current circumstances — that the majority of species distribution records exist as presence‐only data (e.g. from museums and herbaria), and that there is an established need for predictions of species distributions — mean that scientists and conservation managers seek to develop robust methods for using these data. Such methods must, in particular, accommodate the difficulties caused by lack of reliable information about sites where species are absent. Here we test two approaches for overcoming these difficulties, analysing a range of data sets using the technique of multivariate adaptive regression splines (MARS). MARS is closely related to regression techniques such as generalized additive models (GAMs) that are commonly and successfully used in modelling species distributions, but has particular advantages in its analytical speed and the ease of transfer of analysis results to other computational environments such as a Geographic Information System. MARS also has the advantage that it can model multiple responses, meaning that it can combine information from a set of species to determine the dominant environmental drivers of variation in species composition. We use data from 226 species from six regions of the world, and demonstrate the use of MARS for distribution modelling using presence‐only data. We test whether (1) the type of data used to represent absence or background and (2) the signal from multiple species affect predictive performance, by evaluating predictions at completely independent sites where genuine presence–absence data were recorded. Models developed with absences inferred from the total set of presence‐only sites for a biological group, and using simultaneous analysis of multiple species to inform the choice of predictor variables, performed better than models in which species were analysed singly, or in which pseudo‐absences were drawn randomly from the study area. The methods are fast, relatively simple to understand, and useful for situations where data are limited. A tutorial is included.

Addressing longitudinal connectivity in the systematic conservation planning of fresh waters

“…MARS fits a nonlinear function to the relationships between dependent and predictor variables by breaking the range of each predictor into a subset of portions or "knots", and fitting linear relationships for each of them (basis functions). MARS allows the slope of the fitted linear segments between pairs of segments to vary while ensuring that the full fitted function is without (Elith & Leathwick, 2007). The predictive function is finally composed of a series of connected straight line segments, rather than the smooth curve of a GAM.…”

“…A Multivariate Adaptive Regression Splines (MARS) model developed on the 151 sampled planning units was used to predict the probability of occurrence for each species in the unsampled planning units. MARS is a method of flexible non-parametric regression modelling (Elith & Leathwick, 2007). It is useful for modelling complex non-linear relationships between response and explanatory variables with similar levels of complexity to that of a Generalized Additive Model (GAM) (Hastie, 1991).…”

“…Deviance complements AUC because it expresses the magnitude of the deviations of the fitted values from the observations. Analogous to Elith & Leathwick (2007) the full information given in the predictions (raw probabilities of occurrence) Prior to model construction a Principal Component Analysis was carried out on the environmental data to extract a reduced number of independent predictors for MARS models. The presence-absence of 10 species in the whole biological data set was used for fitting the models (n=151 planning units).…”

“…The model was fitted using a code provided by Elith & Leathwick (2007) We allowed first order interactions between predictors in the models, since previous analysis showed a significant improvement in model performance when they were included. For more statistical details see Leathwick et al (2005).…”

Predicting fish species richness and assemblages with climatic, geographic and morphometric factors: A broad-scale study in Chinese lakes

“…Although presence-absence data is simple, previews researches have verified its advantages in modeling species assemblages and yielding more accurate predictions (Cawsey et al, 2002;Elith and Leathwick, 2007). It has been already widely and successfully used in studying fish distribution patterns, species richness and assemblages recently (Ibarra et al, 2005;Park et al, 2006;He et al, 2011).…”