Associations Between Habitat Quality and Body Size in the Carpathian-Podolian Land Snail Vestia turgida (Gastropoda, Clausiliidae): Species Distribution Model Selection and Assessment of Performance

Abstract

Species distribution models (SDMs) are generally thought to be good indicators of habitat suitability, and thus of species’ performance. Consequently SDMs can be validated by checking whether the areas projected to have the greatest habitat quality are occupied by individuals or populations with higher than average fi tness. We hypothesized a positive and statistically signifi cant relationship between observed in the fi eld body size of the snail V. turgida (Rossmässler, 1836) and modelled habitat suitability, tested this relationship with linear mixed models, and found that indeed, larger individuals tend to occupy high-quality areas, as predicted by the SDMs. However, by testing several SDM algorithms, we found varied levels of performance in terms of expounding this relationship. Marginal R2 expressing the variance explained by the fi xed terms in the regression models, was adopted as a measure of functional accuracy, and used to rank the SDMs accordingly. In this respect, the Bayesian additive regression trees (BART) algorithm gave the best result, despite the low AUC and TSS. By restricting our analysis to the BART algorithm only, a variety of sets of environmental variables commonly or less used in the construction of SDMs were explored and tested according to their functional accuracy. In this respect, the SDM produced using the ENVIREM data set gave the best result.

Anthropogenic load іs a leading factor in the morphological variability of Chondrula tridens (Gastropoda, Enidae) in the northwestern Azov Sea region

“…There is a relationship between habitat quality and body size of land snails. Animals reach their largest sizes in optimal conditions (Mammola et al, 2019;Tytar, 2021). The data obtained indicate that the total shell sizes of mollusks from different biotopes were quite homogeneous.…”

Evaluating ecological uniqueness over broad spatial extents using species distribution modelling

“…An ensemble of SDM algorithms could also be used to capture a broader range of possible outcomes for the LCBD predictions. However, given the high performance of BARTs in model comparisons (Konowalik and Nosol 2021, Tytar and Baidashnikov 2021) and the large extent we covered, we do not believe the changes to the LCBD gradients would be strong enough to affect our results in a meaningful way.…”

“…Trees are first constrained as weak learners by priors regarding structure and nodes, then updated through an iterative Bayesian backfitting Markov Chain Monte Carlo (MCMC) algorithm which ultimately generates a posterior distribution of predicted classification probabilities (Chipman et al 2010, Carlson 2020). In the context of species distribution modelling, BARTs showed high performance when compared to other predictive algorithms (Konowalik andNosol 2021, Tytar andBaidashnikov 2021). We first performed BARTs separately for all species and estimated the probability of occurrence in all the sites of our region of interest using the posterior median.…”

Modeling the Distribution of the Chytrid Fungus Batrachochytrium dendrobatidis with Special Reference to Ukraine

“…Second, our model included a set of 16 climate and two topographic variables (the ENVIREM dataset downloaded from http://envirem.github.io (accessed on 20 January 2023); Table 1), many of which were likely to have direct relevance to ecological or physiological processes determining species distributions [49][50][51][52]. The included topographic variables were potentially important as well because they could modify the effect of the climate descriptors.…”