The drivers of dark diversity in the Scandinavian mountains are metric‐dependent

Abstract

QuestionDark diversity refers to the set of species that are not observed in an area but could potentially occur based on suitable local environmental conditions. In this paper, we applied both niche‐based and co‐occurrence‐based methods to estimate the dark diversity of vascular plant species in the subarctic mountains. We then aimed to unravel the drivers explaining (a) why some locations were missing relatively more suitable species than others, and (b) why certain plant species were more often absent from suitable locations than others.LocationThe Scandinavian mountains around Abisko, northern Sweden.MethodsWe calculated the dark diversity in 107 plots spread out across four mountain trails using four different methods: two co‐occurrence‐based (Beals’ index and the hypergeometric method) and two niche‐based (the climatic niche model and climatic niche model followed by species‐specific threshold). We then applied multiple Generalized Linear Mixed‐Effects Models and General Linear Models to determine which habitat characteristics and species traits contributed the most to dark diversity.ResultsThe study showed a notable divergence in the predicted drivers of dark diversity depending on the method used. Nevertheless, we can conclude that plot‐level dark diversity was generally 17% higher in areas at low elevations and 31% higher in areas with a low species richness.ConclusionOur findings call for caution when interpreting statistical findings of dark‐diversity estimates. Even so, all analyses point toward an important role for natural processes such as competitive dominance as the main driver of the spatial patterns found in dark diversity in the northern Scandes.

A roadmap to carefully select methods for dark‐diversity studies

“…In case of poor performance of some methods, it is possible to rely on one or a few more suitable methods. Below, we provide a brief overview of existing dark-diversity methods, and use the data and questions from Hostens et al (2023) to discuss the challenges of various approaches and evaluate their adequacy for the particular case study.…”

“…An obvious limitation of this method is that nature is continuous but habitat types are discrete units, which complicates the assignation of sites into predefined habitat types. Neither expert opinions nor classification of habitat types were readily available for the study area in Hostens et al (2023); therefore, we did not consider this family of methods further.…”

“…Species distribution modelling is another suitable approach, also used by Hostens et al (2023). This method links the presence and absence of different species to ecological conditions, and predictions can be made for sites where the same set of ecological variables is known (Elith & Leathwick, 2009).…”

“…In a recent study published in this journal, Hostens et al (2023) applied the dark-diversity concept and compared the outcomes of different estimation methods of dark diversity. While the authors agree on the usefulness of the concept, they also worry about some methods yielding different results.…”

“…Beside that, we also note that methodological comparisons would benefit if dark-diversity size is standardized similarly (Appendix S1, Figure S1), and if standardized dark-diversity size is related to descriptive parameters that are mathematically not directly related (Appendix S1, Figure S2). Hostens et al (2023) used dark-diversity estimations of plants in a region within the Scandinavian mountains to answer two main questions: '(a) why some locations were missing relatively more suitable species than others, and (b) why certain plant species were more often absent from suitable locations than others'. They used data from 107 square vegetation plots comprising 73 plots of 1 m 2 and 34 plots of 100 m 2 , where records on 49 plant species with at least 10 occurrences were available.…”