Responses to habitat fragmentation across taxa
Except for herptiles, all taxa assessed in our analysis typically accumulate more species across a large number of small patches than a few large patches of the same total area. Conserving small patches will be especially important for conservation of the Earth’s flora, given that the SL > SS was extremely rare when assessing plants (Fig. 2). One reason for this pattern might be that plants were the most diverse taxon in our analysis, with typically more than 100 species in their metacommunities. Communities with more species usually have higher turnover due to niche partitioning (Finke & Snyder 2008; Chesson 2018), and high turnover is an important process determining SS > SL (Deane et al. 2020; Fahrig et al. 2022). In addition, plants might be especially likely to display SS > SL due their tendency to aggregate in space. More specifically, several small patches are more likely to intersect with more species distributions than few large patches when those distributions are clumped in space (May et al. 2019; Fahrig et al. 2022).
We also found that mammals, birds, and invertebrates were more speciose in sets of many small patches than few large ones (Figs. 2,3). This might seem inconsistent with evidence that, for these taxa, occurrence probability is typically higher in large than small patches (e.g., Schultz & Crone 2005; Keinath et al. 2017). Indeed, based on this pattern Timmers et al. (2022) recently concluded that “[c]onservation of threatened bird species in fragmented landscapes should preferably focus on strict protection of large forest ”. However, this type of inference is based on comparing single small to single large patches. Using Timmers et al. (2022) estimates it is in fact more likely to encounter a threatened bird species somewhere in a set of a thousand 1-ha forest patches than in one 1000-ha patch. This example highlights how recommendations on reserve design at the landscape scale must be based on landscape-scale analyses (Fahrig et al. 2019).
In contrast with all other taxa, for herptiles SL > SS was over twice as likely as SS > SL (Fig. 2, 3). We speculate that these species are particularly vulnerable to dispersal mortality in the matrix. Many amphibians and reptiles require disjunct habitats at different times of the year and so must move between them, and in a landscape with many small patches, such movements are likely to take them through the matrix. Being non-volant and slow-moving, they may be more at risk from various mortality factors in the matrix than other taxa. For example, Rytwinski & Fahrig (2012) found that amphibians and reptiles are the taxa most impacted by roads and traffic, likely through road kill. The high dispersal mortality would lead to smaller populations of these species across many small patches than few large ones, increasing the likelihood of extinction. We note, however, that even for herptiles there are more species in several small than few large patches in nearly half of the datasets (4 of 9).
In addition to taxonomic effects, we found negative effects of patch size evenness on the likelihood of finding positive fragmentation effects on biodiversity (Riva & Fahrig 2022) (Fig. 3). In other words, as the sizes of small and large patches in a dataset become more different, and thus more small patches are needed to equal the area of a large patch, it becomes more likely that sets of small patches will consistently harbor more species than a set of a few large patches. This result is important because it contradicts a major theoretical argument used to prioritize large patches in conservation, i.e., that patch-level extinction risk determines biodiversity in sets of patches (Diamond 1975; Riva & Fahrig 2022). Landscape-scale processes, e.g., turnover or metacommunity dynamics (Chase et al. 2020b; Deane et al.2020; Fahrig et al. 2022), appear to be more important than patch-scale extinction risk in determining biodiversity in fragmented landscapes.