Dataset considerations.
Our distribution model (SDM) approach, rather than the exclusive use of point occurrence records, helped to overcome issues of incomplete sampling (Barthlott et al., 2007; Meyer et al., 2015; Brummitt et al., 2021), and provided broader, macroecological insight into how species in lowland, montane, and alpine regions differ in terms of their occupied climatic niche space. In general, while the quality and extent of species sampling (Fig. 2) was good, we were unable to generate SDMs for every American seed plant. With 68,241 species matched to our phylogeny, we covered only ~18% of estimated global seed plant lineages (Stevens, 2001), and less than ~48% of the total estimated American floral diversity (Ullola Ullola et al., 2017). Further, the threshold minimum of twelve records to build reliable SDMs certainly excluded numerous range-limited endemics, which is likely to disproportionately impact representation of Andean alpine lineages, where the degree of alpine endemism is substantial (Hughes and Eastwood, 2006). With these limitations in mind, we have attempted to draw broad, macroecological conclusions that are unlikely to be significantly impacted by loss of models for range-limited endemics. Nevertheless, species excluded as a result of the conservative nature of our data cleaning protocol highlight the need for increased sampling of range-limited endemics to enhance the broader understanding of how such species impact macroecological patterns and conclusions.
Further, although we have grouped communities by elevational categories, important differences between Northern and Southern Hemisphere alpine habitats, such as the degree of seasonality, growing season length, and extent and duration of snow cover, should be taken into account when forming conclusions from this macroecological dataset. For instance, in our dataset, we did not find a correlation between species richness and elevation in alpine-only communities (data not shown), and it may be that there are many avenues by which a region or local community can achieve a high level of biodiversity (Hughes, 2017). Future studies should investigate how more local-scale processes might be influencing the observed macroecological patterns we find here.
Contrasts across elevation belts and specialization .
A species pool defines a collection of organisms in the region. However, another aspect of a species pool is the subset of that potential community species pool that is able to colonize and survive in the local setting (Emerson and Gillespie, 2008). Although myriad factors determine which plant species are able to persist in montane and alpine regions (Körner, 2003), here we investigated a portion of the macroscale abiotic niche space occupied by American seed plants to provide insight into how factors such as temperature and precipitation might impact species distributions (Moles et al., 2014). Compared to non-alpine species, those inhabiting alpine habitats, regardless of whether such species had ranges centered in the alpine belt or elsewhere, showed unique richness patterns (Fig. 2c), occupied a drastically reduced climatic niche space (Fig. 3d), and showed broad temperature but narrow precipitation niche breadth (Fig. 4d). Elsewhere, we have also shown that these alpine species showed distinct patterns of phylogenetic diversity (Figueroa et al., pers. comm. ). In contrast, species in the montane belt were relatively similar to both the species pool overall and lowland taxa in their observed richness patterns, occupied niche space, niche breadth (Fig. 2-4), and phylogenetic diversity (Figueroa et al., pers. comm. ). These findings underscore the unique confluence of factors that define plant diversity at the extreme elevational limits of the alpine belt (Körner, 2003) and reiterate the complexity of defining an appropriate regional pool for such an ecosystem.
These results showed a notable separation between the occupied climatic niche space of alpine and non-alpine communities. That there was not a similarly clear separation between montane and lowland communities raises the question of what distinguishes montane communities from adjacent assemblies. One distinguishing factor is certainly their greater plant diversity (Grytnes and Vetaas, 2002; Grytnes, 2003; Cardelús, Colwell, Watkins, 2006). Additionally, montane environments at tropical latitudes display greater phylogenetic affinity with temperate lowlands (González-Caro et al., 2020), suggesting cooler-climate corridors have been used to track amenable climate, allowing montane species to colonize lowlands (Donoghue, 2008).
In this study, we found that niche breadth tended to be maximal for montane generalists (Fig. 6), indicating the possibility of montane regions representing a transitional zone, in which the greatest variety of forms from other elevations could coexist. This is conceptually analogous to the often-observed mid-elevation peak in species richness (Grytnes and Vetaas, 2002; Grytnes, 2003; González-Caro et al., 2020). However, our results differ subtly but importantly from such observations in two ways. First, our elevational categories are based on habitat type (Körner et al., 2011) and not absolute elevation per se . Second, we demonstrate that montane communities occupy a broad climatic niche space and have greater niche breadth than alpine communities. These niche characteristics are not identical to species richness, though they could be influenced by it.
Concomitantly, montane communities could also be distinguished from alpine ones by factors other than those examined here, such as soil and mineral conditions (Egli and Poulenard, 2016), pest and pathogen distributions (Rasmann et al., 2014), and/or the distributions of dispersal agents. Additionally, although at the regional scale montane communities overlapped in climatic niche space with lowland sites, this did not mean montane community composition matched that of lowland areas occurring at the same latitude. At the same time, the greatly reduced alpine diversity in Central America might suggest a greater role for dispersal limitation for alpine lineages compared to montane ones (Figueroa et al., pers. comm. ), which could also contribute to explaining differences in distributional patterns between alpine and montane communities in this region.
In addition to contrasting the climatic niches of species occurring within different elevation belts, our SDM approach allowed us to incorporate some biological variation in the climate species experience across their ranges, while also quantifying the fraction of each species’ range occurring in different climatic conditions. We could thus distinguish between species with ranges centered in alpine or montane habitat (i.e., high-elevation ‘specialists’) from those whose ranges extended into higher elevations but were centered in different habitats (i.e., high-elevation ‘generalists’). These distinctions influence what constitutes an alpine species (Körner, 2003) and could reflect different adaptive responses and tolerances. Under our classifications, we found that alpine generalists differed significantly inBTEMP from both alpine specialists and non-alpine species (regardless of elevation).
These results might suggest different strategies are needed by alpine generalists and specialists with respect to temperature responses, as the generalist species may encounter a wider range of temperatures than the specialists. At the same time, in this dataset, genera with alpinespecialists had a greater fraction of both alpinegeneralists and montane specialists (Supplemental Fig. S2; Figueroa et al., pers. comm. ). This could indicate that alpine specialists derive mainly from alpine generalists and montane specialists (i.e., diversification has occurred as species encountered novel, higher-elevation habitat), consistent with a ‘montane speciation model’ (sensu Roy, 1997; but see Dagallier et al., 2020). It could also indicate that alpine specialists tend to arise within lineages having greater evolutionary potential to adapt to high-elevation conditions, even if the strategies involved differ among these lineages (Folk et al., 2020; Martínez-Padilla, Estrada, Early, García-Gonzalez, 2017). However, we also acknowledge that our climate analyses incorporated only macroscopic conditions and did not address ways in which species at high elevations find and create microclimatic conditions to enhance survival (Körner, 2003; Ohler, Lechleitner, Junker, 2020).