Introduction
Understanding mechanisms governing spatial patterns of biodiversity at biogeographical scales is a challenging theme in ecology. Spatial isolation between populations can decrease connectivity and limit gene flow and therefore plays a major role in shaping inter-population variability and speciation processes (Orsini et al. 2013, Sexton et al. 2014, Pironon et al. 2016). Comparative analyses of natural systems characterised by spatially discontinuous habitats such as islands separated permanently by saltwater (hereafter referred to as marine islands) with those where isolation can be driven by increasing geographic distance between populations within a comparatively benign landscape matrix (e.g., the mainland) have been encouraged, as they can advance our understanding of the consequences of spatial isolation for phenotypes and genotypes (Haila 2002, Laurance 2008, Santos et al. 2016, Martín-Queller et al. 2017, Patiño et al. 2017, Flantua et al. 2020).
To date, comparing marine island populations to mainland populations has been a classic approach to understanding the drivers of isolation due to the obvious geographic separation of islands from the mainland, particularly for oceanic rather than continental islands (Weigelt and Kreft 2013). On islands, organisms can be subject to strong selection pressure due to a large variety of eco-evolutionary forces that include lowered gene flow, founder effect, genetic drift and high extinction rates due to smaller population sizes, modified abiotic and biotic conditions (Santos et al. 2016, Patiño et al. 2017). These factors have been linked to shifts in body and organ size (the “island rule”, Foster 1964, Lomolino et al. 2013, Benítez-López et al. 2021), decreased dispersal (Burns 2018), slower growth rates and increased life span (Andrews 1976, Lens et al. 2013), and changes in reproductive strategies and behaviour (Covas 2012, Morinay et al. 2013) in island populations. Such changes associated with island populations are known as the “island syndrome” (Whittaker and Fernández-Palacios 2006). In conjunction with such life history, physiological and behavioural changes, spatial equilibrial processes (founder effects, restricted dispersal, small population sizes, higher extinction rates) should theoretically reduce the neutral genetic diversity of island populations in comparison to mainland populations. However, while such patterns are predicted, this observation is not generally applicable across all island systems (Frankham 1998, Woolfit and Bromhan 2005, García-Verdugo et al. 2015).
While islands have been the classic focus of isolation effects, isolation can also emerge on the mainland, due to either large geographic distances between populations (in continuous and recently fragmented habitats, Laurance 2008, McIntyre & Hobbs 1999, Watson 2002), or environmental discontinuities between suitable habitat patches in ecological islands (Csergő et al. 2014, Tapper et al. 2014, Steinbauer et al. 2016). However, mainland isolation is likely to differ from classic marine island isolation, as mainland habitat islands lack an abrupt saltwater barrier and experience higher spatial or temporal connectivity (Watson 2002, Driscoll et al. 2013, Ó Marcaigh et al. 2021). As a result of this difference, their analogy with marine islands has been questioned (Flantua et al. 2020). This calls for further comparative investigations to better understand the spatial mechanisms governing the biodiversity of island versus mainland systems, and for the conceptual unification of isolation research across systems (Haila 2002, Laurance 2008, Santos et al. 2016, Patiño et al. 2017, Flantua et al. 2020).
Geographic distance may be key determinant of spatial isolation, as well as an important driver of spatial variability, and it is perhaps the most commonly used metric of geographic isolation (Wright 1943, Orsini et al. 2013, Sexton et al. 2014). But geographic distance is not an exclusive determinant of spatial patterns in phenotypic traits and neutral genetic diversity. While strict isolation by distance emerges due to limits to dispersal and genetic drift (Wright 1943), the role of environmental conditions in fostering spatial population variability may override the direct effects of geographic distance (Kalmar and Currie 2006, Shafer and Wolf 2013, Orsini et al. 2013, Sexton et al. 2016). Environmental heterogeneity, modified biotic interactions and habitat disturbance often shape the course of ecological and evolutionary processes in populations worldwide, and have sculpted much of the individuality of island populations (Kalmar and Currie 2006, Heaney 2007, Triantis et al. 2010, Lens et al. 2013, Weigelt and Kreft 2013, Stuessy et al. 2014, Borregaard et al. 2016). Environmental factors are key determinants of intraspecific body size variation in vertebrate groups globally (Henry et al. 2023). Increasing evidence indicates that even the spatial patterns of neutral genetic diversity are heavily influenced by environmental conditions in addition to the geographic position of populations (Lira-Noriega and Manthey 2014, de Kort et al. 2021). Despite significant advances in understanding these two major drivers of biodiversity at different levels of organisation, global comparative evidence for differential effects in island versus mainland systems is still lacking.
A complicating circumstance is that significant differences in responses may exist across different traits or groups of species, some being more responsive to geographic forces, while others responded more readily to environmental conditions (Sexton et al. 2014, Pironon et al. 2016, Orsini et al. 2013, Henry et al. 2023). For example, genetic diversity of plants responds more readily to geographic, than environmental drivers compared with animals (Sexton et al. 2014). As a result, a series of geographic, environmental and taxonomic factors need to be considered for a better understanding of the links between life histories and spatial isolation (Dupré and Ehrlén 2002, Sutherland et al 2013) and in order to detect the effect of system type (e.g., island or mainland) on inter-population variability (García-Verdugo et al. 2015, De Kort et al. 2021). Due to the difficulties in disentangling these influencing factors, the development of global biogeographic models of population variability has been slow, despite major advances in functional biogeography and population macroecology (overviews in Schrader et al. 2021a, Schrader et al. 2021b, Buckley and Puy 2022, Vasconcelos 2023).
Here we conducted a global meta-analysis of multiple plant and animal populations studied in both island and mainland systems, in which we test how geographic distance and macroclimatic distance relate to phenotypic and neutral genetic diversity variation within the populations of marine island systems and mainland systems (Fig. 1). While neutral genetic diversity results from spatial processes such as gene flow, migration or dispersal, it has mostly indirect effect on fitness through e.g. inbreeding depression or founder effects (Holderegger et al. 2006). In contrast, phenotypic variability is mainly influenced by a mixture of adaptive and plastic responses to the environment, and only partially by neutral (standing) genetic diversity. Therefore, the two measures provide complementary insights into processes underlying spatial diversity patterns. While we expected populations to show greater differences in phenotypic traits and neutral genetic diversity with increasing distances between populations, we predicted that these effects would be stronger within island systems, which show consistent spatial structure compared to the mainland systems. We further expected greater differences in phenotypic traits and neutral genetic diversity with increasing macroclimatic differences between populations, but we did not expect macroclimate effects to differ between the two system types. Finally, we predicted that phenotypic traits would show higher levels of spatial variation across populations compared to neutral genetic diversity, because the former are more strongly influenced by natural selection.