It is not clear why greater species richness is observed in warmer and more humid regions. Since the first description of large-scale patterns of species diversity by Alexander von Humboldt in 1807 (ref. 9), differences in climatic conditions between tropical and extratropical regions have been considered a plausible explanation for global diversity gradients. It is now well established that species richness is often strongly correlated with climatic conditions, namely temperature, water availability and resulting ecosystem productivity1,2,3,4,5,6. Although several causal pathways link climatic variables to species diversity including both current ecology and deep-time evolutionary processes5,10, we lack an understanding of how the geography of climate is associated with diversity patterns. Specifically, it is unknown whether greater diversity in warmer and more humid regions results from the effect of the climate itself, or from the geographic area occupied by these climatic conditions (that is, climate area) and the spatial isolation between similar climatic conditions (that is, climate isolation), both of which could lead to higher diversity.
In an influential piece7, John Terborgh argued that “unusual environmental situations will carry impoverished flora in relation to nearby sites incorporating more usual conditions”. Conversely, more usual or common climatic conditions (that is, those that cover a greater geographic area) will have high species richness7. The relationship between climate area and species richness led to the area hypothesis to explain global patterns of species diversity, originally arguing that climatic conditions occurring in the tropics are more common than those conditions occurring in extratropical regions. The area hypothesis was further popularized and amplified by Michael Rosenzweig8, but the emphasis on climate area was lost. At present, the area hypothesis is considered in a very broad geographic context building on the idea that tropical regions have more land than extratropical regions, thereby offering a reasonable explanation for why greater species richness is observed in the tropics. More recently, the commonness of climatic conditions (that is, the total global extent of a given climatic condition) has been used to explain empirical patterns of species richness11, leading to the formalization of species–area relationship within climate12. In addition to area, the isolation of fragments of a given climatic condition on the surface of the Earth can also be expected to affect species diversity. Speciation rates are expected to increase as a result of reduced gene flow and increased environmental heterogeneity13,14,15. The same expectation applies when biodiversity is analysed within different climates. Climatic conditions with isolated and larger extent across the surface of the globe should reduce extinction rates, facilitate allopatric speciation and shelter biotas that evolved independently, and therefore strongly affect climatic gradients of diversity at the global scale. However, the combined effects of the geography of climate (that is, climate area and isolation) and climatic conditions per se on global patterns of diversity remain unknown.
A substantial challenge in assessing the impact of the geography of climate, specifically climate area and isolation, arises from the necessity to reorient traditional biodiversity studies—typically focused on geographic landscapes—towards the realm of climate space. This pivot demands a fresh perspective on examining the relationships between diversity and climate. Using multidimensional space, defined by climatic conditions, to study biodiversity patterns is not a new concept. It has been previously proposed in the literature16,17,18. Primarily, this approach has been used for the classification of life zones and biomes19,20, as well as in species distribution modelling and its various applications21,22. By contrast, this perspective has been generally overlooked in studies investigating the emergence and maintenance of large-scale diversity patterns11,12. Here we investigate tetrapod diversity patterns thoroughly in climate space, disentangling the effects of the geography of climate and climate conditions per se, on diversity–climate relationships. We anticipate finding a higher number of species in climatic conditions that both cover large geographic areas and exhibit characteristics of isolation or fragmentation. We believe that these findings will hold true regardless of the inherent differences in thermal physiologies, such as endothermy and ectothermy, among tetrapod groups. These expectations are aligned with empirical evidence showing the effect of geographic area and isolation on geographic patterns of species richness independent of species’ thermal and metabolic physiologies8,13,15,23. The shift towards understanding diversity as a pattern in climate space, driven by a process in climate space, can reveal insights into how climate is structured globally and help face the challenges imposed by climate change.
We used the first two axes of a principal component analysis of 12 global-scale climate variables to define a two-dimensional orthogonal climate space (Fig. 1a–c) that represents thermal and water availability limits to species distribution. Each axis of the climate space was then divided into climate cells of equal climate intervals (Fig. 1c; results robust to interval size, Supplementary Information). Thus, each climate cell represents several geographic locations that fall within a specific climate interval. Likewise, several geographic locations belong to a unique climate cell (Fig. 1d). This connection between climate and geographic space, referred to as Hutchinson’s duality18, allows geographic information to be mapped in climate space and vice versa. Using this approach, we computed the geographic extent of a climatic condition that we refer to as climate area (Fig. 1f). As a climate condition is scattered on the surface of the planet, we identified the individual fragments of a climate condition (that is, regions within a climate cell that are geographically isolated from other regions within the same climate cell) and measured the average geodesic distance among climate fragments, which we refer to as climate isolation (Fig. 1e). Climate area and climate isolation represent the geography of global climate. We also computed the within-climate-cell average of the first two principal components for each climate cell to evaluate the effect of the climate itself. Finally, for each climate cell we counted the number of species that fall within that climatic condition (Fig. 2) using range distribution data for terrestrial amphibians, reptiles, mammals and birds, and investigated the effect of the climate, and its area and isolation on richness patterns and species composition of each tetrapod group.
a,b, Scores of the first and second principal components (PC1 and PC2) mapped across the world. c, Climatic information represented in a two-dimensional coordinate plane in which the x axis and y axis are represented by PC1 and PC2 and each point represents the climatic information of each geographic location (n = 13,312 geographic cells). d, Using the gridded climatic space created with PC1 and PC2 (c), the duality between climate and geographic spaces is shown by matching colours between climate and geographic grids, in which similarity in colour tone indicates similarity in climatic condition. The duality between geographic and climate space refers to the relationship between geographic and climate space: a given climatic condition (climate cell in climate space) is observed in several geographic locations, and several geographic locations belong to a unique climate condition. e, Climate area representing the sum of land surface area of a given climatic condition (that is, climate cell). f, Climate isolation representing the average geodesic distance among climate fragments (that is, geographic cells connected to each other that occur within a single climate).
Source data
a–d, Number of species that fall within a climatic condition counted for birds (n = 180 climate cells; a), mammals (n = 178 climate cells; b), amphibians (n = 174 climate cells; c) and reptiles (n = 165 climate cells; d). PC1 and PC2 represent the first and second axes of a principal component analysis of 12 global-scale climate variables (Supplementary Information).
Source data
For each tetrapod group, we fitted a Poisson-distributed generalized additive mode to account for nonlinearity in model residuals with species richness in climate space as the response variable, and the geography of climate (that is, climate area and isolation) and climate itself (that is, first and second principal components) as predictors (Extended Data Table 1). Our model explained nearly 90% of the variation in richness for birds (proportion of null deviance = 0.90, adjusted R2 = 0.92), mammals (proportion of null deviance = 0.90, adjusted R2 = 0.91), amphibians (proportion of null deviance = 0.88, adjusted R2 = 0.90) and reptiles (proportion of null deviance = 0.88, adjusted R2 = 0.91)–results hold consistent when nonlinearity is addressed through polynomial regressions (Supplementary Information). These results indicate that for all tetrapod groups, the geography of climate and climate itself can explain most of the variability of tetrapod richness in climate space. For all groups (Extended Data Figs. 1–3), partial residual richness (that is, richness not explained by other predictors) increases with area (Fig. 3a) and with isolation (Fig. 3b), even though climates occurring on opposite poles are isolated, their extreme conditions are suitable to only a few species. Thus, the geographic distribution of polar climates does not effectively isolate a substantial number of species. Partial residual richness is positively related to the first principal component (Fig. 3c) that is composed mostly by temperature variables and has a hump-shaped relationship with the second principal component that is defined by a balance of energy and water availability (Fig. 3d). The explanatory power of the geography of climate and climate itself is very similar for all groups (Fig. 4). In terms of proportion of null deviance, climate area contributes about 10%, whereas climate isolation accounts for roughly 5%. Comparatively, the first and second principal components contribute around 13% and 2%, respectively. Shared explanatory power within the geography of climate totals about 12%, whereas within climate itself it is around 2%. Thus, when considering both isolated and shared contributions, the impact of the geography of climate is nearly double that of climate itself (Fig. 4). The remaining 45% of shared explanatory power comes from overlapping contributions of geography of climate and climate itself. Using different resolutions and combinations of variables to define climate space did not qualitatively change the results (Supplementary Information).
a,b, Partial residual richness versus climate area and isolation. c,d, Partial residual richness versus climate itself here represented by the first two principal components. The grey shading around each regression line represents the 95% confidence interval. Identical patterns are observed for other tetrapod groups (Extended Data Figs. 2 and 3).
Source data
For each tetrapod group, we separated how much of the variability in species richness on the climate space is explained exclusively by climate area and climate isolation (geography of climate) and principal components one and two (climate itself). We also computed the joint contribution within the geography of climate and climate itself and between the geography of climate and climate itself.
Source data
Our results reinforce previous findings that the geographic extent of climatic conditions is positively associated with species diversity7,11,12, but further highlight that only 10% of the variation in richness comes exclusively from climate area alone. The mechanisms underlying the species–area relationship in climate space are different from those proposed for geographic space because similar climatic conditions are not continuously distributed across geography. In geographicl space, a continuous large area often leads to higher environmental heterogeneity and consequently different species exploring different ecological opportunities24. However, scattered climatic conditions in climate space, that sum up to define climate area, represent several geographic locations within the same type of environment. As a result, within a given climate, there is little environmental heterogeneity (assuming climate variation influences environmental heterogeneity) that could lead to specialization for different environmental conditions. Therefore, the finding of more species occurring with homogeneous climatic conditions that occupy larger geographic extents is probably attributable to capacity rules25. Climatic conditions that cover more extensive land areas are believed to support more individuals, leading to larger populations. Larger populations, in turn, are associated with increased rates of speciation and reduced rates of extinction5,25,26.
Another aspect of the geography of climatic conditions is that a larger climate area does not necessarily translate to larger continuous habitat. In fact, climate area is strongly correlated with the number of climate fragments (that is, geographic cells connected to each other that occur within a single climate; Pearson’s r = 0.95; Fig. 5a and Extended Data Fig. 4). However, the correlation between the number of climate fragments or climate area and climate isolation (measured as the average distance between climate fragments) is weak, with Pearson’s correlation coefficients of 0.23 and 0.11, respectively. These observations highlight the importance of characterizing climate isolation in addition to the area of climatic conditions when macroecological patterns are analysed in climate space.
a, The relationship between climate area and the number of climate fragments (n = 164 climate cells; Extended Data Fig. 4). b, The relationship between average absolute latitudes of each climate cell in climate space and climate isolation (n = 1,028 climate cells) at a resolution of 60 equal intervals (Extended Data Fig. 5). In b, τ expresses the quantiles in which quantile regressions are fitted and OLS represents the ordinary least squares regression (dashed line). c,e, The relationship between the geography of climate and the turnover component of taxonomical β-diversity (n = 164 climate cells; Extended Data Figs. 6–8). d,f, The relationship between the geography of climate and the nestedness component of β-diversity (n = 164 climate cells; Extended Data Figs. 6–8). The grey shading around regression lines represents the 95% confidence interval.
Source data
The mechanism underlying the association between species richness and climate isolation is probably linked to climate isolation influencing gene flow among diverging populations. Over deep time, at the global scale, climate change, continental drift and mountain uplift can affect the spatial connection among similar climatic conditions. Populations of a species dispersing to follow their optimum climatic conditions expand, contract and fragment their geographic distribution within climatic conditions27. The isolation of populations within spatially disconnected climatic conditions increases the chances of allopatric speciation events and founder speciation events caused by long-distance dispersal27,28. Therefore, at the global scale we expect that locations where climatic conditions are more isolated will also have more species, which is consistent with our findings. In addition, isolated climatic conditions shelter biotas that independently evolved within that climatic condition making it likely that different species pools will be sampled. This is in fact evidenced by the increase in the turnover of species composition with the increase of climate isolation (Fig. 5c).
Notably, the most conspicuous and discussed pattern of climate isolation at the global scale is that of two broad climatic zones in high latitudes, with northern and southern polar and temperate zones being disjunct and separated by a large contiguous tropical zone8,29. However, our results show that tropical climates, spread across several continents, are more isolated than polar and temperate climates. There is an envelope relationship between the distance of similar climates (distance among climate fragments) and latitude (Fig. 5b and Extended Data Fig. 5), with a tendency of shorter distances among similar climates at higher latitudes (Fig. 5b, ordinary least squares model) and greater variability in geographic isolation of similar climates in tropical zones. The general tendency of greater isolation in lower latitudes increases the probability of independent pools of species evolving within warmer and more humid climates spread in a mosaic of similar but isolated climatic conditions. We find that climates occupying large and isolated areas have increased community differentiation with both area and isolation of climate (Fig. 5c,d and Extended Data Figs. 6–8), indicating that ecological communities occurring within larger and isolated climates have high replacement of species among communities. The likely mechanism behind this pattern is that independent pools of species evolve as a consequence of dispersal limitation and historical changes in the geography of climate30. In addition, climates occupying small and connected areas have a nested community structure with fewer species that are a subset of richer communities within climates (Fig. 5e,f and Extended Data Figs. 6–8). This pattern probably emerges because of lower dispersal limitation within smaller and connected climatic conditions. These results demonstrate how the area and isolation of climate strongly capture changes in community composition even within homogeneous climatic conditions. Thus, climate area and isolation capture patterns of community differentiation across the globe.
The area hypothesis to explain latitudinal diversity gradients emerged through observations that climatic conditions that occur in tropical environments are more common than climatic conditions that occur in extratropical regions7,8. However, with the observed reduction in climate isolation at higher latitudes, it becomes evident that not only the commonness of climatic conditions, but also the geographic distribution and isolation of similar climatic conditions, need to be taken into account. Here we propose that the area hypothesis to explain global-scale patterns of species diversity should be modified into an area–isolation hypothesis because not only do lower-latitude climates have larger geographic extent, but climate isolation decreases towards the poles. Thus, tropical climates, characterized by large areas that are both fragmented and isolated, tend to have more observed species. Such climatic structures could promote greater speciation rates through capacity rules and reduced gene flow. Our results show: (1) increase of species richness and turnover with climate area and isolation; (2) larger isolation of tropical climates; and (3) the degree of fragmentation of larger climates call for a revision of the area hypothesis. Although climates occurring in the tropics are more common, these climates are also more fragmented and isolated.
Even though our model, considering all geographic features of climate, explains a large fraction of the variation in tetrapod diversity, the remaining 10% of unexplained variation showed interesting patterns when projected to climate and geographic space (Fig. 6). Along with the model’s goodness of fit, residual patterns are consistent among tetrapod groups (Fig. 6). These residual patterns highlight an important aspect of analysing species diversity directly in climate space. Historical contingencies of different regions with the same climatic conditions are disregarded by combining the presence and absence of all species in those regions, regardless of whether the regions are clustered or scattered around the globe. However, geographic patterns of model residuals can give insights into why more or fewer species are observed under certain climatic conditions after controlling for the effects of the geography of climate and climate itself.
a–d, The residual richness is mapped in climate and geographic space for birds (a), mammals (b), amphibians (c) and reptiles (d). Blue areas express locations where fewer species are predicted than expected by model predictors and red areas express locations where more species are predicted than expected by model predictors.
Source data
With few exceptions, parts of mountainous climates across all continents have more species than expected by the climate and climate geography. However, the spatial resolution in our global-scale study may not be enough to precisely characterize mountainous climates. These climates usually occupy a small geographic extent (25% of terrestrial landmass31) when compared to lowlands and yet, especially in tropical environments, are home to approximately 90% of tetrapod species31. Tropical mountains usually have more species than expected by climate itself (that is, positive residuals31,32) and the same occurs when the geography of climate is also considered (Fig. 6). Using the present climate to define the effect of climate on mountainous regions misses one important aspect, namely the great importance of mountains for biodiversity refugia under climatic oscillations31. Therefore, both the area and isolation of climate and climate per se are not enough to explain the building of mountain diversity, especially in tropical mountains such as the Andes.
Lower species richness than expected by climate itself and the geography of climate (that is, negative residuals) are observed in many arid regions across the world with few exceptions. In South America, the eastern dry diagonal connecting Caatinga, Cerrado and Chaco and the western dry diagonal connecting Patagonia, Monte, Prepruna dry Pruna and the Atacama Desert, and in Africa, part of the Sahara Desert and the Somalian Desert, among other dry regions (Fig. 6), show a very clear pattern of fewer species than expected by the geography of climate and climate itself. Contrary to mountainous regions, dry regions occupy a large geographic extent, which given the expected species–area relationship, should support a greater number of species. However, their extreme climates act as a barrier for lineages that are not capable of surviving in such arid environments33. The difficulty of adapting to extreme conditions33, even if these conditions are common on the surface of the planet, could explain why fewer species are observed in these regions than expected by climate itself and the geography of climate.
Here we demonstrated that the present geography of climate and climate itself can explain a large fraction of the tetrapod diversity and that the isolated effect of the geography of climate almost doubles the effect of climate itself. However, species richness might not be associated only with the present climate, but also with past climate and past geographic structures of climate. It should be expected that over deep time some climatic conditions were more common than others, that connections and disconnections of climatic conditions occurred and that some conditions appear and disappear across millions of years. Therefore, exploring the dynamics of the geography of climate over deep time is a natural next step for studying diversity–climate relationships in climate space. The same rationale for past climatic dynamics can also be used for future climate change. Ongoing climate changes may alter the commonness of climatic conditions as well as their connection and isolation. For example, climate velocity, representing the direction and speed that species move to maintain their current climatic condition under climate change34,35, is largely affected by climate connectivity36 and can benefit from better understanding of the effect of the geography of climate on biodiversity patterns. If many species have their optimum climate within a climatic condition that contracts its geographic extent with climate change, then competition might increase, potentially influencing species coexistence. In addition, the contraction of the geographic extent of a climatic condition imposes evolutionary pressure for individuals to shift their optimum to other similar climatic conditions that are either less saturated with species or have expanded their geographic extent on the surface of the planet. Finally, connections and disconnections of climate affect the ability of species to disperse between similar climatic conditions affecting the isolation among populations and consequently influencing their gene flow. Therefore, understanding how the geography of climate is associated with biodiversity is key to better understand and mitigate the impacts of climate change on biodiversity.
Building on these insights, we underline the urgent necessity to incorporate the geography of climate into studies examining the impact of climate change on biodiversity patterns. This crucial element has been largely disregarded. Recognizing the demonstrated connection between these geographic factors and species richness, it is critical that we unravel their temporal dynamics for effective biodiversity conservation. We must look beyond mere changes in climate and also consider their commonness and isolation across the planet. Neglecting these factors may result in unexpected climate change impacts on biodiversity, stemming from our failure to track changes in the geography of climate.
Key considerations that emerge include whether our current conservation efforts are inadvertently biased towards protecting common climates, thereby neglecting rarer climates that may harbour unique species. Furthermore, we must understand how ongoing climate change might influence the prevalence and isolation of various climatic conditions. Climate change could fragment previously continuous climates, intensifying the challenges for species to disperse and maintain their preferred climatic conditions. Likewise, if a previously extensive climate shrinks owing to climate change, this could influence species coexistence.
In essence, to mitigate the impacts of climate change on biodiversity, it is paramount that we deepen our understanding of the geography of climate and its shifts over time. This approach could represent a substantial contribution to conservation biology, providing more comprehensive and effective strategies for biodiversity preservation.