An Update of Wallace’s Zoogeographic Regions of the World

See allHide authors and affiliations

Science  04 Jan 2013:
Vol. 339, Issue 6115, pp. 74-78
DOI: 10.1126/science.1228282


Modern attempts to produce biogeographic maps focus on the distribution of species, and the maps are typically drawn without phylogenetic considerations. Here, we generate a global map of zoogeographic regions by combining data on the distributions and phylogenetic relationships of 21,037 species of amphibians, birds, and mammals. We identify 20 distinct zoogeographic regions, which are grouped into 11 larger realms. We document the lack of support for several regions previously defined based on distributional data and show that spatial turnover in the phylogenetic composition of vertebrate assemblages is higher in the Southern than in the Northern Hemisphere. We further show that the integration of phylogenetic information provides valuable insight on historical relationships among regions, permitting the identification of evolutionarily unique regions of the world.

Biogeographic and bioclimatic regions are the fundamental units of comparison in many broad-scale ecological and evolutionary studies (1, 2) and provide an essential tool for conservation planning (3, 4). In 1876, Alfred Russel Wallace published the first map of global terrestrial zoogeographic regions (5), which later became the cornerstone of modern biogeography (3). Using existing knowledge of his time (6), mostly on the distributions and taxonomic relationships of broadly defined vertebrate families, Wallace divided the world into six terrestrial zoogeographic units largely delineated by what we now know as the continental plates. Despite relying on limited information and lacking a statistical basis, Wallace’s original map is still in use today.

Wallace’s original zoogeographic regionalization scheme considered ancestral relationships among species, but subsequent schemes generally used data only on the contemporary distributions of species without explicitly considering phylogenetic relationships (79). Phylogenetic trees contain essential information on the evolutionary relationships of species and have become increasingly available in recent decades, permitting the delineation of biogeographic regions as originally envisioned by Wallace. The opportunity now exists to use phylogenetic information for grouping assemblages of species into biogeographic units on a global scale. In addition to permitting a sound delimitation of biogeographic regions, phylogenetic information allows quantifying phylogenetic affinities among regions (e.g., 10). Newly developed statistical frameworks facilitate the transparent characterization of large biogeographic data sets while minimizing the need for subjective decisions (11).

Here, we delineated the terrestrial zoogeographic realms and regions of the world (12) by integrating data on the global distributions and phylogenetic relationships of the world’s amphibians (6110 species), nonpelagic birds (10,074 species), and nonmarine mammals (4853 species), a total of 21,037 vertebrates species [see (13) for details]. Pairwise phylogenetic beta diversity (pβ) metrics were used to quantify change in phylogenetic composition among species assemblages across the globe. Analyses of combined taxa pβ values identified a total of 20 zoogeographic regions, nested within 11 larger realms, and quantified phylogenetic relatedness among all pairs of realms and regions (Fig. 1, figs. S1 and S2, and tables S1 and S2). We also used pβ to quantify the uniqueness of regions, with the Australian (mean pβ = 0.68), Madagascan (mean pβ = 0.63), and South American (mean pβ = 0.61) regions being the most phylogenetically distinct assemblages of vertebrates (Fig. 2). These evolutionarily unique regions harbor radiations of species from several clades that are either restricted to a given region or found in only a few regions.

Fig. 1

Map of the terrestrial zoogeographic realms and regions of the world. Zoogeographic realms and regions are the product of analytical clustering of phylogenetic turnover of assemblages of species, including 21,037 species of amphibians, nonpelagic birds, and nonmarine mammals worldwide. Dashed lines delineate the 20 zoogeographic regions identified in this study. Thick lines group these regions into 11 broad-scale realms, which are named. Color differences depict the amount of phylogenetic turnover among realms. (For more details on relationships among realms, see the dendrogram and NMDS plot in fig. S1.) Dotted regions have no species records, and Antarctica is not included in the analyses.

Fig. 2

Map of evolutionary uniqueness for terrestrial zoogeographic regions of the world based on data for 21,037 species of vertebrates. Evolutionary uniqueness is calculated as the mean of pairwise pβ values between the focal region and all other regions. Colors indicate the degree to which each region differs from all other regions based on mean pairwise pβ. Regions colored in dark red are the most evolutionarily unique. Dotted regions have no species records, and Antarctica is not included in the analyses.

Our combined taxa map (Fig. 1) contrasts with some previously published global zoogeographic maps derived exclusively from data on the distribution of vertebrate species (8, 9, 11). The key discrepancy between our classification of zoogeographic regions and these previous classifications is the lack of support for previous Palearctic boundaries, which restricted this biogeographic region to the higher latitudes of the Eastern Hemisphere. The regions of central and eastern Siberia are phylogenetically more similar to the arctic parts of the Nearctic region, as traditionally defined, than to other parts of the Palearctic (fig. S2). As a result, our newly defined Palearctic realm extends across the arctic and into the northern part of the Western Hemisphere (Fig. 1 and fig. S1). These results bear similarities with the zoogeographic map of (11) derived from data on the global distribution of mammal families. In addition, our results suggest that the Saharo-Arabian realm is intermediate between the Afrotropical and Sino-Japanese realms [see the nonmetric multidimensional scaling (NMDS) plot in fig. S2]. Finally, we newly define the Panamanian, Sino-Japanese, and Oceanian realms [but see the Oceanian realm of Udvardy in (14) derived from data on plants].

Our classification of vertebrate assemblages into zoogeographic units exhibits some interesting similarities with Wallace’s original classification, as well as some important differences (fig. S3). For example, Wallace classified islands east of Borneo and Bali in his Australian region (fig. S3), which is analogous to our Oceanian and Australian realms combined (Fig. 1 and fig. S1). In contrast, we find that at least some of these islands (e.g., Sulawesi) belong to our Oriental realm, which spans Sundaland, Indochina, and India (Fig. 1 and fig. S1). Moreover, our Oceanian realm is separate from the Australian realm and includes New Guinea together with the Pacific Islands (14), whereas Wallace lumped these two biogeographic units into the Australian region. Wallace further argued that the Makassar Strait between Borneo and Sulawesi, now known as “Wallace’s Line” (15), was a major barrier to dispersal that greatly inhibited exchanges between the Australian and Asian land masses. Much debate subsequently arose regarding the precise location of the principal faunal divide between Wallace’s Oriental and Australian realms (15) (see fig. S3 for an illustration of Wallace’s original line). Our combined taxa analyses lend the strongest support to the hypothesis of Weber (16), who positioned this boundary east of Sulawesi, corresponding to the zoogeographic boundary separating our Oriental and Oceanian realms (Fig. 1 and fig. S1). However, our taxon-specific geographic delineation for birds is more consistent with Wallace’s line than Weber’s line (Fig. 3A and figs. S3 and S4A).

Fig. 3

Maps of terrestrial zoogeographic regions of the world based on data for (A) amphibians (6110 species), (B) birds (10,074 species), and (C) nonmarine mammals (4853 species). Color differences depict the relative amount of phylogenetic turnover among regions within each taxonomic clade. (For more details on relationships among regions, see the dendrogram and NMDS plots in fig. S4, A to C.) Dotted regions have no species records, and Antarctica is not included in the analyses.

The delineation of and relationships among our zoogeographic regions differ among taxa (Fig. 3 and fig. S4), and we find more regions for mammals (n = 34 regions) than for amphibians or birds (both n = 19 regions). A comparison of pβ matrices across the three vertebrate taxa reveals that amphibian assemblages located in the northeastern Arctico-Siberian, southern African, and Madagascan regions are more phylogenetically distinct than those of birds or mammals for the same regions (fig. S5). Moreover, the Australian region harbors more phylogenetically distinct assemblages of amphibians and mammals relative to birds (fig. S5). Using a partial Mantel test [see (13) for details on this analysis], which accounts for geographic distances among species assemblages (17), we find that global pβ values for birds and mammals are more strongly correlated (r = 0.68, P < 0.001) than for amphibians and birds (r = 0.39, P < 0.001) or amphibians and mammals (r = 0.43, P < 0.001). These results might partly reflect a major episode of diversification early in the evolutionary history of amphibians (18). Alternatively, differences in spatial patterns of phylogenetic turnover among vertebrate classes might result from lower dispersal ability (19) and greater sensitivity of amphibians to environmental conditions (20). Interestingly, previous comparative studies documented similar incongruence in the diversity and distribution of amphibians relative to that of birds and mammals (21, 22).

The contrast between our zoogeographic regions with regions based only on distributional data (Fig. 4) demonstrates the consequences of incorporating phylogenetic information in the delineation of zoogeographic units. Relative to expectations based on turnover of species, spatial turnover in the phylogenetic composition of assemblages of species is generally low in the Northern Hemisphere, whereas the opposite is true in the Southern Hemisphere (Fig. 4A). In particular, amphibians exhibit low spatial turnover in phylogenetic composition relative to their turnover in the composition of species between the North American and Eurasian regions (Fig. 4B; also compare fig. S4A with fig. S6A). Higher phylogenetic uniqueness in the Southern than in the Northern Hemisphere is consistent with long-term isolation having left a pervasive signature on species assemblages, where oceanic barriers have limited dispersal between continents (23, 24). In the Northern Hemisphere, the newly defined boundaries of the Palearctic realm might reflect the continuous presence of nonglaciated tundra in eastern Siberia and Beringia (25), whereas the subtle differences in the phylogenetic composition of assemblages over the Northern Hemisphere as a whole might be a consequence of a high degree of connectivity and range dynamics. Low rates of extinctions resulting from greater climatic stability in the Southern Hemisphere could also have contributed to this pattern by allowing species that belong to ancient clades to persist through time (26, 27).

Fig. 4

Combined (A) and taxon-specific (B to D) maps illustrating the degree of phylogenetic turnover relative to the turnover of species observed among zoogeographic regions based on data for species of (A) amphibians, (B) birds, and (C) nonmarine mammals. The color scale depicts the degree to which faunal turnover between the regional assignment of the focal grid cell and the regional assignment of all other grid cells results from differences in pβ relative to beta diversity. Red colors indicate regions with a high degree of phylogenetic differentiation relative to compositional differentiation, whereas blue colors indicate the opposite. Dotted regions have no speciesș records, and Antarctica is not included in the analyses.

Our maps of zoogeographic realms and regions provide a broad overview of the distribution of the world’s amphibians, birds, and nonmarine mammals, allowing the identification of geographic areas harboring distinct evolutionary histories [see (28) for links to downloadable maps of zoogeographic realms and regions for projection in GIS (geographic information systems) mapping software and Google Earth]. These maps reflect major advances made in recent decades regarding our knowledge of the distribution and phylogeny of vertebrates and can be used to elucidate the forces and historical events responsible for the formation of the biogeographic realms and regions we recognize today. Our delineation of the zoogeographic realms and regions of the world, and especially that of the realms, appears robust to the type and quality of distributional and phylogenetic data used [see (13) for details]. Inclusion of additional phylogenetic information on branch length or improved resolution of the phylogenetic trees has the potential to facilitate a finer delineation of regions within our realms. The inclusion of data (when they become available) on reptiles, invertebrates, and/or plants may also affect the boundaries of our realms and regions and the relationships among them. Nevertheless, the maps presented here delineate robust zoogeographic units for vertebrates that can be scaled within specific continents and/or taxonomic clades. Due to these qualities, our analytical approach and zoogeographic maps provide a baseline for a wide variety of comparative ecological, biogeographic, evolutionary, and conservation-based studies (3, 22, 29).

Supplementary Materials

Materials and Methods

Figs. S1 to S11

Tables S1 to S5

Appendices S1 and S2

References (30729)

References and Notes

  1. A. R. Wallace and his contemporary W. L. Sclater used the terms “region” to denote six main zoogeographical units at a global scale and “subregion” to denote finer scale subdivisions. Wallace’s and Sclater’s regions and subregions are roughly equivalent to the realms and regions proposed here. The work of Sclater was published in The Geographical Journal (1894–1897).
  2. Materials and methods are available as supplementary materials on Science Online.
  3. M. Maechler, P. Rousseeuw, A. Struyf, M. Hubert, K. Hornik, Cluster: Cluster analysis basics and extensions, R package version 1.14.2 (2012).