Underpinning global biogeographical schemes with quantitative data


Drawing lines on maps is a central and important part of what we biogeographers do. The function of the lines is usually that they indicate regions where there is a significant difference between the biotas on either side, due to differences in their history or ecology. But we have always been hampered by the fact that the differences, especially where they are between major units of the globe, have usually been qualitative, not quantitative. They therefore cannot indicate degrees of difference, and cannot provide a hierarchical system of relationships between the regions.

An important paper by Kreft & Jetz (2010) shows how the recent production of global species range maps of mammals, new multivariate techniques, and improved computer power have now made it possible to address this problem. After dividing the map into an equal-area grid, the distribution data were first converted into a taxon per grid cell incidence matrix. The authors then assessed nine different hierarchical clustering algorithms at various taxonomic levels (families, genera and species), finding that UPGMA (unweighted pair-group method using arithmetic averages), an agglomerative clustering algorithm, consistently provided the best performance in delimiting biogeographical transition zones. In this approach, each grid cell is paired with the one with which it has most faunistic similarity; this process is repeated into progressively higher-level clusters until they all form one group. This method therefore provides a continuous hierarchical system of relationships between the units, which can be shown as a dendrogram.

Their results in general are striking in providing quantitative confirmation of the main outlines of the mammalian biogeographical regions, and are extremely similar to Wallace’s classic scheme for mammal biogeography. But there are also some differences. At first sight, the most surprising is that, at both genus and family level, Madagascar is closest to Australia/New Guinea. But, in fact, this is not because there is any deep linkage between them (the resemblance is at a very low level of similarity), but is because they are both very different from the other regions, each having a very high level of endemism at higher taxonomic levels than those investigated by Kreft and Jetz. So, they are the two regions left over after everyone else has found a partner! But a more real and important difference is that, at all three taxonomic levels, the Sahara and the Arabian Peninsula, plus adjacent parts of the Middle East, cluster with Africa, rather than with Eurasia as Wallace had suggested.

Kreft & Jetz (2010) rightly comment that no single method will be adequate in providing clear and appropriate data for all the various aspects of historical and ecological biogeography, but they suggest that, in combination with the growing number of well-resolved complete phylogenies, they may facilitate new, perhaps more analytical, evolutionary interpretations. But, before we become over-optimistic about the future, we must note that, inevitably, they have started with the easiest group, the mammals. The much greater number of taxa in such groups as flowering plants or insects will make similar analyses much more difficult, as will our much less detailed knowledge of their species range maps – for example, it has been commented that ‘no full Amazonian distribution of any plant species is completely known’ (Bush & Lovejoy, 2007, p. 1292, my italics)!

Kreft & Jetz (2010) also present an analysis for bats vs. non-flying mammals, and note that the North American bats showed their strongest association with South America, whereas their earth-bound relatives are more closely linked to Eurasia. They suggest that this is because the bats’ greater ability to disperse over longer distances (and to cross ocean barriers) allowed them to extend their ranges to South America in a fashion that was impossible for other mammals. This greater dispersal power is also reflected in Procheş' (2005) observation that cluster analysis of bat distributions showed that, in the transition zone between the Oriental and the New Guinea/Australia regions, bat distributions are closer to those of plants than to those of other animals. Procheş suggested that his results give hope that a common biogeographical scheme for all organisms is within reach, while commenting that the current separation between plants and animals is probably one of the least productive divisions, because both groups contain such a diversity of characteristics of size, dispersal abilities and climatic requirements.

I touched on the relevance of dispersal ability to biogeographical patterns nearly 10 years ago (Cox, 2001), when I pointed out that the mammals are unusual, even within the animal kingdom, for their very limited ability to disperse across an ocean barrier. I therefore suggested that it would be better to refer to Wallace’s regions as the ‘mammal biogeographical’ regions, rather than as the ‘zoogeographical’ regions, as at least some members of many other animal groups had adaptations for crossing such barriers, and consequently had patterns of distribution more similar to those of plants. I therefore felt that it was useful to retain two basic patterns of global biogeography, one for groups with a high dispersal ability (such as flowering plants), and one for those with low dispersal ability (such as mammals). But the results of Kreft & Jetz (2010) also warn us that, as our knowledge of patterns of distribution and our ability to interpret them increases, we are likely to be faced with a much greater variety of resolved patterns than merely these two extremes. I myself cannot see the purpose and utility in selecting one particular global pattern as the one against which all others must be compared. But, if others see such a need, it would surely be best to choose that of mammals. That is because, most of them being quite unable to cross ocean gaps, their distribution reflects that most dramatic and obvious of all ecological transition points – that between the margins of the continents and the sea, and therefore between land and water and their totally different ecological requirements. This demarcation is therefore also one that has greatly affected the patterns of distribution of many other organisms. The differences between their patterns and those of mammals is likely to be mainly the result of the extent to which they have developed dispersal mechanisms to overcome this major barrier. Finally, that land/sea transition is both clear and familiar on all maps. (To select this pattern would also, appropriately enough, remind global biogeographers of their Wallacean roots!)

An extremely different suggestion for a single biogeographical scheme has been made by Morrone (2002). His methodology is based on the panbiogeographical system of identification of generalized tracks linking areas that contained some common biotic elements, together with the rejection of dispersal across barriers as a possible factor involved. Instead, the method posits that an ancestral form was already in existence in all of the areas concerned and later, by vicariance after barriers between them formed, differentiated into the taxa we see today. It also leads to the suggestion that, where today’s continents contain areas that differ in their biotas, this is because these continents are ‘composite’ biogeographically rather than natural areas – whereas most biogeographers would accept this as the results of different climate and ecology. Morrone (2002) accordingly erected an ‘Austral kingdom’, which includes the Andean part of South America, the Cape region of South Africa and the southern and eastern parts of Australia, but not the rest of those continents. I have (Cox, 1998) argued at length against the whole basis of the panbiogeographical method, and I do not believe that Morrone’s suggestion would be a suitable model for a single global biogeographical scheme. I was therefore not surprised to find no trace of Morrone’s system in Kreft and Jetz’s results.

In another paper in this issue of Journal of Biogeography, Esselstyn et al. (2010) tackle some problems in what is perhaps biogeographically the most complex area of the world. The biogeography of the region between the present-day coastlines of Southeast Asia and Asia has been complicated, not only by the gradual approach of Australia and its biota from the east, but also by changes in the pattern of islands there. These changes were caused by tectonic activity along the southern edges of the Greater Sunda Islands, and also by Pleistocene changes in sea level, leading to cycles of fragmentation and reunion of patches of land. Esselstyn et al. focus on the vertebrate biogeography of Palawan, an elongate island on the northern edge of the Southeast Asian (Sunda) continental shelf, extending from Borneo towards the Philippines. Palawan is separated from Borneo by a channel that is currently about 140 m deep. It is also a geologically composite island, part of which rafted away from near Taiwan c. 30 Ma.

Previously, it had been thought that the vertebrate fauna of Palawan was basically similar to that of Borneo, but Esselstyn et al. (2010) found that, of 39 Palawan species or populations, 17 are sister to Philippine lineages, while others have links with both, or with the Sunda Islands and/or other islands. It would appear that Palawan has played multiple biogeographical roles, depending on its varying geographical links, sometimes being merely an extension of Borneo, sometimes (or for some taxa) acting as a colonization route into the Philippines, and perhaps even at some time having been an isolated oceanic island receiving colonists by an over-water route. Such are the increasingly detailed possibilities that biogeographers have to take into account when they analyse such faunas in increasing detail!

Editor: Robert Whittaker