The roles of geological history and colonization ... - Wiley Online Library

0 downloads 0 Views 558KB Size Report
Lawrence R. Heaney1*, Joseph S. Walsh, Jr1,2 and A. Townsend Peterson1. 1Field Museum of Natural History, 1400 South. Lake Shore Drive, Chicago, USA, ...
Journal of Biogeography (J. Biogeogr.) (2005) 32, 229–247

ORIGINAL ARTICLE

The roles of geological history and colonization abilities in genetic differentiation between mammalian populations in the Philippine archipelago Lawrence R. Heaney1*, Joseph S. Walsh, Jr1,2 and A. Townsend Peterson1

1

Field Museum of Natural History, 1400 South Lake Shore Drive, Chicago, USA, and 2 Undergraduate Program in Biological Sciences, Northwestern University, Evanston, IL, USA

ABSTRACT

Aim To test hypotheses that: (1) late Pleistocene low sea-level shorelines (rather than current shorelines) define patterns of genetic variation among mammals on oceanic Philippine islands; (2) species-specific ecological attributes, especially forest fidelity and vagility, determine the extent to which common genetic patterns are exhibited among a set of species; (3) populations show reduced within-population variation on small, isolated oceanic islands; (4) populations tend to be most highly differentiated on small, isolated islands; and (5) to assess the extent to which patterns of genetic differentiation among multiple species are determined by interactions of ecological traits and geological/geographic conditions. Location The Philippine Islands, a large group of oceanic islands in Southeast (SE) Asia with unusually high levels of endemism among mammals. Methods Starch-gel electrophoresis of protein allozymes of six species of small fruit bats (Chiroptera, Pteropodidae) and one rodent (Rodentia, Muridae). Results Genetic distances between populations within all species are not correlated with distances between present-day shorelines, but are positively correlated with distances between shorelines during the last Pleistocene period of low sea level; relatively little intraspecific variation was found within these ‘Pleistocene islands’. Island area and isolation of oceanic populations have only slight effects on standing genetic variation within populations, but populations on some isolated islands have heightened levels of genetic differentiation, and reduced levels of gene flow, relative to other islands. Species associated with disturbed habitat (all of which fly readily across open habitats) show more genetic variation within populations than species associated with primary rain forest (all of which avoid flying out from beneath forest canopy). Species associated with disturbed habitats, which tend to be widely distributed in SE Asia, also show higher rates of gene flow and less differentiation between populations than species associated with rain forest, which tend to be Philippine endemic species. One rain forest bat has levels of gene flow and heterozygosity similar to the forest-living rodent in our study.

*Correspondence: Lawrence R. Heaney, Field Museum of Natural History, 1400 S. Lake Shore Drive, Chicago, IL 60605, USA. E-mail: [email protected] Present address: A. Townsend Peterson, Natural History Museum, The University of Kansas, Lawrence, KS, USA.

Main conclusions The maximum limits of Philippine islands that were reached during Pleistocene periods of low sea level define areas of relative genetic homogeneity, whereas even narrow sea channels between adjacent but permanently isolated oceanic islands are associated with most genetic variation within the species. Moreover, the distance between ‘Pleistocene islands’ is correlated with the extent of genetic distances within species. The structure of genetic variation is strongly influenced by the ecology of the species, predominantly as a result of their varying levels of vagility and ability to tolerate open (non-forested) habitat. Readily available information on ecology

ª 2005 Blackwell Publishing Ltd www.blackwellpublishing.com/jbi

229

L. R. Heaney et al.

(habitat association and vagility) and geological circumstances (presence or absence of Pleistocene land-bridges between islands, and distance between oceanic islands during periods of low sea level) are combined to produce a simple predictive model of likely patterns of genetic differentiation (and hence speciation) among these mammals, and probably among other organisms, in oceanic archipelagos. Keywords Biogeography, Chiroptera, differentiation, diversification, ecological traits, gene flow, genetic variation, geology, Philippines, Rodentia.

INTRODUCTION The search for an understanding of the causes of differentiation and diversification among island populations has been an intellectual crucible in evolutionary biology. From the original ruminations of Darwin and Wallace on the geographic circumstances of speciation to observations of natural selection in action, islands have provided a wealth of insight for biologists (e.g. Grant, 1998; Hall & Holloway, 1998; Whittaker, 1998; Avise, 2000; Schluter, 2000). In particular, the presence of high levels of endemism, and the processes and circumstances that produce those endemic species, has attracted much attention. Clearly, part of the great appeal of island systems is their relative simplicity: on islands, terrestrial populations are discretely bounded, gene flow is likely to be limited between them, island areas and between-island distances are easily measured, and island communities tend to have fewer species than mainland communities. Perhaps this apparent simplicity has tended to wed biologists to simple explanations for island phenomena. Biogeographers, in particular, have often divided into camps preferring either historical (e.g. Rosen, 1975) or ecological (e.g. MacArthur & Wilson, 1967) explanations of island phenomena. The extensive bodies of work by both groups convinces us that both ecological and historical factors must be important in generating intraspecific and interspecific diversity in island settings and that integration of the two perspectives is essential. We agree with recent authors that historical and ecological factors should be treated as complementary variables, rather than competing hypotheses, in explaining patterns of geographic variation within species (e.g. Bermingham & Moritz, 1998; Whittaker, 1998; Heaney, 2000; Lomolino, 2000; Zink et al., 2000; Riddle & Hafner, in press). Recent developments, such as the use of DNA-based population phylogenies, have been useful in resolving questions regarding the importance of historical and ecological factors in influencing patterns of geographic variation in single species (e.g. Avise, 2000). However, the particular history of any single taxon will necessarily represent only a portion of general patterns and causes of differentiation within any large, historically complex region; this means that general patterns 230

often cannot be perceived, and general hypotheses often cannot be tested, based on single species (Powers et al., 1991; Riddle & Hafner, in press). A multi-species comparative approach should be most useful for detecting the role of ecological and historical factors in generating patterns of variation and differentiation, helping to distinguish their relative importance, and determining the manner in which they interact to produce phylogenies and patterns of biological diversity (e.g. Zink et al., 2000; Hewitt, 2001; Ricklefs & Bermingham, 2001; Arbogast & Kenagy, 2001; Riddle & Hafner, in press). The Philippine archipelago is an exceptional theatre in which to investigate the roles of past history and current ecology in structuring geographic variation. The 7000 islands originated as a set of de novo oceanic islands [with the exception of one group that was united with mainland Southeast (SE) Asia] of varying ages and geological histories, as summarized below. It is an area of high biotic diversity and exceptional endemism that is in critical need of conservation (Myers, 1988; Wildlife Conservation Society of the Philippines, 1997; Heaney & Regalado, 1998; Mittermeier et al., 1999; Holloway, 2003; Mey, 2003). While it is noteworthy that at least 111 of the 170 native species of terrestrial mammals (64%) are endemic (Heaney et al., 1998), it is still more striking that 24 of 84 genera (29%) are endemic, implying much in situ diversification, and phylogenetic studies suggest that several large endemic clades are present among fruit bats and murid rodents (Heaney & Rickart, 1990; Heaney, 2000; Steppan et al., 2003). Each oceanic island that has remained continuously isolated from its neighbouring islands is a unique centre of mammalian endemism, with 25–80% of the nonvolant mammals endemic, even on islands of only a few hundred square kilometres. Similar patterns are evident among butterflies (Holloway, 2003) and trichopteran insects (Mey, 2003). The manner in which this diversification has arisen among Philippine mammals over evolutionary time, and the ecological means by which it has been maintained, have been the subject of diverse studies of biogeography, diversity gradients, systematics, and population biology (e.g. Heaney, 1986, 1991a, 2000, 2001; Heideman & Heaney, 1989; Heaney & Rickart, Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

Geology, ecology, and genetic differentiation in Philippine mammals 1990; Rickart et al., 1991, 1993; Heaney et al., 1998, 1999; Steppan et al., 2003). Among these was an initial examination of patterns of genetic differentiation in two species of fruit bats (Peterson & Heaney, 1993); this analysis showed that Cynopterus brachyotis, a species widespread in South East (SE) Asia that occupies disturbed anthropogenic habitats, had high levels of heterozygosity, high levels of gene flow, and low levels of genetic differentiation. In contrast, Haplonycteris fischeri, a Philippine endemic that occurs in primary rain forest, had low levels of heterozygosity, low levels of gene flow, and high levels of genetic differentiation. Further, we found that the two species showed significantly similar geographic patterns of genetic differentiation between populations, and that those patterns were strongly influenced by the extent of island

shorelines during Pleistocene periods of low sea, but not by current shorelines. In this study, we extend those observations by increasing the number of species (from 2 to 7) and the number of islands (from 6 to 11). We include six fruit bat species because they are speciose, abundant, and generally easily captured (Heideman & Heaney, 1989). Several of these species are endemic to (but widespread within) the archipelago, maximizing the likelihood that general patterns could be detected. The 11 islands (Fig. 1) represent many (although not all) of the distinct areas of endemism in the Philippines, and a range of areas and degrees of isolation; the number was limited by the availability of suitably fresh frozen tissues. It should be noted that not all species occur on all 11 of the islands, and in a few cases we

Figure 1 Philippine archipelago. Extent of late Pleistocene landmasses (areas delimited by present 120 m bathymetric contour) shaded (redrawn from Heaney, 1986). Black dots indicate the origin of samples used in this study. Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

231

L. R. Heaney et al. lacked tissues for study from reference islands where the species is known to occur. We also include Rattus everetti, a Philippine endemic murid rodent that is the only non-volant small mammal that occurs widely within the archipelago; fortunately, it is sufficiently abundant that we were able to obtain several adequate samples. We used protein electrophoresis in this study, an analytical tool of great power that is technically difficult and not often used currently, but highly appropriate for investigation of the issues of concern to us here. Because protein electrophoresis requires relatively large amounts of fresh or freshly-frozen tissue, our sample sizes for the fruit bats are sometimes small. However, we have endeavoured to use statistical analysis cautiously to avoid falsely detecting patterns, especially in a multi-species comparison, with the result that those patterns that we describe here are likely to be robust. We believe that such information on population genetics is crucial to developing an understanding of the role of geography in influencing the process of diversification in this highly biodiverse oceanic archipelago. As part of this study, we test two hypotheses that we proposed in the earlier paper. The first is that late Pleistocene shorelines of oceanic islands (i.e. the maximum extent of dry-land islands) are as important in structuring patterns of intraspecific variation in the Philippines as they are in delimiting regions of interspecific diversity. Previous investigation of patterns of species distribution, diversity, and endemism in Philippine mammals have shown that shorelines of islands that existed during Pleistocene periods of low sea level form the primary boundaries between the highly distinct faunal regions in the archipelago (25–80% endemism among non-volant mammals and 7–22% among fruit bats; Heaney, 1986, 1991a, 1993, in press; Heaney et al., 1998). We predict that these geographic barriers will also manifest themselves as significant partitions of genetic variation between island populations common to all seven species in this study. The second hypothesis is that ecological attributes of species determine the extent to which common patterns are exhibited. We expect that basic knowledge regarding distribution and habitat preferences of these species derived from extensive field work in the Philippines (e.g. Heideman & Heaney, 1989; Heaney et al., 1998, 1999; Rickart et al., 1991, 1993) will provide indications of levels of gene flow and consequent degree of differentiation between populations. We also address three questions unresolved in the previous study: (1) whether small, isolated island populations tend to show reduced within-population variability, (2) whether small, isolated island populations tend to be more genetically differentiated, and (3) whether the tendency to develop genetically distinct populations, which we consider to be an intrinsic component of the process of speciation, is correlated with, and can be predicted from, readily measured ecological and geographic/geological parameters. Study species and site This study examines seven species of Philippine mammals that fall into three general ecological and geographic patterns: (1) 232

three species widespread in SE Asia, all of which are primarily associated with disturbed habitat; (2) three species that are endemic to the Philippines but widespread within the oceanic archipelago and are primarily associated with forest (although with variation, as noted below); and (3) one species that is a Philippine endemic associated with forest, but restricted to one Pleistocene island (Heaney et al., 1989, 1999; Heideman & Heaney, 1989; Heaney, 1991a; Rickart et al., 1991, 1993). The three fruit bats to which we refer as ‘widespread species’ are found throughout SE Asia and are common in disturbed anthropogenic habitats in the Philippines. They are C. brachyotis, a small (30–35 g) frugivore; Macroglossus minimus, a small (15–20 g) nectarivore; and Rousettus amplexicaudatus, a medium-sized (70–100 g) frugivore. In the Philippines, these three bats forage in orchards, other agricultural areas, and disturbed secondary forest. They are most common at lower elevations and are usually absent in montane rain forest. Rousettus amplexicaudatus is most common in clearings and orchards, and are known to regularly fly long distances (> 20 km night)1) to forage, often across open water (e.g. Rickart et al., 1993). Macroglossus minimus is usually found in association with wild or domestic abaca or banana (Musa spp.) in open secondary forest or agricultural areas, but also feed on mangroves that grow in patches in estuaries. Cynopterus brachyotis prefers agricultural areas or secondary forest, and is rare in primary rain forest except on one small island (Maripipi) which lacks Philippine endemic fruit bats (Rickart et al., 1993). In contrast, the second cluster of three species are endemic to the Philippines, but occur nearly throughout the archipelago. They are members of endemic genera, generally are common in relatively undisturbed rain forest, and are variable in their presence in heavily disturbed areas lacking good canopy cover. Haplonycteris fischeri, a small (15–20 g) frugivorous bat, is the most habitat-restricted mammal in this category; it is common beneath the canopy in primary rain forest, scarce in secondary forest, and absent in open agricultural areas. It is often the most abundant fruit bat in mature forest at middle elevations. Ptenochirus jagori, a medium-sized (70–90 g) frugivorous bat, is known to move farther than H. fischeri (Heideman & Heaney, 1989) and prefers primary rain forest but can maintain populations even in degraded secondary rain forest. Rattus everetti, a large (230–420 g) endemic Philippine rodent and the only nonvolant mammal in this study, prefers disturbed forest and tolerates primary forest, but is absent away from forest (Rickart et al., 1993; Heaney et al., 1999). The third group is represented by one species of small (35–40 g) frugivorous bat, P. minor. It is restricted to a single Pleistocene island (Greater Mindanao; see below), is common in primary or good secondary lowland rain forest, tolerates second-growth, and is scarce outside of forest. The Philippine archipelago is an especially interesting arena for investigating biological diversification (Heaney, 1986, 1991a,b; Mitchell et al., 1986; Packham, 1996; Hall, 1998, 2002; Steppan et al., 2003). In brief, the first of the extant Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

Geology, ecology, and genetic differentiation in Philippine mammals islands appeared as dry land during the early Oligocene, but most of the extant islands have originated since the late Miocene, and many during the Pliocene. One portion of the archipelago, Palawan and associated small islands, probably was joined to the Asian mainland during the middle or early Pleistocene, and is part of the Sunda Shelf biogeographic region, rather than the Philippine region (Esselstyn et al., in press). The rest of the archipelago has never had a dry-land connection to SE Asia, having arisen as a set of de novo islands from the ocean floor, most often far to the SE of its current location, as a result of tectonic and volcanic activity, gradual coalescence, and variable but progressive uplift. Most current topography is probably close to its maximum level. Since the archipelago sits on an uplifted platform bracketed by deep trenches, the depths between some islands are relatively shallow. Thus, as Pleistocene sea level rose and fell in concert with the development of continental ice sheets, certain groups of islands have experienced repeated cycles of coalescence and fragmentation. During the most recent period of low sea level in the Pleistocene, c. 18,000 years ago (Fairbanks, 1989; Siddall et al., 2003), many current islands merged into four large islands: Greater Luzon, Greater Mindanao, Greater NegrosPanay, and Greater Palawan (Fig. 1). The present configuration of islands represents a phase of fragmentation due to high sea level. Each of the ‘Pleistocene islands’ shown in Fig. 1 originated as a de novo oceanic island between c. 25 and 0.5 Ma, and (with the exception of Palawan) has had no dryland connection to other islands or continents. We call them ‘Pleistocene islands’ because they reached their largest size during the late Pleistocene periods of low sea level, not because they originated during the Pleistocene. Eleven present-day islands from seven Pleistocene islands are represented in this study (Fig. 1). During the late Pleistocene period of low sea level, Luzon, Catanduanes, and Polillo coalesced into Greater Luzon; Leyte and Biliran coalesced with modern Mindanao into Greater Mindanao; and Fuga and Barit also coalesced into a single landmass. Negros was part of Greater Negros-Panay, while Mindoro, Sibuyan, and Dalupiri each stood alone and have remained unconnected to other islands. Each major Pleistocene island has been documented as a centre of endemism. Greater Luzon and Greater Mindanao, for example, have 70–80% endemism, and Mindoro, Greater Negros-Panay, and Sibuyan each have 40–50% endemism, among native non-volant mammal species (Heaney, 1993, in press). MATERIALS AND METHODS Sampling was conducted in and near large tracts of forest (relative to the size of the islands); on all islands, most deforestation near our sites dates from the last 10–30 years. Thus, our estimates of genetic variation should not reflect the effects of habitat destruction. Bats were collected in mist nets and euthanized with lethal doses of sodium pentobarbital. Rats were collected in Victor snap traps. Tissues were harvested immediately and frozen in liquid nitrogen, and later stored at Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

the Field Museum in an ultracold freezer at )80 C. Voucher specimens were prepared and deposited at the Field Museum, the Philippine National Museum, and the United States National Museum of Natural History. Protein electrophoresis protocols for C. brachyotis and H. fischeri are described in Peterson & Heaney (1993) and all data included here for these species are from that study. For the remaining species, equal portions of heart, liver, and skeletal muscle tissue were homogenized in a 1 mM disodium EDTA/100 mM Trizma base/0.2 mM NAD, NADP, and ATP buffer, centrifuged for 45 min at 12,000 rpm, and supernatants drawn into capillary tubes for storage. Samples were electrophoresed for 4–6 h on 12% starch gels, depending on the specific analysis desired. Gels were sliced horizontally, and each slice stained using specific protein assays from Shaw & Prasad (1970) and Harris & Hopkinson (1978). Each sample was scored at 32 presumptive genetic loci (enzyme commission numbers from International Union of Biochemistry and Molecular Biology, 1992): AAT (2.6.1.1, 2 loci), ACN (4.2.1.3, 2 loci), ACP (3.1.3.2), ADH (1.1.1.1, 3 loci), AK (2.7.4.3, 2 loci), ATA (2.6.1.2), CK (2.7.3.2, 2 loci), EST (3.1.1.1, 2 loci), G3PDH (1.1.1.8), G6PDH (1.1.1.49), GDA (3.5.4.3), GPI (5.3.1.9), ICD (1.1.1.42, 2 loci), LDH (1.1.1.27, 2 loci), MDH (1.1.1.37, 2 loci), MPI (5.3.1.8), NP (2.4.2.1), PEP (3.4.11; 5 loci, corresponding to PEP-A, -B, -C, -D, -S), PGD (1.1.1.44), PGM (5.4.2.2), PK (2.7.1.40), and SOD (1.15.1.1). For each species, all individuals were analysed on the same gel. To assure correct assignment of homologies, reference individuals were included at multiple points on each gel. Analyses were performed in BIOSYS-1 (Swofford & Selander, 1981). Allele frequencies and three measures of within-population variation – mean observed heterozygosity (Hobs), number of alleles per locus (NALL), and percentage of loci polymorphic (POLY; 5% criterion) – were calculated. Departure from Hardy–Weinberg equilibrium was tested by 3 methods (the chi-square goodness-of-fit test, and an exact probability test (Haldane, 1954); because our samples were sometimes small, we also used a chi-square test with Levene (1949) correction for small sample sizes) (Table 1; Appendix S1 in Supplementary Material). Tests of association between measures of within-population variation (Hobs, NALL, POLY) with island area and isolation were performed with linear regression and non-parametric Spearman rank correlation. Tests for differences between species in levels of withinpopulation variation were performed with analysis of variance and non-parametric Kruskal–Wallis tests. Fixation indices (F-statistics; Wright, 1951, 1965) were used to summarize the distribution of genetic variation within and between populations. Confidence limits on estimates of FST were established by jack-knifing over loci as recommended by Weir & Cockerham (1984); we employed a relatively conservative experimentwise error rate (a ¼ 0.001). Hierarchical fixation indices (Wright, 1978) and variance components (Cockerham, 1969, 1973; Weir, 1990) were calculated based on the Pleistocene connections among islands as follows: (Luzon, Catanduanes, Polillo) (Leyte, Biliran) (Negros) (Mindoro) 233

L. R. Heaney et al. Table 1 Genetic variation within populations (Hobs, observed heterozygosity) and island areas Widespread

Island

Area (km2)

Luzon Negros Mindoro Leyte Catanduanes Polillo Biliran Sibuyan Fuga Dalupiri Barit Mean Hobs Mean NALL Mean POLY

104,688 12,704 9736 7213 1430 606 497 448 93 62 5

Endemic

Narrow endemic

Rousettus amplexicaudatus

Macroglossus minimus

Cynopterus brachyotis

Ptenochirus jagori

Haplonycteris fischeri

Rattus everetti

Ptenochirus minor

0.145 0.111 – 0.115 0.118 0.091 0.090 0.103 0.090 0.075 0.103 0.104 1.36 25.5

0.058 0.051 0.050 0.042 0.025 – 0.033 0.063 – – – 0.046 1.20 14.3

0.065 0.083 – 0.060 0.071 – 0.097 0.040 – – – 0.069 1.45 26.2

0.022 – – 0.032 0.024 – 0.022 0.034 – – – 0.027 1.20 10.3

0.054 0.056 – 0.000 0.036 – 0.036 0.022 – – – 0.034 1.20 19.0

0.025 – – 0.048 0.017 – 0.025 – – – – 0.029 1.20 11.5

– – – 0.027 – – 0.028 – – – – 0.028 1.15 12.0

Summary lines include mean Hobs, mean NALL (number of alleles per population), and mean POLY (proportion of loci polymorphic per population).

(Sibuyan) (Dalupiri) (Fuga, Barit). Estimates of overall gene flow between populations (Nm) were derived from the approximation FST ¼ 1/(1 + 4Nm) as recommended by Slatkin & Barton (1989), and also computed by the private alleles method of Slatkin (1985) using the correction for sample size of Slatkin (1985) as clarified in Slatkin & Barton (1989). Both methods provide reasonable estimates of Nm; Slatkin & Barton (1989) found that these two methods of estimating Nm bracketed the true values of Nm in some simulations. Gene flow among pairs of populations (M; Slatkin, 1993) was calculated by the GST (Nei, 1973) and theta (1; Weir & Cockerham, 1984) methods, using a program by Slatkin (1993). We present the results for M as calculated by the GST method (Nei, 1973, 1977) because estimates of gene flow most commonly reported in the literature are calculated by this method (e.g. it is the method used by BIOSYS-1; Swofford & Selander, 1981). M between Pleistocene islands is simply the mean of all estimates along that track. Cavalli-Sforza & Edwards’ (1967) arc genetic distance (as preferred by Wright, 1978) and Nei’s (1978) unbiased genetic distance were calculated (Appendix S2). Geographic distances were measured as nearest shore-to-shore distances from bathymetric charts (Department of Defense, Defense Mapping Agency charts, Part 2 – Hydrographic Products, Region 9 – East Asia) at the smallest practical scales; distances between Pleistocene shorelines were estimated using the 120 m bathymetric contour from the same charts (Appendix S3). To test the correspondence of genetic and geographic distances, we used permutation-based matrix correlation tests (Mantel, 1967; Dietz, 1983). These tested the proportionality of Cavalli-Sforza and Edwards’ arc genetic distance matrices and geographic distance matrices (between both present-day and inferred Pleistocene shorelines), using a FORTRAN 234

program (MATCORR.EXE) by Dietz (1983). We present the results of the Spearman rank correlations for matrices rather than the more commonly used Mantel test (which uses Pearson product-moment correlation; Mantel, 1967) because the Spearman test is less sensitive to the actual distance measure used (Dietz, 1983) and it is more appropriate when there is less certainty about the reliability of close ranks (Sokal & Rohlf, 1981). The ‘small oceanic islands’ in this study (Sibuyan, Barit, Dalupiri, and Fuga) did not coalesce with large Pleistocene islands at the last glacial maximum. Each originated as a de novo oceanic island and has remained continuously isolated, as described above. Peterson & Heaney (1993) found that populations of Cynopterus and Haplonycteris on these small, isolated oceanic islands showed weakly reduced withinpopulation variability, and predicted that reduced variability would be found in other species. To test this hypothesis, we compared Hobs, NALL, and POLY for each species between large Pleistocene island populations and small oceanic island populations using analysis of variance and the non-parametric Kruskal–Wallis test. Peterson & Heaney (1993) also found that populations on the small, isolated oceanic islands were consistently distinct from other islands in the Philippines, and predicted that this pattern would be evident in other species. To test this hypothesis, we examined genetic distance and M. All tracks leading to small oceanic islands were distinguished from all tracks that did not include an oceanic island (i.e. tracks within and between large Pleistocene islands). For each species, small oceanic island tracks were compared with large Pleistocene island tracks by the Kruskal–Wallis test, which utilizes a conservative number of degrees of freedom. Throughout our analyses, there are instances in which a given hypothesis is tested across several species. In such cases, Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

Geology, ecology, and genetic differentiation in Philippine mammals it is possible that no individual test of the hypothesis will prove statistically significant although a significant overall effect exists. The growing field of meta-analysis (e.g. Gurevitch et al., 1992) provides many methods for combining results from different studies. We restrict ourselves to Fisher’s relatively simple method of combined probabilities (Fisher, 1954; Sokal & Rohlf, 1981) for testing for overall significance. In our application of the method, each species is treated as an independent test of the hypothesis under consideration, e.g. that Hobs is related to log(area). The sum across species of the natural logarithms of the P-values for each such regression or rank correlation is multiplied by )2. This value is distributed as a chi-square with 2 k degrees of freedom, where k is the number of separate tests and probabilities (in this case, the number of species). Relationships between populations were explored through phenetic and phylogenetic methods using the genetic distance measures noted above. UPGMA dendrograms (Sneath & Sokal, 1973) and Distance Wagner trees (Farris, 1972) were generated in BIOSYS-1 (Swofford & Selander, 1981). Fitch (Fitch & Margoliash, 1967) trees were generated in PHYLIP (Felsenstein, 1989). A maximum parsimony analysis of allele frequencies was conducted using the FREQPARS program of Swofford & Berlocher (1987). All of these methods yield similar topological results, and so the UPGMA dendrograms using Cavalli-Sforza and Edwards’ arc genetic distance are presented by convention (Fig. 3). RESULTS Departures from Hardy–Weinberg equilibrium The largest number of departures from Hardy–Weinberg equilibrium was detected by the Levene (1949) correction for small sample sizes to the chi-square test. This method detected 17 departures from Hardy–Weinberg equilibrium, of 225 possible polymorphic loci-populations. Many fewer were detected by the chi-square test for goodness-of-fit and the exact probability test. All departures were heterozygote deficiencies of small degree and were concentrated in a few loci [e.g. G6PDH (5), NP (3), and ICD (2)]. Removal of these loci does not qualitatively affect any of the conclusions presented here. Population dendrograms Two consistent trends are apparent in the UPGMA population dendrograms (Fig. 3). First, the small oceanic island populations (Sibuyan, Dalupiri, Fuga, and Barit) are often the most strongly differentiated from other populations. Sibuyan, in particular, appears most genetically distinct in three of the five taxa that are found on that island. The other trend that emerges is that Biliran and Leyte, the pair of islands with the shallowest ocean depth between them (< 10 m), are the least differentiated in two of six taxa and are not well differentiated in a third. The presumed Pleistocene hierarchy of island Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

relationships, however, emerges clearly only in the case of R. everetti, the only non-volant mammal in this study. It should be noted that cluster analysis of island populations based on genetic distances constitutes an exploratory technique and is not appropriate for testing the hypothesis that Pleistocene shorelines form significant partitions of genetic variation (Sneath & Sokal, 1973). Variation within populations: geological correlates Genetic variability within populations is theoretically related to effective population size (Wright, 1931). Smaller populations are expected to lose genetic variation via genetic drift more rapidly than larger populations, thus achieving a lower standing level of variation due to mutation-drift balance. This expectation has received much attention in the conservation biology literature (e.g. Soule´ 1976, 1987; Frankham, 1995, 1996; Lande, 1995; Berry, 1998), yet how often this phenomenon is important in wild populations is not clear. To test whether this theoretical relationship is present in these Philippine mammals, we used island area, which varies over six orders of magnitude in this study, and the logarithm of island area, as reasonable proxies for effective population size (P. minor, represented by only two populations, was not considered in this analysis). In only one of the six species was a significant regression of within-population variation and island area, detected (similar results were obtained for all three measures of within-population variation; only the results for Hobs are reported). Rousettus amplexicaudatus, a widespread SE Asian fruit bat, displayed significant regressions of Hobs vs. island area and the logarithm of island area {Hobs ¼ 0.098 + 0.000(area), r ¼ 0.775, P ¼ 0.008; Hobs ¼ 0.071 + 0.005[log (area)], r ¼ 0.735, P ¼ 0.015}. Ptenochirus jagori displayed a significant regression of the number of alleles on the logarithm of island area {NALL ¼ 0.772 + 0.138[log(area)], r ¼ 0.963, P ¼ 0.008}, but NALL is a highly sample-size-dependent measure and there was also a significant regression of sample size (N) on island area for this species {N ¼ )14.883 + 7.964[log(area)], r ¼ 0.938, P ¼ 0.019}, so we exercise caution in interpreting this result. Fisher’s method of combined probabilities for all species combined across measures did not yield any significant overall results for relationships of Hobs, NALL, and POLY with area or log(area). Non-parametric methods, including Spearman rank correlation (Sokal & Rohlf, 1981), did not show more significant associations of within-population variability measures and area than would be expected by chance for all taxa other than R. amplexicaudatus, nor were any significant overall trends detected using combined probabilities from non-parametric tests. We also considered the possibility that isolation of island populations may have an effect on within-population variability through lowered frequency of migration. Schmitt et al. (1995) showed that heterozygosity within island populations of the fruit bat C. nusatenggara in Indonesia was correlated with distance to the nearest large source population, and Hisheh 235

L. R. Heaney et al. et al. (1998) found similar results for Eonycteris spelaea in the same area. The analogy with patterns of interspecific diversity is clear: more isolated islands are predicted to receive fewer colonists and tend to have fewer species than islands closer to source pools of colonists (MacArthur & Wilson, 1967). Similarly, one might expect that more isolated islands would receive fewer migrants to augment their genetic diversity. Various indices of isolation (distance to the nearest island, distance to the nearest large island, distance to the nearest Pleistocene island, and distance to the nearest large Pleistocene island) were used in our study. No more significant results than expected by chance were obtained for regressions of measures of within-population variability on isolation indices, for partial regressions of within-population variability on area [and log(area)] plus isolation indices, nor for overall trends across species from combined probabilities. Hence, our data indicate that neither island area nor isolation of island populations from sources of migrants have a substantial overall effect on standing genetic variation within populations in this system, in spite of variation in island area over six orders of magnitude. Variation within populations: ecological correlates We have documented that the widespread SE Asian species that occur in the Philippines usually occupy disturbed habitats, while the Philippine endemic species prefer primary rain forest (although some can maintain populations in disturbed forest; e.g. Heideman & Heaney, 1989; Heaney et al., 1998, 1999; Rickart et al., 1991, 1993). Our previous study of C. brachyotis and H. fischeri found that the widespread C. brachyotis, which prefers disturbed, open habitat, showed high levels of variation within populations, and the endemic species (H. fischeri), which preferred closed-canopy primary forest, showed low levels of variation within populations (Peterson & Heaney, 1993). We postulated that this trend constituted a general pattern. To test this proposition, we performed nested analyses of variance on Hobs, NALL, and POLY in our seven species (Table 1), where the levels are ecological category (widespread/ habitat tolerant species vs. endemic/forest-associated species plus single-Pleistocene-island endemics), species within ecological category, and island populations as the replicates within species. For all three measures, both levels were significant at P < 0.01. This result indicates that, while significant heterogeneity exists within ecological categories, the widespread SE Asian species, which prefer disturbed habitat, display significantly higher levels of within-population variability than species endemic to the Philippines, which prefer primary forest, as predicted. Since heterozygosity and other measures of within-population variation usually do not meet the assumptions of parametric statistics (they are skewed positively in this case), we also performed non-parametric statistical tests. Because no appropriate non-parametric analogue for nested anova is available, we took two approaches. First, we performed a Kruskal–Wallis test using each island population of each 236

species as an independent observation and tested whether or not widespread SE Asian species had greater within-population variability than endemic species and the single-island Pleistocene endemic. For all three measures, all tests are significant (P < 0.01; Table 2). Second, a more conservative approach was applied, treating each species as an independent observation. Species means of the three measures of genetic variation were used to test the hypothesis that widespread species (which prefer disturbed habitat) have higher levels of within-population genetic variation than endemic species (which prefer primary forest; Table 2, with mean values in Table 1). Hobs was significantly higher in widespread species (P < 0.05), and NALL and POLY were nearly so (0.10 < P < 0.05), again supporting the prediction. Variation between populations: geological correlates Wright (1943) provided the theoretical underpinnings for a simple notion, that populations closer to one another should be more similar to one another than populations farther apart due to the homogenizing effects of gene flow. The quantitative theory of ‘isolation by distance’ is mathematically complex and can be difficult to test in its particulars, but numerous authors, notably R. Sokal and colleagues (Sokal & Wartenberg, 1983; Sokal, 1988; Livshits et al., 1991), have relied on nonparametric, permutation-based, matrix correspondence techniques to test for the presence of an isolation by distance pattern. We used these methods to test statistically the proportionality of matrices of genetic distances and geographic distances between populations. We compared matrices of genetic distance (results for Cavalli-Sforza and Edwards’ arc genetic distance are presented in Table 3; other genetic distance measures yielded similar results) with matrices of geographic distance between present-day islands. The tests for all species yield a similar and rather surprising result: matrices of genetic distances between island populations are not correlated with the matrix of geographic distances between present-day islands in any of the six species, nor is an overall

Table 2 Comparisons of genetic variation within populations between widespread SE Asian species and Philippine endemics Hobs

NALL

POLY

n

(a) Widespread SE Asian Philippine endemic P-value

641 179 0.001

573 247 0.007

570 250 0.004

23 17

(b) Widespread SE Asian Philippine endemic P-value

18 10 0.034

16.5 11.5 0.079

17 11 0.077

3 4

Sum of ranks and significance values from Kruskal–Wallis test on observed heterozygosity (Hobs), number of alleles per locus (NALL), and proportion of loci polymorphic (POLY), using (a) islands as observations, and (b) species as observations. Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

Geology, ecology, and genetic differentiation in Philippine mammals Table 3 Spearman rank correlation (rs) tests of proportionality of genetic and geographic distance matrices between presentday islands and between Pleistocene islands

Species

Present distance

Pleistocene distance

rs

rs

P-value

P-value

Widespread species Rousettus amplexicaudatus 0.088 0.332 Macroglossus minimus 0.027 0.472 Cynopterus brachyotis )0.280 0.811

0.339 0.057 0.505* 0.039 0.627 0.100

Endemic species Ptenochirus jagori Haplonycteris fischeri Rattus everetti

0.556* 0.033 0.718 0.061 0.828 0.333

0.553 0.083 0.149 0.279 0.829 0.125

)2[Eln(P)]

15.816

31.438*

Combined probability

P > 0.100

P < 0.005

*Significant correlations and combined probabilities (P < 0.05).

significant correlation detected using combined probabilities (Table 3). However, another relevant pattern requires testing. Patterns of mammalian interspecific diversity in the Philippines are strongly influenced by the boundaries of islands that existed during maximal late Pleistocene sea-level lowering, which represent maximum coalescence during the history of the archipelago (Heaney, 1986, 1991a, 1993, in press; Steppan et al., 2003). This pattern of interspecific diversity suggested another test: we calculated matrix correspondence between genetic distance matrices and the matrix of geographic distances between Pleistocene shorelines, inferred from the 120 m bathymetric contour. Spearman rank correlation yielded significant results for two of six taxa (M. minimus and P. jagori) and nearly significant results (0.10 > P > 0.05) for three additional taxa (R. amplexicaudatus, C. brachyotis, and H. fischeri) for these Pleistocene distances (Table 3). Using combined probabilities across species, we conclude that genetic distances are not correlated with distances between present-day shorelines, but are correlated with distances between late Pleistocene shorelines. It appears, therefore, that the geological history (including both long-term isolation between some islands and Pleistocene coalescence among others) has created a common pattern of geographic variation between populations across the six species, partitioned by late Pleistocene shorelines. Specifically, the geographic distance between permanently isolated geo-historical units is positively correlated with increasing genetic distance. Variation between populations: ecological correlates A final prediction about the extent of the common pattern of differentiation between populations will be addressed here. We believe that the presence of the widespread species on small, isolated islands where no endemic species are found (e.g. Barit, Dalupiri, and Fuga) and the observation that widespread Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

Table 4 Estimates of gene flow (Nm)* and genetic differentiation (FST)  among islands

Species

Nm

NmPA

FST

FST* + 99.9% confidence limits

Widespread species Rousettus amplexicaudatus Macroglossus minimus Cynopterus brachyotis

2.25 2.13 2.02

3.93 3.03 7.29

0.100 0.105 0.110

0.101 ± 0.015 0.105 ± 0.007 0.110 ± 0.063

Endemic species Ptenochirus jagori Haplonycteris fischeri Rattus everetti

2.84 0.16 0.23

5.60 0.95 2.47

0.081 0.606 0.522

0.081 ± 0.007 0.603 ± 0.152 0.520 ± 0.156

Pleistocene endemics Ptenochirus minor

5.85

2.95

0.041

0.041 ± 0.015

*Nm, calculated from FST, after Slatkin & Barton (1989); NmPA, calculated by the method of private alleles (Slatkin, 1985), corrected for sample size.  FST, from BIOSYS-1 output [calculated in the manner of Nei (1977)]; FST*, jack-knifed estimate of FST, after Slatkin (1985), ±jack-knifed confidence limits (a ¼ 0.001).

species will readily cross clearings but endemic species are rarely found far from primary rain forest or out from beneath good canopy cover (except P. jagori, as noted below), should result in lower levels of gene flow between populations of the endemic species on different islands. If this is correct, population genetic theory predicts that endemic species will display higher levels of between-population differentiation than widespread species. To test this prediction, we examined estimates of gene flow (Nm) and fixation indices (FST). Estimates of gene flow generally conform to our predictions (Table 4). The widespread species exhibit high levels of gene flow (about two individuals per generation) and two of three endemic species exhibit low levels of gene flow (around 0.2 individuals per generation). The endemic species P. minor displays very high levels of gene flow, but this is not surprising, since it occurs only within a single Pleistocene island and the results of the previous section indicate that Pleistocene shorelines are the significant boundaries between regions of genetic differentiation. The more widespread Philippine endemic, P. jagori, a member of the same genus, however, has rather high levels of gene flow, as might be expected of a species that is known to be relatively tolerant of disturbance, although not to the extent of the widespread species. We note that estimates of gene flow may be based on past, rather than current, gene flow, and that the time to reach genetic equilibrium is largely a function of effective population size. Thus, these values should be taken as average long-term estimates, rather than instantaneous estimates. With respect to levels of between-population differentiation, FST is fairly low for widespread species, c. 0.1 (Table 4). Two of three endemic species (H. fischeri and R. everetti) show significantly higher levels of between-population differentiation, about five times higher than widespread species. The 237

L. R. Heaney et al. geographically restricted P. minor displays very low levels of differentiation within its single Pleistocene island (although we have few sampling sites), a trend that we discuss below. The more widespread endemic species, P. jagori, presents an illustrative exception to the other endemic species. It displays low levels of within-population variation, but unlike the other endemic species, its habitat tolerances are moderately broad (i.e. it often forages in open habitats and readily maintains populations in secondary forest), and it displays high levels of overall gene flow. We interpret this result as evidence that it is the not the categorization of a species as an endemic per se that is usually associated with low gene flow, but rather with the usual (but not ubiquitous) tendency of endemic species to have low tolerance for disturbed, open habitat, since P. jagori apparently has the broadest habitat tolerance of any endemic Philippine fruit bat (although less than the widespread SE Asian species). Morphological data (Walsh, 1998) show a similar but less pronounced trend, in that levels of withinpopulation variation in P. jagori are similar to those of other endemic species, while levels of between-population variation resemble those of widespread species. Wright’s (1978) hierarchical F-statistics (Table 5) show how overall variation between populations may be apportioned into variation between Pleistocene islands and variation between present-day islands within Pleistocene islands. Two of three Philippine endemics show the overwhelming proportion of their between-population variation structured by the bound-

Table 5 Wright’s (1978) hierarchical F-statistics illustrating genetic differentiation among Pleistocene islands vs. present-day island populations

Species

Level FXY

Widespread species Rousettus amplexicaudatus GPI ISL Total Macroglossus minimus GPI ISL Total Cynopterus brachyotis GPI ISL Total Endemic species Ptenochirus jagori

Haplonycteris fischeri

Rattus everetti

GPI ISL Total GPI ISL Total GPI ISL Total

Variance Percentage component variance

0.003 0.045 0.048 0.021 0.040 0.061 0.070 0.021 0.088

0.0113 0.1449 0.1562 0.0280 0.0536 0.0816 0.0855 0.0226 0.1081

7 93 100 34 66 100 79 21 100

0.029 0.020 0.048 0.537 0.100 0.583 0.492 0.030 0.507

0.0223 0.0152 0.0375 0.7818 0.0675 0.8493 0.8279 0.0253 0.8532

59 41 100 92 8 100 97 3 100

GPI, among Pleistocene islands; ISL, among present-day islands within Pleistocene islands.

238

aries of Pleistocene islands (H. fischeri, 92%; R. everetti, 97%). They exhibit almost no variation between present-day islands within Pleistocene islands. This lack of differentiation between populations within Pleistocene islands is consistent with the observation of low FST in P. minor, which occurs only on a single Pleistocene island. The third endemic species, P. jagori, also displays over half of its variation between populations at the between-Pleistocene-island level (59%), consistent with its moderately broad habitat preferences. Two of three widespread species also display substantial proportions of their betweenpopulation genetic variation at the between-Pleistocene-island level (C. brachyotis, 79%; M. minimus, 34%). In contrast, R. amplexicaudatus shows relatively little of its betweenpopulation genetic variation structured according to the boundaries of Pleistocene islands (7%). Overall patterns Two less commonly employed modes of analysis will be discussed here to help further illuminate the pattern and extent of genetic differentiation in these Philippine mammals. Cockerham’s (1973) extension of Wright’s method of variance components is used to show the complete breakdown of total genetic variation (Table 6), including variation within present island populations which Wright’s hierarchical F-statistics do not illustrate. Additionally, Slatkin’s (1993) M-statistics are used to estimate gene flow between all pairs of islands (Fig. 2). Some clear trends emerge. First, almost no differentiation is found between present-day island populations within Pleistocene islands for these species (Table 6). Second, there is an overall pattern of genetic differentiation between Pleistocene island groups. Two of three endemic species exhibit substantial proportions of their total genetic variation at the betweenPleistocene-island level (H. fischeri and R. everetti). Most of the widespread species also show more variation between Pleistocene islands than between present-day islands within Pleistocene islands, as does the endemic species, P. jagori. Only R. amplexicaudatus appears to display a pattern of variation in which Pleistocene shorelines do not form the primary partitions of genetic variation. Several further generalizations about patterns of gene flow between populations as inferred from allozyme data can be discovered by estimating gene flow (Nm) between all pairs of populations using M-statistics (Slatkin, 1993; Fig. 2). The first general pattern that emerges is that levels of gene flow are higher between Leyte and Biliran than between Luzon and Catanduanes for most species, including all endemic species. This pattern of genetic similarity is consistent with two factors: (1) Leyte and Biliran are presently closer to one another than are Luzon and Catanduanes and may experience higher rates of present-day gene flow; and (2) Leyte and Biliran are separated by a much shallower ocean channel than are Luzon and Catanduanes (Heaney, 1986) and genetic similarity may reflect the more recent separation of Leyte and Biliran. The second general pattern that emerges is that gene flow to Sibuyan appears almost uniformly attenuated (except in M. minimus). Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

Geology, ecology, and genetic differentiation in Philippine mammals Table 6 Cockerham’s (1973) method of complete variance partitioning. Each level is expressed as a percentage of the total variance Widespread Rousettus amplexi-caudatus (16/29)

Endemic

Narrow endemic

Macroglossus minimus (15/30)

Cynopterus brachyotis (9/14)

Ptenochirus jagori (15/29)

Haplonycteris fischeri (10/14)

Rattus everetti (10/26)

Ptenochirus minor (7/29)

3.6 3.4 93.0 100.0

6.7 2.6 90.7 100.0

3.8 1.4 94.8 100.0

30.7 0.8 68.5 100.0

23.9 1.3 74.8 100.0

– 3.5 96.5 100.0

All loci (monomorphic loci calculated as 100% at lowest level) GPI 1.4 1.8 4.3 ISL 2.6 1.7 1.7 W/I 96.0 96.5 94.0 Total 100.0 100.0 100.0

1.9 0.7 97.4 100.0

21.9 0.6 77.5 100.0

9.2 0.5 90.3 100.0

– 0.8 99.2 100.0

Variable loci only GPI 2.4 ISL 4.8 W/I 92.8 Total 100.0

Results are tabled separately for all loci and for variable loci only. Numbers in parentheses indicate the number of variable loci over the total number of loci for each taxon. Negative variance components are treated as equal to zero. GPI, between Pleistocene islands; ISL, between present-day islands within Pleistocene islands; W/I, within present-day islands.

Figure 2 Estimates of gene flow between selected populations [M after Slatkin (1993)]. Estimates of gene flow between Pleistocene islands are averages of all estimates along that track. Abbreviations are: BIL, Biliran; CAT, Catanduanes; LEY, Leyte; LUZ, Luzon; NEG, Negros; SIB, Sibuyan. Philippine endemic species are on the right, non-endemics are on the left.

This could be the result of smaller or more isolated islands being more difficult targets for migrants to hit, although the decrease in levels of migration does not appear to be sufficient Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

to influence levels of within-population variability, as noted below. Finally, endemic species display lower levels of gene flow between Pleistocene islands than within Pleistocene 239

L. R. Heaney et al. islands, with one exception (H. fischeri: Greater Luzon/Greater Negros-Panay). Widespread species show no clear differences in estimates of gene flow within and between Pleistocene shorelines. This result indicates that the permanent oceanic barriers between Pleistocene islands are of greater consequence, especially for endemic species, than recent seawater barriers within Pleistocene islands which have been inundated by sea-level rise during the past 18,000 years. This is true in spite of the fact that some of the more recently formed water gaps within Pleistocene islands are larger than distances between Pleistocene islands (Appendix S3). These overall patterns show the influences of both geological and ecological processes: the Pleistocene history of the Philippine archipelago has created a common pattern of significant differentiation between, but not within, Pleistocene islands. Further, widespread species associated with disturbed habitats show less variation between and more variation within populations than species endemic to the Philippines associated with primary rain forest. Differentiation of small oceanic island populations In general, the populations on the small, isolated oceanic islands in this study (Sibuyan, Barit, Dalupiri, and Fuga) displayed only weakly reduced within-population variability as might be expected due to their small size and isolation. Hobs, NALL, and POLY were significantly lower in oceanic island populations only for R. amplexicaudatus, and combining probabilities across the five relevant species did not detect a significant overall result. There was, however, a consistent trend for small oceanic island populations to be more strongly genetically differentiated than large Pleistocene island populations. Cavalli-Sforza and Edwards arc genetic distance and M were significantly higher and lower, respectively (both P < 0.01), along tracks leading to small oceanic islands than along other island tracks in three of five species (C. brachyotis, H. fischeri, and R. amplexicaudatus) and nearly so in one species (P. jagori; both P  0.10). Combining probabilities yields a significant overall result for both measures (P < 0.001) and we conclude that tracks to small, isolated oceanic islands are characterized by higher genetic distances and lower estimates of gene flow than tracks between large Pleistocene islands. DISCUSSION Geological history vs. ecology? This investigation of genetic variation in Philippine mammals makes two general points clear. First, geological history is of paramount importance in structuring patterns of variation between populations of the seven species we studied. The geological history of the Philippine archipelago has been characterized by the long-term uplift and short-term (Pleistocene) coalescence and fragmentation of groups of islands. The maximal late Pleistocene shorelines define biogeographic units 240

with no history of dry-land connections to other islands or continents. Populations of mammals on modern islands within ‘Pleistocene islands’ have been separated from one another for c. 18,000 years; these show little or no genetic variation between them. In contrast, populations on islands separated by permanent barriers to dispersal have developed substantial genetic differentiation. This pattern is concordant with patterns of interspecific diversity in Philippine mammals; late Pleistocene shorelines delimit faunal regions for mammals with high levels of endemism among non-volant mammals (up to 80%) and moderate levels of endemism among fruit bats (up to 21%; Heaney, 1986, 1991a, 1993, in press). The importance of Pleistocene history in structuring genetic variation within mammalian species in the islands of Wallacea has also been addressed by Schmitt et al. (1995), Hisheh et al. (1998), and Maharadatunkamsi et al. (2003). Their examination of differentiation in the fruit bats C. nusatengarra and E. spelaea from the Lesser Sunda Islands of Indonesia concluded that patterns of genetic distances between island populations are associated with recent colonization from west to east along the island arc (we note they are also consistent with the isolation-by-distance effect), and are more closely correlated with distances between Pleistocene shorelines than with distances between present shorelines, although the overall level of population subdivision that they recorded was markedly lower than in this study. In contrast to Schmitt et al.’s (1995) and Hisheh et al.’s (1998) results, we detected little evidence of a relationship of island area (which is presumably related to effective population size) or island isolation (which is presumably related to levels of gene flow) to amounts of variation within populations. Similarly, Juste et al. (2000) found that, of several insular populations of an African fruit bat, Eidolon helvum, the most isolated of these showed much greater genetic differentiation than the others, in association with reduced gene flow. Second, the ecological attributes of individual species influence the extent to which, and the manner in which, common historical signals are expressed. Our knowledge of natural history and distributions of these mammals successfully predicted whether or not they would demonstrate large or small degrees of genetic differentiation between populations (including, in most respects, the ecologically intermediate P. jagori) and whether or not they would display relatively high or low levels of variation within populations. This result has important implications for understanding evolution – any attempt to make general statements about rates of genetic change between populations must take into account the ecological attributes of the species in question. We also stress that these insights into the common regional pattern of variation and particularly into the differences in expression of that pattern are unlikely to have been gained by the examination of a single species. Generalizations about the patterns of geographic variation and their causes are best achieved through comparative studies such as this one, and similar studies (e.g. Caccone, 1985; Waples, 1987; Brumfield & Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

Geology, ecology, and genetic differentiation in Philippine mammals Capparella, 1996; Zink et al., 2000; Riddle & Hafner, in press). Thus, we conclude that geological history and ecological traits produce interactive processes, and must be considered simultaneously for accurate conclusions about general genetic processes in nature to be reached. Small islands + isolation = speciation? One of the motivations for studying geographic variation is the identification of especially distinctive populations. To the degree that the origin of new species is an extension of divergence between populations within species, identifying divergent populations can provide insight into the processes of speciation. In his classic volume, Mayr (1963) identified ‘geographical isolates’ as those populations most likely to produce new species; discussing island populations in particular, he characterized geographical isolates as typically inhabiting islands of small area and differing environmental conditions, having low population sizes, possessing distinctive morphologies or behaviours, and being spatially or temporally separated from other populations by barriers to gene flow. For the Philippine mammals in this study, populations inhabiting the small, isolated oceanic islands (Sibuyan, Barit, Dalupiri, and Fuga) that were not connected as part of any large Pleistocene island appear to qualify as geographical isolates. These small oceanic islands are much smaller in area than the large Pleistocene islands and have never had dry-land connections to them. All of them except Sibuyan also have depauperate mammal faunas (Heaney et al., 1998, L.R. Heaney et al., unpubl. data). These island populations do not show evidence of substantially reduced genetic variation within any given species. We assume that effective population size is limited by island area. While these small oceanic islands are indeed small by most standards, they do not appear to be small enough to erode within-population genetic variability as has been suggested for other small island populations of mammals (Berry, 1986, 1998). Using the population density estimates of Heideman & Heaney (1989) for fruit bats on Negros Island as first approximations, it seems likely that bat populations on even the smallest of our islands number at least in the thousands – far above the numbers typically believed to cause noticeable loss of genetic variation (Lande & Barrowclough, 1987). Although the populations on small, isolated islands do not show reduced genetic within-island variation within a given species, they are more genetically distinct. For each species, all oceanic island populations possess unique alleles, and the population of H. fischeri from Sibuyan is fixed for one unique allele (Appendix S1). Further, the genetic distances between populations evident in Fig. 3 are greater between Pleistocene islands than within them, and are typically greatest to the small oceanic islands. Whether this genetic distinctiveness is due to selective or neutral factors is not evident from our data; certainly, it is conceivable that the differing biotic and abiotic environmental conditions on these small oceanic islands may impose different selective pressures on these populations. However, the rate of fixation of genes due to genetic drift is a Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

Figure 3 UPGMA dendrograms of island populations based on genetic distance (Cavalli-Sforza & Edwards (1967) arc distance). Abbreviations are: BAR, Barit; BIL, Biliran; CAT, Catanduanes; DAL, Dalupiri; FUG, Fuga; LEY, Leyte; LUZ, Luzon; MRO, Mindoro; NEG, Negros; POL, Polillo; SIB, Sibuyan. All dendrograms drawn to same scale.

function of the inverse of effective population size (Kimura, 1983) and it may be that these small oceanic island populations are sufficiently small to have accelerated rates of fixation of new alleles relative to the large island populations. It is unlikely that the genetic distinctiveness of small isolated island populations is solely a reflection of the phylogenetic history of these populations. Peterson & Heaney (1993) discussed the importance of establishing population-level phylogenies for understanding the biogeographic history of these taxa in the Philippines. They objected to interpreting UPGMA dendrograms, which are the results of cluster analyses of distance data utilizing mid-point rooting, as representations of phylogeny, preferring that estimates of phylogenetic relationships be based on parsimony analyses of character data using outgroup comparisons. Additional objections to the use of allozymes in phylogenetic analyses are that allozyme frequencies are temporally unstable (Gaines et al., 1978; Crother, 1990) and that transitions between electromorphs 241

L. R. Heaney et al. cannot be reliably ordered (Lewontin, 1991). As noted by Peterson & Heaney (1993), the distinctive small oceanic island populations are very unlikely to represent basal divisions vs. other populations in a phylogenetic sense due to their geographic setting (well away from likely avenues of colonization) and to the fact that they are among the geologically most recent of the Philippine islands (Steppan et al., 2003). This implies, for example, that the long branches associated with Sibuyan Island (Fig. 3) are due to rapid evolution on Sibuyan, not to earlier isolation than other populations. Our interpretation could be tested by use of intraspecific phylogenies generated from DNA sequences, as currently being developed by T. Roberts (pers. comm.). Gene flow: endemism or habitat requirements? This report yields two additional insights particularly relevant to our understanding of mammals and the Philippine biota. First, it seems an intuitively obvious expectation that a nonvolant mammal should have rates of dispersal across water barriers orders of magnitude lower than those of a volant mammal. It is therefore remarkable that the endemic rodent R. everetti shows levels of gene flow between Pleistocene islands comparable to that of the endemic fruit bat H. fischeri. One likely explanation for this unanticipated result is that the tolerance for disturbed, open habitat by H. fischeri is far less than that of R. everetti (Rickart et al., 1993; Heaney et al., 1998, 1999): although it is far more effective for small mammals to fly over water barriers than to swim across them, some bat species may cross-water gaps less often than some rodents. This inference suggests the hypothesis that habitat affinity may be as important as mode of dispersal in accounting for variation in levels of gene flow across different taxa. Second, the pattern of variation within and between populations exhibited by P. jagori highlights distinctive aspects of its historical and ecological traits. It is an endemic species that is common in good quality rain forest habitat, and it displays low levels of within-population variation. However, it also persists well in degraded forest and often flies in cleared areas, and it displays high gene flow and little variation between populations. This combination of low within- and between-population variability is often considered to be the earmark of a recent colonizer. It is possible that P. jagori was formerly more restricted in distribution, perhaps on a single Pleistocene island like its sister-taxon, P. minor, and has only recently spread throughout the Philippine archipelago. The data presented here cannot resolve this question, and even if the colonization of the whole of the Philippine archipelago by P. jagori has been a relatively recent event, some differentiation between populations has occurred. The Sibuyan population appears relatively distinct based on the allozyme data presented here (Fig. 2), and morphological data (Walsh, 1998) indicate that on the islands of Greater Mindanao, where P. jagori is sympatric with the smaller P. minor, P. jagori has evolved much larger body size than is found in its other island populations. Investigations currently underway on the 242

phylogenetic relationships of populations of Ptenochirus and related cynopterine fruit bats by T. Roberts using DNA sequencing (pers. comm.) may provide additional insight into the historical association of genetic variation and geographic distribution in these taxa. A simple geographical/ecological model of genetic differentiation in oceanic archipelagos The previous analyses suggest that patterns of genetic variation in these Philippine mammals are strongly influenced by two ecological variables, namely their vagility (especially the ability to fly) and the ability to tolerate open habitats (which in the Philippines are synonymous with heavily disturbed habitats). These variables interact simultaneously in any given species to largely determine its colonizing ability. These two variables, in turn, are associated with varying levels of genetic differentiation, as discussed further below. Among our study species, C. brachyotis and M. minimus are moderately strong fliers, and R. amplexicaudatus is a very strong flier; all three prefer open habitat, and we characterize them as having high colonizing ability. One of the endemic bats (P. jagori) prefers closed-canopy, primary forest, but is a strong flier and maintains populations in open, disturbed forest well; we characterize it overall as having moderate colonizing ability. Two species, H. fischeri and R. everetti, have low colonizing ability, but for different reasons: H. fischeri can fly, but is a relatively weak flier and rarely will fly out from under good canopy cover, while R. everetti does well in disturbed forest (although not in intensive agricultural areas) but cannot fly. Ptenochirus minor prefers good canopy cover but tolerates second growth, and so may have better colonizing ability overall than the prior two species, but its restriction to a single Pleistocene island (Greater Mindanao) implies limited abilities. We further note that C. brachyotis, M. minimus, and R. amplexicaudatus occur on even the most isolated of Philippine islands, whereas P. jagori and R. everetti are absent from some of the most isolated islands (including Barit, Batan, Dalupiri, and Fuga) and H. fischeri is absent from those islands and also from less isolated Siquijor and Camiguin (north of Mindanao; Lepiten 1997; Heaney et al., 1998), providing direct empirical evidence of their limited dispersal abilities. We have found three categories of Philippine islands with respect to geological history: (1) the current islands, many of which aggregated during Pleistocene periods of low sea level (as shown in Fig. 1); within such aggregates, these share very similar faunas; (2) large Pleistocene islands that are surrounded by deep water, which have been isolated continuously throughout their existence but are not very distant from one another; each has a very distinctive fauna, with 40–80% endemism (Heaney, 1998, 2000); and (3) Pleistocene islands that are especially small and isolated by great distance, some of which are depauperate. The last two categories differ primarily in degree, but in the Philippines, they are fairly discrete, as can be seen in Fig. 1. Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

Geology, ecology, and genetic differentiation in Philippine mammals We summarize the two variables of colonizing ability and geographic isolation for our seven study species along two axes in Fig. 4. We present each as a categorical variable rather than a continuous variable because, while conceptually each is continuous, our data demonstrate that they appear to behave in the Philippines in a discontinuous fashion because of historical circumstances; in general, the world is often neither homogeneous nor continuous with respect to these variables. Our data (Table 2, Fig. 2) show C. brachyotis, M. minimus, and R. amplexicaudatus to have consistently high gene flow both within and between Pleistocene islands (Nm of 4–14), although C. brachyotis has Nm as low as 1.2–1.7 on some isolated islands (such as Sibuyan). P. jagori, as expected, has Nm at intermediate levels (3–5.5), and H. fischeri and R. everetti have Nm values always far less than 1.0 between Pleistocene islands. The converse of gene flow, genetic differentiation, is calculated from the same data set as gene flow, and is effectively its inverse, so that species with high gene flow have low differentiation (FST values), and the converse (Tables 2 and 4). In other words, in the absence of high levels of gene flow (defined as Nm less than 1.0 by Wright, 1931), substantial genetic differentiation takes place. Our data demonstrate that these two variables have combined to have a strong association with rates of genetic differentiation, as we show in the third axis in Fig. 4. Within a Pleistocene island (Fig. 4, bottom row), none of our study species showed genetic differentiation. On separate but adjacent Pleistocene islands (Fig. 5, middle row), H. fischeri and

Figure 4 Graphical model of the interaction of colonizing ability and degree of isolation on the extent of genetic differentiation by six species of fruit bats and one murid rodent in the oceanic Philippines. Isolation is low (within islands that merged during Pleistocene periods of low sea level), medium (between islands separated continuously by deep water, but by channels that are narrow), or high (between islands separated continuously by deep water and broad channels). Cb, Cynopterus brachyotis, Hf, Haplonycteris fischeri, Mm, Macroglossus minimus, Pj, Ptenochirus jagori, Pm, Ptenochirus minor, Ra, Rousettus amplexicaudatus, Re, Rattus everetti. Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

R. everetti show moderately high genetic differentiation, but the other species very low differentiation. On distant oceanic islands (Fig. 5, top row), H. fischeri and R. everetti again show heightened differentiation due to low colonizing ability. Somewhat to our surprise, C. brachyotis shows moderate differentiation on isolated islands, perhaps indicating that our estimates of colonizing ability are too high. We found the species with the highest colonizing ability (R. amplexicaudatus) to show virtually no genetic differentiation on distant Pleistocene islands, one that is smaller and a somewhat weaker flier (M. minimus) to show slightly more, and a species with medium colonizing ability (P. jagori) to show somewhat greater genetic differentiation on the same islands. Most crucially for the study of diversification, a comparison of species on a given set of isolated oceanic islands shows that species with high colonizing ability (R. amplexicaudatus and M. minimus) exhibit little genetic differentiation, species with moderate colonizing ability (P. jagori and apparently C. brachyotis) exhibit a moderate amount of genetic differentiation, and species with little colonizing ability (H. fischeri, R. everetti) on the same islands exhibit high genetic differentiation. In other words, the likelihood of any given species developing substantial genetic differentiation increases as it reaches progressively more isolated islands, but a species with ecological traits that reduce its colonizing ability (e.g. an aversion to flying out from beneath the canopy or an inability to fly) will differentiate on less isolated islands than a species with ecological traits that increase the rate of gene flow. This simple description of patterns of interaction between colonization ability and isolation, and the degree of genetic differentiation that is associated, may be used as a model to make predictions for other Philippine mammals, for other Philippine organisms, and for organisms in other oceanic archipelagos. For example, we predict that insectivorous bats in the Philippines will show similar patterns: widespread species that do well in disturbed habitats, such as those in the genera Taphozous, Miniopterus, Myotis, Pipistrellus, and Scotophilus should be very similar to C. brachyotis, M. minimus, and R. amplexicaudatus, but Philippine endemic species in the genera Rhinolophus and some Hipposideros that live in primary forest should be similar to Haplonycteris and Ptenochirus. If there are animals with still less dispersal ability than H. fischeri and R. everetti, perhaps such as the many native shrews and mice, or frogs with very low dispersal, we predict that they will show even lower levels of colonizing ability and higher rates of genetic differentiation; this prediction is at least generally supported by the findings of Brown & Guttman (2002) and Steppan et al. (2003). We predict that Philippine birds will show very similar patterns to these bats overall, with widespread species that favour disturbed habitats having genetic patterns that resemble those of C. brachyotis, M. minimus, and R. amplexicaudatus, whereas those that require primary forest and/or that have limited flight will have genetic patterns that resemble those of H. fischeri. In this study, we found all species to have high colonizing ability and low differentiation within Pleistocene islands, but we note that some other taxa may have 243

L. R. Heaney et al. far lower dispersal abilities and be influenced by more subtle isolating barriers (such as different types of forest), and therefore show higher rates of genetic differentiation within a given modern island (perhaps including frogs and some montane non-volant small mammals, flightless montane invertebrates, etc. that live in montane and mossy forest; Heaney & Rickart, 1990; Heaney, 2001). We predict their relative levels of gene flow and differentiation will be structured similarly to the bats but on a smaller geographic scale, with equivalent ecological correlates regarding habitat fidelity and vagility. Finally, we predict that species on the oceanic islands of Wallacea as a whole, and in other oceanic archipelagos, will show similar patterns of association between colonizing ability, isolation, and differentiation. Ideal test cases would include the birds (Lack, 1976; Ricklefs & Bermingham, 2001) and lizards (Losos & Schluter, 2000; Harmon et al., 2003) of the Caribbean, and the fauna of the Sea of Cortez (Case et al., 2002). CONCLUSION For much of the last three decades, most studies of biological diversity in island ecosystems have approached the topic from either an ecological or an historical/geological perspective, typified by the ecologically-oriented equilibrium model of MacArthur & Wilson (1967) and by the historically-oriented vicariance model (e.g. Rosen, 1975). Moreover, there is generally a tendency for biological research to be deliberately conducted in very simple systems so that the impact of specific processes may be finely parsed. This study has been useful in demonstrating that the question of ‘geological history or ecology’ is based on a false premise that only one or the other is a significant factor; in this case, both are highly important, and we predict that both will be found to be equally important in all other oceanic archipelagos. In this case, current genetic patterns are profoundly influenced by historical (geological) factors, but ecological features of the various species, such as habitat association and vagility, strongly affect rates of gene flow and, therefore, degree of divergence among populations. Indeed, we found that habitat association may in some cases be even more important in determining patterns of genetic variation within species than whether a species is volant or non-volant. Few of our findings could have been derived from studying any single one of the seven species, and few could have been derived had we worked in a simpler set of islands. While very specific and narrowly-oriented studies of simple island systems or single species may be useful and often necessary in adding clarity, studies that emphasize broad contexts, multiple species, and complex regions are likely to be essential for discovering broad patterns and processes. Indeed, this study would have been strengthened by including even more species and islands. The complexity of global biological diversity will best be understood by combining historical and ecological questions and perspectives, and considering them for many species simultaneously; the challenge is not to determine which single model or process predominates, but rather to determine how the processes interact under varying circumstances, and how 244

narrow/focused models can be integrated into comprehensive models that accurately portray the very real complexity of living systems (Case & Cody, 1982; Heaney, 1986, 2000; Whittaker, 1998; Zink et al., 2000; Hewitt, 2001; Arbogast & Kenagy, 2002; Riddle & Hafner, in press). ACKNOWLEDGEMENTS We gratefully acknowledge the assistance of many colleagues and institutions with the field portion of this study, especially A. C. Alcala, D. S. Balete, C. Catibog-Sinha, R. I. Crombie, C. Custodio, R. Fernandez, P. C. Gonzales, P. D. Heideman, J. S. H. Klompen, M. Laranjo, M. V. Lepiten-Tabao, W. Pollisco, E. A. Rickart, C. A. Ross, B. R. Tabaranza, Jr, R. C. B. Utzurrum, the Protected Areas and Wildlife Bureau of the Philippines, the Department of Environment and Natural Resources of the Philippines, Silliman University, the Haribon Foundation, and the Philippine National Museum. Permits were provided by the Philippine Department of Natural Resources. We thank E. A. Rickart, P. D. Heideman, T. J. McIntyre, R. S. Thorington, Jr, R. S. Hoffmann, M. J. Carleton, J. H. Brown, G. G. Musser, J. M. Bates, and S. J. Hackett, who have all played important roles in the development of the ideas presented here. We thank John Bates, Shannon Hackett, Trina Roberts, and two anonymous reviewers for their helpful comments on earlier drafts of the manuscript. These studies were supported in part by the US National Science Foundation (BSR-8514223), the Smithsonian Institution Office of Fellowships and Grants, the Ellen Thorne Smith and Barbara Brown Funds of the Field Museum, and the John D. and Catherine T. MacArthur Foundation’s World Environment and Resources Program (90-9272A). SUPPLEMENTARY MATERIAL The following material is available from http://www.blackwell publishing.com/products/journals/suppmat/JBI/JBI1120/ JBI1120sm.htm Appendix S1 Allele frequencies for Philippine mammals. Appendix S2 Genetic distance matrices. Appendix S3 Matrix of geographic distances between islands. REFERENCES Arbogast, B.S. & Kenagy, G.J. (2001) Comparative phylogeography as an integrative approach to historical biogeography. Journal of Biogeography, 28, 819–825. Avise, J.C. (2000) Phylogeography, the history and formation of species. Harvard University Press, Cambridge. Bermingham, E. & Moritz, C. (1998) Comparative phylogeography: concepts and applications. Molecular Ecology, 7, 367–369. Berry, R.J. (1986) Genetics of insular populations of mammals, with particular reference to differentiation and founder effects in British small mammals. Biological Journal of the Linnean Society, 28, 205–230. Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

Geology, ecology, and genetic differentiation in Philippine mammals Berry, R.J. (1998) Evolution of small mammals. Evolution on islands (ed. by P.R. Grant), pp. 35–50. Oxford University Press, Oxford. Brown, R.M. & Guttman, S.I. (2002) Phylogenetic systematics of the Rana signata complex of Philippine and Bornean stream frogs: reconsideration of Huxley’s modification of Wallace’s Line at the Oriental-Australian faunal zone interface. Biological Journal of the Linnean Society, 76, 393–461. Brumfield, R.T. & Capparella, A.P. (1996) Historical diversification of birds in northwestern South America: a molecular perspective on the role of vicariant events. Evolution, 50, 1607–1624. Caccone, A. (1985) Gene flow in cave arthropods: a qualitative and quantitative approach. Evolution, 39, 1223–1235. Case, T.J. & Cody, M.L. (1982) Synthesis: pattern and process in island biogeography. Island biogeography in the Sea of Cortez (ed. by T.J. Case and M.L. Cody), pp. 307–341. University of California Press, Berkeley. Case, T.J., Cody, M.L. & Ezcurra, E. (eds) (2002) A new island biogeography of the Sea of Cortez. Oxford University Press, Oxford. Cavalli-Sforza, L.L. & Edwards, A.W.F. (1967) Phylogenetic analysis: models and estimation procedures. Evolution, 21, 550–570. Cockerham, C.C. (1969) Variance of gene frequencies. Evolution, 23, 72–84. Cockerham, C.C. (1973) Analysis of gene frequencies. Genetics, 74, 679–700. Crother, B.I. (1990) Is ‘some better than none’ or do allele frequencies contain phylogenetically useful information? Cladistics, 6, 277–281. Dietz, E.J. (1983) Permutation tests for association between distance matrices. Systematic Zoology, 32, 21–26. Esselstyn, J.A., Widmann, P. & Heaney, L.R. (in press) The mammals of Palawan Island, Philippines. Proceedings of the Biological Society of Washington, in press. Fairbanks, R.G. (1989) A 17,000-year glacio-eustatic sea level record: influence of glacial melting rates in the Younger Dryas event and deep-ocean circulation. Science, 342, 637–642. Farris, J.S. (1972) Estimating phylogenetic trees from distance data. American Naturalist, 106, 645–668. Felsenstein, J. (1989) PHYLIP (phylogeny inference package), Version 3.2 manual. University of Washington, Seattle, WA. Fisher, R.A. (1954) Statistical methods for research workers, 12th edn. Oliver and Boyd, Edinburgh. Fitch, W.M. & Margoliash, E. (1967) Construction of phylogenetic trees. Science, 155, 279–284. Frankham, R. (1995) Inbreeding and extinction: a threshold effect. Conservation Biology, 9, 792–799. Frankham, R. (1996) Relationship of genetic variation to population size in wildlife. Conservation Biology, 10, 1500– 1508. Gaines, M.S., McClenaghan, L.R., Jr & Rose, R.K. (1978) Temporal patterns of allozymic variation in fluctuating populations of Microtus ochrogaster. Evolution, 32, 723–739.

Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

Grant, P.R. (1998) Evolution on islands. Oxford University Press, Oxford. Gurevitch, J., Morrow, L.L., Wallace, A. & Walsh, J.S. 1992. A meta-analysis of competition in field experiments. American Naturalist, 140, 539–572. Haldane, J.B.S. (1954) A test for randomness of mating. Journal of Genetics, 52, 631–635. Hall, R. (1998) The plate tectonics of Cenozoic SE Asia and the distribution of land and sea. Biogeography and geological evolution of SE Asia (ed. by R. Hall and J.D. Holloway). Backhuys Publishers, Leiden. Hall, R. (2002) Cenozoic geological and plate tectonic evolution of SE Asia and the SW Pacific: computer-based reconstructions, model and animations. Journal of Asian Earth Sciences, 20, 353–431. Hall, R. & Holloway, J.D. (eds) (1998) Biogeography and geological evolution of SE Asia. Backhuys Publishers, Leiden. Harmon, L.J., Schulte, J.A., II, Larson, A. & Losos, J.B. 2003. Tempo and mode of evolutionary radiation in Iguanian lizards. Science, 301, 961–964. Harris, H. & Hopkinson, D.A. (1978) Handbook of enzyme electrophoresis in human genetics. North Holland Publishing Co., Amsterdam. Heaney, L.R. (1986) Biogeography of mammals in SE Asia: estimates of rates of colonization, extinction, and speciation. Biological Journal of the Linnean Society, 28, 127–165. Heaney, L.R. (1991a) An analysis of patterns of distribution and species richness among Philippine fruit bats (Pteropodidae). Bulletin of the American Museum of Natural History, 206, 145–167. Heaney, L.R. (1991b) A synopsis of climatic and vegetational change in Southeast Asia. Climatic Change, 19, 53–61. Heaney, L.R. (1993) Biodiversity patterns and conservation of mammals in the Philippines. Asia Life Sciences, 2, 261–274. Heaney, L.R. (2000) Dynamic disequilibrium: a long-term, large-scale perspective on the equilibrium model of island biogeography. Global Ecology and Biogeography, 9, 59–74. Heaney, L.R. (2001) Small mammal diversity along elevational gradients in the Philippines: an assessment of patterns and hypotheses. Global Ecology and Biogeography, 10, 15–39. Heaney, L.R. (in press) Conservation in oceanic archipelagos. Frontiers of biogeography, recent advances in the geography of nature (ed. by M.V. Lomolino and L.R. Heaney). Sinauer Associates, Sunderland. Heaney, L.R., Heideman, P.D., Rickart, E.A., Utzurrum, R.B. & Klompen, J.S.H. (1989) Elevational zonation of mammals in the central Philippines. Journal of Tropical Ecology, 5, 259– 280. Heaney, L.R. & Regalado, J.C. (1998) Vanishing treasures of the Philippine rain forest. The Field Museum, Chicago, IL. Heaney, L.R. & Rickart, E.A. (1990) Correlations of clades and clines: geographic, elevational, and phylogenetic distribution patterns among Philippine mammals. Vertebrates in the tropics (ed. by G. Peters and R. Hutterer), pp. 321–332. Museum Alexander Koenig, Bonn.

245

L. R. Heaney et al. Heaney, L.R., Balete, D.S., Dolar, L., Alcala, A.C., Dans, A., Gonzales, P.C., Ingle, N., Lepiten, M.V., Oliver, W., Rickart, E.A., Tabaranza, B.R., Jr & Utzurrum, R.C.B. (1998) A synopsis of the mammalian fauna of the Philippine Islands. Fieldiana Zoology New Series, 88, 1–61. Heaney, L.R., Balete, D.S., Rickart, E.A., Utzurrum, R.C.B. & Gonzales, P.C. (1999) Mammalian diversity on Mt. Isarog, a threatened center of endemism on southern Luzon Island, Philippines. Fieldiana New Series, 95, 1–62. Heideman, P.D. & Heaney, L.R. (1989) Population biology and estimates of abundance of fruit bats (Pteropodidae) in Philippine submontane rainforest. Journal of Zoology (London), 218, 565–586. Hewitt, G.M. (2001) Speciation, hybrid zones and phylogeography – or seeing genes in space and time. Molecular Ecology, 10, 907–913. Hisheh, S., Westerman, M. & Schmitt, L.H. (1998) Biogeography of the Indonesian archipelago: mitochondrial DNA variation in the fruit bat, Eonycteris spelaea. Biological Journal of the Linnean Society, 65, 329–345. Holloway, J.D. (2003) Biological images of geological history: through a glass darkly or brightly face to face? Journal of Biogeography, 30, 165–180. International Union of Biochemistry and Molecular Biology (1992) Enzyme nomenclature. Academic Press, San Diego, CA. Juste, J., Ibanez, C. & Machordom, A. 2000. Morphological and allozyme variation of Eidolon helvum (Mammalia: Megachiroptera) in the islands of the Gulf of Ghana. Biological Journal of the Linnean Society, 71, 359–378. Kimura, M. (1983) The neutral theory of molecular evolution. Cambridge University Press, Cambridge. Lack, D. (1976) Island biology illustrated by the land birds of Jamaica. University of California Press, Berkeley. Lande, R. (1995) Mutation and conservation. Conservation Biology, 9, 782–791. Lande, R. & Barrowclough, G.F. (1987) Effective population size, genetic variation, and their use in population management. Viable populations for conservation (ed. by Soule´, M.), pp. 87–123. Cambridge University Press, Cambridge. Lepiten, M.V. (1997) The mammals of Siquijor Island, central Philippines. Sylvatrop, 5, 1–17. Levene, H. (1949) On a matching problem arising in genetics. Annals of Mathematics and Statistics, 20, 91–94. Lewontin, R. (1991) Twenty-five years ago in genetics: electrophoresis in the development of evolutionary genetics: milestone or millstone? Genetics, 128, 657–662. Livshits, G., Sokal, R.R. & Kobyliansky, E. (1991) Genetic affinities of Jewish populations. American Journal of Human Genetics, 49, 131–146. Lomolino, M.V. (2000) A call for a new paradigm of island biogeography. Global Ecology and Biogeography, 9, 1–6. Losos, J.B. & Schluter, D. (2000) Analysis of an evolutionary species-area relationship. Nature, 408, 847–850. MacArthur, R.H. & Wilson, E.O. (1967) The theory of island biogeography. Princeton University Press, Princeton.

246

Maharadatunkamsi, Hisheh, S., Kitchener, D.J. & Schmitt, L.H. (2003) Relationships between morphology, genetics and geography in the cave fruit bat Eonycteris spelaea (Dobson, 1871) from Indonesia. Biological Journal of the Linnean Society, 79, 511–522. Mantel, N. (1967) The detection of disease clustering and a generalized regression approach. Cancer Research, 27, 209– 220. Mayr, E. (1963) Animal species and evolution. Belknap Press, Cambridge, MA. Mey, W. (2003) Insular radiation of the genus Hydropsyche (Insecta, Trichoptera: Hydropsychidae) Pictet, 1834 in the Philippines and its implications for the biogeography of Southeast Asia. Journal of Biogeography, 30, 227–236. Mitchell, A.H.G., Hernandez, F. & De La Cruz, A.P. (1986) Cenozoic evolution of the Philippine archipelago. Journal of Southeast Asian Earth Sciences, 1, 3–22. Mittermeier, R.A., Myers, N. & Mittermeier, C.G. (1999) Hotspots, earth’s biologically richest and most endangered terrestrial ecoregions. CEMEX, Mexico City. Myers, N. (1988) Environmental degradation and some economic consequences in the Philippines. Environmental Conservation, 15, 205–213. Nei, M. (1973) Analysis of gene diversity in subdivided populations. Proceedings of the National Academy of Science USA, 70, 3321–3323. Nei, M. (1977) F-statistics and analysis of gene diversity in subdivided populations. Annals of Human Genetics, 41, 225– 233. Nei, M. (1978) Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics, 89, 583–590. Packham, G. (1996) Cenozoic SE Asia: reconstructing its aggregation and reorganization. Tectonic evolution of Southeast Asia, Vol. 106 (ed. by R. Hall and D. Blundell), 123–152. Geological Society Special Publication, The Geological Society, London. Peterson, A.T. & Heaney, L.R. (1993) Genetic differentiation in Philippine bats of the genera Cynopterus and Haplonycteris. Biological Journal of the Linnean Society, 49, 203– 218. Powers, D.A., Lauerman, T., Crawford, D. & DiMichele, L. (1991) Genetic mechanisms for adapting to a changing environment. Annual Review of Genetics, 25, 629–659. Rickart, E.A., Heaney, L.R. & Utzurrum, R.C.B. (1991) Distribution and ecology of small mammals along an elevational transect in southeastern Luzon. Journal of Mammalogy, 72, 458–469. Rickart, E.A., Heaney, L.R., Heideman, P.D. & Utzurrum, R.C.B. (1993) The distribution and ecology of mammals on Leyte, Biliran, and Maripipi Islands, Philippines. Fieldiana: Zoology, New Series, 72, 1–62. Ricklefs, R.E. & Bermingham, E. (2001) Nonequilibrium diversity dynamics of the Lesser Antillean avifauna. Science, 294, 1522–1524.

Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

Geology, ecology, and genetic differentiation in Philippine mammals Riddle, B.R. & Hafner, D.J. (in press) The past and future roles of phylogeography in historical biogeography. Frontiers of biogeography (ed. by M.V. Lomolino and L.R. Heaney). Sinauer Associates, Sunderland. Rosen, D.E. (1975) A vicariance model of Caribbean biogeography. Systematic Zoology, 24, 431–464. Schluter, D. (2000) The ecology of adaptive radiation. Oxford University Press, Oxford. Schmitt, L.H., Kitchener, D.J. & How, R.A. (1995) A genetic perspective of mammalian variation and evolution in the Indonesian archipelago: biogeographic correlates in the fruit bat Cynopterus. Evolution, 49, 399–412. Shaw, C.R. & Prasad, R. (1970) Starch-gel electrophoresis of enzymes: a compilation of recipes. Biochemical Genetics, 4, 297–320. Siddall, M., Rohling, E.J., Almogi-Labin, A., Hemleben, C., Meischner, D., Schmelzer, I. & Smeed, D.A. (2003) Sea-level fluctuations during the last glacial cycle. Nature, 423, 853– 858. Slatkin, M. (1985) Rare alleles as indicators of gene flow. Evolution, 39, 53–65. Slatkin, M. (1993) Isolation by distance in equilibrium and nonequilibrium populations. Evolution, 47, 264–279. Slatkin, M. & Barton, N. (1989) A comparison of three indirect methods for estimating average level of gene flow. Evolution, 43, 1349–1368. Sneath, P.H.A. & Sokal, R.R. (1973) Numerical taxonomy. W. H. Freeman, San Francisco. Sokal, R.R. (1988) Genetic, geographic, and linguistic distances in Europe. Proceedings of the National Academy of Science USA, 85, 1722–1726. Sokal, R.R. & Rohlf, F.J. (1981) Biometry, 2nd edn. W. H. Freeman and Co., New York. Sokal, R.R. & Wartenberg, D.E. (1983) A test of spatial autocorrelation analysis using an isolation-by-distance model. Genetics, 105, 219–237. Soule´, M.E. (1976) Allozyme variation, its determinants in space and time. Molecular evolution (ed. by F.J. Ayala), pp. 60–77. Sinauer Associates, Sunderland. Soule´, M.E. (1987) Viable populations for conservation. Cambridge University Press, Cambridge.

Steppan, S.J., Zawadski, C. & Heaney, L.R. (2003) Molecular phylogeny of the endemic Philippine rodent Apomys (Muridae) and the dynamics of diversification in an oceanic archipelago. Biological Journal of the Linnean Society, 80, 699–715. Swofford, D.L. & Berlocher, S.H. (1987) Inferring evolutionary trees from gene frequency data under the principle of maximum parsimony. Systematic Zoology, 36, 293–325. Swofford, D.L. & Selander, R.B. (1981) BIOSYS-1: A FORTRAN program for the comprehensive analysis of electrophoretic data in population genetics and systematics. Journal of Heredity, 72, 281–283. Walsh, J.S., Jr (1998) Geographic variation in Philippine fruit bats (Mammalia: Pteropodidae) and systematics of the cynopterine section. Unpublished PhD Dissertation, University of Chicago, Chicago, IL, 257pp. Waples, R.S. (1987) A multispecies approach to the analysis of gene flow in marine shore fishes. Evolution, 41, 385–400. Weir, B.S. (1990) Genetic data analysis. Sinauer Associates, Inc. Publishers, Sunderland. Weir, B.S. & Cockerham, C.C. (1984) Estimating F-statistics for the analysis of population structure. Evolution, 38, 1358–1370. Whittaker, R.J. (1998) Island biogeography: ecology, evolution and conservation. Oxford University Press, Oxford, 285pp. Wildlife Conservation Society of the Philippines (1997) Philippine red data book. Bookmark, Manila, Philippines, 240pp. Wright, S. (1931) Evolution in Mendelian populations. Genetics, 16, 97–159. Wright, S. (1943) Isolation by distance. Genetics, 28, 114–138. Wright, S. (1951) The genetical structure of populations. Annals of Eugenics, 15, 323–354. Wright, S. (1965) The interpretation of population structure by F-statistics with special regard to systems of mating. Evolution, 19, 395–420. Wright, S. (1978) Evolution and genetics of populations, Vol. 4. Variability within and among natural populations. University of Chicago Press, Chicago, IL. Zink, R.M., Blackwell-Rago, R.C. & Ronquist, F. (2000) The shifting roles of dispersal and vicariance in biogeography. Proceedings of the Royal Society of London Series B – Biological Sciences, 267, 497–503.

BIOSKETCHES Lawrence Heaney is Curator and Head of the Division of Mammals at the Field Museum. His research is focused on the evolution, ecology and conservation of mammal diversity in island and island-like ecosystems, especially in Southeast Asia and the western USA. Joseph Walsh is a Lecturer in the Undergraduate Program in Biological Sciences at Northwestern University and a Research Associate in Zoology at the Field Museum. He has worked on the systematics and biogeography of Southeast Asian fruit bats and is currently involved in restoration of tallgrass prairie and oak savannas in northern Illinois. Townsend Peterson is Associate Professor in Ecology & Evolutionary Biology, and Curator of Ornithology in the Natural History Museum, University of Kansas. His research focuses on ecological and historical factors shaping species’ geographic distributions.

Editor: Philip Stott Journal of Biogeography 32, 229–247, ª 2005 Blackwell Publishing Ltd

247