Abstract
Patterns of endemism in the Neotropics have been explained by restriction of forest to ‘refugia’ in arid cold-stages of the Quaternary (Haffer J (1969) Speciation in Amazonian forest birds. Science 165: 131–137). The palaeoecological record, however, shows no such forest contraction. We review palaeoecological and phylogenetic data on the response of Neotropical taxa and communities to climatic changes of the Cenozoic. Solar insolation varies over this period with latitude and geography, including shifts in opposite directions between high and low latitudes. In the Neotropics, distribution and abundance patterns originate on a wide range of timescales through the Cenozoic, down to the currently dominant precession forcing (20 kyr). In contrast, distributions and abundances at higher latitudes are controlled by obliquity forcing (40 kyr). The patterns observed by Haffer (1969) are likely derived from pre-Quaternary radiations and are not inconsistent with palaeoecological findings of continuous forest cover in major areas of the Neotropics during the Quaternary. The relative proportions of speciation processes have changed through time between predominantly sympatric to predominantly allopatric depending on the prevailing characteristics of orbitally forced climatic changes. Behaviour of Neotropical organisms and ecosystems on long timescales may be influenced much more by precessional forcing than by the obliquity forcing that controls high-latitude climate change and glaciations.
Introduction
Hypotheses relating to Neotropical ‘refugia’ are well known and pervade the literature of low-latitude species diversity (Hill and Hill, 2001). The region is particularly diverse today, and appears to have had a long history of high diversity (Wilf et al., 2003). How that diversity relates to Cenozoic (last 65 Myr) climate history has never been established, although part of the explanation of the latitudinal gradient in diversity may relate to long-term cooling (Hawkins et al., 2006). Even Darwin appears to have had trouble with tropical diversity and climate change, writing about ‘the arctic forms’ and their ‘long southern migration and re-migration northward’ (Darwin, 1859: 368), but not discussing the consequences of this behaviour for low-latitude ecosytems. An original hypothesis, first put forward by Haffer (1969), to explain Neotropical high diversity was based on studies of present-day patterns of bird endemism in the Amazon basin. Noting that there appeared to be pockets of high levels of endemism in mid-altitude sites around the perimeter of the Amazon basin, Haffer (1969) proposed that these were due to refugial isolation and subsequent allopatric speciation during the cold-stages of the Quaternary (last 1.8 Myr) ice ages. He argued that during the aridity-limited environment of the cold-stages, habitats able to support forest bird populations were restricted to mid-altitude locations where moisture levels would have been higher because of orographic precipitation. In the lowland Amazon basin, in contrast, the arid climate would have resulted in the replacement of forests by a ‘sea of savanna’. The prolonged isolation of fauna and flora during Quaternary cold-stages in mid-altitude refugia and their separation from other populations by a ‘sea of savanna’ resulted in allopatric speciation and ultimately pockets of endemism.
This hypothesis of Haffer (1969) was originally used to explain pockets of bird endemism in the Amazon basin, but became a widely utilised model for low-latitude environments that has been used a number of times to explain centres of endemism for other groups within the Amazon basin including lizards, frogs, plants and butterflies (Figure 1). One of the underlying problems with the Haffer (1969) hypothesis, however, is circularity in his line of argument. He used evidence from the current distribution of endemic bird species in the Amazon basin to identify regions of refugia, and then proposed that isolation in these refugia was responsible for their high levels of endemism. Independent evidence to support the hypothesis that during cold-stages of the Quaternary the lowland Amazon basin turned into a ‘sea of savanna’ was lacking, and still is (Bush et al., 2002; Colinvaux et al., 1996, 2000, 2001; Mayle et al., 2004, 2009), so it has been impossible to verify this line of reasoning, and the hypothesis has been controversial (Knapp and Mallet, 2003).
Late-Quaternary ‘refugia’ in the Neotropics. Areas shaded grey and black are reconstructed forest ‘refugia’ based on distribution patterns of, respectively, birds and Anolis lizards. After Simpson and Haffer (1978).
Our understanding of how climates of the Cenozoic (last 65 Myr) have changed is now much deeper than when Haffer (1969) was writing, and we have commensurately more data (palaeoecological and phylogenetic) on the distribution of organisms and how they have responded to those changes. Here, we review the relevant components of the Cenozoic climatic record, develop from them hypotheses about how organisms might have responded, and point out where this climate history differs from the higher latitude history that formed the basis of the Haffer (1969) argument. Then we review evidence for how organisms actually did respond to these climate changes and, finally, return to Haffer (1969) and subsequent debate for an analysis of those ideas in the light of what appears to be happening on the ground.
Cenozoic climates
Current understanding of the net effect of all the components of the Cenozoic climate record has been summarized by Zachos et al. (2001), using ocean sediment oxygen isotope records (Figure 2). These data show that the Earth has been, overall, cooling throughout the Cenozoic, and that as global cooling has intensified in the late Cenozoic, the amplitude of superimposed oscillations, brought about by orbital variations, has increased. It would be reasonable to assume, from these results, that the cooling over the Earth has been about 15°C during the Cenozoic, although not evenly distributed by latitude or season.
Cooling of the Earth through the Cenozoic, based on oxygen isotopic data. Black bars to the left show the extent of glaciation in the Antarctic (right) and Northern Hemsiphere (left). After Zachos et al. (2001).
High-latitude insolation is dominated by obliquity (40 kyr) variations while low-latitude insolation is dominated by precession (20 kyr) variations (Figure 3). The annual radiation received at the equator is greater than at higher latitudes, and is also more variable (Willis et al., 2009: figure 1). These variations are independent of each other, and are not always in phase: notable periods occur when low-latitude insolation is increasing while high-latitude insolation is decreasing (e.g. 95 ka) and vice versa (e.g. 150 ka). Similar patterns persist back in time (Berger, 1978), and must be permanent features of Earth history. The impact of insolation changes is a function of geography (Short et al., 1991), as well as latitude, and it follows that we cannot assume that climatic changes and trends will always be the same in amplitude and timing at low latitudes as at high latitudes. Consequently we cannot assume that environmental (for example East African lake levels: Trauth et al., 2010) or organism responses have always been the same globally in amount or direction.
Deviations of solar radiation from their
For the period of the last (high latitiude) glacial–interglacial oscillation, modelling of the Earth’s climates by Braconnot et al. (2007) show the kinds of differences that might be expected between snapshots at the last glacial maximum (LGM: 22–19 ka) and today. Globally, the LGM was characterized by much cooler temperatures at high northern latitudes (more than 20°C cooler than today in areas at and near the Laurentide and Scandinavian ice sheets). Temperatures were also cooler in tropical latitudes, but to a lesser extent. Precipitation values are more difficult to estimate, and model results vary, but much of the Earth was probably drier than today. There may have been more rainfall in the tropics, especially over the oceans, but most terrestrial tropical regions would have been drier, in all models of Braconnot et al. (2007). In the Neotropics, the pattern of LGM precipitation was spatially complex, with sediment and speleothem records indicating precipitation significantly higher than present on the Bolivian/Peruvian Altiplano (Baker et al., 2001) and near the Atlantic seaboard of Brazil (Cruz et al., 2009; Wang et al., 2007), while pollen data from tropical forest-savanna ecotones in the continental interior show precipitation lower than present during the LGM (Mayle et al., 2000; Whitney et al., 2011).
The LGM may not be a typical period in Earth history, but the concidence of negative trends in precession and obliquity may mean that it was an especially significant period, determining species distributions and abundances under relatively extreme conditions.
Potential organism response
For much of the Cenozoic, Neotropical climates were likely warmer and, probably, wetter than today and the amplitude of climatic fluctuations driven by Milankovitch oscillations were lower than today (Zachos et al., 2001). The overall amplitude of global climate oscillations appears to have been shorter (40 kyr), before the dominant periodicity lengthened to 100 kyr (Crowley and Hyde, 2008; Shackleton, 2000) as climates cooled and continental ice-sheets developed at about 700 ka. The pre-Quaternary Cenozoic would, therefore, have had less in the way of multimillenial-scale climatic variability to force distribution and abundance changes, and it is therefore likely that there was much more stability of communities over longer periods of time (millions of years) during the Paleogene (65–23 Ma) and Neogene (23–2 Ma). At low latitudes it is probable that climate was dominated by strong precessional forcing (20 kyr) of insolation while obliquity-forcing (40 kyr) was more significant at higher latitudes (and possibly even dominant globally). Stability of ecosytems implies that, relative to modern environments, the proportion of sympatric to allopatric speciation is likely to have been greater.
As Neotropical climates cooled, especially within the last 5 Ma, and the amplitude of oscillations increased (probably at all timescales), movement of species distributions and abundance changes should have developed and become more frequent. With the 20-kyr precession-forcing dominant at low latitudes, the pace of climatic change, and hence shifting of populations, should have been as fast as it has ever been. Over this period, the proportion of allopatric speciation relative to sympatric speciation should have increased, because of the movement of populations and formation of new communities. Increased amplitude of environmental change would also have brought about more change to the physical environment, such as sea-level changes, shifts of forest distribution with altitude and perhaps latitudinally in response to changing rainfall patterns. Changes in concentrations of atmospheric CO2 (Bennett and Willis, 2000; Huang et al., 2001; Street-Perrott et al., 1997) and ultraviolet (UV-B) radiation (Flenley, 2007; Willis et al., 2009) would have affected distributions and abundances of plants, to at least as great an extent as changes in temperature and precipitation. Modelling suggests that the CO2 changes would have been particularly important in the tropics (Woillez et al., 2011). These changes in the physical environment should also have brought about increased possibilities for isolation of populations, and hence also allopatric speciation.
The last 700 kyr may have brought about a slight lessening of the pressure on populations in the tropics, because the extreme development of northern ice sheets appears to have had a modulating effect on global temperatures, increasing the importance of a 100-kyr oscillation (Crowley and Hyde, 2008; Shackleton, 2000), seen globally through aspects such as sea-level change and atmospheric composition (Jansen et al., 2007). This tendency to a lengthening of the periodicity of global climatic change should have given communities more stability over time and reduced the frequency with which community turnover and isolating events took place. Hence, the proportion of speciations that were sympatric may have increased slightly, but not to the levels seen during much of the Paleogene and Neogene, when climatic oscillations were less variable (Zachos et al., 2001).
It follows that speciations that took place during most of the Cenozic may have happened by different mechanisms than those during the later part (Quaternary), and that observations on behaviour, including speciations, within the Quaternary may not be characteristic for the earlier part of the Cenozoic (Paleogene and Neogene).
Periods when insolation changes at high and low latitudes were out of phase may have been especially dynamic, because opposite mode forcing would have tended to push species either towards each other, or pulled them apart (depending on the relative direction of insolation changes). At low latitudes, the dominant periodicity of insolation change is the 20 kyr precession oscillation (Figure 3). Other factors being equal, if this translates into temperature changes, the effect would be a cooling, with greatest impact at higher altitudes and serving to force higher altitude taxa to lower altitudes, mixing into lower-altitude ecosytems. At high latitudes, the dominant periodicity is the 40 kyr obliquity oscillation, and the formation of sub-continental ice-sheets means that there the resultant cooling cannot be encompassed within these regions because the available altitudinal ranges are too small relative to the magnitude of cooling (Bennett et al., 1991). These changes then serve to force higher-latitude taxa to lower latitudes, mixing into low-latitude ecosystems.
Low-latitude, low-altitude ecosystems can thus be thought of as ‘arks’: storehouses towards which taxa are pushed during negative insolation trends, and from which they can leave during positive insolation trends. As the trends for precession and obliquity are not always in phase (seen as the curves for 10°N and 80°N, respectively, in Figure 3), periods when taxa from high latitudes or higher altitudes move towards ‘arks’ vary, and similarly periods when they can move out. The LGM was unusual in that both trends were in the same direction: movement downward to lower altitudes (forced by precession) and movement to lower latitudes (forced by obliquity) could occur simultaneously.
Observed Neotropical organism response
Fossil evidence
A detailed record of floral changes during the early and mid Cenozoic has been obtained from the Andean Neotropics by Jaramillo et al. (2006). Their data indicate a diversity maximum during the Eocene (exceeding Holocene diversity), and then declining diversity towards the early Miocene, following trends in global temperature (Figure 2; Zachos et al., 2001). The temporal resolution of the record is not sufficiently detailed to show diversity changes at the scale of the Earth’s orbital variations. Long pollen sequences from the Colombian Andes (Hooghiemstra, 1984), and non-Neotropical mid-latitude sequences with fine temporal resolution around 3 Ma (Sniderman et al., 2007; Willis et al., 1999) show changes in vegetation at similar frequency to orbital changes.
A number of key pollen sequences have been obtained from Quaternary sedimentary sequences in the Neotropics (for reviews, see Mayle et al., 2004, for the Neotropics and Morley, 2007, for tropical forests globally) indicating a far more complex response by flora and vegetation to the cold-stages of the Quaternary than previously thought. In the Amazon basin, for example, there are now a number of sedimentary sequences containing fossil pollen that detail vegetation changes over the cold stages of the Quaternary. These include a record spanning the last 170 kyr from lake Pata in northwest Amazonia (Bush et al., 2002; Colinvaux et al., 1996) and a 50-kyr record from the Amazon fan sediments (Haberle and Maslin, 1999). The latter sequence is of particular interest because it represents a basin-wide record of vegetation change in the Amazon basin. In no sequence from the Amazon basin is there evidence for the establishment of a ‘sea of savanna’ in the lowland basin during the cold stages (precessional insolation minima) but rather that it remained forested throughout. The composition of the forest was different with many more higher altitude taxa than at present, suggesting that the total opposite to the Haffer model occurred; trees migrated downwards from higher-altitude sites towards the lowland basin during cold-stages (Colinvaux et al., 2000; Ledru et al., 2007) rather than the other way around. Overall depression of treelines at highest altitudes was of the order of 1000 m (e.g. Flenley, 1979; Hooghiemstra and van der Hammen, 2004), although movements of taxa within forests may have been much less than that. By comparison, the global fall in sea level at LGM was about 120 m (Jansen et al., 2007), much less than the treeline depression, indicating a general compression of the altitudinal range of forests.
In order to understand why forest taxa moved down into the basin rather than upwards, it is necessary to examine evidence from palaeoclimate proxies. Reconstruction of Amazonian climate during the LGM using fossil proxies such as stable isotopes (Huang et al., 2001), gas concentrations in fossil groundwater (Stute et al., 1995) and tropical corals (Guilderson et al., 1994) support the suggestion that LGM temperatures in the Amazon basin would have been up to 5°C cooler than present. Fossil proxies also indicate that there was a significant reduction in atmospheric CO2 concentrations to as low as 180–200 ppm (Cowling et al., 2001), and consequently lower concentrations with increasing altitude, leading to greater CO2 stress (Bennett and Willis, 2000). However, the fossil evidence does not support the suggestion that the Amazon basin became much drier.
Palaeolimnological evidence obtained from lake Pata in northwest Amazonia (Bush et al., 2002), reveals that lake levels there were relatively high before the LGM and that humid conditions existed with, some estimates suggest, precipitation that was even higher than present day (Baker et al., 2001; Mayle et al., 2004). Thus the climatic factors most limiting to vegetation during the cold-stages in Amazonia would have been reduced precipitation, possibly a consequence of precessional forcing (Figure 3) and a CO2-depleted atmosphere. Since the impact of both on the vegetation would have increased with altitude (Bennett and Willis, 2000; Huang et al., 2001), it is highly probable that these were the principal mechanisms responsible for restricting many higher-altitude Andean forest taxa to the lowland Amazon basin during the LGM. Together with a greater extension of dry forest taxa through the Amazon basin (Pennington et al., 2000), this resulted in a forest type in the Amazon basin during the LGM that has no present-day analogue (Colinvaux et al., 2000).
A non-analogue forest in the Amazon lowland basin during the cold-stages of the Quaternary is particularly interesting because it would have had exactly the opposite effect to that predicted by Haffer (1969). Rather than the cold-stages of the ice-ages resulting in small pockets of isolated forest creating conditions for allopatric speciation, the greater mixing of forest communities in the lowland Amazon basin would rather have resulted in increased opportunities for genetic mixing and interbreeding. At the LGM, much colder conditions at higher latitudes (forced by the obliquity minimum) coincided with complex and variable temperature and precipitation conditions at higher altitudes of low latitude (forced by the coincident precession minimum north of the equator, or maximum south of the equator), greater maximum opportunity for such mixing. As typical species longevities are much longer than the period of time since the LGM (Stanley, 1985), it is possible to infer that LGM diversities at tropical low altitudes must have been somewhat higher than before the LGM because of the simultaneous mean movement of taxa towards low latitudes and low altitudes.
So did the cold-stages of the Quaternary have no real long-term impact on the distribution of forested vegetation in the Neotropics? Although the majority of studies for the Neotropics have been carried out in the rainforest, work in Neotropical seasonally dry forests and ecotonal boundaries of the forest/savanna and upper cloud forest/open grassland suggest that this is not necessarily a representative picture for the whole of the Neotropics. Recent work (e.g. Behling and Hooghiemstra, 2000; Mayle, 2004; Mourguiart and Ledru, 2003; Whitney et al., 2011) indicates that in these ‘periphery’ forests there was far greater dynamism of vegetation formations in response to the cold-stages of the Quaternary than in the core of the Amazon basin. Much of the area now occupied by forest edge in Bolivia was occupied by savanna during the LGM so the forest/savanna boundary was north of its present location in the LGM by as much as 30 km, resulting in greater expanses of savanna relative to today (Mayle et al., 2004). This back and forth movement is likely to have been a frequent occurrence through Quaternary climatic oscillations. The effect that this had on population distributions, even if somewhat different from the mid-altitude refugial localities suggested by Haffer (1969), is only now starting to emerge through genetic evidence (Pennington et al., 2004).
Genetic evidence
Genetic studies are providing exponentially increasing amounts of data on the timings of lineage splits in relation to environmental change. Rull (2008) compiled data from more than 300 Neotropical studies up to 2006, and found that species origins of a wide diversity of groups took place continually from the early Cenozoic through to the Quaternary, rather than being concentrated during the Quaternary or any other particular time. Individual studies, therefore, may indicate a wide range of possible timings, with a diversity of conclusions as to their timings in relation to environmental changes. Thus, studies based on rattlesnakes (Wüster et al., 2005), parrots (Eberhard and Bermingham, 2004) and a number of plant taxa (Pennington et al., 2004) apparently support a Quaternary isolation model. All three studies are based on taxa that presently occupy dry-formation vegetation including savannah woodland and seasonally dry tropical forest. The Neotropical rattlesnake, Crotalus durissus, for example, presently occupies primarily seasonally dry landscapes from Mexico to Argentina, but avoids the rainforests of Central and tropical South America. Studies on the mitochondrial DNA gene sequence of this species indicate that there are distinct genetic differences between populations from north and south of the Amazon rainforest and that this split occurred at approximately 1.1 Ma. This result therefore suggests that, at some time in the early Quaternary, Amazonian rainforests may have become fragmented or shrunk considerably to allow intermixing of these now disjunct and genetically distinctive populations.
A similar conclusion was reached by Eberhard and Bermingham (2004) in a study of a species complex of Neotropical parrots Amazona ochrocephala. This complex includes 11 named subspecies that are distributed from Mexico to the Amazon basin and are usually found associated with dry or deciduous woodland, gallery forest and savannah woodland. A number of different populations were sampled across their present-day distribution and their genetic differentiation was determined using mitochondrial DNA. Results suggest that taxa in this complex are very closely related with short genetic distances among the different subspecies indicating a relatively recent and rapid colonisation, probably from South America. The habitat preference of these birds is for lowland open-vegetation formations and they avoid continuous moist forest, so it is proposed that their colonisation into Central America occurred during stages of the Quaternary when open wooded landscapes were more prevalent (Eberhard and Bermingham, 2004).
Pennington et al. (2004) focused on a number of taxonomically unrelated plant groups presently found in the seasonally dry tropical forests. These are currently located in a number of isolated areas within the Neotropics and have a high level of endemism. They are disjunct now, and were probably even more disjunct at the LGM (Werneck et al., 2011). Molecular analyses of these taxa indicate that whilst the South American populations probably diversified during the late Cenozoic with lineage splits occurring at 20–11 Ma, a much more recent diversification pattern is apparent in the Central American populations. Here, most species appear to have originated during the Quaternary with splitting ages of 1.2–1 Ma or less. The conclusion reached by this study was that Quaternary climate fluctuations resulted in the intermixing and then isolation of these seasonally dry vegetation formations in Central America, leading to allopatric speciation. Such results again do not support a continuous moist lowland rainforest through all of the cold-stages of the Quaternary.
In contrast to these examples, however, are a wealth of studies from diverse groups that do not indicate evidence for Quaternary refugial isolation and genetic diversification. The primary splits of Heliconius butterflies, Hyla frogs (Chek et al., 2001), Pseudomyrmex ants (Ward, 1999) and Trinomys rodents (Lara and Patton, 2000) were pre-Quaternary. Lessa et al. (2003) found that, in contrast to higher-latitude taxa, Neotropical mammals show little or no demographic expansion during the Quaternary. In none of these studies was convincing evidence provided to support a Quaternary refugial isolation and diversification model. To date most work has been carried out on dating of lineage splits (for a review see Moritz et al., 2000) and for many taxa previously assumed to be of Quaternary age, it is apparent that their lineage splits occurred much earlier, supporting the earlier conclusion of Stanley (1985).
Phylogenetic data of Atta ants have been used to carry out tests of the several competing hypotheses for the origins of Amazonian diversity (Solomon et al., 2008). It was found to be difficult to distinguish between Miocene marine incursion events and possible late-Quaternary habitat fragmentation. Late-Quaternary fragmentation was indicated most strongly, however, for species of open ground, which is in the opposite sense to that predicted by Haffer (1969).
Increasingly it is becoming clearer that the fossil and genetic evidence for a ‘sea of savanna’ fragmenting the lowland rainforests during Quaternary cold-stages does not exist. If there was fragmentation, then this appears to have had little impact on the genetic diversity of lowland rainforest populations. However, at the margins of the rainforest, in regions currently occupied by dry forest and savanna, there does appear to have been much more dynamism during the Quaternary. Evidence suggests that, in these environments, isolation of populations may well have resulted in genetic diversification. A similar conclusion was reached by Hooghiemstra and van der Hammen (1998) who suggested that in different parts of the Amazon basin the biogeographical importance of Quaternary refugia might have been markedly different.
Discussion
Haffer’s (1969) model built on recognition that species diversity was not evenly distributed across the Neotropics. Subsequent work has largely borne this out, with the identification of several biodiversity hotspots along Andean slopes and southeastern Brazil (Myers et al., 2000) in a pattern that bears more than a passing resemblance to that of Haffer (1969). However, no evidence has ever been presented that shows that the diversifications that created these low-latitude hotspots happened during the Quaternary. On the contrary, most evidence suggests that these diversifications were early- or mid-Cenozoic phenomena (Morley, 2007), with increasing phylogenetic data indicating patterns of diversification that are more complex than would be expected under a late-Quaternary refugia model (e.g. McKenna and Farrell, 2006; Solomon et al., 2008; Whinnett et al., 2005; but see Bennett and Provan, 2008). Consequently, explanations for their existence must be sought in the Paleogene and Neogene, not in the Quaternary (Morley, 2007). Discussion of the nature of LGM vegetation in Amazonia may therefore be irrelevant to that particular aspect. On the other hand, current understanding of Cenozic climates suggest warmer and more stable climates during most of the Cenozoic. The radiations that created the biodiversity hotspots may therefore be the consequence of lineage splitting under these conditions and thus very likely sympatric diversifications.
This understanding of patterns of diversifications in the Neotropics differs from that found at higher latitudes. The palaeoecological and genetic record of North America and Europe is particularly well developed (e.g. Davis and Shaw, 2001; McLachlan et al., 2005; Magri et al., 2006, 2007) and there the existence of major ice sheets during the LGM, and presumably during previous, equivalent, cold-stages, meant that large areas of both continents have flora and fauna that are entirely immigrant since the end of the last glacial period. The species involved have older histories, with genetic data showing divergence of lineages during the pre-Quaternary Cenozoic (e.g. Denk et al., 2005; Lang et al., 2007), but the time and place of lineage splitting cannot now be identified and, in any case, must have taken place under physical and biotic conditions different from the Holocene.
What happened to this higher latitude flora and fauna during high-latitude cold-stages (obliquity insolation minima)? Most data indicate that most tree taxa were located in the southern parts of their respective continents (e.g. the Mediterranean peninsulae, in the case of Europe: Bennett et al., 1991; Huntley and Birks, 1983), and that there have been repeated re-occupations of the northern parts of the continents. Some taxa, more than was appreciated earlier, may have persisted in central Europe at low densities (Bhagwat and Willis, 2008; Stewart and Lister, 2001; Willis and van Andel, 2004), but the overall picture remains one of back and forth movement of entire distributions and the total formation and break-up of communities in the process.
The Neotropical low-latitude situation is different. Species were always able to persist regionally, although there may have been movement within regions. Broadly, forest remained forest. Large areas of recognisable tropical rain forest persisted in Amazonia throughout the Quaternary (Colinvaux et al., 2000, 2001; Morley, 2007). In detail there were changes, but these changes overlie other patterns (such as Paleogene and Neogene diversifications), rather than replacing them. Responses of organisms to Paleogene and Neogene climates include continuing distribution changes following diversification. Dick et al. (2003, 2007) have shown that tree species have dispersed across the Atlantic from Africa to the Americas and then spread within the Neotropics, leaving phlyogeographic patterns that are still evident. Wüster et al. (2005) showed from molecular phylogenetic data that rattlesnakes have spread into northern South America gradually over the last 2 Myr, after the connection of the Panama isthmus. A similar pattern has been observed for a parulid warbler (Vilaça and Santos, 2010), and can be expected for other taxa. Modelling indicates that organismal responses to Quaternary climate fluctuations may have been complex, and a function of historical contingencies (Colwell and Rangel, 2010).
The history of modern species and communities in northern North America (and northern Europe) is largely a record of events of the last 10–14 kyr (the most recent negative obliquity insolation trend), to which events of earlier periods have contributed little more than a reduced species pool, and no spatial patterns. In contrast, we should expect that any complete history of the species and community dynamics of taxa found at any place in the Neotropics will be a complex mixture of processes from different timescales (Table 1), all of which have persisted in some form through to the present with recognisable spatial patterns that derive from these different processes.
Patterns and processes controlling species distributions in the Neotropics.
Conclusions
Several conclusions emerge from this examination of the biogeographical history of Neotropical communities over Cenozoic time:
The pattern of solar insolation changes during the Cenozoic was complex. The dominant periodicity varies with latitude, and the timing of insolation maxima and minima also vary. As a result, organism response varied in nature, rate and direction over time. Periods of time when high latitudes experience insolation maxima while low latitudes have minima are especially interesting (Figure 3). It follows that a simple model of ‘cold-stage’ and ‘warm-stage’ is not globally applicable: a model of 20 kyr precessional forcing at low latitudes (in opposite sense either side of the equator) and 40 kyr obliquity forcing at high latitudes, out of phase, corresponds better to reality. The LGM is a rare period when both trends operated in the same direction simultaneously (Figure 3).
In the light of recent fossil and molecular evidence, the behaviour of species and communities through the Cenozoic is far more complex than originally envisaged. There is little evidence to support the original Haffer (1969) model for a ‘sea of savanna’ replacing the lowland Neotropical rainforest. Rather, the majority of the rainforest remained but the composition of this forest altered to include both montane and dry forest species. However, significant contraction of the rainforest probably did occur at its boundaries to be replaced by savanna and/or seasonally dry forest. These margins may well have provided refugia for species of dry, open environments.
Haffer’s (1969) recognition of areas of higher biodiversity may relate to sympatric speciations during the early and mid Cenozoic, when climates were warmer, probably also wetter, and more stable over longer periods of time than today. His hypothesis that Amazonia must have been largely deforested during the Quaternary does not necessarily follow from the patterns he identified.
Challenges to Haffer’s (1969) hypothesis on the basis that Amazonia was forested during the Quaternary (e.g. Colinvaux et al., 2000, 2001) may have been mis-aimed. Haffer’s patterns do not necessarily require that Amazonia be deforested during the Quaternary (although that was his interpretation), as they might have arisen much further back in time as a combination of the complexities of radiations and distribution and abundance changes.
Despite the apparent genetic differences, there is little evidence to indicate that length of time in isolation is sufficient for allopatric speciation to occur. Molecular evidence indicates that lineage splits for many (possibly most) modern sister taxa occur well before the Quaternary (see above) and there is little fossil or genetic evidence to indicate the emergence of new species in pace with Quaternary ice ages. In fact, from the fossil record, the Quaternary ice ages appear to have been a time of redistribution of populations resulting in local extinction rather than speciation (Bennett, 1990, 2004; Willis and Niklas, 2004). However, length of time under near-stable conditions might be responsible for an increased proportion of sympatric speciation during much of the Cenozoic.
Pre-Quaternary patterns of diversification and spread are still evident, in situ, in the biota of the Neotropics, in contrast to the biota of higher latitudes. Biota of the tropics are thus complex mixtures of taxa that radiated and spread over a wide range of timescales, and probably at different rates.
The relative proportions of sympatric and allopatric speciation may have shifted during the Cenozoic, with greater frequences of sympatric speciation during the Paleogene and Neogene than the Quaternary, while climates were more stable.
Footnotes
Acknowledgements
We thank Francis Mayle and an anonymous referee for helpful comments on an earlier draft. KDB also thanks audiences at talks at ATBC (Suriname, 2008), Universities of Göttingen and Massey (2009) and VI Southern Connection Congress (Bariloche, 2010) for their contributions to discussion of neotropical refugia.
Funding
This work was carried out with a Leverhulme Trust grant (F08 773/E) that supported SAB, and a Royal Society Wolfson Research Merit Award to KDB, which are gratefully acknowledged.