The latitudinal diversity gradient (LDG) is one of the most well recognized and best documented of the global patterns in biotic systems. Diversity patterns relating to latitude apply to biomes on land and in water and they apply to life forms that range from bacteria and viruses to trees and vertebrates, including both ectotherms and endotherms (Willig et al., 2003; Guernier et al., 2004; Hillebrand, 2004; Pommier et al., 2007). Furthermore, the LDG appears to have existed more than 300 million years ago (Stehli et al., 1969; Powell, 2007). The long-term generality of this pattern across taxa suggests a unitary set of primary controlling factors that have been extant throughout the history of life. Other well-recognized regional to global scale biodiversity patterns include the positive relationship with bioregion area and island area (MacArthur & Wilson, 1967), water availability (e.g. Gentry, 1988) and productivity (Gillman & Wright, 2006; Cusens et al., 2012). The influence on diversity due to the interplay between water availability and temperature is also well recognized (O'Brien, 1998; Francis & Currie, 2003; Hawkins et al., 2003).
There is evidence that tropical climates have existed for longer and were more extensive than temperate climates for much of Earth's history (Mittelbach et al., 2007), suggesting that area (Rosenzweig & Abramsky, 1993) and evolutionary time (Rohde, 1992) influence current diversity. By contrast, the long standing energy–richness (or more individuals) hypothesis, as described by Hutchinson (1959), Brown (1981) and Wright (1983), explains the LDG as a function of productivity. This hypothesis suggests that greater productivity can support more individuals and therefore a greater number of minimum-viable-populations of species. However, key predictions of this theory appear to be contradicted by empirical evidence (e.g. Currie et al., 2004) and productivity appears to have an effect on richness beyond that expected on the basis of the number of individuals (Hurlbert & Jetz, 2010). Recently, a model that integrated area and productivity over geological time together with temperature succeeded in explaining 77% of current-day species richness of the four major terrestrial vertebrate groups (Jetz & Fine, 2012).
Hypotheses based on differential rates of diversification are also gaining increasing support (Mittelbach et al., 2007). The idea that global species richness gradients derive from origination and diversification asymmetries that favour tropical climates has been with us in one form or another for more than 130 years (Wallace, 1878; Fisher, 1930; Rensch, 1959; Stehli et al., 1969; Rohde, 1992; Jablonski, 1993; Briggs, 2004). One of several potential explanations for such an asymmetry is the evolutionary speed hypothesis (ESH; Rensch, 1959), elaborated as the effective evolutionary time hypothesis by Rohde (1992).
This implicit assumption of the ESH is equivalent to the main tenet of the tropical niche conservatism hypothesis (TNCH; Wiens & Donoghue, 2004). Tests that demonstrate climatic limits to clade expansion are therefore also consistent with the ESH. The TNCH differs from the ESH, however, because it assumes that the rate of diversification is independent of latitude (Wiens et al., 2006). In summarizing the TNCH, Wiens & Donoghue (2004) suggest that greater clade origination occurred in the tropics prior to 30–35 Ma due to larger tropical extent (i.e. the species–area hypothesis). Although niche conservatism, along with other limitations to dispersal, might explain how the LDG is maintained, in isolation it does not explain how the LDG originated. A separate mechanism such as that relating to area, or stability, must be invoked to explain greater origination of clades in the tropics.
The ESH involves three mechanistic steps between environmental temperature, rate of genetic evolution, rate of diversification and species richness. Two additional indirect relationships are therefore also predicted; that is, between temperature and richness and between temperature and speciation (Fig. 1). Here we begin by reviewing the empirical evidence for and against the relationships predicted by the ESH hypothesis and then we discuss developments that enable us to propose a link between environmental temperature, water availability and area via a common causative mechanism under an integrated theory of evolutionary speed. Thus, we present a model of evolutionary speed that might influence speciation rates and general patterns of diversity. However, in presenting this model we recognize that overlaying such a mechanism there will be other processes that have important, and in some cases dominant, influences on speciation and the maintenance of diversity. Nonetheless, this model provides a synthetic context and focus for future empirical testing.
Prediction 1: Rates of genetic evolution depend on ambient temperature
Greater rates of genetic evolution involving several different genes have been found at lower (warmer) latitudes for a broad range of taxa, including rain forest plants (Wright et al., 2006; Gillman et al., 2010), angiosperm families (Davies et al., 2004), marine foraminifera (Allen et al., 2006), marine fishes (Wright et al., 2011), amphibians (Wright et al., 2010), turtles (Lourenço et al., 2013), mammals (Gillman et al., 2009) and birds (Gillman et al., 2012) (Table 1). Most of these studies have employed large data sets that replicate comparisons between closely related pairs of species from a diverse array of families and involve a range of different genes from both nuclear and mitochondrial genomes. By contrast, Bromham & Cardillo (2003) did not find a significant latitudinal effect, although one component of their study was only marginally non-significant.
Table 1. Empirical studies that have tested for relationships between environmental variables and rates of genetic evolution
|Taxa||Extent||Gene(s)||Variables ||P-values|| n ||Reference|
| ||Global||cyt b||Elevation ||0.006||29||Gillman et al. (2009)|
|Birdsa||Global||cyt b||Latitude ||0.36||33||Bromham & Cardillo (2003)|
| || || || |
| ||Global||cyt b||Elevation/latitude ||0.004||30||Gillman et al. (2012)|
|Hummingbirds||Regional||DNA hybridization||Elevation||(0.0001)b||(24)b||Bleiweiss (1998)|
18 families (caudates and anurans)
|Global||12S, 16S||Latitude ||0.01–0.02||78||Wright et al. (2010)|
|Turtles||Global||RAG1, RAG2 and c-mos||Latitude ||0.050d||224||Lourenço et al. (2013)|
|COX1, ND4 and cyt b||Latitude ||0.001||224|| |
| ||Global||12S, 16S, cyt b||Depth||0.026||42||Wright et al. (2011)|
| ||Global||ITS ||Latitude ||0.0001||45||Wright et al. (2006)|
| ||Global||18S ||Latitude ||0.031||36||Gillman et al. (2010)|
| ||Australia||ITS||Aridity||0.003||30||Goldie et al. (2010)|
| ||Not given||18S, rbcL, atpBc||Temperature||0.001||86||Davies et al. (2004)|
Greater within-species genetic divergence in lower latitude populations has also been demonstrated among vertebrates (Martin & McKay, 2004; Adams & Hadly, 2013), plants (Eo et al., 2008) and birds (Chek et al., 2003) and greater phenotypic differentiation among bird populations at lower latitudes has been reported (Martin & Tewksbury, 2008). Genetic divergence among populations is important if such divergence is a necessary precursor to speciation.
In addition to these latitudinal patterns, faster rates of genetic evolution have been found among birds, amphibians and mammals occurring at lower elevations (Bleiweiss, 1998; Gillman et al., 2009; Wright et al., 2010) and among fishes occupying shallower water (Wright et al., 2011) (Table 1). Mean temperatures are lower at higher latitudes and elevations and at greater ocean depths. Seasonal temperature variation also increases with latitude and elevation on land but it decreases below the thermocline in oceans and peaks at mid-latitudes, not high latitudes, in surface ocean waters (Clarke & Gaston, 2006). Range size increases with latitude but not with elevation. Therefore, although other factors such as population size may account for the patterns observed, we suggest that temperature, or a temperature-related variable such as productivity, is linked with rates of genetic evolution. Increased resource concentration has been shown to produce faster rates of genetic evolution in microbes (Stevens et al., 2007). However, further studies that manipulate temperature and measure rates of genetic change among microbes would be able to test for the independent effect of temperature. In conclusion, the evidence to date is consistent with the prediction that rates of genetic evolution are generally faster in warmer (or more productive) environments, but is not conclusive. For a discussion of putative mechanisms that might underpin these relationships see Gillman & Wright (2013).
Prediction 2: Diversification rates depend on rates of genetic evolution
Given that genetic evolution is necessary for speciation in allopatric populations (Martin & McKay, 2004) and necessary for both reproductive isolation and speciation in sympatric populations, it might be expected, with other factors being equal, that greater underlying rates of genetic evolution lead to greater rates of speciation.
Correlations between both diversification rate and genetic evolution, and between current species richness and genetic evolution in nuclear, mitochondrial and chloroplast genes have been demonstrated (Guo & Ricklefs, 2000; Barraclough & Savolainen, 2001; Jobson & Albert, 2002; Eo & DeWoody, 2010; Duchene & Bromham, 2013). However, the assumption that speciation is dependent on genetic evolution is not without challenge. Correlations between species level genetic evolution and species richness have often been interpreted as accelerated rates of evolution in response to speciation, rather than the converse (e.g. Webster et al., 2003). This alternative interpretation is based on the hypothesis that reduced population sizes (assumed to occur during speciation events) may cause a spike in the number of nearly neutral non-synonymous mutations (dN) (i.e. those that result in amino acid replacement).
Investigation of this question has taken five approaches: (1) testing the assumption under nearly neutral theory (Ohta, 1992) that rates of genetic evolution are faster in small populations; (2) testing for the relative strengths of correlations among rates of genetic evolution, ambient temperature and speciation rate (Davies et al., 2004); (3) testing the prediction that if greater rates of speciation increase substitution rates due to the birth of new species with small populations, then greater variation in the rate of substitution will also occur among species (Lancaster, 2010); (4) testing the prediction, under nearly neutral theory, that reduced population sizes during speciation will have increased the rate of nearly neutral, non-synonymous, substitutions among clades that have undergone faster rates of diversification (Lanfear et al., 2010) (note that synonymous substitution rate is predicted not to be affected by population size under nearly neutral theory); and (5) testing the prediction, under nearly neutral theory, in circumstances where genera have more species at temperate latitudes rather than less, that rates of genetic evolution are also faster among the temperate species (Wright et al., 2006).
Are rates of genetic evolution faster in smaller populations?
Under nearly neutral theory deleterious mutations with selection coefficients so small that they act like neutral mutations are fixed at greater rates in smaller populations where purifying selection is thought to be less efficient. Therefore, genetic evolution is expected to be faster in smaller populations. Several studies spanning three decades have tested for a relationship between population size and rates of genetic change and although there has been some empirical support for nearly neutral theory, overall the results have been equivocal (Woolfit & Bromham, 2005). Hawks et al. (2007), for example, show increasing rates of genetic evolution with increasing human population size. The study with greatest replication (n = 70), used pairs of distantly related species from islands and continents and found no overall difference in the rate of genetic change between small and large populations (Woolfit & Bromham, 2005). Recently, Wright et al. (2009) used 48 pairs of closely related bird species (mostly sister species) of contrasting population sizes and found, contrary to expectations under nearly neutral theory, that bird species with smaller populations had experienced a slower pace in genetic evolution than those with larger populations. Nearly neutral effects probably occur, but there is a lack of evidence suggesting that they predominate and thereby underpin the relationship between rates of diversification and genetic evolution.
What are the relative strengths of correlations among rates of genetic evolution, ambient temperature and speciation rate?
Using a selected angiosperm species from each of 86 sister families, Davies et al. (2004) found relationships between: temperature and species richness; temperature and rates of substitution and; substitution rate and species richness. However, although the relationship between substitution rate and species richness was significant (P = 0.004), substitution rate dropped out of the regression model as a significant predictor variable of species richness when the model was simplified. Davies et al. suggest this indicates that temperature independently influences species richness and the rate of genetic evolution, but that the rate of genetic evolution does not influence species richness. However, the rate of genetic evolution within each sister family in that study was represented by just one species using only the third codon position. A single branch length from each family may be adequate to represent the dependent variable when regressed with temperature. However, it is unclear, given the diversity of branch lengths among species within angiosperm families, whether or not a single branch length from each family is sufficiently free from error to adequately characterize the predictor variable in a regression with speciation rate, where strength of correlations are to be used for comparative purposes.
Do clades with greater speciation rates contain greater variation among species in the rate of genetic evolution as expected if speciation causes a short-term spike in substitutions?
Using fossil-aged lineages, Lancaster (2010) found a positive correlation between substitution rates and rates of diversification among 13 clades of angiosperms as expected if rates of DNA evolution influence speciation. However, she did not find a correlation between substitution rate variation and diversification rate. She therefore rejected the hypothesis that speciation itself had caused an increase in substitution rate.
Is non-synonymous genetic change elevated relative to synonymous change among clades that have undergone faster rates of diversification?
If reduced population size leads to accelerated genetic evolution during speciation via nearly neutral effects it is predicted that the ratio of non-synonymous to synonymous (dN/dS) substitutions should be elevated in more rapidly diversifying clades, while synonymous substitutions should not be affected. Contrary to this prediction, Lanfear et al. (2010) found a positive correlation between the rate of diversification of birds and the rate of synonymous substitutions, but no relationship with dN/dS, suggesting that speciation was not the cause of the increase in rate of genetic evolution. Similar results have been found among Eutherian mammals for nuclear DNA, but not for mitochondrial DNA, and no significant correlations for either nuclear or mitochondrial DNA were found for non-Eutherian mammals (Goldie et al., 2011).
Do species in temperate climates have faster rates of genetic evolution than those in tropical climates, in contrast to the typically faster rates in the tropics, when the genera are more species rich at temperate latitudes?
Wright et al. (2006) tested this prediction among plant species and found that, contrary to the expectation under nearly neutral theory, temperate species had slower rates of genetic evolution even when they belonged to genera that were more species rich at higher latitudes.
Taken together these studies provide little support for the hypothesis that reduced population size during speciation is the cause of the association between rates of genetic evolution and diversification. By contrast, results from empirical studies are generally, but not always, consistent with the alternative hypothesis that the rate of genetic evolution influences the rate of diversification.
Prediction 3: Rates of origination depend on latitude and ambient temperature
Empirical studies testing for associations between net diversification rates and latitude have been conducted for a diverse range of taxa using a variety of approaches. The fossil record shows lower average ages of taxa and greater rates of diversification towards the equator (Stehli et al., 1969; Stehli & Wells, 1971; Jablonski, 1993; Buzas et al., 2002; Crame, 2002; Jablonski et al., 2006).
Further evidence comes from phylogenetic data. Sister clades with a common ancestral node are, by definition, of equivalent evolutionary age and therefore the number of extant species within each clade can be used to compare rates of diversification. Comparisons involving sister clades have demonstrated faster rates of net diversification at lower latitudes for passerine birds, swallowtail butterflies (Cardillo, 1999), ferns (Schneider et al., 2013) and angiosperms (Davies et al., 2004; Jansson & Davies, 2008), whereas three studies, involving Old World primates (Bohm & Mayhew, 2005), herbivorous insects (Farrell & Mitter, 1993) and damselflies (McPeek & Brown, 2000), failed to find statistically significant relationships. However, the latter studies lacked statistical power due to the small sample sizes employed (n = 9, 5 and 1, respectively).
In order to overcome limitations due to a lack of available sister clades of equivalent age, several studies have estimated diversification rates using clade ages derived from node depths of DNA-based phylogenies. Using this technique, negative relationships between latitude and net diversification rates have been found for birds (Cardillo et al., 2005; Ricklefs, 2005, 2006), New World palms (Svenning et al., 2008), squamate reptiles (Ricklefs et al., 2007) and amphibians (Wiens, 2007). Tests using subsets of amphibia (Hylidae & Ranidae) did not, however, produce significant results (Wiens et al., 2006, 2009) and nor did a recent test for a relationship between latitude and diversification among mammal genera (Soria-Carrasco & Castresana, 2012). A positive relationship between temperature and diversification rate was found for plethodontid salamanders, whereas a negative relationship was found for carnivoran mammals (Machac et al., 2012). However, because genetic evolution is generally faster at lower, warmer latitudes, studies that derive divergence dates using DNA-based phylogenies and then use these to estimate diversification rates, suffer from a bias that underestimates the rate of diversification at lower latitudes. Similarly, greater genetic divergence among pairs of species from lower latitudes has been interpreted as indicating slower rates of origination at these latitudes (Weir & Schluter, 2007). Again, this result is a likely artefact of faster rates of genetic evolution at lower latitudes (Gillman et al., 2009, 2011).
Both palaeoecological and phylogenetic studies have the potential to under-represent tropical taxa due to greater sampling effort within temperate latitudes and the relative paucity of fossil deposits in the tropics (Johnson, 2003; Tobias et al., 2008). This data bias can be assumed to have had a conservative influence on the studies reported above, the majority of which provide evidence of increasing net diversification towards the equator. However, net diversification rates provide little insight into the relative contributions of origination, dispersal and extinction.
Studies that differentiate between these components are few, but faster rates of origination at lower latitudes have been found among marine bivalves (Flessa & Jablonski, 1996; Jablonski et al., 2006), 26 invertebrate orders (Jablonski, 1993; Martin et al., 2007) and marine foraminifera (Allen et al., 2006). Higher latitudes appear also to be net receivers of taxa dispersing ‘out of the tropics’ (Jablonski et al., 2006). This leakage of taxa from low to high latitudes may therefore act to ameliorate the LDG.
Although extinction appears, in many cases, to contribute to net diversification asymmetries favouring the tropics (Jablonski et al., 2006), and other areas of high diversity such as California (Lancaster & Kay, 2013), extinction rates among some taxa have been greater in tropical regions (Roy & Pandolfi, 2005; Martin et al., 2007; Powell, 2007). Among marine bivalves, extinctions rates were found to be highest in temperate regions and lowest in Polar Regions during the last 5 million years (Valentine et al., 2008). Thus, the extinction pattern is not always consistent with the LDG and would therefore not appear to be the primary mechanism driving net diversification heterogeneity, although it may contribute to the pattern in many cases.
Although it is a parsimonious explanation that latitudinal patterns in diversification rates are due to a negative covariance of temperature with latitude (relationship 3, Fig. 1), this is not necessarily the case as other variables related to latitude may underpin patterns of diversification. However, a recent study of biodiversity patterns over the last 540 million years using fossil marine invertebrates, while controlling for sampling effort, found strong temporal associations between temperature and biodiversity and temperature and origination rates (Mayhew et al., 2012).
There have been alternative mechanisms suggested to account for latitudinal gradients in origination. Schemske (2002) suggested that biotic interactions are more important for evolution and origination in the tropics and because biotic interactions are also more spatially variable in the tropics, geographically isolated populations face different evolutionary pressures and this enhances divergence into different species. However, this hypothesis relies on high pre-existing diversity at low latitudes in order for there to be greater heterogeneity in biotic selection pressure. Therefore, it is unclear as to how this mechanism could act as a primary generator of diversification gradients. Such a mechanism may, nonetheless, contribute to a positive feedback system that enhances or maintains diversity (Fig. 1). Dynesius & Jansson (2000) suggested that at higher latitudes more extreme climate changes in the past might have favoured selection for strong dispersal capabilities and large range sizes, because these traits better enable species to track favourable conditions as they shift across latitudes. In turn, such traits could reduce speciation at high latitudes by enhancing gene flow and favouring more generalized adaptations.
In summary, there is strong evidence that a broad range of life forms have experienced faster net rates of diversification at lower latitudes. Evidence from marine taxa suggests that this differential is likely to be due to greater rates of origination in the tropics. Extinction rates, rather than consistently contributing to net diversification differentials, might in some cases have been greater at lower, not higher, latitudes. Similarly, range shifts tend to be tropical to polar, thus acting to counter the LDG rather than enhancing it. However, there is a clear need for further work to explore whether or not there is a relationship between temperature and diversification that is independent of latitude, because there are alternative hypotheses that might explain latitudinal gradients in diversification that do not invoke average temperatures.
Prediction 4: Species richness is dependent on diversification rate
Several studies (reviewed above) have found rates of origination that are consistent with the LDG (e.g. Cardillo et al., 2005; Jablonski et al., 2006) and that are therefore also consistent with this prediction (relationship 4 in Fig. 1). However, if diversity patterns derive from rates of diversification then patterns in diversity that are either independent of, or contrary to, the LDG should also be associated with the same atypical pattern in the tempo of diversification. Such tests are few. However, Ricklefs et al. (2006) report that mangrove species richness patterns, which could not be attributed to climatic variation, were nonetheless related to rates of origination. Marine bivalves, nanoplankton and radiolarians that have had greater rates of origination at mid-latitudes, rather than at low latitudes, also have atypical species richness maxima at mid-latitudes (Allen et al., 2006; Krug et al., 2007). Despite similar climates, the species richness of Gladiolus is an order of magnitude greater in the Cape of South Africa than in the Mediterranean Basin where diversification has been three to five times slower (Valente et al., 2011). These studies are therefore consistent with the prediction that diversity patterns are dependent on rates of origination. However, a high diversification rate among carnations has also been found in temperate Europe where plant diversity is not high (Valente et al., 2010), suggesting there is a need for more investigation in this area.
Prediction 5: Species richness is dependent on ambient temperature
The ESH was formulated to explain the positive relationship between species richness and latitude. However, it assumes that the important latitudinal variable controlling species richness is ambient temperature and therefore it predicts a positive monotonic relationship between temperature and species richness (relationship 5 in Fig. 1). At this point the ESH runs into the problem that the observed relationship between ambient temperature and species richness tends to be unimodal if richness is measured across a gradient that includes environments where water becomes a limiting factor. As temperature increases, species richness tends to increase up to a maximum, beyond which water deficits then begin to depress richness (O'Brien, 1998; Francis & Currie, 2003; Hawkins et al., 2003, 2005; Field et al., 2005). Given that water and warm temperatures are both necessary for high species richness, with some exceptions such as reptiles (e.g. Whittaker et al., 2007), it suggests that productivity or energy flux rather than ambient temperature is the primary factor that limits species richness. Indeed, for plants, species richness is best predicted by actual evapotranspiration (AET) (Kreft & Jetz, 2007) and generally increases with productivity at both fine and coarse grains and at all but very small spatial extents (Gillman & Wright, 2006, 2010). Both water and temperature have also been found to be important for species richness patterns in birds, mammals, amphibians (Kalmar & Currie, 2007; Whittaker et al., 2007; McCain, 2009) and invertebrates (e.g. Kaspari et al., 2000; Hawkins & Porter, 2003). Productivity may therefore be a better predictor of total species richness than solar energy (e.g. Honkanen et al., 2010). A recent meta-analysis of 273 published studies (Cusens et al., 2012) demonstrates that for both terrestrial and freshwater animals, the predominant relationship between species richness and productivity at all spatial scales is positive. In general, warm dry climates have low productivity and are species poor and warm wet climates are productive and species rich.
If evolutionary speed is an important factor in determining species richness patterns we would then expect to find that evolutionary speed is also dependent on water in environments where water is a limiting factor. Indeed, this was the observed result in a recent study that compared rates of genetic evolution between 28 xeric Australian outback plant species with congeneric rain forest species at similar latitudes (Goldie et al., 2010). Therefore, the ESH requires modification to a construct that predicts a positive relationship between both available water and temperature with rates of genetic evolution, rather than just between temperature and rates of genetic evolution. Water and temperature may affect rates of genetic evolution via a combined influence on productivity. Alternatively, they may have independent effects (Fig. 2) or the relationship may not be causal. If productivity has a causal effect it can be further predicted that variables other than temperature and water, such as soil nutrient status, that influence productivity will also be associated with the speed of genetic evolution.
Figure 2. An integrated evolutionary speed model in which temperature, water and the number of reproducing individuals are all primary influences on the rate of evolutionary speed and genetic diversity acting via biome level rates of mutation and the commensurate rate at which mutations with positive selection coefficients are produced. Water and temperature might have independent influences on rate of genetic evolution or they might act through their combined influences on productivity. If productivity has a causal influence on rates of genetic evolution then other factors that influence productivity will also be predicted to influence these rates. Area is proposed to influence rates of genetic evolution via the total number of reproducing individuals. Secondary influences on evolutionary speed and genetic diversity are proposed due to rates of biotic and abiotic environmental change and via environmental heterogeneity.
Download figure to PowerPoint