Extinction Drivers: Animals Versus Plants

Extrinsic Versus Intrinsic Factors

Explaining why some species appear predisposed to higher extinction risk than others is an important goal for conservation research (McKinney 1997). The five main extinction drivers include habitat loss, climate change, increased pollution, resources over-exploitation and invasive species (Millennium Ecosystem Assessment 2005), and all are linked directly or indirectly to anthropogenic pressures. These drivers parallel Jarred Diamond's 'evil quartet' (Diamond 1984, 1989), but with the more recent addition of climate change, and Diamond additionally included the possibility of extinction cascades in which secondary extinctions follow the loss of key species, for example, due to the disruption of ecosystem processes. We can further simplify this list into extrinsic (e.g. climate change) and intrinsic factors (e.g. ecological traits such as population density and species life-history traits such as body size and gestation length) (Cardillo et al. 2005). Extrinsic factors might help explain geographic variation in extinction risk, whereas intrinsic factors might better explain taxonomic patterns; however, highest risk is where driver intensity associated with extrinsic factors overlaps with species intrinsic vulnerability. In addition, species are increasingly likely to be exposed to multiple drivers, and this will likely further exacerbate risk of extinction (Brook et al. 2008).

Extinction Drivers in Animals

Correlates of extinction risk in animal kingdom have been explored extensively using data from the IUCN Red List (Bennett and Owens 1997; Russell et al. 1998; Purvis et al. 2000a, b; Cardillo 2003; Cooper et al. 2008) with particular attention to mammals (Russell et al. 1998; Cardillo et al. 2005, 2008; Isaac et al. 2007; Huang et al. 2012), perhaps the best-studied higher taxonomic group. Across studies, high extinction risk is generally associated with large body size, long generation times and small geographic range sizes (Bennett and Owens 1997; Russell et al. 1998; Purvis et al. 2000a; Cardillo 2003; Fisher and Owens 2004; Cooper et al. 2008). Conversely, species at low risk of extinction are small, reproduce rapidly, and have a wide niche breadth.

We know, for example, that mammals that are at risk of extinction are, on average, an order of magnitude heavier than non-threatened species (IUCN 2003). The size-selectivity of extinction risk is not unique to the current extinction crisis; past mass extinction events, such as that of the late Pleistocene, were also biased towards larger species (Martin 1967; Johnson 2002). During the late-Pleistocene – early-Holocene extinction event, there was a mass extinction of much of the mammalian megafauna, resulting in a loss of several complete ecological guilds and their predators (Cione et al. 2003). Size selectivity in extinction risk has been long-recognised (e.g. Pimm 1991; Lawton 1995; Pimm et al. 1988; Cardillo and Bromham 2001; Johnson 2002), and there are many potential explanations. Large-sized mammals might be more extinction-prone because of generally lower average population densities (Damuth 1981), putting them at greater risk from stochastic population dynamics. High risk in large bodied mammals might also reflect the negative correlation between intrinsic rates of population increase and body mass (Fenschel 1974), and thus longer recovery times following population declines. There might also be an increased propensity for humans to exploit larger species (Bodmer et al. 1997; Jerozolimski and Peres 2003). The relationship between species traits and extinction risk, however, is not straightforward, because of the complex interaction between intrinsic and extrinsic drivers, and different clades might have very different predictors (e.g. see Cardillo et al. 2008).

Cardillo and colleagues (2005) demonstrated that risk in small-sized mammals (<3 kg) was largely determined by extrinsic factors including the size and location of geographical ranges. However, the predisposing factors in the larger size class include both intrinsic species properties (e.g. population density, neonatal mass and litters per year) and extrinsic factors. Such fine-scaled analyses can help address whether extinctions are linked to 'bad genes' or 'bad luck' (Raup 1993; Bennett and Owens 1997). For mammals, it appears that extinction in small bodied species is more likely a case of bad luck, driven by extrinsic factors. For larger bodied species, bad genes, that is, genes controlling intrinsic traits such as body size and life history are additional aggravating factors promoting extinctions.

Compared to vertebrates, the distribution and drivers of extinction risk in invertebrate communities has been poorly explored. However, a recent study estimated that one-fifth of invertebrate species may be threatened with extinction, with freshwater species at particular high risk (Collen et al. 2012). Collen and colleagues suggested that the greater threat to freshwater species was predominantly driven by agricultural pollution and dam construction, invasive species and waterborne diseases. More generally, and perhaps unsurprisingly, species that are less mobile and with limited geographic ranges, such as freshwater mollusks, tend to be at higher risk (Collen et al. 2012). In marine ecosystems, however, the market values of some invertebrates correlate strongly with their risk of extinction, e.g. invertebrate species considered luxury seafood (Purcell et al. 2014), providing an exception to the general trend for greater threat to be observed in larger-sized species.

The phylogenetic distribution of extinction risk in mammals has also been of much interest. In mammals, it has been suggested that species subtending from longer phylogenetic branches, and thus representing greater unique evolutionary history, are at higher risk of extinction (Russell et al. 1998; Purvis et al. 2000a). This pattern matches to Wilson's (1961) 'taxon cycle', which predicts that older species would have higher extinction probabilities as species expand and contract in their geographical distributions over their evolutionary lifetimes. Although, as originally described, the taxon cycle referred to the distribution of species on islands (ants on islands in Melanesia), the concept has been extended to include species on continents (e.g. see Ricklefs and Bermingham 2002). Alternatively, it might simply echo the pattern of historical extinctions, in which older species represent survivors of once more diverse clades (Purvis et al. 2000a). However, the precise relationship between extinction risk and evolutionary age remains debated (Verde et al. 2013). Further, patterns of extinction risk in plants appears to show an opposite trend, with higher risk associated with young species in species rich (Schwartz and Simberloff 2001; Meijaard et al. 2007) and more rapidly diversifying clades (Davies et al. 2011), suggesting that predictors of extinction in plants might be very different to those for mammals.

Extinction Drivers in Plants

Species extinction in the plant kingdom is predominantly the result of habitat loss, for example through deforestation. Tropical forests, which cover less than 7 % of the world land area, contain over 50 % of global biodiversity (Dirzo and Raven 2003), but these unique habitats are being destroyed at unprecedented rates (Laurance 1991; Achard et al. 2002) as a result of rapid human population growth and economic development. In tropical Asia and Africa, over 40 % of the primary forests is already lost (Wright 2005). This drastic reduction in forest cover has had a devastating impact on plant diversity (Sodhi and Brook 2006). Although there is some evidence that, globally, recent rates of deforestation are slowing, we likely owe a large extinction debt due to the time lag between habitat loss and species losses predicted from the reduction in area. Thus, even should we be successful in preserving the remaining forest cover, many species might still be predicted to be lost over the following decades as habitats return to a new, lower diversity, equilibrium state. This extinction event will likely be exacerbated by the effects of ongoing climate change as local climate conditions shift and species are forced to either adapt to new conditions or track climate space (Willis et al. 2008).

Plant responses to environmental change are difficult to predict. With warming, plants might adapt by shifting their phenologies – the timing of life history strategies – for example flowering earlier and losing leaves later (Parmesan 2007). Recent work indicates significant phylogenetic conservatism in flowering phenology (Davies et al. 2013), suggesting that there might be some evolutionary constraints to species adaptive responses. If the velocity of climate change is high, species may not have the necessary time to adjust their phenological responses. Alternatively, species might track suitable climates, for example by shifting their distribution northwards or towards higher elevations (Sandel et al. 2011). Species already restricted to high elevation biomes might then be particularly vulnerable as increased warming may result in the reduction of suitable habitat and, at the extremes, complete habitat loss. In the biodiversity hotspots of the Eastern Arc Mountains of East Africa, species at higher elevations already tend to be more threatened, perhaps reflecting recent climate shifts (Yessoufou et al. 2012). Species that are unable to adapt their phenology or track climate through space will be most vulnerable to extinction. In data from Thoreau's woods in Concord, MA, spanning 100 years, it is already possible to detect declines in populations among species that have failed to shift their phenologies to match recent climate change (Willis et al. 2008, 2010). These data also revealed phylogenetic structure in species responses, suggesting evolutionary conservatism not only in flowering times, but also plasticity in flowering times (see also Davies et al. 2013).

As for animals, there has been much work aimed at identifying intrinsic lifehistory traits that predispose some plant species towards extinction (Sodhi et al. 2008). However, investigating the correlates of extinction risk within the plant kingdom has proven somewhat more challenging, as key traits frequently differ between studies (Walker and Preston 2006; Sodhi et al. 2008). In addition, traits explain only a small proportion of the variation in extinction risk and, with the exception of geographic range size, we have yet to reveal any single strong correlate equivalent to, for example, body size in mammals. Life-history traits that have been found to correlate with plant extinction include pollination syndrome (e.g. wind or animal mediated), sexual system, habit, height, and dispersal mode (Sodhi et al. 2008). For tropical angiosperms, these traits can explain ~10 % of extinction risk (Sodhi et al. 2008), whereas equivalent models of intrinsic drivers for mammals can explain up to 30 % of the variation in extinction risk (Cardillo et al. 2008). However, even for mammals, explanatory power tends to be lower when exploring predictors across disparate clades (Cardillo et al. 2008), reflecting clade specific sensitivities to different drivers. Perhaps, therefore, it is unsurprising that in flowering plants, a group containing up to 500,000 species, predictive models are often poor.

An alternative avenue of exploration has considered the importance of evolutionary history in models of extinction risk (Sodhi et al. 2008; Davies et al. 2011). In plants, there is increasing evidence that a species evolutionary history might be more important than its life history in explaining extinction risk. As mentioned above, threatened terrestrial plants generally fall within species-rich clades (Schwartz and Simberloff 2001; Pilgrim et al. 2004) that represent recent radiations (Davies et al. 2011). However, when we look at the distribution of extinction risks across plant families, species-poor and especially monotypic families also appear to contain species at higher risk of extinction (Vamosi and Wilson 2008). It is therefore possible that mechanistic explanations for variation in extinction risk differ between old and young clades. Old and species-poor families may represent remnants of once more diverse clades, with species vulnerabilities associated with intrinsic life history traits and long generation times, as in mammals. In contrast, extinction risk in younger, still diversifying clades, may be more closely linked to the speciation process, with high extinction risk more closely associated with traits driving speciation, such as small geographic range size and short generation times.