Introduction
The Anthropocene is characterized by an unprecedented rate of biodiversity loss driven by a number of anthropogenic stressors including climate change, pollution, habitat loss, overexploitation and the transmission of invasive species (Young et al. 2016). As populations decline in the face of these stressors the need for conservation intervention becomes increasingly important. However, conserving small populations is complicated as declining population size increases the risk of detrimental demographic processes driving populations inexorably towards extinction (Fagan & Holmes 2006). For example, individual fitness in many species is expected to decrease with population size due to Allee effects (Berec et al. 2007) and a loss of genetic diversity (Saccheri et al. 1998; Blomqvist et al. 2010); demographic stochasticity influences small populations by increasing the annual variability in population growth rate (Fagan & Holmes 2006), which is particularly problematic in small populations as major fluctuations could lead to their extinction (Gilpin & Soulé 1986; Caughley 1994); moreover, small populations are also especially vulnerable to direct extirpation from external drivers of mortality, such as environmental stochasticity and random catastrophes (Caughley 1994). The concurrent presence of these processes is thought to lead to self-reinforcing, rapid and catastrophic downward spirals to extinction, so-called ‘extinction vortices’ (Gilpin & Soulé 1986), during which there may be little prospect of the population recovering even with intense conservation effort (Palomares et al. 2012). To identify the populations most at-risk and to make informed conservation decisions, we need to understand the factors that determine the robustness of a population to extirpation by the extinction vortex.
Fagan and Holmes (2006) empirically corroborated, albeit with a small database of population extirpations, several preexisting hypotheses of the extinction vortex; specifically, that (i) time to extinction scales to the logarithm of population size, indicating that as a population declines its time to extinction decreases at an increasing rate, (ii) geometric growth rate declines as extinction nears, due to declining individual fitness and (iii) annual variability in population change increases as extinction nears, attributable to an increasing influence of stochastic factors.
A species’ intrinsic and ecological traits are often key predictors of extinction risk (Gaston & Blackburn 1995; Purvis et al. 2000; Cardillo et al. 2008), with geographic range size, life-history speed, and degree of specialization emerging as persistent indicators (Chichorro et al. 2019). However, with few exceptions (Duncan & Young 2000; Brashares 2003; Koh et al. 2004), real-life extinctions have rarely been used to infer extinction proneness in relation to biological traits (Brook et al. 2008). Similarly, using a lab-based experiment, Godwin et al. (2020) were the first to explicitly investigate how variation in a specific behavioral trait (mating pattern) can result in differential population vulnerability to the extinction vortex; though analogous studies have not been carried out on real-life population data.
Unfortunately, many of the traits identified as important predictors of extinction risk are difficult to measure, particularly in populations which are already severely reduced, meaning that it is necessary to use proxy measures of these intrinsic ecological traits. Body size is a particularly important trait, associated with a suite of intrinsic, ecological and anthropogenic factors that are frequently invoked in studies relating to extinction risk such as life-history speed (Johnson 2002), population density (Fa & Purvis 1997; Davidson et al. 2009), and the level of exploitation by humans (Owens & Bennett 2000; Ripple et al. 2016, 2019). Furthermore, because of the ease of obtaining body size data and the significance of body size as a correlate of many hard-to-record population traits, it is arguably the most readily available trait available among taxa. This enhances the potential utility in predicting how small populations will respond without having to obtain more cryptic information with time-consuming and expensive data-collection procedures.
Smaller-bodied species are generally more fecund with greater intrinsic rates of growth, meaning they can recover from perturbations more quickly (Brook & Bowman 2005) and spend less time at small population sizes where there is a large threat of extirpation (Allen et al. 2017). However, slower life history speed in larger-bodied species is linked to greater resistance to both environmental (Millar & Hickling 1990; Peltonen & Hanski 1991; Sinclair 2003; Saether et al. 2013; Yeakelet al . 2018) and demographic (Jeppsson & Forslund 2012; Saether et al. 2013) stochasticity. Greater susceptibility to stochastic processes implies that populations of smaller-bodied species can be abruptly reduced to a point where the risk of extinction is acutely high (Schoener et al. 2003; Allen et al. 2017). The relationship between body size and robustness against the extinction vortex thus depends on the relative importance of population growth rate versus susceptibility to stochastic elements.
Here, we assess – for the first time – whether body size can interact with underlying demographic processes to influence the dynamics of a population in the region of an extinction event, building upon the analysis of Fagan and Holmes (2006) but with a much larger database. To do this requires cases of populations monitored through to extirpation, negating the need to designate quasi-extinction thresholds, which could result in erroneous interpretations of extinction dynamics (Fagan & Holmes 2006). We use a global database of vertebrate population time series, supplemented with body size data from various life history databases to identify 55 populations where extirpation has been observed. We find support for the three aforementioned predictions of the extinction vortex (Gilpin & Soule 1986; Fagan & Holmes 2006) and evidence that deterioration in population dynamics, due to the extinction vortex, takes place at a faster rate in smaller-bodied species.