Threat information and extinction risk

Improving comparative extinction risk analysis

‘Extinction risk’ is a powerful notion. If one species’ extinction risk is way higher than another species then there’s a strong argument to give it a greater share of the limited resources available for conservation. New research is suggesting, however, that there’s much we can do to improve the way we calculate extinction risk. A common approach for determining the likelihood of a species going extinct within the foreseeable future involves something like this:

1. Take a group of species of interest (eg, stinkbugs)

2. Look at their levels of endangerment using some metric (eg, ‘extinct’ vs ‘still here’ or their IUCN Red-List status)

3. Gather some information about those species (eg, how many offspring they have per year, how much they weigh), and

4. Search for correlations between their level of endangerment (2) and this species information (3) using your favourite statistical method (generally known as a ‘comparative extinction risk analysis’).

Such analyses have been instrumental in highlighting that the risk of species’ extinction is not random. For example, a common pattern is that larger-bodied, longer-lived and more ecologically specialised species are at greater risk (Purvis et al., 2000).

However, despite the potentially useful application of comparative extinction risk analyses for conservation practice and policy, they have generally had little influence in the real world (Cardilllo and Meijaard 2012). One reason for this under performance may be that few of these studies seem to incorporate information on the threats affecting species’ survival (eg, habitat loss, invasive species, overhunting). Rather, many seem to focus on life-history and ecology as factors that predispose species to decline or extinction. While this can lead to general management conclusions (eg, protect big predatory fish), it may not tell you why they are under threat in the first place (eg, overfishing) or how you might go about protecting them.

“If you decide that comparative extinction risk analysis could be a tool of use in your extinction-fighting toolbox, you should do the best you can to quantify and represent threats before drawing any management conclusions.”

In other words, a species’ risk of extinction is shaped by the interactions that arise between its intrinsic traits and the extrinsic factors to which it may be exposed, such as threats. Recognising those interactions could help pinpoint better management options.

In our paper, we investigate whether threats really are under utilised in extinction risk studies and whether this actually matters from a conservation management perspective (Murray et al., 2014). The answers are: yes and yes!

First, we reviewed almost 100 studies and found that the majority (63%) did not include threat variables at all, even though most (59%) of the threat variables that have so far been employed have been significant predictors of extinction risk. Despite this, we did find an increasing trend over time in the number of studies that include threats to explain patterns of extinction risk, which is promising (see Figure 1).

Second, we investigated the value of threat information in extinction risk analyses by comparing predicted IUCN Red List classifications from extinction risk models with and without threat variables. We made comparisons using two study systems of differing taxonomic and geographic scales: a global dataset on mammals and a continental dataset on amphibians.

We found that including threat variables only modestly improved our overall ability to predict threatened species. However, models with and without threats disagreed considerably on the identity of threatened species (11% and 5% for amphibians and mammals, respectively, translating to dozens and hundreds of species).

Third, we examined the potential ramifications of this disagreement for a conceivable management end-use. As a hypothetical example, we shortlisted species to receive increased conservation attention on the basis of their relative ‘endangerment’, predicted from our model. We found that models with and without threats disagreed on 20-60% of the species that would be on the shortlist, depending on the proportion of all species included in the list. So, if you decide that comparative extinction risk analysis could be a tool of use in your extinction-fighting toolbox, our study suggests you should do the best you can to quantify and represent threats before drawing any management conclusions.

More info: Martina Di Fonzo or Kris Murray


Cardilllo M & E Meijaard (2012). Are comparative studies of extinction risk useful for conservation? Trends in Ecology and Evolution 27:167-171.

Murray KA, LD Verde Arregoitia, A Davidson, M Di Marco & MMI Di Fonzo (2014). Threat to the point: improving the value of comparative extinction risk analysis for conservation action. Global Change Biology 20:483–494.

Purvis A, P-M Agapow, JL Gittleman & GM Mace (2000). Nonrandom extinction and the loss of evolutionary history. Science 288:328-330. DPoint78__Page_09_Image_0001

Figure 1: Trends in extinction risk publications. (a) The proportion of studies considering threats has increased through time, against a background of more publications per year. (b) In step with more publications per year, models of extinction risk have also become more variable rich with fairly consistent inclusion of threat or other extrinsic factors since 2000; however, on average, models remain dominated by intrinsic trait variables (From Murray et al., 2014).

Leave a Reply