The X-Team


Can we define vulnerability to extinction risk of understudied species based on the known parameters of well studied taxa? 


Species diversity is balanced by extinction and speciation rates, but there is variation across the tree of life due to different rates in different species. Extinction has been defined in a phylogenetic context as the termination of a branch in a phylogenetic tree; that is, a branch that has no living descendants and cannot therefore be reconstructed from data obtained from contemporary species 21. Normally extinction rate is balanced by speciation events, but from fossil and geological records we know that there were five times in which extinction rate was higher than in any other geological interval—near the end of the Ordovician, Devonian, Permian, Triassic and Cretaceous Periods (See Table 1)1. These are known as the “Big Five” mass extinction events. However, these events display distinctly different patterns of diversity changes. For some there is a decrease in extinction rate followed by a similar decrease in speciation rate (Triassic and Devon) while others were driven predominantly by extinction (Ordovian, Permian and Cretaceous) 2.

Tab.1 The ‘Big Five’ mass extinction events 1

Event Proposed causes
The Ordovician event, ended, 443 Myr ago; within 3.3 to 1.9 Myr 57% of genera were lost, an estimated 86% of species. Onset of alternating glacial and interglacial episodes; repeated marine transgressions and regressions. Uplift and weathering of the Appalachians affecting atmospheric and ocean chemistry. Sequestration of CO2.
The Devonian event, ended, 359 Myr ago; within 29 to 2 Myr 35% of genera were lost, an estimated 75% of species. Global cooling (followed by global warming), possibly tied to the diversification of land plants, with associated weathering, paedogenesis, and the drawdown of global CO2. Evidence for widespread deep-water anoxia and the spread of anoxic waters by transgressions. Timing and importance of bolide impacts still debated.
The Permian event, ended, 251 Myr ago; within 2.8 Myr to 160 Kyr 56% of genera were lost, an estimated 96% of species. Siberian volcanism. Global warming. Spread of deep marine anoxic waters. Elevated H2S and CO2 concentrations in both marine and terrestrial realms. Ocean acidification. Evidence for a bolide impact still debated.
The Triassic event, ended, 200 Myr ago; within 8.3 Myr to 600 Kyr 47% of genera were lost, an estimated 80% of species. Activity in the Central Atlantic Magmatic Province (CAMP) thought to have elevated atmospheric CO2 levels, which increased global temperatures and led to a calcification crisis in the world oceans.
The Cretaceous event, ended, 65 Myr ago; within 2.5 Myr to less than a year 40% of genera were lost, an estimated 76% of species. A bolide impact in the Yucata´n is thought to have led to a global cataclysm and caused rapid cooling. Preceding the impact, biota may have been declining owing to a variety of causes: Deccan volcanism contemporaneous with global warming; tectonic uplift altering biogeography and accelerating erosion, potentially contributing to ocean eutrophication and anoxic episodes. CO2 spike just before extinction, drop during extinction.

Recent publications and reviews have suggested that planet Earth is entering the 6th  mass extinction caused largely by human effects 1,12. This conclusion is primarily based on evidence gathered from a few well-studied lineages - e.g. mammals, birds, fish, and amphibians. Not only are we entering an extinction event, the rate at which species are going extinct is much greater than past events. For mammals, amphibians, birds, and reptiles extinction rates have been shown to be as fast or faster than the rates that produced the “Big Five” extinctions in the past.

However, many questions about the patterns and processes of extinction remain, especially in understudied groups comprising the majority of living organisms (Fig.1). Can we define the vulnerability to extinction risk of understudied species based on the known parameters of well-studied taxa? Can what we know about well-studied taxa be generalized to other groups to determine if we are in fact facing a global mass extinction event? 


Figure 1 Summary of number of animal species in each Red List Category in each taxonomic class.

Proposal: White Paper

For this study we propose to develop a model that takes into account the parameters used to determine extinction risk levels for different well-studied organisms to determine vulnerability to extinction risk of understudied taxa. We discuss the opportunities and challenges of applying this knowledge to these understudied groups and point to future work that could provide useful data for examining extinction processes. 

Several factors (mainly anthropogenic threats) are proposed as the main causes of the present decline of global populations (Table 2). The rate and extent of human-mediated extinctions are debated, but there is general agreement that extinction rates have soared over the past few hundred years, largely as a result of accelerated habitat destruction and the global expansion of the human population during the twentieth century 19. The current and future extinction rates are estimated using a variety of measures such as species–area models and changes in the World Conservation Union’s (IUCN) threat categories over time. Based on the global assessment of all known species, 31% known amphibians, 12% of known birds, and 20% of mammal species are currently listed by the IUCN as under threat. But what about understudied lineages? Can we use phylogenetics together with the information we have for well-studied groups, to access the vulnerability to extinction risk for understudied species?

Tab.2 Factors that are involved in recent species decline

Factor Causes Process Species
Habitat loss Primarily anthropogenic No place to live! all taxa 10 11
Habitat fragmentation Human development Separation of existing populations amphibians 12, mammals
Ocean acidification Carbon emission Prevention of mineralization and loss of symbionts coral and other skeletonized invertebrates
Global warming Greenhouse gases Sensitivity to temperature coral
Emerging infections Climate change and natural causes Decrease in population size amphibians 13, coral, honeybees, mammals 14
Invasive species Human introduction Competition and predation increase mammals 15 16, plants
Overhunting Antropogenic (food or recreation) Decrease in population size mammals 17, fish, birds 18


Past studies looking to determine whether we are facing an ongoing mass extinction or to account for differences in extinction vulnerability among taxa have faced several difficulties, including incomplete information about species biology. Here we propose a phylogenetic-based model that may allow the prediction of vulnerability to extinction risk of understudied groups, the identification of relevant factors that should be considered for studies and the identification of the taxa where more data is needed. Our proposed method provides a way of identifying species vulnerability to extinction even in those clades lacking significant biological parameters information or can function as a proxy when data on that particular species are not available or are very difficult to gather. Considering the increased severity of current anthropogenic pressures, having a way to predict vulnerability is indeed important, especially when data regarding the extinction risk of a group can come too late to take effective measures toward their conservation.

Proposed Work

Measuring extinction combines the magnitude as the percentage of species that have gone extinct and rate of extinction which is the number of extinctions divided by the time over which the extinctions occurred that can be observed in the fossil record as well as in recent examinations. Parameters like Range size, Migration rates, Genetic diversity, Trophic level, Climatic variables, Life history, Functional traits, Body size, Measurement of heterozygosity, Larval survival and growth, Egg-hatching rate, Female heterozygosity have been used to describe the decline or possible extinction risks of different animal spezies (Table 3). Based on these parameters together with the knowledge about the traits of the well studied animals we will give a model to determine the risk levels for different underestimated species in different domains of life. Grandcolas et al. (2011) summerized a rapid overview of the literature that has shown that many different research fields have a tendency to use non-heritable and extrinsic traits in a phylogenetic context as taxon age, geographical distribution, associates (parasites, symbionts, etc.), and bioclimatic modelled niches. The real problem is that there is no way to know which kind of bias will be introduced if an extrinsic surrogate (extinction risk instead of body size, life histories, etc.) is used in a phylogenetic perspective, and the results are simply not interpretable 9. Therefore the idea of the model used here is to use parameters of the life history of single species together with the biogeografical distribution in the same bios and to studying those risks by considering the appropriate phenotypic heritable traits in a phylogenetic perspective.

Tab.3 - Summary of parameters that are used to model extinction

Parameter Range size Genetic diversity Trophic level Climatic variables Life history Functional traits Body size
Reference 3 4 4 4 5 6 4 7 4 7 8

Evolutionary biologists have developed a number of statistical methods to model the evolution of different morphological, behavioral, and genetic features of organisms across phylogenies22. These models are used to infer ancestral traits as well as the traits in groups of the phylogeny for which data on them are not available. We propose to adapt these methods in order to model the evolution of parameters and processes affecting extinction risk.

Like other organismal traits, risk factors and parameters used to assess extinction risk are likely to be correlated across a phylogenetic tree. We will develop a statistical framework based on existing models developed to describe the evolution of traits on a phylogeny22, 23. The model will account for the rate of evolution of the trait across the phylogeny as well as the degree of correlation between traits (Figure 2). Since trait evolution is likely to be heterogeneous across the phylogeny we will also incorporate heterogeneity in rates of evolution and correlation in traits across the phylogeny. Using these methods we will infer the extinction parameters and processes that are likely to be important to a particular group and assign a confidence interval around these estimates.

We will parameterize the model using existing data on extinction risks in different taxa and conduct simulations to assess the effectiveness of the method. In order to parameterize the model we will compile the parameters used to define extinction risk in each group as well as the processes thought to affect extinction rates. These parameters and processes will be mapped onto a phylogeny and the rates of evolution and correlation between traits will be estimated. Using these data we will parameterize our model of trait evolution within particular groups of the phylogeny for which data are available. We will then apply the model to infer the likely parameters and processes that will affect extinction risk in other parts of the phylogeny. Simulations will be conducted in order to examine several features of the model:

  1. Item 1 We will exclude known trait information from different branches of the phylogeny when parameterizing the model to see how accurately we can recover the values of the traits by extrapolation.
  2. Item 2 We will use simulated data to examine the effects of sparse data coverage over the phylogeny in order to determine the amount of data needed to make accurate inferences.
  3. item 3 We will determine the ability to confidently assess extinction risks from distant branches of the phylogeny when data for closely related taxa is unknown. This final step will be important in determining the taxonomic resolution of our method.

Figure 2 In order to model the evolution of extinction risk it will be necessary to have data on each of the different parameters and proccesses important to extinction risk (trait 1 to trait N). Each trait will have it's own rate of evolution measured by the amount of change over a given unit of distance. Traits may be correlated with one another which can provide additional information for a model of their evolution. The model of evolution may vary across the tree thus it will be important to incorporate heterogeneity in the model in different parts of the phylogeny.


A main assumption that is implicit in this model is that traits that are chosen for the model are correlated with extinction risks and thus are able to inform them. This does not imply that these traits are treated as heritable phenotypic character20, but rather that there is a correlation between traits and extinction risk that can be extrapolated to make inferences in understudied lineages. Although this is admittedly significant, the main aim of this paper is to guide future research into the sixth mass extinction event and not to make broad assumptions about the adaptive qualities of traits or their affect on extinction risks.


As with any model, it is imperative to understand its caveats. The most important to consider for our model are the following:

  1. Item 1 We need a detectable pattern of trait distribution throughout the tree to draw any meaningful conclusions. However, a random pattern would suggest that none of the traits that are used are able to predict extinction risk among taxa.
  2. Item 2 One of the goals of the model is to inform our understating of understudied species, but the model parameters will still be limited by the absence of data that surrounds them. Significantly, trait information is not available for most groups of organisms, and the data that is available is not evenly distributed.
  3. Item 3 Our current estimates of species diversity in many major lineages are admittedly very poor. Even the definition of what constitutes a biological species is heavily debated for most taxa, and essentially non-existent for many others (i.e. bacteria). These factors will influence the total branch lengths in the model, and we need to be aware that changes in the phylogeny may alter the conclusions.

Expected Results

The accuracy of the model will be highly correlated with the taxonomic level used in the analysis. It is expected that groups that split earlier will diverge less in their traits. Therefore, the model will perform better analysing species within a genus than genera within a family, as those divisions tend to reflect time for diversification24. Another important aspect of the model are the traits chosen. Some traits will be important in predicting extinction risk for all groups whereas some will be specific to a few taxa. For instance, body size is thought to be the main caracteristic determining the taxon vulnerability in mammals25. Several traits associated with extinction risk correlate with body mass and above the threshold of three kilograms the species' extinction risk increase considerably. Nevertheless, for amphibians the extinction probablility seams to be correlated to other traits, such as habitat fragmentation and infections (Table 2), and the use of body mass as a predictor of vulnerability would generate unrealistic results. It is also important to take into account the taxonomic level when choosing the traits to be used in the model. For example, when analysing a clade within mammals where all species are either below or above the threshold, the body mass fails to address the species' vulnerability and other traits should be used instead. Another confounding source is homoplasy. As those traits are not shared by a commom ancestor they cannot be included in the model and have to be analysed in each species separately.
With that being said, we expect that this model will enable us to determine the actual vulnerability to extinction of several understudied groups and most important will guide future studies and conservation programs towards understanding the biology of these organisms. Moreover, with this methological approach one can access the extinction risk of organisms that are currently endangered (i.e. cannot be manipulated) by studying its sister-group.

Future Considerations

It is clear that models are limited by the data that informs them and it will be imperative to gather more data of all kinds for undersampled organisms. However, we anticipate that one important aspect of this model is that it will inform directed sampling of species for future predictions about extinction, both for taxa as well as characters. For example, it will identify understudied lineages that if preferentially sampled will significantly increase the predictive ability of the model. Also, it will highlight traits that have a high probability of informing the model, and should be the focus of future efforts for understanding extinction. Finally, it is important to consider that analyses at different scales should be considered in the future. For example, applying this model to a particular community (i.e. considering all taxa in the terrestrial tropics) or at different taxonomic levels (considering only vertebrates) may yield different results. These effects should be explored both with empirical data as well as simulations.

1. Barnosky, A. D., Matzke, N., Tomiya, S., Wogan, G. O. U., Swartz, B., Quental, T. B., Marshall, C., et al. // Has the Earth’s sixth mass extinction already arrived?// Nature 471, 51–57, 2011.
2. Bambach, R. K., Knoll, A. H., & Wang, S. C. Origination , extinction , and mass depletions of marine diversity. Paleobiology30(4), 522-54, 2004.
3. He, F. // Area-based assessment of extinction risk Reports. // Ecology, 93(5), 974-980, 2012.
4. Purvis, A., Gittleman, J., Cowlishaw, G., and Mace, G. // Predicting extinction risk in declining species // Proc. R. Soc. Lond. B, 267, 1947-1952, 2000
5. Thomas, C., Cameron, A., Green, R., Bakkenes, M., Beaumont, L., Collingham, Y., Erasmus, B., Siqueira, M., Grainger, A., Hannah, L., Hughes, L., Huntley, B., Jaarsveld, A., Midgley, G., Miles, L., Ortega-Huerta, M., Peterson, A., Phillips, O. & Williams, S. // Extinction risk from climate change // Nature 427, 154-148, 2004
6. Benito, B., Martínez-Ortega, M., Muñoz, L., Lorite, J. and Peñas, J. // Assessing extinction-risk of endangered plants using species distribution models: a case study of habitat depletion caused by the spread of greenhouses // Biodiversity and Conservation, Volume 18, Number 9, Pages 2509-2520, 2009
7. Dullinger, S. et al. // Extinction debt of high-mountain plants under twenty-first-century climate change // Nature Climate Change 2, 619–622, 2012.
8. Cardillo, M., Mace, G., Gittleman, J., and Purvis, A. // Latent extinction risk and the future battlegrounds of mammal conservation // PNAS 103(11), 4157-4161, 2006
9. Friedman, M., Sallan, L. C. Five hundred million years of extinction and recovery: a phanerozoic survey of large-scale diversity patterns in fishes. Palaeontology 55(4), 707-742, 2012.
10. Cheke, A, and Hume, J, // Lost Land of the Dodo: An ecological history of Mauritius, Réunion and Rodrigues // T & A D Poyser, London, 2008.
11. Pimm, S. and Raven, P. // Extinction by numbers // Nature 403, 843-845, 2000
12. Wake, D., Vredenburg, V. // Are we in the midst of the sixth mass extinction? A view from the world of amphibians. // PNAS 105(1), 11466-73, 2008.
13. Cheng, T., & Rovito, S. (2011). Coincident mass extirpation of neotropical amphibians with the emergence of the infectious fungal pathogen Batrachochytrium dendrobatidis. Proceedings. Biological sciences / The Royal Society, 108(23)
14. Flannery, T. // The rats of Christmas past. Australian Natural History.// 23(5):394-400, 1990
15. Turvey, S. & Helgen, K. // Boromys offella. In: IUCN (2010). IUCN Red List of Threatened Species // Version 2010.4.
16. Hardy, M. // The extinct mink from the Maine shell heaps // Forest and Stream 61:125, 1903
17. Flannery, T., Schouten, P. // A Gap in Nature: Discovering the World's Extinct Animals // William Heinemann, London. ISBN: 0434008192 (UK edition), 2001
18. Duncan, R., Blackburn, T. & Worthy, T. // Prehistoric bird extinctions and human hunting // Proc. R. Soc. Lond. B 269, 517-521, 2002
19. Sodhi, N.S., B.W. Brook & C.A.J. Bradshaw. 2009. Causes and consequences of species extinctions. Pp. 514-520 in Princeton Guide to Ecology (S. A. Levin, ed.). Princeton University Press.
20. Grandcolas, P., Nattier, R., Legendre, F. & Pellens, R. Mapping extrinsic traits such as extinction risks or modelled bioclimatic niches on phylogenies: does it make sense at all? Cladistics 27, 181–185 (2011).
21. Ricklefs, R.E. Estimating diversification rates from phylogenetic information. TRENDS in Ecology and Evolution 22, 601-610, 2007 .
22. Pagel, M. Inferring the historical patterns of biological evolution. Nature 401, 884-887, 1999.
23. Wiens, J.J. Estimating rates and patterns of morphological evolution from phylogenies: lessons in limb lability from Australian Lerisa lizards. Journal of Biology, 8:19, 2009.
24. Uyeda, J.C.; Hansen, T.F.; Arnold, S. J. and Pienaar, J. The million-year wait for macroevolutionary bursts. PNAS 108(38) 15908-15913, 2011.
25. Cardillo, M.; Mace, G.M.; Jones, K.E.; Bielby, J.; Bininda-Emonds, O.; Sechrest, W.; David, C.; Orme, L. and Purvis, A. Multiple Causes of High Extinction Risk in Large Mammal Species. Science 309, 1239-1241, 2005.
26. GATES, R.D.; BAGHDASARIAN, G. and MUSCATINE, L. Temperature stress causes host cell detachment in symbiotic cnidarians: implications for coral bleaching. Biol. Bull. 182:324-332, 1992.


Ana Paula Vieira (Masters student, UFRJ-MN)
Christopher Graves (PhD candidate, Brown University, USA)
Alex Hubbe (PhD candidate, IBUSP-Genética, Brazil)
Hannah E. Marx (PhD candidate, University of Idaho, USA)
Mariana Nery (PhD candidate, Universidad Austral de Chile, Chile)
Jennifer M. Schmidt (PhD candidate, Friedrch-Schiller-Universität Jena, Germany)
Mauro T. C. Sugawara (Undergraduate student, IBUSP-Ecologia, Brazil)
Alejandra Vasco (Post-Doc, The New York Botanical Gardens, USA)

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License