Article

# Spending limited resources on de-extinction could lead to net biodiversity loss

• Nature Ecology & Evolution 1, Article number: 0053 (2017)
• doi:10.1038/s41559-016-0053
Accepted:
Published online:

## Abstract

There is contentious debate surrounding the merits of de-extinction as a biodiversity conservation tool. Here, we use extant analogues to predict conservation actions for potential de-extinction candidate species from New Zealand and the Australian state of New South Wales, and use a prioritization protocol to predict the impacts of reintroducing and maintaining populations of these species on conservation of extant threatened species. Even using the optimistic assumptions that resurrection of species is externally sponsored, and that actions for resurrected species can share costs with extant analogue species, public funding for conservation of resurrected species would lead to fewer extant species that could be conserved, suggesting net biodiversity loss. If full costs of establishment and maintenance for resurrected species populations were publicly funded, there could be substantial sacrifices in extant species conservation. If conservation of resurrected species populations could be fully externally sponsored, there could be benefits to extant threatened species. However, such benefits would be outweighed by opportunity costs, assuming such discretionary money could directly fund conservation of extant species. Potential sacrifices in conservation of extant species should be a crucial consideration in deciding whether to invest in de-extinction or focus our efforts on extant species.

Technological advances are reducing the barriers to resurrecting extinct species or their close genetic proxies, allowing de-extinction to be considered as a biodiversity conservation tool 1,2 . Arguments in favour of de-extinction include necessity, driven by the rapid rate of species and habitat loss 3,4 , an ethical duty to redress past mistakes 5 , as well as potential technological and ecological knowledge that could stem from de-extinction programmes 4 . Counter-arguments include high risk of failure due to difficulties of cloning for some species 6 , technical risks inherent in re-introductions 7,​8,​9 , loss of culture in resurrected animal species 8 , and lack of remaining habitat for some species 10,11 , as well as negative consequences for extant species, including reduced incentive for traditional conservation 12 , and ecological impacts of introducing long-absent or genetically modified species 12 .

The relative cost versus benefit for biodiversity is fundamental to the debate surrounding de-extinction. Assuming species are resurrected to be released into former habitats, the cost of de-extinction includes the process of producing initial founder populations, translocating individuals, then monitoring and managing new wild populations. If conservation funds are re-directed from extant to resurrected species, there is risk of perverse outcomes whereby net biodiversity might decrease as a result of de-extinction 12,13 . Although private agencies might fund the resurrection of extinct species out of technical or philanthropic interest, the subsequent ongoing management of such species (many of which would face the same threats that made them extinct) would fall on government agencies, as commonly occurs with extant threatened species. Alternatively, if private agencies are willing to provide new funding for post de-extinction management, there could be additional benefits to species sharing habitats or threats.

Here, we test the potential impact of establishing and sustaining wild populations of resurrected extinct species (or proxies of such species) on the conservation of extant species. Specifically, we use long-term conservation programmes for extant analogue species in New Zealand (NZ) and the Australian state of New South Wales (NSW), to infer potential conservation actions for resurrected species, and predict the impact of resurrected species programmes on conservation of extant species. We use these datasets because they contain detailed prescriptions and costs of actions designed to achieve population recovery for most of the extant threatened species requiring specific management actions in either jurisdiction. We estimate the net number of extant species that can be conserved, using the following scenarios: (1) establishment and maintenance of resurrected species become the burden of government conservation programmes, and (2) establishment and maintenance of resurrected species populations are funded externally using non-public resources. In Scenario 1, the use of government resources on resurrected species results in less funding for extant species programmes, but provides potential benefits for species that share actions with resurrected species. In Scenario 2, there are also potential benefits to extant species conservation programmes through shared conservation actions. However, there are potential opportunity costs, if private agencies use resources they could otherwise have used on conservation of extant species. Our analysis assumes that species would be resurrected to be re-introduced into their former habitats, rather than for other potential reasons, such as research or public display.

Because little is known about the costs of producing viable initial populations of resurrected species, we do not consider this in our analysis, and assume it is covered by a private agency. Instead, we focus on the long-term cost of conservation for resurrected species, assuming that such species would have small founder populations that require conservation actions similar to those required for extant threatened species.

Although there is considerable uncertainty regarding the necessary conservation actions for many extinct species should they be resurrected 7 , we assume that such species would share many actions with closely related extant species that share habitats, threats and ecological roles. Therefore, from among the endemic, fully extinct species of our study areas, we chose focal extinct species whose taxonomy, range, habitat, life history and threats were similar to an extant threatened analogue species. Among 70 recently extinct (AD 1000 to present) species in NZ, we found 11 for which we could assign reasonable analogues (Supplementary Table 1). For NSW, we considered 29 recently extinct species, and found 5 with reasonable extant analogues. Our inferred conservation programmes for the extinct species (assuming they were resurrected), were the same as for their analogue extant species, with the addition of captive breeding and translocation costs, based on average costs of captive breeding and translocation from extant species of the same taxonomic group (for example, bird, amphibian). Although cost and shared actions were not criteria for choosing our focal species, our chosen group represented a broad range in terms of estimated costs and the number of extant species with shared actions (Supplementary Table 1). Note that using analogues in this way probably underestimates the cost, along with the risk of failure, and that we have assumed that actions could be completely shared between resurrected and extant analogue species. It is unlikely that an effective conservation programme for a resurrected species would completely share actions with that of an extant species. Our analysis also includes the largely untested assumption that technical barriers to creating initial populations of these species can be overcome 6,14 . Thus, our results should be regarded as optimistic in favour of the net benefits of resurrected-species conservation programmes.

To assess the potential influence of resurrected species on extant species conservation, we incorporated the proposed programmes for resurrected species into threatened species project prioritization protocols developed for the New Zealand Department of Conservation and the NSW Office of Environment and Heritage (see ‘Methods’ for details). Costs of shared actions (for example, predator control that benefits several species sharing a site) were shared among prioritized species recovery projects. Thus, if private funding covers the cost of actions for a resurrected species, the cost of the same actions for any other species (including the resurrected species’ analogue) would also be covered, potentially allowing more species to be conserved within a given budget.

In Scenario 1, where resurrected species become the burden of governments, we subtracted the budgets for resurrected species conservation programmes from realistic baseline budgets for NZ (NZD\$30 million 15,16 ) and NSW (AUD\$4.65 million 17 ), and set the cost of any specific actions that were shared in location and time with other species (for example, predator fence on a shared habitat patch) to zero. We compared the number of extant species that could be prioritized for funding in this scenario with the number of extant species that would normally be prioritized with the same baseline budget. We did this for each resurrected species considered individually (cost of only one focal species subtracted), as well as all resurrected species (11 for NZ and 5 for NSW) considered together.

For Scenario 2, where resurrected species programmes are entirely externally sponsored, we determined potential benefits for conservation of extant species by setting the cost of any shared actions between resurrected and extant species to zero, and re-ran the prioritization algorithms with our baseline budgets. To determine the opportunity cost associated with this scenario, we added the cost of actions for resurrected species’ sponsorship programmes to the baseline budget for extant species prioritization, then determined the number of species that would be prioritized for conservation if such funding could be used on extant instead of resurrected species.

For both NZ and NSW species, the number of extant species that could be prioritized for conservation was generally lower in Scenario 1, where resurrected species become the burden of the government (Fig. 1, red bars). This suggests a potential long-term net loss of biodiversity if conservation efforts are shifted towards resurrected species. For NZ, there were potential net gains associated with a single resurrected species, Coenocorypha chathamica. This is because the conservation prescription for this species contained many shared actions with 39 extant species that inhabit its former habitat on Chatham Island. Shared costs for some of these species allowed more to be prioritized than in the baseline scenario. However, for NSW the estimated conservation costs for two extinct species are greater than the most recent baseline budget estimate, suggesting that the government budget would have to be drastically increased if conservation of either species were publically funded. Given that the NZ and NSW algorithms are designed to efficiently conserve species using limited resources, and to account for shared costs, it is possible that the impact of government payment for resurrected species programmes could be greater in other jurisdictions where spending is less efficient.

In Scenario 2, where the conservation costs of resurrected species are covered by an external agency, lowered costs of shared actions would allow more extant species to be prioritized for conservation (Fig. 1, yellow bars). However, the potential biodiversity benefits are outweighed by the opportunity costs of not applying the same funding to extant species (Fig. 1, blue bars).

For both scenarios, including the costs of all resurrected species together amplified the results (Supplementary Table 2). For example, government-funded conservation for all 11 focal extinct species in NZ would sacrifice conservation for nearly three times as many (31) extant species. External funding for conservation of the five focal extinct NSW species could instead be used to conserve over eight times as many (42) extant species.

It is probable that including the costs of producing viable initial populations of these species would greatly increase our estimates for the sacrifices in extant species conservation. The costs for such programmes are difficult to project, but they are likely to be substantial. A programme aimed at using stem cell technology and surrogates to prevent the extinction of the northern white rhinoceros (which would have fewer technical hurdles than resurrecting species from only preserved materials) has been estimated to cost several million dollars 2 . In addition, our analyses make the generous assumption that we would have perfect analogues for the extinct species, such that conservation programmes for resurrected and extant species would share costs. Breeding and husbandry of resurrected species before and after reintroduction could well be more expensive and more prone to failure than for extant species, because we typically know less about the behaviour and physiology of extinct species. Reintroduction of locally extirpated species would likely have considerably lower risk and costs, given better knowledge of the ecology and physiology of such species.

Debates regarding the merits of de-extinction tend to centre on either ethical or biological arguments. Ethical arguments often focus on the potential of de-extinction to right past wrongs, versus the ‘moral hazard’ arising from diminished motivation to conserve extant species, if it is assumed that extinctions can be reversed sometime in the (potentially distant) future 12 . Biological arguments often focus on the relative benefit to biodiversity. For example, there may be conservation gains through applying technical lessons learned in the process of attempting de-extinctions 4 . There could also be gains through restoration of ecosystem processes that were provided by the extinct species 18 . For example, extinct ‘ecosystem engineers’, such as woolly mammoths or passenger pigeons, could potentially be resurrected in attempts to restore their lost functional roles 19 . In addition, resurrected species could act as ‘flagships’ to promote conservation 5 , and potentially increase resources for management of extant threatened species.

However, there is considerable risk in assuming that resurrected species would fill these intended roles. Resurrected ecosystem engineers would be introduced into environments that have been much altered by humans, and they could fail to thrive in these new circumstances 7,19 . Resurrecting populations large enough for such species to fill their former roles could also prove very challenging 19 . Conversely, there may be biodiversity losses if resurrected species become invasive or spread disease 12,20,21 . Experience with extant iconic species also suggests a high risk that iconic species resurrected as ‘flagships’ could draw resources away from programmes for extant species 22 , or even create self-reinforcing biases whereby the public profile of resurrected species and resources spent on them would synergistically increase, at the expense of non-iconic extant species 23 .

More fundamentally, de-extinction could lead either to biodiversity gains via resurrection of the extinct species themselves and shared conservation actions with extant species, or to losses through missed opportunities to allocate resources to extant species. Conservation resources are scarce 24 , necessitating careful allocation of funds 25 . Our analysis strongly suggests that resources expended on long-term conservation of resurrected species could easily lead to net biodiversity loss, compared with spending the same resources on extant species. If the risk of failure and the costs associated with establishing viable populations could also be calculated, estimates of potential net losses or missed opportunities would probably be considerably higher. Given this considerable potential for missed opportunity, as well as the risks inherent in assuming a resurrected species would fulfil its role as an ecosystem engineer or flagship species, it is unlikely that de-extinction could be justified on grounds of biodiversity conservation.

## Methods

Focal extinct species were chosen based on taxonomic relatedness as well as having ranges, habitats, threats and life-history strategies shared with extant analogue species. We chose fully extinct species (not just locally extirpated) that went extinct after AD 1000, assuming that feasibility of resurrection (for example, availability of genetic material, knowledge of life history and physiology) would be prohibitively low for species that went extinct before this time. However, we did not assess the availability of genetic material in our focal species, nor consider the feasibility or cost of producing viable initial populations.

### Species conservation projects and prioritization algorithms

Species conservation projects for all species in the NZ and NSW datasets (including the analogue species) were determined using information gathered from threatened species experts (>100 experts for NZ, ~250 for NSW). The projects include the specific actions (including specific location, timing and cost) considered necessary to ensure ~95% probability of each species’ persistence over 50 years.

The prioritization algorithms rank species by the cost-effectiveness of their conservation projects, using the following equation: $E i = B i × S i C i$ where E i is the cost effectiveness of the conservation project for species i; B i is the benefit of the project to the species, defined as the difference between estimated probabilities that a species will be secure in 50 years with and without the project; S i is the estimated probability of success for the conservation project; and C i is the total cost of all actions for the project, across all sites. The NZ algorithm uses an additional parameter that estimates a species’ evolutionary distinctiveness (see ref. 15 for details). Costs of actions are shared among prioritized species recovery projects. For example, the cost of predator control at a site that benefits two prioritized species sharing the site is reduced by 50% for each of the two species.

The algorithm begins with all species ranked, then eliminates the lowest-ranked species sequentially until the set first-year budget is reached. As species are removed from the ranks, cost sharing is updated for the remaining species. Species that are no longer prioritized no longer share costs with those that remain. Additional details regarding the algorithm are given in refs 15,26 .

### Data availability

The data and code for the NZ and NSW prioritization protocols have been deposited in the Dryad Digital Repository at http://dx.doi.org/10.5061/dryad.3qn55 and http://dx.doi.org/10.5061/dryad.p86t5, respectively.

How to cite this article: Bennett, J. R. et al. Spending limited resources on de-extinction could lead to net biodiversity loss. Nat. Ecol. Evol. 1, 0053 (2017).

## References

1. 1.

et al. First birth of an animal from an extinct subspecies (Capra pyrenaica pyrenaica) by cloning. Theriogenology 71, 1026–1034 (2009).

2. 2.

Stem-cell plan aims to bring rhino back from brink of extinction. Nature 533, 20–21 (2016).

3. 3.

Extinction need not be forever. Nature 492, 9 (2012).

4. 4.

& What if extinction is not forever? Science 340, 32–33 (2013).

5. 5.

The ethics of de-extinction. NanoEthics 8, 165–178 (2014).

6. 6.

Mammoth 2.0: Will genome engineering resurrect extinct species? Genome Biol. 16, 228 (2015).

7. 7.

, & Reintroducing resurrected species: selecting DeExtinction candidates. Trends Ecol. Evol. 29, 140–147 (2014).

8. 8.

From dinosaurs to dodos: Who could and should we de-extinct? Front. Biogeog. 6, 20–24 (2014).

9. 9.

et al. Assessing the benefits and risks of translocations in changing environments: a genetic perspective. Evol. Appl. 4, 709–725 (2011).

10. 10.

, , & US protected lands mismatch biodiversity priorities. Proc. Natl Acad. Sci. USA 112, 5081–5086 (2015).

11. 11.

et al. De-extinction potential under climate change: extensive mismatch between historic and future habitat suitability for three candidate birds. Biol. Cons. 197, 164–170 (2016).

12. 12.

IUCN/SSC Guiding Principles on Creating Proxies of Extinct Species for Conservation Benefit: Version 1.0 (International Union for Conservation of Nature, 2016).

13. 13.

Going the way of the dodo: de-extinction, dualisms, and reframing conservation. Wash. Univ. Law Rev. 92, 849–906 (2015).

14. 14.

Pathways to de-extinction: How close can we get to resurrection of an extinct species? Funct. Ecol. (2016).

15. 15.

et al. Balancing phylogenetic diversity and species numbers in conservation prioritization, using a case study of threatened species in New Zealand. Biol. Cons. 174, 47–54 (2014).

16. 16.

et al. Effect of risk aversion on prioritizing conservation projects. Cons. Biol. 29, 513–524 (2015).

17. 17.

More Plants and Animals to be Saved from Extinction: Saving Our Species 2016–2021 (New South Wales Office of Environment and Heritiage, 2016).

18. 18.

et al. Resolving lost herbivore community structure using coprolites of four sympatric moa species (Aves: Dinornithiformes). Proc. Natl Acad. Sci. USA 110, 16910–16915 (2013).

19. 19.

, , & A mammoth undertaking: harnessing insight from functional ecology to shape de-extinction priority setting. Funct. Ecol. (2016).

20. 20.

Is it right to reverse extinction? Nature 509, 261 (2014).

21. 21.

, , & Rewilding is the new Pandora’s box in conservation. Curr. Biol. 26, R87–R91 (2016).

22. 22.

& Funding extinction? Biological needs and political Realities in the allocation of resources to endangered species recovery. BioSci. 52, 169–177 (2002).

23. 23.

, , & What drives policy decision-making related to species conservation? Biol. Cons. 142, 1370–1380 (2009).

24. 24.

et al. Financial costs of meeting global biodiversity conservation targets: current spending and unmet needs. Science 338, 946–949 (2012).

25. 25.

, & Setting conservation priorities. Ann. NY Acad. Sci. 1162, 237–264 (2009).

26. 26.

, & Optimal allocation of resources among threatened species: a project prioritization protocol. Cons. Biol. 23, 328–338 (2009).

## Acknowledgements

J.R.B. was supported the Natural Science and Engineering Research Council of Canada (NSERC) and the Australian Research Council (ARC) Centre of Excellence for Environmental Decisions (CEED). H.P.P. was funded by an ARC Laureate Fellowship and CEED.

## Affiliations

1. ### Department of Biology, Carleton University, 1125 Colonel By Drive, Ottawa, Ontario K1S 5B6, Canada

• Joseph R. Bennett
2. ### Science and Policy Group, Department of Conservation, 70 Moorhouse Avenue, Addington, Christchurch 8011, New Zealand

• Richard F. Maloney
3. ### School of Biological Sciences, University of Canterbury, Private Bag 4800, Christchurch 8140, New Zealand

• Tammy E. Steeves
4. ### New South Wales Office of Environment and Heritage, 59 Goulburn Street, Sydney, New South Wales 2000, Australia

• James Brazill-Boast
5. ### University of Queensland, ARC Centre of Excellence for Environmental Decisions, School of Biological Sciences, St Lucia, Queensland 4072, Australia

• Hugh P. Possingham
6. ### Conservation Science, The Nature Conservancy, 245 Riverside Drive, West End, Queensland 4101, Australia

• Hugh P. Possingham
7. ### University of Otago, Department of Zoology, 340 Great King Street, Dunedin 9016, New Zealand

• Philip J. Seddon

## Authors

### Contributions

J.R.B., R.F.M. and P.J.S. designed the study. J.R.B., R.F.M. and J.B.-B. analysed the data. J.R.B. wrote the paper, with input from all other authors.

### Competing interests

The authors declare no competing financial interests.

## Corresponding author

Correspondence to Joseph R. Bennett.

## PDF files

1. 1.

### Supplementary information

Supplementary Tables 1,2; Supplementary Discussion