Estimated dates of recent extinctions for North American and Hawaiian birds

https://doi.org/10.1016/j.biocon.2009.11.026Get rights and content

Abstract

Series of sighting records – the years in which a species has been recorded – can be used to infer whether species have gone extinct, and when extinctions occurred. We compiled sighting records for 52 rare bird species, subspecies, and distinct island populations from North America and Hawaii, 38 of which proved adequate for such analyses. Using a data set that combines non-controversial sight records with those for which physical evidence exists, no populations were judged likely to be extant, including those that have not been declared extinct. The ‘alalā was the only species with a 95% confidence interval around the extinction date that extended beyond 2009, suggesting that this population is the least unlikely to be extinct. Although all are probably extinct, populations were ranked according to their likelihood of persistence, so that any future searches can be prioritized to minimize the risk that resources are spent on extinct species. Estimated extinction dates spanned the period from the 1840s–2000s, with evidence for a peak in the early 1900s. On average, only about 4 years passed between a species’ last sighting and its estimated extinction date, and the 95% confidence intervals around extinction dates extended 9–26 years after the last sighting. Long gaps between sightings were very rare. Mean and median gap sizes between consecutive sightings within sighting records were 2.5 and 0 years, respectively. Gaps between the last and penultimate sightings were smaller than average gap sizes earlier in sighting records. Finally, a non-parametric method that can be calculated with more limited data proved a weak substitute for using more complete sighting records.

Introduction

Demonstrating beyond all doubt that a species is extinct is essentially impossible because there is always some chance, however small, that searches have been inadequate (Solow, 1993a, Reed, 1996). The rediscovery of species after decades in which they were not reported (Keith and Burgman, 2004, Butchart et al., 2006) and the continued discovery of species new to science, even in well studied groups such as birds (Peterson, 1998), illustrate the ease with which species can go undetected. Any judgment of extinction, therefore, is inherently probabilistic (McInerny et al., 2006).

Determining the point at which one should cease efforts to re-find species that are putatively extinct (“extaille”, sensuBanks, 1976) is a problem that has been long recognized (e.g., Hodge, 1911). Premature declarations of extinction are of great concern to conservation biologists, who do not want to mistakenly end protection efforts (Collar, 1998). On the other hand, continuing protection efforts after a species’ demise would divert important resources away from those species that continue to be in need of protection (Chadès et al., 2008). Identifying extinction dates also can provide insights into the extinction process (Caughley, 1994, Ferraz, 2003, Roberts and Solow, 2003). Consequently, decisions about whether a species has gone extinct have many repercussions (Diamond, 1987, Butchart et al., 2006), including how limited conservation funds are used; how decisions for taxa with different needs are balanced; how conservation assessments are made; and how research is prioritized.

Given the large number of putatively extinct species, and the difficulty and expense of conducting sufficiently thorough searches to ensure a high degree of confidence that a species truly is extinct (e.g., Scott et al., 1986, Scott et al., 2008, Reynolds and Snetsinger, 2001), Butchart et al. (2006) developed a framework for categorizing the level of confidence that a species no longer persists. This method considers evidence for and against extinction and the time since the species was last reliably reported, and uses information such as how well a decline has been documented, severity of known threats, search effort, and ease of detection. Quantifying components of this model would increase the ease with which species can be directly compared, which might facilitate difficult choices about resource allocation.

For example, quantitative methods for estimating extinction likelihood and predicted year of extinction from series of sighting data are available (e.g., Solow, 2005, Rivadeneira et al., 2009). These methods do not address all components of the Butchart et al. (2006) model, and should not be used in isolation when making conservation decisions. They do, however, provide a basis for quantifying one axis of investigation, while making explicit assumptions about some of the others (Roberts et al., in press). With these methods, the probability of detection and amount of search effort, for instance, are not assumed to be constant over the period of investigation (i.e., they can differ from year to year; Solow, 2005). Most of the methods assume only that there is no long-term trend in these variables (one exception is that of Solow, 1993b, which assumes a systematic decline in detectability as extinction is approached). The use of these methods, thus, provides one step towards a more quantitative assessment of whether and when species have gone extinct.

Most applications of these methods have focused either on illustrating new theoretical developments or have examined individual, high profile species (e.g., Roberts and Solow, 2003, Solow et al., 2006). Applying these methods systematically to quantify extinction patterns across suites of species is less common, probably because the necessary data sets are difficult to create (though see Burgman et al., 1995, McInerny et al., 2006, Patten et al., in press), but might have more practical value than isolated case studies. We have compiled temporal sequences of sighting records for all bird taxa (species, subspecies, or distinct island populations) from the United States and Canada that are presumed or suspected to have gone extinct during the past 200 years. We selected this set of populations because birds are generally well known, and because there has been considerable professional and amateur interest in finding rare species in this region for a long time. In addition, several species from this region have gone unseen for decades (North American Bird Conservation Initiative, 2009), but have not been declared extinct. Previously, we used a version of this data set to evaluate the underlying distributional assumptions of different methods for inferring the extinction parameters (Vogel et al., 2009). We concluded that, for our suite of species, a simple model that assumes a stationary Poisson process (Solow, 1993a) is more appropriate than the alternatives (e.g., Solow, 1993b, Solow, 2005, Roberts and Solow, 2003).

Here, we use the stationary Poisson method to estimate persistence probabilities and extinction dates for each population in order to determine which, if any, are likely to be extant and worthy of additional searches. Using the resulting information, we then examined the temporal patterns of North American and Hawaiian bird extinctions over the past two centuries, and tested whether there is evidence that the extinction rate has changed. Next, we determined the typical lag between the last sighting date and the estimated extinction date for a population, and between sequential sightings in extant populations, both of which provide insights into how long one should wait before inferring extinction. We also tested the hypothesis that as a species approaches extinction its inter-sighting interval changes. Finally, we examined whether a non-parametric alternative to the extinction model that we used, which can be calculated with only the two most recent sightings (Solow and Roberts, 2003), provides a good approximation of the results we obtained from our entire data sets.

Section snippets

Species selection and data compilation

We collated sighting information for 52 rare bird species, subspecies, and distinct island populations (see Supplementary material). We limited our evaluation to presumptive extinctions during the last two centuries because those prior to this time are unlikely to have a time series of good sighting records. We included a species/population if it has been designated as extinct by the US Fish and Wildlife Service (US Fish and Wildlife Service, 2006), is thought by many to be extinct, or has not

Results

Of the 52 populations for which we collated data, we had sufficient information to conduct analyses for 32 using the physical evidence sighting records, and 38 when independent expert opinion records were added. Here, we present results for the larger set of 38 populations, noting results based on just physical evidence sightings only when they differed qualitatively. Estimated extinction dates for the 38 populations ranged from 1846 to 2006, with the upper 95% bounds on these estimates ranging

Discussion

Based on our analyses, all of the populations studied are very likely extinct. The confidence interval around the estimated extinction date for the ‘alalā extends beyond the current time, suggesting that this is the population that is most likely to be extant. Even in this case, however, the possibility of persistence is not high, and if confirmed sightings are not forthcoming by 2014 it would be reasonable to conclude with a high level of confidence that the population is extinct. Of the

Conclusions

Sighting record models, coupled with standardized methods for accepting observations as valid, provide a method for systematically evaluating the likelihood that species persist and for prioritizing search efforts for potentially extinct species. Our results suggest that all of the populations that we studied are probably extinct, but we identify those species for which additional searches would most likely be worthwhile. In assessing extinction likelihood, more detailed analyses incorporating

Acknowledgements

We thank P. Pyle and R. Pyle for access to their data base on Hawaiian bird sightings, a preview of their monograph, and answering far more questions than it was reasonable for us to ask; G. Budney, P. Collins, A. Engilis, K. Garrett, J. Hill, J. Jeffrey, M. Medler, M. Patten, D. Pratt, T. Pratt, M. Scott, R. Shallenberger, and J. Woods for additional information on various species; L. Bevier and D. Sibley for many long discussions about these issues; the many museums that have contributed data

References (49)

  • R.C. Banks

    Extation

    Science

    (1976)
  • C.J. Bauer et al.

    Rediscovery of the Cherry-throated Tanager Nemosia rourei in southern Espirito Santo, Brazil

    Bird Conserv. Int.

    (2000)
  • BirdLife International, 2009. World Bird Database. BirdLife International, Cambridge....
  • M.A. Burgman et al.

    Inferring threat from scientific collections

    Conserv. Biol.

    (1995)
  • M.A. Burnham et al.

    Model Selection and Multimodel Inference: A Practical Information-theoretic Approach

    (2002)
  • S.H.M. Butchart et al.

    Going or gone: defining ‘Possibly Extinct’ species to give a truer picture of recent extinctions

    Bull. BOC

    (2006)
  • G. Caughley

    Directions in conservation biology

    J. Anim. Ecol.

    (1994)
  • I. Chadès et al.

    When to stop managing or surveying cryptic threatened species

    Proc. Natl. Acad. Sci.

    (2008)
  • N.J. Collar

    Extinction by assumption; or the Romeo error on Cebu

    Oryx

    (1998)
  • J.M. Diamond

    Extant unless proven extinct? Or, extinct unless proven extant?

    Conserv. Biol.

    (1987)
  • J.C. Eames et al.

    The rediscovery of Gurney’s Pitta Pitta gurneyi in Myanmar and an estimate of its population size based on remaining forest cover

    Bird Conserv. Int.

    (2005)
  • C.S. Elphick

    How you count counts: the importance of methods research in applied ecology

    J. Appl. Ecol.

    (2008)
  • G. Ferraz

    Rates of species loss from Amazonian forest fragments

    Proc. Natl. Acad. Sci.

    (2003)
  • J.W. Fitzpatrick et al.

    Ivory-billed woodpecker (Campephilus principalis) persists in continental North America

    Science

    (2005)
  • J.W. Fitzpatrick et al.

    Response to comment on “Ivory-billed woodpecker (Campephilus principalis) persists in continental North America”

    Science

    (2006)
  • P.M. Gorreson et al.

    Status and trends of native Hawaiian songbirds

  • P. Hahn

    Where is that Vanished Bird?

    (1963)
  • I.J. Harrison et al.

    The quiet crisis: a preliminary listing of the freshwater fishes of the world that are extinct or “missing in action”

  • C.F. Hodge

    The passenger pigeon investigation

    Auk

    (1911)
  • D.A. Keith et al.

    The Lazarus effect: can the dynamics of extinct species lists tell us anything about the status of biodiversity?

    Biol. Conserv.

    (2004)
  • J. Köhler et al.

    New amphibians and global conservation: a boost in species discoveries in a highly endangered vertebrate group

    BioScience

    (2005)
  • D.L. Leonard

    Recovery expenditures for birds listed under the US Endangered Species Act: the disparity between mainland and Hawaiian taxa

    Biol. Conserv.

    (2008)
  • G.J. McInerny et al.

    Significance of sighting rate in inferring extinction and threat

    Conserv. Biol.

    (2006)
  • K.S. McKelvey et al.

    Using anecdotal occurrence data for rare or elusive species: the illusion of reality and a call for evidentiary standards

    BioScience

    (2008)
  • Cited by (0)

    1

    Present address: Durrell Institute of Conservation and Ecology, School of Anthropology and Conservation, Marlowe Building, University of Kent, Canterbury, Kent CT2 7NR, UK.

    View full text