Comment Relation of Minimum Viable Population Size to Biology, Time Frame, and Objective

J. MICHAEL REED∗ § AND EARL D. MCCOY† ∗Department of Biology, Tufts University, Medford, MA, 02155, U.S.A. †Department of Integrative Biology, University of South Florida, Tampa, FL, 33620, U.S.A.

Shoemaker et al.’s (2013) estimate of the minimum vi- probability of extinction escalates dramatically over 40 able population size (MVPS) of the bog turtle (Glyptemys generations (Fig. 1). Given the long generation length, muhlenbergii) is 3 orders of magnitude below popu- 20–30 years, of the bog turtle, Shoemaker et al.’s result lation sizes typically viewed as minimally viable (Reed is, therefore, not surprising; but, it is inconsistent with et al. 2003; Trail et al. 2007). Even Flather et al. (2011), typical conception of MVPS, which assumes a population who question the utility of MVPS for conservation plan- will be self-sustaining. Similarly, Brook et al. (1999) con- ning, concede that long-term persistent populations will cluded a high probability of persistence for 50 years for require thousands of individuals. Rather than conclude a population of 18 Whooping Cranes (Grus americana). their result is incorrect, Shoemaker at al. suggest that Because Whooping Cranes can live for 35 years, it is MVPS has been overestimated for long-lived species. Al- unlikely that any pair of birds that can reproduce would though we agree in principle that small populations, es- fail to produce a lineage that persisted 50 years; but, 18 pecially of long-lived species, have conservation value, individuals do not necessarily constitute a viable popu- we do not agree that Shoemaker et al.’s result provides lation. Taking this reasoning to the absurd, the single evidence that small populations are viable. This position female Yangtze giant soft-shell turtle (Rafetus swinhoei) is inconsistent with existing empirical evidence of persis- living in the Changsha Zoo in Vietnam would constitute tence of small populations (Simberloff & Gibbons 2004; a viable population within a 50-year time frame because Fagan et al. 2005; Fagan & Holmes 2006). We suggest it has lived for more than 80 years. The definition of that Shoemaker et al.’s estimate is a gross underestimate, viability must make biological sense. emanating from their definition of viability,whichistoo The first published paper that formalized MVPS was by narrow, not biologically meaningful, and ignores factors Shaffer (1981), who pointed out that there is no single, such as demographic and environmental stochasticity, biologically driven definition of MVPS. Shaffer (1981) loss of genetic variability, and catastrophes. Their con- arbitrarily proposed 99% probability of persisting 1000 clusion that populations of long-lived species can be or- years as a standard to protect species from demographic, ders of magnitude smaller than currently believed does genetic, and environmental stochasticity, including catas- a disservice to species conservation and raises the long- trophes, but acknowledged that the probability and time term discussion in conservation biology about whether frame could be more or less stringent. Shaffer called for we should be focusing on the minimum for management. a discussion among resource managers to decide on rea- Shoemaker et al. used too short a time frame, 100 sonable practical criteria. Thus, MVPS estimates should years. Frankham and Brook (2004) and O’Grady et al. use a probability of persistence and time frame set a pri- (2008) identify the importance of setting the proper time ori to meet socially and biologically defined goals (Reed frame for conservation questions and point out that gen- et al. 2006). In an assessment of MVPSs of a suite of verte- eration length is often the appropriate scale. It can be brates, Reed et al. (2003) used a definition of 99% prob- shown easily that even a strongly declining population ability of persistence for 40 generations in an attempt to can be viable over the short time frame of 10 genera- standardize and thus facilitate MVPS comparisons with a tions (ࣙ100 years for long-lived species) and that the single time frame across taxa with different life histories.

§email [email protected] Paper submitted July 5, 2013; revised manuscript accepted September 21, 2013. 867 Conservation Biology, Volume 28, No. 3, 867–870 C 2014 Society for Conservation Biology DOI: 10.1111/cobi.12274 868 Minimum Viable Population Size

Figure 1. Mean population size over time of a population that is declining rapidly (e.g., due to habitat loss or overharvest) based on the results of an exponential population growth model with an additive stochastic ∗ component: xt+1 = xt + xt+1,wherext+1 = rt xt + δxt+1 and δxt+1 was drawn from a uniform, random distribution between −1and1(rt is −0.1, −0.15, and −0.2 to depict different degrees of severe population decline). Each simulation was run 75 times with an initial population size of 20 females. Standard deviations shown for r =−0.1 and −0.02. The probability of extinction, P(ext)t, for 750 runs of each model is shown for t = 10 generations and t = 40 generations.

Shoemaker et al., in contrast, evaluated a 90% probability a bog turtle population against environmental change or of persistence for 3–5 generations. Consequently, their catastrophe. results cannot be compared easily with studies that con- An alternative method for defining a viable population clude MVPSs are thousands of individuals (Trail et al. is to have a nondeclining population trajectory (Reed 2010). et al. 1998). Here, one must decide what size decline is Catastrophes and systematic environmental change important and detectable, and variance in population size were not considered in Shoemaker et al.’s estimate of dictates both how small a decline can be detected and MVPS. Consequences for the bog turtle of both could be how many years one can project into the future (Brook et severe. If, for example, the species exhibits temperature- al. 2006). This approach removes the need to set a time sensitive sex determination (unknown at present), then frame a priori, but it requires one to decide how small climate change could alter sex ratios, with potentially a decline one wants to be able to reject. It is important profound demographic repercussions. The potential ef- to keep in mind, however, that the time frame must be fects of quasi-predictable and unpredictable forces can long enough to determine whether the detected decline be included in population viability assessments and are is worrisome; if the decline is part of a long-term cycle, a well known to increase MVPS (Gerber & Hilborn 2001; short downturn in population size might not matter (Hill Wilcox & Elderd 2003). Although their study lasted 9 & Hagan 1991). years, this is only a fraction of a bog turtle’s generation Another method for estimating MVPS that does not length, so variance in demographic parameters will be require decisions about time frame is to estimate the underestimated, thus underestimating MVPS. For exam- population size necessary to maintain genetic variabil- ple, Reed et al. (2003) found that doubling the length of ity at the quantitative trait level. This is the 500 portion time a species was studied increased MVPS by 67%. An of the so-called 50/500 rule, referring to the effective alternative way to account for catastrophe is to require a size (Ne) of a closed population (Jamieson & Allendorf minimum range size larger than the area covered by the 2012). Because the ratio Ne/N for natural populations largest event that could affect a population (Reed 1992). tends to be very low for long-lived species (Frankham A population of 15 breeding females is not likely to buffer 1995), Shoemaker et al.’s 15 females would result in an

Conservation Biology Volume 28, No. 3, 2014 Reed & McCoy 869

Ne of 3–6. Consequently, 50% of the population’s genetic most populations are small, any estimate of MVPS can variability would be lost from a closed population in 4–8 be used to sacrifice some of these populations. If the 1 t generations (Ht = 0.5 from Ht = H0(1 − ) [Hedrick estimate is large, then resource managers may suggest 2Ne 2009]). Even if Ne = 15, inbreeding rate also changes as that populations below that size might not be worth pro- a function of 1/2Ne, so any genetic load would be rapidly tecting (alluded to by Clements et al. 2012). This is the expressed in a closed population. O’Grady et al. (2006), situation Shoemaker et al. may have been trying to avoid, for example, showed that ignoring inbreeding depression by pointing out that these small populations can serve overestimates survival prospects. For the Ne = 500 rule important short-term conservation goals. However, if the of thumb, population size for a closed population would estimate is small, then decision makers may suggest that have to be on the order of 2500–5000 if bog turtle Ne/N any population above that size could be reduced in size. ratio is similar to that found for many vertebrates. Al- though Shoemaker et al. acknowledge that they ignored evolutionary potential, they nevertheless stuck with their Acknowledgments conclusion about MVPS. Another possible method of estimating MVPS focuses We thank B. Noon and 2 anonymous reviewers for com- on the minimum area that a population needs to inhabit menting on earlier versions of this manuscript. in order escape environmental catastrophes (Reed 1992). Assuming some standard population density, multiplying by that area would give an estimate of MVPS. If population Literature Cited density changes with area in some standard manner— higher at small area and lower at larger area—a related Brook, B. W., J. R. Cannon, R. C. Lacy, C. Mirande, and R. Frankham. method identifies the area at which a relatively abrupt 1999. Comparison of the population viability analysis packages GAPPS, INMAT, RAMAS and VORTEX for the whooping crane (Grus change in density occurs, which could be multiplied by americana). Animal Conservation 2:23–31. population density to estimate MVPS (Smallwood 2002; Brook, B. W., L. W. Traill, and C. J. A. Bradshaw. 2006. Minimum viable McCoy & Mushinsky 2007). population sizes and global extinction risk are unrelated. Ecology Finally, one could estimate MVPS from the perspective Letters 9:375–382. of a species’ function in an ecosystem; that is, is a species Clements, G. R., C. J. A. Bradshaw, B. W. Brook, and W. F. Laurance. 2012. The SAFE index: using a threshold population target to mea- abundant enough to perform particular ecological func- sure relative species threat. Frontiers in Ecology and the Environ- tions? This definition was proposed by Conner (1988), ment 9:521–525. and ecological function has been used as an alternative Conner, R. N. 1988. Wildlife populations: Minimally viable or ecologi- to demographic metrics of viability (Estes et al. 2010; cally functional? Wildlife Society Bulletin 16:80–84. Galetti et al. 2013). Shoemaker et al.’s MVPS is unlikely Estes, J. A., M. T. Tinker, and J. L. Bodkin. 2010. Using ecological func- tion to develop recovery criteria for depleted species: sea otters to meet any desired ecosystem function bog turtles might and kelp forests in the Aleutian Archipelago. Conservation Biology fulfill. 24:852–860. No method can provide the perfect estimate of MVPS Fagan, W. F., and E. E. Holmes. 2006. Quantifying the extinction vortex. or minimum area needed to ensure a specific, reasonable Ecology Letters 9:51–60. chance of persistence for some period of time. Compari- Fagan, W. F., C. M. Kennedy, and P. J. Unmack. 2005. Quantifying rarity, losses, and risks for native fishes of the Lower Colorado River son of the results from several methods should, however, Basin: implications for conservation listing. Conservation Biology converge on some rough estimate: tens or hundreds or 19:1872–1882. thousands. We suspect that Shoemaker et al.’s (2013) Flather, C. H., G. D. Hayward, S. R. Beissinger, and P. A. Stephens. low estimate for bog turtles would not be similar to any 2011. Minimum viable populations: Is there a “magic number” for derived by alternative methods. Although we have sug- conservation practitioners? Trends in Ecology & Evolution 26:307– 316. gested that Shoemaker et al.’s MVPS estimate is too low Frankham, R. 1995. Effective population size/adult population size ra- to promote long-term self-sustaining populations, we do tios in wildlife: a review. Genetics Research 66:95–107. wish to reinforce their point that small populations have Frankham, R., and B. W. Brook. 2004. The importance of time scale in potential value. Small populations are of value as parts conservation biology and ecology. Annales Zoolici Fennici 41:459– of metapopulations, as temporary refugia, as stepping 463. Galetti, M., et al. 2013. Functional extinction of birds drives rapid evo- stones between larger populations, and as genetic reser- lutionary changes in seed size. Science 340:1086–1090. voirs (Mushinsky et al. 1997). Gerber, L. R., and R. Hilborn. 2001. Catastrophic events and recovery Although Shoemaker et al.’s goal, to support protec- from low densities in populations of otariids: implications for the tion of isolated, small populations, as well as large pop- risk of extinction. Mammal Review 31:131–150. ulations, is laudable, we are concerned that it will con- Hedrick, P. W. 2009. Genetics of populations. 4th edition. Jones & Bartlett Publishers, Sudbury, MA. tribute to apparent attempts to undermine MVPS criteria Hill, N. P. and J. M. Hagan. 1991. Trends of some northeastern North for endangered species protection (Walter 1990; Wilhere American landbirds: a half-century of data. Wilson Bulletin 103:165– 2008). When a landscape has been fragmented, such that 182.

Conservation Biology Volume 28, No. 3, 2014 870 Minimum Viable Population Size

Jamieson, I. G., and F. W. Allendorf. 2012. How does the 50/500 at thirty: conserving biodiversity in human-dominated landscapes. rule apply to MVPs? Trends in Ecology & Evolution 27:578– Island Press, Washington, D.C. 584. Reed, J. M., C. S. Elphick, and L. W. Oring. 1998. Life-history and viability McCoy, E. D., and H. R. Mushinsky. 2007. Estimates of minimum patch analysis of the endangered Hawaiian stilt. Biological Conservation size depend on method of estimation and the condition of the habi- 84:35–45. tat. Ecology 88:1401–1407. Shaffer, M. L. 1981. Minimum viable populations for species conserva- Mushinsky, H. R., E. D. McCoy, and D. S. Wilson. 1997. Patterns of tion. BioScience 31:131–134. gopher tortoise demography in Florida. Pages 252–258 in J. van Shoemaker, K. T., A. R. Breisch, J. W. Jaycox, and J. P. Gibbs. 2013. Abbema, editor. Proceedings of International Conference on Con- Reexamining the minimum viable population concept for long-lived servation, Restoration, and Management of Tortoises and Turtles. species. Conservation Biology 27:443–452. Turtle and Tortoise Society, New York. Simberloff, D., and L. Gibbons. 2004. Now you see them, now you O’Grady, J. J., B. W. Brook, D. H. Reed, J. D. Ballou, D. W. Tonkyn, and R. don’t!—population crashes of established introduced species. Bio- Frankham. 2006. Realistic levels of inbreeding depression strongly logical Invasions 6:161–172. affect extinction risk in wild populations. Biological Conservation Smallwood, K. S. 2002. Habitat models based on numerical compar- 133:42–51. isons. Pages 83–95 in J. M. Scott, P. J. Heglund, M. L. Morrison, J. O’Grady, J. J., D. H. Reed, B. W. Brook, and R. Frankham. 2008. B. Haufler, M. G. Raphael, W. A. Wall, and F. B. Samson, editors. Extinction risk scales better to generations than to years. Animal Predicting species occurrences. Island Press, Washington, D.C. Conservation 11:442–451. Trail, L. W., C. J. A. Bradshaw, and B. W. Brook. 2007. Minimum viable Reed, D. H., J. J. O’Grady, B. W. Brook, J. D. Ballou, and R. Frankham. population size: a metaanalysis of 30 years of published estimates. 2003. Estimates of minimum viable population sizes for vertebrates Biological Conservation 139:159–166. and factors influencing those estimates. Biological Conservation Trail, L. W., B. W. Brook, R. R. Frankham, and C. J. A. Bradshaw. 2010. 113:23–34. Pragmatic population viability targets in a rapidly changing world. Reed, J. M. 1992. A system for ranking conservation priorities for Biological Conservation 143:28–34. Neotropical migrant birds based on relative susceptibility to ex- Walter, H. S. 1990. Small viable population: the Red-tailed Hawk of tinction. Pages 524–536 in J. M. Hagan, III, and D. W. Johnston, Socorro Island. Conservation Biology 4:441–443. editors. Ecology and conservation of Neotropical migrant landbirds. Wilcox, C., and B. Elderd. 2003. The effect of density-dependent catas- Smithsonian Institution Press, Washington, D.C. trophes on population persistence time. Journal of Applied Ecology Reed, J. M., H. R. Akc¸akaya, M. Burgman, D. Bender, S. Beissinger, and 40:859–871. J. M. Scott. 2006. Critical habitat. Pages 164–177 in J. M. Scott, D. Wilhere, G. F. 2008. The how-much-is-enough myth. Conservation Bi- D. Goble, and F. W. Davis, editors. The Endangered Species Act ology 22:514–517.

Conservation Biology Volume 28, No. 3, 2014