A total of 200 listed terrestrial vertebrate species out of 240 had completed recovery plans [3]. We found 59% of completed recovery plans specifying a current population size and 14.5% specifying a variance for the population size estimate. Of the recovery plans that stipulated a population size as downlisting criteria, 24% reported variance for the current population size. Of the recovery plans that stipulated a population size for delisting criteria, 20% reported a variance for the current population size.

The percentage of recovery plans that provide a current population size did not differ significantly by decade (χ² = 2.10, p-value = 0.55), but the percentage of recovery plans that provided uncertainty or an estimate of variance for the current population size did (χ² = 16.04, p-value = 0.001) with the 2000s and 2010s reporting more variance (Figure 1). The percentage of recovery plans that provide a current population size was significantly different between terrestrial vertebrate Classes (χ² = 9.57, p-value = 0.02), and the percentage of recovery plans that provided uncertainty or an estimate of variance for the current population size also differed between animal Classes (χ² = 37.80, p-value < 0.01) (Figure 2). More recovery plans for the taxa Class Aves and Mammalia reported more estimates of current population size and variance for population size in comparison to Reptilia and Amphibia (Figure 2).

We found that 43% of the recovery plans stipulated a population size as a recovery criterion. The percentage of avian recovery plans (59%) that stipulated a population size as a recovery criterion was significantly higher compared to reptiles (38%), mammals (31%) and amphibians (24%) (χ² = 18.05, p-value < 0.01). Also, the percentage of the most recent recovery plans written in the 2010s (61%) that stipulated a population size as a recovery criterion was higher compared to the 2000s (41%), 1990s (42%) and 1980s (40%), however this difference was not significant (χ² = 6.57, p-value = 0.08).

Gerber & Hatch [4] found that recovery plans for endangered and threatened species have improved since 1998 because recovery criteria and quantitative data are more frequently included in recovery documents, although they found that the number of quantitative recovery criteria did not increase significantly for animals. We found that since the 1990s the number of recovery plans for terrestrial vertebrates have improved by incorporating more uncertainty into their estimate of population size (Figure 1). Also, in comparison to Gerber & Hatch [4], we found a higher percentage of recent recovery plans (2010s) for animals have included more quantitative recovery criteria (i.e., population size), albeit not significantly.

Overall, the ESA has been making a positive difference for species at risk of extinction [12] and we identified improvements in the reporting of quantitative data in recovery plans for terrestrial vertebrates. However, less than a quarter of listed terrestrial vertebrates (23%) recovery plans reported variance or uncertainty with their population size estimates. This may be the result of data limitations and inadequate resources for robust survey efforts [12]. Tear et al. [8] recommended that population estimates “should use a central tendency (mean, median, and mode) and/or the upper confidence interval to incorporate a greater margin of error and create more robust estimates for stipulated recovery criteria.” This requires the need for more rigorous survey efforts that incorporate a measure of uncertainty. For example, the recovery plan for the Marbled Murrelet (Brachyramphus marmoratus) recommended that the development of more robust consistent survey efforts be conducted before a current population size is defined [13].

Tear et al. [6,8] suggested that population size estimates should incorporate error or uncertainty when stipulating recovery criteria so species are not delisted prematurely. One approach we recommend for developing more detailed and robust recovery criteria when defining population size is by incorporating uncertainty into calculations of a minimum detectable difference between current population size and specified recovery criteria for delisting [14]. A minimum detectable difference (MDD) represents the smallest difference or change that would be statistically significant when comparing different samples depending on the variance of the samples and a defined level of uncertainty [14]. The equation below can be used to determine MDD (indicated with σ):

In this equation, n is the specified sample size (i.e., number of surveys used to estimate population size), s²_p is the variance in the population size estimates, and t_α,v and t_β(1),v are applied by using a one-tailed t-value at a significance level of say 0.10. The calculation of a MDD can then be modified to determine how much larger a recovery criteria for population size must be to ensure 90% confidence for recovery. If delisting were to be considered for a listed species, a biologist could calculate MDD, based on past survey data, and then add the MDD to the species’ recovery criteria. If the current population size is greater than the sum of MDD and the species’ recovery criteria then the biologist would be >90% confident the species had achieved the recovery criteria. This could provide biologists more confidence to downlist or delist a species based on best available science, rather than strong political pressure.

For example, in order for the Whooping crane (Grus Americana) to be considered for downlisting, one of the objectives is for three self-sustaining populations of cranes, of relatively similar size, to total 360 birds over 10 years [15]. Based on the Whooping crane recovery plan [15], the total annual population size in the last 10 years have been 219, 248, 244, 255, 256, 270, 292, 317, 335 and 343. This gives us an n = 10, s²_p = 1,756, a 2-tailed value = 1.83 and a one-tailed value = 1.37 for 90% confidence at nine degrees of freedom. This produces an MDD of 46.98 or 47. Thus, a crane biologist would be >90% confident the recovery criteria was achieved when the total crane population size was 407. However, this calculation would have to be revisited as the Whooping crane populations got closer to the actual recovery criteria.

In order to calculate a MDD for endangered species, robust survey efforts need to be either developed or re-visited for each species. Unfortunately for many species, there is a lack of resources and funding available to government agencies to obtain these data [12]. Many have recommended that species recovery be part of a dynamic adaptive management process where field biologists and other government personnel work more with academics and other research scientists to pool resources and obtain required quantitative data [4,12,16,17,18]. This would be especially true for reptiles and amphibians.

However, academics may have no incentive to help with the recovery process because faculty promotion may be based on attaining grant money and authoring numerous publications rather than being part of a recovery team. This may be the case for some large research based institutions, however there are numerous benefits academic faculty can take advantage of by being part of a recovery team. As well as professional development, faculty promotion can also be based on excellence in teaching and community service. Being part of a recovery team is an excellent form of community outreach and service, especially if faculty help lead local citizen science groups. For teaching excellence, faculty can get students involved in the recovery process as an applied teaching platform that involves active learning, service learning, project based learning and cooperative learning within their courses including opportunities for student internships and research. This type of hands on student learning can lead to better student evaluation scores [19], which can also be used in the faculty promotion process. Thus, if implemented properly, recovery planning can benefit academic faculty members.

Other recovery team members that are focused and represent diverse entities such as federal/state/local governments, non-government conservation organizations, local stakeholders, etc., have also been shown to improve the recovery plan process [12,16,17]. Non-academic based incentives that can help maintain recovery team diversity may include opportunities to secure congressional funding for activities involved in restoration programs, the mobilization of resources from different entities to apply for matching grants, technical assistance on stewardship activities, being part of a recovery process or a success story for marketing purposes, and/or the development of a government review process that rewards outreach and collaborative behaviors through awards and thank you letters [20].