Recommendations for Improving Recovery Criteria under the US Endangered Species Act Authors: Daniel F. Doak, Gina K. Himes Boor, Victoria J. Bakker, Williams F. Morris, Allison Louthan, Scott A. Morrison, Amanda Stanley, & Larry B. Crowder This is a pre-copy-edited, author-produced PDF of an article accepted for publication in BioScience following peer review. The version of record [Doak, Daniel F., Gina K. Himes Boor, Victoria J. Bakker, William F. Morris, Allison Louthan, Scott A. Morrison, Amanda Stanley, and Larry B. Crowder. "Recommendations for Improving Recovery Criteria under the US Endangered Species Act." BioScience 65, no. 2 (February 2015): 189-199] is available online at: https://dx.doi.org/10.1093/biosci/biu215. Doak, Daniel F., Gina K. Himes Boor, Victoria J. Bakker, William F. Morris, Allison Louthan, Scott A. Morrison, Amanda Stanley, and Larry B. Crowder. "Recommendations for Improving Recovery Criteria under the US Endangered Species Act." BioScience 65, no. 2 (February 2015): 189-199. DOI: https://dx.doi.org/10.1093/biosci/biu215. Made available through Montana State University’s ScholarWorks scholarworks.montana.edu 1 Recommendations for improving recovery criteria under the United States Endangered 2 Species Act 3 4 Doak, D.F.1 5 Environmental Studies Program 6 University of Colorado, Boulder 7 Attn. 215 8 Boulder, CO 80309 9 daniel.doak@colorado.edu 10 303-919-6231 11 12 Himes Boor, G.K. 13 Ecology Department 14 Montana State University 15 P.O. Box 173460 16 Bozeman, MT 59717 17 gkhimesboor@montana.edu 18 406-994-1821 19 20 Bakker, V.J. 21 Ecology Department 22 Montana State University 23 P.O. Box 173460 24 Bozeman, MT 59717 25 vjbakker@gmail.com 26 406-589-4399 27 28 Morris, W.F. 29 Department of Ecology and Genetics 30 Uppsala University 31 Norbyvägen 18D 32 75236 Uppsala 33 Sweden 34 and 35 Biology Department 36 Duke University 37 Box 90338 38 Durham, NC 27708 39 bill.morris@ebc.uu.se 40 Phone: 011-46-18-471-2861 41 42 1 Authors are listed in order of their contributions to the paper 1 Louthan, A. 43 Environmental Studies Program 44 University of Colorado, Boulder 45 Attn. 215 46 Boulder, CO 80309 47 allisonmlouthan@gmail.com 48 (307) 766 3493 49 50 Morrison, S.A. 51 The Nature Conservancy 52 201 Mission St., 4th Floor, San Francisco, CA 94105 USA. 53 smorrison@tnc.org 54 415 963 6603 55 56 Stanley, A. 57 Wilburforce Foundation 58 2034 NW 56th St, Ste 300 Seattle, WA 98107-3127 59 Amanda@wilburforce.org 60 Office: 206.632.2325 x114 61 62 Crowder, L. 63 Center for Ocean Solutions 64 99 Pacific Street, Suite 555E 65 Monterey, CA 93940 66 Larry.Crowder@stanford.edu 67 831 333 2099 68 69 2 Abstract 70 Recovery criteria, the thresholds mandated by the Endangered Species Act that define when 71 species may be considered for downlisting or removal from the endangered species list, are a key 72 component of conservation planning in the U.S. We recommend improvements in the definition 73 and scientific justification of recovery criteria, addressing both data-rich and data-poor 74 situations. We emphasize the distinction between recovery actions and recovery criteria, and 75 recommend the use of quantitative population analyses to measure impacts of threats and to 76 explicitly tie recovery criteria to population status. To this end we provide a brief tutorial on the 77 legal and practical requirements and constraints of recovery criteria development. We conclude 78 by contrasting our recommendations with other alternatives, and describing ways that academic 79 scientists can contribute productively to the planning process and to endangered species 80 recovery. 81 Introduction 82 Over the past 20 years, ecologists and conservation biologists have conducted multiple 83 reviews of the United States Endangered Species Act (ESA) focused on legal, policy, and 84 especially scientific elements of the Act’s implementation (e.g. Boersma et al. 2001, Foin et al. 85 1998, Gerber and Hatch 2002, Gibbs and Currie 2012, Hoekstra et al. 2002, Lawler et al. 2002, 86 Morris et al. 2002, Moyle et al. 2003, Scott et al. 2005, Tear et al. 1993, 1995). These reviews 87 have found numerous shortcomings in the effectiveness and scientific basis of recovery plans 88 and recovery criteria and have suggested just as many remedies. In response to these academic 89 reviews and to court decisions interpreting the ESA, the two government agencies that 90 implement the Act (US Fish and Wildlife Service [USFWS] and National Marine Fisheries 91 3 Service [NMFS] – henceforth the “Services”) have continued to update their procedures for 92 recovery planning (NMFS and USFWS 2010). 93 Despite these efforts, recent reviews of the ESA’s implementation have still found little 94 improvement in key metrics of scientific rigor, including the clear articulation and biological 95 justification of recovery criteria (Himes Boor 2014, Neel et al. 2012). This situation prompted us 96 to convene a workshop to find pragmatic ways to improve this central part of ESA recovery 97 planning. To increase the odds that our recommendations would have traction, we sought to 98 understand the viewpoints of representatives from many parts of the conservation community 99 and to focus on one key element of the ESA – recovery criteria and their use – rather than 100 conducting a general critique of the Act or its implementation. 101 We focus on recovery criteria for three reasons. First, they specify the conditions under 102 which a species may be considered for downlisting (being moved from endangered to threatened 103 status) or delisting (removing from ESA protection), thereby defining what characteristics the 104 Services expect a population to exhibit once it reaches a state of recovery. Criteria thus serve as a 105 structuring element for a recovery plan as a whole and guide the actions of government agencies 106 and other entities. Second, the ESA stipulates that recovery criteria be “measurable and 107 objective” and that delisting decisions be based on “the best scientific and commercial data 108 available” (16 U.S.C. §§ 1533); both requirements inject a primary role for science, although 109 exactly how recovery standards are to be defined or supported is left unclear. Finally, a vast 110 amount has been written about assessing extinction risk, establishing targets for healthy 111 populations in the face of harvest and habitat loss, analyzing the consequences of population size 112 and connectivity for inbreeding, and other topics directly relevant to setting recovery thresholds. 113 4 Thus, recovery criteria appeared to be a relatively tractable target for improving the scientific 114 implementation of the ESA. 115 While we see a critical role for science in setting recovery criteria, defining what 116 “recovery” should mean for a population or species involves more than scientific analysis. In 117 particular, the risk of partial or complete failure (i.e., extinction) we as a society are willing to 118 accept and the degree to which we try to restore species to former numbers, distributions, and 119 ecological functions blend into matters legal and ethical. These decisions are often made in part 120 by biologists, but we emphasize that they are not objective biological decisions, and that they 121 require careful attention (Box A). 122 We begin with a brief tutorial on recovery planning, emphasizing the development of 123 criteria. Even though all of us have read or reviewed numerous plans, served on recovery teams, 124 or both, we nonetheless did not appreciate the practical constraints that several key legal and 125 administrative rulings impose on how recovery plans must be written. Given our advocacy of 126 increased involvement of academics in recovery planning, this description of “everything you 127 (should have) always wanted to know about recovery planning, but were too ignorant to ask” is 128 especially germane. 129 Legal and policy context 130 Recovery plans describe the biology of the species and its threats, develop a strategy for 131 attaining recovery, outline actions needed to carry out the strategy, and detail the criteria by 132 which attainment of recovery (Table 1) can be assessed. While a bevy of requirements and 133 recommendations shape how recovery criteria are developed (NMFS and USFWS 2010), a 134 handful of rules and legal decisions are also of key importance. The only explicit guidelines in 135 the ESA regarding recovery criteria and actions are that recovery plans must “to the maximum 136 5 extent practicable,” contain “objective, measurable criteria which, when met, would result in a 137 determination, in accordance with the provisions [of the ESA], that the species be removed from 138 the list,” and “a description of such site-specific management actions as may be necessary to 139 achieve the plan’s goal for the conservation and survival of the species” 16 U.S.C. § 140 1533(f)(1)(B). The ESA definition of endangered (“in danger of extinction throughout all or a 141 significant portion of its range”) highlights the role of extinction risk and spatial distribution in 142 defining recovery but otherwise provides little guidance for recovery criteria, and in fact injects 143 additional need for policy clarification for undefined terms such as “in danger of” and 144 “significant portion of its range” (Carroll et al. 2010, Vucetich et al. 2006). The Services’ 145 Recovery Planning Guidance (NMFS and USFWS 2010), intended to provide more explicit 146 guidelines for recovery planning and to outline policy directives, indicates that they do not 147 consider the measureable and objective requirement to mean that criteria must be quantitative 148 (Section 5.1.8.3). The Guidance document defines recovery actions to be all activities “necessary 149 to achieve full recovery of the species” as well as “the monitoring actions necessary to track the 150 effectiveness of these actions and the status of the species” (NMFS and USFWS 2010). 151 One aspect of the Services’ approach to recovery criteria stems from the ESA 152 requirement that prior to listing the Services must conduct a formal review to assess the extent to 153 which the species is affected by five specific “threat factors”: A) Destruction, modification, or 154 curtailment of habitat or range; B) Overutilization; C) Disease or predation; D) Inadequacy of 155 existing regulation; and, E) Any other natural or manmade factors. A species can only be 156 removed from the list when none of the five factors threatens or endangers it. The courts have 157 ruled that recovery criteria must address all five threat factors, and measure whether they have 158 been ameliorated (Fund for Animals v. Babbitt: 903 F. Supp. 96 (D.D.C 1995)). The Services 159 6 interpret this ruling literally and recommend that plan writers formulate separate recovery criteria 160 targeted at each threat factor (GAO 2006, NMFS and USFWS 2010). The Services also suggest 161 that demographic criteria (which we use in the sense of any estimates of population status: i.e., 162 population size, trends through time, demographic rates, genetic factors, spatial distribution, or 163 population viability indices) be listed separately from “threat-based” criteria (NMFS and 164 USFWS 2010). 165 A final aspect of real-world recovery planning worth highlighting is that relatively few 166 plans are written by recovery teams of agency and non-agency experts. About half are written by 167 only one or a few agency personnel or contractors (D. Crouse, USFWS, pers. comm.). This 168 limited authorship demonstrates that resources (expertise, time, and money) for writing recovery 169 plans are even more restricted than is widely recognized. 170 Current approaches to defining recovery criteria 171 How do these requirements and constraints affect the formulation of recovery criteria? 172 Even very recent plans differ greatly in the number, range, format, quantity, and degree of 173 specificity of their recovery criteria (see Appendix A for examples of criteria from different 174 plans, including many of those referred to in this section). For example, some plans contain only 175 demographic criteria, such as the short-tailed albatross (Phoebastria albatrus) plan, whose sole 176 delisting criterion stipulates requirements for population size, growth rate, and spatial 177 distribution of the population. 178 However, most recent plans also, or primarily, use threat-based criteria that specify 179 control or reduction of threats. The level of threat reduction required can vary in specificity and 180 may or may not be linked explicitly to demography or viability. For example, one delisting 181 criterion for the Vermillion darter (Etheostoma chermockz) requires the attainment of very 182 7 specific water quality standards for turbidity over 10 consecutive years under a specified 183 sampling regime. In contrast, the Sei whale (Balaenoptera borealis) threat-based recovery 184 criteria are more general, requiring that each threat identified in the plan, such as reduced prey 185 abundance due to climate change, anthropogenic noise, ship collisions, and gear entanglement, 186 continue “to be investigated and any necessary actions being taken to address the issue are 187 shown to be effective or this is no longer believed to be a threat.” 188 Some threat-based criteria essentially consist of actions, including administrative or 189 monitoring directives focused on specific threats. For example downlisting criteria for the 190 smalltooth sawfish (Pristis pectinata) stipulate that public education programs about the species 191 and the prohibitions against harming it be in place. Similarly, delisting criteria for the Kemps 192 Ridley sea turtle (Lepidochelys kempii) include establishment of a network of monitoring sites. 193 Occasionally, threats are accounted for by weighing their impacts on demographic 194 processes. For example, delisting criteria for the Gila trout (Oncorhynchus gilae) focus solely on 195 the number of populations and occupied streams because these metrics were determined by 196 quantitative analysis to best demonstrate resilience to the effects of catastrophic fires, the 197 primary proximal threat to the species. More generally, the Gulf Coast jaguarundi (Puma 198 yagouaroundi cacomitli) plan calls for habitat loss, degradation, and fragmentation to be reduced 199 to the point that the species is no longer in danger of extinction. Similarly, the Wyoming toad 200 (Anaxyrus baxteri) plan calls for chytridiomycosis infections rates to be maintained at levels that 201 ensure long-term sustainability of the population. 202 Other demographic criteria take the form of “viability criteria” that are either direct 203 measures of a population’s risk of extinction or quasi-extinction (e.g., 5% risk of extinction 204 within 100 years) or demographic measures (e.g., population size or trend) that have been shown 205 8 to directly relate to a target recovery threshold, commonly extinction risk. For example, one 206 delisting criterion for island fox (Urocyon littoralis) is based on extinction risk, as calculated 207 from population size and mortality rates. This criterion also details the time period, quasi-208 extinction threshold, and number of years of consistently meeting the risk threshold required 209 before recovery is declared. This plan also explicitly states that the analyses of risk can and 210 should be updated as more data become available. Many more variations on demographic- and 211 threat-based criteria exist among recent plans (Appendix A). 212 Regardless of their content, the ESA mandates that recovery criteria be measurable, but 213 there is no history of this mandate being interpreted in the narrowest, most literal sense. Rather, a 214 wide variety of measures, most of which are indirect and imprecise in the sense that they require 215 statistical extrapolation from partial information (e.g., population sizes estimated from mark-216 recapture analyses, indirectly assayed threat abatement standards, estimated genetically effective 217 population sizes, and probabilities of future extinction) have all been included in plans. 218 Some plans specify that additional evaluation, such as monitoring, population viability 219 analyses (PVA), or threat assessment will be needed to develop or clarify criteria that are not 220 immediately measureable. For example, some plans (e.g., Mariana fruit bat Pteropus mariannus 221 mariannus; Bexar County karst invertebrates; dwarf lake iris, Iris lacustris) state as criteria 222 specific viability targets for a PVA yet to be developed. Others (e.g., gentian pinkroot, Spigelia 223 gentianoides, scaleshell mussel, Leptodea leptodon, Guthrie’s ground plum, Astragalus 224 bibullatus, Puerto Rican parrot, Amazona vittata) merely state criteria stipulating that future 225 analyses must show populations are “viable,” without defining viability. Many threat-based 226 criteria also call for additional analyses to specify target levels. For example, the criteria may 227 state that habitat adequate in extent, quality, and quantity will be identified and protected (e.g., 228 9 plan for Florida manatee, Trichechus manatus) or that a threat will continue to be investigated 229 and ameliorated until it is no longer limiting recovery (e.g., entanglement for Sei whales, or 230 water flows for Florida manatee). 231 Common problems with current recovery criteria 232 We see two problems with the way criteria are often framed and justified. First, many 233 plans fail to link the recovery criteria, either demographic or threat-based, to some objective 234 definition of population recovery. In other words, many plans do not clearly articulate how 235 meeting recovery criteria will result in a population that is at low risk of extinction or otherwise 236 deemed to be “recovered.” This issue has a considerable history in critiques of recovery plans 237 (Gerber and Hatch 2002, Schemske et al. 1994) and continues to be a problem in even the most 238 recent plans (Neel et al. 2012). 239 A second, but related, problem is the conflation of recovery criteria and recovery actions. 240 While these two aspects of a plan are described as distinct elements in the ESA (Table 1), in 241 practice many plans include what would commonly be considered actions (Salafsky et al. 2008) 242 among their recovery criteria. For example, many plans include criteria requiring establishment 243 of monitoring programs or other biological studies (Appendix A). We heard from both Service 244 personnel and conservation NGOs that recovery plan writers may seek to highlight the 245 importance of actions by listing them as criteria and that funding may be more available for 246 actions that are listed as criteria. Still, we view this mixing of actions and criteria as problematic. 247 Recovery criteria should reflect something about the status of the species itself (e.g., population 248 size or distribution, rate of population growth, rate of mortality from some threat) that indicates 249 that it has reached a state of recovery, while recovery actions are what managers do to achieve 250 and evaluate recovery (Table 1). 251 10 Recommendations for improved recovery criteria 252 Regardless of the exact degree of risk that a plan’s recovery criteria embrace – part of the 253 societal decisions that underlie any plan – a scientifically defensible plan should include 254 recovery criteria establishing that the species is safe from extinction or extreme declines for the 255 moderate-term future or that the species is likely to maintain an even higher number or wider 256 geographical distribution deemed necessary for it to play its proper ecological role. Such criteria 257 must account for existing and anticipated or potential future threats (Salafsky et al. 2008), 258 including climate change effects, and shifting regulatory and threat landscapes faced by delisted 259 species (Soulé et al. 2005). The broad set of analytical methods used to judge whether a 260 population or set of populations meets such a standard is usually called population viability 261 analysis (PVA). While we use this acronym, we emphasize that it is something of a misnomer, as 262 these tools very often are used to do much more than simply assess the risk of extinction or near 263 extinction of populations. In the context of recovery criteria, they can and should be used to 264 judge the likelihood of sustaining a wide range of desired attributes of a recovered species, 265 including number and density of individuals, number and geographic distribution of populations, 266 and fulfillment of ecological functioning. 267 Within this broad suggestion, we offer three more specific recommendations: 268 Recommendation 1: The central recovery criteria should be quantitative, biologically-based, and 269 clearly justified. To the greatest extent possible, criteria should be quantitative, focused on traits 270 of the species itself rather than external factors, and based on clear scientific reasoning. To 271 ensure this direct link between criteria and species biology, plans should have a distinct section 272 that outlines the biological justification for each criterion, with evidence of how the quantitative 273 standards are objectively linked to a clearly stated definition of recovery (Box B). Given the 274 11 ambiguity in the ESA regarding what recovery is, this recommendation serves to facilitate both 275 an unambiguous statement of how recovery is defined for a species and how the specified criteria 276 demonstrate that the species has a high probability of remaining in this “recovered” state. Both 277 the definition and rationale are essential to ensure that the connections between available 278 information about the species and the plan’s recovery criteria are transparent to the public and to 279 plan reviewers. We recognize that many other, ancillary criteria will often be included in plans 280 that address less direct aspects of recovery and population management, but without inclusion of 281 criteria that are directly related to biological recovery, a plan is not scientifically defendable. 282 Recommendation 2: All plans should include demographic criteria. Plans should include one or 283 more demographic criteria (criteria focused on population number, dynamics or demography) 284 and state how analyses have been (or will be) done to tie these criteria to the probability of 285 populations meeting specific quasi-extinction risk thresholds or other indices of population 286 health (Box B). If adequate data are available at the time a plan is written, plan developers 287 should conduct analyses of population viability and identify quantitative population metrics, such 288 as population size, population trends over a specified time period, and/or geographical 289 distribution that indicate the population has an acceptably low risk of falling below recovery 290 thresholds. If the data are not in hand to support such analyses when a plan is written, criteria can 291 state the thresholds and risks that are deemed acceptable, and recovery actions can specify 292 collection of the data that will be needed to assess when that criterion has been met (Fig. 1). Both 293 of these approaches are preferable to setting arbitrary demographic thresholds that have no clear 294 link to a species’ ecosystem role or its future viability (Schemske et al. 1994, Tear et al. 1995). 295 As noted above, these approaches have already been taken in some approved plans (e.g., Sei 296 whale, Mariana fruit bat), and have been advocated by NMFS scientists (Demaster et al. 2004) 297 12 and others (Himes Boor 2014), so they are not untested nor too uncertain to pass muster under 298 the ESA. In practice, many of the best plans take a combined approach, defining demographic 299 standards that predict a certain safety from falling below desired thresholds, but also stipulating 300 further data collection to refine the link between numbers and safety, which will in general 301 involve use of some type of PVA (Appendix C). 302 Recommendation 3: Threat-based criteria should derive from the population consequences of 303 threats. A plan that has only threat-based criteria, unlinked to population trends or demographic 304 measurements, is difficult or impossible to defend scientifically. When quantitative estimates of 305 the impacts of threats on demographic processes or population growth rates are available, the 306 level of threat reduction stipulated as a goal for recovery should be based on their population-307 level effects, in the context of other threats and the species’ life history. As the classic case of the 308 loggerhead sea turtle (Caretta caretta) shows, such analyses are necessary to correctly prioritize 309 among different threats and gauge the threat reduction needed to achieve self-sustaining 310 populations (Crouse et al. 1987, Crowder et al. 1994), in part because threat factors themselves, 311 let along specific levels for their abatement, are inherently difficult to crisply and defendably 312 define. We recommend that the goals of threat abatement set as recovery criteria – that is, needed 313 for removal of a species from ESA protection -- be expressed in terms of the level of threat 314 reduction needed for population viability. Specifically, the impacts of current and anticipated 315 future threats (including loss of ESA protections) should be included in population models so 316 that interactive effects of multiple threats, or threat reductions, are folded into an overall 317 assessment of viability (see Appendix B). One option, already taken in some plans (e.g., black-318 footed ferret, Mustela nigripes), is to specify that if the population has reached demographic 319 thresholds that indicate recovery, then threats have been adequately abated. Due to ESA-related 320 13 legal rulings, such demographic thresholds must be justified in the context of threats. Moreover, 321 the criteria should specify that any new information about the demographic impacts of threats 322 and the expected impact of regulatory changes after delisting be incorporated when assessing 323 whether the population is recovered. While accurately anticipating novel or changing threats is 324 not trivial, our approach incorporates this uncertainty into a framework that is flexible and 325 requires any new threats to be controlled to the levels necessary to achieve population safety. 326 If the demographic impacts of a threat cannot be adequately quantified when a plan is 327 written, one alternative is to define criteria addressing this threat in terms of viability (Box B). In 328 these data-poor situations (Fig. 1), this would involve a two-pronged approach that takes 329 advantage of the requirement for plans to define actions as well as criteria. First, recovery criteria 330 would specify that the threat must be low enough to allow the population to meet a specific 331 viability standard. Second, recovery actions would include activities that lower threat levels and 332 also collect data to quantify the demographic or population-level responses to these threat 333 reductions. 334 This approach to threat reduction can also effectively address conservation-reliant 335 species. Managers are increasingly aware that many endangered species will require 336 conservation measures in perpetuity (Goble et al. 2012). Well-executed PVA analyses can take 337 into account future threat management scenarios, including the effects of delisting on regulatory 338 mechanisms needed to ensure that essential management continues. In our view, assessing 339 whether even the seemingly non-biological threat factor D (“inadequacy of existing regulation”) 340 has been sufficiently ameliorated requires a population perspective (e.g., will laws limiting future 341 harvest allow the species to sustain numbers above desired population thresholds?). In some 342 cases, a realistic consideration of a species’ biology and future threat scenarios (e.g., climate 343 14 change, regulatory changes) may preclude recovery criteria that are attainable in the foreseeable 344 future; nevertheless, such a determination would be a successful outcome of quantitative 345 analyses and of the ESA, rather than a failure (Doremus and Pagel 2001). 346 Implications of these recommendations 347 Our recommendations contrast with the Service’s current guidelines on viability-based 348 criteria, which state that such criteria should be ancillary to “traditional population and listing 349 factor-based recovery criteria” because, they state, PVAs rely on estimates of vital rates and on 350 assumptions about threat conditions and their effects on demographic rates (NMFS and USFWS 351 2010; as noted elsewhere, PVAs can be based on many other kinds of data). Yet, “traditional” 352 criteria not linked to PVA are also based on guesses or assumptions about population processes, 353 including demographic rates, as well as assumptions about threat conditions and their effects on 354 demography, with the important difference that these assumptions and estimates are often 355 unclear, implicit, and indirect. This lack of transparency in the estimates and assumptions linking 356 traditional criteria and population health is their key weakness. In viability-based criteria, 357 assumptions about the effects of threats on recovery are explicitly stated, which allows for 358 updating of criteria as assumptions are tested and additional data are collected. 359 Following our recommendations will make criteria more scientifically and legally 360 defensible and more aligned with the already-developed conservation planning literature (e.g., 361 Salafsky et al. 2002 & 2008). In particular, our recommendations seek to create a scientifically 362 justifiable approach that can accommodate the diverse situations of different listed species (Fig. 363 1). For some species, large, long-term data sets are available, the effects of threat factors have 364 been experimentally estimated, and adequate financial resources to support management are in 365 hand. For most species, none of these advantages exist, and a recovery plan can count on only 366 15 modest monitoring and analysis efforts, which make rigid numerical recovery criteria set at the 367 time the plan is written impractical and indefensible. The approach that we suggest can 368 accommodate both these extremes, without resorting to weak generalizations or guesswork. 369 Further, they are designed to be flexible enough to allow recovery criteria to stay relevant in the 370 face of shifting threat conditions such as climate change, exotic species, and land use change. 371 Just as importantly, an emphasis on recovery criteria that are tied to population status, 372 rather than to amelioration of specific threats, can give the Services flexibility to change 373 management tactics if new threats arise after the recovery plan is written. Using demographic 374 criteria, the degree of threat abatement needed can be directly tied to the ultimate goal of 375 recovery, and when new information indicates that more, or less, attention to a given threat is 376 needed, the criteria can accommodate this updated information. 377 Finally, having to show that recovery criteria actually mean that a population is relatively 378 safe from extinction or from dropping to a low level that impedes its functional role in an 379 ecosystem may mean that some species are not removed from the list as quickly. We underscore, 380 however, that this is not a valid objection to these recommendations. If we are slower to remove 381 species from ESA protections because we cannot say with an acceptable degree of certainty that 382 they are indeed recovered, that is the scientifically-justifiable, legally-required, and 383 precautionary outcome. That said, making clearer statements of how recovery is defined should 384 also mean faster delisting of some species, as well as making recovery actions more targeted and 385 de-listing decisions less contentious. 386 In considering our first and most fundamental recommendation, it is important to address 387 several aspects of PVA and related population analysis tools. First, this is not a recommendation 388 to adopt hopelessly complex approaches to viability assessment. Population analyses can be 389 16 quite simple, even when applied to spatially complex situations (see Appendix C for examples); 390 this recommendation does not require mountains of data or cutting-edge analysis, nor is it 391 designed to be a job creation program for population modelers. What it does require is a clear 392 statement of what risk of population deterioration is deemed acceptable, and why the recovery 393 criteria proposed would indicate that a species has likely met this goal. The need to define such 394 clear standards is the most fundamental advantage of taking this approach to recovery criteria 395 development. 396 Second, implementing these recommendations does not require that PVA and other 397 population analysis methods be flawless. The strengths and weaknesses of predicting population 398 fates have been thoroughly dissected in the conservation literature (Beissinger and Westphal 399 1998, Coulson et al. 2001, Ellner et al. 2002, Ludwig 1999). However, the core shortcomings of 400 PVA as a predictive tool are shared with all other predictive methods. Some may argue that, 401 because they are based on analyses more complex than simple statistics, viability-based criteria 402 may be less palatable to policy-makers and managers. But this objection applies to many types of 403 scientific evidence used in legal and social contexts, such as genetic analyses used in criminal 404 cases or the formulation of ecotoxicological standards in pollution control, and in this case can 405 be addressed by clear explanation of the details of the data and assumptions used to estimate 406 population viability and its uncertainty. 407 Finally, with regard to the use of population analysis methods to judge recovery, the 408 limitations of PVAs must be judged against the shortcomings of alternative methods for 409 determining recovery. We do not see a good argument for the use of criteria justified mostly or 410 solely by expert opinion as opposed to standards based on actual analysis of population status 411 and dynamics. Another potential option would be to adopt IUCN listing criteria (IUCN 2012). 412 17 However, we believe that this would be a poor way to improve recovery planning. While their 413 adoption would standardize recovery criteria, IUCN benchmarks were designed as a one-size-414 fits-all system for global priority setting across all taxa and multiple conservation situations, and 415 as such do not take into account species-specific biology and threat conditions. With that said, 416 our recommendations are not incompatible with the IUCN approach, since one of the 417 requirements for moving a species to a lower IUCN threat level is the completion of a 418 quantitative analysis to evaluate its risk of extinction. 419 Implementing the recommendations 420 Criticism of ESA implementation is easy, but practical improvements likely to be 421 adopted given the Services’ legal, political, and budgetary constraints are hard. Based on our 422 conversations with Service personnel, we offer these suggestions for how to implement our 423 recommendations. 424 First, we suggest that the recovery planning guidelines be revised to provide clear 425 guidance to recovery plan authors on why and how to set quantitative, scientifically defensible 426 criteria. We have tried to describe as lucidly as possible how such criteria could be formulated 427 (Box B; Appendix C). 428 Second, we suggest that the Services develop mechanisms to encourage both natural and 429 social scientists from academia to contribute their expertise and time to the process of developing 430 recovery criteria. Writing a well-articulated, objective, and defensible plan would seem nearly 431 impossible without input from individuals with multiple perspectives and expertise, including 432 those with: A) An understanding of the legal and regulatory sideboards of recovery planning; B) 433 Knowledge of the species and its ecosystem, as well as the threats the species faces and their 434 biological impacts; C) Knowledge of the political, social, and land-use settings where the species 435 18 occurs; and, D) Expertise in analytical and modeling methods necessary to define and evaluate 436 ‘recovery’ in a scientifically defensible way. For high-profile species, it is easier for the Services 437 to assemble recovery teams that include members with each of these types of expertise. But the 438 many species for which plans are written by individuals or small teams will often not have the 439 benefit of this complete set of knowledge and skills. This is not a trivial obstacle to improving 440 recovery planning. 441 One possibility to redress this limitation is for university biologists to incorporate 442 recovering planning into their teaching. For example, graduate students in a population ecology 443 course could construct, parameterize, and use population models to craft demographically-based 444 threat reduction actions and recovery criteria. If adequate data are not available, students and 445 faculty could work with plan writers to design effective recovery actions to collect the data 446 needed to define recovery. Close coordination with the Services in such efforts is essential so 447 that the contributions of academic partners are useful to the planning process. A different 448 approach to achieve the same end would be to find funding for postdoctoral researchers or other 449 individuals outside the Services to contribute expertise that could allow the Services to more 450 rapidly produce defensible plans. An added benefit of either scenario is that a cohort of young 451 scientists will gain real-world experience at the intersection of conservation science, practice, 452 and policy, and thereby foster their careers in conservation. Experts on planning, policy, social 453 science, and environmental law could likewise be tapped to work on other elements of recovery 454 planning. 455 Finally, the Services are required to review the status of each listed species every five 456 years, including the evaluation of new information and threats that can trigger a revision of an 457 outdated recovery plan (NMFS and USFWS 2010). We urge the Services to create openings for 458 19 non-agency experts to participate in these reviews, including updating population assessments in 459 light of new data. This phase of the recovery process presents another opportunity for early-460 career scientists to make substantive contributions to conservation practice. 461 Conclusions 462 We believe we have presented practical and important ways to enhance the scientific 463 integrity of the recovery planning process. Similarly, we think that creating ways to better tap the 464 expertise, time, and enthusiasm of scientists outside of the Services can be a means to implement 465 these recommendations and overcome very real constraints faced by the Services in writing 466 strong recovery plans. For that external involvement to be efficient and effective, however, the 467 Services must be open to working with outsiders, and scientists must understand the needs and 468 constraints inherent in ESA implementation. 469 Although we have focused here on recovery planning under the United States ESA, many 470 other nations have similar legislation with provisions for endangered species recovery. While 471 there is a parallel set of proposed approaches to endangered species assessment and recovery 472 planning in other jurisdictions, these proposals and critiques are similar to those of the US ESA – 473 there are many suggestions but little evidence of on-the-ground improvement (Mooers et al. 474 2010, Salafsky et al. 2008, but see Salafsky and Margoluis 1999, . The general approaches we 475 suggest here can help improve the management of threated species elsewhere, and may also have 476 application to other aspects of ESA planning, such as critical habitat designation. With our 477 emphasis on defining clear standards by which to judge recovery, and requiring that recovery 478 criteria and threat reductions be explicitly linked to these measures of population safety, our 479 recommended approach will help ensure that recovery plans more effectively and efficiently 480 guide recovery of imperiled species. 481 20 Acknowledgements 482 We gratefully acknowledge the contributions of T. Abbott, C. Ambrose, C. Carol, D. Crouse, M. 483 Neel, L. Rabin, J. Tutchton, and S. Wolf, all of whom participated in our October 2012 484 workshop but who could not, or chose not to, be authors of this paper. Nonetheless, they 485 provided key perspectives and information and deserve more than a standard acknowledgement. 486 The Wilburforce Foundation and The Nature Conservancy provided funding and the Gordon and 487 Betty Moore Foundation hosted the workshop. W.F. Morris was supported by the Swedish 488 Science Council. 489 490 491 21 Literature cited 492 493 Beissinger SR, Westphal MI. 1998. On the use of demographic models of population viability in 494 endangered species management. Journal of Wildlife Management 62: 821-841. 495 Boersma PD, Kareiva P, Fagan WF, Clark JA, Hoekstra JM. 2001. How good are endangered 496 species recovery plans? Bioscience 51: 643-649. 497 Carroll C, Vucetich JA, Nelson MP, Rohlf DJ, Phillips MK. 2010. Geography and recovery 498 under the US Endangered Species Act. Conservation Biology 24: 395-403. 499 Coulson T, Mace GM, Hudson E, Possingham H. 2001. The use and abuse of population 500 viability analysis. Trends in Ecology & Evolution 16: 219-221. 501 Crouse DT, Crowder LB, Caswell H. 1987. A stage-based population model for loggerhead sea 502 turtles and implications for conservation. Ecology (Washington D C) 68: 1412-1423. 503 Crowder LB, Crouse DT, Heppell SS, Martin TH. 1994. Predicting the impact of turtle excluder 504 devices on loggerhead sea turtle populations. Ecological Applications 4: 437-445. 505 Demaster DP, Angliss R, Cochrane J, Mace P, Merrick R, Miller M, Rumsey S, Taylor B, 506 Thompson G, Waple R. 2004. Recommendations to NOAA Fisheries: ESA Listing Criteria 507 by the Quantitative Working Group. NOAA Technical Memorandum NMFS-F/SPO-67. 508 Doremus H, Pagel JE. 2001. Why listing may be forever: Perspectives on delisting under the US 509 Endangered Species Act. Conservation Biology 15: 1258-1268. 510 Ellner SP, Fieberg J, Ludwig D, Wilcox C. 2002. Precision of population viability analysis. 511 Conservation Biology 16: 258-261. 512 22 Foin TC, Riley SPD, Pawley AL, Ayres DR, Carlsen TM, Hodum PJ, Switzer PV. 1998. 513 Improving recovery planning for threatened and endangered species. Bioscience 48: 177-514 184. 515 GAO (Government Accountability Office). 2006. Endangered Species: Time and Costs Required 516 to Recover Species Are Largely Unknown. GAO-06-463R Endangered Species Recovery. 517 Report no. 518 Gerber LR, Demaster DP. 1999. A quantitative approach to Endangered Species Act 519 classification of long-lived vertebrates: application to the north Pacific humpback whale. 520 Conservation Biology 13: 1203-1214. 521 Gerber LR, Hatch LT. 2002. Are we recovering? An evaluation of recovery criteria under the 522 U.S. Endangered Species Act. Ecological Applications 12: 668-673. 523 Gibbs KE, Currie DJ. 2012. Protecting endangered species: Do the main legislative tools work? 524 PLoS One 7. 525 Gilpin ME. 1987. Spatial structure and population vulnerability. Pages 125-139 in Soule ME, ed. 526 Viable populations for conservation. Cambridge, UK: Cambridge University Press. 527 Goble DD. 2009. The Endangered Species Act: What we talk about when we talk about 528 recovery. Natural Resources Journal 49: 1-44. 529 Goble DD, Wiens JA, Scott JM, Male TD, Hall JA. 2012. Conservation-reliant species. 530 BioScience 62: 869-873. 531 Himes Boor GK. 2014. A framework for developing objective and measurable recovery criteria 532 for threatened and endangered species. Conservation Biology 28: 33-43. 533 Hoekstra JM, Clark JA, Fagan WF, Boersma PD. 2002. A comprehensive review of Endangered 534 Species Act recovery plans. Ecological Applications 12: 630-640. 535 23 IUCN. 2012. IUCN Red List Categories and Criteria. Version 3.1. Second edition. Gland, 536 Switzerland and Cambridge, UK: IUCN. Report no. 537 Lawler JJ, Campbell SP, Guerry AD, Kolozsvary MB, O'Connor RJ, Seward LCN. 2002. The 538 scope and treatment of threats in endangered species recovery plans. Ecological 539 Applications 12: 663-667. 540 Ludwig D. 1999. Is it meaningful to estimate a probability of extinction? Ecology (Washington 541 D C) 80: 298-310. 542 Mace GM, Lande R. 1991. Assessing extinction threats - toward a reevaluation of IUCN 543 threatened species categories. Conservation Biology 5: 148-157. 544 McElhany P, Ruckleshaus MH, Ford MJ, Wainwright TC, Bjorkstedt. EP. 2000. Viable 545 salmonid populations and the recovery of evolutionarily significant units. NOAA 546 Technical Memorandum NMFS-NWFSC-42. 156 pp. 547 Mooers AO, Doak DF, Findlay CS, Green DM, Grouios C, Manne LL, Rashvand A, Rudd MA, 548 Whitton J. 2010. Science, policy, and species at risk in Canada. Bioscience 60: 843-849. 549 Morris WF, Bloch PL, Hudgens BR, Moyle LC, Stinchcombe JR. 2002. Population viability 550 analysis in endangered species recovery plans: Past use and future improvements. 551 Ecological Applications 12: 708-712. 552 Moyle LC, Stinchcombe JR, Hudgens BR, Morris WF. 2003. Conservation genetics in the 553 recovery of endangered animal species: A review of US endangered species recovery plans 554 (1977-1998). Animal Biodiversity and Conservation 26: 85-95. 555 Neel MC, Leidner AK, Haines A, Goble DD, Scott JM. 2012. By the numbers: How is recovery 556 defined by the US Endangered Species Act? Bioscience 62: 646-657. 557 24 NMFS and USFWS. 2010. Interim endangered and threatened species recovery planning 558 guidance, Version 1.3. 559 Regan T, Taylor B, Thompson G, Chochrane J, Merrick R, Nammack M, Rumsey S, Ralls K, 560 Runge M. 2009. Developing a structure for quantitative listing criteria for the U.S. 561 Endangered Species Act using performance testing, Phase 1 report. NOAA Technical 562 Memorandum NMFS-TM-NMFS-SWFSC-437. U.S. Department of Commerce, 563 Southwest Fisheries Science Center. 564 Rohlf DJ. 1991. Six biological reasons why the Endangered Species Act doesn't work - and 565 what to do about it. Conservation Biology 5: 273-282. 566 Salafsky N, Margoluis R. 1999. Threat reduction assessment: A practical and cost-effective 567 approach to evaluating conservation and development projects. Conservation Biology 13: 568 830-841. 569 Salafsky N, et al. 2008. A standard lexicon for biodiversity conservation: Unified classifications 570 of threats and actions. Conservation Biology 22: 897-911. 571 Schemske DW, Husband BC, Ruckelshaus MH, Goodwillie C, Parker IM, Bishop JG. 1994. 572 Evaluating approaches to the conservation of rare and endangered plants. Ecology 75: 584-573 606. 574 Scott JM, Goble DD, Wiens JA, Wilcove DS, Bean M, Male T. 2005. Recovery of imperiled 575 species under the Endangered Species Act: the need for a new approach. Frontiers in 576 Ecology and the Environment 3: 383-389. 577 Soulé ME, Estes JA, Miller B, Honnold DL. 2005. Strongly interacting species: conservation 578 policy, management, and ethics. BioScience 55: 168-176. 579 25 Tear TH, Scott JM, Hayward PH, Griffith B. 1993. Status and prospects for success of the 580 Endangered Species Act; a look at recovery plans. Science 262: 976-977. 581 —. 1995. Recovery plans and the Endangered Species Act; are criticisms supported by data. 582 Conservation Biology 9: 182-195. 583 U.S. Fish and Wildlife Service [USFWS]. 2013. ESA basics: 40 years of conserving endangered 584 species. USFWS Endangered Species Program, Arlington, Virginia. Available from 585 http://www.fws.gov/endangered/esa-library/pdf/ESA_basics.pdf (accessed July 2013). 586 Vucetich JA, Nelson MP, Phillips MK. 2006. The normative dimension and legal meaning of 587 endangered and recovery in the US Endangered Species Act. Conservation Biology 20: 588 1383-1390. 589 590 26 TABLES 591 592 Table 1: Key definitions under ESA 593 ESA protects species listed under the act as endangered or threatened: 594 Endangered: “In danger of extinction throughout all or a significant portion of its range" 595 (16 USC § 1532). 596 Threatened: "Likely to become an endangered species within the foreseeable future 597 throughout all or a significant portion of its range" (16 USC § 1532) 598 ESA requires the development of recovery plans whose purpose is “to restore a species to 599 ecological health” (USFWS 2013a). Several closely related concepts form the foundation of a 600 recovery plan: 601 Recovery or Recovery goal: ESA’s “ultimate goal is to ‘recover’ species so they no 602 longer need protection under the ESA” (USFWS 2013). Thus, at a minimum, “recovery” 603 means the species is not in danger of extinction in the foreseeable future. Translating this 604 to the terms of quantitative conservation biology, recovery is the attainment of the 605 conditions by which the species is viable over a long time frame. According to the 606 Services, “some recovery planning efforts may attempt to set goals higher than those 607 needed to achieve delisting of the species” (NMFS and USFWS 2010). An example of 608 such a goal might be reaching densities and distributions that allow it to fulfill key 609 ecological roles. 610 27 Recovery objective: The Services use recovery objectives to link the recovery goal and 611 criteria, stating “recovery objectives are the parameters of the goal, and criteria are the 612 values for those parameters” (NMFS and USFWS 2010). 613 Recovery criteria: The conditions that signify recovery has been attained. As stated by 614 the Services, “recovery criteria are the values by which it is determined that [a recovery] 615 objective has been reached…” (NMFS and USFWS 2010). Thus, a clearly stated concept 616 of recovery might be 95 percent probability of persistence over 100 years. 617 Recovery actions: The steps the Services or other managers take to manage the species 618 to achieve the goal of recovery. As stated by the Services, recovery actions are the steps 619 “that will alleviate known threats and restore the species to long term sustainability. 620 These actions might include (but are not limited to) habitat protection, limitations on 621 take, outreach, research, control of disease, control of invasive species, controlled 622 (including captive) propagation, reintroduction or augmentation of populations, and 623 monitoring actions” (NMFS and USFWS 2010). 624 625 626 627 628 28 FIGURE CAPTIONS 629 Figure 1. Formulating the path to recovery for threatened and endangered species is influenced by the degree of knowledge of threats 630 and of population demography and distribution. We present general guidelines for developing demographic and threat-based recovery 631 criteria for listed species based on the initial levels of knowledge about the species and its threats. All completed recovery plans, 632 including those listed here as examples, are available at: http://www.fws.gov/endangered/species/recovery-plans.html 633 634 29 Box A. Sociopolitical factors influencing recovery criteria 637 Multiple analyses have shown that sociopolitical factors have strong influences on many aspects 638 of ESA implementation, including recovery criteria (Goble 2009, Vucetich et al. 2006). Two 639 crucial components of recovery criteria that are particularly influenced by social and policy 640 considerations are: 641 Portion of range to which a species should be restored. The ESA calls for a species to be 642 listed if it is endangered or threatened in all or a Significant Portion of its Range (SPR), and thus 643 delisting should specify the geographic area to which healthy populations must be restored. 644 Despite ongoing debate about the meaning of SPR (Carroll et al. 2010, Vucetich et al. 2006), the 645 issue of where endangered species must or should be restored is clearly influenced by the 646 sociopolitical setting and constraints imposed by feasibility and societal desirability. Within 647 existing recovery plans, the extent of occupied range for recovered populations is typically 648 addressed through viability needs. Similarly, USFWS recently issued guidance on SPR, 649 clarifying that a portion of the range is considered significant if “its contribution to the viability 650 of the species is so important that, without that portion, the species would be in danger of 651 extinction” (76 Fed. Reg. 237 (December 2011), pp. 76987-77006). The viability-based approach 652 to recovery criteria we advocate neither requires nor precludes broader definitions of SPR arising 653 from the policy arena. 654 Acceptable risk of extinction. Under the ESA, recovery implicitly means a species is not in 655 danger of extinction (Table 1), but any population has some possibility of extinction and the ESA 656 does not quantitatively define acceptable vs. unacceptable risk. Several authors have advocated 657 for normative standards for acceptable extinction risk (e.g., Gerber and Demaster 1999, Gilpin 658 1987, Mace and Lande 1991), and NMFS documents have proposed some guidelines (Demaster 659 31 et al. 2004, McElhany et al. 2000, Regan et al. 2009). Similarly, IUCN has established extinction 660 risk levels for its categories of endangerment (IUCN 2012). 661 Nonetheless, the acceptable risk of extinction for a recovered species has so far been determined 662 on a case-by-case basis. We surveyed plans from 2009 to the present, and show below the 663 combinations of extinction risk and time horizons for species for which both risk and horizon 664 were defined in recovery criteria. We also indicate IUCN viability standards. Across plans, there 665 is high variation, but also a negative association between time horizon and extinction risk 666 (Spearman rank correlations -0.59 and -0.83 [p<0.02] for delisting and downlisting, 667 respectively), further exacerbating the high variance in acceptable extinction risk across plans. 668 Society is willing to accept a higher extinction risk for some species (upper left) than for others 669 (lower right). One striking trend was how few of these plans (only 6 of 23) employed quasi-670 extinction thresholds, with the majority using complete extinction in defining risk. While we do 671 not propose or advocate for any universal standards for risk here, viability-based recovery 672 32 criteria are compatible with the establishment of either universal or taxon-specific standards 673 arising from the policy arena. 674 675 Box A Figure. Viability criteria used to assess recovery are highly variable across recovery 676 plans. A common way to assess population viability is by the risk that a population will become 677 extinct, or fall below a specified quasi-extinction threshold, over some time horizon. In ESA 678 recovery plans published between 2009 and 2013, a range of acceptable risks of extinction or 679 quasi-extinction that would allow delisting or downlisting were used, and there was a similarly 680 wide range of time horizons employed. Viability standards defined for different IUCN categories 681 of risk are also shown; for a given time horizon, ESA criteria are generally more demanding than 682 those used by the IUCN. 683 684 685 33 Box B. Illustrative wording for recovery criteria 686 We present the following templates for demographic, threat-based, and combined criteria that 687 follow the recommendations outlined in the text as well as recommendations made in Himes 688 Boor (2014). They are presented as illustrative examples; many other criteria could be 689 formulated that meet the standards set out in our recommendations. 690 Demographic criterion with adequate data: 691 Estimated intrinsic growth rate for the entire population must meet or exceed _ with _% 692 probability of certainty for more than _ years based on our analyses that such growth will 693 result in a population with less than _% probability of quasi-extinction within _ years (see 694 Appendix _ for analysis details). 695 Demographic criterion with inadequate data: 696 The species as a whole should have <_% probability of extinction within _ years and 697 each individual population should maintain a probability of extinction <_% within _ 698 years. The viability models should be peer reviewed and must take into account 699 uncertainty in parameter estimates and future scenarios, including potential impacts of 700 climate change and threat factors _ and _. The data to complete such an assessment 701 should meet the standards outlined in Section _ of this recovery plan. 702 Threat-based criterion with adequate data: 703 Threat _ must be reduced to an estimated (based on the upper _% confidence interval) _ 704 units per year across the entire species current range and must remain at or below that 705 level for _ years. We estimate that this reduction will result in a _% increase in vital rate 706 _, thus allowing a population growth rate consistent with less than an _% probability of 707 extinction within _ years (see Appendix _ for the detailed analysis). 708 34 Threat-based criterion with inadequate data: 709 Threats _ and _ should be reduced such that their cumulative impact on the species is no 710 longer threatening its viability and the population has greater than _% probability of 711 persistence for more than _ years. The model developed to estimate viability should be 712 peer reviewed and must take into account uncertainty in parameter estimates, future 713 management scenarios, and threat impacts. 714 Combined demographic and threat-based criterion: 715 Estimates of total population size must meet or exceed _ breeding individuals with _% 716 probability of certainty for more than _ years, based on our analyses demonstrating that a 717 population maintaining that number of breeding individuals has less than _% probability 718 of quasi-extinction within _ years and has overcome threats _ and _. 719 720 Each criterion for which a model (e.g., PVA or threats analysis) is used, should also specify the 721 section of the recovery plan containing the detailed model description, including all model 722 assumptions and justifications. Criteria should also be accompanied by references to the section 723 of the recovery plan describing explicit methodologies for collecting data and estimating 724 parameters, including acceptable levels of uncertainty surrounding estimated parameters. This 725 will ensure the appropriate data are collected for the desired analysis. 726 35