Using ecological theory to develop recovery criteria for an endangered butterfly

Recovery criteria (RC) serve the important purpose of determining when an endangered species can be delisted, or removed from protection under the Endangered Species Act (ESA). Although delisting is the ultimate goal for recovering all threatened species, it has been a controversial process, because delisted species may lose some protection provided by the ESA, making them susceptible to the same causes of decline that resulted in their initial listing. RC are designated by the US Fish and Wildlife Service (USFWS) in the recovery plan of an endangered species, and the ESA mandates that RC should be based on “objective, measurable criteria.” How to designate RC has been a thorny problem since the ESA was enacted in 1972. In this issue of BioScience, we find conflicting views of how RC should be determined offered by Doak and colleagues and by Wolf and colleagues. The authors of both articles champion the application of quantitative RC in place of the heterogeneous approach currently used. But here the similarities end. Doak and colleagues argue that RC should be based on “demographic criteria,” emphasizing estimates of the risk of extinction from a population viability analysis (PVA) that projects population size for decades to 100 years (or more) into the future. Under this approach, recovery plans would be required to include RC tied to the probability of meeting specific extinction risk or demographic thresholds from models. In contrast, Wolf and colleagues espouse a less data-intensive and broader set of methods, based on the ecological principles of representation, resiliency, and redundancy (the “3Rs”). The 3Rs would be evaluated quantitatively or qualitatively using multiple approaches for setting recovery targets, such as the percentage of historic range, population size, the number and spatial distribution of populations, and the risk of extinction from a PVA when adequate data are available. Both articles provide important insights into the shortcomings of past efforts to delineate RC and discussion of the concepts for delineating RC. But both run head on into the same knotty problems of developing quantitative criteria for RC that relate to population viability and the absence of risk standards.

Use of population viability analyses (PVAs) in endangered species recovery planning has been met with both support and criticism. Previous reviews promote use of PVA for setting scientifically based, measurable, and objective recovery criteria and recommend improvements to increase the framework's utility. However, others have questioned the value of PVA models for setting recovery criteria and assert that PVAs are more appropriate for understanding relative trade-offs between alternative management actions. We reviewed 258 final recovery plans for 642 plants listed under the U.S. Endangered Species Act to determine the number of plans that used or recommended PVA in recovery planning. We also reviewed 223 publications that describe plant PVAs to assess how these models were designed and whether those designs reflected previous recommendations for improvement of PVAs. Twenty-four percent of listed species had recovery plans that used or recommended PVA. In publications, the typical model was a matrix population model parameterized with ≤5 years of demographic data that did not consider stochasticity, genetics, density dependence, seed banks, vegetative reproduction, dormancy, threats, or management strategies. Population growth rates for different populations of the same species or for the same population at different points in time were often statistically different or varied by >10%. Therefore, PVAs parameterized with underlying vital rates that vary to this degree may not accurately predict recovery objectives across a species' entire distribution or over longer time scales. We assert that PVA, although an important tool as part of an adaptive-management program, can help to determine quantitative recovery criteria only if more long-term data sets that capture spatiotemporal variability in vital rates become available. Lacking this, there is a strong need for viable and comprehensive methods for determining quantitative, science-based recovery criteria for endangered species with minimal data availability. Uso Actual y Potencial del Análisis de Viabilidad Poblacional para la Recuperación de Especies de Plantas Enlistadas en el Acta de Especies En Peligro de E.U.A.

Simple SummaryThe objective of our study was to evaluate the mention of uncertainty (i.e., variance) associated with population size estimates within U.S. recovery plans for endangered animals. To do this we reviewed all finalized recovery plans for listed terrestrial vertebrate species. We found that more recent recovery plans reported more estimates of population size and uncertainty. Also, bird and mammal recovery plans reported more estimates of population size and uncertainty. We recommend that updated recovery plans combine uncertainty of population size estimates with a minimum detectable difference to aid in successful recovery. United States recovery plans contain biological information for a species listed under the Endangered Species Act and specify recovery criteria to provide basis for species recovery. The objective of our study was to evaluate whether recovery plans provide uncertainty (e.g., variance) with estimates of population size. We reviewed all finalized recovery plans for listed terrestrial vertebrate species to record the following data: (1) if a current population size was given, (2) if a measure of uncertainty or variance was associated with current estimates of population size and (3) if population size was stipulated for recovery. We found that 59% of completed recovery plans specified a current population size, 14.5% specified a variance for the current population size estimate and 43% specified population size as a recovery criterion. More recent recovery plans reported more estimates of current population size, uncertainty and population size as a recovery criterion. Also, bird and mammal recovery plans reported more estimates of population size and uncertainty compared to reptiles and amphibians. We suggest the use of calculating minimum detectable differences to improve confidence when delisting endangered animals and we identified incentives for individuals to get involved in recovery planning to improve access to quantitative data.

We asked two basic questions about endangered species recovery plans: (1) Have recovery plans improved over the last decade? (2) Are important features of recovery plans biased toward plants or animals? We answer these questions in the context of a large national study aimed at statistically summarizing key features of recovery plans and how those plans use science. In addition, we asked if the status of endangered species tends to be improving. Overall, the U.S. Fish and Wildlife Service (USFWS) is improving in its use of science in recovery plans. For example, based on the increased number of monitoring tasks included in more recent plans, the USFWS is becoming more concerned with the need to monitor changes in endangered species populations and their habitats. Similarly, specific recovery criteria, such as the number of years that a species needs to maintain a target population size, are more often being included in recovery plans. However, several features of recovery plans have not shown improvement, such as the tasks recommended to address major threats and the limited influence of focal species' biology on the selection of recovery criteria and monitoring protocols. Recovery plan features tend to be biased toward animals in several ways. Among animals, more tasks are recommended to address limitations in current biological information, to address major threats to the species, and to enhance public relations than is the case for plants. Finally, ∼30% of species for which the current recovery plan is older than 1990 are increasing in abundance, a marked improvement over those species for which the recovery plan was only recently written. These data suggest that the process of listing species and writing recovery plans is working. In addition, animal populations are more often increasing in abundance than plant populations.

Legal classification of species requires scientific and values-based components, and how those components interact depends on how people frame the decision. Is classification a negotiation of trade-offs, a decision on how to allocate conservation efforts, or simply a comparison of the biological status of a species to a legal standard? The answers to problem-framing questions such as these influence decision making in species classifications. In our experience, however, decision makers, staff biologists, and stakeholders often have differing perspectives of the decision problem and assume different framings. In addition to differences between individuals, in some cases it appears individuals themselves are unclear about the decision process, which contributes to regulatory paralysis, litigation, and a loss of trust by agency staff and the public. We present 5 framings: putting species in the right bin, doing right by the species over time, saving the most species on a limited budget, weighing extinction risk against other objectives, and strategic classification to advance conservation. These framings are inspired by elements observed in current classification practices. Putting species in the right bin entails comparing a scientific status assessment with policy thresholds and accounting for potential misclassification costs. Doing right by the species adds a time dimension to the classification decision, and saving the most species on a limited budget classifies a suite of species simultaneously. Weighing extinction risk against other objectives would weigh ecological or socioeconomic concerns in classification decisions, and strategic classification to advance conservation would make negotiation a component of classification. We view these framings as a means to generate thought, discussion, and movement toward selection and application of explicit classification framings. Being explicit about the decision framing could lead decision makers toward more efficient and defensible decisions, reduce internal confusion and external conflict, and support better collaboration between scientists and policy makers.

The minimum viable population (MVP) size has been compared for a wide range of organisms in conservation biology, but a limited number of studies investigated it for freshwater fishes, which exhibit diverse life history strategies. In this study, the MVP size and population growth rate of 36 fish species in the Yangtze River were estimated and compared with their life-history traits. The results indicated that the MVP size ranged from 42 to 320 individuals, and instantaneous per-capita population growth rate ranged from 0.009 to 0.188 per year. MVP size and population growth rate were significantly associated with three life history traits: the age at maturity, generation time, and fecundity. Long-lived species with delayed maturation, long generation time, and high fecundity had a greater MVP size and a lower population growth rate than short-lived species. Therefore, our results emphasize a need for prioritizing our conservation effort more on long-lived species.

Revisions allow the recovery planning process for threatened and endangered species to be flexible and responsive to new information or changes in the status of a species. However, the Endangered Species Act defines neither firm criteria that trigger revision of recovery plans nor clear guidelines about how plans should be revised. Consequently, the effect of revisions in the recovery planning process is unknown. We examined how species and recovery plan attributes influenced the likelihood that a plan would be revised and how the content of plans changed with revision. Vertebrate species with designated critical habitat were nearly four times more likely to have their recovery plans revised than were invertebrates or plants without designated critical habitat. Nonetheless, recovery priorities assigned by the U.S. Fish and Wildlife Service (USFWS) did not influence the likelihood of plan revision. Paired comparisons between original and revised versions suggested that knowledge of species biology and status had improved, and that recognition of threats had increased since the original plans were written. However, these improvements did not lead to recovery criteria or monitoring actions that were more clearly justified. We recommend that recovery plan authors strive to maximize benefits from improved biological information by defining management actions and goals that are more biologically justified. We also urge the USFWS to establish a consistent priority system for recovery plan revisions that affords consideration to listed species of all taxa and emphasizes revisions for those species most likely to benefit.

The Lower Keys marsh rabbit (LKMR; Sylvilagus palustris hefneri) is endemic to the Lower Keys of Florida and exists as a metapopulation in patches of salt-marsh–buttonwood transition zone, freshwater wetlands, and coastal beach berm vegetation (Forys and Humphrey 1996). Local LKMR populations interact through dispersing individuals, and patches undergo periodic local extinctions and recolonization (Forys and Humphrey 1999a). Much of the LKMR’s habitat has been lost or fragmented by human development over the past several decades, prompting the United States Fish and Wildlife Service (USFWS) to list the subspecies as endangered in 1990 (USFWS 1990). Knowledge of the distribution of occupied and potential LKMR habitat patches is critical to the design of recovery strategies for the subspecies. Historically, LKMRs occupied most, if not all, larger islands from Big Pine Key to Boca Chica Key, and may have occurred on Key West as well (DePourtales 1877, Layne 1974, Howe 1988, Lazell 1989; D. Stevenson, Felix Environmental Services, personal communication; Fig. 1). Although rapid habitat loss occurred in the 1970s and 1980s (USFWS 1997), little information is available on the LKMR’s distribution prior to 1988. The oldest known written records are anonymous notes dated 1968–1987 written on blue line photographs discovered at the National Key Deer Refuge in 2003. These records documented presence of LKMRs at 20 sites on Big Pine Key. Howe (1988) conducted the first published survey of the LKMR metapopulation by surveying 13 patches noted by J. Lazell (The Conservation Agency, unpublished data) and 3 additional patches. He reported that rabbits were absent from 4 of these patches, no longer occurred on Cudjoe Key, and were threatened by human activities in remaining patches. Forys (1995) completed a more comprehensive survey from 1991 through 1993. She identified 59 patches of occupied and potential (i.e., unoccupied, but apparently suitable) LKMR habitat. This survey included all but 2 of the patches noted by Howe (1988). LKMRs consistently occupied 19 of these patches over the course of her study, occupied 23 other patches in 1 survey, and did not occupy 17 patches. In 1995, Forys et al. (1996) discovered an additional 19 and 47 occupied and potential patches, respectively. Forys (1995) and Forys et al. (1996) found no extant populations between Big Pine and Sugarloaf keys. Based on these surveys and demographic data collected by Forys (1995), a population viability analysis (PVA) model suggested the metapopulation had a 100% probability of extinction in the next 50 years (Forys and Humphrey 1999b). Researchers have not formally surveyed the range-wide status and distribution of the LKMR since the mid-1990s. Managers charged with the recovery of this species need current data. Therefore, we surveyed potential LKMR habitat throughout the Lower Keys from 2001 through 2005. Specifically, our objectives were 1) to update the distribution and occupancy status of LKMR habitat patches, 2) to document patterns in patch occupancy from 2001 through 2005, and 3) to draw comparisons to previous formal and informal surveys.

We calculated a genetically based minimum viable population size of the red-cockaded woodpecker (Picoides borealis) using a formula derived from Hill (1972) and life history data from a long-term study. Based on published criteria for maintenance of genetic variability, a red-cockaded woodpecker population must contain 509 breeding pairs to be considered viable. It is likely that no existing population contains 509 breeding pairs. Genetically based estimates of population viability may not be valid, but if they are adopted as in the recovery plan for the red-cockaded woodpecker, the area required for a viable population would be >25,450 ha. The estimates of population size and area required for a viable population are considerably higher than those contained in the species' recovery plan (U.S. Fish and Wildl. Serv. 1985). J. WILDL. MANAGE. 52(3):385-391 A viable population is self-sustaining over a long period. Population viability is a critical issue in conservation, but how viability should be assessed is unclear. The most commonly used methods are (1) determining the probability, based on demography and population size, that a population will become extinct over a predetermined number of years (Shaffer 1983, Shaffer and Samson 1985); (2) determining the size of naturally occurring, stable populations of a species (Shaffer 1981); and (3) determining if a population is large enough to maintain its genetic variability (Franklin 1980, Lehmkuhl 1984). These 3 methods address different aspects of viability and are not interchangeable. Genetic variability contributes to viability because it provides potential for populations to adapt to changing environments. An excessive loss of genetic variability therefore reduces the chances of a population persisting. In small populations, loss of genetic variability may be caused by inbreeding (Ballou and Rails 1982, Ralls and Ballou 1983) or genetic drift (Wright 1931, 1948). Franklin (1980) suggested that an effective population size (N,) of 50 would be adequate to avoid losses from inbreeding and an N, of 500 would avoid loss of genetic variability caused by drift. Frankel and Soule (1981) and Frankel (1983) provided further support for these recommendations. There are, however, serious problems in determining viable population size using genetic criteria. Although the relationship between N, and loss of genetic variability is fairly well understood, the relationship between genetic variability and population viability is not. Franklin's (1980) recommendations are based on the former relationship. Moreover, N, required to maintain genetic variability may vary from the standard of 500 individuals because of differences in inherent variability among species, demographic constraints, or evolutionary history of a population's structure (Frankel 1983, Lande and Barrowclough 1987). Therefore, Franklin's (1980) estimate of 500 should be considered an approximation subject to a variety of errors in any application. Furthermore, maintainin a certain N, does not guarantee the longterm preservation of genetic variability. Consider a s ochastic event, such as a hurricane or severe winter, that kills a large portion of the breeding population and causes poor reproduction the following year. Even if the population recovers to its original size within 2 years, the loss of genetic variability caused by the stochastic event is not recovered immediately (Franklin 1980). Such rare events are unlikely to be incorporated into calculations of N,. These considerations lead some to favor demographically based estimates of minimum viable popl ion size over genetically based estimates (Shaffer 1983), and to argue against the use of genetic models of population viability in conse vation (W. R. Dawson et al., Report of the advisory panel on the spotted owl, Natl. Audubon Soc., unpubl. rep., 1986). Although we share these reservations about genetic models of population viability, we recognize certain realities. Genetically based estimates of minimum viable population size are already being incorporated into recovery plans for endangered species. Because the National

Animal responses to habitat boundaries will influence the effects of habitat fragmentation on population dynamics. Although this is an intuitive and often observed animal behavior, the influences of habitat boundaries have rarely been quantified in the field or considered in theoretical models of large scale processes. We quantified movement behavior of the Fender's blue butterfly (Icaricia icarioides fenderi) as a function of distance from host-plant patches. We measured the butterfly's tendency to move toward habitat patches (bias) and their tendency to continue to move in the direction they were already going (correlation). We found that butterflies significantly modify their behavior within 10–22 m from the habitat boundary. We used these data to predict large scale patterns of residence time as a function of patch size, using three dispersal models: homogeneous response to habitat, heterogeneous response to habitat, and heterogeneous response to habitat with edge-mediated behavior. We simulated movement for males and females in eight patch sizes (0.1–8 ha) and asked how residence time varies among the models. We found that adding edge-mediated behavior significantly increases the residence of Fender's blue butterflies in their natal patch. Only the model with edge-mediated behavior for females was consistent with independent mark–release–recapture (MRR) estimates of residence time; other models dramatically underestimated residence times, relative to MRR data.

Although population viability analysis (PVA) can be an important tool for strengthening endangered species recovery efforts, the extent to which such analyses remain embedded in the social process of recovery planning is often unrecognized. We analyzed two recovery plans for the Mexican wolf that were developed using similar data and methods but arrived at contrasting conclusions as to appropriate recovery goals or criteria. We found that approximately half of the contrast arose from uncertainty regarding biological data, with the remainder divided between policy-related decisions and mixed biological-policy factors. Contrasts arose from both differences in input parameter values and how parameter uncertainty informed the level of precaution embodied in resulting criteria. Policy-related uncertainty originated from contrasts in thresholds for acceptable risk and disagreement as to how to define endangered species recovery. Rather than turning to PVA to produce politically acceptable definitions of recovery that appear science-based, agencies should clarify the nexus between science and policy elements in their decision processes. The limitations we identify in endangered-species policy and how PVAs are conducted as part of recovery planning must be addressed if PVAs are to fulfill their potential to increase the odds of successful conservation outcomes.

Abstract: The U. S. Endangered Species Act ( ESA ) requires that recovery plans establish “objective, measurable” criteria on which to base listing decisions. Recovery plans are more effective if these criteria are clearly linked to the biology of species of interest. We reviewed recovery plans for 27 listed insect species. Recovery criteria for threatened and endangered insects were poorly linked to species biology. We used population viability analysis to develop quantitative recovery criteria for insects whose population sizes can be estimated, and applied this framework in the context of a recently listed butterfly, the Fender's blue ( Icaricia icarioides fenderi ). We used a simple diffusion‐approximation approach developed to estimate extinction risk from count‐based census data. The method assumes that future trends can be predicted based on current conditions and that observer error is minimal. Of 12 sites we surveyed for at least 8 years, only one population had a high likelihood of persistence. Given observed variation in population growth rate ( σ 2 = 0.79 ) and an initial population size of 300 butterflies, a minimum average growth rate ( λ ) of 1.83 would be needed for there to be a 95% probability that a population would survive 100 years. As the number of independent sites increased, the minimum λ required to have a 95% probability that at least a population at one site would survive 100 years declined. For downlisting, we recommend that three independent sites be maintained or restored in each region of the Fender's blue, with minimum average growth rates of at least λ = 1.55 over 10 years. Given that 41 insect species are listed under the ESA, development of quantitative recovery criteria would be useful and feasible for at‐risk insect species whose population sizes can be estimated.

Biodiversity is severely threatened by habitat destruction. As a consequence of habitat destruction, the remaining habitat becomes more fragmented. This results in time-lagged population extirpations in remaining fragments when these are too small to support populations in the long term. If these time-lagged effects are ignored, the long-term impacts of habitat loss and fragmentation will be underestimated. We quantified the magnitude of time-lagged effects of habitat fragmentation for 157 nonvolant terrestrial mammal species in Madagascar, one of the biodiversity hotspots with the highest rates of habitat loss and fragmentation. We refined species' geographic ranges based on habitat preferences and elevation limits and then estimated which habitat fragments were too small to support a population for at least 100 years given stochastic population fluctuations. We also evaluated whether time-lagged effects would change the threat status of species according to the International Union for the Conservation of Nature (IUCN) Red List assessment framework. We used allometric relationships to obtain the population parameters required to simulate the population dynamics of each species, and we quantified the consequences of uncertainty in these parameter estimates by repeating the analyses with a range of plausible parameter values. Based on the median outcomes, we found that for 34 species (22% of the 157 species) at least 10% of their current habitat contained unviable populations. Eight species (5%) had a higher threat status when accounting for time-lagged effects. Based on 0.95-quantile values, following a precautionary principle, for 108 species (69%) at least 10% of their habitat contained unviable populations, and 51 species (32%) had a higher threat status. Our results highlight the need to preserve continuous habitat and improve connectivity between habitat fragments. Moreover, our findings may help to identify species for which time-lagged effects are most severe and which may thus benefit the most from conservation actions.

Sound science is a conservation mantra, but “sound science” apparently means different things to different people. According to a coalition of 19 prominent environmental groups, the Sound Science for Endangered Species Planning Act (HR 4840) is “a stealth attempt to undermine science and to weaken the Endangered Species Act” (the coalition’s letter of 9 July 2002 to members of Congress is available at www.earthjustice.org/news/display.html? ID=398f Gordon Smith (R–OR) has introduced a similar bill in the Senate. Both opponents and proponents of the bill agree that listing decisions should be made on the basis of the “best available scientific and commercial data,” as already required under the ESA. But HR 4840 would go further: It would require that the secretary of the Department of the Interior establish criteria for scientific and commercial data, studies, and other information used for listing determinations. Bill opponents think such a requirement amounts to dictating the use of “specific kinds of scientific information in every situation” and “limiting the scientific tools available to scientists to evaluate the threats facing species and people.” In making listing decisions, the US Fish and Wildlife Service (USFWS) and the National Marine Fisheries Service would be compelled to give peerreviewed or field-tested data greater weight than other information that may, bill opponents say, constitute the best available science (population viability analysis, for example, and modeling). Bill cosponsor Greg Walden (R–OR) has advocated that every endangered species decision undergo a National Academies–like review of the scientific data. Referring to the National Academy of Sciences report that found insufficient support for the government-mandated shutoff of water to the Klamath basin to avoid jeopardizing endangered species, Walden said, “The crisis forced on the people of the Klamath Basin is a tragic example of why peer-reviewed science is essential.” Although the legislation would allow the USFWS to pay reviewers “if funds are available,” the bill authorizes no additional funding for these reviews. With ESA implementation funding already stretched to the limits, it is unlikely that the USFWS would be able to convene independent review boards unless scientists were to contribute their time and expertise free of charge. Gary Frazer, chief of the USFWS Division of Endangered Species, estimates that each review—if managed by the USFWS— would cost an average of $100,000. He added that, if contracted to the National Academy of Sciences, that cost would probably rise to $160,000. Negotiations between George Miller (D–CA) and Nick Rahall (D–WV), with ESA challenger Richard Pombo (R–CA), to address some of these concerns were unsuccessful. Miller and Rahall acknowledge that there are some problems with ESA implementation, and both introduced legislation addressing some of the same issues. At a press conference convened on 10 July by opponents of HR 4840, Miller said that while better review is needed, the bill is unworkable and conflates political science and sound science. Organized by Earthjustice, a nonprofit public interest law firm, more than 300 scientists have signed a letter opposing the sound science bills. Speaking on their behalf at the 10 July press conference, David Blockstein, senior scientist for the National Council on Science and the Environment and a member of the AIBS Public Policy Review Committee, said that the Sound Science for Endangered Species Planning Act, in contrast to its title, “is likely to be detrimental to both endangered species and to science.” “It would be a major mistake to place restrictions on what types of data should be considered,” Blockstein explained, because “to do so would undermine the credibility of science and also the credibility of any decisions that would be made based on compromised science.” Some scientists, however, do support the bill. Testifying at a hearing on 16 March, Raymond Dueser, associate dean of the College of Natural Resources at Utah State University, offered the opinion that the proposed changes to the ESA have “the potential to improve the operation of the ESA in significant ways,” although he also expressed concern that preference for empirical data might minimize the role of models. George Miller suggested that HR 4840 is unlikely to reach the House floor for a vote. Nevertheless, he said, “We treat every attack as serious.” Should it ever come to a vote, members of Congress might heed the expert assessment of the National Academies of Science, which in 1995 said that “there has been a good match between science and the ESA.”

Although low back pain is a common medical condition that often progresses to become a chronic problem, little is known about the likelihood of recovery from chronic low back pain (CLBP). This study aimed to measure the risk of recovery from CLBP based on low back pain intensity and back-related functioning measures reported by participants within a pain research registry over 12months of observation and to consider the implications for osteopathic medicine. A total of 740 participants with CLBP in the Pain Registry for Epidemiological, Clinical, and Interventional Studies and Innovation in the United States were studied between April 2016 and October 2021. Inception cohorts for pain recovery and functional recovery were assembled from the participants who did not meet the recovery criteria at registry enrollment. The pain recovery criterion was having a score of≤1/10 on a numerical rating scale for low back pain intensity, and the functional recovery criterion was having a score of≤4/24 on the Roland-Morris Disability Questionnaire. A total of 737 and 692 participants were included in the inception cohorts for pain recovery and functional recovery, respectively. Participants provided follow-up data at quarterly encounters over 12months to determine if they achieved and maintained a pain or functional recovery from CLBP over the entire period of observation. Logistic regression was utilized to identify factors associated with recovery. The mean age of the participants at baseline was 52.9years (SD, 13.1years) and 551 (74.5%) were female. No participant reported a pain recovery that was maintained over all four quarterly encounters, whereas 16 participants (2.3%; 95% CI, 1.2-3.4%) maintained a functional recovery. Having high levels of pain self-efficacy (OR, 17.50; 95% CI, 2.30-133.23; p=0.006) and being Hispanic (OR, 3.55; 95% CI, 1.11-11.37; p=0.03) were associated with functional recovery, and high levels of pain catastrophizing (OR, 0.15; 95% CI, 0.03-0.65; p=0.01) and having chronic widespread pain (OR, 0.23; 95% CI, 0.08-0.66; p=0.007) were inversely associated with functional recovery. The findings for pain self-efficacy and Hispanic ethnicity remained significant in the multivariate analysis that adjusted for potential confounders. The absence of pain recovery and the low likelihood of functional recovery observed in our study suggests that osteopathic physicians should embrace a biopsychosocial approach to CLBP management and work with patients to set realistic expectations based on more pragmatic outcome measures, such as those that address health-related quality of life. The findings also suggest the potential importance of patient education and counseling to enhance pain self-efficacy.