Why Bonytail and Razorback Sucker?

We restrict further coverage here to bonytail and razorback sucker because of their persistence and precarious state in the Lower Colorado River. Bonytails are rare and thus little studied, so there are few contributions to a review of the species' biology (Pacey and Marsh 1998). Razorback suckers, on the other hand, have been studied extensively, and their life history and ecology are treated in several excellent reviews (Minckley et al. 1991, Marsh et al. 2015, Bestgen et al. 2020) and in the species' SSA (= species status assessment; USFWS 2018). Dramatic reduction in distribution and population sizes of both species led the Lower Colorado River MultiSpecies Conservation Program (LCR MSCP¹) to prioritize these fishes for conservation management in the Lower Colorado River mainstem, and actions on their behalf have been actively and aggressively implemented. We do not deal with Colorado pikeminnow (formerly known as Colorado squawfish) because it was absent from the Lower Colorado River when the program was implemented and remains so. Furthermore, there were objections to stocking Colorado pikeminnow that would be fully protected in the Lower Colorado River under the Endangered Species Act (ESA); these objections centered largely around potential influence on river operations and existing sport fisheries of a large, endangered, predatory species. As a result, “experimental nonessential” status for Colorado pikeminnow was rejected despite that designation for fish stocked into the Salt and Verde rivers in central Arizona (USFWS 1985, Hendrickson 1993). Humpback chub also is not a covered species because it historically and currently occurs only in Grand Canyon upstream of the LCR MSCP geographical area. Finally, flannelmouth sucker, while an LCR MSCP–covered species monitored by the program, is not a species of concern and not actively managed. Bonytail, humpback chub, Colorado pikeminnow, and razorback sucker all are actively managed in the Upper Colorado River basin under 2 recovery implementation programs covering that portion of the species' range (USFWS 1987a, 1991b).

Photo 1.

Bonytail, Gila elegans. B. Gratwicke, Smithsonian Conservation Biology Institute.

Applied Fisheries Management

Traditional fish management tools that could be used to benefit bonytail and razorback sucker include population control, stocking to augment existing populations or create new ones, habitat manipulation to improve existing conditions or create new places for the fish to live, regulation, and public education, with the last 2 actions targeted at human uses and users. We summarize below how these tools have been applied (or not) in the Lower Colorado River.

Photo 2.

Razorback sucker, Xyrauchen texanus. © J. Sartore/Photo Ark.

Selective harvest, removal, or chemical treatment to control nonnative predators to reduce impacts on native species could benefit native fishes, but it has been applied only locally in select, isolated habitats in the Lower Colorado River basin. In the Upper Colorado River basin, the exact opposite was practiced. The Green River poisoning was implemented in 1962 to eliminate native fishes, primarily minnows (Cyprinidae) and suckers (Catostomidae) in advance of an anticipated sport fishery for trout in the newly created Flaming Gorge Reservoir (Holden 1991). Since then, mechanical removal of predatory nonnative fishes is actively practiced in Grand Canyon and the Upper Colorado River basin to benefit native species. These 2 programs, the Green River poisoning and current predator control, half a century apart, serve as an example of the gradual change that began in the 1980s in management agency mindset, at least superficially, from “sport fish at all costs” to “native fish despite the cost.”

Bonytail and razorback sucker were brought early into the federal hatchery system where efficient methods of artificial propagation and production were devised (e.g., Toney 1974, Hamman 1982, 1985). Repatriation was the management emphasis for both species—a necessary action because naturally recruiting wild populations were unknown. Bonytail repatriation has occurred primarily in the Colorado River and adjacent habitats. Other than deadend stocking into a few urban ponds and lakes, no effort has been directed toward its expansive historical habitat in the Gila River basin. Between 1991 and the end of 2022, bonytail stocking totaled 483,224 fish: 98,172 into Lake Mohave and lakeside backwaters (the mouths of inundated desert washes separated from the reservoir by windgenerated gravel berms), 284,392 into Lake Havasu and adjacent habitats, 99,591 into the Lower Colorado River mainstem and its connectives downstream of Lake Havasu, and 1069 into inland impoundments and isolated backwaters with no river access. Bonytail repatriations into open waters have produced disappointing results (i.e., no evidence of longterm survival, reproduction, or recruitment; Humphrey et al. 2016, Heishman et al. 2022), while stockings into habitats free of nonnative fishes have been encouraging because survival, reproduction, and recruitment occur. Studies of life history and ecology were facilitated by longterm survival in Cibola High Levee Pond (Photo 3, Appendix 1.1) in Arizona and California (Mueller 2006) and in constructed native fish ponds at Imperial National Wildlife Refuge (NWR) (Imperial Ponds hereafter; Photo 4 and Appendix 1.2) near Yuma, Arizona (e.g., Kesner et al. 2008). Management at both sites comprised provision of spawning substrate, repeated stockings, and various attempts to control nonnative fishes. Work at Imperial Ponds included attempts to maintain suitable habitat conditions, specifically adequate summer oxygen. Successful lifecycle completion and establishment of multiple year classes were documented at both sites. Ironically, bonytail is easily propagated artificially and reproduces prolifically in captivity and in sites absent nonnative fishes. However, attempts to establish new populations through stocking into open waters or closed systems where nonnative fishes are present have consistently failed due to high poststocking mortality from avian and piscine predation that present a suite of mitigation challenges (e.g., Carpenter and Mueller 2008). Establishing wild populations of bonytail and finding alternatives to hatchery production and repatriation to perpetuate this critically imperiled species are among the most urgent native fish conservation challenges in western North America.

Photo 3.

Cibola High Levee Pond, La Paz Co., Arizona, and Riverside Co., California. An example of a seminatural, offchannel habitat on a river channel. C. Ehlo, USFWS, November 2016.

Photo 4.

Imperial Ponds, Yuma Co., Arizona. An example of a wholly artificial, offchannel pond complex. Ponds are numbered sequentially 1 to 6 from top to bottom in this photo. U.S. Bureau of Reclamation, October 2009.

Razorback sucker management in the Lower Colorado River is summarized in a series of papers published over 4 decades (Minckley 1983, Minckley et al. 1991, Marsh et al. 2015, Bestgen et al. 2020). Most activities paralleled those directed toward bonytail, and the 2 species often were cotreated in the same habitats. Razorback sucker was initially brought into captivity in 1974 when a sample of wild fish from Lake Mohave was taken to the federal hatchery at Willow Beach, Arizona, where it was subjected to propagation studies. Initial successes led to a decision to transport, in 1981 and 1982, the first brood stock to the USFWS native fish facility at Dexter, New Mexico, now named Southwestern Native Aquatic Resources and Recovery Center (SNARRC). Hatchery production and rearing were ramped up quickly to support an ambitious program of repatriation to Gila River basin streams facilitated by a memorandum of understanding between Arizona Game and Fish Department and USFWS that provided for 10 years of stocking into central Arizona waters in lieu of listing under the ESA. Despite stocking millions of young fish, the program failed to establish any new populations because longterm survival was absent and no reproduction occurred (Marsh and Brooks 1989, Minckley et al. 1991, Schooley and Marsh 2007). Consequently, the species was listed as endangered in 1991 (USFWS 1991a). At the same time, thousands to tens of thousands of hatcheryproduced fish were distributed to Arizona, California, and Nevada, to aquariums and zoos, and to federal facilities to support broodstock development, propagation and rearing, research, and stocking into captive and “natural” sites. Initial repatriations to the Lower Colorado River channel were in 1980, followed by stockings into lakes Havasu (1986), Mohave (1992), and Mead (1995). Between 1991 and the end of 2022, 725,837 razorback suckers had been stocked: 296,875 into Lake Mohave and lakeside backwaters, 191,370 into Lake Havasu and adjunct habitats, 228,154 into the Lower Colorado River mainstem and its connectives downstream of Lake Havasu, and 9438 into inland impoundments and isolated backwaters with no river access.

Native fish management goals under auspices of the 50year LCR MSCP initiated in 2005 include stocking 620,000 bonytails and 660,000 razorback suckers into the Lower Colorado River, maintenance of a razorback sucker broodstock and genetic refuge (in Lake Mohave), and creation or protection of 360 acres (146 ha) of habitat for bonytail, flannelmouth sucker, and razorback sucker (LCR MSCP 2004). Through federal fiscal year 2022, 131,987 bonytails and 273,676 razorback suckers had been stocked and 158 acres (64 ha) of habitat, referred to as “conservation areas,” had been created or protected (James Stolberg, Bureau of Reclamation, personal communication, January 2023). Stocking implementation is reviewed, evaluated, and updated periodically (e.g., LCR MSCP 2020).

Habitat manipulation to benefit bonytail and razorback sucker has largely been restricted to offchannel connectives and isolated, adjacent waters; no such actions have been implemented in the river channel proper or its reservoirs. Reasons for this include a lack of practical strategies to carry out such activities and limited support from the states, which preferentially support recreational sport rather than native nongame fisheries; the two activities are not compatible (Clarkson et al. 2005). Ephemeral and perennial isolated and open coves along the shorelines of lakes Mohave (e.g., Appendix 2) and Havasu have been modified by deepening, berm construction, netting emplacement, and installation of wind turbines to help maintain suitable temperature and dissolved oxygen (DO). These sites support seasonal growout and lifecycle completion (reproduction and recruitment) and represent sources of fish for repatriation and opportunities for study; several have been used for research on bonytail and razorback population genetics and demography (Appendices 1, 2).

Other manmade sites (e.g., golf course ponds, urban lakes, and stock waters) also have been used for holding or growout, and some are also important study areas. For example, Cibola High Levee Pond (Appendix 1.1), an isolated segment that was cut off when the river was channelized and riprapped in the latter 1960s, represents an important site of observation, experimentation, and discovery (e.g., Mueller et al. 2005). In addition, native fish ponds at Imperial Ponds (Appendix 1.2), initially created for waterfowl and later reconfigured into 6 small lakes, can support razorback sucker and bonytail in either mono or biculture situations where drivers of population dynamics such as survival/mortality, recruitment, growth, etc., can be investigated. Opportunities to examine effects on native fishes of vegetation removal or enhancement, provision of cover, control of avian or other predators, or other factors are available and awaiting application both at Cibola and Imperial Ponds.

Harvest of bonytail and razorback sucker by any means is prohibited under the ESA and state rule, except for scientific purpose and incidental take. Thus, imposition of other, restrictive regulations such as quotas, seasons, sizes, etc., are unnecessary management tools. Neither species is known for its desirability as a recreational species, and neither is especially susceptible to angling, although there are infrequent reports of fishermen bringing in a carcass, usually hooked in the case of bonytail or snagged in the case of razorback sucker, for examination. Angler reports can be an important source of information, and these fish are generally known to those who fish along the Lower Colorado River and its reservoirs, but we suspect most are unwilling to bring in angled fish because they wish to avoid misperceived consequences of harming an endangered species.

Public education is a powerful management tool (e.g., Fien et al. 2001) that can be used to aid in the conservation of bonytail and razorback sucker. Each of the basin states prominently features bonytail, razorback sucker, and other native species on its website, in its licenserelated regulations, and as part of promotional literature and “trinkets.” Information and education (I&E) are integral aspects of state and federal agency services, and programs like the LCR MSCP have dedicated if not wellfunded I&E. A public school program for kindergarten through third grade, “Sharing Tails,” combines a live, inperson or virtual presentation with crafts and learning games to foster knowledge of native fishes and an interest in their conservation. These outreach efforts have been successful in making a segment of the general public aware of bonytail and razorback suckers, but unfortunately, these fishes are still regarded by many as expendable. Many anglers are interested in their preferred sportfish and seem to lack an appreciation and acceptance of species like bonytail and razorback sucker for their unique place and role in the natural world. Resources of time and money directed toward these fish may be perceived by some as wasted when those same resources could have been applied toward management of sport species (Clarkson et al. 2005). Public education toward development of a positive conservation ethic represents our ultimate challenge, and one that must be met to preserve species like Lower Colorado River native fishes and the biodiversity to which they contribute.

Ephemeral Pond Studies

Having a robust estimate of N_e/N_c for Lake Mohave razorback sucker was an important step forward, but it was not clear that a similar relationship of demography and genetic diversity in OCHs could be achieved for either species. To address this knowledge gap, a series of experiments was initiated where reproductive adult bonytails and razorback suckers were stocked separately into isolated, ephemeral backwater ponds adjacent to Lake Mohave (Appendix 2) to quantify population genetic and demographic parameters. Prior to stocking, fish were sexed, PITtagged, and genotyped using microsatellite markers (Dowling et al. 2014, 2019, Osborne et al. 2018). Roughly 100 male and female parents were stocked into 2 isolated backwaters for each species (Appendix 2), bonytail into Nevada Egg and North NineMile (Osborne et al. 2018) and razorback sucker into Arizona Juvenile (AJ) and Dandy (Dowling et al. 2019).

Harvest data (i.e., fish removed at the end of study) from these shortterm (<6month) experiments were available for razorback sucker stocked in AJ and Dandy (2010–2018) and provided definitive survival information as the OCHs dried during lake drawdowns (Dowling et al. 2019). Harvest was considered complete as ponds were completely dewatered and all fish were removed through netting and seining. Across 9 iterations of the experiment, survival was significantly lower for stocked males than for stocked females. The difference in survival between sexes was partially but not entirely attributed to variation in size at release, with males stocked on average at smaller sizes than females. This sexually dimorphic survival was consistent among years even though survival over the experiment varied temporally, within a backwater among years, and spatially, between backwaters.

Lowest harvest percentages were 6% and 20% for males in Dandy (2016) and AJ (2017) respectively, and 16.7% and 30% for females in Dandy (2018) and AJ (2017), respectively. Highest harvest percentages for males were 68% (2014) for AJ and 65% (2010) for Dandy, while female harvest peaked at 75% (2014) in AJ and 84% (2010) in Dandy. Mean harvest percentages across all years between the 2 backwaters were similar despite the yeartoyear variation, with male harvest percentage at 35% and 31% for AJ and Dandy respectively, and female harvest percentage at 53% and 59% for AJ and Dandy, respectively. Remote sensing of PITtagged fish during these experiments provided individualized tracking data for surviving fish in each OCH. This technique was proven effective in contacting 90% to 100% of fish that were later harvested (Dowling et al. 2021). These experiments each occurred over the course of <1 year, so longterm survival and reproductive trends of backwater stocked fish could not be assessed. Results have genetic implications in that skewed sex ratios reduced N_e/N_c and N_b/N_c (N_b = annual effective number of breeders).

Microsatellite markers were characterized for razorback sucker pond samples from 2010 to 2015 and bonytail pond samples from 2014 to 2016. Genotypes from stocked adults and their progeny (Osborne et al. 2018, Dowling et al. 2019) were used to assess maintenance of genetic diversity (e.g., heterozygosity, allelic richness; Appendix 3), determine parentage of recruits, estimate reproductive success and its variance among individual parents (i.e., VRS), and calculate the annual effective number of breeders (N_b), which is a quantity related to N_e (Waples et al. 2011). Progeny were assigned to parents using a statistical algorithm (Jones and Wang 2010), and VRS was calculated for males and females separately (Osborne et al. 2018). For comparison, values are standardized by dividing VRS by mean family size. These data also allowed estimation of VRS for both species.

For bonytail, parentage could be assigned to more than 83%–92% of offspring sampled from North NineMile (2014–2016) and Nevada Egg (2014, 2016) and 73% of individuals sampled from Nevada Egg in 2015. The ratio N_b/N_c was higher for bonytail than for razorback sucker and ranged from 0.72 to 0.86 (Osborne et al. 2018). In 5 of 6 cases, more than 85% of bonytail adults contributed at least 1 offspring to the annual cohort and had correspondingly low VRS. One OCH experienced poor water quality and partial recruitment failure where fewer than 30% of stocked parents contributed to the annual cohort and VRS was relatively high for both sexes (Osborne et al. 2018).

For razorback sucker, the ratio N_b/N_c ranged from 0.09 to 0.68 (unpublished data), which is comparable to N_e/N_c ratios obtained for Lake Mohave (Turner et al. 2007, Dowling et al. 2014). Between 13% and 70% of females contributed at least one offspring to the progeny pool, but reproduction failed in 2 of 5 years in one ephemeral backwater pond. High levels of VRS among ponds and years for razorback sucker are at least partly explained by variation in early mortality of stocked fish. Sexbiased (i.e., male) mortality skews the sex ratio and thereby increases VRS and lowers N_b/N_c. When considered across species, years, and backwaters, experimental results supported the following conclusions: (1) bonytail exhibited higher N_b/N_c and lower VRS than razorback sucker in nearly all cases, (2) razorback suckers performed at least as well in backwater habitats as they did in Lake Mohave in terms of N_b/N_c and VRS, and (3) VRS was attributable to variance in productivity across ponds and years for both species. Rearing in OCHs should preserve legacy diversity at least as well for bonytail as for razorback sucker. However, spatial (e.g., among different backwaters) and temporal (e.g., within backwaters across years) variation in adult survival and reproductive success for both species are still largely unexplained.

Permanent Pond Studies

The shortterm experiments in ephemeral pond OCHs followed progeny for 1 year but not recruitment to the adult population. As envisaged by Minckley et al. (2003), the OCH plan calls for agestructured populations that can supply genetically diverse fish to the river mainstem. Questions remain about how well genetic diversity is retained across the life history to recruitment for both species (addressed for Lake Mohave razorback sucker in Carson et al. [2016]). In response, experimental studies have been expanded to include 2 large, permanently filled OCHs: Yuma Cove backwater at Lake Mohave and Imperial Ponds.

Yuma Cove backwater is deep enough to retain water all year, making it an ideal location for establishing a recruiting population. Razorback suckers (100 of each sex) were stocked in 2013 to track multiyear trends in genetic and demographic parameters. Fish were hatcheryreared from wildcaught larvae and at least 2 years old at stocking. Remote PITtag sensing was conducted nearly continuously for the first 7 years (February 2013 through March 2019; Dowling et al. 2019). These data were used to track individual survival and incorporated with capture and tagging data to estimate population size using markrecapture methods. As with the ephemeral pond studies for AJ and Dandy, poststocking survival was significantly higher for females than for males, even after accounting for differences in release size (Kesner et al. 2019). Razorback suckers stocked in 2013 and known to be alive by February 2014 included 21 of 100 males and 78 of 100 females. Apparent annual survival increased significantly in subsequent years, with 19 males and 75 females surviving through 2017. Additional fish stocked in 2014 and 2015 experienced higher poststocking mortality, with fewer than 10 from each stocking of 100 fish (50 of each sex in each year) surviving beyond the first year postrelease. This likely was due to the reproductive success of the initial stocking cohort in 2013 and 2014. The loss in overall abundance after the 2013 stocking was replaced by natural recruitment by 2016, and the small OCH maintained an adult population of approximately 400 fish through 2021.

Genetic characterization, parentage assignment of larvae and juveniles, and patterns of reproductive success for razorback sucker were completed for Yuma backwater from 2013 to 2015 using microsatellites as discussed above for ephemeral ponds (Dowling et al. 2019). Parentage assignments worked well for the initial year class of 2013 (97% of parents assigned correctly); however, assignments became more difficult in 2014 and 2015 (77% and 84%, respectively). This likely reflects reproduction by recruits (i.e., fish produced in the backwater), with microsatellite markers providing insufficient statistical power to assign parentage due to shared ancestry of stocked adults and related and uncharacterized recruits.

To further evaluate the longerterm dynamics of survivorship and reproduction in these habitats, the 6 Imperial Ponds were each stocked in 2017 with 150 male and 150 female adult bonytails or razorback suckers (total of 3 ponds for each species—900 individuals per species). After the initial stockings, substantial reproduction at varying levels occurred in all bonytail ponds. Parentage analysis using microsatellite markers revealed high (90%–96%) reproductive success and consequently large genetic effective size (N_e = 252, calculated using the sibship assignment method of Jones and Wang [2010]) for bonytail in one pond—consistent with results from ephemeral ponds. Reproduction was more limited in remaining bonytail ponds, but genetic diversity was preserved between parental and progeny generations, and there was no increase in measures of inbreeding (e.g., F_IS, IR; Appendix 3) between adults and progeny. One bonytail pond demonstrated potential issues with using closed habitats for reproduction including limited reproductive contribution of adults, small effective population size, and a subsequent increase in inbreeding measures between parental and progeny generations. Reproduction was limited in razorback sucker ponds for unknown reasons, preventing analyses like those performed for bonytail.

The larger size of ponds and the establishment of agestructured populations at Imperial Ponds pose some additional challenges including sampling sufficient larvae and recruits (i.e., potential parents of subsequent generations) to facilitate parentage analysis and to accurately estimate reproductive success of adults. Alternative measures of genetic “risk” (i.e., increases in inbreeding) could be used to monitor the genetic health of offchannel populations that leverage the power of nextgeneration sequencing (NGS) methods of genetic monitoring (e.g., genotypinginthousands sequencing [GTseq]; Campbell et al. 2015) developed recently for both bonytail and razorback sucker. Baseline values for these measures could be obtained from the time of stocking and thresholds developed that trigger management actions such as movement of fish between habitat complexes. Genomic measures of inbreeding include multilocus heterozygosity (Coltman et al. 1999), internal relatedness (Amos et al. 2001), and relatedness (Wang 2022) (Appendix 3).

Ecological and Demographic Considerations—SelfSustaining and Resilient Populations

A selfsustaining population according to Scott et al. (2005) “should be able to remain stable or increase over time without human assistance to reproduction or dispersal in the wild.” They add that this condition would be violated by frequent translocations to compensate for human barriers to dispersal or losses due to predation, disease, or other mortality factors but not by occasional translocations to maintain genetic diversity, the latter being an expectation of the OCH approach advocated here.

What a selfsustaining population of bonytail or razorback sucker in an OCH might look like depends on the timescale over which the population is being assessed, that is, whether the assessment is based on a shortterm “snapshot” or integrated over many generations spanning decades. In a stable population, the proportion of each age group (cohort, age class, or life stage) at any given time is the same as the proportion of that group over the lifespan of the cohort (e.g., Royce 1984). In an idealized stable population, the dynamic rates of survival/mortality, recruitment, and growth are constant for each age group, and abundances of individual age groups and the population as a whole do fluctuate over time. This idealized stable state is how a population could look if these rates were integrated over many years; however, intrinsic and environmental factors constantly affect populations, causing the ideal state to rarely be seen in the shortterm. Dynamic rates and numbers within age groups may vary dramatically between years and at times even drop to zero, for example, when one or more year classes of these longlived fishes are lost due to reproductive or recruitment failure. Consequently, we cannot define a stable population of bonytail or razorback sucker in OCHs. This is, in part, because stochastic factors affect populations, but also because opportunities for study are limited to a few habitats, and these habitats also must serve both as study opportunities and as production sites, the latter function effectively disrupting any likelihood of stability. Sufficient observational data collected over many years are necessary to estimate dynamic rates and age group abundances or to refine models that describe a selfsustaining population.

Natural systems are considered resilient when they have the capacity to resist and recover from environmental perturbations (e.g., Capdevila et al. 2020). While large populations theoretically are more resilient than smaller ones, other factors play a role in determining their persistence. It is unknown if this intrinsic capacity is compromised in these relatively small populations of at most hundreds to thousands of bonytails or razorback suckers that occupy OCHs compared with historically much larger populations in geographically expansive systems. It is virtually certain that these large populations were not homogeneous with respect to demographic rates over space and time. Catastrophic changes can extirpate a naturally occurring or captive population, regardless of size. However, because of the mitigating effects of management, captive populations are not subject to disturbances such as temporary excursions into unfavorable ranges of physicochemical conditions or disruption resulting from stochastically mediated collapse of one or more year classes. Ponddwelling populations under the OCH concept are expected to vary temporally in abundance and yearclass strength, but we predict that maintenance of sufficient (likely hundreds, but as yet unquantified) numbers of genetically diverse (similar to baseline, also unquantified) reproductive adults will impart demographic resilience resulting in recovery and persistence following disturbance. These fish should be maintained in an unspecified number of OCHs to spread the risk of local extirpation and recruitment failure.

Genetic and Demographic Considerations—Maintenance of Genetic Diversity

Maintenance of genetic diversity within populations remains one of the most important considerations when managing imperiled species and was critical in formulation of the OCH concept applied to razorback sucker and bonytail (Minckley et al. 2003). Approaches for characterization of genetic diversity have evolved rapidly in the 20 years since, with newer molecular approaches allowing finescaled examination of genetic diversity (reviewed in Turner et al. 2020). These methods now are broadly applied in population genetics studies, including for these 2 species, and are providing essential information for conservation and management applications (see above).

While most early studies of genetic diversity were focused on neutral (i.e., nonadaptive) variation, the importance of adaptive genetic traits was recognized long ago (Lynch 1996, Lynch and Lande 1998) and is even more important today (Teixeira and Huber 2021). Consideration of adaptive variation has largely been theoretical; however, NGS methods have provided evidence for local adaptation of populations (e.g., Hess et al. 2013, Fitzpatrick et al. 2015) and identification of adaptive genetic variants (e.g., Hohenlohe et al. 2013). These features are important considerations because they can help explain differences in how populations respond to local environmental effects.

Population size is an important consideration when managing imperiled species due to the wellknown relationships between number of individuals and adaptive potential (Hoffmann et al. 2017) and between fitness and genetic diversity (Reed and Frankham 2003), with increased population size, distribution, and concomitant levels of genetic variation providing greater resiliency. Concerns over small population size and associated effects of inbreeding in hatcheries are well recognized (Wang et al. 2001, Christie et al. 2014) and have been reported for razorback sucker (Dowling et al. 1996) but not for bonytail.

Concerns about performance of hatchery fish and their impacts on wild populations have also been identified (Heath et al. 2003, Fraser 2008, Osborne et al. 2020). For example, Auld et al. (2021) identified divergence in mating cues between wild and hatcheryreared coho salmon (Oncorhynchus kisutch) that could impact wild populations. Also, while hatcheryreared steelhead trout (Oncorhynchus mykiss) grow faster in the hatchery than offspring of wild fish, their survival is lower under wild conditions (Blouin et al. 2021). Concerns like these have led to the development of alternative approaches to management of endangered species, such as the OCH plan, and these are useful across species with diverse life history strategies (Osborne et al. 2020).

The OCH plan requires management of populations of bonytail and razorback sucker in isolated ponds and other appropriate habitats, with the goal that fish in OCHs represent and maintain existing diversity in these species. To mitigate impacts of inbreeding in small, finite populations, Minckley et al. (2003) called for frequent transfer of sexually mature recruits among OCHs and from the riverine population to OCHs (and vice versa) to maximize parental contributions to reproduction in the metapopulation. This approach is key to the plan and is feasible for razorback sucker but is not possible for bonytail under current conditions because survival in the river is virtually nil and sufficient repatriates are not available to be collected. Hence, novel strategies, yet to be identified and developed, are required to manage population genetics of bonytail. Because bonytails mature at a younger age compared to razorback suckers, they may need to be moved more often to obtain similar levels of genetic structuring within and among subpopulations. Note, however, that some differentiation (via genetic drift) is expected among managed populations and is appropriate as long as individuals from these populations retain the ability to interbreed without loss of fitness.

Demographic and genetic processes are strongly linked in population theory (Wright 1969) and in practical implementation of the OCH program. Management actions that address demographic issues like small population size, skewed sex ratio, and high VRS should also positively affect genetic performance. Under ideal conditions, an OCH should harbor as many reproductively capable adults as can be sustained through natural recruitment. The majority of, if not all, reproductive adults should contribute offspring to annual (or generational) cohort(s) of progeny, and an agestructured population should develop after initial stocking. Additionally, a genetically representative progeny should survive to recruit to the population at a rate that sustains adult numbers and genetic diversity. In reality, the number of fish that can be sustained depends on environmental factors like habitat size, water physicochemistry, and temporal variance in habitat quality. Demographic factors are important determinants of population size and levels of genetic diversity maintained in a particular OCH population. Environmental and demographic factors thus are interconnected drivers of population abundance and genetic diversity. Metrics associated with these factors track how close an actual OCH is to the theoretical ideal or desired target conditions.

A key metric that records the combined influences of environmental and demographic factors on genetic diversity is VRS. VRS is inversely proportional to the ratio N_e/N_c or N_b/N_c, where N_c is the number of reproductively capable adults in an OCH. When VRS is relatively low (i.e., near minimum values of VRS that follow a Poisson distribution with x^– = 1 offspring per parent per generation), the expected ratio of N_e/N_c ≈ 1 (Nunney and Elam 1994). The biological interpretation is that most, if not all, adults contribute offspring to the next generation or cohort (in the case of N_b). For annual species, N_e = N_b, and contributions to the offspring pool are inclusive and evenly distributed across parents when VRS is low. Bonytail and razorback sucker are characterized by agestructured populations with overlapping generations where N_e ≠ N_b (Waples et al. 2011). When measured across generations, low VRS implies that nearly all reproductively capable adults have contributed to the progeny pool at least once over their lifetimes.

VRS can take on very high values in some fishes, including bonytail and razorback sucker. Both species are longlived, attain relatively large body sizes, and devote considerable mass and energy to reproduction (Charnov et al. 2001). Females produce thousands of small eggs, and a single mating pair can potentially contribute all or most offspring to the progeny pool. When one or a few mating pairs contribute the majority of offspring that recruit to the next generation, VRS becomes very large, and N_e/N_c is much smaller than 1. This means that the OCH population harbors much less genetic diversity than ideal conditions.

As discussed above, VRS in OCHs was low for bonytail, with the majority of individuals contributing at least one offspring (Osborne et al. 2018). Conversely, standardized VRS in razorback sucker differed greatly across 2 OCHs tested, and across years within 1 OCH. VRS was particularly high among females, and N_b/N_c was correspondingly low (Dowling et al. 2019, unpublished data). Differences in female body size, skipped spawning, costs of previous reproduction, or poor water quality may explain annual VRS (i.e., VRS measured for parents of an annual cohort), but more study is needed to identify specific causes. Whatever the cause, VRS could be lower when measured over a generation (instead of annually) as more females are expected to successfully contribute to the offspring pool over time.

Based on these results, there are 2 key inferences for management. Bonytail appears to perform close to ideally as long as water quality and other environmental conditions are favorable. Razorback sucker, on the other hand, may need multiple years of spawning to ensure that most parents contribute to the generational progeny pool to attain the same level of genetic representation per parent compared with bonytail. It also is likely that skipped spawning, higherthanexpected male mortality, or costs of previous reproduction contribute to the higher VRS observed in razorback sucker. For permanent OCHs, management action should be considered when N_b/N_c (in bonytail) or N_e/N_c (razorback sucker) is small (e.g., <0.2). Environmental factors such as poor water quality have led to adult mortality and recruitment failure in one or more backwaters. If poor water quality is chronically observed, the OCH should be taken offline to avoid negative effects on the total effective size of the collection of OCHs. When demographic factors are suspected, such as densitydependent changes in fecundity or age and sizerelated mortality, then additional stocking and monitoring of reproduction could be conducted.

OCH Stocking, Harvest, Removal, and Transfer

The total number of OCHs required to meet program goals depends on variance in productivity among OCHs and desired level of genetic diversity measured in terms of N_e across all OCHs. To enhance performance of the OCH program as a whole, Minckley et al. (2003) described the need for periodic exchange of adult individuals between OCH populations and the river mainstem, and across OCHs. However, the plan did not specify when, or at what rate, individuals should be exchanged. Transfer of adults from an OCH to the mainstem bolsters riverine populations and prevents exceedance of carrying capacity in donor OCHs. The number and types (i.e., size class, age, or sex) of fish transferred out of an OCH depends on adult densities, yearclass strength, and sex ratio. This is a complicated issue as illustrated below. Generally, when densities in a donor OCH are high, more adult fish should be transferred out. For example, adult fish from a dominant cohort may be preferentially selected for transfer to maintain the desired age structure in the donor OCH population or to retain the original, unrelated adults for the longlived razorback sucker. Likewise, it is prudent to transfer more females out of the population if sex ratio in a donor OCH is strongly female biased. This action restores a 1:1 sex ratio that is optimal for reducing VRS, increasing N_e/N_c, and reducing loss of genetic diversity. Because of low survival, exchange of bonytails with the river is illadvised. Fish from OCHs could be used to establish new populations, support repatriation efforts into habitats where higher survival is expected, or for other conservation uses. In contrast, the plan as designed should work well for razorback sucker.

Demographic and genetic exchange of individuals across OCHs also achieves important benefits to overall performance in terms of maintenance and enhancement of genetic diversity. For example, low spawner densities relative to carrying capacity, age structure imbalance, and skewed sex ratios are important reasons, in theory (e.g., Nunney 1996), to transfer adult fish from donor OCHs or the river mainstem to a recipient OCH to achieve sustained recruitment and productivity. Periodic transfer of adults across OCHs simultaneously ameliorates effects of local genetic differentiation and losses of genetic diversity and fitness due to local genetic drift and inbreeding. In theory, as few as 1 adult migrant per generation is sufficient to counter differentiation and inbreeding effects (Whiteley et al. 2015). In practice, values of N_b/N_c (bonytail) or N_e/N_c (razorback sucker) <0.2 indicate that fish should be moved at a higher rate because low values indicate that risks of detrimental genetic effects are correspondingly high (Osborne et al. 2018, Dowling et al. 2019). Actual numbers, rates, and demographic characteristics of transferees should be determined based on regular demographic and genetic monitoring of OCH and river mainstem populations (Tallmon et al. 2004). Monitoring data would be translated to an exchange schedule that meets specific demographic and genetic goals to be developed for a particular OCH, and for the collection of OCHs, in an adaptive management context. Costs of demographic and genetic monitoring and population manipulations should be lowered as initial sampling, analysis, and exchange schedules are refined and optimized.

An Example from Razorback Sucker in Yuma Cove Backwater

An example of applying this approach comes from conservation efforts in Yuma Cove backwater. This habitat was fishless when 200 razorback suckers were stocked in 2013. Since then, a selfsustaining population has been observed and monitored in the backwater. In that time, the site has provided a realworld example of the tradeoffs between OCH size, population size, and maintenance of genetic diversity. Through a combination of annual sampling and remote PITtag sensing, demographic and genetic trends in a selfsustaining population of razorback sucker were monitored (Dowling et al. 2021). Within a few years of initial stocking, the population had reached more than 400 adults and hundreds to thousands of youngofyear (Dowling et al. 2021). Poor survival of stocked fish in 2014 and 2015, along with negligible survival of youngofyear fish, suggested resource limitation. An initial attempt to reduce overall biomass was conducted in autumn 2019 when 80 adult and 2877 youngofyear razorback suckers were removed. An additional 100 razorback suckers were stocked the following spring to increase genetic diversity. Twentytwo of the 100 fish stocked were observed alive via remote PITtag sensing through 2021, indicating that poststocking survival of these fish was lower than survival observed for the initial stocking in 2013, but higher than stockings in 2014 and 2015—a small but significant indicator that resources were not being overused and thus limiting, at least temporarily.

The apparent overabundance of razorback sucker in Yuma Cove backwater provided an opportunity to implement a transfer of razorback sucker from an OCH and measure its impact on survival in OCH populations. This test removal relied on size classes to identify firstgeneration recruits from second or thirdgeneration recruits, targeting the transfer of smaller adults that were likely second and thirdgeneration fish. It was considered more desirable to retain originally stocked fish and firstgeneration recruits than more recent (i.e., younger) fish in the OCH because it was assumed these individuals would have greater genetic diversity than offspring of those firstgeneration recruits. As the population matures, the size classes of different generations will overlap, and it will become impossible to selectively harvest the secondplus generations based on size alone. A general OCH harvest plan will need to rely on both demographic and genetic information and modeling. The plan would need to specify a method to estimate the resultant reduction in overall biomass and the impact those removals would have on survival, spawning biomass, and genetic diversity.

The Offchannel Habitat Concept and Species Recovery

According to the ESA, recovery of a threatened or endangered species is attained when protections of the act are no longer required (16 U.S.C. 15311544). For some species, recovery is not achievable, but for others, specific quantitative biological criteria are defined to determine when delisting or downlisting is appropriate. Recovery metrics may be revised and refined as new scientific information is gathered, but revision and refinement may take decades despite available information (e.g., Propst 2020).

Bonytail was listed as endangered in 1980 (USFWS 1980). A recovery plan was published in 1987, updated in 1990, and amended in 2002 (USFWS 1987b, 1990, and 2002a, respectively). The status of bonytail in 1990 was endangered, and it was described as “very rare” with very low to no recruitment. The immediate recovery goal at the time was extinction prevention; longerterm goals leading to down and delisting, along with appropriate criteria, were not addressed, but 2003 was given as a recovery date for bonytail in the Upper Colorado River basin. That date had come and gone by the time a recovery plan amendment and supplement were published; the species was still endangered, and only a few wild adults remained. The recovery objective, as in all such plans, was down and delisting. Quantitative criteria required establishment of “genetically and demographically viable, selfsustaining populations.” In 2002, downlisting was considered possible by 2015 and delisting by 2023. No naturally recruiting populations of bonytail exist today so the species remains critically endangered.

Razorback sucker was listed as endangered in 1991 (USFWS 1991a). A recovery plan was published in 1998 (USFWS 1998) and amended in 2002 with addition of quantitative goals and criteria (USFWS 2002b). These plans are summarized here and below. Status of razorback sucker in 1998 was endangered, with the species occurring only in remnant populations. Four years later in 2002, it remained endangered and was described as “in serious jeopardy.” In 1998, downlisting was considered possible by 2010 and delisting as soon as 2020. Little changed and in 2002 dates for downlisting and delisting were estimated as 2020 and 2023, respectively. Both the 1998 and 2002 criteria required populations to be genetically and demographically viable and selfsustaining. No clear statement of the overall status of razorback suckers can be gleaned from the species' status assessment (USFWS 2018); however, the document suggests qualitative improvement in the condition of existing populations from historical (i.e., 1988) to current (2018) times in both Upper and Lower Colorado River basins. Nonetheless the species inarguably remains biologically imperiled and endangered despite approaching a half century of directed management.

No population of either bonytail or razorback sucker today meets the minimum recovery criterion of being genetically and demographically viable and selfsustaining. While progress has been made, bonytail is functionally extinct in nature. No “wild” razorback sucker population today is known to be selfsustaining, with most populations dependent upon stocking of hatcheryproduced fish. Established populations of predatory and competitive nonnative species continue to thrive and expand, while new species may appear at any time (Propst et al. 2020). In reality, neither bonytail nor razorback sucker is likely to ever meet current recovery criteria, and both species are destined to remain conservation reliant (Goble et al. 2012) until or unless novel and effective management strategies are developed and fully implemented to overcome currently intractable threats. Stocking of hatchery produced, or wildproduced and hatcheryreared (in Lake Mohave) individuals will continue as primary management actions, but no data are available to suggest that these efforts will lead to establishment of selfsustaining populations. Implementation of the OCH concept would result in viable, selfsustaining populations that would not be possible in open waters. These captive populations, although requiring management to maintain diversity and desired age group structure, would provide a continuous source of genetically diverse fish for exchange with natural waters. Although populations of bonytail and razorback sucker have been successfully established in several existing OCHs (Appendices 1, 2), these add relatively few individuals to the total number of fish in the Lower Colorado River. Thus, the need for continued implementation of the OCH concept is urgent. This scenario may be as close as we can get to “recovery” in the foreseeable future.

Selection of Offchannel Habitats

When the OCH concept was originally proposed, Cibola HLP provided one of the few practical examples of if and how it would work. Since then, there have been several attempts to implement the concept, with varied success. In general, the concept itself has proven sound, but the devil has been in the details. Larger ponds, as predicted by Minckley et al. (2003), have been more difficult to manage, and thus far no selfsustaining population of bonytail or razorback sucker has established in an OCH pond >5 surface acres. Unfortunately, 5 surface acres may not be large enough to contain a population that can maintain both a stable age structure and genetic diversity if such ponds rapidly overpopulate (e.g., 1 to 2 years after establishment).

OCH size is not the only factor that has impacted the effectiveness of larger ponds. Another consideration is the capability to drain a site, but incorporation of this feature appears at odds with the suggested water supply in the OCH concept. A passive water supply, through either subsurface flow or through porous shoreline substrates, is preferable to river water pumped through a filter because the latter cannot reliably preclude direct invasion of nonnatives. However, presence of passive flow can make it difficult or logistically impossible to fully drain and chemically renovate an OCH. Further, the proximity to the water source required to allow passive flow may increase the risk of nonnative introduction through direct transfer. Finally, both passive flowthrough and active filtered pumping do not assist in water temperature amelioration that can be necessary to prevent fish kills during the hottest summer months. Well water is the only secure source that can effectively lower water temperatures and prevent incursion of nonnative species, but this source is dependent on availability of reliable electrical power and on humans to ensure timely water delivery.

Use of design features such as boulders to provide cover for bonytail and gravel spawning substrate for razorback sucker generally have been supported by data collected at OCH research sites. Riprap was used readily by bonytail in Cibola HLP, but boulders emplaced for that purpose in the Imperial Ponds were not as useful because interstitial spaces provided by the thin veneer of rock were too small and quickly filled in with sediment. Recruitment of razorback sucker in Imperial Ponds followed the addition of spawning substrates, and remote PIT scanning on those substrates indicated their use during the known spawning period for razorback sucker. However, recruitment at Imperial Ponds has not reached levels seen in smaller isolated habitats such as Cibola HLP and Yuma Cove backwater. Nonnative fish establishment in Imperial Ponds also likely contributed significantly to failed recruitment, but pond size, complexity, depth, and shoreline profile all may contribute as well. It is difficult to isolate the impact of each factor without the appropriate experiments, which would require a larger number of OCHs than are available at this time.

Development and creation of OCHs is only part of the solution, as not all such habitats work equally well for these purposes. Offchannel ponds along the shores of Lake Mohave (e.g., Appendix 2) in the past have been used to rear bonytail and currently are used to rear razorback sucker to improve their growth, with varying levels of success (Minckley and Thorsen 2007, Ward et al. 2007). As discussed above, there was variation in reproduction and survivorship among ponds and years in studies of reproductive success in bonytail and razorback sucker (Osborne et al. 2018, Dowling et al. 2019). There also was considerable variation in adult survival and reproductive success among the 6 habitats at Imperial Ponds (J. Lantow, Bureau of Reclamation, personal communication). For bonytail, reproduction and recruitment regularly occurred in 2 ponds (2 and 6; photo 4); however, fish kills occurred in 2020 (pond 5) and 2022 (pond 2). Razorback sucker reproduction and recruitment have occurred consistently in Yuma Cove backwater since 2013; however, this has been true for only 1 of the 3 Imperial Ponds occupied by this species, and there were significant fish kills in 2 ponds in summer 2022. It is unclear whether these results would differ if bonytails were stocked into ponds where razorback sucker mortality occurs or vice versa.

Given the high probability that at least some ponds developed for the OCH program will be unsuitable for their intended purpose, it would be valuable to develop tools that allowed identification of OCHs that are likely to be useful before limited resources and effort are invested. This would require detailed study of biotic and abiotic conditions of ponds, allowing comparison of successful and failed ponds to identify conditions that are important determinants of survival, reproduction, and recruitment of these species. Unfortunately, this approach is fraught with difficulty because it likely is a combination of abiotic and biotic features that causes failure of a pond, and it would be difficult to tease out the important factors experimentally. Individual OCH quality is likely to vary from year to year, so this regular monitoring of OCHs is essential.

Development of NGS methods and their application to ecological studies provide a possible solution to this problem because this methodology has been used to characterize biotic communities and their response to abiotic and biotic effects. Several studies have used metabarcoding of water and/or sediment samples. Metabarcoding is an NGSbased method where DNA or RNA is sequenced to identify many specific taxa in a water or sediment sample to assess the role of various organisms as bioindicators of habitat quality. Ankley et al. (2022) used these methods in experimental freshwater mesocosms exposed to an oil spill and monitored the change in the zooplankton community. They found that change in zooplankton species diversity from 3 days prior to an oil spill to 38 days after was a useful indicator for characterizing the ecological response of communities to environmental changes. Emilson et al. (2017) conducted a survey of macroinvertebrates across environmental stream gradients and found that DNA metabarcoding provided equivalent results to traditional morphological approaches and concluded that barcoding of invertebrates could provide a useful tool for broadscale biomonitoring across watersheds.

Biomonitoring of bacterial communities provides an alternative approach for the purposes of assessing the quality of OCHs used to establish selfsustaining populations of bonytail and razorback sucker. Aylagas et al. (2021) assessed the use of DNAbased monitoring of bacterial communities in aquatic sediments, finding that bacteria can function as bioindicators of environmental stressors. Pilgrim et al. (2022) conducted similar characterization of the periphytonassociated bacterial community among localities, identifying bacterial taxa that were associated with changes in total phosphorus and nitrogen concentrations. These results highlight the potential value of bacteria as bioindicators for characterizing environmental differences among OCHs, possibly allowing identification of environmental factors associated with failure and suitable habitats for use in this program.

Filling the Gaps—Demographic and Ecological Information Needed for Implementation

When fully implemented, the program will contain many OCHs integrated into a large metapopulation. Management goals will focus on enhancing demographic and genetic performance in individual OCHs and for the network as a whole. Yet almost everything we know about offchannel habitats comes from study of the relatively few ephemeral and permanent habitats described above. From that information, we used metapopulation theory and intuition to propose how the collective or integrated OCH program will perform, but this proposal will need testing through rigorous monitoring of environmental and demographic factors within each OCH. The integrated OCH program will consist of many individual OCHs, each ostensibly managed to maximize abundance, persistence, and genetic diversity. Metapopulation theory tells us that abundance and genetic diversity are maximized for the program as a whole when OCHs produce roughly equal numbers of recruits per spawner per generation (Nunney 1996), the adult population is stable and agestructured (Nunney 1993), and density is maintained just under carrying capacity through time. In a given year, some habitats will be more productive than others, and some might fail entirely with local extirpation. High variance in productivity and high extirpation risk in space and time diminish the number of adults and genetic diversity in the program as a whole (e.g., Vucetich et al. 1997). Therefore, identifying, monitoring, and mitigating ecological factors that drive differences in productivity among individual OCHs should ultimately lead to more adult fish, higher genetic diversity, and better recruitment for the program as a whole. Individual OCHs that consistently perform poorly should be removed from the program, preferably in concert with introduction of consistently productive OCHs into the program. Generally speaking, the number of OCHs needed to meet program goals must be increased when variability in productivity and extirpation risk across individual OCHs is high.

While population genetic parameters are relatively easy to quantify, monitoring of demographic factors among adults can only be done readily through remote PITtag scanning technology. This approach contributes little to our knowledge of direct sources of larval and juvenile fish mortality and other ecological factors that underlie recruitment success in individual OCHs. We do not know the nature of density dependence and its influence over the entire life cycle of these fishes, particularly on recruitment success. We also do not know how natural and anthropogenic selection operate in individual OCHs. Parentagebased tagging approaches, such as those employed in the management of Pacific salmon (e.g., Steele et al. 2019), should be employed to monitor VRS and to test for nonrandom survivorship of particular genotypes in individual OCHs, especially when compared genetically to natural and hatchery populations throughout the Colorado River basin.

Preliminary observations at Imperial Ponds and Yuma Cove backwater on Lake Mohave raised a number of immediate issues and questions, including some of those identified in Minckley et al. (2003), that should be addressed to maximize the productivity of OCHs. These include (1) cannibalism of eggs/larvae in bonytail and what habitat modifications can be made to ponds to reduce this impact, (2) densitydependent effects in confined backwater habitats and how these function in determining carrying capacity, (3) variation in function among OCHs (e.g., differences in bonytail and razorback sucker reproduction between sites at Imperial Ponds, razorback sucker in ephemeral ponds), and (4) periodic loss of entire populations due to waterquality issues in OCHs both at Lake Mohave and Imperial Ponds. The last requires either redundancy in the number of ponds so systemwide numbers of adults and progeny can be maintained despite such losses, or a need for reliable surveillance at least of temperature and dissolved oxygen so that losses can be mitigated or eliminated entirely.

TABLE 2.

Recovery goals for bonytail (USFWS 1987b, 1990, 2002a).

TABLE 3.

Recovery goals for razorback sucker (USFWS 1998, 2002b).

Evaluation of Goals and Outcomes

LCR MSCP conservation goals and objectives are quantitatively defined as number of surface acres of water and number of fish stocked. These are easily measured and tracked, and the program appears on a trajectory that will arrive at its designated endpoints in a timely manner (e.g., LCR MSCP 2021). Recovery goals and objectives quantified as population criteria (Tables 2, 3) are another matter entirely. Assessing progress requires estimation of population census size, and progress toward attainment has been slow. Goals promulgated in the earliest recovery plans for bonytail and razorback sucker were qualitative and called for prevention of species extinction. Bonytail has been maintained in captivity and there is an ongoing hatcherybased stocking program that has failed to establish any wild population; thus, quantitative recovery criteria are meaningless for this species. The Lake Mohave Native Fishes Work Group set its sights on restoring a population of 50,000 adult razorback suckers in Lake Mohave that was genetically representative of the historical population in that reservoir (Mueller 1995, Dowling et al. 1996), and that goal was among the downlisting criteria in the 1998 but not the 2002 recovery plan. Intensive effort over more than 3 decades has succeeded in maintaining a population of about 10% of that goal (Marsh et al. 2015, Saucier et al. 2022), and genetic diversity thus far has been maintained (Dowling et al. 2014). Establishment of one or more additional populations of similar size in the Lower Colorado River downstream of Lake Mohave would be a serendipitous outcome of the LSCMSCP, but those fish currently are derived from hatchery production and do not have the same high level of genetic diversity as the Mohave stock. Given the virtually limitless availability of larvae from Lake Mohave, the program might consider exclusive use of these fish in lieu of artificial propagation.

Assessing success of an OCHbased conservation program as envisioned here also requires quantitative studies to estimate key population parameters discussed earlier (at the least numerical abundance and recruitment) and genetic characteristics that can be compared with historical information and with current goals and objectives. Genetic tools have been used to characterize several OCH populations and have provided insight into important genetic and demographic parameters that are essential for evaluating the success of individual OCHs (Dowling et al. 2014, 2019, 2021, Osborne et al. 2018). Especially important are estimates of N_e and N_b generated for both bonytail (Hedrick et al. 2000, Osborne et al. 2018) and razorback sucker (Turner et al. 2007, Dowling et al. 2014), and estimates of relatedness (Appendix 3), which have yet to be generated. This information is especially valuable in a longterm monitoring situation where population census data also are available (Dowling et al. 2014) and will allow assessment of the impact of management actions on genetic and demographic parameters.

A critical aspect of the OCH program is adaptive management, or a periodic assessment that (1) provides data to compare with goals and objectives as described earlier and (2) serves also as the feedback mechanism for positive change. For example, NGS approaches have been developed to provide more accurate estimates of genetic diversity parameters (see Appendix 3), providing information necessary for the maintenance of selfsustaining populations that are viable in the long term. Such metrics will be used to guide management actions, such as harvest of individuals from ponds as they become overpopulated and movement of individuals among OCHs to maintain genetic diversity and decrease relatedness of individuals.

Importance of Redundancy

For the OCH program to be successfully implemented and to continue to preserve the genetic integrity of razorback sucker, redundancy must be an integral part of the system. This is important because experience has shown that individual OCHs are subject to failure (i.e., loss of contained fish) for several reasons, and situations can easily be envisioned in which multiple OCHs might be lost at the same time and for similar reasons. Redundancy, therefore, must extend beyond simply having multiple OCHs at individual conservations areas (e.g., there are 6 [soon to be 7] sites at Imperial Ponds, and 2 areas under development are to have 4 and 7 sites, respectively; J. Stolberg, Bureau of Reclamation, personal communication). The LCR MSCP planning area extends along the Lower Colorado River corridor approximately 450 miles (724 km) from Separation Rapids at the head of Lake Mead to the international boundary with Mexico, and along this path the stream flows through reaches of variable geology, hydrology, local climate, plant and animal community type, land use, and political jurisdiction. To achieve redundancy, OCH sites also must be distributed throughout the system to reflect this inherent variability, both to protect existing genetic diversity and to provide a system in which selection can proceed along a natural trajectory.

Conservation Reliance – The Final Outcome

Aggressive management of razorback sucker in the Lower Colorado River began more than 4 decades ago in the early 1980s (Minckley et al. 1991). In that time, the Lake Mohave population of razorback sucker has dwindled from tens of thousands of wild fish to about 5000 repatriates, and bonytail has become functionally extinct in the wild. Neither species recruits naturally in historical habitats, and persistence of both is entirely dependent on stocking. In context, the LCR MSCP has a finite life of 50 years, about the life span of a razorback sucker and twice that of a bonytail. The program commenced in 2005 and will end in 2055 (at this writing into year 18 with 32 remaining). Valuable lessons have been learned about these fishes, and there is enough understanding of their respective life histories to enable managers to perpetuate them, at least in the relatively short term. Unfortunately, the primary factors responsible for their imperiled status—presence of predatory nonnative fishes and habitat alteration—remain in full force with no likelihood of amelioration. It therefore is clear that bonytail and razorback sucker will be dependent on human intervention for their continued survival as conservationreliant species. Unlike the LCR MSCP, the ESA will be in effect in perpetuity unless eliminated by Congress, and bonytail and razorback sucker will continue to have protections and conservation management under its authority. And, as the ESA was the root of the legal requirement under which the current LCR MSCP was created, so may a future iteration of the MSCP be created under which endangered species conservation may be continued until recovery is achieved, politically if not biologically, or the 2 species wink out of existence.