The need for constructing endangered fish habitats that conform to climate‐driven flow changes in a western U.S. river

Warmwater fish habitat in the San Juan River of the southwestern United States has been reduced by over 30% as a result of water depletion, reservoir inundation, and cold‐water dam releases combined with drought‐related changes in hydrology. This reduction and a suite of other factors have contributed to declines in native fish populations including the federally endangered Colorado Pikeminnow (Ptychocheilus lucius) and Razorback Sucker (Xyrauchen texanus). Conservation efforts for these species include determining flow needs; protecting, managing, and augmenting habitats; and stocking hatchery fish. But the young of stocked fish have low survival due largely to a paucity of nursery habitat not being reformed and maintained under current conditions. Flow recommendations for Navajo Dam releases designed to mimic the river's natural hydrograph have not been met due to water shortages, and the desired outcomes of increased channel complexity and enhanced fish habitat have not been observed. Forecasted hydrology that includes ongoing drought shows that achieving the flow targets through further dam reoperations is unlikely. Mechanical construction of early life‐stage habitats is a highly recommended complement to flow management for offsetting the effects of flow reduction and habitat loss. Habitats with features that are effective and resilient under a range of flows are important in counterbalancing the effects of climate change.

and constructed fish habitats (Beechie et al., 2012;Jaeger et al., 2014).Declining catchment runoff in regulated rivers could further reduce the effectiveness of environmental flows that have lower priority in most years (Pennock et al., 2022).These conditions have prompted the need to design and mechanically construct sites at varying elevations that inundate at reduced and managed flows as part of an overall conservation strategy for imperiled fish species (BioWest, 2005;Chen & Olden, 2017;Large & Petts, 1994;Valdez et al., 2019).
The San Juan River of the southwestern U.S. has undergone substantial geomorphic and hydrologic changes with consequent effects on ecological processes and native fish populations.The San Juan is a major tributary of the Colorado River that forms one of the three major subbasins of the Upper Colorado River Basin (upper basin; Figure 1), contributing about 15% of flow volume (Bennett et al., 2019).Dams and water diversions in the San Juan control floods and provide irrigation and domestic water supplies such that present water uses combined with climate change have resulted in annual flow reductions of about 30% below historical levels (Bennett et al., 2019).These flow reductions have led to channel narrowing, reduced in-channel fish habitats, and reduced lateral connectivity with floodplains.Also, reservoir inundation of the lower reach and cold hypolimnetic dam releases in the upper reach have left only about 362 km of warmwater fish habitat that supports small

Research Impact Statement
Ongoing drought in western U.S. rivers has reduced the efficacy of fish habitats, raising the need for large-scale construction of habitats that conform to climate-driven flow changes.
F I G U R E 1 Map of the Colorado River and the three major subbasins of the upper basin.
The San Juan River Basin Recovery Implementation Program (SJRIP) was established in 1992 to protect and recover the CPM and RBS in the San Juan River, while water development proceeds in compliance with applicable federal and state laws (www.coloradori verre covery.org).
Conservation of these species includes research to determine species flow needs, and protection, management, and augmentation of habitats (SJRIP, 2016).These conservation efforts have struggled to address the flow-to-habitat relationships that provide river flows necessary to sustain various life stages of these species.Flow recommendations in 1999 formulated releases from Navajo Dam to mimic the natural hydrograph for the benefit of the endangered and native fish species in the available warmwater habitat (Holden, 1999).In the ensuing 15 years, these flow targets were not regularly met, and the expected levels of channel complexity and enhanced native fish habitat were not observed.
Subsequent evaluation of the flow recommendations showed that further modifying dam releases could slightly increase the ability to meet the flow targets, but it was unlikely that the recommendations could be met with the existing and forecasted hydrology under ongoing drought and climate change conditions (SJRIP, 2018a).
Seven years of research (1991)(1992)(1993)(1994)(1995)(1996)(1997) found low numbers of wild CPM and no RBS in the San Juan River (Holden, 2000), and hatchery fish have been stocked since 1994 to augment the populations.Some stocked fish are reproducing but survival of the young of both species is low (USFWS, 2018(USFWS, , 2020) ) largely because of a paucity of nursery habitat that is not being reformed and maintained by present flows (Barkalow et al., 2020;Farrington et al., 2020).To augment and enhance fish habitats, the SJRIP is designing and constructing riverine features that will support early life stages of the species (Bliesner & Lamarra, 2007;Keller-Bliesner Engineering, 2018).This paper explores the historical and present condition of fish habitats in the San Juan River and proposes large-scale construction of early life-stage habitat, with sites designed to become inundated at available river flows, as one tool to support conservation of CPM and RBS.We compare the San Juan River with other river reaches used by CPM and RBS to contextualize habitat design and construction appropriate for the contemporary San Juan.A number of factors contribute to low survival and recruitment of CPM and RBS in the San Juan River-including low numbers of spawning adults (Diver & Wilson, 2018), encroachment of fish habitats by non-native vegetation (Bassett, 2015), predation and competition from non-native fishes (Franssen et al., 2015), high emigration of young (Cathcart et al., 2017(Cathcart et al., , 2018)), and a paucity of available rearing habitat (Farrington et al., 2020)-all of which are being explored by the SJRIP (SJRIP, 2016).We recommend that mechanical construction of early life-stage habitats be implemented to complement flow management in the San Juan River given the present geomorphic state of the river and the reduced flow from ongoing drought.

| Historical zoogeographical setting
Understanding the zoogeographical setting of the Colorado River is important in understanding the historical and contemporary ecological roles of the principal tributaries and their habitats in the life histories of the CPM and RBS.Geological evidence suggests that the Upper Colorado River has been in its present course for more than 5 million years and historically flowed into one or more closed basins south of the upper end of present-day Grand Canyon (Luchitta, 1990).About 5 million years ago (late Miocene/early Pliocene), the river began carving its way through the Colorado Plateau forming the Grand Canyon, and the upper basin joined with the more dispersed streams of the lower basin in the last 2-3 million years (McKee et al., 1967).Fish species that evolved primarily in the upper basin (including the CPM) were mostly large riverine forms, and those that evolved in the lower basin (likely the RBS) were associated with small streams and lacustrine habitats (Miller, 1961;Minckley et al., 1986).

| Colorado Pikeminnow
Archeological evidence suggests that the CPM evolved as a highly migratory piscivore that ranged from tributaries of the upper basin as far downriver as the estuarian zone in the Gulf of California (Miller, 1961).Biochemical genetics show evidence of multiple stocks of CPM in tributaries of the upper basin with genetic panmixia indicating constant interchange of genes among the stocks or subpopulations (Morizot et al., 2002).This hypothesis of multiple interconnected stocks is supported by long-distance movements and strong fidelity of adults in contemporary subpopulations to particular tributaries and reflects the development of powerful selection mechanisms that evolved over thousands of years (Irving & Modde, 2000;Tyus, 1990).
The CPM that used the San Juan River historically were likely one stock that inhabited the mainstem Colorado River through Glen Canyon and spawned in the San Juan, similar to the portion of the contemporary Green River population that resides in the mainstem and spawns in the tributary Yampa River (Bestgen et al., 2007).The disruption of migratory pathways by dams and the loss of large reaches of habitat to reservoir inundation starting with Hoover Dam in 1935 likely triggered the decline of the San Juan stock of CPM, as well as others in the lower basin.The few remaining wild fish in the San Juan River likely represent a remnant stock of fish persisting as long-lived adults with periodic successful reproduction, but these fish presently lack the full complement of habitats otherwise found in rivers with self-sustaining populations (USFWS, 2020).

| Razorback Sucker
Contemporary collections and taxonomic analyses of RBS indicate historical long-term residence in interconnected lake habitats formed by lava flow dams in western Grand Canyon in the last 1.8 million years and by melting Pleistocene glaciers in the last 12,000-15,000 years (Hubbs & Miller, 1953;McKee et al., 1967;Minckley et al., 1986).At least 13 major lava dams impounded the Colorado River through the Grand Canyon with some persisting at least 250,000 years before eroding (Rugg & Austin, 1998).These dams created a series of large lakes, with the most recent ~400,000 years ago.Shoreline deposits show that the largest dam at Lava Falls Rapid impounded a lake 90 m above the high-water level of present-day Lake Powell and backed water for a distance of over 480 km to Moab, Utah (Duffield, 1997).
The evolution of the RBS is likely closely linked to these early and extensive lacustrine habitats of the lower basin.The present diversity of mitochondrial DNA in RBS from Lake Mohave is remarkably high, indicating that the contemporary population is comprised of direct descendants of a large, diverse, panmictic population that historically inhabited the lower basin.Whereas populations in the upper basin are less diverse and possess fewer unique genotypes, suggesting that the RBS thrived in the early lakes and abundant floodplains of the lower basin before dispersing to other parts of the Colorado River system (Dowling et al., 1996).
The evolutionary history of the RBS helps to explain its contemporary predisposition to riverside floodplains and lacustrine environments.
The species has been able to survive and reproduce in artificial reservoirs in the last 85 years (lakes Mohave, Havasu, and Mead) before experiencing sharp declines due to predation of young by non-native fishes and a lack of early life-stage habitats (USFWS, 2018).The evolutionary and genetic evidence indicate that the RBS requires floodplain-like nursery habitats that are absent from the contemporary San Juan River.
The numbers of young RBS hatched in the San Juan River are low likely in part to a paucity of nursery floodplain habitat.However, there is some recruitment in low water years when low velocities and warm water temperatures provide some shallow-water habitats (Barkalow et al., 2020).
Spawning habitat may also be limited for the RBS in the San Juan River.A high level of hybridization with the Flannelmouth Sucker (Catostomus latipinnis) (23.7% of larvae and 100% of juveniles) may be due to small numbers of RBS spawning with aggregations of Flannelmouth Sucker in limited habitat (Mussmann et al., 2022).Notably, survival of the young hybrids appears higher than that of young RBS (Mussmann et al., 2022), indicating different and distinct habitat needs.

| The contemporary San Juan River
Changes in land and water uses in the last century have reduced overall San Juan River flows and narrowed and simplified the river channel (Bennett et al., 2018(Bennett et al., , 2019)).Man-made reservoirs have also inundated reaches of riverine habitat important to the CPM and RBS.Glen Canyon Dam and Navajo Dam were completed in 1963, and by the mid-1980s, Lake Powell had inundated the lower 87 km of river, Navajo Reservoir had inundated 43 km of the upper river, and Navajo Dam releases had cooled the river 69 km downriver to the Animas River confluence (Figure 2).These changes reduced potential range and habitat of the CPM and RBS in the San Juan by about 31%, from about 523 to 362 km (Animas River confluence down to Lake Powell inflow near Clay Hills; Holden, 2000).Lake Powell also inundated the mainstem Colorado River for about 145 km upriver and 110 km downriver of the San Juan River confluence, eliminating much of the connecting mainstem habitat that was likely occupied by CPM and RBS.When lake elevation is below about 1097 m, deltaic sediments are exposed and a waterfall forms at Piute Farms Canyon, creating a barrier about 87 km upriver of the historical Colorado River confluence that prevents fish that move downriver from returning upriver (Cathcart et al., 2018).The Piute Farms waterfall is not a feature of the historical river, having been formed as the river carved a new channel through Lake Powell sediments during 1989-1995 and 2001-present.Of the three recovery units of CPM (USFWS, 2002a), the San Juan River has the least available longitudinal habitat.Wild populations of CPM remain in about 1754 km of the two other subbasins, including 1278 km in the Green River and 476 km in the Upper Colorado River (Miller, 2014).These reaches are considerably longer than the 362 km of habitat available in the contemporary San Juan River.The San Juan also has the steepest channel gradient of any major tributary used by CPM and RBS.For the 362 km of river from the confluence with the Animas River to Lake Powell, mean gradient is 1.91 m/km and maximum is 4.0 m/km (Holden, 1999), whereas mean gradients of the Green and Colorado rivers are 1.26 and 1.42 m/km, respectively (Figure 3).The San Juan is narrower, steeper, and shallower than the Green and Upper Colorado rivers.This geomorphology, combined with non-native vegetation anchoring the shorelines, limits the formation of backwaters used as nursery habitats by young CPM and the nursery floodplains used by young RBS.Historical photographs of the lower San Juan River and the mainstem Colorado River through Glen Canyon (Inskip, 1995) provide evidence that before these areas were inundated by Lake Powell, they contained low-gradient, sand-bed channels with backwaters and tributary inflows characteristic of nursery habitats used by CPM and RBS in other rivers.

| Flow recommendations
The 1991 Biological Opinion for the Animas-La Plata Project, a water project in southwestern Colorado and northwestern New Mexico, contained a reasonable and prudent alternative (RPA) identified by the U.S. Fish and Wildlife Service (USFWS) to avoid the likelihood of jeopardizing the CPM and RBS.The RPA included a 7-year research period, reoperation of Navajo Dam, and a commitment by the U.S. Bureau of Reclamation (USBR) and others to develop and implement the SJRIP (USFWS, 1991).This opinion reflected the USFWS's belief that the small F I G U R E 2 Map of the San Juan River from Navajo Dam downriver to the Lake Powell inflow, with river miles (RM) as distances upriver from Clay Hills near the Lake Powell inflow.

F I G U R E 3
Channel gradients of the San Juan, Green, Yampa, and Colorado rivers.Gradients of individual rivers from Google Earth.Profiles and locations are color coded.numbers of wild CPM and RBS in the San Juan River represented historical populations that had declined because of human alteration of the hydrology, and that these species could be recovered in the San Juan by restoring its flow and habitat (Gosnell, 2001).
By the mid-1990s, it was estimated that the population of wild CPM in the San Juan River was less than 100 individuals and possibly fewer than 50 (Holden, 1999).The RPA's research period (1991)(1992)(1993)(1994)(1995)(1996)(1997) was intended to quantify relationships of flow to fish habitat, but direct relationships were not derived primarily because only 22 young and 13 adult CPM and no RBS were collected (Holden, 2000).To augment the existing populations and to better understand habitat use and flow needs, nearly six million young hatchery CPM and over 200,000 juvenile and adult RBS were stocked in the San Juan from 1994 to 2020.
In 1999, the SJRIP adopted year-round flow recommendations for the San Juan River that biologists thought would benefit the native fishes (Holden, 1999).The flow recommendations were designed to provide water at specified times, quantities, and durations while maintaining the authorized purposes of Navajo Dam.The primary high flow targets were flow rate (e.g., 5000 ft 3 /s [142 m 3 /s]), minimum duration (e.g., 14 days), and annual frequency (e.g., 33% of years).Secondary targets addressed maximum duration between occurrences (e.g., not to exceed 6 years without reaching 97% of target flow).It was assumed that flow targets would be met if the future hydrology of the San Juan River was statistically the same as the modeled period of record , but the ensuing hydrology was drier than the modeled period and the flow targets were not met (SJRIP, 2018a).
Drought in the San Juan River subbasin resulted in less available water than the modeled period of record, reducing the ability to achieve the higher primary and secondary flow targets.Aside from continual dam releases to maintain base flows, the only high releases were of short duration, and these impeded the ability to recover reservoir storage and further decreased the availability of water for future long-duration releases and elevated baseflows (SJRIP, 2018a).During this period of drought, there was a general deterioration of desirable backwater and secondary channel habitats that was attributed to the lower flows, associated dam operations, and increasing impacts of non-native vegetation that stabilized the river banks and reduced habitat complexity (Lamarra & Lamarra, 2016).
Following a review of the 1999 flow recommendations and new water calculations in 2014, an adjustment was made to the dam operating rules under a new decision tree.This adjustment revealed that it was unlikely the recommendations could be met under the existing and forecasted hydrology that included drought and climate change (SJRIP, 2018a).

| Impacts of climate change on hydrology
Studies of the present and future effects of climate change on the Colorado River system, and specifically the San Juan River subbasin, show the likelihood of a continued decline in water supplies.A multi-model framework analysis found that the subbasin could experience significant disruptions to water deliveries (−12% to −48% for the drier models) and shortages of 53%-73% that exceed manageable thresholds, potentially affecting the subbasin as well as other regions that receive water and energy from the San Juan River (Bennett et al., 2019).
Hydrologic modeling (Precision Water Resources Engineering, 2020) also shows that climate change impacts to overall water supplies for Navajo Reservoir and the San Juan River could range from 2% to 8% increases in a mid-low impact scenario (25 percentile impacts observed from over 200 projections) to significant decreases of 21%-26% in a mid-high impact scenario (75 percentile).A center impact scenario (50 percentile) suggests decreases ranging 10%-16% for the period extending to the year 2050, with slightly higher impacts to the year 2070.
While these results show the possibility of increasing overall water supplies under some climate projections, the downward trends of the center impact scenario and most other projections are a strong signal for decreasing future water supplies in the San Juan River subbasin.
These modeling scenarios also show that average annual Navajo Dam outflows could decrease significantly, with the center impact scenario showing decreases of 19% in the 2050 time period and 23% in the 2070 period, while the mid-high impact scenario shows a 33% decrease (Precision Water Resources Engineering, 2020).This translates to a frequency of achieving the high flow targets (spring peak releases) in only about 25% of years under the center impact scenario and about 15% of years under the mid-high impact scenario.Modeling also suggests that Navajo Reservoir spring peak inflows could shift to earlier in the year by about 3 weeks for the 2050 period and by about another week for the 2070 period, potentially changing the timing of available water for spring high flow releases.These analyses confirm the low likelihood of achieving the higher targets of the flow recommendations into the future.

| Channel simplification
Reductions in channel area and island numbers have occurred in recent decades in the San Juan River.This is in contrast to studies that show a fairly consistent and diverse channel geomorphology for the San Juan River in the early 1900s.Two studies using aerial imagery (Bassett, 2015;Bliesner & Lamarra, 2000) found that island area in the mid-1930s was similar at about 1619 and 1862 ha, respectively, a difference of 15%.
Channel area was just over 5868 ha compared to less than 5787 ha, a difference of 1.8%, and the number of islands was 616 compared to 865.
Channel and island areas were relatively consistent between the two studies, but island counts varied probably because different criteria were used to define islands and a different scale was used during feature digitization.
Annual mapping has documented large decreases in island counts and channel area throughout the Colorado River system starting in the mid-1900s, including the San Juan River subbasin.The most common mechanisms for island losses appear to be the closing off of secondary channels by vegetation encroachment and sedimentation, and abandonment by downcutting of the main channel (Bassett, 2015).In the 1930s, riverine vegetation was primarily stands of cottonwoods but by 1994, non-native vegetation accounted for 67% of all riparian vegetation that was primarily invasive Russian olive (Elaeagnus angustifolia) and tamarisk (Tamarix spp.) (Bliesner & Lamarra, 2000).These vegetative changes together with altered flow regimes led to channel simplification of the San Juan, where habitats essential to native fishes are less prevalent.Significant decreases in low-velocity habitats were seen in the mid-1990s and early 2000s that further contributed to the decline of the native fishes (Holden, 1999).These geomorphic changes reduced the areas of backwaters, embayments, and secondary channels in the San Juan River that provide low-velocity habitats for larvae and young fishes, including CPM and RBS (Farrington et al., 2020).

| DISCUSS ION
There was general acknowledgment at the onset of the SJRIP of the low numbers of wild CPM and RBS and the limited quantity and quality of key native fish habitats in the San Juan River (Gosnell, 2001;Holden, 1999).Members of the SJRIP embraced flow management as the key conservation paradigm to support and enhance the habitat features needed for reestablishing self-sustaining populations of these species in the San Juan.Flow recommendations were designed to mimic the natural hydrograph through releases from Navajo Dam, with the assumption that flow targets would benefit the native fishes and continually reshape and diversify the habitat.The 7-year research period that preceded the flow recommendations was unable to unambiguously derive flow-to-habitat relationships for the San Juan River because the low numbers of wild CPM and the absence of RBS precluded direct measures of habitat use.The high flow targets were not met due to ongoing drought and fish habitat continued to degrade through channel simplification and increased substrate embeddedness.To better understand species habitat use, and to augment the populations, nearly six million hatchery CPM and RBS were stocked in the San Juan from 1994 to 2020.
Notwithstanding this substantial stocking, survival and recruitment of wild-produced fish beyond their first year of life have been low (USFWS, 2018, 2020), and studies show a bottleneck limiting development to adulthood (SJRIP, 2018b).Reproduction of CPM has been documented annually since 2013 with increasing catches of larval fish over time, but recruitment beyond the first year of life is limited (Farrington et al., 2020).Similarly, the population of RBS in the San Juan is comprised primarily of stocked fish and although there is annual reproduction, there is little evidence of survival and recruitment of young (Barkalow et al., 2020).A population viability analysis for CPM concluded that based on present survival and recruitment rates, the San Juan population would be extirpated in about 30 years (approximate species longevity) following cessation of stocking (Miller, 2014), revealing the low viability of the population in the absence of stocking.
Studies show that young CPM and RBS use low-velocity habitats in the San Juan River, including backwaters and side channels (e.g., Farrington et al., 2020).These habitats are not common features of the San Juan (Lamarra & Lamarra, 2016) and they occur in lower abundances than in the Green and Upper Colorado rivers (Miller, 2014), where wild populations of these species are self-sustaining.Low survival and recruitment of wild-produced CPM and RBS in the San Juan River appear to be inextricably linked to the paucity of suitable backwaters and the absence of riverside floodplains.One of the working hypotheses developed by the SJRIP to explain identified recruitment bottlenecks for CPM and RBS is that "availability of rearing habitats affects rates of recruitment" (SJRIP, 2018b).A comprehensive conservation strategy for these endangered fishes should include affirmative management to develop and enhance suitable rearing habitats as a complement to flow management and other efforts in view of this recruitment bottleneck.
Mechanical construction of nursery habitats is highly recommended for the San Juan River to enhance present habitat availability and to counterbalance the effects of reduced flows from ongoing drought.A well-designed plan is essential for mechanically constructing habitats that are strategically located and can function under a variety of expected flow conditions and also meet life-history needs.The following sections discuss the need for constructing nursery habitats in the San Juan River and offer considerations for habitat construction, based on experiences from other river basins.

| Colorado Pikeminnow
While subadult and adult CPM in the Green and Upper Colorado rivers typically use runs, eddies, and pools (USFWS, 2002a; Table 1) that are among the most common habitats in the San Juan (Bliesner & Lamarra, 2007), the larvae and age-0 stages require a different and more limited habitat type.This is characterized as a backwater that is a small water feature surrounded on three sides by land with a narrow, typically downriver connection to the main channel.The preferred sandy backwaters develop during receding flows as scour channels at the toe and periphery of large main-channel sand bars, or at the lower end of isolated side channels (Muth et al., 2000).They are formed at the end of spring runoff within the active channel and are not floodplain features (Figure 5; USFWS, 2002aUSFWS, , 2020)).
Habitat studies of the San Juan River show that backwaters are among the least numerous habitats present and comprise the least habitat area (Bliesner & Lamarra, 2007).This indicates that there are likely insufficient quantities of this nursery habitat for larvae and age-0 CPM in the San Juan, particularly in the upper predominantly rocky reaches.On a longitudinal scale, the numbers of backwaters are fewer than 5 per mile upriver of the Stateline (RM 120), and fewer than 10 per mile downriver, a pattern that is consistent with the notion that nursery habitat is principally in the lower river (Bliesner & Lamarra, 2007).The numbers of backwaters in the Green and Upper Colorado rivers are two to three times greater than in the San Juan River (Miller, 2014).

| Razorback Sucker
Subadult and adult RBS in the Green and Upper Colorado rivers typically use deep runs, eddies, large backwaters, and flooded off-channel habitats (USFWS, 2002b; Table 1).These are among the most common habitats in the San Juan River, except for the absence of floodplains (Bliesner & Lamarra, 2007) that are used by the larval and age-0 stages in other rivers.Riverside floodplains (USFWS, 2018) form in bottomlands, low-lying wetlands, and oxbow channels that flood ephemerally and connect to the main channel during spring runoff (Figure 5).These areas provide warm temperatures, low-velocity habitats, and high levels of food in the form of small invertebrates.The density of zooplankton in floodplains of the Green River can be 29 times greater than in backwaters and 157 times greater than in the main channel (Mabey & Shiozawa, 1993).The high productivity of floodplains is important as laboratory studies show that RBS larvae need to feed within 8-19 days of hatching to avoid starvation (Papoulias & Minckley, 1990).The large numbers of RBS found in the Lake Powell inflow (Cathcart et al., 2018) show the species' propensity for lacustrine habitats.

TA B L E 1 Habitats of Colorado Pikeminnow and Razorback
Sucker by life stage in occupied rivers (Green and Upper Colorado) compared to the San Juan River (USFWS, 2002a(USFWS, , 2002b)).

Colorado Pikeminnow
Adult movement and migration • Green: long-distance movements by wild adults to spawning sites (up to 950 km roundtrip); high fidelity for sites For the nearly 30 years that the SJRIP has been in existence, a great deal has been learned about the geomorphology, hydrology, and biology of the San Juan River, including the life histories of the endangered fishes.The low survival of young and their low recruitment is compelling evidence that the San Juan lacks the full complement of conditions necessary to support self-sustaining populations of CPM and RBS.The time has come to recognize that the river in its present state is not entirely suitable for conservation of the species largely because of a paucity of nursery habitats and the inability of the drought-driven hydrology to shape and maintain those habitats.We recommend a greater focus by the SJRIP toward a habitat restoration program for early life stages of CPM and RBS by utilizing mechanical habitat construction and maintenance as a compliment to flow management measures.Increasing the availability of these habitats will provide places for CPM and RBS to grow past the sensitive larval life stage and recruit to the adult populations.
To date, mechanical treatments in the San Juan River have removed non-native vegetation and sediment plugs from secondary channel inlets and old secondary and tertiary channels.Over 27 ha of aquatic habitat were restored between 2008 and 2014 (Bassett, 2015).Preliminary monitoring indicates that mechanical treatments may be producing the intended results on a local level, including similar fish population density and species composition in restored channels and adjacent natural channels (Franssen et al., 2015).Also, preliminary designs and pilot projects (e.g., Keller-Bliesner Engineering, 2018) show that mechanical creation and enhancement of nursery habitats is possible.These types of projects need to be expanded and implemented on a large scale.While the extent of mechanical construction of habitats in the San Juan River to support self-sustaining populations will likely be considerable and costly, a large-scale habitat restoration program offers the most likely path to success as part of an integrated conservation strategy for the species.
As implemented in other basins (see "Constructed habitats in other rivers" subsection), constructed habitats should be carefully designed through use of site-specific hydrologic and geomorphic data and sound design principles (Seavy et al., 2009).These habitats must be strategically located to effectively entrain drifting larvae and provide stable productive environments at a range of available flows (Figure 6).This means constructed habitats will need to be designed and placed so that there are habitats available at both low and high flows.Local mechanical disturbances (e.g., bulldozers) should also be considered to remove non-native vegetation and break up substrate embeddedness.

| Constructed habitats in other rivers
Mechanical construction of early life-stage habitats has effectively offset some effects of reduced flows and has improved survival and recruitment of native species in other river basins.One of the most extensive habitat restoration programs in the southwestern U.S. is in the Middle Rio Grande, New Mexico (https://webap ps.usgs.gov/MRGESCP/).Since 2003, the Middle Rio Grande Endangered Species Collaborative Program has created or restored about 1600 ha of habitat under a water depletion-neutral framework that mechanically modifies banklines, islands, and old floodplains to create backwaters, embayments, terraces, and depressions that inundate at discharges of 500-3500 ft 3 /s (14-99 m 3 /s).Nearly 300 sites ranging from 0.4 to 5 ha in size have been constructed as features designed to seasonally inundate and entrain eggs and larvae of the endangered Rio Grande silvery minnow (Hybognathus amarus), a small-bodied cyprinid that relies on spring floodplains as nursery habitats for larvae (Valdez et al., 2020).This habitat restoration program involves floodplain connections and habitats designed with sloping features that become effective under a range of flows.Different habitats are constructed at different locations and various elevations to become inundated at available flows.Managed releases in coordination with the Rio Grande Compact Commission have enhanced inundation of restored floodplains and significantly increased Rio Grande silvery minnow density in the annual population census (Valdez et al., 2019).
In one of the largest habitat restoration programs in the mid-western U.S., the Platte River Recovery Implementation Program (https:// platt erive rprog ram.org/about/ land-plan) has focused substantial conservation efforts on the creation and maintenance of off-channel nesting habitats adjacent to the river to support habitat use and productivity of the endangered interior least tern (Sternula antillarum) and threatened piping plover (Charadrius melodus).The Program has also used mechanical methods such as disking and herbicide application to create and maintain unobstructed channel widths that are ≥650 feet and unforested corridor widths that are ≥1100 feet for endangered whooping crane (Grus americana) resting and feeding areas, in part due to recognition that short-duration high flow releases, as envisioned, would not alone create and/or maintain the desired in-channel features.
Similar large-scale habitat restoration is taking place in the lower Colorado River, where a consortium of interests has implemented a 50-year program that includes expansive mechanical recontouring of the river channel and adjoining floodplains to create habitat for numerous imperiled aquatic species (BioWest, 2005).Among many conservation activities, the Lower Colorado River Multi-Species Conservation Program (LCRMSCP, 2022) seeks to create more than 8000 acres of riparian, marsh and backwater habitat and introduce more than 1 million combined RBS and bonytail (Gila elegans) along 400 miles of the Colorado River from Lake Mead in Nevada to the U.S.-Mexico border (https:// www.lcrms cp.gov/).Many aspects of this program are similar to the challenges seen in the San Juan River.
A number of habitat restoration projects have also been implemented in the Pacific Northwest to provide habitats for early life stages of various salmon species (Oncorhynchus spp.).Mechanically constructed side channels have helped mitigate chronic mortality in parr and smolts (Gibeau et al., 2020), and watershed protections with stream habitat restoration have increased survival and recruitment of young and contributed to recovery of local salmon populations (Battin et al., 2007).Basin-wide restoration of riparian vegetation and channel width has also helped to offset impacts of reduced flows, cooling summer water temperatures by about 1.8-3.5°Cand mitigating climate change impacts to threatened salmon populations (Justice et al., 2017).Restoring lateral floodplain connectivity has also helped to diversify stream habitats and provide salmon smolts with warm protective food-rich environments (Beechie et al., 2012).
Proper planning and engineering are important in scaling habitat features to flow magnitude(s) at which the features are expected to function consistent with species life histories, and to provide protection from exceptionally high flows from floods and storms (Wilby et al., 2010).
In summer of 2021, the gate structure, inlet channel, outlet structure, and dirt berm of the Phase III Wetland Project on the San Juan River was damaged and eroded by monsoonal rainstorm floods of up to 7000 ft 3 /s (198 m 3 /s) in late July (McKinstry & Matta, 2023).Such floods can damage project structures, result in extra construction costs, and delay benefits to species.This illustrates the challenges in planning, constructing, and maintaining habitat in a river like the San Juan with large periodic flood events.
One negative aspect of creating more low-velocity habitats is that these also provide habitat for non-native species that can prey on and compete with native species.Enhanced inundation of riverside floodplains in the Green River increased densities of non-native fishes, but the expanded habitats and improved production and survival of young RBS have helped to offset the impacts of predation and have increased population numbers (USFWS, 2018).
One of the SJRIP recovery elements is habitat protection, management, and augmentation, and a few habitat projects have already been designed and completed.We do not provide a prescription of mechanical habitat creation for the San Juan River in this paper.Instead, we advocate The CPM is a long-distance migrator and its newly hatched larvae drift long distances and have a specific need for small sandy backwaters.
The RBS is innately linked to riverine lacustrine habitats and newly hatched larvae also drift long distances with a specific need for productive riverside floodplains.Backwaters are presently limited in the San Juan River and riverside floodplains are absent.

| Flow management alone cannot provide and enhance native fish habitats
The higher flow targets of the flow recommendations are unlikely to be achieved with reductions in water supply due to ongoing drought and climate change.Flow management alone cannot achieve suitable habitat conditions for CPM and RBS in the San Juan River because of the paucity of nursery habitats and the steep river gradients and non-native vegetation that limit the formation of low-velocity habitats and connecting floodplains.

| Mechanical construction of nursery habitats is highly recommended
Early life-stage habitats of CPM and RBS are inadequate in the contemporary San Juan River and construction of these habitats is needed.
Low-velocity in-channel features such as backwaters and side channels are necessary for larval and age-0 CPM, and large riverside floodplains are necessary for larval and age-0 RBS.These habitats can also be enhanced with local mechanical disturbances that remove non-native vegetation and disrupt substrate embeddedness.These nursery habitats are the most commonly used by the young of these species in other parts of the Colorado River system that support self-sustaining populations.

| Constructed habitats should conform with available flows
Large-scale mechanical construction of nursery habitats that conform to climate-driven flow changes is recommended to improve survival and recruitment of CPM and RBS in the San Juan River.Properly designed and constructed habitats will facilitate inundation of habitats at critical times for early life stages under a variety of flows.The evolutionary histories of CPM and RBS and the total habitat needed for recovery must be considered when designing and constructing these habitats.

| Refinement of conceptual designs and pilot studies of constructed habitats
We recommend further efforts to develop and refine conceptual designs and pilot studies to better assess approaches for constructing nursery habitat features that can withstand extreme flow events and have increased efficacy under low flow conditions.We also recommend studies looking at utilization of these constructed habitats by non-native species and minimizing predation on CPM and RBS.

| Habitat/flow relationships
We recommend assimilation and synthesis of all available habitat information with reconciliation of species and life-stage-specific quantitative relationships between river flow, habitat, and fish abundance.Determining these relationships was a purpose of the 7-year research period, but low numbers of CPM and RBS in the San Juan River precluded quantifying these with certainty.Large numbers of CPM and RBS have been stocked into the San Juan and the habitats used by these fish and their availability should be determinable.

| Quantification of habitat needed for recovery
We recommend quantifying the amount of habitat necessary to achieve the species recovery goals for downlisting and delisting the CPM and RBS (USFWS, 2002a(USFWS, , 2002b)).Achieving these numbers will require certain amounts of habitat, considering survival and recruitment rates of the hatchery-stocked fish and their progeny.Integrated population models can help determine the amount of habitat necessary to recover the species in the San Juan River and can provide stakeholders with an understanding of the future investment needed for recovery.

| Evaluation of flow recommendations
It has not been demonstrated that available flows can effectively create the desired habitat complexity in the San Juan River without complementary mechanical habitat construction.We recommend an evaluation of the flow recommendations using the principles of adaptive management to be informed by the likely availability of future flows.The flow recommendations should be viewed as a large-scale field experiment with the flow targets as hypotheses to be evaluated and possibly revised through ongoing habitat and fish responses.This approach will help inform water management prescriptions and policies for the San Juan River.
Influence of drought and climate change on river hydrology 2.4.1 | Historical drought Dam construction and water projects have decreased the flow of the San Juan River since the early 20th century, and prolonged drought has exacerbated water supply shortages in the last two decades (Figure 4).The present drought in the Colorado River has persisted since 2000 with the period from 2000 through 2020 as the driest 21-year period in more than 100 years of record-keeping, and one of the driest in the past 1200 years based on paleohydrology data (USBR, 2021).Average monthly discharge near Bluff, UT (U.S. Geological Survey [USGS] #09379500) was 2409 ft 3 /s (68 m 3 /s) for the period before construction of Navajo Dam (1929-1962) and 2205 ft 3 /s (62 m 3 /s) for the period after construction (1963-1999).Persistent drought in the 21st century (2000-2020) has further reduced flow at Bluff to an average discharge of 1348 ft 3 /s (38 m 3 /s) for a 44% decline from the pre-dam period.Average annual water volume has decreased from 1,745,200 acre-ft (2.152 Gm 3 ) in 1929-1962 to 976,957 acre-ft (1.205 Gm 3 ) in 2000-2020, also a 44% decline.F I G U R E 4 Total annual runoff volume (gray bars) and 5-year moving average (solid line) in billions of cubic meters (Gm 3 ) for the San Juan River near Bluff, Utah (USGS #09379500), 1929-2020.Average annual volumes are shown for1929-1962, 1963-1999, and 2000-2020.

F
Sandy alluvial backwater used as nursery habitat by larvae and age-0 Colorado Pikeminnow in the Green River, Utah (left).Typical floodplain of the Green River, Utah, used as nursery habitat by larvae and age-0 Razorback Sucker (right).

F
Hypothetical river cross-section illustrating floodplain habitats to be inundated in spring, sloping features that inundate at a range of flows, and separate backwaters and embayments.amore focused and expanded effort on mechanical construction of nursery habitats, including backwaters for CPM and riverside floodplains for RBS, based on a model-driven computation of the amounts of habitat needed to produce the numbers of fish required for species recovery.Young CPM and RBS have specific habitat needs