Characterization of Pallid Sturgeon (
 Scaphirhynchus albus
 ) Spawning Habitat in the Lower Missouri River

Acipenseriformes (sturgeons and paddlefish) globally have declined throughout their range due to river fragmentation, habitat loss, overfishing, and degradation of water quality. In North America, pallid sturgeon (Scaphirhynchus albus) populations have experienced poor to no recruitment, or substantial levels of hybridization with the closely related shovelnose sturgeon (S. platorynchus). The Lower Missouri River is the only portion of the species’ range where successful reproduction and recruitment of genetically pure pallid sturgeon have been documented. This paper documents spawning habitat and behavior on the Lower Missouri River, which comprises over 1,300 km of unfragmented river habitat. The objective of this study was to determine spawning locations and describe habitat characteristics and environmental conditions (depth, water velocity, substrate, discharge, temperature, and turbidity) on the Lower Missouri River. We measured habitat characteristics for spawning events of ten telemetry‐tagged female pallid sturgeon from 2008–2013 that occurred in discrete reaches distributed over hundreds of kilometers. These results show pallid sturgeon select deep and fast areas in or near the navigation channel along outside revetted banks for spawning. These habitats are deeper and faster than nearby river habitats within the surrounding river reach. Spawning patches have a mean depth of 6.6 m and a mean depth‐averaged water‐column velocity of 1.4 m per second. Substrates in spawning patches consist of coarse bank revetment, gravel, sand, and bedrock. Results indicate habitat used by pallid sturgeon for spawning is more common and widespread in the present‐day channelized Lower Missouri River relative to the sparse and disperse coarse substrates available prior to channelization. Understanding the spawning habitats currently utilized on the Lower Missouri River and if they are functioning properly is important for improving habitat remediation measures aimed at increasing reproductive success. Recovery efforts for pallid sturgeon on the Missouri River, if successful, can provide guidance to sturgeon recovery on other river systems; particularly large, regulated, and channelized rivers.


| INTRODUC TI ON
Nearly all species of Acipenseriformes (sturgeons and paddlefish) are considered highly threatened globally; 24 of the 27 species are listed by the International Union for the Conservation of Nature (Cooke, Paukert, & Hogan, 2012) (https ://www.iucnr edlist.org/). Sturgeon and paddlefish use large freshwater river systems for spawning, often migrating hundreds of kilometers (km) (Auer, 1996;Bemis & Kynard, 1997). Reasons for the decline of sturgeon worldwide include fragmentation of habitat by dams, commercial fishing, and the degradation of habitat and water quality (Haxton & Cano, 2016;Rochard, Castelnaud, & Lepage, 1990).
The pallid sturgeon (Scaphirhynchus albus) is a long-lived, largeriver obligate species, native to the swift, turbid waters of the Missouri River, the Middle to Lower Mississippi River and some large tributaries from Montana to the Gulf of Mexico (Jordan et al., 2016;Kallemeyn, 1983). Pallid sturgeon populations declined through the 1900s and the species was listed as endangered under the U.S. Endangered Species Act in 1990(USFWS, 1990. Habitat loss, altered flow regimes, degraded water quality, and hybridization with the closely related shovelnose sturgeon (S. platorynchus) were identified as the major threats to species survival and recovery. Until recently, researchers have been unable to detect reproduction and recruitment (Dryer & Sandvol, 1993;USFWS, 2014). Given small population sizes with few reproductive adults, the apparent paucity of natural reproduction and recruitment in the Lower Missouri River (LMOR) nearly 30 years after listing is concerning from a species conservation perspective (Steffensen et al., 2019), (Figure 1).
In addition to longitudinal fragmentation by dams, simplification of habitats, and loss of lateral connectivity to the floodplain, it has been hypothesized that specific habitats, such as food producing and rearing habitats for early life stages, and spawning habitat for adults, may have become limited or less than suitable for sufficient population growth . A recent integration and analysis of information related to pallid sturgeon reproductive ecology indicated that rehabilitation of spawning habitat should be among priority management actions to avoid jeopardy to the species in the Missouri River .
Previously, the location of pallid sturgeon spawning habitats in the LMOR were unknown and their characteristics undescribed (DeLonay, Chojnacki, Jacobson, Albers, et al., 2016;. Spawning had been F I G U R E 1 Map of the Missouri River basin, Lower Missouri River, and ten pallid sturgeon spawning locations assumed to occur over coarse substrates in or adjacent to the main river channel (Becker, 1983;Mayden & Kuhajda, 1997), but assumptions were based on analogies to known spawning habitats of other sturgeon species rather than documentation of pallid sturgeon spawning (Dryer & Sandvol, 1993). Most sturgeon species exhibit upstream migratory behavior (Auer, 1996) and spawn in rivers on hard substrates. Sturgeon have been documented spawning over gravel, cobbles, boulders, bedrock, sand, and artificial substrates such as wood pilings, often immediately downstream from a dam (Bruch & Binkowski, 2002;Du et al., 2011;Krykhtin & Svirskii, 1997;Paragamian, 2012;Parsley, Beckman, & McCabe, 1993;Sulak & Clugston, 1998). Spawning habitat hydraulics vary with species and river system; however, many sturgeon species commonly aggregate and spawn in habitat patches with a wide range of depths and relatively high flow velocity (Baril, Buszkiewicz, Biron, Phelps, & Grant, 2017;McAdam et al., 2018;Smith, Smokorowski, & Power, 2017;Wyman et al., 2017). Sturgeon eggs become adhesive several minutes after fertilization and the common inference is that functional spawning substrate for pallid sturgeon is also likely coarse, hard, rock material which allows for stability in high-velocity environments where currents prevent sedimentation (Detlaff, Ginsburg, & Schmallhausen, 1993;Laustrup, Jacobson, & Simpkins, 2007).
In high-velocity environments, fertilized eggs may be transported some distance downstream before adhering to coarse substrates or being entrained in interstitial spaces. Lack of knowledge of specific habitat conditions necessary for successful spawning and reproduction of pallid sturgeon limits the ability to define conservation or engineering criteria for habitat protection and rehabilitation efforts (Baril et al., 2017;McAdam et al., 2018;Wang, Xia, & Wang, 2012).
In 2005, the Comprehensive Sturgeon Research Project (CSRP) was initiated to improve the fundamental understanding of reproductive ecology of the pallid sturgeon to inform river-and species-management decisions on the Missouri River. Using acoustic telemetry, from 2007 to 2015, CSRP biologists documented 33 pallid sturgeon spawning events on the LMOR and Platte River, a Missouri River tributary (DeLonay, . LMOR spawning locations are distributed over 950 km of the river from within a few km of Gavins Point Dam to 325 km upstream from the confluence with the Mississippi River in Missouri. Spawning has been detected or inferred through combinations of intensive (hourly to daily) and extensive (weekly to monthly) manual telemetry tracking of reproductive sturgeon. Spawning has been validated through recapture and surgical assessments and depth and temperature information from data-storage tags implanted inside telemetered sturgeon (DeLonay, Chojnacki, Jacobson, Albers, et al., 2016). Of the 33 female sturgeon that spawned, ten were tracked intensively to precise locations during spawning events where we made habitat measurements of depth, velocity, and substrate in the spawning patch and surrounding reach. Wildhaber et al., 2007). In this paper we describe the first observed spawning events for pallid sturgeon in an open river system, and present measurements of sturgeon spawning habitat conditions in a river hundreds of km downstream from main-stem dams.

| S TUDY ARE A
The LMOR flows downstream 1,300 km from Gavins Point Dam on the Nebraska-South Dakota border to its junction with the Mississippi River near St. Louis, Missouri (Figure 1). Gavins Point Dam, constructed from 1952Dam, constructed from -1957, is downstream from four other large reservoirs in the main-stem Missouri River system and serves to regulate flows for navigation in the 1,200 km downstream from Sioux City, Iowa. Flow alteration on the LMOR has resulted in a reduction in peak flows and an increase of summer low flows (Galat & Lipkin, 2000;Jacobson & Galat, 2008). With increasing distance downstream from Gavins Point Dam, the LMOR achieves a more natural hydrograph with spring rises occurring in most years as large tributaries such as the Platte River (river km 957), the Kansas River (river km 591), and other tributaries enter the LMOR (Galat & Lipkin, 2000;Jacobson & Galat, 2008). The pre-dam, pre-channelization LMOR system was multithreaded and consisted of shifting sandbars, vegetated islands, eroding banks, and backwaters (Jacobson & Galat, 2006). Channelization of the LMOR to maintain a navigation channel and to stabilize banks decreased the river width by 50%-66% and altered river habitats substantially (Funk & Robinson, 1974;Hallberg, Harbaugh, & Witinok, 1979). Channelization included the installation of coarse rock for bank revetment on the banks of outside bends, construction of river-training dike structures in the river on inside bends, and the construction of levees above the banks for flood control.
These alterations resulted in a relatively narrow and deep, singlethreaded river with few emergent sandbars or islands (Ferrell, 1995). Sand is the prevalent substrate in the LMOR and forms variably sized, migrating dunes, particularly in the navigation channel Reuter, Jacobson, Elliott, Johnson, & Delonay, 2008).

| ME THODS
As used in this article, spawning refers to release of eggs by reproductive female fish. Spawning, then, is a necessary condition, but is not necessarily sufficient for reproductive success and recruitment. Egg release must occur in the presence of reproductive males; eggs need to be fertilized; fertilized embryos need to incubate without being subject to excessive mortality related to predation, sediment deposition, or water quality impairment; free embryos need to hatch into the drift; and larvae need to survive to recruit to the population Wildhaber et al., 2007). Any of the steps in this cascade may be limiting to reproduction and recruitment, but our emphasis is on spawning as egg release and the habitat conditions that influence spawning. We use the term spawning habitat to indicate the general hydraulic and substrate conditions associated with spawning. We use the term spawning patch to indicate the best achievable delineation of spawning habitat through acoustic telemetry during an actual spawning event; hence spawning patches denote subsets of spawning habitat. The term spawning location refers to the location on the river measured upstream from the junction of the Missouri and Mississippi Rivers to the nearest 10th of a kilometer (or mile) to the centroid of the spawning patch. We use spawning reach to indicate the river context around spawning patches; reaches are usually the scale of a bend.
We identified pallid sturgeon spawning locations using acoustic telemetry (MAP RT-A dual port acoustic receiver with directional LHP_1 hydrophones; transmitter model# MM-M-16-50 [77-kHz, 80 × 16 mm, 35 g in air]; Lotek Wireless Inc.) over a 6-year period (2008-2013) (DeLonay, Chojnacki, Jacobson, Albers, et al., 2016;. Reproductive condition of tagged female sturgeon was assessed in the early spring through capture and egg biopsy. A few males were tagged and tracked as part of this study, but due to the large spatial scale of the study and low numbers of fish available for telemetry, female reproductive sturgeon were the main focus of this study. Individual females (and occasionally, males) were manually tracked by boat (hourly to daily) to spawning reaches over weeks to months by field crews from March through June. We recorded coordinates of fish positions using sub-meter differential global positioning systems (DGPS), and a customized ArcPad (ESRI, Redlands, California) mapping application .
Multiple lines of evidence, including behavior before and during the event in addition to post-event recapture and surgical assessments were used to determine when spawning events occurred.
Previous studies indicated pallid sturgeon spawning migrations generally followed an upstream movement pattern over long (>100 km) distances, initiated when water temperatures began to rise above 10°C. Upstream migrating fish generally selected relatively shallow, low-velocity water along inside bends, presumably to minimize energetic expense (McElroy, DeLonay, & Jacobson, 2012). Migration typically culminated with a change in behavior involving short bursts of activity (spawning) in deep, fast water near an outside bend's revetted bank usually at the sturgeon's upstream-most location or migration apex . For the purposes of this study, a spawning event was considered to have occurred once the upstream movement ceased followed by characteristic short bursts of activity in deep, fast water near a revetted bank. Fish telemetry points were recorded with varying intensity during spawning events using manual tracking. We have only included the telemetry points classified as spawning points in this paper and analysis. Spawning telemetry points were recorded over a 1 to 3-day period and include some 24-hr observations when conditions permitted tracking to continue through the night.
The accuracy of manual acoustic telemetry points in the LMOR georeferenced with DGPS varies spatially and temporally with depth, river discharge, and ambient noise associated with sediment transport. With acoustic telemetry, site-specific conditions such as large amounts of bedrock or bank revetment can cause reverberation and multipath of acoustic signals degrading the precision of fish locations near these substrates. Given the sub-meter precision of DGPS and a range of environmental conditions and depths, the accuracy of the telemetry locations in the LMOR is generally 5 m or better in the horizontal direction. The ten spawning events reported in this paper involve female pallid sturgeon intensively tracked during the spawning event, and in all cases, spawning was validated through recapture and reproductive assessments that included the use of ultrasound and/or surgical examination of the ovaries. Dual-frequency identification sonar (DIDSON/ARIS, Sound Metrics, Corp.) was also used opportunistically to validate spawning behaviors.
We used a variety of hydroacoustic tools to map 0.65-2.3 km long reaches centered on spawning patches defined by telemetry positions; the reach-scale maps provide habitat availability. The telemetry fish positions provided habitat use at the patch scale.
Spawning reaches were usually mapped within a few days and at a discharge within 10% of that observed during the spawning event.
We used a multibeam echosounder, and an acoustic Doppler current profiler (ADCP) georeferenced with a high-resolution real-time kinematic global positioning system (RTK GPS) to map depths and velocities. A sidescan sonar georeferenced with sub-meter DGPS was used to map substrates. Hypack (Xylem, Inc.) survey software was used for navigation and habitat data collection.
High-resolution bathymetry data were collected using a Reson 7125 SeaBat multibeam echosounder (Reson Teledyne Marine), an inertial motion unit, and RTK GPS along longitudinal transects.
Multibeam surveys did not include regions within dike fields to prevent equipment damage. Multibeam data were edited using Hypack's Hysweep software, and soundings were exported and gridded to generate maps with a 1-m cell size.
Velocity and depth data were collected using a 1200-kHz Rio Grande ADCP (Teledyne Marine) and RTK GPS (Gaeuman & Jacobson, 2005;Reuter et al., 2008). The ADCP was deployed from a rigid moving-boat mount over cross-sectional transects with a 20-m spacing perpendicular to flow and measured depths as shallow as 0.8 m. ADCP mapping extended across most of the river width, including inside the dike fields, and for this reason the ADCP-derived depth values from each of the 4 ADCP beams were used in the depth-use and availability analysis. Depth-averaged velocities measured by the ADCP were used to evaluate velocities at patch and reach scales. Maps of average water column velocity and ADCP-derived depth data were interpolated and gridded with a 5 m cell size using methods described in detail in Reuter et al.,

2008.
Maps of depth-slope were generated from ADCP-derived gridded depth maps using the ArcGIS (ESRI) slope algorithm to calculate the maximum slope of the depth grid within a 3 by 3 cell matrix, in units of degrees. A benthic terrain classification was applied using concepts developed from the Topographic Position Index and Benthic Terrain Modeler (Lundblad et al., 2006;Weiss, 2001).
The benthic terrain classification developed for the Missouri River uses depth and slope information to classify the river into crests, depressions, slopes, and flat areas (Reuter, Jacobson, Elliott, & DeLonay, 2009). Substrate type, classified as sand, revetment, or bedrock, was interpreted from both sidescan sonar imagery and multibeam data.
We inferred substrate characteristics in spawning reaches from visually interpreted, mosaicked sidescan sonar data collected in the spawning reach with a 900-khz Marine Sonic Sidescan Sonar towfish (Marine Sonic Technology, Ltd., White Marsh) (Elliott, Jacobson, & DeLonay, 2004;Reuter et al., 2008). We also interpreted substrate from bed textures in gridded and point cloud multibeam data.

| RE SULTS
All 10 pallid sturgeon spawning reaches mapped as part of this work were in the LMOR downstream from Gavins Point Dam between river kms 325 to 934 (Figure 1, Table 1). The upstream-most reach was located just downstream from the confluence with the Platte River on the Nebraska-Iowa border; three reaches were located upstream from the Kansas River; two were located near the Kansas River; and four reaches were clustered near Boonville, Missouri in Central Missouri (Figure 1, Table 1). In most years spawning occurred from late April to mid-May, with the earliest spawn date occurring on March 31, 2012 and the latest on May 19, 2011 (Table 1). There were between 8 and 51 telemetry points classified as spawning points and used for analysis at each spawning location (Tables 2,3). Fish were recaptured as quickly as possible for surgical evaluation, over half of the fish were recaptured between 0-4 days of spawning, and two of the fish evaded recapture until several months after spawning occurred (Table 1).
Spawning occurred over a broad range of discharges, ranging from 1,011-3,823 cubic meters per second, with flow percentiles in the 40th to 91st range for the post-dam flow record, as measured at U.S. Geological Survey streamflow-gaging stations closest to spawning locations (Table 1) TA B L E 1 Summary of water temperature, discharge, substrate, depth, depth-averaged velocity, and turbidity at pallid sturgeon spawning locations in the Lower Missouri River. (

| D ISCUSS I ON
Pallid sturgeon spawn in deep, high-velocity patches on outside bends adjacent to bank revetment in the contemporary LMOR.
Spawning patches are highly modified habitats at the base of, or adjacent to, coarse bank revetment installed during 20th century channelization. The deep, high-velocity habitats near coarse substrates that comprise present-day, pallid sturgeon spawning habitat are common on the LMOR (Bulliner, Elliott, & Jacobson, 2017;Reuter et al., 2008). Deep, high-velocity habitats and coarse substrates were less common in the pre-dam, pre-channelized LMOR, which was characterized by a dynamic, multi-threaded sandy channel, bare sandbars, vegetated islands, eroding banks, and backwaters. Although there are no quantitative habitat TA B L E 2 Depth in spawning reaches and at telemetry fish locations (spawning patches) recorded during pallid sturgeon spawning events Dam and Sioux City, Iowa are slower and much shallower than spawning habitats selected by pallid sturgeon in the channelized LMOR (Erwin, Jacobson, & Elliott, 2017;Reuter et al., 2009). Prechannelization, coarse substrates exposed in the river were rare and associated with glacial deposits, coarse-bedded tributary inputs, like the mouths of tributaries draining the Ozark Plateau including the Osage and Gasconade Rivers, or deposits associated with local bedrock outcrops (Laustrup et al., 2007;Reuter et al., 2008 shovelnose sturgeon on the LMOR; unlike our study, the shovelnose sturgeon studies did not differentiate spawning habitat from migratory habitat (Bonnot et al., 2011;Reuter et al., 2009).
The channelized LMOR's artificially emplaced bank revetments and gravels may provide poor quality or non-functional habitats for spawning and embryo incubation. Interstitial spaces in bank-revetment or associated gravels may not be adequate for embryo adhesion and incubation. Constant movement of sandy bedforms in and adjacent to spawning patches presents a potential disturbance to embryo deposition and incubation. Repeat  Spawning in the contemporary LMOR occurs in a relatively narrow zone between stable coarse substrate on revetted banks and a highly dynamic, mobile sand bed. The fate of embryos under these conditions is unknown, and has been assumed that embryos deposited on sand will become buried by dunes within hours due to high sediment transport rates .
Laboratory studies have shown low survival and development rates for white sturgeon embryos buried in sand (Kock, Congleton, & Anders, 2006). Emerging laboratory data for pallid sturgeon, however, indicate with the exception of complete burial by sand some survival can be expected (Chojnacki, George, & DeLonay, 2017).
Field assessments of embryo deposition in LMOR spawning patches remain a challenge because of difficulties associated with identifying precisely where embryos are released and how far they may be transported before adhering to substrate. Furthermore, the time it takes embryos to become adhesive in a turbid, turbulent river environment is unknown. Turbidity conditions during spawning events were variable; moderate compared to the high range of variability in turbidity that can occur on the LMOR, and lower than pre-dam and channelization historical measurements from 1907 (Blevins, 2006). Laboratory, flume, and mesocosm studies of embryo adhesion characteristics and transport could resolve some of these uncertainties.
High-resolution telemetry using two-dimensional acoustic arrays, or the implementation of event-based tags to record egg expulsion events could potentially be used to document spawning and embryo deposition with enough precision to determine if they are deposited on bank revetment, in gravel or sand at the base of bank revetment, on bedrock, on gravel associated with bedrock outcrops, or within migrating sand dunes. Pallid sturgeon also spawn near the river bed where flow separation and complex turbulence associated with sand dunes and revetment create highly variable velocity fields.
ADCP's similar to those used in this study are not capable of measuring velocities in the bottom-most 25-50 cm. Improved instrumentation is needed to more precisely measure patch conditions near the substrate to more accurately assess habitat selection and embryo transport and fate.
Despite low numbers and poor recruitment, the LMOR is the only portion of the pallid sturgeon's range where spawning, reproduction, and recruitment of genetically pure pallid sturgeon has been documented (Jordan et al., 2019). Intensive study of known spawning patches, however, has provided little evidence of successful reproduction -measured as hatch of fertilized embryos and downstream dispersal of free embryos. Sampling for drifting free embryos downstream from the ten suspected spawning patches captured only one potential hybrid pallid sturgeon x shovelnose sturgeon free embryo .
This was immediately downstream from the May 16-19, 2011 spawning location near river km 348. Lack of documented reproductive success at these patches is consistent with the hypothesis that they may be of poor quality and lack some characteristics necessary for survival from embryo deposition through hatch.
The Pallid Sturgeon Effects Analysis and Missouri River Science and Adaptive Management Plan introduced the hypothesis that spawning habitat quality (among other potential limitations) may be limiting reproduction and recruitment on the LMOR; subsequently the Missouri River Recovery Program has prioritized the design and construction of improved spawning habitat for pallid sturgeon . Very little design guidance exists for functional spawning habitat in the LMOR. Spawning habitat construction and remediation efforts, particularly for lake sturgeon, have shown variable rates of success (McAdam et al., 2018). It is unclear the extent these rehabilitation efforts translate to LMOR and to pallid sturgeon, although there is value in lessons to be learned from less successful or failed restoration projects (Baril, Biron, & Grant, 2019). Repeated aggregations and spawning of pallid sturgeon have been documented on the Lower Yellowstone River in a reach that is wider and shallower than the LMOR, and which consists of mostly sand substrate (DeLonay, Chojnacki, Jacobson, Albers, et al., 2016). Low numbers of pallid sturgeon free embryos have been Management Plan as science priorities to improve understanding of spawning and the role of spawning habitats in reproduction and recruitment .
In summary, this effort presents increased understanding in spawning behavior and the locations of spawning habitats used by pallid sturgeon in an open river system many km downstream from main-stem dams. Downstream from LMOR spawning locations, hundreds of km of free-flowing river on the LMOR and Mississippi Rivers are potentially available for larval development and drift. Although information gaps remain, this is the first study to present detailed quantitative measurements and maps of sturgeon spawning locations and habitat in a large, regulated, and channelized river, hundreds of km below a dam.