Streamflow Reductions and Habitat Drying Affect Growth, Survival ...

Transactions of the American Fisheries Society 139:1566–1583, 2010 Ó Copyright by the American Fisheries Society 2010 DOI: 10.1577/T09-143.1

[Article]

Streamflow Reductions and Habitat Drying Affect Growth, Survival, and Recruitment of Brassy Minnow across a Great Plains Riverscape JEFFREY A. FALKE*1 Department of Fish, Wildlife, and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523-1474, USA

KEVIN R. BESTGEN Larval Fish Laboratory, Department of Fish, Wildlife, and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523-1474, USA

KURT D. FAUSCH Department of Fish, Wildlife, and Conservation Biology, Colorado State University, Fort Collins, Colorado 80523-1474, USA Abstract.—Flow alterations caused by reservoir storage, groundwater pumping, diversions, and drought are widespread in North American Great Plains streams and have altered and fragmented habitats and reduced native fish biodiversity. Early life stages of fish are particularly sensitive to altered flow regimes, and reduced growth and survival may negatively affect the persistence of native species and assemblages. We investigated how growth and survival of brassy minnow Hybognathus hankinsoni larvae in the Arikaree River, Colorado, varied among three 6.4-km river segments that differed in hydrology and how climate influenced drying rates of spawning and rearing habitats in these segments over 3 years. We found that brassy minnow spawned in backwater habitats within a discrete period from mid-April to late May, based on otolith increment analysis. The timing of spawning and growth of larvae were influenced by climate and the hydrologic context of the river segment. Brassy minnow spawned 2 weeks earlier under warm, dry conditions in 1 year, and both growth rates and survival were significantly lower than during two wetter years (growth: 0.25 mm/d versus 0.30 and 0.41 mm/d; survival: 0.8391/d versus 0.894 and 0.897/d). For cohorts of larvae in individual backwaters, survival was higher in spawning habitats that were larger and that dried more slowly, and among cohorts that hatched in the middle of the spawning period under a moderate thermal regime. Overall, we found that brassy minnow spawning and recruitment were strongly influenced by habitat drying driven by interactions among stream geomorphology, groundwater pumping, and climate across multiple spatial scales. We suggest that conservation efforts explicitly consider the adaptations of this fish to harsh environments and focus on providing flows to maintain the spawning, rearing, and refuge habitats that are critical to brassy minnow population persistence.

Water abstraction for human use is a major problem affecting rivers and streams worldwide, and has resulted in losses of natural flow periodicity, increased channel drying, and severed upstream–downstream linkages (Benke 1990; Malmqvist and Rundle 2002). For stream fishes, these changes in flow have caused habitat loss and reduced dispersal opportunities among habitats. Successful reproduction and survival of early life stages of stream fishes are strongly influenced by hydrologic variability (Starrett 1951; Schlosser 1985; * Corresponding author: [email protected] 1 Present address: Department of Fisheries and Wildlife, Oregon State University, ) U.S. Environmental Protection Agency, 200 Southwest 35th Street, Corvallis, Oregon 97333, USA. Received August 12, 2009; accepted June 5, 2010 Published online October 4, 2010

Mion et al. 1998) and as a result are particularly sensitive to altered hydrologic regimes (Scheidegger and Bain 1995; Freeman et al. 2001; Bestgen et al. 2006). Water withdrawals are an especially important issue for streams in the semiarid Great Plains of the Midwestern United States, where water diversions and agricultural irrigation pumping are widespread (Gutentag et al. 1984; McGuire et al. 2003; Falke et al., in press b). Pumping has caused high rates of groundwater decline in the Ogallala formation of the High Plains aquifer that underlies much of the western Great Plains, resulting in reduced flows in streams that are hydrologically connected to groundwater. Groundwater is a key factor for fish persistence in dryland streams because it maintains base flows and connections among habitats along the riverscape, especially during drought. Furthermore, groundwater moderates

1566

HABITAT DRYING EFFECTS ON BRASSY MINNOW

high water temperatures during summer, and prevents complete freezing during harsh winters (Labbe and Fausch 2000; Scheurer et al. 2003). Therefore, knowledge of factors that influence growth, survival, and recruitment of native fishes in groundwater-fed plains streams is important for their conservation as habitats suffer further drying. Spawning and early life history stages in fish are critical periods that may serve as bottlenecks that limit recruitment to reproductively mature life stages (May 1973; Ludsin and DeVries 1997; Halpern et al. 2005). In streams, discharge, water temperature, and precipitation are important mechanisms limiting survival of larval fish (Crecco and Savoy 1984; Mion et al. 1998; Leach and Houde 1999), and evidence suggests that these same factors affect survival of larval fish in Great Plains streams (Harvey 1987; Durham and Wilde 2006, 2009). Moreover, stream fishes are often influenced by factors that interact across multiple spatial and temporal scales, especially in dynamic plains streams (Lohr and Fausch 1997; Labbe and Fausch 2000; Fausch et al. 2002). For example, spawning and recruitment may be driven simultaneously by processes that create and destroy spawning and rearing habitats at local scales, as well as transport processes that influence fish dispersal across segment scales (Scheurer et al. 2003). Understanding these multiscale controls on spawning, growth, and survival of early life stages of plains fishes also will be critical for planning their conservation (see Dudley and Platania 2007). Although adapted to harsh conditions common to Great Plains streams, native fishes are often near their threshold of physicochemical tolerance (Matthews 1987; Matthews and Zimmerman 1990), which may partly explain why habitat fragmentation and loss have led to population declines (Fausch and Bestgen 1997; Hubert and Gordon 2007). For example, the brassy minnow Hybognathus hankinsoni is listed as a threatened species in Colorado because its distribution and abundance have declined (Scheurer 2001; CDOW 2007), and populations are projected to decline more in the future (Falke et al., in press b). One cause of these declines could be recruitment failure owing to habitat fragmentation and loss from altered hydrology. Additionally, there is evidence that brassy minnow use multiple habitat types for spawning and early rearing, juvenile and adult growth, and refuge during harsh periods, which could be hampered by increased intermittency (Scheurer et al. 2003). However, basic early life history and spawning habitat requirements for this species are unknown, as are the factors that influence population regulation at the larval stage. Our primary objective was to describe habitat and

1567

life history dynamics of brassy minnow to generate information needed to protect populations in Great Plains rivers that are being threatened by groundwater pumping, increased habitat fragmentation, and climate change. To accomplish this, we compared brassy minnow spawning and rearing, and growth and survival of larvae, in three river segments that differed markedly in hydrology and drying due to different geomorphology and effects of groundwater pumping (Scheurer et al. 2003; Falke et al., in press b). Comparisons were made across 3 years that differed in climate, and were made at both the local scale and among segments, to assess recruitment processes that operate at multiple scales across the riverscape. Within this framework, our specific objectives were to (1) measure the rate of spawning habitat drying and investigate the importance of shallow alluvial groundwater for maintaining those habitats at the local scale, (2) determine the period of hatching and the ontogeny of larval recruitment from backwater spawning habitats to main-channel rearing habitats, (3) relate survival of larvae in backwaters to local habitat characteristics, and (4) measure growth and survival of larvae across segments and years and relate these to hydrology at the segment scale and climate during a multiyear drought. Finally, we discuss implications of continued anthropogenic habitat drying (see Falke et al., in press b) for brassy minnow recruitment in Great Plains streams. Methods Study area.—Our study area was restricted to the lower half of the Arikaree River basin, Colorado, because segments with the potential for perennial streamflow and fish habitats occur only in the downstream 110 km of the basin (Figure 1). We sampled larval fish and spawning habitats within three 6.4-km segments selected to represent a gradient in intermittency and connection to groundwater, which are described in detail in Scheurer et al. (2003) and (Falke et al., in press b). In the upstream segment, the alluvium through which the river flows is hydrologically connected to groundwater from the underlying High Plains aquifer. There are no alluvial irrigation wells nearby that influence annual drying, and as a result, even in dry years this segment supports perennial flow and alternating runs and deep pools. Beaver Castor canadensis have created large pools in some reaches. In contrast, the alluvium in the middle segment rests on bedrock and is hydrologically disconnected from the groundwater aquifer (i.e., there is no lateral groundwater input). In addition, three alluvial irrigation wells nearby increase channel drying, so this segment is largely intermittent during summer months (Squires 2007). Owing to heterogeneity in local

1568

FALKE ET AL.

FIGURE 1.—Map of the study area in the Arikaree River basin in northeastern Colorado showing the locations of the three 6.4-km study segments (US ¼ upstream, MS ¼ middle, and DS ¼ downstream) and Black Wolf Creek. The solid lines in the lower basin indicate where the potential for perennial streamflow remains (based on Falke et al., in press b). The locations of all 59 available spawning habitats (43 channel margins and 16 backwaters) sampled across 3 years in the three sections are shown to the right. The locations of 12 thermographs monitored over 2 years and six groundwater wells monitored in 1 year in the upstream segment are also shown. A cluster of three backwaters close together in the middle segment is indicated.

scale geomorphology, the upper portion of the middle segment is interspersed with deep, well-developed pools, whereas the lower portion is wide and shallow, with few pools. Finally, in the downstream segment bedrock also underlies the alluvium, and 11 alluvial wells in and near the segment cause complete channel drying by early summer. Surface flow in this segment is transient, occurring only during early spring in wet years, and following summer storms that fill the alluvium. In addition, a perennial tributary, Black Wolf Creek (Figure 1), often sustains a short reach of flowing habitat in the middle of this segment. Sampling spawning and refuge pool habitats.—We investigated the effects of seasonal drying on spawning habitats for brassy minnow along the three study segments and in the lower 1 km of Black Wolf Creek during spring and summer 2005–2007. In late May 2005 and 2006, and late March 2007, all potential spawning and rearing habitats in each segment were identified, classified into backwater and channel-

margin habitats, and georeferenced using a Global Positioning System (GPS). We focused our sampling effort on backwaters and channel margins, two vegetated habitat types in which brassy minnow are reported to spawn (Copes 1975; Scheurer et al. 2003). Backwaters were relatively large, deep, off-channel habitats connected to the main channel but with little or no flow. Channel-margin habitats were relatively small, shallow, flowing areas at margins of the main channel where higher spring flows inundated terrestrial vegetation. Each spawning habitat identified was surveyed weekly or biweekly through the first or second week of July. In backwater habitats during each sampling occasion we measured surface area, maximum depth, conductivity, and ambient water temperatures at the surface and just above the substrate. Surface area (m2) was calculated as length times average width along three evenly spaced transects. Maximum depth was measured with a stadia rod (cm), and conductivity (lS) was measured using a Yellow Springs Instruments


Systems, Inc., model 85 multimeter. Surface and substrate water temperatures (nearest 0.18C) were recorded with a digital thermometer (Versatuff Plus 396, Cooper-Atkins Corp.). Water temperature was also measured hourly from January 2006 to July 2007 using thermographs (HOBO Water Temp Pro version 1, Onset Corp.) installed in backwater and mainchannel habitats (upstream segment, n ¼ 7; middle segment, n ¼ 3; and Black Wolf Creek, n ¼ 1; Figure 1). A thermograph installed in the downstream segment was frequently dewatered and so those data were not included in analyses. All pool habitats for juvenile and adult fish along the entire upstream and middle segments were identified and georeferenced in late July each year, during the period of lowest flow. In the upstream segment, a subset of pools (n ¼ 31 of 172 in 2005, 29 of 180 in 2006, and 19 of 218 in 2007) were randomly selected from two pool size categories (small and large) based on the distribution of pool volumes (m3; Falke et al., in press b). In the middle segment, all pools were sampled in all years (n ¼ 9 in 2005, n ¼ 27 in 2006, n ¼ 29 in 2007). No pools were ever present in summer in the downstream segment. In August 2005, we installed six groundwater monitoring wells (see Falke et al., in press b for details) spaced evenly along the upstream segment (Figure 1) to investigate the relationship between groundwater stage and spawning habitat in the wettest segment. Logistical constraints prevented installing wells in the other segments. The wells were approximately 10 m from the stream channel. Depth to groundwater (cm) was measured using a steel tape during spawning habitat surveys. We summed the area for two to four backwaters near each of four wells and compared the percentage of backwater area remaining with the groundwater stage weekly for five consecutive weeks. We also recorded the main-channel connectivity of the reach in which the well and backwaters were located (flowing ¼ connected pools, intermittent ¼ disconnected pools; there were no dry reaches) during the final week of sampling. No backwaters were located near the two most upstream wells. Sampling larval and adult fish.—Larval brassy minnow were sampled concurrently with all previously identified spawning and rearing habitats beginning in late May 2005 and 2006, and late March 2007, and ending the first or second week of July. In 2005, larvae were collected from both channel margins and backwaters using aquarium dip nets (20 3 16 cm, 250 lm mesh) during daytime spawning habitat surveys. In 2006 and 2007, larvae were sampled using dip nets in shallow backwaters (,30 cm maximum depth) and channel margins, whereas larvae were

1569

sampled in deeper backwaters at night using floating quatrefoil-type light traps (design modified from Killgore 1994). Four 4-mm entrance slots allowed larvae to enter an inner chamber that consisted of four 7.5-cmdiameter Plexiglas tubes 14 cm long. Traps were attached at the top to a 6-cm-thick Styrofoam ring and contained yellow glow sticks to attract fish larvae. Upon retrieval, trap contents were flushed into a lower plastic bowl from which water drained through screened holes. Traps were deployed at fixed locations in backwaters at dusk for approximately 2 h. Laboratory experiments (see next section) showed light traps were effective over an area of at least 22 m2, so multiple traps were used in backwaters greater than 22 m2. A paired comparison of dipnetting and lighttrapping using 2007 data showed that estimates of brassy minnow size structure and relative abundance collected using the two gear types were similar, so we pooled the data (Falke et al., in press a). We also sampled young-of-year (age-0) and adult brassy minnow in refuge pools within study segments in August of 2005 through 2007 to quantify recruitment and population structure. No pools were present in the downstream segment during this period. Brassy minnow were collected using three-pass depletion seining (4.8-mm mesh), except in 2005 when two passes were made per pool. Pools were blocked at both ends using block nets (same mesh size) to prevent fish movement. All seining passes were conducted from upstream to downstream. All brassy minnow were enumerated separately for each pass; a subsample collected in each segment was measured to obtain total lengths (TL, mm) and fork lengths (FL, mm), and all fish were released. Light-trap validation.—We tested the efficacy of light traps to capture larval fish in a laboratory experiment (Foothills Fishery Facility, Colorado State University). We released a known number of 4-d-old fathead minnow Pimephales promelas larvae (n ¼ 30, approximately 5 mm standard length [SL]) at four fixed distances from a light trap, and counted the number captured after 2 h. Fathead minnow is a common associate of brassy minnow in Arikaree River light-trap samples, is commonly captured in light-trap sampling in other locations (e.g., Green River, Utah; K. R. Bestgen, unpublished data), and has similar spawning time, spawning habitat use, and size at hatching as does brassy minnow (Falke et al., in press a), so it was judged to be a useful surrogate cyprinid for these trials. Two fiberglass raceways (3.0 3 0.45 m) were painted matte black to simulate the dark substrate in Arikaree River backwater habitats, and to reduce reflection that could interfere with larval attraction. The light trap was placed approximately 0.37 m from the end of the

1570

FALKE ET AL.

raceway, and for each treatment an impermeable plastic divider was inserted into the raceway at one of four release distances: 0.37, 1.13, 1.88, and 2.63 m. Experiments were conducted at dusk. Treatments were randomly assigned and two replicates of each treatment were conducted over 4 d total. Validation of brassy minnow otolith daily rings.— Knowledge of time of first otolith increment deposition and the frequency of increment deposition are critical for reducing bias in back-calculation of age from otolith microstructures (Campana and Neilson 1985). We validated our estimates of hatching dates for fieldcollected brassy minnow by spawning and rearing brassy minnow in the laboratory and evaluating otolith increment deposition over time. We collected sexually mature brassy minnow adults from a pond on the Foothills Campus of Colorado State University using minnow traps. These were progeny of fish originally collected from the Arikaree River. Only fish that showed secondary sexual characteristics (males with yellow pigmented fins, females with distended abdomens) were retained. We dry-stripped eggs from females and combined them with milt from one or two males in a petri dish, after which a few milliliters of water were added. The adhesive eggs were water hardened in about 1 h, allowed to incubate in the petri dish, and hatched in approximately 3 d. We split the larvae into two batches, and continued to rear them under two temperature regimes. The first was constant at ambient room temperature (208C), whereas the second fluctuated between 188C and 248C on a diel cycle (see Bestgen and Bundy 1998). The second treatment simulated the thermal regime in the Arikaree River during which brassy minnow hatch and rear and was conducted to facilitate identification of daily rings, which are thought to be accentuated by natural diel temperature fluctuations (Bestgen and Bundy 1998). Larvae in both treatments were reared under a natural photoperiod (14.5 h light : 9.5 h dark). They were initially fed a diet of ground flake food, and then switched to newly hatched brine shrimp Artemia spp. once the larvae were large enough to capture and consume them. For both treatments, we recorded embryo development and preserved a series of four fish per day from each treatment in 100% ethanol for the first 30 d after hatching. For each preserved specimen, we measured TL (mm) and SL (mm), extracted otoliths, and counted daily rings as described in the next section. Larval brassy minnow age and growth.—We measured length at age, hatching date, and growth for larval brassy minnow collected in the field during 2005–2007. After measuring TL and SL using digital

calipers (60.01 mm), right sagittal otoliths were dissected from fish and stored in a drop of immersion oil on a standard microscope slide. Increments in otoliths from fish greater than 17 mm SL were difficult to read, so they were ground and polished using standard techniques (Stevenson and Campana 1992). A compound light microscope at 4003 magnification was used to count daily rings in otoliths. Two readers made independent estimates of the total number of daily rings for each otolith. Estimates were compared, and those that exceeded 10% difference between the estimates were discarded. Final ages for individual larvae were the average of the two estimates, and hatch date was estimated as the collection date minus the estimated age in days. Growth of larvae was calculated by subtracting length at capture from average length at hatching (4 mm SL; see Validation of Brassy Minnow Otolith Daily Rings in Results) and dividing by age (d) to obtain growth per day. We calculated growth as the change in mean SL estimated using larvae grouped in 5-d age increments. Growth was calculated for brassy minnow larvae in backwater habitats in the middle and upstream segments from 2005 to 2007. For 2007, we used only the subsample of fish that was aged to make comparisons among years (see Hatching Date Distributions in Results). We included only fish up to 40 d old, or the age by which we predicted that habitat switching had occurred (see Brassy Minnow Habitat Transition in Results), to make valid comparisons among years for a single habitat type, namely backwaters. Backwaters most likely become unsuitable rearing habitats as they dry, environmental conditions deteriorate (e.g., owing to high temperature and low dissolved oxygen), and food resources are depleted. We hypothesized that brassy minnow larvae would move from backwaters to main-channel habitats at some size threshold to avoid habitat desiccation and reduced growth and survival. To assess this habitat shift, we calculated the change in mean catch per unit of effort (CPUE; fish/min) of brassy minnow larvae in backwater and channel-margin habitats in the upstream segment from mid-April to late July 2007. Statistical analyses.—We compared hatching dates, growth, and survival of brassy minnow larvae among individual backwaters, and among segments and across years, and related them to habitat drying during spring and summer. We estimated the rate of drying for individual backwater habitats in the upstream and middle segments each year from the slope of a linear regression of backwater area (m2) as a function of time (d), using measurements collected over the season. We then compared mean backwater drying rates of


individual habitats among segments and years using two-factor analysis of variance (ANOVA). If differences were detected, we used Tukey’s honestly significant difference (HSD) test for multiple comparisons. We recognize that comparing variation among reaches with individual habitats within reaches serving as replicates could violate independence assumptions of ANOVA. To address this issue, we conducted a randomized permutation test (Manly 2007) with no independence assumptions and tested the same main effects and interaction as the ANOVA. Results of the permutation test were identical to those of the ANOVA. Based on these results we felt justified that the ANOVA analysis did not influence our results in substantial ways to warrant a different analysis. In a concurrent study, we found that brassy minnow larvae occupied backwater spawning habitats at high abundance, whereas they were found at low abundance in channel-margin habitats (Falke et al., in press a). Therefore, we modeled the influence of backwater spawning-habitat characteristics on survival of cohorts of brassy minnow collected in 2007. The sampling period was too short and sample sizes were too small during 2005 and 2006 to allow this analysis. After age assignment (see Hatching Date Distributions in Results), we assigned larvae collected in individual backwaters to cohorts based on an overall hatchingdate distribution. Hatching dates for 98% of larvae were between 16 April and 21 May 2007. Subsequently, we split this 36-d period into three consecutive 12-d periods, and assigned larvae to early, middle, or late cohorts. We then calculated daily survival rates for each cohort in each backwater using catch curves. The catch curves were calculated from the descending limb of increment-frequency histograms (Essig and Cole 1986). Daily survival rate (S) was calculated as S ¼ eZ, where e is the base of the natural logarithm and Z is the slope of the catch curve. We modeled cohort survival as a function of backwater habitat characteristics and cohort-specific predictors for larvae collected in 2007. Daily survival rates (proportions) were arcsine- and square-root transformed to meet the assumptions of linear models. Backwater characteristics were rate of drying across weeks (see above; RATE), and backwater area (AREA) and maximum depth (DEPTH) during the first week of habitat sampling. We also included the abundance of larvae (total number collected in an individual backwater across all weeks; ABUN) to investigate density dependence. Cohort-specific predictors were mean hatching date (MHD) for the cohort, and cumulative growing season degree-days (GSDD). The GSDD were calculated by summing the average daily temperature (8C) from January 1 to the mean hatching

1571

date of the cohort. We chose GSDD as our thermal metric because it can be calculated for individual cohorts, represents the thermal conditions to which each cohort was exposed prior to hatching, and applies across the range of hatching dates and individual larval sizes within a cohort. Temperature data were derived from the thermograph nearest each backwater (Figure 1). We also included a quadratic term for MHD and GSDD because we hypothesized survival might be lower early or late in spawning period. Finally, we included a categorical variable identifying the segment (upstream or middle) in which the cohort hatched. We constructed a set of 22 a priori candidate models that contained sets of explanatory variables of potential biological significance (Burnham and Anderson 2002). We used Akaike’s information criterion (AIC) to rank models by comparing each of the candidate models simultaneously. The AIC scores were adjusted for bias due to small sample size (AICc), and Akaike weights (wi) were calculated. To account for model uncertainty, we used model averaging to calculate parameter estimates and variances from models in the confidence model set (wi . 0.05) and made inferences based upon these estimates. All analyses were conducted using Proc GENMOD in SAS version 9.2 (SAS Institute). Finally, we combined all larvae collected across segments within each year to compare interannual differences in survival of brassy minnow larvae in the Arikaree River. For 2007 we used only the subsample of fish that was aged (see Hatching Date Distributions in Results). Therefore, these can be considered estimates of survival for brassy minnow larvae within the Arikaree River basin as a whole. We used analysis of covariance (ANCOVA) to test for differences in the slope (i.e., daily survival rate) of the descending limb of increment-frequency histograms using year as the covariate. If differences among years were detected, we used Tukey’s HSD test for multiple comparisons to evaluate in which year S differed. The significance level was set at a ¼ 0.05 for all analyses. Results Habitat Availability and Drying There was a drought in eastern Colorado during 2000 to 2007 (Falke et al., in press a). Precipitation and mean annual flows were highest for 2005 (53.2 cm and 0.05 m3/s, respectively), lowest during 2006 (32.8 cm and 0.02 m3/s), and intermediate during 2007 (33.0 cm and 0.04 m3/s). Flows in spring 2007 were higher than in other years owing to abundant snowfall in December 2006. Fifty-nine individual spawning habitats were measured in the three segments and Black Wolf Creek during the 3-year study (backwaters, n ¼ 16; channel

1572

FALKE ET AL.

TABLE 1.—Characteristics of brassy minnow spawning habitats during 2005–2007 in three 6.4-km segments of the Arikaree River and lower Black Wolf Creek in the Arikaree River basin, Colorado. Numbers of weeks and habitats (BW ¼ backwater, CM ¼ channel margin) sampled, total and mean 6 SE area of backwater habitats, and proportion of each habitat type that dried by mid-July each year are shown. The number of each habitat type varied each year as they were created and destroyed and as flow changed. Number of habitats sampled Segment Upstream

Middle

Downstream and Black Wolf Creek

Backwater area (m2)

Year

Number of weeks

CM

BW

Total

2005 2006 2007 2005 2006 2007 2005 2006 2007

4 5 6 4 5 6 4 5 6

5 2 5 7 0 14 10 5 10

9 8 9 5 3 5 0 0 0

710 460 880 430 210 940 0 0 0

margins, n ¼ 43; Table 1), although the number present varied each year. No backwaters were ever present in the downstream segment or Black Wolf Creek. Backwater habitats were clustered in the lower half of the upstream segment and the upper third of the middle segment. The number and total area of backwater habitats varied among years and corresponded to climate conditions. Total backwater area was lowest in the dry conditions of 2006, but higher snowfall during December 2006 resulted in the highest total backwater area during spring and early summer 2007. In 2007, total backwater area in the upstream and middle segments was about twice and four times the amount available in 2006, respectively. We found that many backwater and channel-margin habitats dried completely by mid-July in all segments and years (Table 1). Due to their shallow ephemeral nature, most (60–100%) channel-margin habitats dried by the end of sampling each year. Fewer backwaters dried completely, and fewer dried in the upstream segment than in the middle segment (0–25% versus 33–80%). Backwaters in the middle segment dried faster (mean ¼ 17.84 m2/week, SE ¼ 4.56) than those in the upstream segment (mean ¼ 5.69 m2/week, SE ¼ 0.92) when all 3 years were combined (ANOVA: F ¼ 12.88, P ¼ 0.001). Based on Tukey’s multiplecomparisons tests, backwaters dried fastest in 2007 (mean ¼ 18.74 m2/week, SE ¼ 6.15) when the two segments were combined (versus 2005: mean ¼ 7.95, SE ¼ 1.79, P ¼ 0.031; versus 2006: mean ¼ 6.29, SE ¼ 1.97, P ¼ 0.045). There was no difference in the rate of backwater drying between 2005 and 2006 (P ¼ 0.931). Across segments and years, the rate of drying was fastest in the middle segment in 2007. Although the total area of backwater habitat available was greatest

Proportion of habitats dried

Mean

CM

BW

6 6 6 6 6 6

1.00 1.00 1.00 1.00

0.22 0.25 0.00 0.80 0.33 0.60

64 42 97 33 69 188

21 13 23 11 12 59

1.00 0.50 0.60 0.50

during spring 2007 in the middle segment, this area dried rapidly. Main-Channel and Backwater Thermal Regimes We recorded hourly temperatures from January 2006 through July 2007 for five backwaters and six mainchannel habitats in the upstream and middle segments and Black Wolf Creek (Table 2). Mean temperatures were higher during the warm dry conditions in 2006 (e.g., mean main-channel June temperature ¼ 24.58C) versus 2007 (mean June temperature ¼ 22.18C) for both backwater (t-test: t ¼ 2.57, P ¼ 0.030) and mainchannel (t ¼ 2.57, P ¼ 0.002) habitats. Backwater habitats were warmer during winter months (January– March), whereas main-channel habitats were warmer during spring and summer months (April–June). Based on mean temperatures during January through June 2006, backwater habitats were approximately 18C cooler in the upstream segment versus middle segment (t ¼ 2.57, P ¼ 0.002). However, no difference was apparent during 2007 (t ¼ 2.57, P ¼ 0.586). The highest recorded temperatures occurred in the main channel of Black Wolf Creek in late June through July of both years (2006 maximum ¼ 33.58C, 2007 maximum ¼ 28.78C). Average late summer temperatures (July–August) in main-channel habitats during 2006 were 21.78C (SE ¼ 0.4) in the upstream segment, 22.78C (SE ¼ 0.3) in the middle segment, and 23.68C (SE ¼ 0.4) in Black Wolf Creek. During fall 2006 (September–October) mainchannel temperatures were similar among segments and averaged 13.18C (SE ¼ 1.1). However, in early winter (November–December) main-channel temperatures in the upstream segment were moderate (average ¼ 6.38C, SE ¼ 0.5), whereas temperatures in the middle segment and Black Wolf Creek were colder (1.58C, SE

1573


TABLE 2.—Mean 6 SE monthly water temperatures (8C) based on hourly thermograph data collected from backwater (n ¼ 5) and main-channel (n ¼ 6) habitats in the upstream and middle study segments (Figure 1) of the Arikaree River, in 2006 and 2007. 2006 Month

Backwater

Jan Feb Mar Apr May Jun

5.3 5.2 10.5 12.9 16.0 18.3

6 6 6 6 6 6

0.6 0.7 0.7 0.5 0.7 0.5

2007 Main channel 3.2 3.2 9.3 14.4 21.5 24.5

6 6 6 6 6 6

¼ 0.1). These patterns underscore the importance of groundwater in moderating main-channel temperatures, especially during the dry conditions of 2006. Groundwater and Spawning Habitat Backwater habitats in the upstream segment were fed by groundwater and decreased in area as groundwater levels declined during spring and early summer 2006, the driest year sampled. Additionally, the relationship between backwater spawning habitat area and shallow near-channel groundwater stage reflected reach-scale habitat connectivity. In three intermittent reaches in the upstream segment, backwater area and groundwater stage declined linearly over time (Figure 2). The percent of backwater area remaining at the end of June was 27, 15, and 57% of the original area in these reaches. Conversely, in a flowing reach, groundwater stage and backwater area increased over time, the latter to 123% of the original area at the beginning of June. We attribute this increase to beaver activity, because their dams impounded the stream as flows decreased during June and increased shallow groundwater levels. Light Trap Validation We found no significant difference in the number of larval fathead minnow collected in light traps among release distances (ANOVA: F ¼ 3.11, P ¼ 0.15). Overall, light traps captured two-thirds of the available larvae (mean ¼ 20 fish/trap, SE ¼ 0.94). These data indicate that the light trap design was effective at capturing larval fish to at least a distance of 2.63 m, or within a circle 22 m2 in area. Validation of Brassy Minnow Otolith Daily Rings Age estimation using otoliths from 47 brassy minnow larvae reared in the laboratory for 1–30 d corresponded well with known ages. However, increments in otoliths of larvae reared at constant ambient temperature (208C) were difficult to distinguish, similar to results reported for Colorado pikeminnow Ptycho-

0.7 0.7 0.6 0.6 0.6 0.5

Backwater 4.2 4.9 9.9 10.7 15.7 17.3

6 6 6 6 6 6

0.6 0.6 0.6 0.6 0.5 0.7

Main channel 1.7 2.2 8.6 13.0 19.8 22.1

6 6 6 6 6 6

0.7 0.7 0.6 0.7 0.7 0.5

cheilus lucius by Bestgen and Bundy (1998). Therefore, we analyzed only brassy minnow larvae raised in the fluctuating thermal regime, which was similar to natural conditions. Fish were approximately 4.0 mm SL at hatching, and fish 1 d old had one clear increment, indicating that the first daily increment in brassy minnow is deposited on the day of hatching. Overall, daily increments were easy to distinguish, and blind increment counts were nearly identical to known age. A linear regression of estimated age on known age was highly significant (r2 ¼ 0.97, P , 0.001), and the 95% confidence interval (CI) of the slope (0.928 6 0.05) nearly overlapped a value of 1, suggesting a 1:1 relationship. Linear regressions of SL (mm) on age (SL ¼ 0.319 [age] þ 3.409, r2 ¼ 0.94, SE of intercept ¼ 0.171, SE of slope ¼ 0.012), otolith diameter (OD; lm) on age (OD ¼ 1.867 [age] þ 5.972, r2 ¼ 0.98, SE of intercept ¼ 0.638, SE of slope ¼ 0.045), and SL on OD (SL ¼ 0.171 [OD] þ 2.377, r2 ¼ 0.97, SE of intercept ¼ 0.144, SE of slope ¼ 0.005) for known age fish were all highly significant (P , 0.001). These results indicate that counts of daily rings in field-collected brassy minnow larvae provided accurate and precise estimates of age. Hatching Date Distributions A total of 4,505 brassy minnow larvae were captured during 3 years of sampling and more than 85% were captured in the upstream and middle segments (Table 3). Brassy minnow were not captured in Black Wolf Creek in 2005 nor in the downstream segment in 2006 when it was entirely dry during sampling. A total of 514 brassy minnow larvae were aged, including all larvae collected in 2005 (n ¼ 191) and 2006 (n ¼ 168) and a subsample of larvae collected in 2007 (n ¼ 155). For 2007, we randomly selected two individuals from each segment and Black Wolf Creek within each 1-mm length-class from 4 to 35 mm SL. There were few large individuals, so some length-classes had fewer than two fish from each segment. The subsampling in 2007 enabled us to develop age–length relationships that we

1574

FALKE ET AL.

FIGURE 2.—Percentage of total backwater area remaining by reach (n ¼ 2–4 backwaters each; solid lines, left y-axis), and relative groundwater stage (dashed lines, right y-axis) for five consecutive weekly measurements starting 1 June 2006 in the upstream segment, Arikaree River. Right y-axis values vary among plots.

used to assign ages to the rest of the fish collected. The slope of the SL versus estimated age (d) relationships was different in Black Wolf Creek than in the upstream, middle, and downstream segments (ANCOVA: F ¼ 337.67, P , 0.01), so we fit two separate equations, one for Black Wolf Creek (BWC) and one for the other three segments (SEG), as follows:

BWC : Age ¼ 2:757ðSLÞ 9:81 ðn ¼ 37; r 2 ¼ 0:95Þ SEG : Age ¼ 2:159ðSLÞ 4:49 ðn ¼ 118; r 2 ¼ 0:94Þ Across segments and years, the majority of brassy minnow larvae hatched between mid-April and late May (Figure 3). A subset of larvae hatched in early June in the downstream segment and Black Wolf Creek

TABLE 3.—Number of brassy minnow larvae collected and number aged from otoliths each year from three 6.4-km segments of the Arikaree River and the lower 1 km of Black Wolf Creek from 2005 to 2007. Year

Upstream

Middle

Downstream

Black Wolf Creek

Total collected

Number aged

2005 2006 2007 Total

103 42 1,393 1,538

83 58 2,166 2,307

5 0 151 156

0 68 436 504

191 168 4,146 4,505

191 168 155 514


1575

FIGURE 3.—Hatching date distributions based on otolith daily increments for brassy minnow larvae collected from the three study segments and Black Wolf Creek in the Arikaree River basin during 2005–2007. Hatching dates for larvae in 2007 were estimated from a subsample of fish (n ¼ 155), and this distribution was then applied to all fish. Bars represent 1 d.

in 2007. Brassy minnow spawned earlier, especially in the upstream segment and Black Wolf Creek, during the hot dry conditions of 2006 compared with 2005 and 2007. Brassy Minnow Habitat Transition Trends in CPUE data supported the hypothesis that brassy minnow larvae transitioned from backwater habitat to main-channel margins when backwater conditions deteriorated (Figure 4). Brassy minnow CPUE increased quickly in backwater habitats during spawning in early May. Thereafter, CPUE declined in backwaters and increased in channel-margin habitats. During June, CPUE of brassy minnow was higher in channel margin habitats than in backwater habitats. According to our hatching date estimates and relative growth estimates (see below), age of these larvae ranged from 30 to 40 d and length from 17 to 22 mm SL during this transition from backwaters to channel margins.

FIGURE 4.—Mean 6 SE catch per unit effort of brassy minnow larvae during spring and summer 2007 in backwater and channel-margin habitats in the upstream segment of the Arikaree River.

1576

FALKE ET AL.

TABLE 4.—Summary of model selection statistics for the top four multiple linear regression models (Akaike weight [wi] . 0.05) and the global model of survival of 25 cohorts of brassy minnow larvae in 14 backwater spawning habitats in the Arikaree River in 2007. Abbreviations are as follows: L-L ¼ the log-likelihood; DAICc ¼ the difference in the corrected Akaike information criterion (AICc) value for a particular model compared with the top-ranked model; K ¼ the number of parameters, including the intercept and residual variance; GSDD ¼ growing season degree-days; and MHD ¼ mean hatching date (day of the year). Model Rate, Rate, Rate, Rate, Rate,

a

L-L

2

area, depth, GSDD , GSDD area, GSDD, GSDD2, MHDb, MHD2 GSDD, GSDD2 area, GSDD, GSDD2 area, depth, abundance, segment, GSDD, GSDD2, MHD, MHD2 (global model)

Influence of Spawning Habitat Characteristics on Cohort Survival The highest ranked model (wi ¼ 0.27) suggested that the rate of backwater drying, initial area and depth, and the linear and quadratic terms for GSDD had the most influence on survival of 25 cohorts of brassy minnow in 2007 (Table 4). However, there was considerable support for three other models (range of wi ¼ 0.17– 0.23), one of which included the linear and quadratic terms for mean hatching date, so estimates from those top models were model-averaged. Survival was predicted to be higher in backwaters that dried more slowly and that were initially larger and deeper (Table 5). Survival was also higher for cohorts that hatched in the middle of the spawning period (Figure 5) and at a moderate number of GSDD (median ¼ 810). For example, when all other parameters are held constant, reducing depth by 25% reduced daily survival by approximately 0.01/d. Over 30 d, this decrease in daily survival results in a 26% increase in mortality (0.9930 ¼ 0.74/d). Models including the segment factor were not included in top models, probably because most variation due to segment (e.g., faster drying rates in middle than upstream) was accounted for by other variables. Nevertheless, a simple t-test comparing

51.24 53.26 47.19 48.80 54.30

AICc

DAIC

wi

K

81.90 0.00 0.27 7 81.53 0.37 0.23 8 81.20 0.70 0.19 5 80.92 0.98 0.17 6 66.29 15.61 ,0.01 11

arcsine- and square-root-transformed mean cohort survival between the two segments showed that survival was higher in the upstream segment (0.84/d) than in the middle segment (0.68/d; t ¼ 2.55, P ¼ 0.03). Relative Growth and Survival among Years Growth of brassy minnow larvae in backwaters differed among years with different climate conditions. The slopes of mean SL as a function of age were significantly different among years (ANCOVA: F ¼ 158.63, P , 0.001) and the growth rate each year was significantly different from the others (Tukey’s HSD test: all P 0.05). The slopes (mean 6 SE), which can be interpreted as growth rates (mm/d), were highest in 2007 (0.41 6 0.02), moderate in 2005 (0.30 6 0.03), and lowest in 2006 (0.25 6 0.03). Survival of brassy minnow larvae also differed among years with different climate conditions when data were pooled over all habitat types and segments. Estimates of daily survival (S, 1/d) were 0.894, 0.839 and 0.897 for 2005, 2006, and 2007, respectively, which were different among years (ANCOVA: F ¼ 158.63, P , 0.001). Survival was significantly lower in the dry year 2006 than in the wetter years 2005 and 2007 (both P , 0.001) based on Tukey’s HSD post

TABLE 5.—Model-averaged parameter estimates, unconditional SE values, and lower and upper 95% confidence limits (CLs) for covariates predicting survival of 25 cohorts of brassy minnow larvae in 14 backwater spawning habitats in the Arikaree River in 2007. Results are based on the top four multiple-regression models, which were responsible for most (86%) of the collective model weight (see Table 4). Covariate

Parameter estimate

Lower 95% CL

Upper 95% CL

Intercept Rate Area Depth GSDD GSDD2 MHD MHD2

5.45 6 3.24 0.0025 6 0.00086 0.00011 6 0.000062 0.00044 6 0.000044 0.0061 6 0.0026 9.82 3 107 6 2.30 3 108 0.035 6 0.0029 0.00014 6 0.000044

10.80 0.0039 0.0000054 0.00036 0.0018 1.02 3 106 0.030 0.00021

0.13 0.0010 0.00021 0.00051 0.010 9.45 3 107 0.039 0.000064


1577

ter pumping, and climate, which operate at multiple spatial and temporal scales over the riverscape and influence persistence and abundance of brassy minnow across stream reaches. Here, we synthesize information from this research and related studies (e.g., Scheurer et al. 2003; Falke et al., in press a and b) to describe the early life history of brassy minnow and discuss how groundwater pumping, drought, and climate change may influence habitat drying and prospects for persistence and conservation of this plains stream minnow. Brassy Minnow Spawning

FIGURE 5.—Estimated survival of brassy minnow larvae (yaxis) as a function of the rate of backwater habitat drying (xaxis) and mean cohort hatching date (z-axis) in the Arikaree River. Parameters (Table 5) were estimated using multiple linear regression and averaged from the top four models ranked according to AICc.

hoc comparisons. No difference in survival rates between 2005 and 2007 could be detected (P ¼ 0.286). Recruitment among Years and Segments Based on length-frequency histograms, size structure of brassy minnow populations varied among years, and probably was influenced by survival of age-0 fish to late summer, and adults overwinter (Figure 6). The histograms indicated that age-0 fish were a maximum of 55 mm FL and age-1 and older fish were 55 mm FL or larger in August each year. Age-0 fish were relatively abundant in 2005 and 2007 during August but were in low abundance in 2006 in both segments. Low survival during the late-spring–early-summer larval stage probably contributed to a low number of age-0 fish in August 2006. In 2005, many age-1 and older fish were present in the upstream segment, but few were collected in the middle segment. However, during 2006 adult fish were present at high relative abundance in both segments, suggesting high survival of age-0 brassy minnow produced in 2005. Relative abundance of adults in 2007 was low in both segments, indicating low survival of age-0 fish and adults that remained by August 2006. Discussion Our research showed that brassy minnow spawning, growth, survival, and recruitment were strongly influenced by habitat drying among different segments of the Arikaree River. In turn, drying is driven by interactions among stream geomorphology, groundwa-

In previous research, we found brassy minnow spawning was not initiated by any obvious hydrologic cue, but instead commenced once water temperature exceeded a critical threshold (Falke et al., in press a). We and others have found that brassy minnow spawn in shallow, vegetated backwater habitats at about 670 cumulative growing season degree-days, a condition typically met by mid-April in the Arikaree River, and continued for about a month (Copes 1975; Scheurer et al. 2003; Falke et al., in press a). We found that backwaters are benign habitats that remain warmer than main-channel habitats during winter and early spring but stay cooler during late spring and early summer. Brassy minnow eggs are adhesive and adults probably attach them to vegetation or other structure in spawning habitats, which keeps embryos aerated and away from the potentially smothering silt substrate common in low-velocity backwaters. Eggs develop quickly and larvae hatch within 3 d at 4.0 mm SL. Adhesive eggs and rapid incubation and growth of larvae are reproductive characteristics that allow brassy minnow to thrive in harsh unpredictable environments (Fausch and Bestgen 1997). Backwater habitats in the Arikaree River are critical spawning habitats for brassy minnow. We found that the number of backwaters was dependent on the hydrologic context of the segment and interannual climate variation. As a result, these dynamic habitats potentially control (i.e., limit) the amount of spawning across the riverscape and from year to year. The number and size of backwaters was greatest in the perennial upstream segment and least in the highly intermittent downstream segment and varied among 3 years that ranged from moderately wet to very dry. In the two study segments that supported backwater habitats, area was highest in 2005 and 2007, both years with relatively wet spring seasons, whereas it was much lower in the extremely dry conditions of 2006. The peak of brassy minnow spawning occurred from mid-April to late May, a period in which backwaters were abundant in the upstream segment across the 3

1578

FALKE ET AL.

FIGURE 6.—Length-frequency histograms of adult brassy minnow collected in August 2005 to 2007 in pools in the upstream and middle segments of the Arikaree River.

years, but where drying rates were rapid in the middle segment, even in the relatively wet conditions of 2007. The most extreme habitat conditions were found in the downstream segment and Black Wolf Creek where spawning and rearing habitats were highly ephemeral; yet a few brassy minnow larvae were found in these habitats when they were available, as a result of both adult spawning and larval drift. Segment-scale transport processes are probably important and allow brassy minnow to colonize backwaters that are formed far from spawning locations. Moreover, owing to the

patchy distribution of spawning habitats across both space and time, brassy minnow must quickly colonize spawning habitats from other sources, indicating that source-sink population dynamics are probably important in the Arikaree River basin (Falke and Fausch 2010). Brassy Minnow Survival, Growth, and Recruitment Survival of brassy minnow larvae in backwater habitats was influenced by complex interactions between habitat drying that determined habitat size,


and water temperature driven by groundwater connectivity and climate. At the local scale, we found that the rate of habitat drying, habitat area and depth, growing season days, and timing of hatching were important predictors of cohort survival. Larvae in fast drying habitats, which were probably small and relatively shallow, had lower survival compared with backwaters that dried more slowly and were larger and deeper. Survival of brassy minnow larvae was highest in large, deep, backwater habitats, which are associated with groundwater input and most of which occurred in the upstream segment. For example, survival of larvae was predicted to decrease 0.01/d with a 25% decrease in backwater depth, all other factors being equal. Survival of larvae in large backwater spawning habitats is probably higher because physicochemical conditions in these habitats are more benign than in small backwaters, and larval prey resources may be greater (Cushing 1975, 1990). Temperature controls larval metabolic rates, which directly affect growth and ultimately condition of larvae (Blaxter 1992). We found that the number of growing season degree-days before hatching was an important predictor of larval survival. Survival of larvae that hatch in the middle of the spawning period was highest compared with those hatched early and late, which is probably due to moderate water temperatures that conferred the advantage of adequate growth conditions. Owing to higher metabolic costs, cohorts spawned in relatively warm backwaters late in the season are probably unable to meet minimum food requirements during the critical period following yolk sac absorption, resulting in lower survival. Additionally, larvae that hatch late in the spawning period are more likely to be trapped in rapidly drying spawning habitats as physicochemical conditions deteriorate and connections with the main channel are severed. Across years, growth and survival of larvae were lowest in 2006, the driest year, based on our analyses of growth rates of larval cohorts, catch-curve analyses, and length-frequency histograms in August. Low juvenile recruitment in 2006 and low overwinter survival probably resulted in few spawning adults in 2007. However, despite low overwinter survival, adult brassy minnow were able to recolonize spawning habitats (Falke et al., in press a) and successfully spawn, as evidenced by the high abundance of brassy minnow larvae collected in spring 2007. Clearly this species has the capability of producing large yearclasses when spawning habitat is plentiful, such as in 2007. Maintenance of those backwaters during early summer is critical for producing recruits from those young, but is a separate process linked to groundwater availability, which controls stream drying and back-

1579

water persistence. Overall, these results indicate that environmental conditions during spawning (i.e., climate) were important mechanisms that influenced brassy minnow population structure. The catch curves we used to estimate survival of brassy minnow larvae are based on several restrictive assumptions: (1) mortality rate is constant with age, (2) larvae recruit to the sampling gear at a similar size, (3) detectability is equal across size, and (4) for larvae, increment counts correspond to actual age in days (Robson and Chapman 1961; Ricker 1975; Essig and Cole 1986). Recruitment to the sampling gear was probably not an issue, because larvae were captured at all ages including as young as 1 d old. Also, detectability was high (.0.8) across the range of sizes we modeled in the catch-curve analysis (Falke et al., in press a), and increment counts were validated as accurate. As such, these three assumptions hold. Mortality rate was probably not constant across ages because larvae are susceptible to different mortality factors from the early to late larval periods. However, our objective was to use survival rates generated from the catch curves in a relative, not absolute, measure, so violating this assumption was unlikely to strongly bias our conclusions. Dispersal of Age-0 Brassy Minnow Transport of brassy minnow larvae into the downstream segment was apparent, based on evidence from age estimates and distributions of hatching dates. During 2005, five brassy minnow larvae were collected in channel-margin habitats in the downstream segment that were wet for only 2 weeks following a precipitation event. These five larvae were estimated to have hatched before the habitats became available (i.e., before the rainstorm), indicating that they must have hatched elsewhere. The most likely source was passive drift from upstream habitats because the larvae were too small to swim from downstream (mean SL ¼ 10.1 mm). Secondly, in 2007 the peak of hatching in the downstream segment was about a week earlier than in Black Wolf Creek (Figure 3), even though this tributary provided most of the flow that created the habitat in the main river, and the two locations were less than 1 km apart. Habitat conditions in the downstream segment were most likely not conducive for brassy minnow spawning because no backwater habitats were available. These results suggested larvae collected earliest in the downstream segment were not from Black Wolf Creek, but instead had drifted from habitats upstream. Dispersal of age-0 brassy minnow may occur at two stages due to two different mechanisms. First, larvae may be passively dispersed by high flow events, and

1580

FALKE ET AL.

FIGURE 7.—Conceptual model of generalized trends in spawning, rearing, and refuge habitat availability for early life stages of brassy minnow within river segments with differing flow regimes across a Great Plains riverscape. Arrows indicate the direction and relative magnitude (arrow width) of recruitment and survival within these habitats and segments. Also displayed are the population viability and population type expected to be found within each type of segment. Conditions within each segment type are expected to become drier (i.e., to shift to the right) with continued groundwater pumping, drought, and climate change (Falke et al., in press b).

the distance over which they travel probably depends on the magnitude and duration of discharge. During spring 2005, we found larvae in ephemeral channelmargin habitats in the downstream segment; these larvae were older than the ephemeral habitats and apparently had been displaced from upstream. Under this scenario, passive dispersal is an important driver of source-sink dynamics. However, larvae may also be passively dispersed into unsuitable habitats that dry quickly, such as the lower reaches of the Arikaree River (Falke et al., in press b), and suffer high mortality from desiccation or starvation. Second, active dispersal from backwaters to mainchannel rearing habitats by young brassy minnow is probably owing to declining resources in backwaters caused by drying or competition with larvae of other species, or both. (Falke et al., in press a) found that larvae of other plains fish species (e.g., central stoneroller Campostoma anomalum and fathead minnow) use backwater habitats during the same period as brassy minnow. Regardless of the mechanism, we found a clear pattern in the decline of brassy minnow abundance in backwater habitats over time, concurrent with increased abundance in channel-margin habitats. Once in the main channel, larvae need to find suitable

refuge pools as habitats continue to dry during summer. In the upstream segment, where deep pools are common, refugia are probably easy to access. However, in river segments with few or no pools (e.g., middle and downstream segments) the active dispersal stage may be prolonged and early life stages may be subject to starvation or predation while they seek suitable refugia. As a result, the dispersal stage is probably critical for recruitment and population regulation of brassy minnow in the Arikaree River. Synthesis and Conservation Implications Habitat drying is a critical factor that influences growth and survival of brassy minnow in Great Plains streams across all life stages (Scheurer et al. 2003; Falke et al., in press a). The quality and quantity of spawning, rearing, and refuge habitats for brassy minnow, and the connections among them, reflect the segments in which those habitats are set (Figure 7). Here we draw general inferences from the large Arikaree River segments we studied across 3 years to provide predictions across life stages and population types (e.g., sources and sinks) for population viability of plains fishes in other streams, which may also be affected by altered streamflows.


Relatively wet segments, such as our upstream study segment, support large populations and have high recruitment and survival of fish across life stages. Alluvial deposits through which these segments flow are hydraulically connected to regional aquifers (Falke et al., in press b). Even during dry years, these segments support many large backwater spawning habitats that allow for successful recruitment of larvae. When larvae reach a threshold in body size, connections among habitats that persist through summer allow them to move into main-channel-margin habitats and then to pools that offer suitable conditions for growth (Scheurer et al. 2003; Falke and Fausch 2010). During winter, moderate main-channel temperatures in abundant deep pools offer refuge from freezing conditions (Labbe and Fausch 2000; Falke et al., in press a). These wet segments probably serve as source populations that provide demographic support for (i.e., allow for the persistence of) populations in sink habitats (Falke and Fausch 2010). Intermittent segments, such as our middle segment, may support high recruitment of larvae in some backwaters during the spring, but harsh conditions in the summer and winter may limit survival of age-0 and older fish (Figure 7). These segments lack lateral alluvial groundwater input owing to their bedrock geology. In dry conditions, such as those in 2006, few backwater habitats occur or persist. Connections to adjacent channel-margin habitats may dry before larvae are large enough to move between these habitat types. Larvae that are successful at moving into the main channel are faced with warm, rapidly drying habitats and may be forced into isolated refuge pools for the summer and winter months, where mortality in frozen pools is probably high and the probability of surviving to spawn the next spring is low. Because a moderate density of high-quality spawning backwaters occurs in intermittent segments, during high flow years these segments also may provide individuals to support poor quality habitats downstream (i.e., sinks). However, during dry years, low connectivity among spawning and rearing habitats, and long dry reaches, preclude larval transport out of intermittent segments. Dry segments, such as the downstream segment, are poor quality habitats. Similar to the intermittent segments, dry segments do not receive lateral groundwater input. As a result, in dry years no habitats exist, and even in wet years no backwater habitats in which brassy minnow prefer to spawn are available (Falke et al., in press a; Figure 7). Channel margins or small tributaries (e.g., Black Wolf Creek) may provide some opportunity for spawning or rearing of larvae that drift from upstream, but water temperatures are very high, channel margins quickly dry, recruitment is low, and

1581

survival past the larval stage is likely to be very low or nonexistent unless streamflow is augmented. Populations in dry segments are mainly supported by input of individuals from upstream populations and, hence, fit the classic definition of sink populations (Falke and Fausch 2010). We have shown that both interannual variability in climate and the hydrologic context of segments along the riverscape have a strong influence on habitat availability and recruitment of brassy minnow in the Arikaree River. Although our data were collected during a drought period, droughts are a common occurrence on the Great Plains, and they may become more frequent and more intense in the future due to climate change (Ojima and Lackett 2002; Falke et al., in press a). However, droughts are natural phenomena to which plains fishes have become adapted over evolutionary time. As we have shown, life history adaptations of plains stream fishes, such as brassy minnow, allow for them to survive and quickly repopulate entire segments following droughts and other natural disturbance. A concern that is more pressing than periodic droughts facing stream fish populations across the western Great Plains is permanent streamflow reduction owing to diversions, reservoir storage, and irrigation pumping. In the Arikaree River, overuse of groundwater resources by irrigated agriculture, when combined with drought, has led to alterations in periodicity and magnitude of flows, which is the likely cause of five species being extirpated from the basin (Falke et al., in press b). For example, the downstream segment, and reaches downstream from that point, formerly supported a relatively diverse native fish community as recently as 1977 (Cancalosi 1980). Currently, these reaches are highly ephemeral or permanently dry, probably owing to irrigation wells that pump directly from the alluvial aquifer. Furthermore, groundwater–fish habitat models predict that at current pumping rates, hydrologic alteration will become more severe in the future resulting in further losses of stream fish habitat availability and habitat connectivity (Falke et al., in press b). The long-term implications for brassy minnow populations are that even conditions in wet and intermittent segments, such as our upstream segment, will continue to shift to drier states resulting in fewer spawning habitats and lower recruitment and survival of fish (Figure 7). Our research underscores the importance of understanding mechanisms by which human actions like groundwater pumping affect population dynamics and processes, even of fishes that are native to relatively harsh environments. As such, we suggest that conservation efforts should focus on increasing streamflows,

1582

FALKE ET AL.

especially during summer, to maintain spawning, rearing, and refuge habitats that are critical to brassy minnow population persistence. Effective conservation will require explicit consideration of the adaptations of this species to harsh environments and the understanding that these adaptations can protect these fish populations only to a point. Above some level of human-caused perturbation, even adaptations to harsh environments are unlikely to protect stream fish populations in Great Plains streams with highly altered flow regimes. As a result, as is the case with the Arikaree River, the risk of extirpation will increase with habitat drying. Acknowledgments We thank K. Bentley, N. Cathcart, A. Ficke, J. Hammer, M. Hill, A. Klug, W. Pate, C. Saunders, and Z. Underwood for their help in the field and laboratory. R. Fitzpatrick and S. Seal provided valuable assistance with collection, spawning, and rearing of brassy minnow for otolith increment validation. C. Myrick generously provided facilities for our light-trap experiment at the Foothills Fisheries Laboratory, Fort Collins, Colorado. We also thank D. Winkelman, D. Durnford, and L. Bailey for their input and comments on study design and analyses. W. Burnidge (The Nature Conservancy, Boulder, Colorado) provided valuable logistical support. This research was funded by a grant to K.D.F. from the Colorado Division of Wildlife, administered by T. Nesler. References Benke, A. C. 1990. A perspective on America’s vanishing streams. Journal of the North American Benthological Society 9:77–88. Bestgen, K. R., D. W. Beyers, J. A. Rice, and G. B. Haines. 2006. Factors affecting recruitment of young Colorado pikeminnow: synthesis of predation experiments, field studies, and individual-based modeling. Transactions of the American Fisheries Society 135:1722–1742. Bestgen, K. R., and J. M. Bundy. 1998. Environmental factors affect daily increment deposition and otolith growth in young Colorado squawfish. Transactions of the American Fisheries Society 127:105–117. Blaxter, J. H. S. 1992. The effect of temperature on larval fishes. Netherlands Journal of Zoology 42:336–357. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information theoretic approach. Springer-Verlag, New York. Campana, S. E., and J. D. Neilson. 1985. Microstructure of fish otoliths. Canadian Journal of Fisheries and Aquatic Sciences 42:1014–1032. Cancalosi, J. J. 1980. Fishes of the Republican River basin. Master’s thesis. Colorado State University, Fort Collins. CDOW (Colorado Division of Wildlife). 2007. Endangered,

threatened, and species of special concern. Available: wildlife.state.co.us/WildlifeSpecies/SpeciesOfConcern/ Fish/FishOfConcern.htm. (September 2008). Copes, F. A. 1975. Ecology of the brassy minnow, Hybognathus hankinsoni (Cyprinidae). University of Wisconsin–Stevens Point, Museum of Natural History, Stevens Point. Crecco, V. A., and T. F. Savoy. 1984. Effects of fluctuations in hydrographic conditions on year-class strength of American shad (Alosa sapidissima) in the Connecticut River. Canadian Journal of Fisheries and Aquatic Sciences 41:1216–1223. Cushing, D. H. 1975. Marine ecology and fisheries. Cambridge University Press, Cambridge, UK. Cushing, D. H. 1990. Plankton production and year-class strength in fish populations: an update of the match/ mismatch hypothesis. Advances in Marine Biology 14:1– 122. Dudley, R. K., and S. P. Platania. 2007. Flow regulation and fragmentation imperil pelagic-spawning riverine fishes. Ecological Applications 17:2074–2086. Durham, B. W., and G. R. Wilde. 2006. Influence of stream discharge on reproductive success of a prairie stream fish assemblage. Transactions of the American Fisheries Society 135:1644–1653. Durham, B. W., and G. R. Wilde. 2009. Population dynamics of the smalleye shiner, an imperiled cyprinid fish endemic to the Brazos River, Texas. Transactions of the American Fisheries Society 138:666–674. Essig, R. J., and C. F. Cole. 1986. Methods of estimating larval fish mortality from daily increments in otoliths. Transactions of the American Fisheries Society 115:34– 40. Falke, J. A., and K. D. Fausch. 2010. From metapopulations to metacommunities: linking theory with empirical observations of the spatial population dynamics of stream fishes. Pages 207–233 in D. A. Jackson and K. B. Gido, editors. Community ecology of stream fishes: concepts, approaches and techniques. American Fisheries Society, Symposium 73, Bethesda, Maryland. Falke, J. A., K. D. Fausch, K. R. Bestgen, and L. L. Bailey. In press a. Spawning phenology and habitat use in a Great Plains stream fish assemblage: an occupancy estimation approach. Canadian Journal of Fisheries and Aquatic Sciences. Falke, J. A., K. D. Fausch, R. Magelky, A. Aldred, D. S. Durnford, L. K. Riley, and R. Oad. In press b. The role of groundwater pumping and drought in shaping ecological futures for stream fishes in dryland river basin of the western Great Plains, USA. Ecohydrology. Fausch, K. D., and K. R. Bestgen. 1997. Ecology of fishes indigenous to the central and southwestern Great Plains. Pages 131–166 in F. L. Knopf and F. B. Samson, editors. Ecology and conservation of Great Plains vertebrates. Springer-Verlag, New York. Fausch, K. D., C. E. Torgersen, C. V. Baxter, and H. W. Li. 2002. Landscapes to riverscapes: bridging the gap between research and conservation of stream fishes. BioScience 52:483–498. Freeman, M. C., Z. H. Bowen, K. D. Bovee, and E. R. Irwin. 2001. Flow and habitat effects on juvenile fish


abundance in natural and altered flow regimes. Ecological Applications 11:179–190. Gutentag, E. D., F. J. Heimes, N. C. Krothe, R. R. Luckey, and J. B. Weeks. 1984. Geohydrology of the High Plains Aquifer in parts of Colorado, Kansas, Nebraska, New Mexico, Oklahoma, South Dakota, Texas, and Wyoming. U.S. Geological Survey, Professional Paper 1400-B. Halpern, B. S., S. D. Gaines, and R. R. Warner. 2005. Habitat size, recruitment, and longevity as factors limiting population size in stage-structured species. American Naturalist 165:82–94. Harvey, B. C. 1987. Susceptibility of young-of-the-year fishes to downstream displacement by flooding. Transactions of the American Fisheries Society 116:851–855. Hubert, W. A., and K. M. Gordon. 2007. Great Plains fishes declining or threatened with extirpation in Montana, Wyoming, or Colorado. Pages 3–13 in M. J. Brouder and J. A. Scheurer, editors. Status, distribution, and conservation of native freshwater fishes of western North America. American Fisheries Society, Symposium 53, Bethesda, Maryland. Killgore, K. J. 1994. Design and application of a larval fish trap. Natural Resources Conservation Service, Wetland Reserve Program, Technical Note, FW-EV-3.1, Washington, D. C. Labbe, T. R., and K. D. Fausch. 2000. Dynamics of intermittent stream habitat regulate persistence of a threatened fish at multiple scales. Ecological Applications 10:1774–1791. Leach, S. D., and E. D. Houde. 1999. Effects of environmental factors on survival, growth, and production of American shad larvae. Journal of Fish Biology 54:767–786. Lohr, S. C., and K. D. Fausch. 1997. Multiscale analysis of natural variability in stream fish assemblages of a western Great Plains watershed. Copeia 1997:706–724. Ludsin, S. A., and D. R. DeVries. 1997. First-year recruitment of largemouth bass: the interdependency of early life stages. Ecological Applications 7:1024–1038. Malmqvist, B., and S. D. Rundle. 2002. Threats to the running water ecosystems of the world. Environmental Conservation 29:134–153. Manly, B. F. J. 2007. Randomization, bootstrap, and Monte Carlo methods in biology. Chapman and Hall, London. Matthews, W. J. 1987. Physiochemical tolerance and selectivity of stream fishes as related to their geographic ranges and local distributions. Pages 111–120 in W. J. Matthews and D. C. Heins, editors. Community and evolutionary ecology of North American stream fishes. University of Oklahoma Press, Norman. Matthews, W. J., and E. G. Zimmerman. 1990. Potential

1583

effects of global warming on native fishes of the southern Great Plains and the southwest. Fisheries 15(6):26–32. May, R. C. 1973. Larval mortality in marine fishes and the critical period concept. Pages 1–19 in J. H. Blaxter, editor. The early life history of fish. Springer-Verlag, New York. McGuire, V. L., M. R. Johnson, R. L. Schieffer, J. S. Stanton, S. K. Sebree, and I. M. Verstraeten. 2003. Water in storage and approaches to ground-water management, High Plains Aquifer, 2000. U.S. Geological Survey, Circular 1243. Mion, J. B., R. A. Stein, and E. A. Marschall. 1998. River discharge drives survival of larval walleye. Ecological Applications 8:88–103. Ojima, D. S., and J. M. Lackett. 2002. Preparing for a changing climate: the potential consequences of climate variability and change – central Great Plains. Colorado State University, Report for the U.S. Global Change Research Program, Fort Collins. Ricker, W. E. 1975. Compilation and interpretation of biological statistics of fish populations. Fisheries Research Board of Canada Bulletin 191. Robson, D. S., and D. G. Chapman. 1961. Catch curves and mortality rates. Transactions of the American Fisheries Society 90:181–189. Scheidegger, K. J., and M. B. Bain. 1995. Larval fish distribution and microhabitat use in free-flowing and regulated rivers. Copeia 1995:125–135. Scheurer, J. A. 2001. Brassy minnow in Colorado plains streams: identification, historical distribution, and habitat requirements at multiple scales. Master’s thesis. Colorado State University, Fort Collins. Scheurer, J. A., K. D. Fausch, and K. R. Bestgen. 2003. Multiscale processes regulate brassy minnow persistence in a Great Plains river. Transactions of the American Fisheries Society 132:840–855. Schlosser, I. J. 1985. Flow regime, juvenile abundance, and the assemblage structure of stream fishes. Ecology 66:1484–1490. Squires, A. 2007. Groundwater response functions and water balances for parameter estimation and stream habitat modeling. Master’s thesis. Colorado State University, Fort Collins, Colorado. Starrett, W. C. 1951. Some factors affecting the abundance of minnows in the Des Moines River, Iowa. Ecology (Washington, D.C.) 32:13–27. Stevenson, D. K., and S. E. Campana. 1992. Otolith microstructure examination and analysis. Canadian Special Publication of Fisheries and Aquatic Sciences 117.