|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 MANIPULATION OF FISH VITAL RATES THROUGH ECOSYSTEM EXPERIMENTATION IN A REGULATED RIVER By COLTON FINCH A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2012
2 2012 Colton Finch
3 ACKNOWLEDGMENTS I thank Dr. William Pine III, the chair of my supervisory committee, for his generous and insightful support, as well as committee members Dr. Carl Walters and Dr. Micheal Allen for their encouragement and mentoring throughout my time at the University of Florida. I thank the United States Bureau of Reclamation and the Unit ed States Geological Survey Grand Canyon Monitoring and Research Center for their financial and logistical support. I thank Brandon Gerig, Mike Dodrill Mike Yard, the United States Fish and Wildlife Service, the Arizona Game and Fish Department and Bri an Dierker for invaluable assistance with data collection and management. I also thank Dr. Robert Ahrens, Dr. Charles Yackulic, Dr. Josh Korman and James Hines for coding and statistical assistance.
4 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 3 LIST OF TABLES ................................ ................................ ................................ ............ 5 LIST OF FIGURES ................................ ................................ ................................ .......... 6 LIST OF ABBREVIATIONS ................................ ................................ ............................. 8 ABSTRACT ................................ ................................ ................................ ..................... 9 CHAPTER 1 INTRODUCTION AND BACKGROUND ................................ ................................ 11 2 INFLUENCING GROWTH RATES THROUGH EXPERIMENTAL FLOW TREATMENTS ................................ ................................ ................................ ....... 14 Introduction ................................ ................................ ................................ ............. 14 Methods ................................ ................................ ................................ .................. 17 Study Site ................................ ................................ ................................ ......... 17 Sampling Techniques ................................ ................................ ....................... 18 Data Analysis ................................ ................................ ................................ ... 20 Results ................................ ................................ ................................ .................... 21 Discussion ................................ ................................ ................................ .............. 22 3 ARTIFICIAL DISCHARGE FLUCTUATIONS AND JUVENILE FISH SURVIVAL RATES ................................ ................................ ................................ .................... 40 Introduction ................................ ................................ ................................ ............. 40 Methods ................................ ................................ ................................ .................. 45 Study Site ................................ ................................ ................................ ......... 45 Sampling Techniques ................................ ................................ ....................... 46 Data Analysis ................................ ................................ ................................ ... 47 Results ................................ ................................ ................................ .................... 51 Simulation Results ................................ ................................ ............................ 51 Flow effects and Juvenile Humpback Chub Apparent Survival ........................ 53 Discussion ................................ ................................ ................................ .............. 55 4 CONCLUSION ................................ ................................ ................................ ........ 72 LIST OF REFERENCES ................................ ................................ ............................... 76 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 85
5 LIST OF TABLES Table page 2 1 Sample sizes of juvenile humpback chub recaptured for growth purposes. ......... 28 3 1 Capture recapture summary table of data provided to MARK for analysis. .......... 60 3 2 Apparent survival models for juvenile humpback chub in Grand Canyon from 2009 ................................ ................................ ...... 61
6 LIST OF FIGURES Figure p age 2 1 Map of study area near confluence of Colorado and Little Colorado Rivers in northern Arizona, containing the Little Colorado River aggregation of humpback chub. ................................ ................................ ............................. 29 2 2 Colorado River discharge August 22nd September 1 Ferry gauge, 98 km upstream from my sampling universe ............................. 30 2 3 Colorado River water temperatures as downstream from Glen Canyon Dam ................................ ............................. 31 2 4 Colora do River water temperatures 1994 2002, 2003 2008, and during this study, with a model of pre Glen Canyon Dam historical temperatures ................................ ................................ ................................ ... 32 2 5 Daily di scharge in the Little Colorado River near the confluence with the Colorado River over the three year period of this study (2009 2012) ............. 33 2 6 Sampling calendar ................................ ................................ .......................... 34 2 7 Detail map of study area. ................................ ................................ ................ 35 2 8 Distributions of the mean daily growth ra tes for juvenile humpback chub in the mainstem Colorado River between July and August ................................ 36 2 9 Distributions of the mean daily growth rates for juvenile humpback chub in the mainstem Colorado River between September and October ................... 37 2 10 Distributions of the mean daily growth rates for juvenile humpback chub in the Little Colorado River between July and August or September ................. 38 2 11 Distributions of the mean daily growth rates for juvenile humpback chub in the Little Colorado River between September and October ........................... 39 3 1 Colorado River mainstem flow treatments over the course of this study ........ 62 3 2 Length frequency histogram of juvenile humpback chub considered in this survival analysis ................................ ................................ ............................. 63 3 3 Apparent survival vs. capture probability simulations using mark/recapture data generated in GENCAPH1 and analyzed using Program MARK: Phi(t) p(.) and Phi(.) p(t) models ................................ ................................ ............... 64
7 3 4 Apparent survival vs. capture probability simulations using mark/recapture data generated in GENCAPH1 and analyzed using Program MARK: Phi(t ) p(t) models ................................ ................................ ................................ ..... 65 3 5 Apparent survival vs. capture probability simulations using mark/recapture data generated in GENCAPH 1 and analyzed using Program MARK: Phi(flow) p(t) models ................................ ................................ ....................... 66 3 6 Apparent survival vs. capture probability simulations usin g mark/recapture data generated in GENCAPH1 and analyzed using Program MARK: Phi(flow) p(flow) models ................................ ................................ ................. 67 3 7 Trip specifi c capture probability of juvenile humpback chub over the thirteen recapture opportunities based on three leading models. ................... 68 3 8 Annual apparent survival estimates for juvenile humpback chub, grouped dependent capture probability, see Figure 3 3). ................................ ............................... 69 3 9 Annual apparent survival estimates for juvenile humpback chub, separated by flow treatments. ................................ ................................ ......................... 70 3 10 Annual apparent survival estimates for juvenile humpback chub, as derived by the constant model (time dependent capture probability, see Figure 3 3). ................................ ................................ ................................ ..... 71
8 LIST OF ABBREVIATION S AZGFD Arizona Game and Fish Department BIA Bureau of Indian Affairs BOR United States Bureau of Reclamation FWS United States Fish and Wildlife Service GCMRC Grand Canyon Monitoring and Research Center LCR Little Colorado River NPS National Park Service NSE Nearshore Ecology, a juvenile native fish research project funded by the Grand Canyon Monitoring and Research Center PIT passive integrated transponder rkm Glen Canyon Dam at Lake Powell USGS United States Geological Survey VIE visible implant elastomer WAPA Western Area Power Authority
9 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science MANIPULATION OF FISH VITAL RATES THROUGH ECOSYSTEM EXPERIMENTATION IN A REGULATED RIVER By Colton Finch December 2012 Chair: William Pine, III Major: Wildlife Ecology and Conservation Managers of fish populations often use growth and survival of juveniles as important descriptors of population health. Juvenile fish are sensitive to environmental perturbations, so monitor ing their vital rates allow s managers in regulated systems such as dam tail waters to directly evaluate the effects of flow operations on fish populations. Despite this direct pathway less is known about juvenile fish compared to adults in these habitats as they are often more difficult to collect due to smaller body size and differential habitat use In this study I evaluated individual growth and apparent survival rate of juvenile humpback chub Gila cypha (<200 mm total length) in the Grand Canyon reach of the Colorado River Arizona for three years. I sampled during typical fluctuating hydropower flows, and during experimental steady dam releases. Growth in juvenile humpback chub declined during steady flows versus fluctuating flows, despite temperatures being roughly equivalent across flow treatments A nnual a pparent survival of juvenile humpback chub initially tagged at <100 m m total length was not significantly different between flow treatments and ranged from 37 67% depending on the model Steady flo w experiments of the magnitude and timing that I observed did not improve growth or apparent survival rates of juvenile humpback chub in Grand Canyon. This
10 suggests that juvenile humpback chub growth and survival is robust to the flow fluctuations observed during 2009 2011 an d more extreme flow treatments such as higher or lower flows, or longer duration experiments, are likely required before differences in these key vital rates are detected
11 CHAPTER 1 INTRODUCTION AND BACKGROUND Humans are ecosystem engineers that often dramatically change landscape attributes and resource availability through land use alterations (e.g. conversion to agriculture or urbanization of landscapes, dewatering of rivers, reservoir construct ion, fire suppression; Foley et al 2005). Humans also introduce novel species that increase competition and predation pressures. If native species do not have serendipitously advantageous morphology or behavior (Gould and Vrba 1982) or are unable to rapidly adapt to the new ecological com munity, these introductions often lead to local extirpation or even extinction (Savidge 1987, Oghutu Ohwayo 1990). The loss of commercially or socially important native species can have tremendous financial, ecological, or social consequences, and legislat ion and public pressure often requires managing agencies to seek remedial action. profile example of human influence at an ecosystem level. Fire suppression on the forested north and sout h rims of Grand Canyon (Fule et al 2002), regulation of the Colorado River due to the construction and operati on of Glen Canyon Dam (Gloss et al 2005), the influx of nearly 5 million human visitors per year (NPS 2012), and the introduction of non native species such as rocky mountain elk (Truett 1996), burros (Ruffner and Carothers 1982), and multiple fish species has followed European settlement of the area. These influences have exacerbated environmental degradation such as overgrazing and the s pread of noxious weeds (Carothers et al. 1976), range reductions in desert bighorn sheep (McKnigh t 1958) and extirpations of four of eight
12 native fish species (Minckley 1991) seven of which are endemic to the Colorado River basin Modern efforts to comba t resource degradation have increasingly incorporated principles of adaptive management. Adaptive management was originally developed in the 1970s and 1980s and has become widely discussed and occasionally instituted across a gamut of natural resource scen arios. The crux of adaptive management is based on reducing uncertainty in inherently complex natural systems in order to improve understanding of system function and resource conditions for management purposes (Walters and Hilborn 1978, Walters and Hollin g 1990). Adaptive management has a mixed history of success and implementation, usually related to opposition in acknowledging uncertainty and the opportunity cost of running ecosystem level experiments where the exact outcome of experimental actions is no t precisely understood. The Glen Canyon Dam Adaptive Management Program (GCD AMP) was instituted through collaboration of various public and private entities involved in resource manage ment within the Colorado River c orridor of Marble, Glen, and Grand Can yons in northern Arizona. Participants include the United States Department of the Interio Bureau of Reclamation ( BOR Monitoring and Research Center (GCMRC), National Park Service (NPS), Arizona Game and Fish Department (AZGFD), Bureau of Indian Affairs (BIA), United States Fish and Wildlife Service (FWS), (WAPA), the seven Colorado River basin states (Colorado, Wyoming, Utah, New Mexico, Nevada, California, and Arizona), five Native American tribes, and recreational
13 interests including boating and fishing enthusiasts. These groups must all reach a consensus on the management directives for Glen Canyon Dam in order to satisfy the Endangered Species Act of 1973 and the Grand Canyon Protection Act of 1992 while still allowing Lake Powell and Glen Canyon Dam to serve as a water supply storage reservoir, flood control structure, and hydropower facility. In t his thesis I will present the results of an ecosystem scale adaptive management experiment implemented by GCD AMP designed to mitigate adverse impacts of river regulation and dam operation on a federally listed (Endangered) native fish species.
14 CHAPTER 2 I NFLUENCING GROWTH RA TES THROUGH EXPERIME NTAL FLOW TREATMENTS Introduction Predation and starvation are two principal causes of mortality that structure animal populations, and in most natural systems neither predation nor starvation can be avoided with out exposure to the other. While they are not always mutually exclusive activities, animals face a critically important and delicate trade off between foraging sufficiently while evading predation (Werner et al 1 983, Carey and Moore 1986, Cowlishaw 1997, Strobbe et al 2011). Fish foraging are na theory helps explain this balance by demonstrating that predation vulnerable juvenile fish are confined to relatively small proportions of available habitat where both refugia and food resources are available, even though this precludes growing at the maximum attainable rate (Werner et al 1983, Walters and Juanez 1993). However, many fish (especially those in temperate environments) must reach a critical mass after their first growing season if they are to survive the winter (Thompson et al 1991, Ludsin and DeVries 1997, Biro et al 2004). In prey sparse environments, these individual s may only succeed by increasing foraging time thus exposing themselves to increased predation risk. Individual fish that survive th is increased foraging time while avoiding the increased predation risk benefit from a positive feedback loop that is advantageous in terms of individual fitness. As an example, f aster growth rates have been shown to g the winter (Ludsin and Devries 1997) as well as increase the food resources available for consumption and reduce the vulnerability to predation of that individual due to gape limitation (Nilsson and Bronmark 2000, Urban 2007). The dual influence of larger fish having reduced overwinter mortality and reduced
15 vulnerability to predators often results in growth rate being used as a surrogate for survival in juvenile fishes (Lorenzen 2006). This theory is often extended to a management context through man agement actions to increase growth rate and ultimately expect improvements in population status of important fish species. The humpback chub Gila cypha is a federally listed (Endangered) cyprinid fish endemic to the Colorado River basin, with the largest e xtant population found in nine aggregations throughout Grand Canyon. More than 90% of the individuals in the Grand Canyon population live near the confluence of the Colorado and Little Colorado Rivers, z and Ryel 1997, for additional references see Goulet and LaGory 2009; Figure 2 1). Most mature individuals participate in a potamodromous spawning migration to the unregulated and seasonally warmed Little Colorado River (LCR) for spawning in the spring, a fter which they return to the mainstem Colorado River (Kaeding and Zimmerman 1983, Valdez and Ryel 1997, Coggins et al 2006a). Humpback chub population trend in Grand Canyon aldez and Ryel 1997, Coggins et a l 2006a), motivating extensive resource assessment and experimental management of Glen Canyon Dam (GCD) to reverse these negative trends in population ( Lovich and Melis 2007 ). Adult humpback chub population demography has been monitored since the late 1980 arily through intensive mark recapture tagging efforts in the Little Colorado River during spring and fall. These monitoring programs provide direct estimates of adult survival and abundance, and reconstructed estimates of juvenile survival (Coggin s et al 2006a, 2006 b). Adult survival rates during the period of population decline (late
16 93%) which suggests that a recruitment bottleneck at earlier life stages may have caused the negative population tre nd. concluded that in river physical conditions such as flows or cold mainstem water temperatures or biological factors such as negative interactions with adult rainbow trout could have significant impacts on juvenile humpback chub. Experimental removal of non native trout was conducted in 2003 2006 and 2009 (Coggins et al. 2011; Yard et al. 2011), and in 2008 a series of discharge experiments began in an attempt to resolve uncertainty between the relationship of river flow and juvenile fish vital rates as part of the Glen Canyon Dam Adaptive Management Program. Extant flow operations from Glen Canyon Dam are designed to follow diel fluctuations in power demand across the southwestern United States, with electricity production increasing and decreasing daily (proportional to dam discharge and flow rate in the downstream Colorado River through Grand Canyon) and causing an artificial odified low components. In contrast, under steady flow conditions, water persists in nearshore areas such as low angle habitat and backwaters and warms on a spatially limited scal e (Korman et al 2006), increasing growth rates for juvenile rainbow trout compared to fluctuating flows (Korman and Campana 2009). Similar growth patterns have not yet been document ed in native fish, although increased temperatures could potentially improve metabolic efficiency and growth of warmwater species such as humpback chub
17 (Coggins and Pine 2010) and would provide important insight into how to operate dams to benefit native f ish. To determine if growth rates of juvenile humpback chub improve during steady flows, GCDAMP initiated a series of constant dam releases for specific time periods (September and October 2008 2012; Figure 2 2) as a contrast to extant fluctuating flows th e remainder of the year. In this paper I will assess the effe ct of steady flows on growth rates of juvenile humpback chub. Methods Study Site The Grand Canyon reach of the Colorado River is the roughly 400 km section bounded downstream by Lake Mead and ups tream by Lake Powell, the first and second largest reservoirs in the United States (Andrews 1991). Average discharge of the Colorado River through Grand Canyon for the past decade (2000 2010) was 351 m 3 /sec as measured at the Phantom Ranch gauging sta tion 145) and 171 km below Glen Canyon Dam. This river reach is contained within the borders of Grand Canyon National Park and is a UNESCO world heritage site, in addition to being listed as a federally protect ed region of cultural, geologic, and biological significance under the Grand Canyon Protection Act of 1992. The Colorado River within Grand Canyon is stenothermic and coo l due to stratification of Lake Powell and hypolimnetic discharges Annual Glen Canyon Dam water release temperatures fluctuate around 2C on average (from 8C to 10C f rom 1994 2002), with occasional temperatures fluctuating annually by as much as 7C (from 8C to 15C in 2011, Figure 2 3). Since the closing of Glen Canyon Dam, t he n ear record maximum temperatures of 2011 still only span about 25% of the original range of annual temperature fluctuation (Figure 2 4). This reduction in seasonal mainstem
18 temperature maximums is believed to be one of the main drivers for the reduction or extirpation of populations of warmwater endemic fishes including humpback chub (Kaeding and Zimmerman 1983, Valdez and Ryel 1997, Clarkson and Childs 2000). The Little Colorado River (LCR) contains the LCR aggregation of humpback chub, and is the largest t ributary of the Colorado River within Grand Canyon National Park. The LCR drains approximately 44,000 km 2 in northern and central Arizona with a mean annual discharge of 11.54 m 3 /sec since 2004 and enters the Colorado River 126 km below Glen Canyon Dam and rkm 100 The LCR is essentially unregulated, although some of the upper basin has been dewatered due to human settlement. Runoff patterns are characteristically bimodal: extended discharges during snowmelt dominated spring flo ods are followed by shorter, stochastic summer flood pulses associated with monsoonal precipitation (Figure 2 5). I considered the LCR as a control system for humpback chub growth studies because it is not affected by the dam operations beyond the confluen ce with the Colorado River, but does experience natural seasonal fluctuations in temperature and discharge As th e only unregulated river in Grand Canyon that still has a sizeable population of humpback chub, it can uniquely offer some distinction between growth effects that may be occurring naturally due to seasonal fluctuations and those that may be occurring due to steady versus fluctuating flow effects. Sampling Techniques I conducted a mark recaptu re study of humpback chub in the mainstem Colorado Rive r from river km 102 to 106 and in the Little Colorado River from the confluence upstream roughly 14 km. I sampled the LCR both independently and as part of the fall US Fish and Wildlife Service mark recapture humpback chub surveys. I sampled 10 to
19 12 days each month of the field season in both rivers (12 trips total, Figures 2 5 and 2 6) during July October 2009 2011. July and August mainstem Colorado River samples during 2009 and 2010 represented typical fluctuating hydropower flows, while September and Oc tober samples from all years represented the fall steady flow experiments. High steady dam discharges during July and August 2011 were exceptional for my study period due to equalization flows between lakes Powell and Mead during a high water year (Figur es 2 5 and 2 6), and provided a summer steady flow contrast compared to fluctuating flows in July and August of the previous two years. I used two gear types to sample the mainstem Colorado River fish community (Figure 2 7) : un baited mini hoop nets (50 cm d iameter, 100 cm long, single 10 cm throat, made of 6 mm nylon mesh, fished for 12 consecutive days over 24 hour intervals) and slow speed boat electrofishing ( see Korman and Campana 2009; unit output was pulsed DC current at 15 20 amps and 200 300 volts, 7 10 seconds per meter of shoreline, repeated 24 to 72 hours apart f or 3 to 5 total passes per trip ). I sampled the Little Colorado River with the same un baited mini hoop nets, deployed in areas likely to yield the highest capture and recapture rates of juvenile humpback chub Electrofishing is not feasible in the LCR due to natural high salinity. I measured, tagged and returned humpback chub to the same location where they were captured. All humpback ch ub >100 mm in total length (TL) received a 134.2 kHz passive integrated transponder (PIT) tag (9 mm long, BIOMARK) with a unique number identifiable to individual fish. Although I sampled the entire fish community as part of the larger sampling project, th is analysis in cludes only humpback chub large
20 ca trip to limit the at large growth period to a certain discharge regime and omit the influence of inter annual, seasonal, or ontogenetic shifts in growth rates. Due to logistical and safety reasons during flash flooding, sampling was not always feasible in the Little Colorado River during s ummer monsoon season (July September) To improve recapture sample sizes for Little Colorado River fish initially captured in July, I included recaptures over the interval of July to August or September in the LCR, whereas mainstem recaptures were limited to fish captured in July and recaptured in August only. Because LCR discharge shifts are more gradual and continuous (fish recaptured in September in the LCR were experiencing similar conditions as August, where the mainstem conditions had markedly changed due to the flow experiment, Figure 2 2), these fish could be considered part of the same treatment type. All growth rates are assumed to be linear for the 15 70 day intervals when fish were at large. Data Analysis I divided recaptured humpback chub into t welve groups based on which season (summer and fall) year (2009, 2010, and 2011) and system (mainstem Colorado River or Little Colorado River) in which they resided (Table 2 1). For instance, if a humpback chub over 100 but less than 200 mm TL was captur ed, measured and released in July 2009 and then recaptured in the same place in August 2009, it became part of the respective river. If any individual fish was not recaptured on the subsequent trip, it was not included in t hat sample. Again, the exception for this is in the Little Colorado River, where fish were allowed to be at large for up to two months (from July to August or Septemb er)
21 I calculated the mean daily growth for each of the 12 humpback chub subsamples (Table 2 1) over the interval during which they were at large, and then performed a 10,000 iteration bootstrap resample with replacement using Program R to determine the distribution of the mean growth rate for that subsample. The assumptions of this approach w ere 1) fish remain within the same river system for the duration of time between capture and recapture, and 2) captured individuals represent a random sample from the overall population. Results Juvenile humpback chub daily growth rates in the mainstem Col orado River during summer fluctuating flows (July to August periods in 2009 and 2010 Figure 2 8 ) were higher than the daily growth rates during fall steady flow experiments in the same year (September to Oc t ober periods in 2009 and 2010, F igure 2 9) despi te similar temperatures across the sampling season (Figure 2 3 ). I expected growth rates to decline in the fall due to decreased solar insolation (Yard et al. 2005) and possibly reduced primary production which would seasonally confound the fall steady fl ow treatments. However, the high water year of 2011 (Figure 2 6) provided an unplanned experiment with a serendipitous contrast. In 2011, spring a nd summer discharges were high and steady compared to 2009 and 2010 and due to required equalization flows bet ween Lakes Powell and Mead. This created a reciprocal growth test of steady flows during the July August summertime periods (in contrast to fluctuating flows the previous two years over the same time period). Although there is some overlap of the bootstrap ped distribution of the sample means, the resulting distributions are disjunct for summer 2011 (steady flow interval) compared to summer 2009 and 2010 (fluctuating flows Figure 2 8 ). This contrast allowed us to determine that individual humpback chub
22 grew more slowly during the steady flow interval (2011) than during fluctuating flow intervals (2009 and 2010), even when they occ upied the same season Flow conditions in the Little Colorado River were not related to dam operation (Figure 2 5), and as I exp ected there were no trends in growth rates for the Little Colorado River related to flow in the mainstem Colorado River (Figures 2 10 and 2 11). I did observe a seasonal decline in mean daily growth rate in 2009 and 2011 from rates of roughly 0.12 mm/day f rom July to August/September down to roughly 0.01 mm/day for the interval between September and October. Of the three years assessed, 2010 growth of juvenile humpback chub in the LCR was relatively high (0.25 mm/day) and did not decline in September and Oc tob er as it had in previous years. Discussion Common policies for improving population status of fishes such as stocking, bag and slot limits, or removal of competing u ndesirable species are difficu lt and costly to use in remote areas or in areas where harvest is precluded for legal or logistical reasons. Ecosystem level controls offer an alternative to labor intensive and expensive hands on techniques for managing fishes and fisheries, especially in regul ated rivers. The current a d aptive m anagement program at Glen Canyon Dam provides unique opportunities to probe the efficacy of achieving ecosystem level improvements in fish vital rates such as growth through flow manipulation. Growth rate is an important indicator of habitat quality and potential recruit m ent and/or population success in fishes because it incorporates a broad range of habitat characteristics such as forage quality, availability of refugia, and density and competition between fish population s within a certain trophic level (Nilsson and Bronmark 2000, Urban 2007). Managing for growth is widely used in aquaculture and
23 food fish resource management to maximize production, and is also prevalent in management scenarios for listed fish species as i t serves as a useful surrogat e for survival (Lorenzen 2006). This study demonstrates that juvenile humpback chub daily growth rates are lower during steady flows than fluctuating flows when both flows occurred in the same season (0.07 mm/day during steady flows from July to August 2011 versus 0.12 mm/day and 0.15 mm/day during fluctuating flows from July August 2009 and 2010, respectively, Figure 2 8). Growth rates are also lower during steady flows than fluctuating flows within the same year (0.08 mm/day and 0.07 mm/day during steady flows versus 0.15 mm/day and 0.12 mm/day during fluctuating flows in 2009 and 2010, respectively; Figures 2 8 and 2 9), even though temperatures were similar across the sampling periods (Figure 2 3). There is some overlap of t he bootstrapped resample of mean daily growth rates, but the distributions were essentially disjunct and demonstrated that juvenile humpback chub grew faster when discharge fluctuated mildly due to power production than when it was held experimentally cons tant. In order to be included in my subsamples, fish had to remain in the same river system between captures (assumption #1). If individuals mixed frequently between captures and thus spent unpredictable amounts of time in both the mainstem Colorado River and the LCR while they were at large, the subsamples that I used to determine the flow growth relationship between river systems would have been flawed. In this study, the average distance of the mainstem Colorado River sampling area from the Little Colora do River is ~3 km, while the average distance from my sample location within the Little Colorado River to the mainstem was over 1 km. Humpback chub,
24 especially adults, are known to travel farther than this on an annual basis for potamodromous spring spawni ng runs (Kaeding and Zimmerman 1983, Gorman and Stone 1999). However, the relatively short time at large (15 to 70 days) of sampled fish, smaller body size and movement potential of juvenile fish, and the fact that they were recaptured in the same river wh ere they were originally handled on the preceding trip supports the assumption that they did not spend significant time in the other river system. Individuals that did move (and experienced mixed environments and growth rates) between the two systems were not likely recaptured within the short time interval and in the same river that would have caused me to include them in that respective growth subsample. The second sampling assumption is that I had a random and unbiased sample of the population. The data would be biased upward an unknown amount if slower growing individuals died or emigrated at higher rates and were excluded from analyses. However, unless the slower growing individuals were disproportionately missing in one of the two river systems or duri ng one of the flow treatments, I have no reason to assume that the comparative results of the subsamples are invalid. Capture probabilities did not differ by flow treatment, so I have no reason to suspect this to be true. In addition, many of the fish excl uded from growth analyses because they were not recaptured on the next trip were recaptured on later trips, implying that subsamples were not decided by size or mortality dependent attributes but were subject to random selection. Water temperature is an important determinate of growth rate in fish that has potential to confound flow experiments (Valdez and Ryel 1997, Korman and Campana
25 2009). Unfortunately temperature is not often within the controls of dam managers and is instead subject to stochastic fluctuation due to environmental and meteorological conditions that drive riverine inputs to reservoirs and reservoir storage levels During the three year period of this study, 2009 and 2010 had similar temperature profiles. In 2011 high runoff into Lake Powell and associated short residence time caused near record discharge temperatures in the Colorado River within Grand Canyon since the closing of Glen Canyon Dam (Figure 2 3). However, the Colorado River historically fluctuated fr om near freezing in winter to almost 30C in summer (Figure 2 4), so the difference only around 10% of the historical thermal range of humpback chub. I thus expected only mild improvements in growth rate for the 2011 July August interval. I actually observed that growth declined compared to same time period in the previous two years, which were fluctuating flow conditions. Due to the heat storage effect of Lake Powell, ma ximum water temperatures do not occur in the Colorado River below Glen Canyon Dam until September or October (Figure 2 4). This has important implications for within year flow tests. Based on temperature, I would expect the periods of highest growth to occ ur during the fall (steady flow) time period, but growth was actually lower during fall steady flows. This further demonstrates that steady flows reduce growth in juvenile humpback chub, even with the added influence of higher water temperatures during ste ady flows periods both within years and across years. Without observing the expected relationship between temperature and growth rate, I suggest food availability in the invertebrate drift as a possible mechanism for the
26 demonstrated flow induced differenc es in mean daily growth rate in juvenile humpback chub. Invertebrate drift in regulated rivers increased during fluctuating flows in the Flathead River of Montana (Perry and Perry 1986) and the Hawea River of New Zealand (Irvine and Henriques 1984). These increases in invertebrate drift are likely present in Grand Canyon as well, and may reduce the metabolic costs of feeding for Colorado River fish as well as allow them to stay closer to predation refuges. During steady flows, insect drift likely returns to a more natural cycle, with pulses in drift density occurring nocturnally to minimize predation and maximize successful dispersal of insects (Waters 1972, Muller 1974), resulting in reduced availability to foraging fish. I also suggest the relationship bet ween discharge and invertebrate colonization and drift as a mechanism to explain the unusual growth conditions I observed in 2011. The July August 2011 interval demonstrated the lowest growth rate for that season of the three years, and the September Octob er period demonstrated the highest growth rate for that season of the three years. This cannot be attributed to steady versus fluctuating flow conditions or temperature, as the entire 2011 sampling period occurred during steady flows (Figure 2 6) and was o f fairly uniform temperature (Figure 2 3). However, it was likely discharge related. High, steady discharges that began in spring 2011 permanently flooded unoccupied substrate s that were either dry or intermittently submerged due to hydropower fluctuations During the lag time before algae, diatoms, and insects colonized this newly submerged substrate, the same amount of emergent food items were effectually diluted in larger volumes of water and reduced foraging efficiency of fish. Once colonization of this new ly submerged substrate occurred drift density likely improved and increased food availability and growth. Then, with
27 reductions in discharge on September 1 st this newly established foodbase may have drifted catastrophically to avoid desiccation and was likely concentrated in the lower volume of water, increasing food availability to fish. This could serve as the possible mechanism that produced the high growth rates I observed during September October 2011. A similar situation was observed i n the re gulated Kootenai River of Montana, where the highest rates of insect drift were also recorded during a lower steady flow immediately following a period of high steady discharges (Perry and Perry 1986). Although these results do not categorically preclude a system wide improvement in growth in the Colorado River during fall 2011 t he absence of higher growth in the LCR duri ng this period suggests that in situ ma instem Colorado River conditions were likely responsible for the growth differences, and my data s uggests that these changes are likely related to discharge Fluctuating hydropower flows can actually be beneficial for juvenile fish growth. Changes in discharge can increase insect drift (Irvine and Henriques 1984, Perry and Perry 1986), as well as the c oncentration of drifting insects, and I suggest this as a mechanism for higher growth during fluctuating flows in the Colorado River. If individual fish are exposed to increased metabolic costs and reduced foraging efficiency d uring experimental steady flo ws their growth rates are reduced, with potential implications for future survival, recruitment, and population growth or stability. The re is likely an ideal fluctuation level to optimize juvenile fish growth rate that is as yet unknown.
28 Table 2 1. Samp le sizes of juvenile humpback chub recaptured for growth purposes. Location: Colorado River Little Colorado River Interval: Jul Aug Sep Oct Jul Aug Sep Sep Oct 2009 28 9 4 55 2010 23 82 20 52 2011 38 61 8 20 Total : 89 152 32 127
29 Figure 2 1 Map of study area near confluence of Colorado and Little Colorado Rivers in northern Arizona, containing the Little Colorado River aggregation of humpback chub.
30 Figure 2 2. Colorado River discharge August 22nd September 10th, 2009 at the L Ferry ga uge, 98 km upstream from my sampling universe. The period before September 1st is representative of normal hydropower operations, while the period after September 1st is representative of the steady flow experiment. Data from USGS instantaneou s data archive. Available: http://nwis.waterdata.usgs.gov/nwis
31 Figure 2 downstream from Glen Canyon Dam. The solid gray line is the mean annual temperature as estimated by the period from 1994 2002 while the dotted gray line is the mean annual t emperature as estimated by the period from 2003 2008. The solid black line reflects temperatures observed durin g each of the three years of this study ( July 2009 July 2012; not an average). Data from USGS instantaneous data archive. Available: http://nwis.waterdata.usgs.gov/nwis
32 Figure 2 4. Colorado River w ater temperatures from 1994 2002, 2003 2008, and during this study (July 2009 July 2012; solid gray, dotted gray, and black lines r espectively ), with a model of pre Glen Canyon Dam historical temperatures (dashed gray line). Gray lines are mean temperatures while black line represents actual annual temperatures Data from USGS instantaneous data archive. Available: http://nwis.waterdata.usgs.gov/nwis
33 Figure 2 5. Daily discharge in the Little Colorado River near the confluence with the Colorado River over the three year period of this study (2009 2012 ). Sampling periods are represented by the diagonally striped boxes. Data from USGS discharge gauge 09402300. Available: http://nwis.waterdata.usgs.gov/nwis
34 Figure 2 6. Sampling calendar. Steady flow experiments occurred in September and October in the years 2009, 2010, and 2012 (not studied), as well as April November 2011. Discharge data available at http://waterdata.usgs.gov/az/nwis/rt
35 Figure 2 7. Detail m ap of study area. kilometers. All available shoreline (excludes rapids) were electrofished in all trips. Hoop effort increased through the study: Site 1 contained 47 hoops on trip 1, 60 hoops on trips 2 12, and 45 hoops on trips 13 and 14, deployed across habitat types according to their availability. Site 2 contained no hoops on trips 1 6, 20 hoops on the upper half on trips 7 12, and 45 hoops on trips 13 and 14, also deployed across habitat types according to their avai lability.
36 Figure 2 8. Distributions of the mean daily growth rates for juvenile humpback chub (100 200 mm total length ) in the mainstem Colorado River between July and August; 10,000 iteration bootstrap resamples with replacement
37 Figure 2 9. Distri butions of the mean daily growth rates for juvenile humpback chub (100 200 mm total length ) in the mainstem Colorado River between September and October; 10,000 iteration bootstrap resamples with replacement
38 Figure 2 10. Distributions of the mean dai ly growth rates for juvenile humpback chub (100 200 mm total length ) in the Little Colorado River between July and August or September; 10,000 iteration bootstrap resamples with replacement
39 Figure 2 11 Distributions of the mean daily growth rates for juvenile humpback chub (100 200 mm total length ) in the Little Colorado River between September and October; 10,000 iteration bootstrap resamples with replacement
40 CHAPTER 3 ARTIFICIAL DISCHARGE FLUCTUATIONS AND JUV ENILE FISH SURVIVAL RATES Introduction dams (WCD 2000) These structures modify flow in riverine ecosystems by storing water in upstream reservoirs and regulating water releases downstream that can facilitate shipping, power production, irrigation, or provide protection to communities by attenuating floods (WCD 2000, Poff et al. 2006, Richter and Thomas 2007). The dams and the ecosystems they create (reservoirs) and modify (rivers) vary hugely in si ze from small run of the river reservoirs to immense, canyon bound reservoirs that can store several years of mean annual flow (Christensen et al. 2004, Anderson and Woosely 2005, Blinn and Poff 2005, Nilsson et al. 2005). While regulated rivers are found across the gamut of world climates, temperatures, and elevations, ecosystem changes that occur following dam construction and river modification are similar and widespread in terms of physical and ecological trade offs compared to unregulated rivers. These changes include sequestration of woody debris and sediment in upstream reservoirs (Stanford and Ward 1991, Kearsley et al. 1994, Schmidt et al. 1998), altered trophic dynamics (Stevens et al. 1997, Kennedy and Gloss 2005), novel fish community composition including introduction of non native species (Mueller and Marsh 2002, Gloss and Coggins 2005, Coggins et al 2011), modified thermal regime (Stanford and Ward 1991, Clarkson and Childs 2000), and timing, magnitude, and frequency of floods with associated connectivity to floodplain habitat in rivers below dams (Poff et al. 1997, Lovich and Melis 2007, Dutterer et al. 2012).
41 Optimizing dam operations to minimize ecological and physical impacts to riverine environments while simultaneously meeting anthropogen ic needs from the river ecosystem is often a common goal of water managers (Anderson and Woosely 2005, Richter and Thomas 2007), although the balance can be difficult to achieve in heavily utilized river systems. For instance, the Colorado River in the wes tern Unite d States is i dentified as the most legislated, most debated, and most litigated river in the entire world. It has more people, more industry, and a more significant economy dependent This riverine ecosystem, including the UNESCO recognized reach in Grand Canyon National Park, is highly altered due to flow modifications from numerous dams and water diversion projects. In an attempt to mitigate detrimental effects of river regulation in the Grand Canyon reach of the Colorado River, the Glen Canyon Dam Adaptive Management Program (GCD AMP) was developed and instituted based on principles of adaptive management (Walters and Hilborn 1978, Walters and Holling 1990) with a commission to mo nitor, preserve, and restore the downstream ecosystem in Grand Canyon. Because of the multi year water storage capacity in Lake Powell and the regulatory structure of the GCD AMP, a unique opportunity exists in the Grand Canyon reach of the Colorado River to experimentally assess various flow policies and identify those which simultaneously benefit the riverine ecosystem while meeting power and water needs of human users. As an example, if flow policies could be experimentally identified that create mainst em Colorado River conditions conducive to improving the population status of the imperiled native fish community without precluding hydropower production, water supply, or recreation, this information would be invaluable to
42 managers of Glen Canyon Dam. The Endangered Species Act, the Grand Canyon all considered under the auspices of the GCD AMP, and optimizing dam discharges to satisfy legal and social requirements is th e principal goal of these unique ecosystem level experimental flow policies. One specific focus of GCD AMP is popul ation recovery of humpback chub Gila cypha a large bodied, morphologically distinct minnow endemic to the Colorado River basin (for addition al references see Goulet and LaGory 2009). Humpback chub are currently federally listed (Endangered) under the Endangered Species Act. The exact reason for declines in humpback chub populations in Grand Canyon (and elsewhere in the Colorado River) are unkn own but generally include: (1) negative interactions with non native fish, (2) loss of essential habitats due to flow, temperature, and sediment input modifications, and (3) non native parasites (Minckley 1991, Valdez and Ryel 1997). Most of the remaining humpback chub live near the confluence of the Colorado River and the Little Colorado River, the largest tributary within Grand Canyon (Figure 2 1). Adult humpback chub migrate from the Colorado River to the Little Colorado River (LCR) in spring to spawn S uccessful reproduction of this warm water species can then occur because t he thermal regime of the LCR resembles the pre dam Colorado River (Kaeding and Zimmerman 1983, Douglas and Mars h 1996, Gorman and Stone 1999). Young humpback chub then rear in the Li ttle Colorado River or disperse into the mainstem Colorado River, where they encounter daily fluctuations in discharge, seasonally colder temperatures due to hypolimnetic dam releases, and higher numbers
43 of nonnative fish than in the Little Colorado River (Kaeding and Zimmerman 1983, Valdez and Ryel 1997, Paukert et al. 2006). Although much is known about life history and vital rates of adult humpback chub, there is a paucity of data available on young of year and juvenile humpback chub in the Colorado Rive r system. Resolving this data gap is critical to humpback chub recovery because recruitment to adulthood is likely impeded by low survival of juveniles. To date, most inferences on juvenile humpback chub abundance and survival are indirect, and come from e stimating the number of juveniles alive in a given year from the number of age 4+ adults found in future years. For example, the number of recruits in 2005 would be estimated by first determining the number of age 4 fish in 2009 and then back calculating how many recruits had to be present in 2005 (with assumptions about survival rates in each year) to have produced that number of a ge 4 adults (Coggins et al. 2006b ). A key shortcoming in this approach is the long time lag between management actions and in formation on juvenile fish response, and the possibility that the response (positive or negative) in the number of fish occurred at any time during the four year lag period. This type of reconstruction was necessary because contemporary sampling technique s and monitoring activities did not capture or mark sufficient numbers of juvenile humpback chub to draw inference on their abundance or survival rates directly. Despite limited data on young humpback chub, many current management actions (e.g., mechanical removal of trout, experimental discharges, artificial floods to build sandbar and backwater habitat) are designed to benefit the juvenile life stage because juvenile fish are generally more sensitive than adult fish to predation (Ward
44 and Bonar 2003), col d water temperatures and thermal shock (Clarkson and Childs 2000, Robinson and Childs 2001, Ward and Bonar 2003), and flow variation (Scheidegger and Bain 1995). Predator removal is still inconclusive (Coggins et al. 2011), and a temperature control device has not been constructed, so current attempts to improve vital rates of juvenile humpback chub has culminated in an ecosystem scale discharge experiment in Grand Canyon. Hydropower is produced at Glen Canyon Dam to match diel fluctuations in power demand across the southwest United States by increasing and decreasing discharges through the turbines as power demand goes up and down (Hughes 1991, Figure 2 2). These regular, rapid changes in discharge resemble ocean tides (Fisher and LaVoy 1972), and are unna tural in the Colorado River (notable exceptions being snowmelt dominated headwater streams which rise and fall with daily temperatures, and the delta at the Gulf of California, Mexico). Fish are known to exhibit lateral movements during fluctuating dischar ges (Bunt et al. 1999) that can increase displacement and energetic costs of larval and juvenile fish (Scheidegger and Bain 1995, Korman and Campana 2009) or expose them to predation due to forced movement from refuge habitat (Werner et al. 1983, Bain et a l. 1988, Walters and Juanez 1993). In contrast, steady flows can produce localized warming of water in low angle habitats and improve growth of juvenile fish (Korman et al. 2006), which is cl osely tied to survival (Rice et al. 1993, Houde 1997, Lorenzen 2 006). To determine if discharge fluctuations affect survival of juvenile humpback chub, GCD AMP implemented a series of steady flow experiments during September and October 2008 2012. In this analysis I evaluate the effects of these annual steady flows ove r a three year period (2009 2012, Figure 2 6) to determine
45 what, if any, effects flow has on apparent survival rates of juvenile humpback chub. This funded to study the habit at use, abundance, and vital rates of juvenile humpback chub within the LCR aggregation. Methods Study Site The Grand Canyon reach of the Colorado River is the roughly 400 km river section bounded downstream by Lake Mead and upstream by Lake Powell, the f irst and second largest reservoirs in the United States, respectively (Andrews 1991). This river reach is contained within the borders of Grand Canyon National Park, is designated a UNESCO world heritage site, and is recognized as a US and international re gion of cultural, geologic, and biological significance under the Grand Canyon Protection Act of 1992. Humpback chub are distributed throughout the Grand Canyon reach of the Colorado River in eight distinct aggregations. The LCR aggregation occupies the lo wer 15 km of the LCR and the associated inflow reach of the Colorado River and contains roughly 95% of the adult humpback chub still found in the lower Colorado River basin (Douglas and Marsh 1996, Valdez and Ryel 1997, Coggins et al. 2006 a ). This section of to as river kilometers (rkm). I sampled a portion of the LCR aggregation of humpback chub from rkm 102.1 to 104.7, below the confluence of the Colorado and Little Co lorado Rivers, which I will refer to as the NSE reach. Although I did sample through to rkm 106.1 as part of the overall NSE study, I did not sample the lower third with all gears or in all years, so it was excluded from this analysis.
46 Sampling T echniques I subdivided the portion from rkm 102.1 to 104.7 into two sites approximat ely 1000 m in length (Figure 2 7 ), which were further subdivided into 25 m habitat sub units to provide a spatially referenced location for all sampling efforts. I sampled both site s in July, August, September, and October for 12 days for 2009 2011. The July and August samples from 2009 2010 represented typical fluctuating hydropower flows, while September and October samples from 2009 2011 were conducted during the fall steady flow experiments (Figure s 2 2 and 2 6). Previously unplanned high steady dam discharges occurred during July and August 2011 as a result of required equalization flows between lakes Powell and Mead (Figure 3 1, see discussion). During 2012 I sampled for nine da ys in both April and July during typical fluctuating flows. I used two gear types to sample the fish community, un baited mini hoop nets (50 cm diameter, 100 cm long, single 10 cm throat, made of 6 mm nylon mesh, fished for 12 consecutive days over 24 hour intervals) and boat electrofishing (pulsed DC current at 15 20 amps, 200 300 volts, 7 10 seconds per meter of shoreline, repeated 24 to 72 hours apart for three passes per trip). Pilot sampling in July 2009 included 48 hoops in the upstream site which exp anded to 60 hoops in Aug Oct. Hoopnet efforts further expanded to include 80 hoops (60 in the upper site, rkm 102.1 to 103.5, and 20 in the lower site, rkm 103.6 to 104.7) in 2010 and 2011. Hoopnet efforts in 2012 included 90 hoops, and were spread more ev enly across both sites, with 45 hoops in each (Figure 2 7). All available shoreline areas in both sites were electrofished during all sampling trips following guidelines in Korman and Campana (2009), which targets small fish with emphasis on maintaining sl ow boat speeds of 7 10 sec/m of shoreline.
47 Following capture, I measured, tagged, and released humpback chub in the same location where they were captured. Individual humpback chub between 40 and 100 mm total length (TL) were given batch marks using visibl e implant elastomer (VIE, Northwest Marine Technologies). These elastomer batch marks were identifiable to trip number, gear, and site. On subsequent recapture, if the humpback chub was still <100 mm TL, it was given an additional VIE batch tag coded accor ding to that respective trip number, gear, and site. Any h umpback chub >100 mm TL received a 134.2 kHz passive integrated transponder (PIT) tag (9 mm long, BIOMARK) with a unique number identifiable to individual fish rather than an additional VIE tag. Thi s analysis only includes humpback chub that received VIE tags (40 100 mm TL) when they were first captured (Figure 3 2). When humpback chub were next captured above 100 mm TL and received a PIT tag they were considered removed from the study population. D ata Analysis In mark recapture experiments it is possible to have confound ed apparent survival rate and capture probability e s timates due to the inability to distinguish between fish that died (low survival) and fish that lived, but were not observed (high survival, low capture probability). Because I was concerned that capture probabilities may be low, thus increasing the risk of confounded apparent survival estimates, I used simulated data to assess the accuracy of the Cormack Jolly breton et al. 1992) for later use in my analyses. In this way I could determine whether accuracy of the model I planned to use from data with known survival and capture probability rates I used GENCAPH1 ( http://www.mbr pwrc.usgs.gov/software/gencaph1.shtml ) to generate mark recapture datasets from virtual populations with known survival rates and capture probabilities and with similar sample sizes to my real data, and then
48 exported this data to Program MARK for analysis. I evaluated the efficacy of MARK at returning precise and accurate estimates under the various simulation scenarios (designed to mimic possible effects of flow treatments) and used these simulations to determine the accu racy of the apparent survival estimates that I would later derive from field data using MARK. I also performed a power assessment of the simulations to predict my ability to detect changes in the real data should they occur. After I generated simulated data, I fit the recaptures only (CJS) model based on field data using program MARK ( http://warnercnr.colostate.edu/~gwhite/mark/mark.htm ). This model estimates two parameters: (1) capture probability on a given trip and (2) apparent survival for the interval between trips. Input data consisted of binary coded capture histories that were equal in length to the number of sampling events ( n= 14). For example, a fish that was captured on the second trip and then not seen until the tenth trip recaptures were ignored in this analysis, both to avoid double counting VIE marked fish (that could not be distinguished from each other as a result of batch marking) and to preserve the requisite binary nature of inputs for the recaptures only MARK model (Table 3 1) Th e basic assumptions of the CJS model are (1) all individuals in the population (both marked and unmarked) have the same probability of capture, (2) all marked individuals have the same probability of dying or permanently emigrating, (3) tags do not increase the probability of death, and (4) tags are not lost or overlooked. I develope d a suite of a priori models to test hypotheses about the probable effects of the annual steady flow experiments. These models were based on biologically relevant hypotheses about humpback chub behavior coupled with discharge
49 information, and were construc ted to consider apparent survival and capture probability as they relate to flow, a modified version of flow to include transition periods (see Model 4, below), year, the interaction of flow and year, time dependent, or as constant rates across the study p eriod, for a total of 36 possible models (Table 3 2 ). In order to derive annual apparent survival rates and because each sampling event spanned 12 days, I calibrated sampling intervals in MARK as fractions of a year. These intervals spanned from the averag e date of capture for humpback chub in a given trip to the average date of capture for humpback chub on the next trip. Models were selected by reviewing the most parsimonious and biologically reasonable model that also had AIC c support. c scores less than four units from the top model were considered strongly supported with limited ability to separate inference (Burnham and Anderson c scores between four and six units from the top model were considered moderately supp c greater than six were considered poorly supported. Here I will further discuss the top five performing models according to these model selection criteria. The annual model (Model 1) considers separate parameters for apparent survival July 2010, July 2010 July 2011, and July 2011 July 2012. Capture probability is estimated by time (uniquely estimated for each trip). This model has three parameters for apparent survival and 13 parameters for capture probability. The flow model (Model 2) has two parameters for survival; one for each flow treatment type (steady or fluctuating flow, Figure 3 1). This model considers September October of 2009 and 2010 and July October 2011 as steady flows, wi th all other time periods being fluctuating flows. In this model capture probabilities were also assumed to
50 be time dependent (uniquely estimated for each trip). This model has two parameters for survival and 13 for capture probability. The constant apparent survival model (Model 3) was the simplest of the five preferred models and assumes apparent survival is constant across all intervals and that capture probability varies by time. Statistical support for this model would suggest that flow or flow*y ear associated terms did not improve model fit, but that flow independent or consistent survival pressures were ubiquitous. The total number of parameters for this model was 14, one for survival and thirteen for time specific capture probability. The inter mediate flow model (Model 4) adds a third survival parameter to the flow model (Model 2) representing an intermediate flow treatment. This third parameter was for periods between our sampling trips when flows transitioned from steady to fluctuating (Septem ber October of 2009 and 2010, as well as October 2010 July 2011, Figure 3 1 ). If any of the different flows have a survival effect, then apparent survival rates during transition time intervals would likely be intermediate because flows were roughly hal f fluctuating and half steady during these intervals. However, since we did not sample across these transition periods, we cannot separate the effects of steady versus fluctuating flows from each other, but instead must model a third rate that estimates ap parent survival during the transition period In this model capture probabilities were also assumed to be time dependent. This model has three parameters for apparent survival and 13 for capture probability, making 16 total parameters.
51 With a flow year mod el (Model 5) I attempted to separate effects of flow and years by starting with Model 2 and then adding a third apparent survival parameter that considers 2011 separate from 2009 and 2010 estimates. This is because river flows during 2011 were substantial ly different than 2009 and 2010 with relatively high, but steady flows during July and August, and lower, but still steady flows in September and October. In this model capture probabilities were assumed to be time dependent. There were three parameters fo r apparent survival and 13 for capture probability, making 16 total parameters. Results Simulation R esults My simulation results demonstrate that the CJS model accurately estimated apparent survival and capture probability across the range of simulated data sets I fit to this model. A key result of these simulations is that the model can generally only fit the data apparent survival estimates are made based on shared data across annual, seasonal, or flow treatment designated repeated samples. Across all cases, models where appare nt survival is estimated independently for each time interval did not converge. In the first simulation I held capture probability constant and varied survival, and vice versa. I manipulated annual survival from 80 98%, and capture probability from 5 3 0% using both upward and downward trends (the Phi(.) p(t) and the Phi(t) p(.) models in MARK, Figure 3 3 ). This simulation demonstrates that MARK is able to derive accurate estimates of apparent survival even if capture probability fluctuates widely (5% to 3 0 % capture probability), but only when apparent survival is constant. When apparent survival is estimated independently for each sampling event (no annual replicates), the
52 potential confounding effects of capture probability and apparent survival become evi dent and the model estimates do not converge or rapidly lose statistical precision (no confidence intervals or wide confidence int ervals, respectively Figure 3 3 ). In the second simulation, I increased survival and capture probability together, and decrea sed them together. I then attempted to estimate unique apparent survival and capture probability estimates for each time step (a Phi(t) p(t) model, Figure 3 4 ). This model performed poorly by failing to accurately estimate apparent survival and capture pro bability when they both varied in the same way and when they were both estimated at every time step. This simulation model does not allow replicates (i.e. the same flow type, but shared across multiple years from real data), but rather estimates unique val ues for all parameters during all sampling periods and over par ameterizes the model The failure of this model reinforces the need to repeat experimental treatments if recapture data are sparse (Table 3 1 ). In the third simulation I mimicked flow dependent increases or decreases in apparent survival, with upward and downward trends in capture probability ( Phi(flow) p(t) model, Figure 3 5 ). This simulation demonstrates that no matter the direction of change in survival rates due to flow conditions, MARK is a ble to accurately distinguish between apparent survival and capture probability because treatments were grouped across years as annual replicates This simulation also demonstrates the MARK can detect changes when capture probability changes by 5% or more, as well as changes in annual apparent survival rates as small as 10% if those changes occur after a suitabl e marked population is present.
53 For the last simulation I modeled survival as being either constant or flow dependent, and capture probability as be ing flow dependent (Phi(.) p(flow) and Phi( flow) p(flow) models, Figure 3 6 ). This last model demonstrates that Program MARK still produces accurate estimates of apparent survival when survival and capture probability are proportionally confounded by flow treatment, as long as the experimental rep licates are allowed However, the simulation demonstrates that if capture probability varies in the same direction and proportion as apparent survival rate (i.e. as a result of flow treatments), that apparent survi val may need to vary by as much as 30% in order for the model to actually detect the change. Flow effects and Juvenile H umpback C hub Apparent S urvival After fitting the 36 possible models for apparent survival and capture probability described above to fi eld data, the three best fitting models were differentiated by less c points, making them quantitatively inseparable based on my model c points, indicating moderate quan t itative support. The remaining c points from the top model ( 22 models) or did not converge upon estimates and were unreliable (9 models, Table 3 2) All five top models estimated capture probability by time (Figure 3 7) This result is not surprising given the vari ation in discharge across all 14 trips (Figure 3 3). Importantly, estimates of capture probability are specific to Nearshore Ecology trips, as opposed to gear and pass specific estimates. This is because I considered a unit of effort to be our combined electrofishing and hoopnet sampling, and because I excluded multiple within trip recaptures of the same individuals due to the deficiencies of batch
54 marking methods. Gear specific capture probability estimates are preferred, b ut recapture data is currently too sparse. As a result these capture probabilities are not directly transmissible to other sampling scenarios unless they incorporate similar arrays of gears and effort. The model estimating unique apparent survival rates ac (with time dependent capture probability) had the most quantitative support (Model 1). This model indicates that the 95% confidence intervals for apparent survival from July 2009 July 2010 and July 2010 July 2011 were 58 71% and 57 7 5%, respectively (Figure 3 8) The fish year from July 2011 July 2012 had lower survival, with 95% confidence intervals of 28 54%. Capture probability estimates varied from 9 25% per NSE trip (multiple passes of both electrofishing and hoopnets Figure 3 7 ). The model estimating two apparent survival rates, one for fluctuating flows and c score of 2.7. Modeled 95% confidence intervals for apparent survival during fluctuating flows and steady flows were 59 78% and 16 65%, respectively (Figure 3 9) Capture probability estimates varied according to sample interval as previously discussed and ranged from 9 26% (Figure 3 7) The next highest performing model estimated a constant apparent survival rate across the entire s ampling period (Model 3), with capture probability estimated by time. Apparent survival 95% confidence intervals ranged from 59 67% (Figure 3 10) with capture probability estimates from 9 25% (Figure 3 7) Quantitative support for this c points behind the top model. The fourth of the five top performing models (Model 4) estimated three apparent survival rates; one for fluctuating flows, one for steady flows, and one for the intervals
55 between sampling trips where flows were neither fluctua ting nor steady, but both. The 95% confidence intervals of those estimates were 57 77%, 13 61%, and 54 88%, respectively. Capture probability estimates ranged from 9 26%. The last model (Model 5) attempted to distinguish between flow effects and the potent ial effects of an unplanned and anomalous 2011 discharge year. This model considers separate apparent survival rates during fluctuating flows, during steady flows of 2009 and 2010, and then considers high steady flows of 2011 separately. The apparent survi val 95% confidence intervals were 56 77%, 10 86%, and 11 63%, respectively. Capture probability ranged from 9 25%, still estimated uniquely for each NSE style trip. Discussion The 2009 2011 fall steady flow experiment from Glen Canyon Dam did not change th e apparent survival rate of juvenile humpback chub compared to fluctuating flow operations. Models that estimated apparent survival by flow treatments (e.g. steady versus fluctuating dam releases) were indistinguishable based on my model selection criteria Estimates of apparent survival within flow treatments all had overlapping 95% confidence intervals, and data were not trending toward statistical significance. In addition, the top five models regardless of whether they included flow (annual and constant models) as a factor of apparent survival were all separated by less than five c points. The lack of contrast in estimates within models coupled with the lack of distinction between models indicates that apparent survival of juvenile humpback chub in G rand Canyon is robust to flow variation of the magnitude that I observed. The lower rate for apparent survival from July 2011 July 2012 indicated by the annual model (Figure 3 8) is not alarming because the trend is not continuous over the
56 three year perio d. Higher uncertainty is typical of the latter intervals in mark recapture experiments, as many fish are still of unknown fate (dead versus not observed). Previous NSE data showed similar trends in annual survival for 2010 when it was the last period sampl ed; a trend which later improved with the addition of 2011 data. I expect numerical increases in 2011 2012 apparent survival rates as well as improvement in precision of those estimates as juvenile chub sampling (ongoing) continues in the future. I was con cerned about bias that may occur in my estimates of survival due to differences in recapture probability versus capture probability. However, Program MARK uses all future captures of any individual fish to inform previous captures (conditioned on capture), so every additional sampling trip reduces potential bias that stems from individuals avoiding recapture. I ignored recaptures within the same trip, so most recapture data that I used in analyses occurred at least 30 days later, after fish had time to mix and return to normal behavior (i.e., no tagging effects). This is reinforced by the fact that I recaptured 39% (1989 of 5154) of juvenile humpback chub on later sampling trips. The suite of simulations I performed is an important demonstratio n of the power of MARK, as it proves that even if you have a large decline in the numbers of recaptured fish during the experimental flow (wh ich I saw in October 2009 in this study), the software can accurately determine if that decline is due to changes in apparent sur vival or in capture probability (Figures 3 3 through 3 6) These simulations also demonstrate the importance of experimental replicates if sample sizes are limited in order to avoid non convergence or highly imprecise estimates. Lastly, if the perturbation
57 that is causing the change in apparent survival also influences capture probability in the same direction and magnitude, annual apparent survival rates may need to change by as much as 30% before Program MARK will return statistically significant results. Direct estimates of survival of juvenile fish are invaluable to managers of endangered species or sport fisheries. Difficulties in mark recapture of small fish unfortunately leave those managers with indirect tools to assess and manage survival such as st ocking regimens and influencing growth rate to improve survival. Managing dam releases to directly improve survival rates of resident fishes on an ecosystem scale offers a more efficient and simple way to improve the status of the species of interest. Alth ough I did not observe an effect of the flow treatments thus far prescribed in Grand Canyon, I have demonstrated that it is possible to use mark recapture techniques with juvenile fish that are below the minimum PIT tag size (<100 mm TL) to provide direct and credible vital rate estimates before fish recruit to adulthood. Previous estimates of survival for juvenile humpback chub were back calculated through recruitment reconstruction as part of the Age Structured Mark Recapture model (Coggins et al. 2006 b ). The ability to directly estimate juvenile survival allows managers to quickly monitor how experimental actions affect vital rates during critical life stages and ultimately move toward humpback chub population recovery. I have also demonstrated that juvenile humpback chub (<100 mm TL) can survive, overwinter, and grow in the mainstem Colorado River. We observed 30.6% (1579 of 5154) of the juvenile humpback chub that we originally VIE tagged below 100 mm TL grow to PIT tag size (>100 mm TL) over the three year course of this sampling project, while per trip capture probabilities were 26% or less. Annual apparent survival
58 rates of the three top models ranged from 37 69% regardless of flows. These models were indistinguishabl c score, which lends credence to a constant survival rate of 63% as estimated by the constant apparent survival model. Humpback chub of PIT tag size that are still juveniles (100 200 mm TL) would likely survive at even higher rates, which has sign ificant population implications. Although the steady flows that I observed during 2009 2011 did not result in improvements to vital rates of juvenile humpback chub, this does not suggest that flow variations are unequivocally beneficial or detrimental to n ative fish populations. A key limitation of this steady flow experiment as designed is that the steady flows occurred during September and October, which were beyond the time period of maximum solar insolation for this reach of Grand Canyon (Yard et al. 20 05 ). Future experimental scenarios could include altering the timing of the fluctuating/steady flow experiments to determine if the lack of change in apparent survival during steady flow periods is a result of the steady flow experiment or seasonal factors Ideal contrast would be steady flows in July and August when solar insolation, nearshore warming, and primary production potential are highest, with fluctuating flows in September and October. Additional experiments varying the volume of water released d uring steady flows would also clarify the possible effects of flow modification (less water warms more quickly). The magnitude of the experimental flow regime compared to the pre dam environment in which humpback chub evolved is a critical consideration in evaluating juvenile humpback chub responses to this flow experiment. The pre settlement Colorado River in Grand Canyon generally reached a historical base flow in fall of less than 85 m 3 /s (Schmidt et al. 2005, Figure 2 4 ), and stochastic monsoonal precip itation
59 from the Little Colorado River and other tributaries would sometimes double the river discharge almost instantaneously. During spring and summer runoff, snowmelt dominated floods peaked at 2400 m 3 /s on a 2 year frequency (Topping et al. 2003). Thes e 100% changes in instantaneous rate over short periods, to well over 1000% changes in discharge annually, were significant environmental factors which shaped the Colorado River ecosystem in which humpback chub and other native species evolved. During my study period, the maximum annual change in discharge was ten percent of the historica l annual fluctuation (Figure 2 4 ). The tempering of natural annual flow variation due to operation of Glen Canyon Dam has allowed humpback chub, which are flexible in habi tat selection (Kaeding and Zimmerman 1983, Stone and Gorman 2006 ) to be robust to the small scale flow fluctuations observed in this study Evaluations of ecosystem experiments such as the Nearshore Ecology project are important to successful ly directing e cosystem management. The ideological crux of Adaptive Management involves reducing uncertainty about complex ecological systems. My data indicates that steady flows of the magnitude I observed in the Colorado River have no appreciable effect on apparent su rvival of juvenile humpback chub (no model distinction ). This information is invaluable in increasing understanding of the flow fish relationship as r egulated rivers continue to be over allocated due to population growth, anthropogenic effects of climate c hange, or even a return to paleoclimate mean discharge (Christensen et al. 2004, Barnett and Pierce 2009) A robust understanding of the relationship between managed flows and life history characteristics of lotic fishes will be necessary to perpetuate the existence of this increasingly imperiled community.
60 Table 3 1. Capture recapture summary table of data provided to MARK for analysis. Number of fish marked Recapture trip number 2 3 4 5 6 7 8 9 10 11 12 13 14 Total Marking trip number 1 266 40 37 12 23 12 14 5 7 5 7 3 3 5 173 2 260 47 20 19 12 15 14 7 4 5 1 1 1 146 3 235 19 20 19 24 20 5 13 4 4 0 0 128 4 85 8 9 6 8 2 6 4 0 1 0 44 5 139 24 29 14 6 8 9 3 0 2 95 6 218 68 20 5 9 5 1 1 2 111 7 359 109 22 27 14 6 6 9 193 8 258 34 34 27 8 6 2 111 9 139 30 19 12 6 5 72 10 519 81 33 23 12 149 11 885 88 75 37 200 12 478 42 26 68 13 829 89 89 Rows and columns represent corresponding sampling trips. Values represent counts of juvenile humpback chub seen during a given trip (column) that were marked in a previous trip (row).
61 Table 3 2. Apparent survival models for juvenile humpback chub in Grand Canyon from 2009 Model name AIC c c No. Par. Phi(year) p(t) 9947.1 0.0 16 Phi(flow) p(t) 9949.8 2.7 15 Phi(.) p(t) 9949.9 2.8 14 Phi(mod_flow) p(t) 9951.2 4.1 16 Phi(year*flow) p(t) 9951.5 4.4 16 Phi(year) p(year) 10042.9 95.9 6 Phi(year*flow) p(flow) 10061.9 114.8 5 Phi(year) p(mod_flow) 10062.9 115.8 6 Phi(year*flow) p(mod_flow) 10063.8 116.7 6 Phi(year) p(year*flow) 10066.5 119.4 6 Phi(year) p(.) 10067.2 120.1 4 Phi(year) p(flow) 10067.5 120.4 5 Phi(mod_flow) p(year) 10078.5 131.4 6 Phi(mod_flow) p(flow) 10079.9 132.8 5 Phi(flow) p(mod_flow) 10080.8 133.7 5 Phi(.) p(year) 10080.8 133.7 4 Phi(mod_flow) p(mod_flow) 10081.9 134.8 6 Phi(flow) p(year) 10082.0 135.0 5 Phi(.) p(mod_flow) 10101.3 154.2 4 Phi(mod_flow) p(year*flow) 10118.2 171.1 6 Phi(flow) p(year*flow) 10121.3 174.2 5 Phi(flow) p(flow) 10123.3 176.2 4 Phi(mod_flow) p(.) 10142.1 195.0 4 Phi(flow) p(.) 10149.9 202.8 3 Phi(.) p(year*flow) 10151.1 204.0 4 Phi(.) p(flow) 10155.0 207.9 3 Phi(.) p(.) 10158.2 211.1 2 Models with non convergence Phi(t) p(t) 9950.3 3.2 26 Phi(t) p(year) 10016.7 69.6 16 Phi(t) p(flow) 10019.2 72.1 15 Phi(t) p(mod_flow) 10020.2 73.1 16 Phi(t) p(.) 10026.7 79.6 14 Phi(t) p(year*flow) 10029.7 82.6 16 Phi(year*flow) p(year) 10071.5 124.4 6 Phi(year*flow) p(year*flow) 10088.1 141.0 6 Phi(year*flow) p(.) 10099.8 152.7 4 Model parameter estimation is symbolized as follows: constant (.), time dependent (t), fish year (year), flow dependent (flow), modified flow dependent (mod_flow), and year flow interaction (year*flow).
62 Figure 3 1. Colorado River mainstem flow treatments over the course of this study ( July 2009 July 2012) Black areas represent discharges that fluctuated daily according to hydropower demands, while gray represents discharges that were held experimenta lly steady.
63 Figure 3 2. Length frequency histogram o f juvenile humpback chub considered in this survival analysis. Fish were considered removed from the population when they were PIT tagged (first capture above 100 mm TL).
64 Figure 3 3. A pparent survival vs. capture probability simulations using mark/recapture data generated in GENCAPH1 and analyzed using Program MARK: Phi(t) p(.) and Phi(.) p(t) models. Note failure to estimate apparent survival for Phi(t) models (estimates are either unbounded o r missing).
65 Figure 3 4. A pparent survival vs. capture probability simulations using mark/recapture data generated in GENCAPH1 and analyzed using Program MARK: Phi(t) p(t) models. Note failure to estimate apparent survival for all models (estimates are missing or unbounded).
66 Figure 3 5. A pparent survival vs. capture probability simulations using mark/recapture data generated in GENCAPH1 and analyzed using Program MARK: Phi(flow) p(t) models. Note model accuracy in estimating apparent survival despit e order of magnitude change in capture probability due to experimental replicates (years).
67 Figure 3 6. A pparent survival vs. capture probability simulations using mark/recapture data generated in GENCAPH1 and analyzed using Program MARK: Phi(flow) p(fl ow) models. Apparent survival estimates are accurate if capture probability effects of flow are present, although some precision is lost with capture probabilities <10%. Field data is strongly unsupportive of p (flow) models over p (t) models (Table 3).
68 Figure 3 7 Trip specific capture probability of juvenile humpback chub over the thirteen recapture opportunities based on three leading models.
69 Figure 3 8 Annual apparent survival estim ates for juvenile humpback chub, grouped by of one year to July of the next year time dependent capture probability, see Figure 3 3)
70 Figure 3 9 Annual apparent survival estimates for juvenile humpback chub, separated by flow treatments.
71 Figure 3 10 Annual apparent survival estimates for juvenile humpback chub, as derived by the constant model (time dependent capture probability, see Figure 3 3).
72 CHAPTER 4 CONCLUSION S It is commonly accepted that the more natural a flow regime, the more likely it is to favor native versus non native fi sh ( Marchetti and Moyle 2001, Schultz et al. 2003, Propst and Gido 2004 ) However, in some systems, river regulation has positively aff ected native fish populations ( Crisp et al. 1983) or certain native species (Welker and Scarnecchia 2003) The Colorado River of the southwestern United States was a muddy an d violent river prone to discharge extremes before construction and operation of both Glen Canyon and Hoover dams Severe droughts drastic ally reduced aquatic habitat and could leave milli ons of fish dead (Minckley 1991, Mueller and Marsh 2002) Flooding likely had less of an impact where fish could escape into floodplains or other refugia (Minckley 1991), but certainly would exacerbate stress and displace some portion of the population, es pecially in canyon bound reaches. While millennia of behavioral and morphological adaptation likely shaped the native fish community to reduce the detrimental impact of these environmental pressures, in a system where natural conditions were physically d angerous or energetically costly, it does not automatically follow that a return to natural flows will trigger positive response s in the native fish community. I was unable to detect changes in apparent survival in juvenile humpback chub Gila cypha (C hapte r 3) during experimental steady flows in Grand Canyon Although the robust to observed hydropower fluctuations as well, and a return to slightly more natural flow s predictably has no significant effect on apparent survival rates. Other factors may be more important than discharge, especially non native fish and water temperature.
73 This is not surprising upon consider ing the state of the fish community in the unregul ated portions of large tributaries to the Colorado River (upper Salt and Verde rivers, Gila River, upper Green River). If a natural hydrograph and lack of hydropower fluctuations were sufficient to exclude invasive species and favor the natives, we would expect the upper portions of these rivers to be devoid of non native fish. This expectation is negated in the face of ubiquitous invasive predators and the common extirpation of native fish fauna despite natural flows. On the other hand, the const ruction of Glen Canyon and Hoover dams and the subsequent cooling of the Colorado River through Grand Canyon may have actually been the timely factor that stemmed the ongoing invasion of warm water non native fish that had already led to the extirpation of native fish in the lower reaches of the Colorado River (Minckley 1991). Rainbow trout and native fish have coexisted for decades in Grand Canyon (and their congeners as sympatriates for millennia in headwater areas), while the families Centrarc hidae, Icta luridae, and Moronidae have almost universally displaced native fish populations where they are introduced. Within the corridor of Grand Canyon near the confluence of the Little Colorado d cliffs abutting the water render discharge changes relatively insignificant Slightly m oderating fluctuating hydropower flows by releasing steady discharges for a two month period in the fall all within a range of less than one pre dam flows predictably did not have a measurable effect on apparent survival rate of juvenile humpback chub
74 Although apparent survival rates did not change significantly between flows I did observe declines in growth rate for juvenile humpback chub during the fall steady flow experiments (C hapter 2) This may be the result of three different interactions: (1) The LCR aggregation of humpback chub occupies a river reach that has r elatively steep angle shorelines compared to other parts of the canyon, and habitat selection or availability for juvenile fish is relatively invulnerable to flow fluctuations. This may also be the reason that the humpback chub population has survived well in this canyon reach despite the influence of Glen Canyon Dam. However, the declines I observed in growth rates during steady flows indicate that this may not be the case. (2) The fall steady flow experiment was relatively short (2 months long) and, while producing conditions that could cause declines in growth rate among individual fish, it did not persist long enough to have a measureable population effect on apparent survival rates before flow returned to the typical modified fluctuating flow regi me. La stly, (3) fish included in these apparent survival analyses were below 100 mm TL (down to 40 mm TL) whereas fish that composed my growth treatment groups were all over 100 mm TL (in order to be tracked individually with PIT tags). Humpback chub of differen t size groups choose different habitats (Stone and Gorman 2006) and likely occupy different functional feeding groups that would cause their growth and survival rates to differ. Further study of survival rates of larger juveniles would help elucidate this possible difference. While experimental steady flows of the magnitude that I observed did not improve vital rates of juvenile humpback chub, this does not automatically condemn all possible steady flow treatments. Shifting the period of steady flows to ear lier in the year
75 when solar insolation and potential warming are highest (Yard et al. 20 05 ), or extending the period of steady flows long enough for the foodbase to respond may have a different effect on the native fish community I encourage managers of G len Canyon Dam and other regulated rivers to persist in innovative experimentation as we all strive to meet societal pressures for resources while preserving ecological integrity for future generations.
76 LIST OF REFERENCES Anderson, M. T. and L. H. Woosley, Jr. 20 05. Water availability for the w estern United States key scientific challenges. U.S. Geological Survey Circular 1261, Denver, Colorado. Andrews, E. D. 1991. Sediment transport in the Colorado River basin. Pages 54 74 in Nation al Resource Council (U.S.) committee to r eview the Glen Canyon Environmental Studies. Colorado River Ecology and Dam Management: Proceedings of a Symposium May 24 25 1990, Santa Fe, NM. National Academy Press, Washington, D.C. Bain, M. B., J. T. Finn, and H. E. Book e. 1988. Streamflow regulation and fish co mmunity structure. Ecology 69 :382 392. Barnett, T. P. and D. W. Pierce. 2009. Sustainable water deliveries from the Colorado River in a changing climate. Proceedings of the National Academy of Science 106 :7334 733 8. Biro, P. A., A. E. Morton, J. R. Post, and E. A. Parkinson. 2004. Over winter lipid depletion and mortality of age 0 rainbow trout ( Oncorhynchus mykiss ). Canadian Journal of Fisheries and Aquatic Sciences 61:1513 1519. Blinn, D. W. and N. L. Poff. 200 5. Colorado River Basin. Pages 483 526 in Benke, A. C. and C. E. Cushing, editors. Rivers of North America. Elsevier Academic Press, Burlington, Massachusetts. Bunt, C. M., S. J. Cooke, C. Katopodis, and R. S. McKinley. 1999. Movement and summer habitat o f brown trout ( Salmo trutta ) below a pulsed discharge hydroelectric generating station. Regulated Rivers: Research and Management 15:395 403. Burnham K. P. and D. R. Anderson 2002 Model selection and multimodel inference. Springer, New York, N ew Y ork Carey, H. V. and P. Moore. 1986. Foraging and predation risk in yellow bellied marmots. Am erican Midland Naturalist 116 :267 275. Carothers, S. W., M. E. Stitt, and R. R. Johnson. 1976. Feral asses on public lands: an analysis of biotic impact, legal cons iderations, and management alternatives. Transactions of the North American Wildlife and Natural Resources Conference 41:396 405. Crisp, D. T., R. H. K. Mann, and P. R. Cubby. 1983. Effects of regulation of the River Tees upon fish populations below Cow G reen Reservoir. Journal of Applied Ecology 20 :371 386.
77 Christensen, N. S., A. W. Wood, N. Voisin, D. P. Lettenmaier, and R. N. Palmer. 2004. The effects of climate change on the hydrology and water resources of the Colorado River basin. Climatic Change 62:337 363. Clarkson, R. W. and M. R. Childs. 2000. Temperature effects of hypolimnial release dams on early life stages of Colorado River basin big river fishes. Copeia 2000 :402 412. Coggins, L. G. Jr., W. E. Pine III, C. J. Walters, D. R. Van Haverbeke D. Ward, and H. C. Johnstone 2006 a Abundance trends and status of the Little Colorado River population of humpback chub. North American Journal of Fisheries Management 26:233 245. Coggins, L. G. Jr., W. E. Pine III, C. J. Walters, and S. J. D. Martel l 2006 b Age structured mark recapture analysis: a virtual population analysis based model for analyzing age structured capture recapture data. North American Journal of Fisheries Management 26:201 205. Coggins, L. G. Jr. and W. E. Pine III. 2010. Develo pment of a temperature dependent growth model for the endangered humpback chub using capture recapture data. The O pen Fish Science Journal 2010 :122 131. Coggins, L. G. Jr., M. D. Yard, and W. E. Pine III. 2011. Nonnative fish control in the Colorado River in Grand Canyon, Arizona: an effective program or serendipitous timing? Transactions of the American Fisheries Society 140 :456 470. Converse, Y. K., C. P. Hawkins, and R. A. Valdez. 1998. Habitat relationships of subadult humpback chub in the Colorado Ri ver through Grand Canyon: spatial variability and implications of flow regulation. Regulated Rivers: Research and Management 14:267 284. Cowlishaw, G. 1997. Trade offs between foraging and predation risk determine habitat use in a desert baboon population Animal Behavior 53:667 686. Department of the Interior (DOI). 2002. Proposed experimental releases from Glen Canyon Dam and removal of non native fish. Environmental Assessment, Salt Lake City, Utah. Available: http://azmemory.lib.az.us/cgi bin/showfile.exe?CISOROOT=/feddocs&CISOPTR=780&filename=781.pdf ( October 2011 ) Douglas, M. E. and P. C. Marsh. 1996. Population estimates/population moveme nts of Gila cypha and endangered cyprinid fish in the Grand Canyon region of Arizona. Copeia 1:15 28.
78 Dutterer, A. C., C. Mesing, R. Cailteux, M. S. Allen, W. E. Pine, and P. A. Strickland. 2012. Fish recruitment is influenced by river flows and floodpla in inundation at Apalachicola River, Florida. River Research and Applications (in press ). Fisher, S. G. and A. LaVoy. 1972. Differences in littoral fauna due to fluctuating water levels below a hydroelectric dam. Journal of the Fisheries Research Board of Canada 29 :1472 1476. Foley, J. A., R. DeFries, G. P. Asner, C. Barford, G. Bonan, S. R. Carpenter, F. S. Chapin, M. T. Coe, G. C. Daily, H. K. Gibbs, J. H. Helkowski, T. Hollway, E. A. Howard, C. J. Kucharik, C. Monfreda, J. A. Patz, I. C. Prentice, N Ramankutty, and P. K. Snyder. 2005. Global consequence s of land use. Science 309 :570 574. Fule, P. Z., W. W. Covington, H. B. Smith, J. D. Springer, T. A. Heinlen, K. D. Huisinga, and M. M. Moore. 2002. Comparing ecological restoration alternatives: Grand Canyon, Arizona. Forest Ecology and Management 170:19 41. Gloss, S. P. and L. G. Coggins. 2005. Fishes of Grand Canyon. Pages 33 56 in Gloss, S. P., Lovich, J. E., and T. S. Melis, editors. The State of the Colorado River Ecosystem in Grand Canyon. Unite d States Geological Survey Circular 1282, Reston, Virginia. Gorman, O. T. and D. M. Stone. 1999. Ecology of spawning humpback chub, Gila cypha in the Little Colorado River near Grand Canyon, Arizona. Environmental Biology of Fishes 55:115 133. Gould, S. J. and E. S. Vrba. 1982. Exaptation a missing term in the science of form. Paleobiology 8:4 15. Goulet, C. and K. LaGory. 2009. Annotated bibliography for the humpback chub ( Gila cypha ) with emphasis on the Grand Canyon population. Report of A rgonne National Laboratory Environmental Division prepared for Western Area Power Administration Colorado River Storage Project Management Center, Salt Lake City, Utah. Available: http://www. ipd.anl.gov/anlpubs/2009/10/65302.pdf ( November 2012 ) Grand Canyon Protection Act (GCPA). 1992. United States Bureau of Reclamation. Reclamation Projects Authorization and Adjustment Act of 1992. Available: http://www.usbr.gov/uc/legal/gcpa1992.html (O ctober 2011 ) Houde, E. D. 1997. Patterns and trends in larval stage growth and mortality of teleost fish. Journal of Fish Biology 51:52 83.
7 9 Hughes, T. C. 1991. Reservoir operations. Pages 207 22 5 in National Resource Council (U.S.) Committee to Review the Glen Canyon Environmental Studies. Colorado River Ecology and Dam Management: Proceedings of a Symposium May 24 25 1990, Santa Fe, NM. National Academy Press, Washington, D.C. Ingram, H., A. D. Tarlock, and C. R. Oggins. 1991. The law and politics of the operation of Glen Canyon Dam. Pages 10 27 in National Resource Council (U.S.) Committee to Review the Glen Canyon Environmental Studies. Colorado River Ecology and Dam Management: Proceedings of a Symposium May 24 25 1990, Santa Fe, NM. National Academy Press, Washington, D.C. Irvine, J. R. and P. R. Henriques. 1984. A preliminary investigation on effects of fluctuating flows on invertebrates of the Hawea River, a large regulated river in New Zealand. New Zealand Journal of Marine and Freshwater Research 18:283 290. Kaeding, L. R. and M. A. Zimmerman. 1983. Life history and ecology of the humpback chub in the Little Colorado and Colorado Rivers of the Grand Canyon. Transactions of the A merican Fisheries Society 112 :577 594. Kearsley, L. H., J. C. Schmidt, and K. D. Warren. 1994. Effects of Glen Canyon Dam on Colorado River sand deposits used as campsites in Grand Canyon National Park, USA. Regulated Rivers: Research and Management 9:137 149. Kennedy, T. A. and S. P. Gloss. 2005. Aquatic ecology: the role of organic matter and invertebrates. Pages 87 101 in Gloss, S. P., Lovich, J. E., and T. S. Melis, editors. The State of the Colorado River Ecosystem in Grand Canyon. United States Geological Survey Circular 1282, Reston, Virginia. Korman, J., M. Kaplinski, and J. Buszowski. 2006. Effects of air and mainstem water temperatures, hydraulic insolation, and fluctuating flows from Glen Canyon Dam on water temperatures in shoreline environments of t he Colorado River in Grand Canyon. Final report for Grand Canyon Monitoring and Research Center, Flagstaff, Arizona. Korman, J. and S. E. Campana. 2009. Effects of hydropeaking on nearshore habitat use and growth of age 0 rainbow trout in a large regulate d river. Transactions of the American Fisheries Society 138:76 87. Lebreton, J. D., K. P. Burnham, J. Clobert, and D. R. Anderson. 1992. Modeling survival and testing biological hypotheses using marked animals: a unified approach with case stud ies. Ecolog ical Monographs 62 :67 118. Lessard, R. B., S. J. D. Martell, C. J. Walters, T. E. Essington, and J. F. Kitchell. Should ecosystem management involve active control of species abund ances? Ecology and Society 10 :1.
80 Lorenzen, K. 2006. Population manage ment in fisheries enhancement: g aining key information from release experiments through use of a size dependent mortality model. Fisheries Research 80:19 27. Lovich, J. and T. S. Melis. 2007. The state of the Colorado River ecosystem in Grand Canyon: Lessons f rom 10 years of adaptive ecosystem management. International Journa l of River Basin Management 5 :207 221. 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. Marchetti, M. P. and P. B. Moyle. 2001. Effects of flow regime on fish assemblages in a regulated California strea m. Ecological Applications 11 :530 539. McCullough, D. A., J. M. Bartholow, H. I. Jager, R. L. Beschta, E. F. Cheslak, M. L. Deas J. L. Ebersole, J. S. Foott, S. L. Johnson, K. R. Marine, M. G. Mesa, J. H. Peterson, Y. Souchon, K. F. Tiffan, and W. A. Wurtsbaugh. 2009. Research in thermal biology: burning questions for coldwater stream fishes. Re views in Fisheries Science 17 :90 115 McKnight, T. L. 1958. The feral burro in the United States: distribution and problems. The Jour nal of Wildlife Management 22 :163 179. Minckley, W. L. 1991. Native fishes of the Grand Canyon region: an obituary? Pages 124 177 in Marzolf, G. R., editor. Colorado River Ecology and Dam Management. National Academy Press Washington, D.C. Mueller, G.A., and P. C. Marsh. 2002. Lost, a desert river and its native fishes: a historical perspective of the lower Colorado River. U nited States Geological Survey Information and Technolog y Report USGS/BRD/ITR 2002 0010, Fort Collins, Colorado. Muller, K. 1974. Stream drift as a chronobiological phenomenon in running water ecosystems. Annual Review of Ecology and Systematics 5:309 323. National Park Service (NPS). 2012. Grand Canyon National Park, Plan Your Visit. Available: http://www.nps.gov/grca/planyourvisit/index.htm ( February 2012 ) Nilsson, C., C. A. Reidy, M. Dynesius, and C. Revenga. 2005. F ragmentation and flow 408. Nilsson, P. A. and C. Bronmark. 2000. Prey vulnerability to a gape size limited predator: behavioral and morphological impacts on northern pike piscivory. Oikos 88:5 39 546.
81 Oghutu Ohwayo, R. 1990. The decline of the native fishe s of Lakes Victoria and Kyoga (e ast Africa ) and the impacts of introduced species, especially the Nile perch, Lates niloticus and the Nile tilapia, Oreochromis niloticus Environmental Biolo gy of Fishes 27:81 96. Paukert, C. P., L. G. Coggins, Jr., and C. E. Flaccus. 2006. Distribution and movement of humpback chub in the Colorado River, Grand Canyon, based on recaptures. Transactions of the American Fisheries Society 135:539 544. Perry, S. A. and W. B. Perry. 1986. Effects of experimental flow regulation on invertebrate drift and stranding in the Flathead and Kootenai Rivers, Montana, USA. Hydrobiologia 134:171 182. Pine, W. E. III, J. Korman, K. E. Limburg, M. S. Allen, and T. K. Frazer. 2009. Nearshore ecology of Grand Canyon fish. Research proposal submitted to United States Geological Survey, Grand Canyon Monitoring and Research Center, Flagstaff, Arizona. Available: http://www.usbr.gov/uc/rm/amp/twg/mtgs/09jun22/NSE_Pine_rev.pdf ( October 2011 ) Poff, N. L., J. D. Allan, M. B. Bain, J. R. Karr, K. L. Prestegaard, B. D. Richter, R. E. Sparks, and J. C. Stromberg. 1997. The natura l flow regime. Biosc ience 47 :769 784. Propst, D. L. and K. B. Gido. 2004. Responses of native and non native fishes to natural flow regime mimicry in the San Juan River. Transactions of the American Fisheries Society. 133:922 931. Reisner, M. 1986. Cadil lac desert. Viking, New York Rice, J. A., T. J. Miller, K. A. Rose, L. B. Crowder, E. A. Marschall, A. S. Trebitz, and D. L. DeAngelis. 1993. Growth rate variation and larval survival: inferences from an individual based size dependent predation model. Canadian Journal of F isheries and Aquatic Sciences 50:133 142. Richter, B. D. and G. A. Thomas. 2007. Restoring environmental flows by modifying dam operations. Ecology and Society 12:12. Robinson, A. T. and M. R. Childs. 2001. Juvenile growth of native fishes in the Little Colorado River and in a thermally modified portion of the Colorado River. North American Journal of Fisheries Management 21:809 815. Ruffner, G. A. and S. W. Carothers. 1982. Age structure, condition, and reproduction of two Equus asinus (Equidae) populat ions from Grand Canyon National Park. T he Southwestern Naturalist 27 :403 411.
82 Savidge, J. A. 1987. Extinction of an island forest avifauna by an introduced sna ke. Ecology 68 :660 668. Scheidegger, K. J. and M. B. Bain. 1995. Larval fish distribution and m icrohabitat use in free flowing and regulated rivers. Copeia 1995 :125 135. Schmidt, J. C., R. H. Webb, R. A. Valdez, G. R. Marzolf, and L. E. Stevens. 1998. Science and values in river restoration in th e Grand Canyon. Bioscience 48 :735 747. Schmidt, L. M, S. P. Gloss, and C. N. Updike. 2005. Overview. Pages 1 16 in Gloss, S. P., Lovich, J. E., and T. S. Melis, editors. The State of the Colorado River Ecosystem in Grand Canyon. United States Geological Survey Circular 1282, Reston, VA. Schult z, A. A., O. E. Maughan, S. A. Bonar, and W. J. Matter. 2003. Effects of flooding on abundance of native and non native fishes downstream from a small impoundment. North American Journal of Fisheries Management 23:503 511. Shannon, J. P., D. W. Blinn, P. L. Benenati, and K. P. Wilson. 2006. Organic drift in a regulated desert river. Canadian Journal of Fisheries and Aquatic Sciences 53:1360 1369. Sogard, S. M. 1997. Size selective mortality in the juvenile stage of teleost fishes: a review. Bulletin of Ma rine Science 60 :1129 1157. Stanford, J. A. and J. V. Ward. 1991. Limnology of Lake Powell and the chemistry of the Colorado River. Pages 75 102 in National Resource Council (U.S.) Committee to Review the Glen Canyon Environmental Studies. Colorado River Ecology and Dam Management: Proceedings of a Symposium May 24 25 1990, Santa Fe, NM. National Academy Press, Washington, D.C. Stevens, L. E., J. P. Shannon, and D. W. Blinn. 1997. Colorado River benthic ecology in Grand Canyon, Arizona, USA: dam, tributary and geomorphical influences. Regulated Rivers: Research and Management 13:129 149. Stone, D. M. 2005. Effect of baiting on hoop net catch rates o f endangered humpback chub. North American Journal of Fisheries Management 25:640 645. Stone, D. M. and O. T. Gorman. 2006. Ontogenesis of endangered humpback chub ( Gila cypha ) in the Little Colorado River, Arizona. The American Midland Naturalist 155:123 135. Strobbe, F., M. A. McPeek, M. De Block, and R. Stoks. 2011. Fish predation selects for reduced foraging activity. Behavioral Ecology and Sociobiology 65:241 247.
83 Thompson, J. M., E. P. Bergersen, C. A. Carlson, and L. R. Kaeding. 1991. Role of size condition, and lipid content in overwinter survival of age 0 Colorado squawfish. Transactions of the American Fisheries Society 120:346 353. Topping, D. J., J. C. Schmidt, and L. E. Vierra Jr. 2003. Computation and analysis of the instantaneous discharg May 8, 1921, through September 30, 2000. United States Geological Survey professional paper 1677 Reston, VA. Truett, J. 1996. Bison and elk in the American southwest Environmental Management 20 :195 206. Urban, M. C. 2 007. The growth predation risk trade off under a growing gape limited predation threat. Ecology 88 :2587 2597. USBR (United States Bureau of Reclamation). 2012. The law of the river. Available: http://www.usbr.gov/lc/region/g1000/lawofrvr.html ( October 2012 ) Valdez, R. A. and R. J. Ryel. 1997. Life history and ecology of the humpback chub in the Colorado River in Grand Canyon, Arizona. Pages 3 31 in Proceedings of the Third Biennial Conference on the Colorado Plateau National Park Service, Flagstaff, AZ. Walters, C. J. and R. Hilborn. 1978. Ecological optimization and adaptive management. Annual Review of Ecological Systems 9:157 188. Walters, C. J. and C. S. Holling. 1990. Large s cale management experiments and learning my doing. Ecology 71 :2060 2068. Walters, C. J. and F Juanes. 1993. Recruitment limitation as a consequence of natural selection for the use of restricted feeding habitats and predation risk taking by juvenile fishe s. Canadian Journal of Fisheries and Aquatic Sciences 50:2058 2070. Ward, D. L. and S. A. Bonar. 2003. Effects of cold water on susceptibility of age 0 flannelmouth sucker to predation by rainbow trout. The S outhwestern Naturalist 48 :43 46. Waters, T. F. 1972. The drift of stream insects. Annual Review of Entomology 17:253 272. Welker, T. L. and D. L. Scarnecchia. 2003. Differences in species comp osition and feeding ecology of C atostomid fishes in two distinct segments of the Missouri River, North Dakota U.S.A. Environmental Biology of Fishes 68:129 141.
84 Werner, E. E., J. F. Gilliam, D. J. Hall, and G. G. Mittelbach. 1983. An experimental test of the effects of predation risk on habitat use in fish. Ecology 6 4:1540 1548. World Commission on Dams (WCD). 2000. Dams and development: a new framework for decision making. The Report of the World Commission on dams. Available: http://www.internationalrivers.org/f iles/world_commission_on_dams_final_report. pdf (September 2011). Yard, M. D., G. E. Bennett, S. N. Mietz, L. G. Coggins, Jr., L. E. Stevens, S. Hueftle, and D. W. Blinn. 2005. Influence of topographic complexity on solar insolation estimates for the Colo rado River, Grand Canyon, AZ. E cological Modelling 183:157 172. Yard, M. D., L. G. Coggins, Jr., C. V. Baxter, G. E. Bennett, and J. Korman. 2011. Trout piscivory in the Colorado River, Grand Canyon: effects of turbidity, temperature, and fish prey availa bility. Transactions of the American Fisheries Society 140:471 486.
85 BIOGRAPHICAL SKETCH Colton Finch was raised on a beef cattle ranch in the Ver de Valley of central Arizona. His interest in the outdoors and the interaction between resource management o rganizations and private consumption of natural resources took him to the College of Agriculture and Life Sciences at the University of Arizona. There he earned his undergraduate degree in Natural Resources Fisheries Conservation and Management and graduated summa cum laude in May 2009. Colton then studied Wildlife Ecology and Conservation at the University of Florida with an e mphasis on fisheries coursework under the t utelage of Dr. William P ine III. He earned his Master of Science from the Universi Canyon is ongoing as owner of the private consulting firm Foothills Ecological Services LLC, where he leads field crews and analyzes data as part of the Natal Origins project, studying the interaction between non native rainbow trout and native fish.