|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 SITE OCCUPANCY AND HABITAT SELECTION OF ENDANGERED HUMPBACK CHUB DURING EXPERIMENTAL FLOW RELEASES FROM GLEN CANYON DAM IN THE COLORADO RIVER IN GRAND CANYON, ARIZONA By BRANDON GERIG A THESIS PRESENTED TO THE GRADUATE SCHOOL O F THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2012
2 2012 BRANDON GERIG
3 To my family and friends
4 ACKNOWLEDGMENTS Funding for this research was provided b y the Bureau of Reclamation. My advisor, Dr. Bill Pine, provided beneficial advice and thoughtful critique throughout the course of my graduate study at University of Florida. I would like to thank my committee members Dr. Mike Allen and Dr. Peter Whaylen for their time and insight. I would like to acknowledge fellow graduate students Michael Dodrill and Colton Finch for their thoughtful discussion and encouragement as we pursued graduate studies together My study could not have been completed without th e dedicated field support from many individuals. I would like to acknowledge research technicians: Jessica Pierson, Jake Hall, Nick Bene, Forrest Hayes and Pilar Wolters, Grand Canyon Boatman: Steve Jones, Brian Dierker, Brian Smith, Pete Weiss, and Matt G ontram, and USGS Grand Canyon Monitoring and Research Center logistical staff: Seth Fel der, Dave Foster and Carol Fritz inger. I would like to thank Arizona Game and Fish staff inclu ding: Aaron Bunch, Brian Clark, Theresa Hunt and Luke Avery I would also l ike to thank USGS Grand Canyon and Monitoring Staff for their guidance, insight and advice througho ut the course of this project. I am particularly indebted to Mike Yard, Glenn Bennett, Ted Melis, Nick Voichick and Ted Kennedy.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 B ACKGROUND ................................ ................................ ................................ ...... 12 2 SITE OCCUPANCY OF HUMPBACK CHUB IN THE COLORADO RIVER IN GRAND CANYON, AZ ................................ ................................ ............................ 15 Introduction ................................ ................................ ................................ ............. 15 Methods ................................ ................................ ................................ .................. 18 Study Area ................................ ................................ ................................ ........ 18 Sampling Procedures ................................ ................................ ....................... 20 Habitat Classification and Covariates ................................ ............................... 21 Description of Occupancy M odels ................................ ................................ .... 22 ................................ ................................ ................ 24 Results ................................ ................................ ................................ .................... 25 Catch Composition ................................ ................................ ........................... 25 Detection Probabilities ................................ ................................ ...................... 25 Hoop net ................................ ................................ ................................ .... 25 Electrofishing ................................ ................................ ............................. 26 Occupancy Probabilities ................................ ................................ ................... 27 Hoop net ................................ ................................ ................................ .... 27 Electrofishing ................................ ................................ ............................. 27 Discussion ................................ ................................ ................................ .............. 28 3 HABITAT SELECTION AND MOVEMENT OF HUMPBACK CHUB ....................... 53 Introduction ................................ ................................ ................................ ............. 53 Methods ................................ ................................ ................................ .................. 55 Study Site ................................ ................................ ................................ ......... 55 Fish handling and surgery methods ................................ ................................ 56 Tracking of telemetered humpback chub ................................ .......................... 57 Habitat mapping ................................ ................................ ............................... 57 Statistical A nalysis Habitat Selection ................................ ................................ 59 Statistical Analysis of Movement ................................ ................................ ...... 61 Results ................................ ................................ ................................ .................... 62 Selection Ratios ................................ ................................ ............................... 62
6 Resource Selection Function ................................ ................................ ............ 63 Movement ................................ ................................ ................................ ......... 64 Discussion ................................ ................................ ................................ .............. 65 4 SYNTHESIS AND CONCLUSIONS ................................ ................................ ........ 80 LIST OF REFERENCES ................................ ................................ ............................... 82 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 89
7 LIST OF TABLES Table page 2 1 Covariates thought to influence probability of occupancy and probability of detection of f ish in Colorado River in Grand Canyon. ................................ ......... 34 2 2 Ranking of site occupancy models used to estimate detection probability of humpback chub captured in repeated hoop net surveys from 2009 2011. ......... 35 2 3 Ranking of site occupancy models used to estimate detection probability of humpback chub in repeated electrofishing surveys from 2009 2011. ................. 36 2 4 Ranking of site occupancy models used to estimate occupancy probability of humpback chub in Grand Canyon. ................................ ................................ ..... 37 2 5 Ranking of site occupancy models used to estimate occupa ncy probability of humpback chub in Grand Canyon. ................................ ................................ ..... 38 3 1 Biological attributes collected from humpback chub implanted with acoustic telemetry transmitters. ................................ ................................ ........................ 69 3 2 C hi square statistics testing (1) the distribution and (2) selection of habitat characteristics of humpback chub during contrasting flow regimes ................... 70 3 3 AIC ranking of mixed model logistic regression models used to determine habitat selection of telemetered humpback chub. ................................ ............... 71 3 4 Coefficient estimates from best AIC ranked mixed model logistic regre ssion for telemetered humpback chub during contrasting flow regimes ....................... 72
8 LIST OF FIGURES Figure page 2 1 Map of study site in the Colorado River in Grand Can yon, A Z. ......................... 39 2 2 Discharge patterns during field sampling in 2009. ................................ .............. 40 2 3 Dischar ge patterns during field sampling in 2010. ................................ .............. 41 2 4 Discharge patterns during field sampling in 2011. ................................ .............. 42 2 5 Aerial image from a G IS showing the typical arrangement of 50 m habitat sub units along shoreline habitats. ................................ ................................ ............ 43 2 6 Patterns of detection probability for different size categories of humpback chub during repeated h oop net surveys in 2009 2011. ................................ ....... 44 2 7 Relationship between probability of detection and mean depth of site during repeated hoop net surveys 2009 to 2011. ................................ .......................... 45 2 8 Patterns of detection probability for stratified by size category of HBC during repeated electrofishing surveys in 2009. ................................ ............................ 46 2 9 Patterns of detection probabilit y for stratified by size category of HBC during repeated electrofishing surveys in 2010. ................................ ............................ 47 2 10 Patterns of detection probability for stratified by size category of HBC during repeated electr ofishing surveys in 2011. ................................ ............................ 48 2 11 Relationship between probability of detection and proportion medium substrate in a site during repeated electrofishing surveys 2009 to 2011. ........... 49 2 12 Relationship between probability of detection and proportion small substrate in a site during repeated electrofishing surveys 2009 to 2011. ........................... 50 2 13 Patterns of occupancy probability stratified by size class and trip during repeated hoop net surveys from 2009 to 2011. ................................ .................. 51 2 14 Patterns of occupancy probability from repeat electro fishing surveys stratified by size class, flow regime and year during 2009 to 2011. ................................ .. 52 3 1 Map of study site in the Colorado River in Grand Canyon, AZ ......................... 73 3 2 Mean, maximum and minimum daily discharge during the fluctuating flow regime and the experimental steady flow regime. ................................ .............. 74 3 3 Mean daily turbidity measurements from July to October 2010 in the Colorado River in Grand Canyon. ................................ ................................ ....... 75
9 3 4 Selection Ratios with 95% Bonferroni confidence intervals of (A.) habitat type, (B .) hydraulic type,(C.) depth category (m) and (D.) substrate category. ... 76 3 5 Kernel density plot of fitted values comparing used versus available locations for the top resource selection func tion. ................................ ............................... 77 3 6 Movement of humpback chub during contrasting flow regimes.. ........................ 78 3 7 Observed distance from shore measurements of tel emetered humpback chub during low, medium and high turbidity levels. ................................ ............ 79
10 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Deg ree of Master of Science SITE OCCUPANCY AND HABITAT SELECTION OF ENDANGERED HUMPBACK CHUB DURING EXPERIMENTAL FLOW RELEASES FROM GLEN CANYON DAM IN THE COLORADO RIVER IN GRAND CANYON, ARIZONA By Brandon Scott Gerig May 2012 Chair: William E. Pine, III Major: Wildlife Ecology and Conservation Conserving native riverine fish communities is in part dependent on how regulated river corridors are managed. In order for dams and hydropower operations to be managed optimally with respect to ecosystem integrity an understanding of the ecological linkages between stream flow conditions, physical habitat and fish community structure is essential. In the Colorado River with in Grand Canyon altered river conditions caused by the installation of Glen Canyon Dam have been implicated in the decline of the humpback chub Specifically, hydropeaking flow schedules or fluctuating flow regimes have caused diel destabilization of nearshore habitats that are thought to be important for humpback chub. Beginning in 2009 a multi year adaptive management experiment was implemented to contrast the current fluctuating flow management policy with an experimental steady flow regime. We examined the habitat use patterns of humpback chub using a patch occupancy approach during 2009 2011 to assess the influence of this experimental steady flow regime on habitat use Additionally, in 2010 I assessed habitat selection for a 30 newly recruited humpback chub implanted with acoustic telemetry tags. Four river trips occurred yearly during July through October 2009 and
11 2010. River sampling surveys were conducted in July and August were during fluctuating flows whereas trips conducted in September and October occurred during steady flows. During this study, I found that site occupancy was most r elated to time varying (e.g. sampling survey, year + flow regime) and size specific covariates for hoop net and electrofishing gears. Habitat covariates received little model support and robust relationships between occupancy and habitat were not observed. For telemetered humpback chub I found that they extensively select eddy hydraulic types while avoiding runs. Tagged fish were also associated with intermediate depths and larger substrates. No effect of the flow experiment was observed. My research sug gests that for both passive and active gears humpback chub did not respond to the steady flow regime. Future research assessing the influence of flow on humpback chub should carefully consider the magnitude and timing of discharge releases while also cons idering the spatial area in which sampling will occur.
12 CHAPTER 1 BACKGROUND This thesis evaluate d the effects of a series of steady flow releases from Glen Canyon Dam on humpback chub in the Colorado River in Grand Canyon, AZ. The completion of Glen C anyon Dam caused a suite of abiotic changes that influenced the ecology of the Colorado River ecosystem (Topping 2003, Gloss and Coggins 2005, Kennedy and Gloss 2005). Prior to Glen Canyon Dam the Colorado River in Grand Canyon exhibited large seasonal ch anges in discharge, turbidity and temperature. In post dam environment, the river has been characterized by elevated base flow, elimination of season al floods, daily fluctuat ion of discharge for hydropower production reduction of sediment inputs, and stab ilized thermal regime (Topping 2003). The changes to the hydrology have had cascading effects on biota. There have been changes to the food base, shifts in the aquatic invertebrate community, increases in the abundance of non native fishes and reductions i n the native fish community. The original Grand Canyon fish community consisted of eight native species and exhibited a high degree of endemism (Minckley and Marsh 2009). Habitat alterations caused by mainstem hydroelectric dams and the widespread intro duction of non native species led to the extirpation of 4 native species (bonytail chub ( Gila elegans ), Colorado pikeminnow ( Ptychocheilus lucius ), razorback sucker ( Xyrauchen texanus )) and roundtail chub ( Gila robusta ) reductions in native fish populatio ns, and increased population sizes of non native fishes in Grand Canyon (Minckley and Marsh 2009). The current Grand Canyon fish community consists primarily of a mixture of native: bluehead sucker ( Catostomus discobolus ), flannelmouth sucker (Catastomous latipinnis), humpback chub (Gila cypha), speckled dace ( Rhinichthys osculus ), and non
13 native: rainbow trout ( Onchorychus mykiss ), brown trout ( Salmo trutta ), black bullhead ( Ameiurus melas ), common carp ( Cyprinus carpio ), fathead minnow ( Pimephales promel as ), plains killifish ( Fundulus zebrinus ) fishes (Gloss and Coggins 2005). The focal species of Colorado River management actions in Grand Canyon is the humpback chub (Coggins et al. 2006 ; Coggins et al. 2011). The humpback chub was listed as a federally endangered species in 1967 and became federally protected in 1973 under the Endangered Species Act. Currently, humpback chub exist in six fragmented populations with in the Colorado River Basin and the largest population occurs in Grand Canyon near the co nfluence of the Little Colorado River and Colorado River (Coggins et al. 2006). The Glen Canyon Dam Adaptive Management Program was formed following the 1996 Record of Decision from the Environmental Impact Statement on the operation of Glen Canyon Dam ( M elis et al. 2006 ). Conservation of native Colorado River Basin fishes, particularly the endangered humpback chub was one of the primary goals of this program. The G len C anyon D am A daptive M anagement P rogram consists of federal, state, tribal and public s takeholders that work together to implement management strategies designed to improve the understanding and management of resources within the Colorado River corridor in Grand Canyon Given that resource response to management is highly uncertain an activ e adaptive management approach is used to couple resource management with research (Melis et al. 2006) This study resulted from formal consultation between the Bureau of Reclamation and U.S. Fish and Wildlife Service to address the implementation of modif ied low fluctuating flow and the effect of this policy action on the endangered humpback chub
14 population in Grand Canyon Within the Glen Canyon Dam Adaptive Management Program stakeholders felt that the continued implementation of modified low fluctuatin g flow negatively influenced native fish and that steady flow releases during certain times of the year would benefit humpback chub and the other members of the extant native fish community. In this study, I utilized a series of experimental steady flow r eleases from 2009 to 2011 from Glen Canyon Dam to relate f dam operations to endangered humpback chub populations in the Colorado River in Grand Canyon As part of this experiment, flow operations from G len Canyon Dam were changed from the typica l operati ng policy of modified low fluctuating flow to steady discharge rates for eight weeks. I utilized the manipulation of Glen Canyon Dam operations to assess the difference between flow regimes (modified low fluctuating flow and steady flow) on habitat use an d movement of humpback chub in the Colorado River in Grand Canyon. Specifically, the research objectives are: 1.) assess nearshore habitat use of humpback chub using site occupancy models during contrasting flow regimes and 2.) d etermine the habitat sele ction and movement patterns of newly recruited adult humpback chub during contrasting flow regime s using acoustic telemetry.
15 CHAPTER 2 SITE OCCUPANCY OF HUMPBAC K CHUB IN THE COLORA DO RIVER IN GRAND CANYON, AZ Introduction F ish habitat use in rivers i s a complex trade off between numerous abiotic and biotic factors such as thermal environment, flow regime, food availability, competition (Werner and Hall 1979 ; Butler 1980), predation risk (Werner et al. 1983 ; Walters and Martell 2004), and shifts betwee n foraging and resting areas. Habitat use and selection often varies with life stage for many organisms and ontogenetic habitat shifts are common ly adopted by fish to utilize specific spawning, nursery and foraging area s (Werner et al. 1988; Walters and Ju anes 1993; Faush et al. 2002; Walters and Martell 2004). Understanding the temporal and spatial patterns of habitat use is a critical need in ecology (Johnson 1980 ; Morrison et al. 2006). Observations of animal habitat use are often used to inform decisi ons about habitat requirements thus informing management policies designed to protect specific habitat types and species (Van Horne 1983). Increasingly, efforts to manage aquatic resources have emphasized habitat protection as an approach to manage ecosys tems v ersus traditional approaches of managi ng single species (Minns et al. 1996 ; Minns et al. 1999). Habitat protection and restoration measures have been mandated in the United States (Magnuson Stevens Fisheries Conservation and Management Act, Endangere d Species Act ) as a primary means to conserve fish populations from decline. However, many of these efforts are not well informed and habitat mitigation and restoration efforts are rarely assessed experimentally thus they often fail to assess
16 whether habit at use and population vital rates change following protection or restoration efforts (Van Horne 1983, Quinn and Kwak 2000, Gunn and Sein 2000 Rosenfeld 2003). R iverine ecosystems have been significantly altered by anthropogenic changes to flow regime t hrough the widespread construction of dams or other control structures (Dynesius and Nilsson 1994 ; Poff et al 1997 ; Richter et al. 1997 ; Bunn and Arthington 2002). Large dams often change the timing, magnitude, frequency, duration and rate of change in h ydrologic conditions within rivers (Poff et al. 1997) and these hydrologic changes are thought to alter riverine processes including ecosystem productivity, sediment transport and temperature all of which are thought to have had negative effects on riverin e ecosystems including fish populations (Nilsson et al. 2005 ; Poff et al. 2007). T he completion of Glen Canyon Dam led to dramatic changes in the hydrology and ecology of the Colorado River ecosystem in Grand Canyon (Topping 2003 ; Gloss and Coggins 2005 ; Kennedy and Gloss 2005). The pre dam environment in Grand Canyon exhibited large seasonal changes in discharge, turbidity and temperature. The post dam environment has been characterized by elevated base flow, elimination of season al floods, daily fluctu at ion of discharge for hydropower production reduction of sediment inputs, and stabilized thermal regime (Topping 2003). Along with changes to the physical environment there were large scale changes to the biotic community. Trophic structure shifted from allochthonous to autochthonous production and reduced aquatic invertebrate diversity and abundance (Kennedy and Gloss 2005; Gloss and Coggins 2005). Native fish populations declined as a result of changes to physical habitat and the proliferation of exot ic species (e.g. cold water salmonids) (Yard et al
17 2011). Because of the scientific consensus on the negative consequences of G len C anyon D am on the aquatic ecosystem of the Colorado River in Grand Canyon there has been an increase in restoration efforts designed to manage the dam in a way that mimics elements of the historic flow regime with the intent of positively influencing native fish populations while preserving ecological integrity (Schmidt et al. 1998). T he Glen Canyon Dam Adaptive Management program was implemented in 1996 to address uncertainty in policy actions on the operation of G len C anyon D am on resources within the Colorado River ecosystem. Conservation of native Colorado River Basin fishes, particularly the endangered humpback chub wa s one of the primary goals of the program (Melis et al. 2006). In 2008 an adaptive management experiment was implemented to assess fish population responses to a series of planned steady flow releases from Glen Canyon Dam. G len C anyon D am is typically oper ated as a peak power facility under the modified low fluctuating flow regime in which the dam is peaked daily between 140 (m 3 /sec) to 850 (m 3 /sec) to produce hydroelectric power. Dam operations during September and October were scheduled to operate under a steady flow regime allowing humpback chub populations to be monitored during contrasting flow regime s It was hypothesized that diel fluctuat ions in discharge could influence survival of native fish by destabilizing nearshore habitats that are thought to be i mportant to juvenile humpback chub. Further, it was hypothesized that the steady flow regimes would raise temperature of nearshore areas and improve growth rates. Despite long term monitoring programs which track native fish populations in Grand Cany on most research efforts have focused on humpback chub population status and trend (Coggins et al. 2003) with less attention on habitat use and requirements
18 (Gorman and Ston e 1999; Stone and Gorman 2005). To date, there is only one published study on patte rns of habitat use in humpback chub in the C olorado River in Grand Canyon (Converse et al. 1998) and a little information exists on the role dam operations play in structuring habitat use patterns. The goal of this study is to assess habitat use of humpba ck chub using site occupancy models in the Colorado River in Grand Canyon during contrasting flow regimes to evaluate the impact of a steady flow regime from Glen Canyon Dam on humpback chub. This approach allows the estimation of un biased habitat use or occupancy probabilities by explicitly accounting for heterogeneity in detection. Using this approach, I evaluated the importance of time varying, size specific, site specific and pass specific covariates on patterns of site occupancy while accounting for changes in detectability. This information on habitat use of humpback chub will help guide the management of the Colorado River in Grand Canyon by providing managers with insight into the influence of experimental flows, importance of habitat features and evaluation of sampling methodologies. Methods Study Area This study was conducted w ithin the L ittle C olorado R iver inflow area of the Colorado River in Grand Canyon (Figure 1 1) My study area was 4.5 km in total length and located between Heart Island (RM 125 RKM ) and Lava Chuar Rapid (R K M 131.2 ) The study area was subdivided into three reaches approximately 1500 m in length (3,000 m total shoreline) Each of the study reaches w as separated by small (~1.5 m height ) rapids. Within each reach t he shore line was further broken up into adjacent 50 m nearshore sites using aerial photographs in a geographic information system (Figure
19 2 5) I sampled each reach during 12 river surveys in July, August, September, and October of 2009 t hrough 2011. Discharge P atterns Discharge patterns were continuous ly monitored at 15 min intervals remotely at USGS gauge 09383100 (R K M 165 ). The Colorado River Flow Sediment and Stage model was used to back calculate the difference in average water travel time between USGS gaug e 0983100 a nd the study location (RKM 125 131.2 ) using an unsteady flow model (Wiele and Griffin 1997). As part of th is study July and August surveys during 2009 and 2010 sampling surveys occurred during the m odified low fluctuating flow regime ( MLFF, di el discharges of 250 to 500 m 3 sec in 2009; 300 to 500 m 3 sec in 2010), and September and October sampling survey s were sampled during the low steady flow regime (constant flows of 250 m 3 sec in 2009; 300 m 3 sec in 2010; Figure 1 2 and Figure 1 3). Du e to increases in upper basin water inputs to Lake Powell in 2011 and record low reservoir levels in Lake Mead, discharge patterns in 2011 from GCD were a high steady flow regime (HSF) during July and August (constant flows of 715 m 3 sec) and a medium stea dy flow regime (MSF) during September and October (constant flows of 460 m 3 sec; Figure 1 4). Extensive tributary flooding during August and September 2010 caused increases in discharge that were unrelated to dam operations (Figure 1 3). Daily turbidity m easurements were continuously monitored in 15 minute intervals remotely from an acoustic Doppler profiler at USGS gauge 09383100 (R K M 165 ). An unsteady flow model was used to back calculate water travel time to the study area (Wiele and Griffin 1997). Spi kes in turbidity were caused by tributary floods.
20 Sampling Procedures Hoop nets and boat electrofishing were used to sample humpback chub during this study to evaluate patterns of site occupancy. Eig hty hoop nets were set within designated sites in the f irst and second sampling reach. The hoop net sites were fished during each of the four yearly sampling survey s during 2009 to 2011. Hoop nets were fished for 24 hour passes over 12 consecutive nights during each sampling survey Boat electrofishing was c onducted in sampling reaches one, two and three. The effective sampling area of hoop nets boat electrofishing was estimated to be a 50 m site within 4 m of shore Sampling reach one had 58 sites, sampling reach two had 49 sites and sampling reach three ha d 52 sites. This sampling design allowed for all shoreline areas excluding rapids in my sampling area to be sampled completely. During 2009, e ach site within a reach was repeatedly sampled three times over a five day period for reaches one and two ( 48 hou rs between passes) and over successive nights in reach 3 ( 24 h ours between passes) During 2010 and 2011, e ach site within a reach was repeatedly sampled three times over successive nights ( 24 hours between passes) Electrofishing surveys were conducted using two 6 m Osprey boats outfitted with either a Smith Root Type 12b or Coffelt CPS system powered by a Honda 4500 watt generator to apply a complex pattern of pulsed DC (Speas et al. 2001, Korman et al. 2009). During each electrofishing pass n earshore sites were shocked slowly ( 10 m 1 sec). All electrofishing s ampling began at dark between 1800 hours and 2000 h ours depending on season. All surveys were finished by 0300 h ours All fish captured during hoop net and electrofishing sampling were processed and returned to the site from which they were captured following standard sampling protocols (Persons et al. 2011).
21 Catch data for humpback chub w ere stratified by size class and converted to a binary encounter history of p resence absence observations. Hoop net catches by pass were stratified into three size classes: 40 8 0 mm, 81 150 mm, and 150 250 mm. E lectrofishing catches by pass were stratified into 40 80 mm and 81 150 mm size classes. Size classes were determined by visually analyzing length frequ ency histograms. Humpback chub greater than 150 mm in electrofishing and 250 mm in hoop net surveys were captured too infrequently to be used in occupancy models and excluded from analysis. Habitat Classification and Covariates A suite of site specific c ategorical and continuous habitat covariates were used for modeling humpback chub detection and occupancy probabilities (Table 2 1). Categorical habitat covariates used included habitat type and hydraulic type. Continuous habitat covariates included subst rate proportion, mean depth and mean site depth 2 (Table 2 1). Habitat type in each site was classified in field into four discrete habitat types: cliff, debris fan, talus slope and sandbar following Converse (et al. 1998). Cliff habitats were characterize d as shear walls rising vertically and laterally over the river. Debris fans were characterized as shorelines of large cobble and boulder that were transported into the river corridor by tributaries during flooding. Sandbar habitats were classified as sh orelines of contiguous beach and exposed sand. Talus habitats were classified as shorelines consisting mainly of boulders deposited by rockslides and rockfall. The hydraulic type of a site was mapped in the field and split into two discrete types: eddy f low and downstream flow. Eddies represent areas of recirculating flow that generally occur downstream of channel constricting debris fans. The downstream category represents any portion of the river where the water is traveling downstream.
22 Substrate class es were grouped into three classes: small, medium and large substrates using a modified Wentworth scale. In GIS, a 4 m (electrofishing sites) and 10 m ( hoop net sites) buffer was applied to the river edge and the proportion of each substrate category in th e site was calculated. Substrates classified as small ranged from silt to course sand. Substrates classified as medium represented small gravel to small cobble. Substrates classified as large ranged in size from large cobble to boulder. A bathymetry ma p of the study reach was used to determine the mean elevation of each hoop net and electrofishing site along the river edge Similar to the substrate classification a 4 m and 10 m buffer were applied to electrofishing and hoop net sites. Depth was determin ed as the average shoreline elevation minus the average elevation of the river bottom within each site. Description of Occupancy Models The estimation of occupancy and detection probabilities required the use of presence absence data from repeated hoop n et and electrofishing passes. Multi season occupancy models estimate d probabilities of site occupancy and detection probability using the encounter histories (h i ) within a given site for a given survey (MacKenzie et al. 2006, Donovan and Hines 2007). As a n example, an observed h i =010 represents a site that was sampled for 3 passes in a survey Given incomplete detection, the species was observed only during the second pass The probability of the following h i is: i is the probability that a species in detected in pass i 1 p i is equal to the probability that the species in not detected during pass i An observed detection history, h i = 000 ind icates that during each of the 3 passes the species of interest was not detected. Given that detection probability is less
23 than one this detection history can arise two ways: the species is truly absent or the species is occupying the site yet undetected. The probability of h i =000 is: p 1 )(1 p 2 )(1 p 3 ) is the probability that the species of interest was present at the site but undetected and (1 does not occupy the site of interest. Detection (p maximizing the following model likelihood: which represents the product of each encounter histories (h i ) probability (MacKenzie et al. 2002, MacKenzie et al. 2006, Donovon and Hines 2007) A logit link function was used to model the effect of covariates on occupancy and detection probabilities. The logit link function is specified as: j is the effect of the given covariate x ij o is the intercept of the model. Assumptions of multi season occupancy models included: ( 1) Detectability and occupancy probability of humpback chub is constant among sites or heterogeneity in detection and occupancy can be modeled completely with the inclusion of covariates, (2) detection of humpback chub are independent at each site, and (3) sites are closed to changes in occupancy by humpback chub during the course of a sampling survey (season) I fur ther assumed that assumption three could be relaxed if movement in and out of a site occur ed random ly (Kendall 1999 ; MacKenzie et al. 2006). When random localized movement occurs it is appropriate to restrict our interpretation of occupancy
24 models to the proportion of sites used rather than the proportion of sites occupied (MacKenzie et al. 2006). A priori models were developed to propose mathematical descriptions of factors that influence d site occupancy and detection probab ility of humpback chub (Table 1 1). Following the recommendation of MacKenzie et al. (2006), models were fit for detection probability first and the best model for detection was selected to model site occupancy. Covariates included in models were presume d to be biologically important to occupancy and detection probabilities. Time varying, size specific, site specific and pass specific covariates were considered. Time varying covariates included survey (e.g. 1 12), flow regime (e.g. MLFF, LSF, MSF, HSF ), year (e.g. 2009, 2010, 2011) and year + flow regime (e.g. unique parameter for each year flow regime combination). Size specific covariates included humpback chub size class. Site specific covariates included sampling reach (electrofishing only ), mean site depth substrate proportion (e.g. small, medium, large) site habitat type and site hydraulic type. Mean site depth 2 was also considered to determine if site occupancy and detection exhibited a non linear response to this covariate. Turbidity was consider ed a pass specific covariate and was only modeled for detection probability (Ta ble 1 1). information criteria (AIC, hereafter; Burnham and Anderson 2002). AIC ranks competing models b y com paring maximized log likehoods while applying a penalty for as used to visualize the top model. Model weights (w i ) were calculated to determine the support each model received given the data. Because of model uncerta inty models with AIC w i greater than 0.05 were also
25 used to draw inference from. All occupancy models were fit using Program Presence 3.1 (Hines 2011). Results Catch Composition Eighty hoop net sites were sampled over the course of 12 sampling survey s i n 2009 to 2011. Because of inconsistency in site location and increased number of sites in 2010 and 2011 only 57 sites were included in subsequent analyses for fish captured in hoop net s. A total of twelve fish species were collected in hoop net surveys w hile 15 species were sampled during electrofishing surveys. During hoop net surveys humpback chub (HBC) were the numerically dominate catch in 2009 2011 accounting for 74%, 72.5% and 76% of the total catch (Table 2 2). A total of 158 electrofishing sites were sampled a total of 36 times over 12 survey s in 2009 to 2011. Humpback chub (HBC) accounted for 16% and 10% of the catch in 2009 and 2010 while increasing to 24% of the total catch in 2011. Detection Probabilities Hoop net A total of f ourteen mode ls were constructed to assess heterogeneity in detection across a range of physical, temporal and biological attributes (Table 2 1 ). The best fitting model estimated a range of detection probabilities between 0.05 0.33 and included sampling survey fish s ize, and a quadratic form depth (including mean depth + mean depth squared in model statement) as covariates (Figure 2 6 Table 2 2 ). Detection probabilities were lowest for humpback chub 150 250 mm and highest for chub 40 80 mm and 81 150 mm. Humpback chu b 150 250 mm were detected significantly less than chub between 40 80 mm and 81 150 mm across all sampling
26 survey s with the exception of October of 2009 (Figure 2 6 ). The best fitting model also indicated that the probability of detection was highest at i ntermediate depths (Figure 2 7 ). Detection models that included other habitat covariates (Table 2 1 ) were not supported by the data (Table 2 2 ). Models that did not include sampling survey and fish size as covariates performed poorly (Table 2 2 ). There was no support for models that modeled heterogeneity of detection as a function of year of sampling or flow regime. Electrofishing A total of s ixteen models were constructed to model heterogeneity in detection probability of humpback chub captured during electrofishing surveys (Table 2 3 ). The top detection model included sampling survey sampling reach, humpback chub size class and substrate heterogeneity (proportion of medium sized substrate within the site sampled) as important covariates (Table 2 3 ). Detection probability estimates for this model ranged by time and fish size from the highest values in September of 2011 (range=0.18 0.54) to lowest in October of 2009 (range=0.05 0.22) (Figure 2 8 ; Figure 2 10 ). Detection probabilities for humpback chu b between 40 80 mm captured during electrofishing surveys were always higher than detection probabilities for fish between 81 150 mm across all sampling survey s and sampling reaches. Detection probabilities for both 40 80mm and 81 150 mm humpback chub exh ibited a monotonic decline from sampling reach one to reach three in 2009 through 2011 (Figure 2 8 ; Figure 2 9 Figure 2 10 ). The best fitting model also indicated that there was a positive relationship between the probability of detection and substrate h eterogeneity (proportion of medium substrate within a habitat sampling unit) (Figure 1 1 1 ). The second best fitting model, ranked by AIC was parameterized similarly to the top model except that a negative relationship between detection probability and sub strate homogeneity (proportion small
27 substrate within a site) was observed (Table 2 3 Figure 2 1 2 ). Detection models that included other habitat covariates (Table 2 3 ) were not supported by the data. Models that did not include survey reach and size as covariates fit the data poorly (Table 2 3 ). There was no support for models that modeled heterogeneity of detection as a function of year of sampling or flow regime. Occupancy Probabilities Hoop net A total of t hirteen models were constructed to model h eterogeneity in occupancy ( habitat use) probability of humpback chub captured in hoop net s as a function of habitat temporal and biological attributes. The best fitting model indicated that occupancy probability varied as a function of survey of sampling and size c ategory of humpback chub (Table 2 4 ). Estimated occupancy probabilities were relatively consistent among survey s and size c lass with a mean occupancy probability of 0.80 (Figure 2 1 3 ). During all surveys chub 151 250 mm exhibited the lowest occ upancy probability but this result was not statistically significant. The inclusion of habitat covariates to explain occupancy did not result in a significant improvement in model fit or significant habitat relationships (Table 2 4 ). Electrofishing A t otal of e ighteen models were constructed to model occupancy (use) probability of humpback chub observed in electrofishing surveys. The model best supported by the data indicated that occupancy probability varied as a function of year, flow regime and size category. Overall occupancy probabilities ranged from 0.72 0.90 for chub 40 80 mm and 0.28 0.59 for fish 81 150 mm (Figure 2 12). Occupancy probabilities for humpback 40 80 mm and 81 150 mm declined slightly through time
28 (Figure 2 12). Eight additional of 5 of the top model (Table 1 5 ). Each of these models were parameterized identically to the top model except that they each had a habitat covariate included. The inclusion of habitat covariates in occupancy models did not sign ificantly improve model fit and exhibited weak insignificant habitat relationships (Table 1 5 ). Models that included survey as a covariate provided poor model fit (Table 1 5 ). Discussion During this study, occupancy probabilities were observed to be con sistently high and there was no marked change in occupancy state between the fluctuating flow regime and experimental steady flow regimes. This is interesting because it had been hypothesized that stabilized steady flow would elicit a population level resp onse causing occupancy patterns to change between flow regimes. Because estimates of occupancy probabilities were high, robust relationships between habitat covariates and humpback occurrence were not able to be estimated. Detection probabilities exhibit ed considerable heterogeneity reflecting differences among survey, size class and habitat covariates. The magnitude of discharge releases from G len C anyon Dam between the fluctuating flow regime and steady flow regime s may not have been sufficient to elic it a response in humpback chub populations. The implementation of modified low fluctuating flow in 1996 redu ced the diel range in discharge and reduced the frequency of low flow events (Gloss et al. 2006) effectively minimizing large daily contrasts in dis charge from GCD. Despite differences in discharge during field surveys from 2009 to 2011 the ranges of flows observed in this study were within the operational constraints
29 of modified low fluctuating flow. If the contrasts in flow regimes had been greate r humpback chub responses to steady flow experiment may have been observed. The spatial location of the study area may have contributed to the lack of a response to flow vari ation by humpback chub. Korman ( et al. 2004) modeled humpback chub habitat use over a range of discharges and found that areas which were comprised of steep angle shoreline such as talus and debris fan habitat types were more robust to changes in discharge than low angle homogenous shorelines such as sandbars About 60 % of m y study area was comprised of talus and debris fan shorelines and geomorphically defined by eddy fan complexes The high structural complexity of shorelines in my study area may act as a refuge from discharge fluctuations which allowed humpback chub to be resistan t to changing river stage (Korman et al. 2004). I postulate that the general habitat attributes available to juvenile humpback chub likely did not change enough between the fluctuating flow and the steady flow regimes to necessitate a shift in habitat use While there is no natural analog in river systems to a fluctuating flow regime, minor peaking operations such as modified low fluctuating flow operations from G len C anyon D am may not influence how chub use habitat significantly. However, this could vary spatially and temporally. The timing of the steady flow experiment may have influenced the lack of response by humpback chub to the steady flow regimes. The steady flow regimes prescribed by the Bureau of Reclamation were designed to take place in the fal l of each year. Different experimental designs that alternated the timing of the steady flows (e.g. Fall Steady Flow 2009, Summer Steady Flow 2010, Fall Steady Flow 2011) were not considered. Thus, I was unable to assess how flow regime and season interac t to
30 influence occupancy. The potential benefits of steady flows to humpback chub vary based on the timing and size of discharge releases. During this study measured differences in nearshore temperatures between the fluctuating and steady flow during the 2010 field season were negligible (Ross and Grams 2011) suggesting that the hypothesis of differential warming of nearshore areas was not achieved given the implemented experimental design. A previous steady flow experiment implemented in the summer of 2 000 demonstrated that nearshore water temperatures do warm considerably if the discharge releases are low and the timing allows for maximum absorption of solar radiation (i.e. summer) (Trammel et al. 2002). T o better understand the interaction between G l en Canyon Dam operations and humpback chub populations f uture flow experiments should give careful consideration to the magnitude of discharge release d from G len C anyon D am, the spatial location where fish monitoring will occur and the timing of discharge releases In this study no strong relationships between habitat covariates and site occupancy were established. This suggests that given the parameters measured and modeled, no single variable was a strong predictor of whether a habitat will be occupied by humpback chub. Past research by Converse (et al. 1998) suggested that juvenile humpback chub select ed vegetated, talus and debris fan habitat attribute s while generally avoiding cobble, bedrock and sandy habitats attribute s Differences in observed h abitat relationships between this study and prior research may be a result of multiple factors A bundances of humpback chub have i ncreased substantially in recent years (Coggins et al. 2011). Given the high abundances of chub, habitat relationships may not be a pparent using occurrence data because most available nearshore habitats
31 within my study area were being used. I nference from this research is restricted to a 4.5 km section of the Colorado River that has the highest abundances of chub of any location in Grand Canyon Factors governing humpback chub occupancy may occur at a scale larger than was considered in this study and f uture research on humpback chub habitat use should increase the spatial extent of the study site to improve on the strength of i nference. In the future, the use of occupancy models as an analytical technique may be most useful when combined with a generalized linear modeling approach that estimates habitat covariate influence on humpback chub abundance This nested approach would allow occupancy models to determine the scale at which chub occurrence varies and the generalized linear model would identify factors that are related to fish density. Previous assessments of habitat use of humpback chub in Grand Canyon did not incorporat e an estimation of detectability ; therefore it is uncertain if the observed patterns of catch are in fact differences in habitat use or whether sampling was more effective in particular habitats (Va ldez and Ryal 1995, Converse et al 1998, Hubert and Fabr izio 2007). Detection probabilities in this study varied among humpback chub size category and survey suggesting size selectivity of the gear and temporal variation in humpback chub available for capture. Habitat characteristics including mean site depth and substrate heterogeneity were also observed to influence detection probability depending on the gear type used for sampling. Understanding how detection changes temporally, with fish size and site habitat characteristics will be useful for future and research and monitoring activities.
32 As humpback chub grow they undergo an ontogenetic habitat shift from using nearshore habitats as juveniles to large recirculating eddies as adults (Stone and Gorman 1999, Gerig 2012). Hoop net detectability was increas ed at intermediate depths 5 10 m from shore and most effectively sampled humpback chub size classes between 40 80 mm and 81 150 mm. Larger humpback chub having undergone their ontogenesis may move in and out of the effective sampling area of an individua l hoop net while occupying an area larger than an individual habitat site. Juvenile chub between 40 80 mm which reside in nearshore habitats exhibited higher detection probabilities when sampled using boat electrofishing and habitat sub units with increase d substrate heterogeneity had higher detection probabilities. Sites of increased substrate heterogeneity may offer increased interstitial space for small fish and may act as a refuge from discharge fluctuations and predators. In general, to maximize detec tion probability for humpback chub boat electrofishing should be considered for to sampl e fish under 80 mm while hoop net t ing is most effective for fish between 81 150 mm. Temporal variation in detection probabilities of humpback chub w ere observed betwe en sampling survey s within and between years (Figure s 1 4, 1 6, 1 7). Considering sampling survey as a covariate likely integrates important biotic and physical factors such as season, overriding turbidity and humpback chub abundance that can act in conc ert to influence detection probability. The large reduction in detection probability in October of 2009 may be related to turbidity conditions. Turbidity conditions during the October survey of 2009 were the lowest observed during the course of the study (mean=4 NTU). In response to low turbidity conditions, humpback
33 chub may reduce movement and seek physical cover in talus or debris fan shorelines, move to deep locations outside of the effective sampling area or emigrate from the sampling universe due t o a perceived increase in predation risk caused by clear water (Stone 2011). In contrast to Stone (2011), humpback chub in the mainstem Colorado River may not seek hoop net s as cover during periods of extremely low turbidity. In 2010 and 2011, detection probabilities increased consistently from July to September and plateaued or decreased from September to October. Detectability is influenced by increases or decreases in abundance (MacKenzie and Royale 2005). Timing of emigration of humpback chub from t he L ittle C olorado R iver varies through time and by fish size (Gorman and Stone 1999) and is likely influenced by timing of monsoonal floods which cause small to large flood pulses. Floods of sufficient size likely elicit an emigration event of j uvenile a nd newly recruited humpback chub rearing in the Little Colorado River to move to the mainstem Colorado River An active flood year in the LCR basin in 2010 and smaller events in 2011 may have been responsible for the increase in detection probability that was observed between survey s.
34 Table 2 1. Covariates thought to influence probability of occupancy and probability of detection of fish in Colorado River in Grand Canyon. Predictor Inference to Detection (p) Depth p varies with mean depth of site Substrate small p varies with proportion small subs. in site Substrate medium ies with proportion medium subs. in site P varies with proportion medium subs. in site Substrate large p varies with proportion large subs. in site Turbidity (NTU) p is influenced by turbi dity level Habitat type p varies with habitat type of site Hydraulic type p varies with hydraulic type Reach (EF) p varies spatially by reach for EF Survey surveys p varies temporally between surveys Year p varies with temporal variation in abundance Flow Regime p varies with dam operations
35 Table 2 2 Ranking of site occupancy models used to estimate detection probability of humpback chub captured in repeated hoop net surveys from 2009 2011 Model AIC AIC wt K a Survey +Depth+Depth) 20733 0 0.9998 18 + Survey +Depth) 20751 17.13 0.0002 17 Survey +Large) 20755 21.58 0 17 Survey +Medium) 20772 38.66 0 17 Survey +Habitat) 20773 39.84 0 19 Survey +Small) 20775 41.17 0 17 Survey +Hydr aulic Type ) 20778 44.4 0 17 Survey ) 20783 49.37 0 16 Survey +NTU) 20783 49.42 0 17 20988 254.09 0 11 21058 324.71 0 9 21087 353.63 0 8 21126 392.1 9 0 5 21202 468.54 0 3 a = # of parameters
36 Table 2 3 Ranking of site occupancy models used to estimate detection probability of humpback chub in repeated electrofishing surveys from 2009 2011. Model AIC AIC wt K a +Reach+ Survey +Medium) 9216.44 0 0.7837 18 Survey +Small) 9219.19 2.75 0.1981 18 Survey +Depth) 9225.49 9.05 0.0085 18 Survey +Habitat) 9226.01 9.57 0.0065 20 Survey +Depth+Dep th2) 9227.47 11.03 0.0032 19 Survey +Hydr aulic Type ) 9237.83 21.39 0 18 Survey +Large) 9247.31 30.87 0 18 Survey +NTU) 9248.94 32.5 0 18 Survey ) 9250.49 34.05 0 17 ize+Season) 9329.09 112.65 0 15 9339.02 122.58 0 12 9353.4 136.96 0 6 9431.67 215.23 0 7 9433.62 217.18 0 8 9434.05 217.61 0 4 (.), p(.) 9590.02 373.58 0 3 a = # of parameters
37 Table 2 4 Ranking of site occupancy models used to estimate occupancy probability of humpback chub in Grand Canyon. Model AIC AIC wt K a Size+Survey ), p(Best) 20703.45 0.00 0.5729 31 Si ze+Survey+ Hydr aulic Type ), p(Best) 20704.23 0.78 0.3879 32 Survey ), p(Best) 20708.88 5.43 0.0379 29 Size+Survey+ Small), ,p(Best) 20716.76 13.31 0.0013 26 Size+Survey+ Medium), p(Best) 20718.03 14.58 0.0007 26 2071 9.10 15.65 0.0002 26 Size+Survey+ 20720.99 17.54 0.0002 26 Size+Survey+ Depth), p(Best) 20721.09 17.64 0.0001 26 20723.71 20.26 0 23 20724.12 20.67 0 23 Best) 20728.38 24.93 0 21 20732.05 28.60 0 22 20733.44 29.99 0 18 20736.16 32.71 0 20 Size+Survey+ Habitat), p(Best)* Size+Survey+ Depth+Depth 2 ), p(Best)* a = # of parameters, model did not converge
38 Table 2 5 Ranking of site occupancy models used to estimate occupancy probability of humpback chub in Grand Canyon. Model AIC AIC wt K a 9160.19 0 0.2189 24 each), p(Best) 9160.5 0.31 0.1875 26 9161.35 1.16 0.1226 25 aulic Type ), p(Best) 9161.92 1.73 0.0922 25 9161.98 1.79 0.0894 25 Regime+Size+Small), p(Best) 9162.07 1.88 0.0855 25 9162.11 1.92 0.0838 27 9162.17 1.98 0.0813 25 2 ), p(Best) 9164.17 3.98 0.0299 26 9166.61 6.42 0.0088 21 9181.92 21.73 0 21 9185.77 25.58 0 19 Survey +Size), p(Best) 9201.95 41.76 0 30 9215.15 54.96 0 20 9216.44 56.25 0 18 egime), p(Best) 9216.84 56.65 0 21 9218.93 58.74 0 20 9219.66 59.47 0 23 Survey ), p(Best) 9228.25 68.06 0 29 a = # of parameters
39 Figure 2 1. Map of the study site in the Colorado River in Grand Canyon, AZ. Study site was between Heart Island to Lava Chuar Rapid. White dashed box represents sampling area.
40 Figure 2 2. Discharge patterns during field sampling in 2009.
41 Figure 2 3. Discharge patterns during field sampling in 2010.
42 Figure 2 4. Discharge patterns during field sampling in 2011
43 Figure 2 5. Aerial image from a GIS showing the typical arrangement of 50 m habitat sub units along shoreline habitats.
44 Figure 2 6 Patterns of detection probability for different size categories of humpback chub during repeated hoop net surveys in 2009 2011.
45 Figure 2 7 Relationship between probability of detection and mean depth of site during repeated hoop net surveys 2009 to 2011.
46 Figure 2 8 Patterns of detection probability for stratified by size category of HBC during repeated electrofishing surveys in 2009.
47 Figure 2 9 Patterns of detection probability for stratified by size category of HBC during repeated electrofishing surveys in 2010.
48 Figure 2 10 Patterns of detection probability for stratified by size category o f HBC during repeated electrofishing surveys in 2011.
49 Figure 2 11 Relationship between probability of detection and proportion medium substrate in a site during repeated electrofishing surveys 2009 to 2011
50 Figure 2 12 Relationship between probabi lity of detection and proportion small substrate in a site during repeated electrofishing surveys 2009 to 2011
51 Figure 2 13 Patterns of occupancy probability stratified by size class and trip during repeated hoop net surveys from 2009 to 2011.
52 Figure 2 1 4 Patterns of occupancy probability from repeat electrofishing surveys stratified by size class, flow regime and year during 2009 to 2011.
53 CHAPTER 3 HABITAT SELECTION AN D MOVEMENT OF HUMPBA CK CHUB Introduction Globally, many populations of fish in rivers are endangered as a result of anthropogenic alteration of habitat (Minns et al 1996; Rosenfeld 2003) and effective management of extant populations requires a clear understanding of the relationship between fish and their environment. In river ine systems it is essential to identify environmental or habitat characteristics that define the habitat requirements of river fishes This information is required to identify and screen management and restoration policies to preserve or recover populatio ns of individual species or restore in stream habitats and flows to a desired state. W idespread habitat loss and change caused by large dams has been implicated in the decline of native fish in the Colorado River Basin of the Western United States (Tyus 1 991). Large dams disrupt the natural flow regime by changing the timing, magnitude, frequency, duration and rate of change in hydrologic conditions within rivers (Poff et al. 1997). Changes in flow regime alter riverine processes including ecosystem produ ctivity, sediment transport and temperature which are thought to have negative effects on riverine ecosystems including fish populations. As an example, the installation of Glen Canyon Dam in 1963 significantly altered the hydrology of the Colorado River in Grand Canyon and has been implicated in the decline of the endangered humpb ack chub (Valdez and Ryal 1997; Stone and Gorman 1999). The humpback chub is a large, long lived cyprinid endemic to canyon reaches of the Colorado River Basin (Minckley and Mars h 2009). The humpback chub was one of the first species listed as federally endangered and became federally protected under
54 the Endangered Species Act of 1973. Currently, six isolated populations of humpback chub exist within the Colorado River basin of which the largest population occurs in Grand Canyon (Stone and Gorman 1999; Coggins et al. 2006). Historically, the humpback chub was distributed widely throughout Grand Canyon and recruited successfully in both the Little Colorado and mainstem Colorado R iver ( Valdez and Ryal 1995 ). Current populations of humpback chub in Grand Canyon are found in 9 aggregations, with the largest aggregation located near the confluence of the Little Colorado River and mainstem Colorado River ( Valdez and Ryal 1995 ). A larg e proportion of management and restoration efforts in Grand Canyon are directed at maintaining important river habitat (e.g. sandbars, backwaters) and mimicking natural flow patterns (e.g. artificial floods, seasonal steady flows) to positively benefit nat ive fish, including humpback chub While significant effort has been allocated to the monitoring of physical resources and native fish populations there has been little synthesis of the influence of habitat and flow regime on humpback chub ( but see, Valde z and Ryal 1995, Valdez and Ryal 1997; Converse et al. 1998) in the Colorado River in Grand Canyon. Early investigations of humpback chub populations have primarily focused on inferring habitat relationships from catch per effort indices and radio telemet ry ( Kaeding et. al. 1990; Valdez and Ryel 199 7; Converse et al. 1998). However, no studies to date have quantified humpback chub habitat use in relation to available habitat to determine if and how strongly humpback chub select particular habitats. The ob jectives of this study were to determine the habitat selection and movement patterns of telemetered humpback chub in the Colorado River within Grand Canyon
55 during contrasting flow regimes as part of an adaptive management experiment. Specifically, this st udy was designed to (1) determine if humpback chub select habitat features, (2) assess the influence of an experimental steady flow regime on habitat selection and movement patterns (3) evaluate the influence of overriding turbidity conditions to spatial d istribution and habitat use and (4) provide insight into potential habitat requirements of humpback chub. This research will aid managers in evaluating the importance of habitat features to humpback chub population persistence in Grand Canyon and in infor ming Glen Canyon Dam flow operations and in designing future flow experiments Methods Study Site The Little Colorado River reach of the mainstem Colorado River in Grand Canyon, AZ was extensively sampled during 4 river trips in July, August, September and October of 2010 ( Figure 3 1). July and August river trips occurred during the mean low fluctuating flow regime, and September and October river trips were sampled during an experimental steady flow regime. Current water releases from G len C anyon D am are tightly regulated and are typically operated under a hydropeaking flow schedule. The mean low fluctuating flow regime in this study was characterized by diel discharges ranging from 250 (m 3 /sec) to 500 (m 3 /sec) (Figure 3 2). During the experimental stead y flow regime discharge from GCD was constant at 250 (m 3 /sec) (Figure 3 2). Extensive tributary flooding during August and September caused spikes in the hydrograph that were unrelated to dam operations (Figure 3 2). Daily nephelometric turbidity units ( NTU, hereafter) were continuously monitored in 15 minute intervals remotely from an acoustic doppler profiler at USGS gauge 09383100 (R K M 165 ). The Colorado River
56 Flow Sediment and Stage model was used to back calculate the difference in average water trav el time between USGS gauge 0983100 and the study location (RM 125 131.2 ) using an unsteady flow model (Wiele and Griffin 1997). Spikes in turbidity are the result of monsoonal rains which caused tributary flooding (Figure 3 3). Fish handling and surgery m ethods Thirty two humpback chub between 180 2 45 mm were surgically implanted with uniquely coded PT 3 Sonotronics acoustic telemetry tags (Table 3 1) (PT 3: 8 mm diameter, 19 mm length, 1.5 grams, 45 60 day battery life; Sonotronics, Tuscon, AZ, USA). Hu mpback chub tagged with telemetry tags were captured via hoop net ting. Given the size and body shape of humpback chub and guidelines on bodysize tag size relationships from the literature, I chose 180 mm TL as the smallest size to implant humpback chub wit h an acoustic tag to minimize influence of tagging on survival of fish ( Jepsen et al. 2002 ) All fish selected for surgery were anesthetized with sodium bicarbonate until equilibrium was lost and then non anesthetic water was pumped across the gill membra nes with a gravity pump. The tags were quickly implanted into the fish by making a small incision into the abdominal cavity and following insertion the incision was closed with 2 3 absorbable sutures and cyanoacrylate. Each tagged fish was observed for 2 hours until the immediate effects of anesthesia and surgery were no longer visible and the fish was able to remain upright with normal fin movements. Following recovery, fish were released into the same location they were collected A staggered entry ta gging design was used to increase the temporal resolution of habitat selection and movement for telemetered humpback ch ub (Pollock et al. 1989). All fish were implanted with tags between July September. No fis h were implanted during the October trip as vi able tags were still active from prior trips during October.
57 Tracking of telemetered humpback chub Tracking of sonic tagged fish began 24 h our s post surgery and continued through October or tag failure; whichever occurred first. Relocations recorded within 3 days of surgery were censored from all analyses to allow for a recovery period following tag implantation Telemetered fish were t racked twice daily. Generally, tracking event one occurred during the morning between 0500 and 0700 h ou rs while tracking event two commenced in the afternoon between 1400 and 1600 h ou rs. Tracking began at the upstream boundary of the study reach and ended at t he bottom of the study reach (~ RM 63.2 65.6). Tracking was restricted to this area because of operational constrain ts on our research from permitting agencies. Systematic sampling of telemetered individuals occurred by stopping at f ixed locations approximately 100 m apart. Areas in swift water were floated multiple times to scan the area completely. Telemetry survey s were abandoned in rapids and turbulent areas for safety reasons and signal atten uation. As a telemetered fish position was approached, the gain on the telemetry unit was reduced until the telemetry signal was strong in all directions (omnidirectional). Fish location s w ere entered into a mobile GIS database and attributes of the location including time, depth, distance from shore, habitat type and hydraulic type were recorded. Habitat mapping Relocations of telemetered h umpback chub during each tracking s ession were related to data layers of hydraulic type, habitat type, depth and substrate category u sing both existing information collected through the USGS Grand Canyon Monitoring and Research Center (GCMRC) physical science program and field measurements. The hydraulic type was mapped in the field and split into two discrete categories including eddies and runs. Eddies refer to areas of recirculating flow that generally occur
58 downstream of channel constricting debris fans. Runs represent any portion of th e river where the water was traveling downstream. Field measured hydraulic type was digitized in GIS for the entire study site and clipped to the river edge Habitat type was delineated using existing GIS imagery and habitat classifications were verified i n field. Habitat types were split into six discrete habitat types including backwater, cliff, debris fan, offshore, talus slope and sandbar by applying a 15 m buffer to wetted river perimeter in GIS Backwater habitats are characterized as areas of sand d eposition that are separated from the main river channel. Cliff habitats were characterized as shear walls rising vertically and laterally over the river. Debris fans were characterized as shorelines of large cobble and boulder that were transported into the river corridor by tributaries during flooding. Offshore habitat represented any area that was greater than 15 m from shore. Sandbar habitats were classified as shorelines of contiguous beach and exposed sand. Talus habitats were classified as shore lines consisting mainly of boulders deposited by rockslides and rockfall. Depth was determined in GIS as surface water elevation minus the elevation of the river bottom for a given discharge. Existing bathymetric and channel coarseness maps were used to classify substrate type into 12 discrete categories using a modified Wentworth scale. Substrates c ategories were grouped into three classes: small, medium and large substrates. Substrates classified 1 4 represented small substrates ranging from silt to co urse sand. Substrates classified 5 8 represented small gravel to medium cobble while substrates classified 9 12 represent large substrates including large cobble and boulder. Habitat availability was determined by generating random locations within the s tudy site using the Arc GIS
59 habitat availability was equal to the number of fish relocations during each flow regime. Habitat characteristics were related to available location s in A rc GIS. Statistical Analysis Habitat Selection Following Rosenfeld (2003 ), habitat selection was inferred as differential use of a habitat given the availability of the habitat. To assess habitat selection, I conducted my statistical analysis in tw o parts. First, univariate selection ratios were constructed to compare individual humpback chub relocation points to available locations within th e study area ( Thomas and Taylor 1990 ; Manly et al. 2002; Rodgers and White 2007 ). Secondly, I developed a re source selection function to investigate the influence of habitat characteristics on the relative probability of habitat use by humpback chub with a mixed effects logistic regression. These two approaches provide a complimentary approach to studying habit at selection. Univariate selection ratios allow for selection to be determined for specific habitat classes while the resource selection function allows a probability of use to be modeled as a function of habitat characteristics. Separate selection ratio s were constructed to evaluate if selection differed between mean low fluctuating flow and the experimental steady flow regime. A single resources selection function was developed that allowed for the influence of the fluctuating and steady flow regime to be evaluated. Univariate selection ratios were calculated following a type II study design (Thomas and Taylor 1990 ; Manly et al. 2002). A type II study design compared the frequency of habitat use in a discrete habitat class to the availability of the ha bitat class of interest within the study reach. Habitat classes included hydraulic type, habitat type, depth class and substrate class Prior to constructing selection ratios, two different chi square tests were used to test if humpback chub were uniform ly distributed across
60 habitat classes and determine if habitat selection was occurring. Once selection w as established, selection ratio ( Wi ) intervals were constructed to determine which habitats were being s elected for (Manly et al. 2002; Rogers and Whit e 2007). Selection was indicated by values greater than one while avoidance of habitats is indicated by values less than one (Rogers and White 2007). Selection ratios of one indicated that telemetered chub are not selecting for a particular habitat type ( i.e. use is proportional to available habitat). I calculated Bonferroni 95% confidence intervals to determine if the selection ratio indicated positive, negative or neutral selection (Manly et al. 2002; Rogers and White 2007). Log like li hood chi square t est and selection ratios were calculated using the function wides II in package adehabitat in Program R. Mixed effects logistic regression models were implemented to compare used versus available locations as functions of habitat attributes for telemetered humpback chub. Nine different a priori models were developed to propose mathematical descriptions of covariates that influence habitat selection. In this analysis, the telemetered fish was considered a random effect while hydraulic type, habitat type, de pth, substrate category and flow treatment were fixed effects (Rogers and White 2007). I choose a mixed model procedure because it accounts for non normal distribution of residuals autocorrelation associated with repeatedly sampling the same fish and unev en sample sizes (Gillies et al. 2006) I fit combinations of covariates and compared model fit using AIC (Burnham and Anderson 2002). The model with the lowest AIC was chosen for interpretation of habitat selection Mixed effects logistic regression were implemented using the GLMER function from the lme4 library in program R.
61 Statistical Analysis of Movement To describe the movement patterns of humpback chub mean daily displacement and extent of movement were calculated for the mean low fluctuating flow regime and the experimental steady flow regime. I calculated mean daily displacement as the linear distance moved between successive relocations during daily morning tracking events. A mixed model analysis of variance was used to determine if mean daily displacement varied as a function of flow regime. An i ndividual telemetered humpback chub was considered the random effect while flow regime was the fixed effect (Rodgers and White square statistic was used to assess model significance and a Monte Carlo procedure was used to determine 95% confidence intervals around mean daily displacement. Extent of movement for humpback chub was calculated as the linear distance between the most upstream relocation to the most downstream relocation. I used a one way Kruskal Wallis test to determine if there were differences in extent of movement between flow regimes and a Monte Carlo procedure to determine 95% confidence intervals around extent of movement. To assess the influence of changing dischar ge during the fluctuating flow regime on the spatial distribution of telemetered humpback chub a linear mixed model was used to compare distan ce to shore to discharge level. To determine the influence of turbidity on spatial distribution I used a linear m ixed model to determine if distance from shore varied as a function of turbidity category and a Monte Carlo procedure to determine 95% confidence intervals around distance from shore. Three turbidity categories were considered. Low turbidity ranged from 0 30 NTU, medium turbidity levels ranged from 31 300 NTU and high turbidity levels ranged
62 from 301 10,000 NTU. To evaluate if frequency of habitat use changed with increased turbidity a Pierson chi square test was used. Results I recorded 1034 locations of 32 humpback chub implanted with ultrasonic transmitters during four two week sampling trips from July to October 2010. Tagged humpback chub were relocated 344 times during the fluctuating flow regime and 690 times during the experimental steady flow reg ime, respectively. The mean number of relocations for tagged humpback chub was 25 (SD=12.7) and the mean number of days a telemetered fish was observed was 46 (SD=16.2). The mean size of telemetered chub was 199 mm (SD=19.2) and ranged between 180 mm to 2 45 mm (Table 3 1) During the study two telemetered humpback chub likely suffered mortality or tag loss indicated by a lack of movement with in and between sampling trips. Data from these fish w as censored from analysis The following data represents fish k nown to have stayed within the sampling area as we were not permitted to search for fish outside of our defined sampling reach Only fish with a minimum of 15 relocations were included in subsequent analysis. Selection Ratios Habitat selection was assess ed for 14 humpback chub during the fluctuating flow regime and for 25 humpback chub during the steady flow experiment. During both flow regimes locations used by humpback chub differed significantly (P value <0.05) from availability for all habitat charac teristics except for substrate c ategory during the fluctuating flow regime (Table 3 2 ). Hydraulic type, habitat type and depth class exhibited the strongest patterns of selection (Figure 3 4 ). Humpback chub strongly selected eddy hydraulic types and were used in proportions greater than three times
63 their availability. Runs were selected against in both flow regimes. Habitat types including c liff and debris fan shorelines were positively selected for during both flow regimes. Backwater habitat selection w as highly uncertain during the fluctuating flow regime but strongly selected for during the steady flow experiment. Talus slopes were neutrally selected for during the fluctuating flow regime while being positively selected for during the steady flow regim e. Offshore and sandbar habitats were selected against during both flow regime s Telemetered humpback selected intermediate depths of 4 6 m while generally avoiding shallow (<2 m) and deep areas (> 10 m) during both flow regimes. Tagged fish neutrally s elected substrate class during both flows with the exception of positive selection of the medium sized substrate class during the steady flow regime. Resource Selection Function The top resource selection function, ranked by AIC indicated that a combi nati on of hydraulic type, depth, substrate category and an interaction between hydraulic type and substrate category best explained the probability of habitat selection by humpback chub (Table 3 3). H ydraulic type was the strongest predicator of habitat select ion causing the largest improvement in model fit (Table 3 3) Similar to the selection ratio analysis, eddy hydraulic types were used in much greater proportion to their availability and runs were selected against (Table 3 4). Telemetered chub used depths shallower than would be expected given the distribution of depths and selected s ubstrate categories larger than would be expected given substrate category availability There was a significant interaction effect between substrate category and hydraulic t ype. This interaction indicated that chub use d a wide variety of substrate categories in eddies while being negatively associated with small substrates and positively associated with larger
64 substrates in runs. Models that included a coefficient represent ing the flow experiment did not improve model significantly suggesting that habitat selection patterns did not change between the fluctuating and steady flow regime (Table 3 3). The top mode l effectively distinguished between used and available habitats (F ig 3 5 ). Movement Mean daily displacement observed for telemetered humpback chub was not significantly different between the fluctuating and experimental steady flow regime 2 =0.16, P=0.6867). During the fluctuating flow regime, tagged humpback chub mov ed a mean distance of 92 m per day (95% confidence interval, 67 123 m; Figure 3 6 ) while in the steady flow regime telemetered chub moved a mean distance of 106 m per day (95% confidence interval 91 123 m; Figure 3 6 ). In both flow regimes, long distance daily movements (> 500 m) were rare. Significant differences in extent of movement were not observed between the fluctuating and steady flow regime for telemetered 2 =0.14, P=0.7015). The mean extent of tagged fish during fluctuating flow was 414 m (95% confidence interval, 258 595 m; Figure 3 4 ) and during the steady flow experiment mean extent of movement was 515m (95% confidence interval, 333 743 m; Figure 3 4 ). No relationship was found between humpback chub distance from 2 =1.56, P=0.20) suggesti ng that humpback chub do not alter spatial distribution in response to changing discharge under the discharge levels observed. A significant relationship ( 2 =77 .869, P<0.01) was found between distance from shore and turbidity level. During low turbidity c onditions humpback chub were located further offshore than during high turbidity conditions (Figure 3 7). Similarly, during this study the frequency of nearshore habitat use for all nearshore habitat types increased
65 with increased turbidity levels 2 =63. 693, df=10, P<0.01) This pattern was strongest in backwater, cliff and debris fan habitat types (Table 3 4). Discussion T elemetered humpback chub exhibited strong selection of eddy hydraulic types in both the fluctuating and experimental steady flow reg ime. Nearshore habitats including debris fan and cliff faces located within eddies were also positively selected during both flow regimes. Prior research has demo nstrated that humpback chub use eddy complexes extensively and that the distribution of hump back chub in Grand Canyon maybe related in part to the presence of large eddy complexes (Valdez and Ryal 1995). Kaeding et al. (1990) further postulated that humpback chub rely on large channel obstructions such as debris fans to create velocity refuges v ia eddy complexes. In Grand Canyon, eddies generally occur below rapids caused by channel constricting debris fans. Eddies create areas of recirculating flow adjacent to the main current and increase material retention time as a function of slower average wa ter velocities when compared to runs (Schmidt 1990). I hypothesize that the mechanism causing strong selection of eddies by humpback chub is a result of food availability and reduced energetic expenditure. Residing in depositional environments adjacen t to turbulent main channel flow may allow chub to maintain position in lower velocity water while making foraging attempts at entrained organic matter and invertebrates in the water column. Similar patterns have been observed for cyprinids occurring in s mall stream systems in the s outheastern and s outhwestern United States (Rinne 1991, Freeman and Grossman 1993). Humpback chub during this study made small daily movements and exhibited a restricted distribution. Most movements of tagged humpback chub were made within
66 large eddy complexes. However, movement between eddy complexes and from runs to eddies were observed. Previous studies over both short and long temporal scales agree with my findings suggesting that humpback chub exhibit strong patterns of sp atial fidelity (Valdez & Ryal 199 5 Paukert et al. 2006). Seasonal movements of humpback chub are characterized by potadromous migrations between the mainstem Colorado River and the Little Colorado River between March and May (Gorman and Stone 1999, Coggi ns et al. 2006). This migration was not able to be documented because it occurred outside of my research trips. Long distance movements in excess of 50 km by humpback chub have been observed through long term tagging studies (Paukert et al. 2006). Howeve r, given the low instances of immigration in this study it appears that long distance dispersal maybe relatively rare. Further research assessing movement in a meta population framework should be conducted to determine rates of movement, frequency of long distance movements and the recruitment contribution of humpback chub occupying aggregations outside of the LCR confluence S trong selection of habitat and restricted movement patterns suggest that the presence of eddy hydraulic types may be a habitat requ irement of humpback chub in the Colorado River in Grand Canyon. Humpback chub that occupy the mainstem Colorado River in close proximity to the LCR may be able to meet their requirements for foraging and resting within this limited spatial extent. Futur e research should focus on empirically measuring food availability in different habitat types and consider bioenerg etic models which quantify the energy expenditure of occupying various habitats for humpback chub. Existing long term catch data which inclu des the entire canyon corridor should be evaluated to determine if humpback chub strongly select
67 eddies in downstream areas. If this pattern is exhibited canyon wide, restoration efforts which seek to increase mainstem humpback chub abundances should pote ntially consider canyon stretches with high eddy availability. Humpback chub exhibited similar patterns of habitat selection and movement between the fluctuating flow and experimental steady flow regime. Fluctuations in discharge can influence the quantity and quality of available habitat for riverine fish which may influence patterns of habitat use and movement (Kraft 1972, Moog 1993, Dare et al. 2002) The majority of studies examining the effects of flow on fish have been conducted on stream and river po pulations of salmonids (Quinn and Kwak 2000, Dare et al. 2002, Scruton et. al 2006 Korman and Campana 2009 ) and suggest that during periods of high discharge fish occupy deeper areas and are associated with velocity refugia. H umpback chub monitored in th is study exhibited habitat selection and movement patterns that appeared robust to discharge fluctuations. This study was conducted in a locat ion comprised primarily of large eddy complexes adjacent to complex shoreline habitat These habitats are relative ly invariant to m inor discharge fluctuations allowing telemetered humpback chub habitat selection patterns to be robust to changes in flow regime ( Korman et al. 2004 ). Similar to Gerig (2012), I hypothesize that the magnitude of change in discharge betwee n the fluctuating and experimental steady flow regime was not great enough to illict a detectable shift in habitat selection and movement patterns of humpback chub. Overriding turbidity conditions appeared to have a stronger influence on the spatial distri bution of humpback chub than the experimental steady flow regime. The proportion of relocations in nearshore habitats increased concomitantly with turbidity levels and the
68 spatial distribution of humpback chub shifted closer to shore. In particular, backw ater habitats were never used when turbidity conditions were less than 30. However, as turbidity increased, backwater habitats were used over 50 times. These patterns suggest that NTU levels may shift habitat use patterns of humpback chub which ultimately influence habitat selection. The shift in humpback chub spatial distribution maybe related to foraging behavior. Under low turbidity conditions humpback chub forage offshore on material entrained in eddy currents. As turbidity increases humpback chub u se nearshore habitats and may become actively or opportunistically piscivorous (Valdez and Ryal 1995; Stone and Gorman 2006). Backwater seine surveys and short term tethering experiments which occurred concurrently to the telemetry study found that the fr equency of large fish, including humpback chub captured in backwaters increased with increased turbidity and that under turbid conditions predation risk was observed to be highest (Dodrill 2012). Additionally, on two occasions during tethering experiments humpback chub were observed to be foraging on tethered fish (M. Dodrill, Personnel Communication). Prior research suggests that the abundance of small bodied fish increased with turbidity as a result of dispersal from tributaries during floods (Yard et a l. 2011). Correspondingly, these results suggest that humpback chub shift their habitat use patterns during turbidity fluxes to potentially forage on fish in nearshor e habitats such as backwaters. Future research is needed to quantify diet plasticity in humpback chub and determine the extent to which they are piscivorous in the mainstem Colorado River.
69 Table 3 1. Biological attributes collected from humpback chub implanted with acoustic telemetry transmitters. Trip Length (mm) Weight (g) ID Date Tagged L ast Contact Tracking Duration (d) 1 185 52.73 711050 7/12/2010 9/15/2010 64 1 198 65.26 731070 7/12/2010 9/15/2010 61 1 188 55.46 751090 7/12/2010 8/12/2010 30 1 245 127.39 71990 7/15/2010 8/12/2010 59 1 187 54.54 73950 7/16/2010 8/18/2010 27 1 208 7 6.18 751030 7/16/2010 8/12/2010 26 1 180 48.38 71930 7/17/2010 9/15/2010 54 1 180 48.38 71910 7/20/2010 9/15/2010 57 1 256 146.22 75970 7/20/2010 9/15/2010 55 1 189 56.39 73930 7/21/2010 8/18/2010 52 1 187 54.54 75950 7/21/2010 9/15/2010 55 2 226 98. 87 721060 8/7/2010 9/14/2010 36 2 201 68.42 70980 8/8/2010 9/15/2010 38 2 183 50.96 701040 8/8/2010 9/14/2010 36 2 184 51.84 721000 8/8/2010 10/21/2010 37 2 196 63.22 741020 8/9/2010 10/19/2010 70 2 220 90.86 741080 8/9/2010 9/15/2010 36 2 182 50.09 761040 8/9/2010 10/22/2010 73 2 201 68.42 761100 8/9/2010 9/15/2010 36 2 185 52.73 771050 8/9/2010 10/24/2010 75 2 217 87.02 771110 8/9/2010 9/15/2010 64 3 210 78.51 70900 9/4/2010 10/27/2010 59 3 205 72.79 70920 9/4/2010 9/15/2010 38 3 187 54.54 749 40 9/4/2010 10/27/2010 52 3 215 84.53 74960 9/4/2010 10/27/2010 52 3 185 52.73 76960 9/4/2010 10/27/2010 52 3 185 52.73 76980 9/4/2010 8/12/2010 26 3 211 79.69 77970 9/4/2010 10/27/2010 52 3 210 78.51 72920 9/5/2010 9/15/2010 11 3 205 72.79 72940 9/5 /2010 10/27/2010 51 3 183 50.96 77990 9/5/2010 10/27/2010 51 3 187 54.54 701060 9/5/2010 10/27/2010 51
70 Table 3 2. Chi square statistics testing (1) the distribution and (2) selection of habitat characteristics of humpback chub during contrast ing flow regimes. Distribution Selection Habitat Characteristic Flow Regime df P df P Habitat Type Fluctuating 83.87 60 0.02 402.85 65 <0.001 Steady 164.07 105 <0.001 573.18 110 <0.001 Hydraulic Type Fluctuating 30.32 12 <0.001 539.05 13 <0.001 Steady 32.4 21 0.05 833.27 22 <0.001 Depth Class Fluctuating 190.81 6 0 <0.001 242.56 65 <0.001 Steady 220.08 105 <0.001 595.58 110 <0.001 Substrate Class Fluctuating 26.2 24 0.32 29.34 26 0.26 Steady 64.27 42 0.015 121.59 44 <0.001
71 Table 3 3. AIC ranking of mixed model logistic regression models use d to determine habitat selection of telemetered humpback chub. Model LL a AIC K b Depth+Substrate+Hydr aulic +(Substrate*Hydr aulic Type ) 1073 2157 0 5 Depth+Substrate+Hydr aulic Type+Flow Regime +(Substrate*Hydro) 1072 2159 2 5 Depth+Substrate+Hydr aulic Type +(Depth*Hydr aulic Type ) 1076 2163 6 5 Depth+Substrate+Hydr aulic Type + Flow Regime +(Depth*Hydr aulic Type ) 1074 2165 8 6 Depth+Depth+Substrate+Hydr aulic Type 1080 2171 14 5 Depth+Substrate+Hydr aulic Type 1081 2172 15 4 Hydr aulic Type 1111 2228 71 2 Habitat Type 1296 2607 450 6 Depth+Substrate 1364 2736 579 3 a = L og Likelihood; b = # of parameters
72 Table 3 4 Coefficient estimates from best AIC ranked mixed model logistic regression for telemetered humpback chub during contrasting flow regimes Parameter Estimate SE z value P Intercept 2.05 0.19 10.67 <0.01 Depth 0.07 0.015 4.68 <0.01 Substrate 0.37 0.05 6.93 <0.01 Hydr aulic Type 3.2 0.23 13.62 <0.01 Substrate Hydr aulic Type 0.27 0.07 4.1 <0.01
73 Figure 3 1. Map of the study site in the Colorado River in Grand Canyon, AZ. Study site was betwe en Heart Island to Lava Chuar Rapid White dashed box represents sampling area.
74 Figure 3 2. Mean, maximum and minimum daily discharge during the fluctuating flow regime and the experimental steady flow regime.
75 Figure 3 3. Mean daily turb idity measurements from July to October 2010 in the Colorado River in Grand Canyon.
76 Figure 3 4. Selection Ratios with 95% Bonferroni confidence intervals of (A.) habitat type (B.) hydraulic type,(C.) depth category (m) and (D.) substrate category.
77 Figure 3 5. Kernel density plot of fitted values comparing used versus available locations for the top resource selection function.
78 Figure 3 6. Movement of humpback chub during contrasting flow regimes. (A.) Daily displacement and (B.) spatial extent of telemetered humpback chub during the fluctuating and experimental steady flow regime.
79 Figure 3 7. Observed distance from shore measurements of telemetered humpback chub during low, medium and high turbidity levels.
80 CHAPTER 4 SY NTHESIS AND CONCLUSI ONS This research study documented that the steady flow experiment from G len C anyon D am did not have the anticipated result. Occupancy probabilities were observed to be consistently high and there was no marked change in occupancy stat e between the fluctuating flow regime and the steady flow regime s Similarly, analysis of humpback chub habitat selection and movement suggests that the steady flow regimes did not significantly influence their spatial distribution and movement rates. The lac k of significant response to the flow experiment is important. P revious researchers have noted that it is difficult to derive predictable relationships between hydrologic change and fish response and that fish responses to management actions are often counterintuitive (Pine et al. 2009, Poff and Zimmerman 2010, Branford et al. 2011). The use of the natural flow paradigm to structure how management proceeds in regulated river systems is appealing and this paradigm suggests that as flow regime is increm entally restored incremental changes in ecosystem function will be observed (Poff et al. 1997). However, my observations suggest small changes in flow regime independent of other physical conditions such as temperature did not alter how humpback chub use a nd select habitat. not an important variable in the structuring of lotic environments but points to the need integrate flow dependent variables that are strongly influenced by flow regulation such as turbidity and temperature. In regulated systems where physical processes such as turbidity and thermal regime have been decoupled from the flow regime, small changes for biota because flow, tur bidity and
81 tempera ture do not operate independently. When considering optimal flow policies for fish in regulated rivers it is also important to consider the dynamic suite of factors including temperature, turbidity, food availab ility, competition and predation that intera ct with flow to shape patterns observed in field. Using G len Canyon Dam as a management tool has proved a powerful tool in assessing ecosystem functioning of aquatic resources in Grand Canyon. To maximize learning, careful planning should take place wh en flow experiments are designed. Specifically, it is paramount to consider the seasonal timing and magnitude of the flows when anticipating potential outcomes to the experiment. If this research had been conducted under flows with greater contrast and t he steady flow experiments would have taken place in summer results may have been significantly different. To help evaluate policy actions which assess the relationship between humpback chub and G len Canyon Dam operations research monitoring should contin ue to utilize an active adaptive management approach.
82 LIST OF REFERENCES Branford, M.J., P.S. Higgins, J. Korman, and J. Sneep. 2011. Test of an environmental flow release in a British Columbia River: does more water mean more fish? Freshwater B iology 1 16. Bunn, S.E., and A.H. Arthington. 2002. Basic principles and ecological consequences of altered flow regimes for aquatic biodiversity. Environmental Management 30:492 507. Burnham and Anderson. 2002. Model selection and multimodel inference: A practical information theoretic approach. 2 nd edition. Springer publishing. Butler, R.G. 1980. Population size, social behavior and dispersal in house mice: A quantitative investig ation. Animal Behavior 28:78 85. Carpenter, S.R. 1989. Replication and treatment strength in whole lake experiments. Ecology 70:453 463. Coggins, L.W., W.E. Pine III, C.J. Walters, D.R. van Haverbeke and D. Ward. 2006. Abundance trends and status of the Little Colorado River population of humpback chub. North American Journa l of Fisheries Management 26: 233 245. Coggins, L.W., M.D. Yard, and W.E. Pine III. 2011. Non native fish control in the Colorado River in Grand Canyon, AZ: 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 River through Grand Canyon: spatial variability and implications of flow regulation. River Research and Applications 14 : 267 284 Dare, M. R., W.A. Hubert and K.G. Gerow. 2002. Changes in habitat availability and habitat use and movements by two trout species in response to declining discharge in a regulated river during winter. North American Journal of Fisheries Management 22: 917 928. Donovon, T.M., and J. Hines. 2007. Exercises in occupancy modeling and estimation. http://www.uvm.edu/envnr/vtcfwru/spreadsheets/occupancy/occupancy/htm. Dynesius, M. and C. Nilsson. 1994. Fragmentation and Flow Regulation of River Systems in t he North ern Third of the World. Science 266:753 766.
83 Fausch, K.D., C.E. Torgensen, C.V. Baxter and H.W. Li. 2002. Landscapes to riverscapes: bridging the gap between research and conservati on of stream fishes. Bioscience 52: 483 498. Freeman, M.C., and G.D. Grossman. 1993. Effects of habitat availability on dispersion of a stream cyprinid. Environmental Biology of Fishes 37:121 130. G illies, C.S. M. Hebblewhite S.E. Nielsen M.A. Krawchuk C.L. Aldridge, C.L. J.L. Frair, D.J. Saher, C.E. Stevens, an d C.L. Jerde 2006 Application of random effects to the study of resource selection by animals Journal of Animal Ecology 75 : 887 898 Gloss, S.P., and L.G. Coggins. 2005. Fishes of Grand Canyon. Pages: 34 56 in Editors: S.P. Gloss, J.E. Lovich and T.S. M elis, The state of the Colorado River ecosystem in Grand Canyon. USGS circular 1282. Gunn, J.M. and R. Sein. 2000. Effects of forestry roads on reproductive habitat and exploitation of lake trout ( Salvelinus namaycush ) in three experimental lakes. Canad ian Journal of Fisheries and Aquatic Sciences 57: 97 104. Hines, J.E. 2011. Program Presence 3.1. www.mbrpwrc.usgs.gov/software/presence.html Hubert, W. and M. C. Fabrizio 2007. Relativ e abundance and catch/effort relationships. Chapter 7 in Analysis and Interpretation of Freshwater Fisheries Data. Edited by M. Brown and C. Guy. American Fisheries Society. Jepsen, N., A. Koed, E.B. Thorstad, and E. Baras. 2002. Surgical implantation of telemetry transmitters in fish: how much h ave we learned? Hydrobiologia 483: 239 248. Johnson, D.H. 1980. The comparison of usage and availability measurements for evaluatin g resource preference. Ecology 61: 63 71. Kaeding, L.R., B.D. Burdick, P.A. Schr ader and C.W. McAda. 1990. Temporal and spatial relations between the spawning of humpback chub and roundtail chub in the upper Colorado River. Transactions of the American Fisheries Society 119:135 144. Kennedy, T.A., and S.P. Gloss. 2005. Aquatic ecolog y: the role of organic matter and invertebrates. Pages: 88 101 in Editors: S.P. Gloss, J.E. Lovich and T.S. Melis, The state of the Colorado River ecosystem in Grand Canyon. USGS circular 1282.
84 Korman, J. and S. E. Campana. 2009. Effects of hydropeaking o n nearshore habitat use and growth of age 0 rainbow trout in a large regulated river. Transactions of the American Fisheries Society 138: 76 87 Korman, J., S. M. Wiele and Margaret Torizzo. 2004. Modelling effects of discharge on habitat quality and dispe rsal of juvenile humpback chub (Gila cypha) in the Colorado River, Grand Canyon. R iver Research and Applications 20: 379 400. Korman, J., M. Yard, C.J. Walters and L.G. Coggins. 2009. Effects of fish size, habitat, flow and density on capture probabilitie s of age 0 rainbow trout estimated from electrofishing at discrete sites in a large river. Transactions of the American Fisheries Society 135: 58 75. Kraft, M.E. 1972. Effects of controlled flow reduction on a trout stream. Journal Fisheries Research Board of Canada. 29:1405 1411. Kwak, T.J. 1988. Lateral movement and use of floodplain habitat by fishes in the Kankakee River, Illino is. American Midland Naturalist 120:241 249. MacKenzie D.I., J.D. Nichols, G.B. Lach man, S. Droege, J. A. Royale and C.A. Langtimm. 2002. Estimating site occupancy rates when detection probabilitie s are less than one. Ecology 83: 2248 2255 MacKenzie D.I., J.D. Nichols, J.A. Royale, K.H. Pollack, L.L. Bailey, J.E. Hines. 2006. Occupanc y estimation and modeling: inferring patterns and dynamics of species occurrence. Elsevier, Burlington, Vermont. MacKenzie D.I. and J. A. Royale. 2005. Designing efficient occupancy studies: general advice and tips on allocation of survey effort. Journal of Applied Ecology 83: 1105 1114. Manly, B.F.G. 2002. Resource selection by animals: statistical design and analysis for field studies. Kluwer Academic Publishing. Dordrecht, The Netherlands. Melis, T.S., S.J.D. Martell, L.G. Coggins, W.E. Pine III and M. E. Andersen. 2006. Adaptive management of the Colorado River ecosystem below Glen Canyon Dam, Arizona: using science and modeling to resolve uncertainty in river management. Proceedings from Adaptive management of water resources: AWRA summer specialty conference. Minckley, W.L. and P.C. Marsh. 2009. Inland fishes of the greater southwest: chronicle of a vanishing biota. University of Arizona Press. Tucson, Arizona. Minns, C.K, S.E. Doka, C.N. Bakelaar, P.C.E. Brunette, and W.M. Schertzer. 1999. Inden tifying habitats essential for Pike Esox lucius L. in the Long Point Region of Lake Erie: a suitable supply approach. Pages 363 382 in L. Benaka, editor. Fish
85 habitat: essential fish habitat and rehabilitation. American Fisheries Society, Symposium 22, Bet hesda, Maryland. Minns, C.K., J.R.M. Kelso, and R.G. Randall. 1996. Detecting the response of fish to habitat alterations in freshwater ecosystems. Canadian Journal of Fisheries and Aquatic Sciences 53: 403 414. Moog, O. 1993. Quantification of daily pea k hydropower effects on aquatic fauna and management to minimize environmental impacts. River Resea rch and Applications 8: 5 14. Morrison, M.L. and B.G. Marcot. 2006. Wildlife habitat relationships: concepts and applications. 3 rd Edition, Island Press, Wa shington, D.C. Nilsson, C., C.A. Reidy, M. Dynesius, and C. Revenga. 2005. Fragmentation and flow Science 308: 405 408. Quinn, J.W. and T.J. Kwak. 2000. Use of rehabilitated habitat by brown trout and rainb ow trout in an Ozark tailwater river. North American Journal of Fisheries Management 20: 737 751. 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 rec aptures. Transactions of the American Fisheries Society 135:539 544. Persons, W.R., D.L. Ward, L.A. Avery, and A.C. Burtner. 2011. Standardized methods for Grand Canyon fisheries research 2011, U.S. Geological Survey, Grand Canyon and Monitoring Center. Pine III W.E., S.J.D. Martell, C.J. Walters and J.F. Kitchell. 2009. Counterintuitive responses of fish populations to management actions: some common causes and implications for predictions based o n ecosystem modeling. Fisheries 34: 165 180. Pollack, K.H., S.R Winterstein, C.M. Bunck and P.D.Curtis. 1989. Survival analysis in telemetry studies. Journal of Wildlife Management 53:7 15 Poff, N.L., J.D. Allan, M.B. Bain, J.T. Karr, K.L. Prestegaard, B.D. Richter, R.E. Sparks and J.C. Stromberg. 1997. The Natural Flow Regime. American i nstitute of biological sciences 47: 769 784. Poff, N.L., J.D. Olden, D.M. Merritt, and D.M. Pepin. 2007. Homogenization of regional river dynamics by dams and global biodiversity implications. Proceedings of the National Academy of Sc ience 104: 5732 5737.
86 Poff, N.L., and J.K. Zimmerman. 2010. Ecological responses to altered flow regimes: a literature review to inform the science and management of environmental flows. Freshwater Biology 55: 194 205. Richter, B.D., J.V.Baumgartner R. Wigington, and D.P. Braun. 1997. How much water does a river need? Freshwater Biology 37:231 249. Rinne, J.N. Habitat use of spikedace, Meda fulgida (Pisces:Cyprindae) in Southwestern streams with reference to probable habitat competition by red shin er, Notropis lutrensis (Pices: Cypri nidae). Southwestern Naturalist 36:7 13. Rogers, K. B. and G. C. White. 2007. Analysis of movement and habitat use from telemetry data. Pages 625 676 in M. Brown and C. Guy, editors. Analysis and interpretation of f reshwater fisheries data. American Fisheries Society, Bethesda, Maryland. Rosenfeld, J.S. 2003. Assessing the habitat requirements of stream fishes: an overview and evaluation of different approaches. Transactions of the American Fisheries Society 132: 95 3 968. Schmidt, J.C., R.H. Webb, R.A. Valdez, G.R. Marzolf, and L.E. Stevens. 1998. Science and values in river restoration in the Grand Canyon. Bioscience 48:735 747. Scruton, D.A., C.J. Pennell, M.J. Robertson, L.M.N. Ollerhead, K.D. Clarke, K. Alfreds en, A. Harby, R.S. McKinley. 2005. Seasonal response of juvenile atlantic salmon to experimental hydropeaking power generation in Newfoundland, Canada. North American J ournal of Fisheries Management 25: 964 974. Speas, D. W., C.J. Walters, D.L. Ward and R.Scott Rogers. 2004. Effects of intraspecific density and environmental variables on electrofishing catchability of brown and rainbow trout in the Colorado River. North American Journal of Fisheries Management 24 : 586 596. Stone, D.M. 20 10 Overriding Ef fects of Species Specific Turbidity Thresholds on Hoop Net Catch Rates of Native Fishes in the Little Colorado River, Arizona Transactions of the American Fisheries Society 139: 1150 1170. Stone, D.M. and O.T. Gorman. 2006. Ontogenesis of endangered hump back chub (Gila Cypha) in the Little Colorado River, AZ. American Midland Naturalist 155:123 135 Thomas, D.L., and E.J. Taylor. 1990. Study designs and tests for comparing resource use and availability. The Journal of Wildlife Management 54: 322 330.
87 Topping, D.J., J.C. Schmidt, and L.E. Vierra Jr., L.E. 2003. Computation and analysis of Arizona May 8, 1921, through September 30, 2000. U.S. Geological Survey Professional Paper 1677, 118 pages. Trammell, M. R. Valdez, S. Carothers, and R. Ryel. 2002. Effects of a low steady summer flow experiment on native fishes of the Colorado River in Grand Canyon, Arizona. Final Report of SWCA,Inc to U.S. Geological Survey Grand Canyon M onitoring and Research Center Flagstaff AZ Travnichek, V.H., M.B. Bain and M.J. Maceina. Recovery of a warmwater fish assemblage after the initiation of a minimum flow release downstream from a hydroelectric dam. Transactions of the American Fish eries Society 124:836 844. Tyus, H.M. and C.A. Karp. 1991. Habitat use and streamflow needs of rare and endangered fishes, Green River, Utah. U.S. Fish and Wildlife Service, Vernal, Utah. Valdez, R.A. and R.J. Ryel 1995. Life history and ecol ogy of the humpba ck chub in the Colorado River in Grand Canyon, Arizona. Final Repost of BIO/WEST, Inc. to U.S. Bureau of Reclamation Salt Lake City Utah. Valdez, R.A. and R.J. Ryel. 1997. Life history and ecology of the humpback chub in the Colorado River in Grand Canyo n, Arizona. Final report TR 250 08. BIO/WEST, Inc., Logan, Utah. Walters, C.J. 2007. Is adaptive management helping to solve fisheries problems? Ambio 36: 304 307. Walters, C.J., and F. Juanes. 1993. Recruitment limitation as a consequence of optimal ri sk taking behavior by juvenile fish. Canadian Journal of Fisheries and Aquatic Sciences 50: 2058 2070. Walters, C.J., and S. Martell. 2004. Fisheries Ecology and Management. Princeton University Press, Princeton, New Jersey. Werner, E.E., J.F. Gilliam, D.J. Hall, and G.G. Mittelbach. 1983. An experimental test of the effects of predat ion risk on habitat use in fish Ecology 64:1540 1548. Werner, E.E., and D.J. Hall. 1979. Foraging efficiency and habitat switching in competing sunfishes. Ecology 60: 256 2 64. Werner, E.E. and D.J. Hall. 1988. Ontogenetic habitat shifts in bluegill: the foragi ng rate predation risk tradeoff Ecology 69:1352 1366.
88 Wiele, S.M. and E.R. Griffin. 1997. Modifications to a one dimensional model of unsteady flow in the Colorad o River through Grand Canyon. U.S. Geological Survey, Water Resources Investigation 97 4046. 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, temperatu re and fish prey availability. Transactions of the American Fisheries Society 140: 471 486
89 BIOGRAPHICAL SKETCH Brandon Gerig was born to Mark and Michelle Gerig in April of 1987. During childhood, Brandon grew up in South Bend, IN. Family trips spent in Michigan and Northern Ontario left a strong appreciation and perked his interest in ecology and fisheries. He graduated from Lake Superior State University in 2009 with a B achelor of S cience in f isheries and w ildlife m anagement. While at Lake State Br andon gained experience conducting research and running rivers of the Upper Great Lakes and Alaska. After completing his B.S., Brandon began graduate studies at the University of Florida where he studied the influence of experimental flow regimes on nativ e fish populations in the Colorado River in Grand Canyon. He graduated in May 2012 with a Master of Science Brandon hopes to continue graduate studies as a PhD student or work as a fisheries biologist.