Lake Roosevelt Angler Creel Survey Survey Design and Protocol

Transcription

1 1 Lake Roosevelt Angler Creel Survey 2017 Survey Design and Protocol Contact Info Cas Lawson Fiseries Biologist (509) ext. 204 Elliott Kittel Fiseries Biologist (509) ext. 203 Up-to-date sampling protocol, dataseets, etc., can be accessed at: ttp://spokanetribalfiseries.com/current-fising-trends/ January 1 st, 2017 Version 2.1

2 2 Updates sown in blue text 1/1/17. Protocol Overview Stratified Sampling Sceme A stratified random sampling sceme based on tree seasons and tree reaces will be used to systematically select sampling days and randomly select sites. Te reac strata contain te following sites of interest for sampling: Reac 1: Snag Cove (#31), Cina Bend (#1), Nort Gorge (#2), Evans (#3), Nortport (#32), Colville River (#6), Bradbury Beac (#7), Daisy (#8), Gifford (#9), and Kettle Falls (#5) Reac 2: Hunters (#10), Little Falls Dam (#15), Hawk Creek (#19), Porcupine Bay (#16), Fort Spokane (#17), Seven Bays (#18), and Lincoln (#20) Reac 3: Jones Bay (#21), Hanson Harbor (#22), Keller Park (#28), Crescent Bay (#25), Grand Coulee Dam (#26), Swawilla Basin (#27), and Keller Ferry (#23) Te season strata are defined as follows: Season 1: January 1 Marc 31 Season 2: April 1 September 30 Season 3: October 1 December 31 AM/PM Sampling Sessions Te sampling day will be divided into two (AM/PM) work periods (sessions). Season 1: Te AM session will be from 9:00 AM 3:00 PM, and te PM session will be from 1:00 PM 7:00 PM. Te described session overlap in PM start and AM end times is expected witin te analysis and may not be altered. Season 2: Te AM session will be from 8:00 AM 2:30 PM, and te PM session will be from 2:30 PM 9:00 PM. Tere will be no session overlap during season 2. Season 3: Te AM session will be from 9:00 AM 3:00 PM, and te PM session will be from 1:00 PM 7:00 PM. Te described session overlap in PM start and AM end times is expected witin te analysis and may not be altered. Traditional Access Surveys Traditional access surveys (were te creel clerk spends te entire AM or PM period at one site) will only be conducted at ig-use sites. In Reac 1, te only ig-use site is Kettle Falls (#5). In Reac 2, te ig-use sites include Porcupine Bay (#16), Fort Spokane (#17), Seven Bays (#18), and Lincoln (#20). In Reac 3, tere will be two ig-use sites: Keller Ferry (#23) and Spring Canyon (#24). All angler types (boat and sore) completing teir trips sould be interviewed during traditional access surveys. During te PM session at ig-use sites, if no boat trailers or sore angler parties are present, te

3 3 creel clerks may depart te survey site 30 minutes after sunset. Daily sunset times may be obtained at ttp:// and will be provided to clerks by teir supervisors. Te creel survey assumes tere is negligible angling activity 30 minutes after sunset (particularly during te winter monts). Terefore, te creel clerk will complete te survey form as toug tey ad remained, and will note teir actual departure time as a comment. However, if boat trailers or sore anglers are present, te creel clerks is not expected to remain longer tan one our past sunset and will conduct incomplete trip interviews and provide a comment describing te number of sore anglers and boat trailers remaining at te time of departure. Bus Route Surveys Only low-use sites will be included in te bus routes. All angler types (boat and sore) completing teir trips sould be interviewed. Te following bus routes will be used: Reac 1: Bus Route #94: Snag Cove (#31), Cina Bend (#1), Nortport (#32), Nort Gorge (#2) Bus Route #95: Evans (#3), Colville River (#6), Gifford (#9), Daisy (#8), Bradbury Beac (#7) Reac 2: Bus Route #96: Hunters (#10), Little Falls Dam (#15), Hawk Creek (#19) Reac 3: Bus Route #98: Jones Bay (#21), Hanson Harbor (#22), Keller Park (#28) Bus Route #99: Crescent Bay (#25), Grand Coulee Dam (#26), Swawilla Basin (#27) For all bus routes, a random site on te bus route will be selected as te starting site and ten te above order sould be te order in wic te remaining sites are visited. For example, if during bus route #95 Gifford is selected first, ten te creel clerk would visit Gifford, Daisy, Bradbury Beac, Evans, and ten Colville River. Table 4 sows te bus route wait times at eac site. It is important for surveyors to stay on scedule to elp keep any potential bias negligible. Te randomly selected bus route starting location will influence total circuit travel time. If te clerk is running aead of scedule, te clerk will spend extra time at survey locations (see Appendix B for an example of tis scenario). Te surveyor will not end te survey before te end of te AM or PM session during dayligt ours. During parts of te year, PM surveys may be cut sort by sunset. Wen tis occurs, te creel clerk will discontinue te PM bus route 30 minutes after sunset. Te creel survey assumes tere is negligible angling activity 30 minutes after sunset at low-use sites (particularly during te winter monts). Terefore, te creel clerk will complete te survey form as toug tey ad completed te bus route (assuming zero fising activity at te remaining survey sites), and will note teir actual departure time as a comment. Please reference Appendix B for an example of ow te data seet sould be completed wen sunset cuts a PM bus route sort (Figure B2).

4 4 Survey Effort and Sampling Sceme Witin eac season and reac stratum, a systematic sampling sceme will be used to select survey days. Creel clerks will sample for four consecutive days and ten ave two consecutive days off. For eac reac, a random starting point in tis sequence (i.e. 4 days on, 2 days off) will be selected and assigned to te first day in season 1 (January 1). Te sampling will ten continue following te 4 days on, 2 days off pattern for te rest of season 1 (Marc 31). For example, if te 1 st element of te pattern is randomly selected, ten te clerk would conduct creel surveys on January 1 st, 2 nd, 3 rd, and 4 t, take te 5 t and 6 t off, ten continue surveying on te 7 t, 8 t, 9 t, and 10 t, and ave te 11 t and 12 t off, and so fort until Marc 31 st. Similarly, random starting points witin te 4 days on, 2 days off sequence will be selected and assigned to te first day in season 2 and te first day in season 3. If, due to random selection, more tan four consecutive days are selected between seasons, ten days will be dropped from sampling to ensure only four days are sampled in a row. In general, five days per week (Sunday troug Saturday) will be sampled, resulting in over 600 days of sampling eac year (across te tree reac strata). Once all te survey days are selected, it will ten be determined wic site or group of sites (in te case of a bus route) te clerk will survey. For eac reac, all available bus routes and iguse sites will be considered wit an equal selection probability. Te selection probability for eac ig-use site and bus route is sown in Figure 3. Lastly, te AM/PM sessions will be selected wit equal probability. Data Collection To simplify data collection in te field, species code abbreviations are included at te bottom of te data seet (e.g. Rainbow Trout = RBT). Origin, fin clip, and tag codes are also listed at te bottom of te data seet. Boat and sore anglers will be denoted using B and S codes, respectively. Time sould be recorded in military time (24-our clock). Data seets sould be printed on water resistant paper and must be completed and legible. Incomplete or illegible data seets casts doubt on te collected data. Generally, te creel clerk will intend to collect all data from anglers. Te analysis expects tat nearly 100% of anglers departing a location wit a creel clerk will be interviewed. Terefore, sould te creel clerk encounter many anglers departing at once, minimum complete trip information must be collect from all departing anglers. Minimum trip information includes te number of anglers (in a party), angler start/end time, target species, species and number arvested/released, and presence of adipose fin clips. If time permits, te creel clerk will scan arvested fis for PIT tags, and record zip codes from individual anglers. Biological data collection (e.g. arvested fis lengt, weigt, etc.) is te last priority wen te creel clerk is very busy. If te creel clerk misses interviews, tey must communicate wit teir supervisor to arrange for additional elp. Comments sould be used to provide additional detail or context. Creel clerks will use comments to note missed and declined interviews. Example data seets are included in Appendix B (Figure B1, B2).

5 5 Te supervisor of eac reac will mail all original creel data seets to te Spokane Tribe at te end of eac mont. Scans of all data seets will be provided to te Spokane Tribe by te close of business every Friday. Scans will be ed to and to Data seets will be transcribed into te numeric code input expected by te analysis (described in Appendix B) by te Spokane Tribe. Oter Survey Procedures During access and bus route surveys, te creel clerk is expected to remain on site for te entire dayligt duration of te survey session. Te creel clerk is expected to bring sufficient food and water, and may not leave te site unless an emergency occurs. Emergencies include sudden canges in weater, environmental azards (suc as snow, fire, or smoke), or sudden ealt concerns. If an emergency causes te creel clerk to end te survey, te actual time of departure must be recorded on te data seet (in te Finis Creel field) and te clerk s supervisor must be notified, wo in turn must notify te Spokane Tribe. In te event tat a sceduled survey day cannot be surveyed, clerks are responsible for contacting teir supervisors wo in turn are responsible for contacting Cas Lawson or Elliott Kittel of te Spokane Tribe witin a week of te sceduled survey so a random survey day can be resceduled. If scedule conflicts are known aead of time, supervisors are responsible for contacting Cas or Elliott as soon as possible so a randomized make-up survey can be sceduled. Quality Control Random site cecks of all clerks will be conducted to verify tat te sampling scedule and data entry protocols are followed.

6 6 Study Area Lake Roosevelt is a mainstem impoundment on te Columbia River formed by te construction of te Grand Coulee Dam (USBR 2014). Te reservoir is 151 miles long and offers over 600 miles of soreline to anglers and oter recreationalists (USBR 2014). Te soreline and surrounding landscape can be rugged. Terefore, access to te reservoir by anglers is mostly limited to discrete access points (Figure 1; Brent Nicols, Spokane Tribe of Indians, Fiseries Manager, personal communication). Creel survey metods across te reservoir are identical, but are implemented by tree different agencies (Brent Nicols, Spokane Tribe of Indians, Fiseries Manager, personal communication). Te creel survey in te nortern part of te reservoir (Reac 1) is administered by te Wasington Department of Fis and Wildlife. In te middle and eastern sections of te reservoir (Reac 2), te Spokane Tribe of Indians administers te creel survey. Finally, te Confederated Tribes of te Colville Reservations administers te creel survey in te soutern and western sections of te reservoir (Reac 3). Reac 1 contains 13 access point sites, Reac 2 contains 11 access point sites, and Reac 3 contains 8 access point sites (Table 2). Survey Protocol Interview Procedures Tis protocol utilizes a mix of bus route and traditional access surveys. Table 1 comparatively summarizes te updates to te Lake Roosevelt angler creel survey. Creel clerks will interview all types of anglers, regardless of survey type. Additionally, creel clerks will

7 7 strongly prioritize te collection of complete trip information over incomplete trip information. Effort will be calculated directly from te trip lengt reported by anglers and no incomplete trip information will be used to generate pressure, catc, and arvest estimates. Tis metod will alleviate any bias related to te count-as-you-go metod. However, incomplete trip information will be utilized to bolster collection of biological data (e.g collection of fis lengt, weigt, tissue samples, and tag detections). Te Lake Roosevelt creel survey assumes tat creel clerks will be able to intercept nearly 100% of angler parties completing teir fising trips and tat minimum complete trip information will be collected. Minimum complete trip information includes: number of anglers (in te party), angler start/end time, target species, species and number arvested/released, and presence of adipose (AD) clipped fis. Survey Strata Te current protocol utilizes a stratified sampling sceme were a sampling day and site serves as a sampling unit. Since tree separate entities oversee te implementation of te creel survey, tree reac strata were utilized to simplify allocation of sampling effort. In a cange from te previous creel design, te day type strata were removed due to updates in te sampling sceme (see Survey Effort and Sampling Sceme sections) and te season strata along wit te lengt of te sampling day ave been re-defined based on istorical data. Figure 2 sows every boat party s ending time as well as survey starting and ending times in all reaces based on istorical data collected under te Ceng et al. (2004) protocol. Generally, anglers did not start returning from teir trips until about 9:00 AM during te winter monts. During te

8 8 summer monts, survey starting times may ave prevented clerks from interviewing anglers returning from teir trips before 9:00 AM. It also appears tat survey ending times prevented creel clerks from interviewing all returning boat anglers towards te end of te day in some monts. Furtermore, sunrise and sunset times vary greatly trougout te year (Table 3) and anglers primarily fis during dayligt ours (Brent Nicols, Spokane Tribe of Indians, Fiseries Manager, personal communication). Tis makes te fising day (e.g. te period of te day potentially available for fising by anglers) longer in te summer monts tan in te winter monts. In order to capture te variability in te lengt of te fising day, make te most efficient use of te creel clerks time, and maximize te number of parties completing teir trips te clerk would encounter, te season definitions and sampling day lengt ave been updated to te following: season 1 is defined as January 1 troug Marc 31 wit te sampling day ranging from 9:00 AM 7:00 PM, season 2 is defined as April 1 troug September 30 wit te sampling day ranging from 8:00 AM 9:00 PM, and lastly, season 3 is defined as October 1 troug December 31 wit te sampling day ranging from 9:00 AM 7:00 PM. AM/PM Sessions Te updated protocol divides te sampling day into two work periods: an AM session and a PM session. Tis modification ensures tat a clerk does not ave to sample a site for longer tan a standard 8 our work day. Due to te updates in te lengt of te sampling day, te new protocol varies te session lengt based on season. During seasons 1 and 3, te AM session will be from 9:00 AM 3:00 PM and te PM session will be from 1:00 PM 7:00 PM. During seasons 1 and 3, te AM and PM sessions overlap. Tis overlap is fixed witin te analysis and

9 9 deviation from tese times is not acceptable. In season 2, te AM session will be from 8:00 AM 2:30 PM and te PM session will be from 2:30 PM 9:00 PM. Tere is no session overlap in season 2. Sessions will last for six ours during te winter monts (seasons 1 and 3) and for six and a alf ours during te summer monts (season 2). Six ours of sampling allows te clerk to spend an adequate amount of time at eac site during a winter bus route survey. To cover te longer 13-our sampling day in te summer, te work period is extended to 6.5 ours. Bus Routes and Traditional Surveys Discussions between WEST and LRFEP personnel revealed tat tere is eterogeneity in angler effort among access locations. Wit assistance from LRFEP, tis protocol identifies 7 ig-use and 18 low-use access locations of interest (Table 2). Wit te implementation of te new protocol, traditional access surveys will be conducted only at ig-use locations, and bus route surveys will be conducted only at low-use locations. Tis will allow a surveyor s time to be spent more efficiently and effectively te intention being tat a creel clerk will not spend 6+ ours sitting at a low-use site. Additionally, by aving fewer sites to sample per bus route tan recommended in te Ceng et al. (2004) protocol, creel clerks will be able to spend more time at eac site and less time traveling. For bot survey types, surveyors sould arrive on site by te assigned starting time. Start times are determined by wic season te sampling day falls witin and weter or not te survey is to be conducted during te AM or PM session. Te proposed protocol recommends five bus-routes: two routes in Reac 1, one route in Reac 2, and two routes in Reac 3. In Reac 1, te new protocol defines te first bus route (#94) as Snag Cove, Cina Bend, Nortport, and Nort Gorge, and te second bus route (#95) as

10 10 Evans, Colville River, Gifford, Daisy, and Bradbury Beac. Te single bus route (#96) in Reac 2 consists of Hunters, Little Falls Dam, and Hawk Creek. In Reac 3, bus route #98 contains Jones Bay, Hanson Harbor, and Keller Park and te final bus route (#99) consists of Crescent Bay, Grand Coulee Dam, and Swawilla Basin. A random site in eac bus route will be selected as te starting site and ten te above order sould be te order in wic te remaining sites are visited. For example, if Gifford is selected, ten te creel clerk would visit Gifford, Daisy, Bradbury Beac, Evans, and ten Colville River. Te time eac surveyor spends at eac location during te bus route survey as been determined based on estimates of angler use and conversations wit LRFEP staff (Table 4). It is important for surveyors to stay on scedule staying on scedule will elp keep any potential bias negligible (Jones and Robson 1991). Te randomly selected bus route starting location will influence total circuit travel time. If te clerk runs aead of scedule, te clerk will spend extra time at survey locations (see Appendix B for an example of tis scenario). Te surveyor will not end te survey before te end of te session during dayligt ours. Wen conducting PM session traditional access surveys, te creel clerks may depart te survey site 30 minutes after sunset if no boat trailers or sore angler parties are present. However, if boat trailers or sore anglers are present, creel clerks are not expected to remain longer tan one our past sunset, and will provide a comment describing te number of sore anglers or boat trailers remaining at te time of departure. During PM session bus route surveys, creel clerks may discontinue te bus route 30 minutes after sunset. Daily sunset times may be obtained at ttp:// and will be provided to clerks by teir supervisors. Missed surveys sould be documented, and creel clerks sould collect incomplete trip information from

11 11 remaining sore angler parties. Te creel survey assumes tere is negligible angling activity 30 minutes after sunset (particularly during te winter monts). Terefore, te creel clerk will complete te survey form as toug tey ad remained, and will note teir actual departure time in te appropriate field. Survey Effort Te updated protocol recommends creel clerks sample as muc as possible witin tree constraints: (1) eac agency is required to provide teir clerks two consecutive days off, (2) clerks cannot exceed more tan 40 ours per week, and (3) clerks cannot work more tan five days in a row. Travel times to/from te clerks duty stations to access sites can be significant and are included in te creel clerks work times. Clerks will generally sample for four consecutive days followed by two consecutive days off, wic will result in over 600 days of sampling per year. Te drastic increase in survey effort compared to te previous protocol will improve te precision associated wit catc estimates, particularly for rare species. Sampling Sceme Witin eac season and reac strata, a systematic sampling sceme will be used to select survey days. Creel clerks will sample for four consecutive days and ten ave two consecutive days off. For eac reac, a random starting point in tis sequence (i.e. 4 days on, 2 days off) will be selected and assigned to te first day in season 1 (January 1). Te sampling will ten continue following te 4 days on, 2 days off pattern for te rest of season 1 (Marc 31). For example, if te 1 st element of te sequence is randomly selected, ten te clerk would conduct

12 12 creel surveys on January 1 st, 2 nd, 3 rd, and 4 t, take te 5 t and 6 t off, ten continue surveying on te 7 t, 8 t, 9 t, and 10 t, and ave te 11 t and 12 t off, and so fort until Marc 31 st. As anoter example, if te 3 rd element of te sequence is randomly selected, ten te clerk would conduct creel surveys on January 1 st and 2 nd, take te 3 rd and 4 t off, ten continue surveying on te 5 t, 6 t, 7 t, and 8 t, and so fort until Marc 31 st. As a tird example, if te 5 t element in te sequence is selected, ten sampling would take place January 3 rd troug te 6 t, followed by two days off, resume sampling te 9 t troug 12 t, and so on. Similarly, random starting points witin te 4 days on, 2 days off sequence will be selected and assigned to te first day in season 2 and te first day in season 3. If, due to random selection, more tan four consecutive days are selected between seasons, ten days will be dropped from sampling to ensure only four days are sampled in a row. In general, five days per week (Sunday troug Saturday) will be sampled, resulting in over 600 days of sampling eac year. Once all te survey days are selected, it will ten be determined wic site or group of sites (in te case of a bus route) te clerk will survey. All available bus route(s) and ig-use site(s) witin eac reac will be considered wit equal probability of selection. A diagram sowing te selection probabilities is provided in Figure 3. Lastly, AM/PM sessions will be selected wit equal probability (0.5). Survey Caveats During te winter monts, te survey ending time may be well after sunset. Wen following sunset departure guidelines, te clerk must record te data as toug tey stayed

13 13 te entire allotted time ( Finis Creel sould be te same as te session end time), and note teir actual departure time as a comment. Te survey assumes tat angler activity is negligible 30 minutes after sunset, particularly during te winter monts. During access surveys, if te site is totally inaccessible (due to snow, extreme weater, park closure, etc.), creel clerks will record zero fising activity for te entire assigned survey duration and document teir actual departure time ( Finis Creel sould be te same as te session end time). Te creel clerk may ten conduct a supplementary survey at te next closest access site, including a comment as to wat transpired. If a site is totally inaccessible during a bus route, te clerk may split te remaining survey time among te oter sites. During bus route surveys, te clerk will record te actual amount of time spent at te inaccessible location (e.g. one minute), zero fis fising activity, provide a comment, and move on to te next site unless te inaccessible site is te last site on te bus route. If te last site on a bus route is inaccessible (and no anglers are present), te clerk will record zero fising activity for te remainder of te survey period and document teir actual departure time. For example, consider a scenario were a creel clerk is surveying sites on bus route #96 on April 2 (season 2) and te first site to be visited is Little Falls Dam, followed by Hawk Creek, followed by Hunters. According to Table 2, te creel clerk is supposed to spend 1 our 30 minutes at Little Falls Dam, 1 our at Hawk Creek, and 1 our 30 minutes at Hunters. Te clerk travels to Little Falls Dam and collects creel information for te intended allotted time (1 our 30 minutes). Te clerk ten visits Hawk Creek to discover tat te site is closed to te public and no public veicles are present. Te clerk would record a zero for te fising activity, log te

14 14 amount of time tey stayed at te site (say, 10 minutes), and ten proceed to te next site. Ten, at Hunters, te clerk would spend an additional 50 minutes at te site for a total of 2 ours 20 minutes interviewing anglers. If Hawk Creek ad been te first to be surveyed, ten te remaining time (50 minutes) could be divided between Hunters and Little Falls Dam, wic would result in 1 our 55 minutes spent at Hunters and 1 our 55 minutes spent at Little Falls Dam. In bot of te above cases, te data would be recorded wit te times te clerk actually spent at eac site. However, if Hawk Creek ad been te last site to be surveyed, ten te creel clerk needs to andle tis situation differently since te extra time cannot be distributed to anoter site on te bus route. Tis scenario is similar to ow inaccessible access (ig-use) sites are andled, were te clerk can leave te site early and record a zero for fising activity. As before, te clerk sould record te survey start and finis times ( Start Creel / Finis Creel ) as toug tey ad stayed at te site te entire allotted time and te survey will assume tat tere was no fising activity at te site during te sampling period. Some anglers may decline to be surveyed. If tis appens, te creel clerk sould provide a brief comment on te data seet (e.g. one boat wit two anglers declined survey). Similarly, clerks sould clearly document any missed interviews. Te current survey design assumes tat clerks will be able to interview nearly 100% of angler parties departing te access site. If angler activity is more tan te creel clerk can manage, clerks sould communicate wit teir supervisors to arrange for additional support. Tis information sould be relayed to te Spokane Tribe promptly, and te Spokane Tribe can temporarily provide additional field support or guidance.

15 15 Statistical Analysis Te recommended analysis metods are different from tose previously applied (see Appendix A). In general, te estimators we suggest follow te protocol set fort in Robson and Jones (1989). In comparison, te analysis tool developed by Ceng et al. (2004) appeared to be using statistical metods derived from roving surveys. In te Ceng et al. (2004) protocol, incomplete trip information during teir bus route surveys were used to obtain estimates of sore angler s effort, catc, and catc-rate, wile te statistical metods recommended ere estimate catc and catc-rate using only completed trip information. Since te metodology to estimate catc-rate and catc uses angler s average trip lengt in its derivation, biases associated wit incomplete trips are avoided wen complete trip information is used (Malvestuto 1996). For te most part, te format of te input datasets is similar to te format used for te Ceng et al. (2004) analysis code. An example dataset can be found in Appendix B as well as witin te new analysis tool and Table 5 describes eac column and te possible values. Te new R analysis package provides a function to assist in quality assurance and quality control (QA/QC). Te analysis tool also contains a function to randomly select sampling days and te AM or PM work session for eac reac. Finally, te R analysis tool applies te formulas found in Appendix A. Eac function contains corresponding documentation, wic can be viewed in PDF or HTML formats. Any departures from te proposed protocol sould be considered carefully as tey could result in inaccurate results.

16 16 Discussion As wit any survey, tis survey protocol and analysis make assumptions about angler use and catc. First, te analysis assumes tat no anglers completing teir trips are missed wile te clerk is on site. Any missed or declined surveys must be documented by te creel clerk. Second, te analysis assumes no anglers finis fising before 9:00 AM during seasons 1 and 3 and before 8:00 AM during season 2. Tird, te analysis assumes no anglers are fising after 7:00 PM during seasons 1 and 3 and after 9:00 PM during season 2. Any anglers missed will result in an underestimate of effort and catc. To minimize missing angler interviews, clerks will set up visible creel stations. Te proposed sampling sceme provides te most sampling days witin te clerks work constraints. Altoug te sampling sceme does not follow a traditional systematic sampling sceme (e.g. a simple random sample of sampling days witin eac stratum), tere is no reason to believe tat te sample would be an inaccurate representation of te population. Terefore, te analysis treats te data as if a traditional systematic sampling sceme was used. Te variances for effort, catc-rate, and catc are calculated directly using standard stratified random sampling formulas. Terefore, te calculations assume tat tere is no correlation witin sampling data and tat te error is normally distributed. Additionally, te analysis assumes tat catc-rate and catc are independent to calculate te variance of catc. If tese assumptions are in question, a bootstrap resampling procedure sould be used instead. In summary, for bot survey types, te proposed protocol recommends all anglers completing teir trips be interviewed. Traditional access surveys are to be conducted at ig-use

17 17 sites and bus routes are to be conducted at low-use sites. Season strata definitions now include two 3-mont long winter seasons (season 1 and 3) and one 6-mont long summer season (season 2). Te sampling day was updated to reflect istorical data and account for canging dayligt ours. Te sampling day in eac season is sub-divided into AM and PM sessions to ensure complete coverage of te fising day.

18 18 References Astronomical Applications Department of te U.S. Naval Observatory Sunrise and Sunset, Spokane, Wasington [Data file]. Retrieved from ttp://aa.usno.navy.mil/data/docs/rs_oneyear.pp. Accessed On 11/15/15. Ceng, Y. W., C. Lee, and D. Pavlik Lake Roosevelt 2004 recreational survey design. Lake Roosevelt Fiseries Evaluation Program, Spokane Tribe of Indians, Wellpinit, Wasington. Crone, P. R., and S. P. Malvestuto A comparison of five estimators of fising success from creel survey data on tree Alabama reservoirs. American Fiseries Society Symposium 12: Cocran, W.G Sampling Tecniques. 3 rd Edition. Wiley, New York. Hornsby, F. and S. Ranney LRFEP Creel Survey Analysis Package. R package v2.0. Jones, C. M., and K. H. Pollock Recreational survey metods: estimation of effort, arvest, and released catc. Pages in A. V. Zale, D. L. Parris, and T. M. Sutton, editors. Fiseries Tecniques, 3 rd edition. American Fiseries Society, Betesda, Maryland. Jones, C. M., and D. S. Robson Improving precision in angler surveys: traditional access design versus bus route design. Pages in D. Gutrie, J. M. Hoenig, M. Holliday, C. M. Jones, M. J. Mills, S. A. Moberly, K. H. Pollock, and D. R. Talelm, editors. Creel and Angler Surveys in Fiseries management. American Fiseries Society, Symposium 12, Betesda, Maryland.

19 19 Malvestuto, S. P Sampling te recreational creel. Pages in B. R. Murpy and D. W. Willis, editors. Fiseries tecniques, 2 nd edition. American Fiseries Society, Betesda, Maryland. Pollock, K. H., C. M. Jones, and T. L. Brown Angler survey metods and teir applications in fiseries management. American Fiseries Society, Special Publication 25, Betesda, Maryland. Robson, D., and C. M. Jones Te teoretical basis of an access site angler survey design. Biometrics 45: Tompson, S. K Sampling. 2 nd edition. Wiley, New York. United States Bureau of Reclamation (USBR) Grand Coulee Dam. Available at: ttp:// Accessed on 03/12/14.

20 20 Tables Table 1. Comparison between te Ceng et al. (2004) protocol and te protocol provided by Western Ecosystem Tecnology, Inc. (WEST). Documentation Item Ceng et al WEST Documentation out-ofdate wit current survey metods and incomplete descriptions of analysis metods. No documentation for R analysis tool. Up-to-date documentation wit in-text citations and formulas. Extensive documentation associated wit R analysis tool. Sites for inclusion in sampling frame Some sites were included, even toug tey were not of interest to LRFEP. Sites wit low angler visit probabilities were eliminated from te sampling frame and assumed to not exist. Seasonal strata Sampling Day Lengt Inaccessible sites Low-use versus ig-use sites Seasons did not take full advantage of angler s ending times. Inefficient use of clerk s time in te winter. Inadequate coverage of te fising day during te summer. No ability to select a different site; creel clerks were required to be tere all day. Waiting times at low-use sites could be very long causing creel clerks to sit idle for most of te day. Updated seasons based on istorical data. Updated te sampling day lengt based on istorical data. Te sampling day is divided into two sessions. Provides a protocol for dealing wit inaccessible sites. Traditional access surveys will be conducted only at ig-use sites, reducing te likeliood tat surveyors are sitting idle. Comparatively, low-use sites will be visited for only part of an AM/PM session.

21 21 Table 2. (Cont). Item Ceng et al WEST Number of sampling days 39 days were surveyed eac (4-mont) season per reac. 5 days per week will be surveyed in eac reaces 1 and 2. Data QAQC R tool stopped running if data was formatted improperly. R tool includes a data QAQC function to elp ensure proper data formatting. R tool output output included over 800 files wit almost 50 different.r files required. R package tat as independent elp files and documentation; outputs are limited to wat is necessary.

22 22 Table 2. List of all sites by reac and usage designation (ig-use or low-use). Sites were usage is left blank are not included in sampling frame. A map of tese locations can be found in Figure 1. Reac Code Site Name Usage (ig or low) 29 Incelium Frenc Rocks Snag Cove Low 32 Nortport Low 1 Cina Bend Low 2 Nort Gorge Low 1 3 Evans Low 4 Marcus Island Kettle Falls Hig 6 Colville River Low 7 Bradbury Beac Low 8 Daisy Low 9 Gifford Low 10 Hunters Low 11 Columbia Campground Two Rivers Marina A-Frame Spokane Tribal Boat Launc Little Falls Dam Low 16 Porcupine Bay Hig 17 Fort Spokane Hig 18 Seven Bays Hig 19 Hawk Creek Low 20 Lincoln Hig 21 Jones Bay Low 22 Hanson Harbor Low 23 Keller Ferry Hig 3 24 Spring Canyon Hig 25 Crescent Bay Low 26 Grand Coulee Dam Low 27 Swawilla Basin Low 28 Keller Park Low

23 Table sunrise/sunset times for Spokane, Wasington in Pacific Standard Time. One our sould be added for dayligt time, if and wen in use. (Source: Astronomical Applications Department of te U.S. Naval Observatory) 23

24 24 Table 4. Te number of ours allocated at eac creel survey location on bus route surveys. Surveyors sould spend te number of ours listed below at eac location. Only low-use sites are surveyed during te bus route surveys. A random site in eac bus route will be selected as te starting site and ten te order listed below sould be te order in wic te remaining sites are visited. For example, if Gifford from bus route #95 is selected first, ten te creel clerk would visit Gifford, Daisy, Bradbury Beac, Evans, and Colville River, in tat order. Te travel time listed is te time allotted to complete an entire circuit of te bus route. Reac Bus Route Number Code Site name Season 1 (Jan-Marc) Season 2 (April-Sept) Season 3 (Oct-Dec) Snag Cove Cina Bend Nortport Nort Gorge Travel Time between sites Total time (n ours) spent on bus route Evans Colville River Gifford Daisy Bradbury Beac Travel Time between sites Total time (n ours) spent on bus route Hunters Little Falls Dam Hawk Creek Travel Time between sites Total time (n ours) spent on bus route Jones Bay Hanson Harbor Keller Park Travel Time between sites Total time (n ours) spent on bus route Crescent Bay Grand Coulee Dam Swawilla Basin Travel Time between sites Total time (n ours) spent on bus route

25 25 Table 5. Column names and descriptions of eac variable expected in te input data frame for te creel analysis tool. See Appendix B for an example of te data format. Column Name Description Data Type id Unique interview record number. integer: {1, 2, 3,, n} survey Type of survey (traditional or bus route): integer: {1, 2} 1=traditional (access) 2=bus route date Calendar date of te survey. date: mm/dd/yy mm Mont te survey was conducted: integer: {1, 2,, 12} 1=January 2=February 3=Marc 4=April 5=May 6=June 7=July 8=August 9=September 10=October 11=November 12=December dd Calendar day witin te mont te survey was conducted. integer: {1, 2,, 31} depending on te mont yy Year te survey was conducted. integer: {2004, } wday Day of te week te survey was integer: {1, 2,, 7} conducted: 1=Monday 2=Tuesday 3=Wednesday 4=Tursday 5=Friday 6=Saturday 7=Sunday reac Reac or section of Lake Roosevelt: integer: {1, 2, 3} 1=upper 2=middle 3=lower loc Location or access site number te creel clerk is surveying. Possible codes are listed in Table 2. integer.

26 26 Table 5. (Cont.) Column Name Description Data Type screel Proportion of te 24-our day tat a creel survey was started for a particular location; note tat screel sould always be less tan fcreel. For example, if te survey starts at 10:30 AM PST, ten screel = 10.5 / 24 = Values can be rounded to five significant figures. fcreel ours party.no Proportion of te 24-our day tat a creel survey was ended for a particular location; note tat fcreel sould always be greater tan screel. For example, if te survey ends at 01:30 PM PST, ten fcreel = 13.5 / 24 = Values can be rounded to five significant figures. Proportion of te 24-our day tat a creel survey lasted for a particular location; ours = fcreel screel. A unique party number for eac group interviewed. A party is defined as eiter sore anglers (and associated nonanglers, if applicable) or boat anglers (and associated non-anglers, if applicable). A party sould not contain a mixture of boat and sore anglers. numeric values in te interval: [0, 1] numeric values in te interval: [0, 1] numeric values in te interval: [0, 1] integer: {1, 2, } party.size Party size, including non-anglers. integer: {1, 2, } angler.no A unique angler number for eac angler. integer: {1, 2, } angler.count Total number of anglers in a party. integer: {1, 2, } angler.type stime Type of angler (eiter boat or sore): 0=non-angler 1=boat angler 2=sore angler Te proportion of te 24-our day tat an angler started fising. For example, if te angler started fising at 9:45 AM PST, ten stime = 9.75 / 24 = Values can be rounded to five significant figures. Note tat stime sould always be less tan etime. integer: {0, 1, 2} numeric values in te interval: [0, 1]

27 27 Table 5. (Cont.) Column Name Description Data Type etime Te proportion of te 24-our day tat an angler finised fising. For example, if te angler finised fising at 12:10 PM PST, ten etime = / 24 = Values can be rounded to five significant figures. Note tat etime sould always be greater tan stime. totrs completed Proportion of te day tat an angler fised at a particular location; totrs = etime stime. Weter te party was leaving te location (i.e. a completed trip) at te time of te interview; incomplete trips also include trailer and sore angler counts: 0=no (incomplete trip) 1=yes (complete trip) numeric values in te interval: [0, 1] numeric values in te interval: [0, 1] integer: {0, 1} zipcode Te interviewee s current zipcode 5 digit integer target.sp.1 Te primary species code tat te angler was trying to catc: 0=none 1=rainbow trout 2=kokanee 3=walleye 4=smallmout bass 5=wite sturgeon 6=burbot 7=oter salmonid 8=oter 9=nortern pike 10=yellow perc integer: {0, 1,, 10} target.sp.2 Te secondary species code tat te angler was trying to catc: 0=none 1=rainbow trout 2=kokanee 3=walleye 4=smallmout bass 5=wite sturgeon 6=burbot 7=oter salmonid integer: {0, 1,, 10}

28 28 Table 5. (Cont.) Column Name Description Data Type target.sp.2 (Cont.) 8=oter integer: {0, 1,, 10} 9=nortern pike 10=yellow perc sp.caugt Te species code caugt by an angler: integer: {0, 1,, 19} 0=none 1=atcery rainbow trout 2=unknown origin rainbow trout 3=wild rainbow trout 4=atcery kokanee 5=unknown origin kokanee 6=wild kokanee 7=walleye 8=smallmout bass 9=wite sturgeon 10=burbot 11=nortern pike 12=yellow perc 13=lake witefis 14=brown trout 15=common carp 16=nortern pikeminnow 17=black crappie 18=mountain witefis 19=cinook salmon.r Weter te species caugt was integer: {0, 1} arvested: 1=arvested 0=released tl* Total lengt of te species caugt integer: {0, 1, } rounded to te nearest millimeter. wt* Weigt of te species caugt rounded to integer: {0, 1, } te nearest grams. mark* Type of mark: 0=none 1=floy tag 2=fin clip 3=elastomere 4=fin deformity integer: {0, 1,, 4}

29 29 Column Name Description Data Type fin.clip.type* Fin clip combinations: integer: {0, 1,, 6} 0=none 1=Adipose 2=Rigt pectoral 3=Left pectoral 4=Rigt ventral 5=Left ventral 6=oter tag.number* Te tag series number. numeric *Auxiliary columns not used in analysis

30 30 Figures Figure 1. Lake Roosevelt, Wasington creel survey locations. Eac named location as an associated code and can be found in Table 2.

31 Figure 2. Plot sowing party ending times for boat anglers in all reaces of Lake Roosevelt based on istorical data collected under te Ceng et al. (2004) protocol. Eac black point represents a boat party s ending time. Eac blue plus symbol represents eiter a survey starting or ending time. Survey starting times range from 07:26:24 10:48:00 and survey ending times range from 14:30:00 20:38:24. 31

32 32 Bus Route #94 π c =0.33 Reac 1 Bus Route #95 π c =0.33 Kettle Falls (#5) π c =0.33 Bus Route #96 π c =0.20 Porcupine Bay (#16) π c =0.20 Reac 2 Fort Spokane (#17) π c =0.20 Seven Bays (#18) π c =0.20 Lincoln (#20) π c =0.20 Bus Route #98 π c =0.25 Reac 3 Bus Route #99 π c =0.25 Keller Ferry (#23) π c =0.25 Spring Canyon (#24) π c =0.25 Figure 3. Tis figure depicts a probability tree diagram sowing circuit selection probabilities (π c). Here, a circuit is defined as eiter a single (ig-use) site or a bus route consisting of numerous sites. All available circuits are given an equal probability of selection witin eac reac.

33 33

34 Figure 4. Front and back of Lake Roosevelt angler creel survey data collection seet, Version

35 35 Appendix A Statistical Metods Populations are generally too large or too costly to census completely. As a result, samples are obtained from a population using probability sampling (Tompson 2002) so tat statistical inferences about te properties of a population can be made. Here, as in Ceng et al. (2004), we recommend using a stratified random sample were te sampling unit is a simple random sample of days are selected witin eac stratum. Eac stratum is defined as te days and sites contained witin eac reac and season. Tese strata take advantage of te large spatial scale and limitations of personnel as well as te differences in temporal use trougout a year and witin a week. For te benefit of te reader, a brief overview of te formulas for stratified random sampling is provided. A more detailed description including derivations can be found in Cocran (1977) or Pollock et al. (1994). Overview of Stratified Random Sampling For stratified random sampling te following general notation and formulas are used: te stratum being considered ( = 1, 2,..., L) ; i te sampling unit (i = 1,2,..., N ) ; N te population size for stratum ; n te sample size in stratum ;

36 36 N L = N te total population size; = 1 L n= n te total sample size; = 1 W N N = te fraction of te population in stratum, also called te stratum weigt; y i te value of te i t unit of stratum ; y = n i= 1 n y i is te sample mean for stratum ; (0) s = n 2 i= 1 (y y ) n i 1 2 te sample variance for stratum ; (0) y st N L = y te overall estimate of te population mean; (0) = 1 N L 2 2 s N n Var( y st ) = W te variance of te population mean; (0) = 1 n N y st n = N y te overall estimate of te population total; (0) = 1 2 st Var( y ) = N Var( y ) te overall variance of te population total. (0) st Note tat te standard error of any estimate is te square root of te variance. Te formulas above apply to te daily estimates of catc rate, arvest rate, and average effort.

37 37 Effort Estimation Traditional access surveys are a special case of a bus route survey in wic only one or two sites are visited during a sampling day (Jones and Robson 1991). Separate formulas are not needed for daily estimates obtained from traditional surveys versus estimates obtained from bus route surveys (Jones and Robson 1991; Jones and Pollock 2012). Creel clerks sould obtain complete trip interviews from all anglers, regardless of angler type, as tey leave. Complete trip data, regardless of angler type, is used to estimate catc rate and arvest rate (Malvestuto 1995; Jones and Pollock 2012). Sore angling effort for sampling dayi is estimated using te following equation (Robson and Jones 1989): E Si = T 1 w n m kj j= 1 k= 1 j π c e π k (0) were E Si = estimated sore angler effort for te entire i t fising day (ours), T = survey day lengt (ours), w j = wait time at te t j site in ours ( j 1, 2,, n) =, e kj = total time in ours tat te k t sore angler is at te t =, π k = te j site wile clerk is on-site ( k 1, 2,, m) t probability tat te k angler is in te sample, π c = te probability tat te c t circuit is in te sample. A circuit is defined as eiter a single (ig-use) site or a bus route consisting of numerous sites. Figure 3 contains all possible circuits and teir corresponding probabilities.

38 38 In season 2, π k = 0.5 since te AM/PM sessions are distinct and are selected wit equal probability. In seasons 1 and 3, π k will vary based on te time of day te angler ends teir trip. If an angler ends teir trip between 9:00 AM and 1:00 PM during an AM session or between 3:00 PM and 7:00 PM during a PM session, ten π k = 0.5. However, if an angler ends teir trip between 1:00 PM and 3:00 PM, ten π k = 1since tis time period as a 100% cance of being surveyed. Boat angler effort ( E Bi ) is estimated using te same equation as above, but instead e kj is te total time in ours tat te k t boat angler is at te j t site wile clerk is on-site ( k 1, 2,, m) =. Te total effort between sore anglers and boat anglers is found by adding effort estimates of boat and sore anglers togeter: Ei EBi ESi = + (0) Once daily estimates of effort are calculated, strata totals, means, and variances are calculated using equations (0) troug (0) by substituting E i for y i. For example, from equation (0), mean effort for stratum is calculated as: E n E i i= 1 = = n E (0) and te witin stratum variance of te average effort is (see equation (0)):

39 39 s ( ) = Var E = n 2 i= 1 E (E E ) n i 1 2 (0) Effort can also be a measure of te total number of anglers and can be estimated similar to effort measured in angler ours. By replacing e kj wit anglers for te sampling day can be estimated by te following equation: k j in equation (0), te number of A Si = T 1 w n m j j= 1 k= 1 j π c k π k (1) were A Si = te estimated total number of sore anglers for te i t sampling day and k j = one sore angler at te j t site wile clerk is on-site. To obtain an estimate of te total number of boat anglers ( A Bi ), k j would be defined as one boat angler at te t j site wile clerk is on-site. Te estimated number of sore anglers ( ASi ) plus te estimated number of boat anglers ( A Bi ) provides an estimate of te total number of anglers during te t i sampling day ( i ) A. Once again, te standard stratified random sampling formulas provided at te beginning of te section can be applied to find stratum level and population level estimates (equations (0) troug (0)). Catc Rate and Harvest Rate Estimation Oter variables of interest are catc, catc-rate, arvest, and arvest-rate. Altoug, catc and arvest are important for fisery managers to assess separately, in tis section and te next section, te term catc will be used to generically represent eiter catc or arvest. In te

40 40 formulas provided, arvest can simply be substituted for catc to obtain corresponding estimates of arvest and arvest rate. Moreover, catc-rate can be estimated as eiter catc per angler or catc per our. In te equations below, effort will be in terms of estimated angler fising ours ( E i ). To instead obtain estimates of catc per angler, A i may be substituted for Estimates of catc rate are obtained only from completed trip interviews. Stratum level catc rate is calculated using te ratio of means (Pollock et al. 1994): E i. R n c i n c i= 1 = = = n Ei E n i= 1 R (1) were R is te catc per unit our for te t strata and c i = te estimated catc for te i t sampling day. Crone and Malvestuto (1991) found tat wen daily values of catc and effort were used, precision between te ratio of means estimator and te means of ratio estimator were similar. Te ratio of means estimator is self-weigting, meaning tat individual contributions are weigted by te amount of fising effort and is a better estimator of completed trip data tan te mean of ratios (Malvestuto 1996). Here, c i is calculated by substituting e kj wit equation (0) and can be calculated for eiter boat anglers or sore anglers: ckj from c i = T n 1 w m j= 1 k= 1 j π c c π kj k (1) were c kj t = te catc for te k (sore or boat) angler at te j t site wile clerk is on-site. Te 2 witin strata variance for catc-rate ( ) s is calculated as: R

41 41 s 2 R ( ) = Var R = n i= 1 c i R Ei n 1 2 (1) Catc And Harvest Estimation E by catcrate ( R ), or: Stratum level estimates of catc ( C ) are obtained by multiplying effort ( ) 2 and te variance of average catc ( ) ( ) C C = E * R = C (1) s is calculated as: C s = Var C = E Var( R ) + R Var( E ) + Var( E ) Var( R ) (1) Lastly, overall strata totals and means are calculated using equations developed for stratified random sampling (see equations (0) troug (0)).

42 42 Appendix B Te following is an example of te type of data a creel clerk migt collect on a typical sampling day. A corresponding example of te data is also sown in Table B1. A description of eac column and possible values can be found in Table 5. Example: Te creel clerk from Reac 1 is assigned to conduct an AM bus route survey #95 on 1/9/2016 (Figure B1). Tey arrive at Evans at 09:00 AM Pacific Standard Time (PST) and stay at Evans for te allotted time (1.0 our; see Table 4). Wile at Evans, te clerk encounters one party of two sore anglers finising teir trip. Te anglers report tat tey fised from 5:30 AM to 9:15 AM. Te anglers were oping to catc a rainbow trout, but didn t catc any fis. Te clerk visits te next site along te bus route (Colville River location #6). Te clerk arrives at 10:20 AM and stays te allotted time (0.75 our). Te creel clerk does not encounter any anglers finising teir trip. Ten, at 11:30 PM PST, te clerk arrives at te next site (Gifford location #9) and stays for te allotted time (1 our). At Gifford, te clerk encounters 1 boat angler coming in from teir fising trip. Te angler reports tat tey ad been fising from 8:00 AM to 12:00 PM. Te angler was oping to catc eiter rainbow trout or walleye, but ended up catcing one kokanee and one burbot. Te angler decided to keep te kokanee, but trow back te burbot. Ten, at 12:45 PM PST, te clerk arrives at te next site (Daisy location #8) and stays for te allotted time (1.0 our). At 13:30 PM PST, te clerk talks to an angler wo is still fising

43 43 along te sore and as been fising since 11:00 AM. Te angler is oping to catc a walleye and, so far, as caugt and kept 3 walleye. Finally, te clerk travels to Bradbury Beac (location #7) and arrives at 2:00 PM. Due to te random start location (Evans), te clerk is a little aead of scedule, and will remain at Bradbury Beac until te end of te AM session (3:00 PM), wic is an extra 15 minutes beyond te sceduled time (0.75 our; Table 4). Te clerk encounters two sore anglers wo ad been fising togeter, but ad arrived at te site separately. Te first angler arrived at 9:30 AM and te oter angler arrived at 10:00 AM. Bot anglers finised fising at 2:00 PM. Neiter angler caugt any fis, but tey ad been trying to catc kokanee. After tat, te clerk encounters anoter boating party of people returning from teir trip. Te party consists of two non-anglers and two anglers. Tey report tey ad fised from 9:45 AM to 2:00 PM. One angler caugt no fis and te oter angler caugt and arvested 2 rainbow trout. Upon inspection of te rainbow trout, te clerk noticed tat one rainbow trout was adiposed clipped, and detects and records a PIT tag number from te unmarked fis. Neiter angler was targeting any fis in particular.

44 44 Figure B1. Data seet example for te described bus route #95 sampling scenario. Note tat te random start location will influence total bus route circuit travel time. Te entirety of te session sould be spent sampling, meaning te survey sould not be ended before te end of te session (in tis example, te season 1 AM session end time is 3:00 PM).