DFO L b ary / MPO B bliothèque III II I I II I II 12032 06 An Evaluation of Alternative Methods of Collecting Sportfishing Statistics for the Northwest Territories

J. O. Jacobson, R. D. Cook and R. Sopuck

Western Region Department of Fisheries and Oceans SEP VVinnipeg , Manitoba R3T 2N6

August 1983

Canadian Technical Report of Fisheries & Aquatic Sciences No.1180

2;3 -0"

Fisheries Pêches and Oceans et Océans Came. Canad'ranTéchnical Report of Tisheriçs âr^cl ^4qOatic Sçiéneés These reports contain stleritiÎi^ and technical information that represents an important contributioü to existing knôwlecfge but which for some reason-may not be appropriate for primary scientific (i.e. Journal) publication. Technical Reports are directed primarily towards a worldwide audience and have an international distribution. No restriction is placed on subject matter and the series reflects the broad interests and policies of the Department of Fisheries and Oceans, namely, fisheries management, technology and development, ocean sciences, and aquatic environments relevant to Canada. Technical Reports may be cited as full publications. The correct citation appears above the abstract of each répbrt. Eâch report will be abstracted irrAquatic Sciences and Fisheries Abstracts and will be indexed annually in the Department's index to scientific and technical publications. Numbers 1-456 in this series were issued as Technical Reports of the Fisheries Research Board of Canada. Numbers 457-714 were issued as Department of the Environment, Fisheries and Marine Service, Research and Development Directorate Technical Reports. Numbers 715-924 were issued as Department of Fisheries and the Environment, Fisheries and Marine Service Technical: Reports. The current series name was changed with report number 925. -. ., - Details on the availability of Technical Reports in hard copy may be obtained from the issuing establishment indicated on the front cover.

Rapport technique canadien des sciences halieutiques et aquatiques Ces rapports contiennent des renseignements scientifiques et techniques qui constituent uneconlribution importante:aux connaissances actuelles mais qui, pour une raison ou pôur une âutire, ne semblent pas âppropriés pour la publication dans un journal scientifique. Il n'y a aucune restriction quant au sujet, de fait, la série reflète la vaste gamme des -intérêts et des "Politiques du,Mi4istère des Pêches et des Océans, notamment gestion des pêches, techniques et développement, sciences océaniques et environnements aquatiques, au Canada. Les Rapports techniques peuvent être considérés comme des publications complètes. Le titre exact paraîtra au haut du résumé de chaque rapport, qui sera publié dans la revue Aquatic Sciences and Fisheries Abstrac•ts et qui figurera dans l'index annuel des publications scientifiques et techniques du Ministère. Les numéros 1-456 de cette série ont été publiés à titre de Rapports techniques de l'Office des recherches sur les pêcheries du Canada. Les numéros 457-714, à titre de Rapports techniques de la Direction générale de la recherche et du développement, Service des pêches et de la mer, ministère de l'Environnement. Les numéros 715-924 ont été publiés à titre de Rapports techniques du Service des pêches et de la mer, Ministère des Pêches et de l'Environnement. Le nom de la série a été modifié à partir du numéro 925. La page couverture porte le nom de l'établissement auteur où l'on peut se procurer les rapports sous couverture cartonnée. Canadian Technical Report of

Fisheries and Aquatic Sciences 1180

August 1983

AN EVALUATION OF ALTERNATIVE METHODS OF COLLECTING

SPORTFISHING STATISTICS FOR THE NORTHWEST TERRITORIES

by

J. O. Jacobson, R. D. Cook and R. Sopuckl

Western Region

Department of Fisheries and Oceans

Winnipeg, Manitoba R3T 2N6

This is the 157th Technical Report

from the Western Region, Winnipeg

1 All from Interdisciplinary Systems Ltd., 966 Waverley Street, Winnipeg, Manitoba R3 1 4M5 EDITOR'S NOTE

The preparation of the material for this study was done by Inter- disciplinary Systems Ltd. Environmental and Socio-Economic Consultants of Winnipeg, Manitoba through a contract with the Department of Fisheries and Oceans, Western Region, Winnipeg, Manitoba.

This study is an edited version of their report to the Department entitled "A Statistical Reporting System for Recreational Fisheries in the Northwest Territories". That study contained chapters outlining departmental fisheries management, biological and economic programs, descriptions of existing data bases and two appendices (one describing native food fisheries and another containing an annoted bibliography of fishery management publica- tions) that have not been reproduced here.

P.C. Thompson

ED Minister of Supply and Services Canada 1983

Cat. no. FS 97-6/1180 ISSN 0706-6457

Correct citation for this publication is:

Jacobson, J.0., R.D. Cook, and R. Sopuck. 1983. An evaluation of alternative methods of collecting sportfishing statistics for the Northwest Terri- tories. Can. Tech. Rep. Fish. Aquat. Sci. 1180: iv + 18 p. lii

TABLE OF CONTENTS

ABSTRACT/RESUME ...... INTRODUCTION ...... DATA REQUIREMENTS ...... Classification of sport fisheries ...... Intensive fisheries ...... Intensive/extensive fisheries ...... Extensive fisheries ...... Classification of NWT fisheries by sampling requirements Lodge-related fisheries ...... Road-access fisheries ...... Types of data ...... Biological information ...... Economic information ...... DESCRIPTION AND EVALUATION OF DATA COLLECTION SYSTEMS ... Delayed data collection systems ...... Conclusions ...... On-site data collection systems ...... Conclusions ...... Hybrid data collection systems ...... Conclusions ...... RECOMMENDED DATA COLLECTION SYSTEMS ...... Mail questionnaires ...... Site-specific systems ...... Remote-access lodge fisheries ...... Limited-access day fisheries ...... Multiple-access day fisheries and extended-stay fisheries . RECOMMENDATIONS FOR FURTHER STUDY ...... ACKNOWLEDGMENTS ...... REFERENCES ......

LIST OF TABLES

Table Page

1 Examples of lakes and rivers in each fishery category in the NWT 17 2 Summary of reporting methods used for evaluating fish harvest in the Western United States ...... 17

LIST OF FIGURES

Figure Page

1 Schematic of a statistical reporting system for the sport fishery of the NWT ...... 18 iv

ABSTRACT

Jacobson J.O., R.D. Cook, and R. Sopuck. 1983. An evaluation of alternative methods of collecting sportfishing statistics for the Northwest Terri- tories. Can. Tech. Rep. Fish. Aquat, Sci. 1180: iv + 18 p.

This study, as a survey of statistical methods for collecting sportfish- ing data, outlines the basic data requirements for biological and economic fisheries management. The study examines and evaluates delayed data collec- tion systems, on-site data collection systems and hybrid data collection sys- feras with reference to the sport fisheries management requirements of the Northwest Territories.

The study recommends the use of a controlled diary system to collect catch and effort data at remote-access lodge fisheries: a catch/effort census for collecting data for road access fisheries; mail questionnaires for gather- ing general data and trends in catch and effort for intensive/extensive fisheries, and unstructured field checks for providing general information for extensive fisheries. Finally the study outlines the general methodologies to be followed in order to implement the recommended systems.

Key words: sport fishing; catch/effort data; fisheries management.

RESUME

Jacobson, J.O., R.D. Cook, and R. Sopuck. 1983. An evaluation of alternative methods of collecting sportfishing statistics for the Northwest Terri- tories. Can. Tech. Rep. Fish. Aquat. Sci. 1180: iv + 18 p.

Ce rapport, qui se veut une étude sur les méthodes de collecte de don- nées sur la pêche sportive, fournit les statistiques fondamentales nécessaires à l'aménagement biologique et économique des pêcheries. Le rapport examine et évalue divers systèmes de collecte des données: en différé, sur place et hy- brides, et ce, en fonction des besoins d'aménagement de la pêche sportive dans les Territoires du Nord-Ouest.

Le rapport recommande que l'on utilise un carnet où l'on consigne au jour le jour les données sur les prises et sur l'effort dans les pavillons de pêche isolés; que l'on envoie des questionnaires postaux afin de prélever des données générales et relever les tendances du rapport prise-effort chez les pêcheries intensives et de grande envergure, et que l'on effectue un sondage sur les rapports prise-effort dans les pavillons accessibles par la route ainsi que des contrôles libres sur le terrain pour recueillir des données générales sur les pêcheries de grande envergure. Enfin, le rapport souligne quelles méthodes générales adopter pour implanter les mesures proposées.

Mots-clés: pêche sportive; donn4s sur le rapport prise-effort; aménagement des pêcheries. INTRODUCTION Intensive fisheries

The Department of Fisheries and Oceans Fisheries that meet one or more of •the (DFO) is responsible for managing the fish and following criteria are classified as intensively marine mammal resources of the Northwest Terri- managed lakes: tories (NWT). DFO's management mandate is to protect fish stocks, allocate fisheries - fisheries where any combination of com- resources among alternate users, and evaluate mercial, sport or domestic uses are the economic contribution of fisheries resources straining the capacity of the fish to the Canadian and regional economies. As a resources; result of increased northern development, the - any single use fishery that may be problem of allocating limited fish resources exploited to the point where there is among user groups, some of which are in excess fishing effort; competition with one another, represents a major - fisheries that have existing or immi- challenge to resource managers at DFO. nent resource use conflicts, such as between sport and commercial fishermen. The NWT include about 3.4 million square kilometers, more than one-third of the total of Intensive fisheries having any of the Canada. More than 1.3 million square kilometers above characteristics have management problems of the NWT are freshwater lakes that support a that require immediate action and therefore wide diversity of sport, commercial, and domes- require accurate and timely information. Inten- tic fisheries. Two lakes in particular, Great sive sport fisheries are usually associated with Bear and Great Slave, support most of the either high density road access, high density recreational fishing in the NWT. In addition, remote access lodges, or a combination of the there are domestic fisheries on both lakes and a above. Examples of intensive fisheries in the commercial fishery exists on Great Slave Lake. NWT are shown in Table 1. Several other lakes, such as those in the Hay River-Fort Smith, Yellowknife, and Inuvik areas, Intensive/extensive fisheries are subjected to an intermediate amount of fish- ing pressure and resource use conflicts. Final- Fisheries that meet one or more of the ly, a large number of lakes within the NWT are following criteria are classified as intensive/ isolated and receive little, if any, fishing extensive fisheries: pressure of any type. - where there is an established pattern The orderly development and management of of use by commercial, sport, or domes- the fish resources of the NWT requires timely tic users that does not currently and accurate statistics on stocks, yields, strain the capacity of the fish effort, economic values, and public attitudes resources. towards the fish resource. A statistical - where planned or anticlpated increases reporting system to provide these data must be in use could exceed the capacity of the flexible enough to accommodate the diversity of fish resources. fisheries management requirements that exist in - where resource use conflicts do not the NWT. currently exist and are not anticipated to be a problem in the immediate This report describes the biological and future. economic data requirements for the management of the NWT recreational fisheries and reviews the The same biological and economic informa- literature on data collection systems. From tion is required for intensive/extensive fisher- this base, the report outlines a data collection ies as for intensive fisheries; however, the program to meet existing and potential manage- information does not have to be collected as ment requirements and recommends a pilot program frequently and may not have to be as precise. to develop and best specify data collection sys- Examples of lakes in this category are also tems for the sport fisheries in the territory. included in Table 1.

Extensive fisheries

DATA REQUIREMENTS Fisheries that meet one or more of the following criteria are classified as extensive CLASSIFICATION OF SPORT FISHERIES fisheries:

The fisheries of the NWT have been classi- - where there are commercial, sport or fied into three general types based on the level domestic users, but where the level of of management required. These management cate- fishing effort is very low in relation gories are intensive fisheries, intensive/ to the capacity of the fish resources. extensive fisheries, and extensive fisheries - where there are no present concerns (R. Peet personal communication). The catego- that anticipated expansion of the fish- ries, based on present levels of exploitation, ery will exceed the capacity of the are therefore subject to change if developments fish resources. such as road access change the use of the - where resource use conflicts do not fishery. currently exist and are not anticipated within the next 5 to 10 years. 2

Extensive fisheries only require the docu- lakes. It may also be required for mentation of the locations, type, and general selected areas within lakes. levels of sport fish,ing. Examples of extensive fisheries include most lakes in the District of Stock and harvest are basic biological Franklin and any lakes in the Keewatin or Mac- parameters needed to manage fisheries. Stock kenzie Districts that do not have lodge or com- assessments require estimates of such factors as munity access. length, weight, age and reproductive status. Estimates of harvest, effort, and catch/unit effôrt are used in conjunction with stock CLASSIFICATION OF NWT FISHERIES BY SAMPLING assessments to monitor the exploitation rate of REQUIREMENTS the fishery. They are also used in conjunction with age and length distributions to monitor the The management classification of NWT effects of exploitation of populations, and as fisheries described above is designed to iden- management indicators of the quality of the tify the types of information required to manage fishery. DFO has programs in place to conduct the fishery and the frequency with which the biological stock assessments of NWT fisheries as data should be collected. The design of data required. This study deals with•methods of col- collection programs requires the further classi- lecting catch, effort, and catch/unit effort fication of each of these management classifica- statistics for the sport fisheries in a manner tions, based on the types of sampling techniques compatible with, and complementary to, the that will be required to collect data on yield existing stock assessment program. and effort. Slightly different data collection techniques will be required for each of the fol- Economic information lowing categories. There is a close relationship between the Lodge-related fisheries data required to meet biological management objectives and the data required to meet econo- Lodge-related fisheries are defined as mic management objectives. In addition to the remote access fisheries where anglers are under data defined above other economic data are need- the "control" of the lodge operator. Outcamps, ed on a periodic and specific study basis to such as those on the Tree River, are also con- estimate the economic benefits and costs asso- sidered lodge-related fisheries. Lodge-related ciated with the recreational fishery in the fisheries are characterized by having a well- NWT. These economic benefits and costs, gene- defined and controlled user population. rated at both primary and secondary levels, have been defined for the purpose of this study (Cau- Road-access fisheries vin 1982) as:

Road-access fisheries are defined as any - PRIMARY BENEFITS: These benefits are fishery that has ground access. Examples of the "final" demand for the recreation these fisheries include the Yellowknife and Hay use of the fisheries resource. Final River areas and the community sport fishery at demand can be interpreted as the inter- the Sylvia Grinnell River. These fisheries are face between the final consumer - the characterized by having one or more ground angler - and the supplier of the pro- àccess points, with an undefined and uncon- duct - in this case the public sector. trolled user population. Because access to recreational fishing has traditionally been effectively non- priced, the "final" demand or primary TYPES OF DATA benefits are measured as the willing- ness of anglers to pay. It therefore The objective of "ensuring optimal social may incorporate two elements; what they and economic benefits .. from the utilization actually pay (i.e. nominal licence and conservation of fish ... in the Arctic" fees) and what they would be willing to (Cauvin 1981) implies a degree of complimenta- pay if they had to. rity between biological and economic analyses. - PRIMARY COSTS: These costs include the In this regard, many of the data requirements public supply costs and the opportunity are identical. costs that arise as a result of alloca- ting fisheries resources to recreation- Biological information al rather than commercial or domestic fishing. The basic types of data are defined as - SECONDARY BENEFITS: This is defined as follows: the demand for goods and services such as food, lodging, and gear associated STOCK: Total numbers or weight of fish with recreational fishing. present in the waterbody at a given - SECONDARY COSTS: The overall costs of time. Fish stocks must be monitored providing the ancillary goods and ser- periodically to determine numbers and vices such as food, lodging, and gear stock composition (age and length dis- are defined as secondary costs. tributions) in order to assess the effects of exploitation. In developing economic statistics for the CATCH/UNIT EFFORT: Total catch divided sport fishery, the initial focus should be on by total effort. These data will gene- the primary benefits and costs of those portions rally be required on an annual basis of the sport fishery that require intensive for selected species for selected management. These benefits and costs could be 3

analyzed through the development of biological All of these methods can be characterized production functions, conversion of the biolo- by the time at which the data is collected in gical production functions to economic yield relation to the time at which the actual fishing functions, and the calculation of the net value experience occurred. These differences will of the fishery (Cauvin 1982). influence the effectiveness of the methods. Mail questionnaires and telephone surveys are The biological production functions in delayed data collection methods in which users this methodology consist of establishing the are often questioned several months after the relationship between population size, annual fishing occurred. Active creel censuses are an yield, fishing effort, and harvest success per immediate, on-site data collection method. For angler day. This is exactly the same data the other methods, the delay between the fishing required for biological management. The use of experience and the time the data is collected one set of data to meet both biological and eco- varies between these two extremes. nomic management objectives allows administra- tors to optimize allocation of personnel and resources to achieve specific management objec- DELAYED DATA COLLECTION SYSTEMS tives (Gordon et al. 1973). All sample surveys contain a certain pro- Because there is no market system asso- portion of errors; some of them are sampling ciated with recreational fishing, the develop- errors and some nonsampling errors. Mail ques- ment of economic yield functions for sport tionnaires and voluntary returns of creel census fisheries has always been a particularly reports, logs, diaries, and permits all suffer troublesome problem. Some conservationists and from nonresponse bias. Erdos and Morgan (1970) scientists have opposed the use of economic as cited [on p. 444] in Filion (1980) reported techniques for placing a dollar value on non- that the Advertising and Research Foundation market resources (Seckler 1966). Two tech- recommended that an 80% or better response rate niques, based on the willingness-to-pay concept, be targeted for mail surveys in order to mini- have been developed to compare relative benefits mize the problem of inaccurate population esti- of resources and allocate limited resources mates. This level of response yields a rate of among users. These have become widely accepted non-response that is in line with the rate of techniques for estimating the net primary bene- substitution for well-conducted personal inter- fits of recreational fishing (McConnell and Nor- view studies. At this level, it is thought that ton 1975; Dwyer and Bowes 1978) and a number of the precision of estimates will converge for the other resources (Kreutzwiser 1981; Pearse two methodologies. Standard references, such as 1968). The first willingness-to-pay method, Cochran (1977), on the principles and technique derived from travel costs, is based on estimates of sampling describe how to identify and mini- of demand and value that are generated from mize these errors. Filion (1980) reviewed the observed participation or visitation rates. literature on nonsampling errors and focused on Travel costs in this method serve as a proxy for techniques to identify and minimize nonresponse price. The second method, direct surveys, and response biases in human surveys. involves asking fishermen the maximum amount they would pay for the right to use the resource There is considerable evidence from a (equivalent variation) or the amount that would variety of scientific disciplines that differ- be required for them to forego the right to use ences may exist between the true answer to a the resource (compensating variation). question and the respondent's answer to it. Research in experimental psychology indicates an If the demand curve calculated from this inverse, exponential relationship between an willingness-to-pay data is interpreted as a mar- individual's ability to retain information and ginal benefit function, the net value of the the time elapsed since the event occurred. The fishery can be calculated as the total benefit ability to recall events may be influenced by minus the total primary costs. The primary motivation or interest, emotions or perceptions costs include the public supply costs as well as during the event, and the frequency of use or the opportunity costs associated with allocating recurrence (Hilgard and Atkinson 1967). Cannell fisheries resources to anglers. et al. (1977) concluded that, as the time between the event and the questioning increased, (a) there was increased under-reporting of information about that event, (h) events impor- DESCRIPTION AND EVALUATION OF DATA COLLECTION tant to the respondent were reported more com- SYSTEMS pletely and accurately than those of less importance, and (c) reporting of an event was Various methods of collecting fisheries likely to be distorted in a socially desirable data have been reported in the management and direction. scientific literature over the past few years. These methods, designed to collect the biolo- Ashraf et al. (1975) as cited in Kazes gical and economic data requirements have taken (1980) reported a 25.83% increase in the total a variety of forms, including: active and pas- number of trips in the Canadian Travel Survey sive creel censuses, voluntary creel censuses, using a three month rather than a four month licence reporting stubs, species permits, indus- recall period. Kazes (1980) reviewed the try logs, fishermen diaries, mail question- problem of recall bias in several U.S. naires, and telephone surveys. Reporting government surveys. The U.S. Bureau of the methods used by 13 resource management agencies Census conducted a pretest in Washington, D.C. in the western United States prior to 1974 are to study recall abilities of crime victims. summarized in Table 2. Eighty-six percent recalled the incident within 4

a three month recall period 78% recalled it in survey approach is now used for the marine six months, and 70% recalled it within 12 recreational fisheries survey. This modified months. The U.S. Bureau of Labor Statistics has approach uses a telephone survey to collect data tested the accuracy of recall for purchases of on enumeration of fishermen, trips by mode of durable goods. Large items were under-reported fishing, trips by geographic locations and area by 10% for a three month recall period, by 41% fished. The approach uses an intercept survey for a 4-6 month recall period and by 47% for a to collect data on residence of fishermen, catch 7-12 month recall period. In addition, there per trip, and catch by species (Deuel 1980). was a 54% increase in average reported expenditures using three month vs six month Simpson and Bjornn (1965) compared results recall periods. The National Crime Survey, obtained from postal questionnaires, permits conducted by the Bureau of the Census from 1973 returned voluntarily, random samples of permits, to 1975, reported 16% to 30% more crimes for a and check stations for steelhead trout harvest three month recall period than for a six month in Idaho. Postcard questionnaires had been used recall period. A health interview study to estimate the statewide annual harvest of chi- conducted in 1967 by the Department of Health, nook salmon and steelhead trout since 1954. Education and Welfare found that 98.6% of known While these estimates were statistically precise accident victims recalled the incident within 3 (951 CI-± 101 of estimate), the high harvest months, 90% recalled it within six months, and rate that resulted from these estimates was only 77.7% recalled it within 12 months. unrealistic, and questionnaire estimates of har- vest have ranged from 1.8 to 3.4 times the num- A number of studies that investigated ber of fish counted at check stations. Starting recall and nonresponse bias in fish and wildlife in 1962, each angler fishing for steelhead was surveys found that instead of under-reporting, required to have in their possession a steelhead many of the surveys over-reported harvest. This permit to record the date and place of each may be partially due to the prestige factor steelhead caught. In 1964, harvest estimates reported by MacDonald and Dillman (1968). were compared to check station data for the South Fork of the Salmon River. Estimates based Hiett and Worrall (1977) studied the abi- on a random sample of permits were very close to lity of marine recreational fishermen to esti- check station values (737 vs 726). The estimate mate the nature of their catch and to recall from voluntarily returned permits was 1.4 times their total catch and effort over a period of the random sample estimate, and the question- time. This study showed that: naire estimate was 2.4 that of the random sam- ple. Nonresponse bias only accounted for 161 of - Fishermen were unable to identify spe- this overestimate. Consistent with other stu- cies of fish with a satisfactory degree dies, a much smaller percentage of anglers in of accuracy. the random sample group had fished and caught - Fishermen made errors in estimating fish, and they had a lower mean annual catch weight and length of fish. Weight than those in the voluntary return group. tended to be overestimated for smaller fish and underestimated for larger Carline (1972) and Jones et al. (1977) fish. Length tended to be underesti- reported biased results from passive creel cen- mated, regardless of the size of the suses compared to active interviews. In an fish. In addition, 22% incorrectly attempt to estimate the harvest of brook trout reported the number caught. in Wisconsin, Carline (1972) provided self- - Recall of specific fishing dates was addressed postcards to fishermen on-site, with a poor, although the total number of request to complete and return the form. The fishing trips was often estimated with results demonstrated a significant positive bias reasonable acccuracy. Fishermen were in the postal survey data, with catch rates from able to report dates for trips further the postal returns double the catch rates from back in time than for which they were anglers interviewed on-site. Only 501 to 601 of able to give detailed fishing informa- the postcards were returned, and the discrepancy tion. By the end of 60 days, fishermen between the postal survey data and the on-site were unable to provide detailed infor- interview data was attributed to a failure to mation on approximately one-fourth of return postcards by unsuccessful anglers and their reported trips. anglers who caught only a few fish to return - Fishermen tend to overestimate by a postcards. He also suggested that multiple considerable degree the effort expended reporting on postal cards may have contributed For fishing trips. to the bias. He concluded that the length of - Fishermen tend to overestimate by a time the census covers, the number of waters considerable degree the total catch for involved, the number of species in question, and their fishing trips. the proportion of surveys returned were all important considerations in determining the Hiett and Worrall (1977) concluded that, to reliability of postal survey data. improve accuracy of marine recreational fisher- men survey, as much information as possible Jones et al. (1977) comparing a voluntary should be collected from an on-site survey, and registration system with direct interviews on a that the recall period for reporting trips Texas pond fishery, found that respondents to should be no longer than 60 days. the passive survey reported only 17 1 of the num- ber of fish estimated by the active survey. The Charbonneau (1980) reported that as a weight of fish reported in the passive survey result of the above work and other experience was, however, nearly 70 1 of the weight estimated with the National Survey that a modified two- by the active survey. They concluded that these 5

discrepancies probably resulted from the tenden- them to anglers. Anglers were asked to complete cy of fishermen to report only catches of large the forms after fishing and to deposit them in a fish, to overestimate the size of fish taken, or container in their cabins. Only about a dozen both. The passive survey also provided an esti- forms were completed and returned. The poor mate of total fishing pressure that was less response was attributed to the lack of a person than 10% of results derived from the active on-site who was responsible for ensuring that survey. the program worked, and an inadequate education program to impress upon lodge owners and anglers A number of other researchers have also the importance of the program. found biases in postal surveys of fisheries catch and effort. Jensen (1964) compared postal The DFO, Western Region, tried a similar surveys with random samples of fishing vessels type of voluntary creel census at Great Bear and and with log books of party boats. He found Bear Island lodges (with some returns), and at that harvest estimates from the postal surveys Kasba Lake (with very few returns). Neither of were three to five times higher than harvest these voluntary programs was considered to have estimates derived from the on-site samples. provided adequate information on the fishery.

For the past 13 years, the Michigan The state of Alaska has used a combined Department of Natural Resources has conducted postal survey and creel census program since statewide mail surveys of licensed anglers to 1977 to obtain estimates of effort and harvest estimate sport angler effort, catch, and parti- by fishery and by species (Mills 1981). The cipation. During this period, numerous on-site postal surveys consist of repeated mailings of creel censuses have also been undertaken to questionnaires to random samples of Alaskan assess management actions and to provide an in- sport-fishing licence holders. Creel censuses depth assessment of unique and/or economically consist of on-site interviews of random samples important fishing waters. Department personnel of sport fishermen. The postal surveys provide recognized that both methods were vulnerable to estimates for all the fisheries, while the creel various biases; however, they concluded that censuses are only used on major, selected their mail survey estimates for both catch and fisheries to provide validity checks. The 1980 effort generally are two to five times as high postal survey, which involved 12,000 question- as those obtained from an on-site creel census naires with two follow-up mailings, produced a (Gayle Jamsen, personal communication). total return of 66%. Mills (1981) concluded that the 1980 surveys continued to provide an Murphy (1954) found that the rate of effective method of estimating statewide effort return of the postal surveys influenced the and harvest for major Alaskan fisheries. final results of his surveys. Early respondents to his survey reported a considerably higher Since the early 1950's, the DFO, Pacific catch per unit effort than late respondents on Region, has relied on estimates of sport catch the follow-up returns. He finally received made by fisheries officers using a variety of total questionnaire returns of 84%to 91%, and judgemental techniques (MacGregor 1982). During concluded that at the final higher level of the mid-1970's, several studies (i.e. Argue et return he could find no evidence of bias in the al. 1977) indicated that the annually collected postal survey. statistics on sport fishing catch and effort in DFO's Pacific Region were underestimating the The DFO, Western Region, attempted a actual values by as much as two to four times. voluntary questionnaire survey in the Yellow- Consequently, the traditional techniques were knife and Hay River areas in 1979 to determine abandoned, and in mid-1979 the Tidal Sport- locations fished, number of fish caught and fishing Diary Program was instituted (MacGregor retained, hours fished, and fishing quality 1982). (Falk 1981). Only 16% of the questionnaires handed out were returned. Although the author The diary program, which is still in noted that estimates of catch per angler and operation, consisted of an initial random ques- catch per angler-hour compared favorably with tionnaire survey to British Columbia residents. those determined previously using creel cen- This questionnaire is used to collect informa- suses, he also noted that the catch per angler tion on average days fished and average catch. tended to be higher for the questionnaire The data is used to quantify nonresponse bias method, which is consistent with other studies from the diary fishermen returns; it is not used using passive or voluntary creel censuses. Falk to estimate total catch and effort. A telephone (1981) concluded that although the questionnaire follow-up is used to correct for the nonresponse may provide a broad picture of sport fishing in bias to the questionnaire survey. the study area from which to develop management strategies, it does not provide the information A selected number of sports-fishermen who needed to determine total fishing effort and agree to keep a daily record of fishing activity total harvest. are sent a total of four diaries during the year. Diaries are mailed in for analysis quar- The Manitoba Department of Natural terly, and are returned to the fishermen for Resources attempted a voluntary creel census at their personal use when the data has been tabu- five remote lodges (Fishing, Sasaginnigak, lated. Anglers are requested to record fishing Aikens, Dogskin, and Moar lakes) in 1974 (W. activity as it occurs, including the date of Hayden, Manitoba Department of Natural Resour- each trip, the number of other anglers in the ces, 1495 St. James St., Winnipeg, personal com- party, target species, area fished, type of munication). Recording forms were distributed fishing, and the size categories of the fish to lodge owners who were asked to distribute kept and the total caught. Instructions and 6

maps of the major fishing areas are included in portion of the sampled population (Cochran the front of the diary. 1977). Filion (1980) recommended combining two approaches to minimize problems of nonresponse; Several problems still exist with the maximize response, and correct for bias by diary data. In 1980, about 75% of the people extrapolating across successive mailing waves or sampled responded to the postcard question- weighing responses to follow-ups. naire. Of these respondents, 75% agreed to keep a diary and about one-half of this group return- Recall bias is a major problem of delayed ed their diaries every quarter. Telephone data collection surveys that must be taken into reminders were used to contact fishermen who had account in any decision on methodologies to col- not returned their diaries. Less than one-third lect statistics on catch, yield, effort, and of the diary fishermen who returned their books catch per unit effort. Standard techniques are actually went fishing in 1980, even though they not available to address the nature and magni- intended to at the beginning of the year. The tude of the recall bias. Two mathematical tech- telephone reminders indicated that anglers who niques have been proposed to adjust for recall did not return their diaries usually had not bias in post-season game harvest surveys; how- gone fishing. As a result the diary program ever, Filion (1980) suggests that both tech- still contains a response bias in that anglers niques have limitations and should be used with who keep and return diaries fish more often and caution. He suggested that experimentation with are more successful than other anglers. shorter recall intervals, panels, or even the diary method might yield some effective solu- In the 1980 diary survey, 39% of the ang- tions. lers caught no fish, and 10% of the anglers cau- ght 54% of the total salmon catch. The major Even if the recall bias can be accurately concern with the diary program estimates is the defined, the cost of providing sound statistical high catch per unit effort obtained: approxi- estimates of the data correction factors could mately double that of any other study, including exceed the cost of collecting the original the preliminary questionnaire. A portion of data. Because of the nature of magnitude of the this overestimate was attributed to respondents recall bias problem, we concur with Hiett and overreporting individual catch by using boat Worrall (1977) that as much information as pos- totals. This is being alleviated by asking for sible should be collected from on-site surveys, boat rather than individual catches in subse- and that the recall period for reporting of quent surveys. Most of the overestimates were trips should be as short as possible, preferably caused by respondents who underreported the num- no longer than 60 days. ber of days fished. It was concluded that ang- lers tended not to report fishing trips that Voluntary or passive creel censuses, if were unsuccessful . The 1982 program therefore conducted essentially on-site, could reduce the placed more emphasis on encouraging participants recall bias problem. They do, however, suffer to report unsuccessful days by improving the from a major non-response bias (Falk 1981; Car- design of the diary and the accompanying line 1972; Jones et al. 1977). This problem, instructions. combined with the fact that voluntary or passive creel censuses are not directly capable of pro- MacGregor (1982) concluded that despite viding information on total harvest or total the small number of fishermen involved, the effort, makes them unsuitable as the major com- diary program produced reasonable estimates of ponent of a data collection system in the NWT. annual catch. Solutions to the low estimates of effort are being tested. A major attraction of The diary system as described by MacGregor the diary program is the opportunity to obtain (1982) for the tidal sport fishery in British adequate and inexpensive estimates of sport Columbia has some advantages over the methods catch and effort without placing additional bur- described above. Not the least of these is the dens on field staff. opportunity to obtain adequate and inexpensive estimates of sport catch and effort without Conclusions placing additional burdens on field staff. The design could theoretically eliminate the recall It is apparent from the literature that bias problem; however, the existing implementa- techniques using delayed data collection methods tion of the diary program for the tidal sport can suffer from significant nonresponse and fishery still contains a number of potentially recall bias. serious biases due to poor return rates and the small number of active anglers participating. Nonresponse is a universal problem asso- ciated with surveys that require voluntary Consequently, we recommend that on-site responses from the sampled population (see Coch- surveys, rather than delayed data collection ran 1977). Most of the authors reviewed for surveys, be used to collect lake-specific catch, this study (Carline 1972; Jensen 1964; Simpson yield, effort, and catch per unit effort infor- and Bjornn 1965; Murphy 1954) found that the mation for the management of the road-access rate of return of voluntary questionnaires recreational fisheries in the NWT. Existing influenced the final results of their surveys by diary programs such as the Tidal Sportfishing several hundred percent. Murphy (1954) was the Diary Program are not directly applicable to the only author who could find no bias in his final NWT, primarily because of the problem of allo- results; however, his total questionnaire return cating diaries to sport-fishermen when approxi- rate was 84% to 91%, unusually high for resource mately 50% of the anglers each year are new user surveys. Statistical techniques are avail- (Topolniski 1975), and the problem of collecting able to provide estimates of the nonresponse adequate data on yield and effort for specific water bodies. 7

Because of the potential advantages and pointed out that user participation bias can cost savings of a diary system, however, we occur because certain ségmerits of the population recommend a modified controlled diary program have an excessively high probability of being for the remote access lodge fisheries. The pro- included in the sample. This bias results from posed system would essentially eliminate the some persons staying Tonger at the site on each bias problem in the tidal diary program, elimi- visit than others (length of stay bias) or some nate the diary allocation and specific area data people visiting the site more often than others collection problems, and provide an effective (frequency of use bias). Both forms of bias can system for collecting catch, yield, effort and be eliminated if reliable information on the catch per unit effort statistics. probabilities of selection in the sample survey can be established. This requires that an ana- It is not uncommon in resource management lyst have an independent estimate of the target to monitor wildlife populations for general population and a reliable estimate of the proba- management purposes using surveys that are known bility distribution of the survey interviews to be consistently biased. Mail questionnaires, over the activity groups. Malvestuto et al. therefore, may be useful to monitor general (1979) used a multiple regression model of cli- trends and changes in catch and effort for por- matic variables to explain some of the variation tions of the road-access recreation fishery if associated with creel survey estimates of fish- the response and recall biases can be held con- ing effort, and to provide a means of optimally stant. Such a survey of sport fisheries in the allocating sampling effort prior to impleinenta- NWT could provide trend information for long- tion of a survey. term management, and could be used to identify site-specific areas requiring the intensive, Robson (1960) presented the first detailed accurate surveys recommended above. creel census procedure for achieving unbiased sampling and estimating of fishing success, total effort, and total harvest using a strati- ON-SITE DATA COLLECTION SYSTEMS fied random sample. He subsequently adapted this procedure to a roving creel census (Robson The concept of "creel censuses" has been 1961). Regier (1966) adapted Robson's method to used by fisheries managers for more than half a a field survey, and Hayne (1966) and Malvestuto century to collect management information on et al. (1978) have incorporated nonuniform pro- recreational fisheries. These surveys have tra- bability sampling into the stratified random ditionally been used to collect data on total approach. fishing effort, total catch, catch per unit effort, and species composition in the catch A complete census of the fishermen on a (Moyle and Franklin 1957; Erman 1972; Gard and particular lake or river over the complete fish- Seegrist 1972; Farr and Blake 1979; Panek ing season is usually impossible due to limita- 1981). They have also been used in recent years tions on manpower and funds. Most creel cen- to determine fishermen's attitudes, user pat- suses are therefore designed as sample censuses, terns, and socio-economic attitudes towards and a ratio-type method of estimation is employ- development projects (Sinclair and Morley ed to estimate the total catch (Robson 1960). 1975). The term creel census is used in the This method consists of applying an estimated generic sense in the following discussion to catch rate to the estimated total effort. It is describe any direct interview survey of the based on the fact that the number of fish caught recreational fisherman, including intercept, per fisherman is correlated positively with the location, on-site or creel surveys. number of hours fished, and that total fishing time by all fishermen can be estimated from Mottley and Embody (1940), who evaluated easily obtained random counts. The method pro- several creel census techniques, emphasized the vides a statistically unbiased ratio-type esti- importance of any partial creel census meeting mate of total catch and total fishing effort and the criteria of representativeness, adequacy, of the sampling error variance of the esti- and cost. Carlander et al. (1958) who also mators. reviewed existing creel census methods, stressed the importance of applying detailed sampling Robson (1961) modified this method to techniques to creel surveys and emphasized that accommodate the "roving creel census" in which "only when sampling is done according to some the enumerator moves through the fishing area statistical scheme can reliability of the data interviewing fishermen as they are encountered. be measured and certain biases be avoided". Some unique problems associated with the roving creel census are that the sample size depends on A number of physical and external factors the number of distribution of fishermen, the can influence the outcome of a creel census, sample is systematic rather than random, the including the weather, irregular concentrations probability of interviewing a given fisherman of fishing effort, and variations in availabi- depends on how long he fishes, and usually only lity of species (Best and Boles 1956; Malvestuto incomplete information is available from each et al. 1979). fisherman. The major weakness of the roving creel census using incomplete fishing trips to Sinclair and Morley (1975) described a estimate total catch is that the bias of estima- number of other factors in the survey that could tion depends on the basic nature of the stochas- cause erroneous results due to the effects of tic fishing process which, in general, is not disproportionate representation. The factors known (Thompson 1976). The only sure way to include poor questionnaire design, interviewer make the creel census distribution-free is to error, nonrandom sampling, improper survey station interviewers at access points to obtain design, and response bias. In addition, they information on completed trips. 8

Malvestuto et al. (1978) described an HYBRIMATA COLLECTION SYSTEMS application of the roving creel survey with non- uniform probability sampling. The paper serves The literature clearly identifies major as an example of how a statistically based creel problems using delayed data collection tech- survey can provide information required for niques to collect accurate site-specific catch fishery management as well as information needed or yield information. These problems may be to evaluate the design. Their method is evalua- less severe if these techniques are used to col- ted by testing the assumption that the catch per lect effort statistics, particularly if the time unit effort for incomplete fishing trips is an interval between fishing trips and the question- unbiased estimator of the catch per unit effort naire is limited (Hiett and Worrall 1977). for complete trips. The basic features of the application of nonuniform probability sampling Since effort is the most difficult compo- to the roving creel census are: nent of the on-site creel census to collect, a combination system using a mail questionnaire to - The entire period for which the fishery collect effort data and an on-site survey to is to be surveyed is divided into time collect catch per unit effort data could be con- blocks which should contain a similar sidered for the road access fisheries. There amount of expected fishing pressure. are however, four major problems associated with - Each time block is divided into sam- implementing this combined system to collect pling units such that all of the fish- required recreational fishery effort data in the ing time within a block is contained NWT: (1) nonresponse bias; (2) recall bias; (3) within the sampling units and the units sample frame, and (4) timeliness of information. do not overlap. - Sampling probabilities proportional to The nonresponse bias is a standard problem the amount of fishing expected are of mail questionnaires. Previous information assigned to the sampling units. The for the NWT (Topolniski 1975) indicates that sum of the probabilities within any relatively good response rates might be expected block is 1.0. from a simple questionnaire; however, follow-ups - Sampling units are randomly chosen and sub-sampling of the nonrespondents will be within each block on the basis of the required since, as noted earlier, effort and assigned sampling probabilities. response appear to be correlated in similar sur- - Sampling on the water body is comprised veys. Recall bias for effort is related to the of counts and interviews. Counts timeliness of the questionnaire and the amount should be as instantaneous as possible, of information requested. It would not be rea- and the interview should be representa- sonable to ask potential respondents to provide tive of the water body being sampled. any more detail than for days fished during the survey period. If information was required on Malvestuto et al. (1978) concluded that catch per angler hour, a separate survey would this method appeared capable of detecting chan- be required to provide a statistically sound ges in the quality of fishing small enough for estimate of mean angler hours per day. management purposes. They stressed the impor- tance of measuring precision of the survey so Although there are a limited number of that the efficiency and sensitivity of the sur- fishing licence vendors in the NWT, the problem vey design would be evaluated. of identifying a proper sampling frame to col- lect effort data for a particular species on a Conclusions particular lake during a given year is non- trivial. The current licensing system does not After describing the problems associated lend itself to defining such a specific sampling with recall bias, we concluded that on-site sur- frame and a very large number of questionnaires veys are required to collect statistically would be required to achieve adequate sample sound, site-specific catch, yield, and effort size for a particular lake or river. data. Robson (1960, 1961) and Malvestuto et al. (1978) have adequately described the overall Finally, the timeliness of the question- structure of a creel census that will provide naire is critical to minimizing recall bias. sound statistical estimates of total yield, The existing licence system is currently not total effort, and catch per unit effort. Robson capable of providing the data in the time (1961) pointed out problems that can be asso- required (Topolniski, personal communication). ciated with incomplete interviews during roving Even if it were modified, it is questionable creel censuses and Malvestuto et al. (1978) pro- that it could be timely enough (e.g. within 60 vided a mechanism to test for these problems. days of fishing season) to limit recall bias to an acceptable level. Consequently, we have recommended a creel census for road access fisheries, which will be Conclusions described in detail in a later section that is a combination of the above two approaches. It Mail-out questionnaires with their atten- uses an intercept census to determine catch per dant problems of response, recall and timeliness unit effort as suggested by Robson (1961), and are a relatively poor method of providing sta- incorporates some of the temporal and spatial tistically sound, site-specific catch and effort stratifications suggested by Malvestuto et al. data. Nevertheless carefully designed mail (1978). The design eliminates the length of questionnaires may be an appropriate method of stay and frequency of visit bias discussed by providing general and trend effort data on sport Sinclair and Morley (1975). fisheries on either a territorial or sub-territorial (regional) basis. 9

RECOMMENDED DATA COLLECTION SYSTEMS naire data or other sources as requiring more detailed information, we recommend using one of The NWT contains a wide diversity of sport two systems, depending on the type of fishery. fisheries. Two classification systems have been A modified diary program is recommended to col- developed during this study to characterize lect site-specific data on remote access fisher- these fisheries. The intensive, intensive/ ies, and a catch/effect census is recommended to extensive, and extensive classifications charac- collect site-specific data on road-access terize the resource use pressure on the fishery fisheries when required. Application of these and, indirectly, the frequency and type of data collection systems is described below. management information that is required. The Both systems provide the mechanism to collect lodge-related and road-access classifications site-specific biological data, catch and effort characterize the type and geography of the fish- data, and economic data. ery that directly influence the type of data collection program required. It should be noted that the two data col- lection systems recommended for the intensive/ These two classifications were integrated extensive fisheries are not alternative methods into the analysis of the sport fishery data of collecting the same information; they are requirements for the NWT. Based on this analy- separate and complementary systems. Alterna- sis, a three-tier statistical reporting system, tives to the mail questionnaires for collecting shown in Fig. 1, is recommended to provide the general information and monitoring trends necessary management information. include the unstructured field checks recom- mended for the extensive fisheries, or systems The recommended reporting system utilizes using report forms, permits or diaries. four different types of data collection systems: (1) unstructured field checks are recommended While structured field checks are useful for providing general information on extensive for general information they are not capable of fisheries; (2) mail questionnaires are recom- providing unbiased or consistent data with which mended for gathering general data and monitoring to monitor trends. Report forms or permits trends in catch and effort for intensive/ issued with a licence are a traditional way of extensive fisheries; (3) a modified, controlled collecting resource use or harvest data. Their diary system is recommended for collecting sta- advantages are that they have a relatively high tistically sound, site-specific information on public acceptance and very low cost per response the remote-access lodge fisheries, and (4) a (<$1.00 compared to $1.00-$3.00 per question- catch/effort census is recommended for collect- naire response) (Aney 1974; Jamsen 1971). The ing statistically sound, site-specific informa- report forms or permits have an uncontrolled tion on the road-access fisheries. response for information on specific areas; and they suffer from a very low response rate, high We recommend that the current system of nonresponse bias, and varying rates of recall unstructured field checks continue to serve as bias. In addition, they are primarily designed the data collection program for the extensive to collect harvest and effort information and fisheries of the NWT. Experienced field person- are not well suited for collecting other data nel are familiar with their region and, through such as attitudes or economic data. This lack their public contacts, know what is going on of flexibility, combined with relatively large with the resource and its users. Collection of and uncontrollable biases, precludes their con- general information on type and amount of use, sideration as viable alternatives to mail ques- and trend information on effort, success rates tionnaires for collecting general information and catch can be integrated with public informa- and monitoring trends for the intensive/ tion and law enforcement functions with minimal extensive sport fisheries. The diary system has additional costs. many of the same advantages and disadvantages of the reporting forms and permits discussed We recommend that two data collection sys- above. Because of this, and because of the tems be used for the intensive/extensive sport large turnover in NWT licence holders, the diary fisheries. A system of mail questionnaires system is not a viable alternative to the mail should be instituted to collect general informa- questionnaire for intensive/extensive fisheries. tion on the intensive/extensive sport fisheries and to monitor trends in important biological For the intensive sport fisheries we and economic management parameters. This ques- recommend that a modified and controlled diary tionnaire system should be designed for per- program be used annually to collect site- iodic, annual or special use, with a sample specific data on the remote access fisheries, frame stratified by region or area of recrea- and that a catch/effort census be used to col- tional fishery. This questionnaire system will lect site-specific data on the road access require a redesign of the NWT licence system and fisheries. The site-specific data collection will require an administrative and sampling system for the intensive fishery, therefore, design to minimize and make consistent the res- differs fran the intensive/extensive fishing in ponse and recall bias inherent in such a sys- the frequency of application of data collec- tem. The results of these mail questionnaires tion. Since both systems we are recommending should be used to facilitate long-range, for the intensive fishery include the mechanisms regional management decisions and to identify for collecting site-specific biological, catch areas where accurate site-specific information and effort, and economic data, an annual or is required. periodic mail questionnaire is not required. Mail questionnaires can be used, however, in For intensive/extensive sport fisheries special circumstances where other information that have been identified through mail question- such as user attitudes or demographic characteristics are required. 10

MAIL QUESTIONNAIRES SITE-SPECIFIC DATA COLLECTION SYSTEMS

A mail questionnaire can be designed and The basic statistics required to manage administered such that it provides trend data on the recreational fishery are stock, catch, territorial and regional sport fisheries. Four yield, effort, and catch per unit effort. This major administrative and analysis problems asso- study has focused on methods of collecting ciated with such a survey must, however, be statistics on the latter four. For a given overcome for the mail questionnaire to be a use- fishery, let: ful and reliable management tool. D = Total number of days of the season The present licence system is not capable d = Total number of days sampled per of providing an adequate sample frame. Licences season frequently do not contain adequate address Nj = Total number of users on day j, information and are completed in a manner that j = 1,2 ...d makes most of the information unrecognizable and nj = Total number of users sampled on unusable. The content and design of the day j, j= 1, 2...d licences must be reviewed and a revised licence hij = Number of hours user i used the form developed that enables respondents to write f i shery on day j legibly and to provide the required informa- kijs = Number of species s caught and tion. The Manitoba Wildlife Certificate is a kept by user i on day j useful starting model. An administrative system rijs = Number of species s caught and should also be developed that penalizes vendors released by user i on day j for any spoiled licences. lijs = Number of species s used for a shore lunch by user i on day j The current licence system is not capable of providing a timely sampling frame; delays of Utilizing the above, a census design for a between 18 and 24 months now exist between the given fishery should provide estimates for the time the licences are sold and the final compi- following population parameters: lation of the licence sales. The existing sys- tem must be revised to make licence data avail- T + lijs) + x able following the end of the current fishing F.FF (kijs FF.F rijs season. ijs Us where T = total harvest over the season for all A questionnaire program must be developed species that has clear management objectives and data x = mortality ratio for species s caught requirements and that specifically addresses the and released, and problems of nonresponse and recall bias. The program should deal specifically with the data E = SF hij required, the amount of respondent preparation and involvement required, the design of the ij questionnaire, pre-tests of the questionnaire, and the number and types of follow-up and moti- where E= total fishing effort for the season, vational aides required. Filion (1980) has and = T/E reviewed procedures to increase response and C = harvest per unit effort. concluded that high response rates do not neces- sarily mean added costs but definitely provide similarily: important benefits. He pointed out that it costs no more to administer a well-designed t E^F (kijs + lijs) + x SFF rijs questionnaire than a poorly designed one. The ijs Us questionnaire program must also be designed to minimize and control the effects of recall where t sample harvest for d days, and bias. The major effort should be toward making the effects of the recall bias as constant as t= t/d is the mean sample harvest, and possible from one survey to the next. For exam- ple, the literature reviewed shows that recall e = ^f hij bias is highly correlated with time. The mail ij questionnaire must, therefore, be administered consistently to minimize year to year variabi= where e = sample total fishing effort for lity. Hiett and Worrall (1977) and Kazes (1980) d days, and have provided data to judge the effects on accu- racy of delaying the survey for different time =e/d periods beyond the fishing season. Consistency where é= sample mean fishing effort in design, administration, and interpretation will be the key factor in increasing precision and producing reliable trend information. It is useful to identify four distinct types of recreational fisheries for the purpose Finally, the analysis and interpretation of obtaining census information on T, E, and C: program for the mail questionnaire date must remote-access lodge fisheries, limited-access day fisheries, make it abundantly clear that, until nonresponse multiple-access day fisheries, and recall biases can be either eliminated or and extended-stay fisheries. These classifica- adjusted for, the mail questionnaire can only be tions differ with respect to the methods of used to monitor trends. utilization and, from a statistical view, the ability to enumerate and obtain information from 11

the total population of users at any point in This lodge fishery data collection system time. is designed to obtain total population information so that statistical inference Lodge fisheries are controlled access calculations (e.g., standard errors) should not fisheries where use must be by arrangement with be needed. One possible complication, however, a particular lodge or similar commercial enter- is that some of the required information will be prise. The total population of users can be missing. For example, individual users may fail identified at any point during the season by to turn in a particular day's log or may record enumerating lodges within the study area and only partial information on a few catches. If guests at each lodge. the amount of missing data is small, standard imputation methods can be used. On the other Day fisheries with limited access are hand, if considerable data is missing, fisheries where virtually all users arrive and appropriate statistical methods will have to be depart the same day. The number of access developed to obtain population estimates and points is small enough to allow continuous moni- assess their accuracy. toring on any chosen day; however, monitoring throughout the season is not practical. The population of users for a given lodge fishery Establishing statistical methods at this can, therefore, be measured for the entire point in the design of the reporting system to fishing season, whereas it is only practical to cover all possible patterns of missing data measure the population of users for a limited- would be costly and premature. Instead, the access day fishery on a sample of days. likely patterns of missing data should be determined through a one-year pilot program that For the remaining two classifications, it should be designed to test and refine the basic is not practical to obtain information from the methodology. total population of users at any point in the season. The number of access points for Limited-access day fisheries multiple-access fisheries is too large to allow for continual monitoring, and extended-stay In this category of fishery, complete fisheries require complete coverage on any given information on a single day's fishing activities day or continual monitoring of all access points can be obtained by a creel census that involves throughout the season to obtain information on stationing interviewers, possibly in shifts the total population of users. throughout the day, at the access points to the fishery. As users leave the fishery, the Remote-access lodge fisheries interviewer will record the access point identification, the fishery identification, the Total population information can usually date, and, for each user, the total number of be obtained from lodge fisheries because access fish kept, the number used for shore lunch, the to such fisheries is completely controlled. A number caught and released, and the total total census requires the cooperation of the fishing time. lodge managers, the individual guides and, of course, the users. The basic data collection Because fishing pressure is known to vary system we suggest for the lodge fisheries is a in some areas, depending on the day of the week, modification of the Tidal Sportfishing Diary it is desirable to stratify the sampling units suggested by MacGregor (1982). In this system (days) into two strata, where sl = weekdays and each user records his effort and success in a s2 = weekend days and holidays. The required logbook from which a duplicate carbon copy is information should be collected on a random collected at the end of each day. As an incen- sample of di days from stratum si, where i = tive to participate, the users will be allowed 1, 2. Depending on the length of the season and to retain the original log books as a record of the relative number of users in sl and s2, it their fishing trip. The basic information to be may be desirable to sample s2 exhaustively since collected in the log books is that needed for the fishing pressure in s2 will likely be much computing T, E, and T/E, although additional greater than that in sl. Estimates of T, E, and biological information such as fish lengths or T/E can be obtained by using standard weights may also be collected depending on methodology (Cochran 1977). Let Di denote the specific management or research requirements. total number of days in stratum si; let fi = di/Di, the sampling fraction for stratum One person on the lodge staff should be si, and let wi= Di/D, the weight for paid extra to supervise the data collection pro- stratum i for i = 1, 2. The total harvest T for gram. This person will distribute the logs, the fishery is estimated as: provide instructions to the guides, monitor the progress of the program, and collect the log- book carbons each day. T=^ fi(E^F (kijs + lijs) + x FFS rijs) ijs Lis The guides will be responsible for 2 reminding users to maintain up-to-date and accurate log books, and for conducting periodic, = E Di ti informal accuracy checks. With the cooperation i of the lodge manager, it is hoped that these minimal tasks can be incorporated as part of the where ti is the average harvest for stratum guide's regular duties and will not require si for i = 1, 2. additional incentives. The estimated variance of T is:

12

di = d Dist i / y Dj = y Di (Di - di) s2 t i /di

Assuming optimal allocation, the total sample where size required to achieve a specified value Vo di 1 y ( tij _ fo2 for V(T) is estimated by d = [y wi st.]2 / ( V o + • y s2 t .) 1 D 1

on day in and tij is the total harvest j These estimates can be adapted for E by replac- stratum i. ing st i by s ei , i = 1, 2. In either Similarly, the estimate for total effort E case, prior estimates st i and s ei of the is: stratum standard deviations are required from prior information or pilot study data. . y Diéi Multiple-access day fisheries and extended-stay fisheries where is the mean effort over sampled days in stratum i, (total effort over all users and These are the most difficult types of days) E/di for i = 1, 2, and the estimated fisheries to census because of the difficulty in variance of E is: identifying the population of users at any point in the season. The suggested method is to con- duct two separate censuses to obtain direct, îlch = y Di (Di - di)s 2 ei /di independent estimates of E and C = T/E, and then use the product of these as an estimate of T.

where The effort census: The sampling unit for the effort census is a time by space unit. The tem- di poral component is taken to be one hour of a v 16-hour fishing day. The spatial component will s2 i _ 1 ( ei . _ é02 d i _ 1 t j depend on the characteristics of the fishery in question but it should be constructed so that it can be censused in a single hour. In this case, where eij, is the effort over all users on day a unit census is simply a count of the number of j in stratum i. users in a spatial component during a one-hour temporal component. It is assumed that the The simplest estimates of C, the harvest population of users is relatively stable during per unit effort, is each one-hour unit. In other words, if n users are identified in a particular spatial unit î/î during a one-hour temporal unit, then these users contribute n hours of effort to the The estimated variance of C is total. With this structure, total effort is equivalent to population size. îl (ê) = 1 y D2 1 (1- fi) d. E2 For the effort census, it is desirable to stratify in both space and time. The spatial (S2 + C2 S2 - 2 C s ) strata will depend on the particular fishery, t- e- te- but the areas close to high-use access points are natural candidates for stratification. The where temporal strata should be constructed based on weekday and time of day. Four-hour blocks, di starting at 0600h and ending at 1100h, are good candidates for time of day strata, since fishing stei - 1 Î (t iJ*- fi) (eij éi) pressure is usually highest during the early and late parts of the day.

is the sample covariance between harvest and As an example, suppose that a particular effort in stratum i. fishery has 10 high-use access points and 5 low-use access points. The areas around these Since information on three population 15 access points comprise the spatial units parameters is required, the determination of which are divided into two strata consisting of optimal sample size and allocation will depend the 10 high-use points and the 5 low-use on the relative variance of effort, harvest and points. Suppose further that the season lasts harvest per unit effort in each stratum. In any 10 weeks so that there are 50 weekdays and 20 case, it is reasonable to assume that the daily weekend days. There are then 8 temporal cost of sampling is constant across strata. If strata. The first 4 consist of consecutive harvest is of primary importance, the optimal four-hour blocks of time across the 50 weekdays, allocation for a fixed sample size d is estima- with each stratum containing 4 x 50 - 200 pos- ted using sible hours to be sampled. The second four strata consist of four-hour blocks of time 13

across the 20 weekend days, with each stratum yiJ to be the total harvest per unit effort containing 4 x 20 = 80 hours. In total there for day j in stratum i and following the are 2 x 8 = 16 time-space strata. One stratum expressions for E and V (E) given for the effort consists of the time period 0600h-1000h on week- census. Again, sample size and optimal days at the high-use access points. The same allocations are determined as outlined for stratum contains 4 hours x 50 days x 10 access limited-access day fisheries. points, or 2,000 time-5pace sampling units. A random sample of size 30 from the strata would Total harvest: With the estimates from the require 30 hours of sampling, each hour in the effort and harvest per unit effort surveys, the period 0600h-1000h, across 50 possible weekdays total harvest T can be estimated by and 10 access points. A single sample might require a count of all users around the fourth T=EC high-access point from 0900h to 1000h on the third Tuesday of the season. For stratum i, let For unexploited fisheries, the estimated vari- ance of T is approximated by Ni = Total number of time-space units ni = sample size V(T) = E2V (C) + C2V(E) N = sum of Ni Wi = Ni/N Stratum weight where C and E are independent, and for exploited yij = Total number of users observed in fisheries the j-th unit, j=1, 2...ni fi = ni/Ni V(T) = E2V(C) + C2V(E) + 2CE cov (C,E) The estimate of total effort for the fishery is where C and E are not independent, and where then V(E) and V(C) are obtaihed from the effort and harvest per unit effort surveys.

As an alternative to the surveys proposed above, a roving creel census might be used to where obtain an estimate of harvest per unit effort. However, this assumes that the harvest rate is Ÿi = F yiJ/ni constant throughout the day. Malvestuto and J Davies (1978) indicate that this assumption might be acceptable, but Robson (1960, 1961) the estimated variance of E is suggests the opposite may be true. In view of the uncertainty involved, it seems most reason- V(E) = F Ni(Ni- ni) s2i/ni able to adopt a census method that does not J require such an assumption. where Estimates of sample size and optimal allo- cation for the various surveys require informa- S2, = ni 1 (yi j- Ÿi )2 tion on stratum variances. Because such infor- J mation is lacking, it will be necessary to rely on the results of pilot runs to establish the Sample size and optimal allocation are deter- final census protocol. mined according to the procedure outlined for surveys of limited-access day fisheries. The results from the harvest per unit effort census for multiple-access day fisheries Harvest per unit effort censuses: The pro- and extended-stay fisheries can be used to posed census for éstimating C is the same as obtain estimates of T and E in addition to C. that for limited-access day fisheries except This is done by using the estimation methods for that access points will have to be sampled limited-stay fisheries modified to allow for rather than covered exhaustively. To understand additional strata over access points. However, this census, consider the example used for the a bias may be introduced because of users who do effort survey. There are four strata: high- not arrive and depart the same day. As a very access points on weekdays, high-access points on extreme case, a user who arrives the day before weekend days, low-access points on weekdays, and the season begins and departs the day after the low-access points on weekend days. In this season ends will have a zero probablility of case, the temporal component of a sampling unit being included in any sample. The effort census is a day and the spatial component is an access is intended, in part, to allow for such possibi- point. The four strata listed above contain 10 lities. It will, however, be of interest to x 50 = 500, 10 x 20 = 200, 5 x 50 = 250 and 5 x compare estimates of T and E obtained from just 20 = 100 sampling units, respectively. A unit the harvest per unit effort with those obtained is measured by stationing interviewers, possibly using the procedure detailed above. It is pos- in shifts, at the corresponding access point for sible that the effort census may not be neces- a full day and collecting the information as sary for certain multiple access fisheries. indicated for limited-access day surveys. Of course, sampling units are to be randomly selec- ted within strata. RECOMMENDATIONS FOR FURTHER STUDIES The harvest per unit effort, C, is estimated along with its variance by defining The previous section describes a three- tier statistical reporting system to provide 14 recreational fisheries data for the NWT. Four - identify specific biological and-econo- different data collection systems were recom- mic data requirements for the pilot mended and described. program; - identify and characterize the road These data collection systems were recom- access fisheries in the NWT and select mended on the basis of their ability to provide one limited-access and one multiple- the required information with a minimum of cost access fishery for the pilot program; and effort. Each of the systems described have - develop implementation logistics with some flexibility in their exact application. experienced field staff; Since there is no previous operating experience - develop data forms and accompanying with these systems, their costs and statistical educational materials; effectiveness will have to be determined through - test creel census methods during the pilot programs. We recommend that such pilot 1983 season; programs be developed and conducted in 1983 to - analyze results, test assumptions, and evaluate each of the recommended systems. develop sample size and allocation of effort for future surveys, and The unstructured field checks and reports - evaluate the cost and effectiveness of are the most cost-effective method of collecting each survey, and modify as required for general information on the extensive fisheries. implementation. The existing program should be reviewed and evaluated for its contribution to the reporting The recommended systems are designed pri- system, data requirements should be defined, and marily to provide yield and effort data that can a consistent reporting system developed. be used for both biological and economic analy- ses of the recreational fishery. If additional The design of the questionnaire program data are required to provide estimates of the and its administrative system should be ini- value of specific lodge fisheries, site-specific tiated immediately because of the lengthy admi- surveys will be required to collect the data. nistrative lead times required. A computerized If additional data are required to provide esti- mailing, editing, and analysis program should be mates of the value of specific road-access considered to improve the quality of the product fisheries, the surveys whould be conducted in and minimize labor requirements. Because of the conjunction with creel censuses. Combining relatively small number of recreational fisher- these surveys will have the advantage of sharing men involved, consideration should also be given a previously designed sampling scheme, providing to maintaining, as part of the computerized sys- consistent results, and sharing manpower and tem, a data base to provide additional data on costs. the demography and user characteristics of the recreational fishermen. If additional biological or economic data are required for intensive/extensive fisheries, The diary system should be tested on at the mail questionnaire system may be appro- least two lodges, one at Great Bear Lake and one priate. This information should be collected at Great Slave Lake. The test lodges should be through a special questionnaire, since adding medium to large and should have a maximum mix- questions to existing questionnaires could ture of species available for catch. The test reduce the quality and consistency of the exist- program should consist of the following items: ing data collection program.

- identify specific biological and econo- mic data requirements for the pilot program; ACKNOWLEDGMENTS - develop implementation logistics with experienced field staff; The authors received valuable assistance - draft proposed program and discuss with from a number of professionals during the pre- lodge owners at their annual fall paration of this report. This report would not meeting; have been possible without the data and comments - develop data forms and accompanying provided by staff at the Department of Fisheries educational materials and Oceans, Western Region, Freshwater Insti- - test program on lodges during 1983 sea- tute, particularly Peter Thompson, Daniel Topol- son; niski, Roger Peet, and Robert Moshen0. Keith - analyze results and develop imputation Brickley provided valuable comments on an early or statistical techniques to account draft of the report and Ray Schmidt and Martin for any missing data, and Gerrard assisted in the preparation of the final - evaluate the cost and effectiveness of report. the program, and modify as required for implementation.

The creel census should be tested on at REFERENCES least two road-access fisheries. One of these fisheries should be a limited-access type and ANEY, W.W. 1974. Estimating fish and wildlife the other should be a multiple-access type. It harvest, a survey of methods used. Pro- would be preferable to select road-access ceedings of 54th Annual Conference of Wes- fisheries that had a maximum mixture of species tern Association of State Game and Fish available for catch. The test program should Commissioners: 70-83. consist of the following items: ARGUE, A.W., J. COURSELY, and G.D. HARRIS. 1977. Preliminary revision of Georgia 15

Strait and Juan de Fuca Strait tidal sal- ERMAN, D.C. 1972. A fisherman survey of the mon sport catch statistics, 1972 to 1976, Madison- Fi rehole ri vers in Yellowstone based on Georgia Strait head recovery pro- National Park, 1979. Trans. Am. Fish. gram data. Can. Fish. Mar. Serv. Tech. Soc. 101(2): 352-355. Rep. Ser. PAC/T-77-16: iv + 68 p. FALK, M.R. 1981. 1979 Questionnaire survey on ASHRAF, A. 1975. The methodology of the Cana- sport fishing in the Yellowknife and Hay dian travel survey, 1971. Surv. Metho- River Areas, Northwest Territories. dol. 1(2): 208-227. Department of Fisheries and Oceans. Win- nipeg, Man. (unpublished) BEST, E.A., and H.D. BOLES. 1956. An evalua- tion of creel census methods. Calif. Fish FARR, D.H., and C.M. BLAKE. 1979. A creel cen- Game 42: 109-115. sus of the salmonid fishery in South Sandy Creek, New York. N.Y. Fish Game J. 26(1): CANNELL, C.F., L. OKSENBERG, and J.M. CONVERSE. 1-10. 1977. Striving for response accuracy: experiments in new interviewing techni- FILION, F.L. 1980. Human surveys in wildlife ques. J. Marketing Res. 14: 306-315. management p. 411-453. In S.D. Schemnitz (ed.) Wildlife management techniques CARLANDER, K.D., C.J. DICOSTANZO, and R.J. JES- manual. The Wildlife Society, Washington, SEN. 1958. Sampling problems in creel D.C. census. Prog. Fish-Cult. 20(2): 73-81. GARD, R., and D.W. SEEGRIST. 1972. Abundance CARLINE, R.F. 1972. Biased harvest estimates and harvest of trout in Sagehen Creek, from a postal survey of a sport fishery. California. Trans. Am. Fish. Soc. 101(3): Trans. Am. Fish. Soc. 100(2): 262-266. 463-477.

CAUVIN, D.M. 1981. Strategic overview, Pacific GORDON, D., D.W. CHAPMAN, and T.C. BJORN. and Freshwater Fisheries: An economic 1973. Economic evaluation of sport perspective. Department of Fisheries and fisheries - what do they mean? Trans. Oceans, Winnipeg, Man. (unpublished). Am. Fish. Soc. 102(2): 293-311.

CAUVIN, D.M. 1982. The enhancement of fisher- HAYNE, D.W. 1966. Notes on creel survey for ies in Saskatchewan: A study proposal. Tennessee Cooperative Fishery Unit. Ten- Department of Fisheries and Oceans, Wes- nessee Game and Fish Commission. 25 p. tern Region, Winnipeg, Man. (unpublished) mimeo.

CHARBONNEAU, J.J. 1980. National surveys rela- HIETT, R.L., and J.W. WORRALL. 1977. Marine ted to data needs for recreational fisher- recreational fishermen's ability to esti- ies, p. 37-39. In J.H. Grover (ed.) Allo- mate catch and to recall catch and effort cation of fishery resources. Proceedings overtime. HSR-RR-77-/13cd, Human Science of the technical consultation on alloca- Research, Inc., McLean, VA. 22 p. tion of fishery resources held in Vichy, France, 20-23 April 1980. United Nations HILGARD, E.R., and R.C. ATKINSON. 1967. Food and Agriculture Organization. Remembering and forgetting. p. 314-335. In E.R. Hilgard et al. (ed.) Introduction COCHRAN, W.R. 1977. Sampling techniques. John to psychology, 4th ed. Harcourt, Brace Wiley & Sons, Inc., New York. and World, Inc., New York.

CORKUM, L.D., and P.J. McCART. 1981. A review INTERDISCIPLINARY SYSTEMS LTD. 1978. Effects of the fisheries of the Mackenzie Delta of exploration and development in the and nearshore Beaufort Sea. Can. Baker Lake area. Department of Indian Manuscr. Rep. Fish. Aquat. Sci. 1631: v + Affairs and Northern Development, Ottawa, 55 p. Ont.

DEUEL, D.G. 1980. Survey methods used in the INTERDISCIPLINARY SYSTEMS LTD. 1980. Assess- United States marine recreational fishery ment of the socioeconomic impacts of the statistics programme, p. 82-86. In J.H. arctic pilot project on selected high Arc- Grover (ed.) Allocation of fishery re- tic communities. Arctic Pilot Project sources. Proceedings of the technical Calgary, Alta. consultation on allocation of fishery re- sources held in Vichy, France, 20-23 JAMSEN, G.C. 1971. Michigan's mail creel cen- April 1980. United Nations Food and sus methodology. Mich. Dep. Nat. Resour. Agriculture Organization. Res. Dey. Rep. 252: 5 p.

DWYER, J.F., and M.D. BOWES. 1978. Concepts of JENSEN, P.T. 1964. Landing estimates of Cali- value of marine recreational fishing. fornia's marine recreational salmon fish- Am. J. Agric. Econ. 60(5): 1008-1012. ery. Calif. Fish Game 50(1): 48-52.

ERDOS, P.L., and A.J. MORGAN. 1970. Profes- JONES, F.V., W.D. PEARSON, and L.C. FITZ- sional mail surveys, McGraw Hill Ltd., PATRICK. 1977. Yield estimates derived New York. from active and passive creel surveys of a small pond fishery. Tex. J. Sci. 29(1/2): 41-47. 16

KAZES, K.K. 1980. Recall--Information to be ROBSON, D.S. 1960. An unbiased sampling and used when considering change in the recall estimation procedure for creel census of period for the 1982 National Travel Sur- fishermen. Biometrics 16(2): 261-277. vey. U.S. Dep. Commerce, Bureau of Cen- sus. 2 June 1980. 11 p. Unpublished ROBSON, D.S. 1961. On the statistical theory memorandum. of a roving creel census of fishermen. Biometrics 17(3): 415-437. KELEHER, J.J., and C.G. HAIGHT. 1965. The fall domestic fishery at Snowdrift, Northwest SECKLER, D. 1966. On the uses and abuses of Territories. J. Fish. Res. Board Can. economic science in evaluation of public 22(6): 1571-1573. outdoor recreation. Land Econ. 42(4): 485-494. KREUTZWISER, R. 1981. The economic signifi- cance of the Long Point Marsh, Lake Erie, SIMPSON, J.C., and T.C. B'JORNN. 1965. Methods as a recreational resource. J. Great used to estimate salmon and steelhead har- Lakes Res. 7(2): 105-110. vest in Idaho. 45th Annual Conference of Western Association of State Game and Fish McCONNELL, K.E., and V.J. NORTON. 1975. An Commissioners. economic evaluation of marine recreational fishing: A review. Department of Re- SINCLAIR, W.F., and R.W. MORLEY. 1975. Statis- source Economics, University of Rhode tical bias problem in on-site surveys: Island, Kingston. 38 p. the severity of the problem and its poten- tial for solution. J. Fish. Res. Board MacDONALD, D., and E.G. DILLMAN. 1968. Tech- Can. 32(12): 2520-2524. niques for estimating nonstatistical bias in big game harvest surveys. J. Wildl. SOUTHWARD, G.M., and J. DOUGLAS. 1972. Evalua- Manage. 32: 119-129. tion of a winter steelhead fishery on a western Washington river. Trans. Am. MacGREGOR, M. 1982. The tidal sportfishing Fish. Soc. 101(1): 29-34. diary program: report on the pilot years 1979-1980. Department of Fisheries and THOMPSON, W.A., Jr. 1976. Fisherman's luck. Oceans, Vancouver, B.C. 49 p. + App. Biometrics 32: 265-271.

MALVESTUTO, S.P., W.D. DAVIES, and W.L. SHEL- TOPOLNISKI, D. 1975. Northwest Territories re- TON. 1978. An evaluation of the roving creational fishery, 1974-75 statistics. creel survey with nonuniform probability Canada Department of the Environment, sampling. Trans. Am. Fish. Soc. 107(2): Fisheries and Marine Service, Winnipeg, 255-262. Man. Unpubl. Rep. 17 p.

MALVESTUTO, S.P., W.D. DAVIES, and W.L. SHEL- TOPOLNISKI, D. In prep. Survey of sportfishing TON. 1979. Predicting the precision of lodges in the Northwest Territories. creel survey estimates of fishing effort Can. Tech. Rep. Fish. Aquat. Sci. by use of climatic variables. Trans. Am. Fish. Soc. 108: 43-45. WITHLER, R.E. 1976. A review of the literature concerning domestic fisheries associated MILLS, M.J. 1981. Statewide harvest study - with proposed gas pipeline routes in the 1980 data. Alaska Fish Game Fed. Aid Mackenzie River Valley. Report by Aquatic Fish Restor. Annu. Rep. Progr. 1980-1981, Environments Ltd. for Canadian Arctic Gas Proj. SW-I-A. Study Ltd. 38 p.

MOTTLEY, C.M., and D.R. EMBODY. 1940. An effi- cient sampling method for taking a creel census. Prog. Fish-Cult. 49: 21-22.

MOYLE, J.B., and D.R. FRANKLIN. 1957. Quanti- tative creel census on 12 Minnesota lakes. Trans. Am. Fish. Soc. 85: 28-38.

MURPHY, L.W. 1954. Two-year creel census of the Little Salmon River, Idaho. Trans. Am. Fish. Soc. 84: 200-206.

PANEK, F.M. 1981. The warmwater sport fishery of eastern Lake Ontario. N.Y. Fish Game J. 28(2): 178-190.

PEARSE, P.H. 1968. A new approach to the eva- luation of non-priced recreational re- sources. Land Econ. 44(1): 87-99:

REGIER, D.S. 1966. Fishermen and creel cen- suses. New York Cooperative Fishery Re- search Unit. 10 p. 17

Table 1. Examples of lakes and rivers in each fishery category in the NWT (data supplied by R.W. Moshenko, personal communication).

FISHERY FISHERY TYPE

Intensive Fisheries Great Slave Lake road access, lodges Great Bear Lake remote access, lodges Hay River road access Mosquito Creek road access Marian Lake Boat access, winter access Kasba Lake road access Kakisa Lake road access Dolomite Lake road access Campbell Lake road access Sylvia Grinnell River road access Tree River remote access Prelude Lake road access Prosperous Lake road access

Intensive/Extensive Fisheries Gordon Lake winter road access, lodge Sitidgi Lake remote access Hornaday River remote access, road access Contwoyto Lake remote access Oubawnt Lake remote access, lodge Nueltin Lake remote access, lodge Nonacho Lake remote access, lodge Kasba Lake remote access, lodge Snowbird Lake remote access, lodge Yathkyed Lake remote access, lodge Henik Lake remote access, lodge Robertson River remote access, lodge Back River remote access Stanwell-Fletcher Lakes remote access Kingnait Fiord remote access Pangnirtung Fiord remote access

Extensive Fisheries All other lakes predominantly remote access

Table 2. Summary of reporting methpds used for evaluating fish harvest in the Western United States (from Aney 1974).

Unsupervised Sample Surveys Report Check Species Field Checks Field Mail Home Forms Diaries Stations Other2

Cold-Water 12 12 8 2 2 3 1 O Warm-Water 11 9 7 2 2 3 1 O Anadromous 5 3 5 1 3 0 1 1 Marine 1 4 2 0 0 0 O 1

Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, Wyoming. 2 Fish• passage counts and party boat logs. 18

Extensive Fishery

General/Trend Data —] Unstructul cidFiel l Checks i

Extensive/Intensive Fishery

r- General /Trend •Data Site-Specific Data -1 r- I. r--MalMail Questionnaires As Required iRemOte _I Road Access Access

Modified Diary Catch/Effort Census as Required as Required

Intensive Fishery 1 E Site-Specific Data I

Remote -1 Road I Access Access

Modified rii;177y Catch/Effect Census as Required as Required [..._

Fig. 1. Schematic of a statistical reporting system for the sport fishery of the NWT. DATE DUE DATE DE RETOUR

LOWE-MARTIN No. 1137