Creel census method
Array ( [0] => Neuhold, John M. [1] => Lu, Kuo H. )
Publication Number 8
Utah. State Department of Fish and Game
Array ( [0] => Not Specified )
,_ ........ - ;i '1c,tJtJ . 5 : C r e/tJS7 "- - - - . CREE L / CENSUS METHOD by John M. Neuhold and Kuo H. Lu • _-:-__ PUBLICATION NO. 8 OF THE FEDERAL AID DIVISION OF THE -:-~-:-:--::----:-:-:~T::~--::--:-::JiIiIioi. UTAH STAT EP AR TMENT STATE OF UTAH George Dewey Clyde, Governor FISH AND GAME COMMISSION K. E. Bullock Robert B. Mitchell Golden G. Sanderson Dr. Paul G. Stringham W. Rulon White UTAH STATE DEPARTMENT OF FI SH AND GAME J. Perry Egan, Director /Ie!. ' i 1'1 ).r';9 .z - - ERRATA Figure 2, page 12. The solid lines represent the pres- sure and the dotted lines the rate of success. (Omitted from the discussion) Page 28, line 3 should read: 2 " • •• and Sr equals the root mean square errors of the mean number of boat fisherman hours and the mean rate of fishermen per boat ••• " Page 29, line 26 should r ead: 286,870 t 82,732 fishermen II • •• C hours. II CREEL CENSUS METHOD by JOHN M. NEUHOLD and KUO H. LU Publication No. 8 of the UTAH STATE DEPARTMENT OF FISH AND GAME Submitted for Publication March 1, 1957 Printed by the Utah State Department of Fish and Game Salt Lake City, Utah A contribution of projects F~l~R, and FA~RJ Federal Aid Division, Utah State Department of Fish and Game TABLE OF CONTENTS Acknowledgement List of Figures ........ . List of T ables ...... . Introduction __ _ . Creel Census Design .. Fisherman Counts _ Calcul ation of pressure The Interview ____ .. Calcul ation of n:tte of success Calcul ation of harvest ____ ............... . Evaluation of the Method Literature C ited 4 5 6 6 7 9 17 25 30 .......... 31 . ................ 34 ...... 35 .... 36 ACKNOWLEDGMENTS C redi t is given to Dr. W illiam F. Sigler, H ead of the Department of Wildlife Management, U tah State U niversity, Logan, U tah, who origin .. ally provided the concept of counting on intervals of the average daily fishing curve and to W illiam J. McConnell , who fostered this concept and w as instrumental in the developments o f the m ethod. Thanks are given to Dr. Rex Hurst 0 !the U. S. U. D epartment of Applied Statistics for his valuabl e advice and assistance in editing the manuscript. 5 LIST OF FIGURES 1. Fishing pressure and rate of success on Deer Creek Reservoir during August, 1955. 2. Mean daily rate of success and pressure on four 1955 creel census waters. 3. A correlation between air temperatures and fishing pressure. 4. Seasonal fishing pressure curve on Fish Lake during the 1955 season. S. Correlation between mean days fished per season and rate of success. 6. A shore fisherman situation on which an instantaneous count is made. 7. A shore fisherman situation representing a progressive type count. 8. A biased progressive count situation. 9. A hypothetical lO-day fishing season. LIST OF TABLES 1. Difference in the mean pressures of shore fishermen on Scofield Reservoir and boat fishermen on Fish Lake during June and July, 1956, between inclement and non~inclement weather categories. 2. Differences in mean pressures on Scofield, Deer C reek, Strawberry Reservoirs and Fish Lake during 1955 between weekday and week- end days and holidays. 3. Relation of number of counts of fishermen to accuracy and precision of the estimate in which the total number of fisherman hours per season is 944. 4. The effect of stratifying a sample on the basis of days in the lO-day hypothetical situation of Figure 9. 5. Fisherman counts from Scofield Reservoir during the 1955 season. 6. Scofield Reservoir 1955 creel census rate of success. 7. Summary of mean rate of success estimates for species by boat and shore fishermen on Scofield Reservoir during 1955 creel census. 8. Summary of harvest from Scofield Reservoir during the 1955 season. 6 INTRODUCTION Creel census in a cold water trout fishery is, perhaps, one of the most useful tools availabl e in the management of tha t fishery. This statement is particularly true in situations where the fishery is being maintained, of necessity, by artificial means, and tcue to a lesser extent when the fishery is maintained through natural processes. Such basic information concerning the mortality from fishing in the fish population as a total fishing pressure, rate of fishing success, and tota l harvest can be obtained by the application of creel census. From this information the investigator can derive an assortment of data concerning the fish population, e.g., (I) the extent of harvest of fish, (2) the size of the fi sh population, (3) the effect of a given fishing pressure on a fish population, and (4) the numbers of fish required to replace a mortality caused by fi shing. Creel censuses can be accomplished by two methods: (1) observing the total pressure and harvest for an entire season, or (2 ) estimating the pressure and harvest by sampling of the total population. Observing the tota l pressure and harvest is costly as well as time and effort con ... suming except in some isolated instances. One need not argu e, however, as to the valid ity of the values obtained by this method. Estimates, on the other h and, are done with considerable saving in time, money, and effort but with the subsequent sacrifice of certainty that the total observation offers. Estimates number many in methods and analysis. They range from a vol untary fisherman ... questionnaire return to counting and interviewing fishermen, all with varying degrees of success and reliability. Since an estimate is obtained from sampling a portion of the whole population, great variance in the accuracy and precision can be encountered depending upon the thoroughness of the design and the simplicity of the analysis. A ccuracy can only be atta ined through careful design of the creel census and precision through sound application of statistical principles. The prima ry objective of a creel census in management is the deter ... mination of the harvest or fishi ng mor tality during a specified period. Secondarily, the objectives incl ude sllch varied information as the eco ... nomics of the fishery (a function of the fishing pressure) and further estimates of the fish population (related, in some extent, to the fishing mortality). As stated previously, two methods exist whereby values concerning the fi shing pressure and harvest ca n be obtained: ( 1) observation of all 7 fisherm en and their catches during a specified period of time, and (2) observation of a portion of the fishing pressure and harvest during a speci .. fied period with a final expansion to an estimate of the whole. Observation of the whole provides the mOSt desirable kind of data since no variance in the totals is encountered. Data of this nature repre .. sent the t rue values of the population being enumerated . This type of a creel census, however, is usually im practical in terms of time, money, and effort necessary for accomplishment. Observation of a portion ot the whole (the sample) and a final ex .. pansion to the estimate of the whole can be made only with palpable reservations. Since the sampl e is based on only a part of the total situation, allowances must be made for that portion of the universe not sampled. Thus the error that mayor may not be encountered by expanding the sample to the whole is taken into consideration and qualifies the final statement about the universe. The estimate, jf used properly, can give good and even excellent approximations of the actual population values with an expenditure of money, time, and effort somewhat less than what would be necessary were the universe obtained. The goal of most fishery ma nagers is to conduct as many creel censuses as limited funds will permitj consequently, the estimate becomes, to them, even more important. H owever, estimates that require excessive expendi .. tures gain li ttle over obtaining the actual population va lues. The necessity arises, then, to provide an est imate that will afford a critically accurate value at the least possibl e cost. The object of this paper is to present a method th at can be used under a va riety of conditions with reasonably accurate results. 8 CREEL CENSUS DESIGN C reel census methods have been designed under a number of tech~ niques. Designs range from those that provide substantially accurate esti ... mates of the mean values to those that are palpably inaccurate. The primary basis for providing accuracy to the estimates of the mean values is the consideration of all variables, some of which are cryptically subtle; e.g., the estimate of the average number of fishermen per day for an entire season must consider the length of the possible fishing day (where daylight hours are used as the criterion for the length of the fishing day) progressively and not as an entity, This becomes obvious when the peak of the fishing pressure occurs in June, that time of the year with the longest possibl e day, and the trough of the fishing pressure occurs in December, that time of the year with the shortest possible fishing day. If an average value for the length of the day for the season were used, the low pressure would receive weight equ al to the high pressure with a resulting under~estimate of the total pressure. Precision of the estimate in most creel censuses is not even considered, and yet, without precision the values derived from an estimate are but guesses. Fishery biologists commonly tend to temper what they call the precision of their estimate with assumptions, many of which are un .. qualified; e.g., in a fisherman count along a 20~mile st retch of stream where an attempt is made to count and contact all the fishermen for the sample day, the assumption is made that all fishermen are counted and contacted during the sampl e day or that the number of fishermen missed is probably insignificant. In most si tuations the number of fishermen missed is probably insignificant and probably has little bearing on the accuracy of the estimate of the mean number of fishermen per day. However, if fishermen are missed, they are probably missed in proportion to the in~ tensity of the fi shing pressure and, consequently, would have a variance similar to the variance of the fishermen count. The precision of the mean number of fishermen per day, then, could be and probably is adversely affected. Fishermen numbers, as many other biological populations not under laboratory control, are influenced by a number of non~predictable phe~ nomen a as well as some that are predictable. Factors such as weather and rate of fishing success definitely have an effect on numbers of fisher~ men fishing a given body of water during a season; yet, weather' and fishing success are not predictable. 9 Data coll ected from Scofield Reservoi r and Fish Lake during 1956 show a significant difference in mean numbers of fishermen between counts made during the inclement weather and counts made during non~ inclement weather (T able 1). The counts were made on the basis of water condition (calm, choppy or whi te caps) and on the weather index (clear, less than 50% overcast, greater than 50% overcast, and rai ning or snowing). Andriano fou nd that weather categories, similar to those mentioned, produced one of the greatest d ifferences in numbers of fi she r~ men on Henry's Lake in Idaho of all the variables examined ,l T ABLE 1. Differences in the mean pressures o f sh ore fi shermen on Scofield R eservo ir and boat fishermen on Fish Lake d uring June and July, 1956. be twee n inclem ent and non -inclem en t weather categor ies. Place Fish Lake ........ _ Scofield Reservoir M ean Inclem en t W ea ther 31.451 16.402 Pressure N o n.inclem en t W eather 51. 141 22.64' 1 Both values represe n t mean number of hoats per count. Com puted t·va lue l 2.09 3.45 2 Both values represent mean number of fishermen per COUnt. t-value at 95 0/0 Confiden ce Level 2.04 2.04 3H ypothesis tested is one of equal means (after D ixon and Massey, 195 1, pp. 102 and 103) . Bo th insta nces show a t.value grea te r than would be expected at the 95% confi. dence leve l, and the hypothesis of equal means is rejected. Not onl y does the nature of the weather at the particular body of water u nder surveill ance have a di rect effect on the numbers of fishermen, but the weather at nearby centers of population shows indicat ions of affect- ing fishermen nu mbers. D ata from Bear Lake, Uta h ~Idaho, imply that weather in the cities of Logan and Ogden might have some infl uence on num bers of fishermen even though the weather at the lake was opposite the weather at the mu nicipalities. Additional data from the 1955 Deer C reek Reservoir Creel Census indicate a significant d ifference in mean numbers of fi shermen between days following periods of good success and days following periods of poor success (Fig. 1). The effect appears to be somewhat delayed in that the the rate of success appears to influence fishermen nu mbers after occur~' rence rather than during occurrence. Similarly, the time of day seems to have delayed effect on the daily fishing pressure. The fishermen numbers, except for the opening hours of the day, appear to be influenced by the rate of success in the period preceding, as on Deer C reek Reservoir during the 1955 season (Fig. 2). IPersonal cor res pondence with Don And riano of the Idaho Department of Fish and G a me dated August 22, 1956. 10 • 6 .5 .. U> U> w <.> <.> ::> '" .3 u. 0 w .... .. 0:: .2 . / o Since fishermen fish to catch fish this quasi~correlation probably has validity. Differences in mean pressures between week days and week~end days and holidays also appear to be significant from creel censuses conducted in Utah (Table 2). Best and Boles (1956) implied as much when they sampled all week-end days and holidays of the 1953 Rush C reek and Castle Lake data . I f I / • /0 DAYS / f I I I I " f f I f / I " I \ \ \ 20 OF PRESSURE \ \. \ \ \ \ \ '\ 25 / 50 / 25 / 00 7S " 50 2S 30 Figure 1. Fishing pressure and rate o f success on Deer Cre ek Reservoir during August, 1955 . The so lid line represents the fi shing pressure and th e dotted line the rate of success in fi sh per hour. Note that the fishing pressure tends to lag after the rate of success. The peak in the rate of success occurred after an application of copper sulfate for plankton control. 11 2 W ::; II: W or z 0 0 ~ 0 0 -1 0 -0 3: '" 0 " 0 I ,,- -<~ 0 0 ~ o. 00 0 0 0 ~ 0 0 -10 ",0 '" 0 " 0 " -<0; 0 0 ~ 0 0 0 0 MEAN NUMBER OF FISHERMEN MEAN NUMBER OF FISHERMEN ~ m N 0 m ;;; 0 0 0 0 , 0 0 I '" I -; I '" I I ,. I " -;0 ) W - 0 / '" ;: / '" '" / , / '" 0 / -< " / '" 0 '" ,. '" -<~ '" 0 :>J 0 , , , " < , " " 0 " " " '" , " , " , , ~ o . ~ ;. u. ~ 00 N 0 N 0 0 0 0 0 0 0 0 RATE OF SUCCESS RATE OF SUCCESS MEAN NUMBER OF FISHERMEN MEAN NUMBER OF FISHERMEN ~ ~ 0 ~ m 0 0 0 0 0 , 0 0 , , / I , I I , , , / , <- -10 , , " <,0 , , "- '" '" \ " I 0 \ , r " , , " 0 \ " '" l> , , '" -<~ , > 0 '-/ 0 , / " , , / / , / , , ~ o . N ~ ;. ~ ~ ~ 00 0 N ~ ~ " m 0 0 0 0 0 0 0 0 0 0 0 0 0 RAT E OF SUCCESS RATE OF SUCCESS Figure 2. Mean daily rate of success and pressure on four 1955 crcel census waters. Note that the rate of success in fish pcr hour appears to produce a latent effect on the pressure. A possibility also ex ists that the ratc of success is affecced by increasing numbers of in- experienced fishermen which tends to reduce the ratc of success in the period following. 12 N ~ '" () 0 -n '" r 0 '" '" '" '" :>J < Q '" , 0 ~ 0 0 '" of :>J () '" '" '" '" '" '" '" "' :D < Q :D ~ 0 TABLE 2. Differ ences in m ean p ressures o n Scofield, Deer C r eek , Str awbe rry R eservoirs and Fish Lake during 195 5 be tween weckdays and week~ end days and holidays. M ean Pressure' t-value a t Week-end 95% Days and C omputed C onfide n cc Place W eekdays Holidays t-value: Level Scofield Reservoir 12.89 54.88 8.33 1.97 Deer C reek Rese rvoir .. 14.35 79.42 4.88 1.97 Strawberry Reservoir .... 38.64 114.80 9.95 1.97 Fish Lake ........... 36.72 108.00 6.25 1.97 ' Mean pressure is in terms of boats per count. ~ H ypothesis tested is: two populations have the same mean when the population variance is unknown (after D ixon Massey, 1951, pp. 102 and 103). Note that the computed t-values in each situation are significantly larger than the t-val ue at the 95% confide nce level, requiring the rejection of the hypothesis of equ -< (J) Z (J) fTI l> (J) 0 2 MEAN NUMBER OF FISHERMEN PER COUNT o N ~ ~ m o 0 000 o 0 000 o~~==r=====i=1 '" 0 .. 0 en 0 CD 0 o o ( / / / / Figure 4. Seasonal fishing pressure curve on Fish Lake during the 1955 season (solid line). T h e dotted line represents the means in pressure for 20-day periods. Note that th e pressure tends to decrease s li gh tly as the season progresses. 15 (f) (f) UJ U u ::> (f) ~ o w ~ - .. '-' o ...J w a: o :I: (J) '" z g B oa t C u tt h ro at 49 4 0. 60 1. 98 6. 8 1 10 .0 6 .0 04 0 . 0 20 4 V .> Sh or e C u u h ro at 21 78 0. 12 0. 19 4. 35 12 .5 8 .0 00 1 .0 05 8 B oa t Br ow n 49 4 C .O l 0. 05 6. 8 1 10 .0 6 .0 00 1 .0 20 4 Sh or e B ro w n 2 17 8 0. 00 3 0. 00 3 4. 35 12 .5 8 .0 00 00 1 .0 05 8 B oa t A ll S pe ci es 49 4 4. 5 7 26 .7 0 6. 8 1 10 .0 6 .0 54 0 .0 20 4 Sh or e A ll S pe ci es 21 78 1.1 73 2. 72 4. 35 12 .5 8 .0 01 3 .0 05 8 CA LCULATION OF THE HARVEST An estim ation of the ha rvest is, of course, the primary objective of th is creel census, and is calcu lated by mult iplying the mt'an~tota l fishing pressure in fi sherman hours by the mean rate of success : H = (P) ( R), (9) where H = the harvest of fish; P = the mea n~tota l fishing pressure in fisherma n hours; and R = the mean r:3te of success in fi sh per hour. By substitut ing the value obtained for the mea n pressure by boats and for the mean rate of success from boats for all species, the mean~tota l h ar~ vest for all species can be obtained: H = (156,881) (0.67) , = 105,110 mean total of all spec ies harvested. The p recision is calculated in the same man ner as for the propoga~ tion of error formula, (5) : s,~ = (1 05,1 10) ' [ 39 1,074,1 64 ~ (156,88 1), (1l ,048,112,100) (.01849), = 204,279,590 , and the standard of error is calcu la ted as : s,. --1204,279,590, + .00 11 68) , (.67)' 14,293 fish of a ll species from boats. The confidence limits, then , are c.L. ." 105,110 ± ( 1.96) (14,293) , 105,110 ± 28,014 fish of all species from boats. The correlation between pressu re and rate of success was insignificant. The fid ucial statement remains the same as before : tha t the actu al number of fish caugh t will be encompassed by the expressed limits at the 95% confidence level. The total harvest by species for both the boat and the shore p ressure is obtained in th is same ma nner and is summarized in T able 8. 34 TABLE 8. Summary of harves t from Scofield Reservoir during the 1955 season. Combined tat Limits at T:r Mean 'h, 95 % 'h, 95 % Root Square Standard level of Comb ined level of Fifhinl Species Error Error Confidence Mean Confidence Boat Rainbow 163,666,428 12,793 1.96 90,991 ±25,074 Shore Rainbow 4,565,529 2,138 1.96 31,197 ± 4,190 Total Rainbow 168,231,957 12,971 1.96 122,1 88 ±25,423 Boat Cutthroat 5,218,407 2,284 1.96 13,805 ± 4,477 Shore Cu tthroat 45,806 214 1.96 3,510 ± 419 Total Cutthroat 5,264,213 2,294 1.96 17,315 ± 4,496 Boa t Brown 55,218 235 1.96 455 ± 461 Shore Brown 27 5 1.96 90 ± 10 Total Brown 55,245 235 1.96 545 ± 461 Boat All Species 204,279,590 14,293 1.96 105,11 0 ±28,014 Shore All Species 5,711,854 2,591 1.96 35,097 ± 5,078 Total All Species 209,991,444 14,491 1.96 140,207 ±28,402 EVA LUA TION OF THE METHOD As with any creel census method yet devised, this one, too, has its shortcomings. The limits presented by the examples of this method are somewhat wide, but sufficiently precise to apply to most management problems requiring information on pressure and harvest. The sampling of the examples in this paper was done on the basis of 50 percent of the days with two counts per sample day. Precision can be increased if either m ore days are selected as count days or if more counts are made for each existing count day. Since the estimation of the pressure produces the greatest variance, the number of counts can easily be increased with a smaller variance as the result. Pooling units of homogeneous variances can also reduce the variance. Such units as weather categories or weekdays and weekend days can, occasionally, be used in this manner with a somewhat better final esti~ mate of the variance. Scrutiny of the method under which data on fisherman numbers are taken reveals the possibility of the variance being influenced by the length of time each fisherman fishes in relation to the time he is counted. Analysis of the empirical distribution of the time a fisherman starts and ends fish~ ing with respect to time of day is at present being undertaken on data col .. lected over a number of years from various reservoirs in Utah and will be published at a later date. With only means being represented in the statistical treatment of the data, application of the Normal Theory is considered safe and yielcls ad- vantages to the treatment not otherwise available. Design of the creel census is also an important part of the method. Care must be exercised to avoid bias in all instances. A grea t deal of atten .. 35 tion must also be given to the elimination of assumptions if precIsion is desired. The method presented here has its shortcomings, but the premises upon which it is based have definite value to many phases of management in which knowledge of pressure and harvest are necessary. LITERATURE CITED Best, E. A. and H. D. Boles 1956. An evaluation of creel census methods. California Fish and Game. Vol. 42, No.2, pp. 109-143. Deming, W. Edwards 1943. Statistical adjustment of data. New York. John W iley and Sons. 261 pp. Dixon, W ilfred J. and Frank J. Massey, Jr. 1951. Introduction to statistical analysis. New York, McGraw-Hili. 370 pp. Snedecor, George W. 1946. Statistical methods. Ames, Iowa. The Iowa State College Press. 485 pp. 36
Array ( [0] => Array ( [state] => Utah [province] => ) )
Tue, 05 May 2020 19:04:31 +0000
Array ( [document] => dab58761f3fb37f4490a9dc3887bad19731f6208 [content] => 7aa6cfffc0ff0583d75e6490ee4bdfe85a3b355e [dedupe] => 08eebca0e59fe7bfc37943ec2c91d2bee6272ec7 )
Content: 7aa6cfffc0ff0583d75e6490ee4bdfe85a3b355e | Abstract: 2191fc0c9e5968161f4a6e37d3212af80adca3a0