You are on page 1of 10

Fisheries Management and Ecology

Fisheries Management and Ecology, 2010, 17, 1018

Night sampling improves indices used for management of yellow perch in Lake Erie
P. M. KOCOVSKY & M. A. STAPANIAN
USGS Great Lakes Science Center, Lake Erie Biological Station, Sandusky, OH, USA

C. T. KNIGHT
Ohio Department of Natural Resources, Division of Wildlife, Fairport Harbor, OH, USA

Abstract Catch rate (catch per hour) was examined for age-0 and age-1 yellow perch, Perca avescens (Mitchill), captured in bottom trawls from 1991 to 2005 in western Lake Erie: (1) to examine variation of catch rate among years, seasons, diel periods and their interactions; and (2) to determine whether sampling during particular diel periods improved the management value of CPH data used in models to project abundance of age-2 yellow perch. Catch rate varied with year, season and the diel period during which sampling was conducted as well as by the interaction between year and season. Indices of abundance of age-0 and age-1 yellow perch estimated from night samples typically produced better tting models and lower estimates of age-2 abundance than those using morning or afternoon samples, whereas indices using afternoon samples typically produced less precise and higher estimates of abundance. The diel period during which sampling is conducted will not aect observed population trends but may aect estimates of abundance of age-0 and age-1 yellow perch, which in turn aect recommended allowable harvest. A eld experiment throughout western Lake Erie is recommended to examine potential benets of night sampling to management of yellow perch.
KEYWORDS:

dreissenids, harvest, models, Perca avescens, quotas, trawl.

Introduction Sampling regimens that consider the biology and ecology of the target species, and that are adaptable as sh communities and ecology change are critical to effective sheries management (Smith & Gavaris 1993; McAllister & Pikitch 1997; Folmer & Pennington 2000; Schnute & Haigh 2003). An inappropriate sampling regimen can produce variable and biased estimates of sh abundance (e.g. Korsbrekke & Nakken 1999). Ideally, a sheries monitoring programme should include those procedures and collect samples that are both sucient to answer the questions of interest and ecient in terms of expenditures of resources and time (e.g. Environmental Monitoring and Assessment Program; Larsen et al. 1994; Baker, Peck and Sutton 1997; Stapanian et al. 2007). State, provincial and federal agencies have been using data from bottom trawl surveys to assess the sh community of western Lake Erie since at least 1961

(Stapanian et al. 2007). Other assessments include gillnetting, hydroacoustics and midwater trawling. An important use of data from the bottom trawl surveys is to generate shery-independent indices of recruitment for yellow perch, Perca avescens (Mitchill), which is one of the more economically valuable sheries in the Great Lakes region (Leach 1999). Indices of abundance of age-0 and age-1 yellow perch from agency bottom trawl surveys provide the only data for predicting future abundances of age-2 yellow perch, which is an important biological criterion for setting harvest quotas (e.g. Yellow Perch Task Group 2007). Nearly all bottom-trawl sampling by Lake Erie sheries agencies is conducted during daylight hours. The US Geological Survey (USGS) Lake Erie Biological Station is the only agency that has included nighttime sampling in Lake Erie bottom trawl assessment programmes. Research in the north Atlantic has demonstrated that the diel period during which

Correspondence: Patrick M. Kocovsky, USGS Great Lakes Science Center, Lake Erie Biological Station, 6100 Columbus Avenue, Sandusky, OH 44857, USA (e-mail: pkocovsky@usgs.gov)

10

doi: 10.1111/j.1365-2400.2009.00721.x

Published 2010. The article is a US Government work and is in the public domain in the USA.

NIGHT SAMPLING FOR YELLOW PERCH

11

sampling takes place may affect catchability (Glass & Wardle 1989; Walsh 1991) or indices of abundance or recruitment (Michalsen et al. 1996; Casey & Myers 1998) of some species of sh. In Lake Erie, Stapanian et al. (2007) reported that bottom trawl catches of age0 sh of several species in western Lake Erie, including yellow perch, were greater at night than in the morning and afternoon. Furthermore, Stapanian et al. (2009) demonstrated that a shift to a higher ratio of nighttime to daytime catch of age-0 yellow perch occurred between 1990 and 1991 following increases in water clarity associated with the establishment of dreissenid mussels. These appear to be the rst assessments of the potential eect of sampling at night in western Lake Erie. The results of Stapanian et al. (2009) prompted an examination of whether predictions of abundance of age-2 yellow perch, which are based on indices of recruitment from bottom trawl samples, might be aected by the diel period samples are collected. The objective of this analysis was to determine whether samples collected during particular times of day produced better tting regression models used to estimate abundance of age-2 yellow perch, which are the primary basis for recommended allowable harvest (RAH) used as a guide to establish the total allowable catch (TAC) of yellow perch in western Lake Erie. Data from an ongoing, long-term eld study of shes in western Lake Erie were used to evaluate the effect of diel period of sampling on: (1) catch rate of age-0 and age-1 yellow perch; and (2) the precision and perceived accuracy of the relationship between indices of age-0 and age-1 yellow perch abundance and estimated age-2 abundance. The analysis focused on the period since

the establishment of dreissenids given the results of Stapanian et al. (2009) that a shift in catchability of yellow perch followed the dreissenid invasion. Methods
Field data collection

Data were obtained from USGS assessment surveys conducted in the western basin of Lake Erie, USA. Yellow perch were collected with a 7.9-m (headrope length) semi-balloon bottom trawl with 2.5-cm square mesh body and 5-mm square mesh cod end at three established stations (Fig. 1) during summer and autumn, 19912005. All trawling was conducted with the same vessel and identical gear throughout the entire time period. On two or three consecutive days in summer and autumn, duplicate trawl samples along the same transect were collected at the 3.1-, 4.5- and 6.1-m depth contours during three time periods on each day: morning (beginning at least 30 min after sunrise and ending by 12:00 hours), afternoon (13:00 18:00 hours) and night (beginning at least 30 min after sunset and ending by 01:00 hours the next day). More than 99% of all trawl tows included in this study were of 10-min duration. The few exceptions lasted 5-9 min when nets were caught on debris on the bottom. Boat speed was approximately 3.5 km h)1 during each tow. Summer trawling was conducted between the rst week of August and the rst week in September in each year. Autumn trawl samples were collected from late September to early November. Yellow perch captured were categorised as either age 0, age 1 or age 2 and older based on age-length keys

82.84 W

Lake Erie

41.58 N
Marble head p e ninsula

10

20 kilometers

Figure 1. Study sites (lled circles) in western Lake Erie.

Published 2010. The article is a US Government work and is in the public domain in the USA.

12

P. M. KOCOVSKY ET AL.

developed by Ohio Department of Natural Resources (Ohio Department of Natural Resources, unpublished data). log10(CPH + 1) for age-0 and age-1 yellow perch was calculated for each trawl sample to homogenise variance. The log-transformed CPHs (hereafter catch rates) were used to calculate seven different estimators of each of four indices of abundance of age0 and age-1 yellow perch. The four indices of abundance were summer age-0, summer age-1, autumn age0 and autumn age-1. The seven different estimators were average catch rate among all depths from: (1) all three diel periods; (2) morning only; (3) afternoon only; (4) night only; (5) morning and afternoon only; (6) morning and night only; and (7) afternoon and night only. Hereafter, the rst estimator is referred to as the full sampling regimen and the remaining six estimators are referred to as reduced sampling regimens.
Data analysis

The effects of year, season and diel period on catch rate were tested by two-way ANOVA. Separate ANOVAs were conducted for each age class because of the potential for age-1 catch rate to be autocorrelated with age-0 catch rate the previous year. The interactions between year and season, year and diel period and season and diel period were included in both models. The three-variable interaction term could not be tested owing to insufcient degrees of freedom. Tukeys Honestly Signicant Difference test was performed post hoc for signicant (P < 0.05) main effects that were not included in a signicant interaction. The modelling procedure used to establish the RAH for yellow perch on Lake Erie (Yellow Perch Task Group 2007) was used to assess how index values calculated from data collected during dierent diel periods aected predictions of abundance of age-2 yellow perch in western Lake Erie. The rst step in the modelling procedure included estimating abundance of age-2 yellow perch with a catch-at-age model (Deriso et al. 1985) using data from commercial and recreational harvests and state and provincial agency assessment surveys. Catch-at-age predictions of abundance were estimated using Auto Dierentiation Model Builder (ADMB; Otter Research Ltd 2000), and hereafter catch-at-age modelling is referred to as the ADMB model. The second step uses linear regression to predict the next years abundance of age-2 yellow perch using shery-independent indices of abundance of age-0 or age-1 yellow perch. Past years estimates of age-2 abundance from the ADMB model are used as the dependent variable, and past years indices of

abundance for age-0 or age-1 yellow perch from agency bottom-trawl surveys are used as the independent variables to create a suite of models. Separate models are constructed for each index from each of the several state, provincial and federal assessment surveys, such that only one index is used in each model (i.e. either an age-0 or age-1 index is used in each model, but not both). Current-year indices of abundance are then used to predict next years abundance of age-2 yellow perch (age-1 indices are lagged 1 year and age-0 indices are lagged 2 years). The average prediction of age-2 yellow perch abundance from those models with r2 > 0.5 and signicant (P < 0.05) slopes is used to set the RAH. The RAH is the primary biological estimate used in establishing the TAC, which is the maximum number of sh that can be legally harvested. This process for setting RAH has been in place since 2000. The modelling approach described above was used with a slightly different range of years for the analysis presented here. On Lake Erie, managers use survey data from 1983 to the present for projecting the next years abundance of harvestable yellow perch (Yellow Perch Task Group 2007). Thus, the sample size increases with each additional years data. Data for 19912005 were used in this study for two reasons. First, establishing a xed number of years for the analysis controlled for the eect of sample size on r2 values from regression analysis. Second, catch rate of yellow perch at the sample locations in western Lake Erie shifted toward greater night-time catch rate following 1990 (Stapanian et al. 2009). The effect of diel period of sampling on abundance estimates and t of models used to project abundance of age-2 yellow perch (the second step in the modelling process described above) was examined by comparing abundance predictions and r2 values of models using index values from the full sampling regimen to those using index values from each of the six reduced sampling regimens for each of the four indices of abundance. Index values from the full sampling regimen and the reduced sampling regimens (independent variable) were each regressed against age-2 abundance from the ADMB model for each of the four indices, and the abundance estimate and r2 from each of the reduced sampling regimens was compared with those of the full sampling regimen. A higher r2 for a model was interpreted as improved precision. Finally, ADMB estimates of abundance (from Yellow Perch Task Group 2009) were compared with those predicted from each signicant model (P for slope <0.05) that used data collected at night to determine whether data collected at night predicted higher or

Published 2010. The article is a US Government work and is in the public domain in the USA.

NIGHT SAMPLING FOR YELLOW PERCH

13

lower estimates of age-2 abundance than ADMB estimates. Results

(a) 1000
Morning Afternoon Night

100

Catch rate among years

CPH + 1

Summer catch rate for age-0 yellow perch was the highest at night in 9 of the 15 study years and 7 of the last 8 years (Fig. 2A). Autumn catch rate for age-0 yellow perch was the highest at night in 12 of the 15 study years and in 6 of the last 8 years (Fig. 2B). Summer catch rate of age-1 yellow perch was the highest at night in 9 of the 15 study years and in 6 of the last 8 years (Fig. 3A). Autumn catch rate of age-1 yellow perch was the greatest at night in 13 of 15 study years and in 7 of the last 8 years (Fig. 3B). Thus, night catch rates were higher in most years, especially since 1997, for all indices of abundance. For both age-0 and age-1 yellow perch, year, season, diel period and the interaction between year and

CPH + 1 10 1 1991 (b) 1000


Morning Afternoon Night

1994

1997 Year

2000

2003

100

10
(a) 100 000 10 000 CPH + 1 1000 100 10 1 1991 (b) 10 000
Morning Afternoon Night

Morning

Afternoon

Night

1 1991 1994 1997 Year


Figure 3. Catch rate measured as catch per hour + 1 (CPH + 1) of age-1 yellow perch in bottom trawls during 19912005 at three diel periods in summer (a) and autumn (b) in western Lake Erie.

2000

2003

1994

1997 Year

2000

2003

1000 CPH + 1

100

10

1 1991 1994 1997 Year 2000 2003

season were signicant (P < 0.05) in explaining catch rate (Table 1). The signicant interaction between year and season reects the multiplicative eect of variable year class strength and variation in recruitment to the trawl, particularly in summer. The interactions between season and diel period and year and diel period were not signicant for either age group. For both age0 and age-1 yellow perch, night sampling yielded greater catch rates than either morning or afternoon sampling (Tukeys HSD, P < 0.05). Morning and afternoon catch rate of age-0 or age-1 yellow perch did not differ (P > 0.05). As night catch rate differed from both morning and afternoon catch rates, each diel period was examined individually for comparisons of regression models.
Model precision and predictions of abundance

Figure 2. Catch rate measured as catch per hour + 1 (CPH + 1) of age-0 yellow perch in bottom trawls during 19912005 at three diel periods in summer (a) and autumn (b) in western Lake Erie.

The r2 values of models for predicting the abundance of age-2 yellow perch that included night catch rates

Published 2010. The article is a US Government work and is in the public domain in the USA.

14

P. M. KOCOVSKY ET AL.

Table 1. Results of

ANOVA

of catch rate of yellow perch captured in bottom trawls shed in western Lake Erie, 19912005 Age 0 Age 1 P>F <0.0001 <0.0001 0.0002 0.493 <0.0001 0.3684 MS 2.247 0.481 0.835 0.021 0.227 0.090 0.056 F 39.94 8.55 14.84 0.38 4.03 1.6 P>F <0.0001 0.0068 <0.0001 0.6864 0.0008 0.1094

Source Year Season Diel period Season diel period Year season Year diel period Error

d.f. 14 1 2 2 14 28

MS 2.834 6.834 0.971 0.0576 0.856 0.090 0.079

F 35.69 86.06 12.22 0.73 10.78 1.14

Seasons were summer and autumn; diel periods were morning (0.5 h after sunrise to 12:00 hours), afternoon (12:0118:00 hours) and night (0.5 h after sunset to no later than 01:00 hours the next morning).

tended to be higher than when night catch rates were excluded (Table 2), especially for the age-0 indices. Conversely, whenever only morning, only afternoon or the combination of morning and afternoon catch rates were used, r2 of regression models was usually less than that of models using night catch rate. For both age-0 indices, using morning, afternoon or morning and afternoon CPH usually produced regression models with r2 < 0.5, which was below the threshold for a model to be retained for use in calculating average abundance for establishing RAH. Using only night catch rate always produced regression models that projected lower abundance of age-2 yellow perch compared with all other sampling regimens (Table 2) for all indices except autumn age 0

(morning and night data produced the lowest abundance estimate). With the exception of using only afternoon catch rate for the summer age-1 index, regression models that used only morning, only afternoon or the combination of morning and afternoon catch rate predicted higher numbers of age-2 sh than all other sampling regimens. Thus, when considered from the perspective of averaging of model predictions for the purpose of establishing RAH, including catch rate data collected at night in regression models projecting abundance of age-2 yellow perch tended to drive abundance estimates lower. Regression models that used indices of abundance that included catch rate data collected at night typically produced lower estimates of age-2 yellow

Table 2. Regression parameters and estimated abundance of age-2 yellow perch (millions) for linear regressions predicting abundance of age-2 yellow perch in western Lake Erie from indices of abundance from the full sampling regimen [sampling in the morning (M), afternoon (A) and night (N)] and from six reduced sampling regimens using data from one or more diel periods Reduced sampling regimens Index Summer age 0 r2 Slope Y intercept Age-2 estimate r2 Slope Y intercept Age-2 estimate r2 Slope Y intercept Age-2 estimate r2 Slope Y intercept Age-2 estimate Full regimen 0.57 4.74 1.75 2.95 0.83 10.00 0.58 17.76 0.60 8.23 )2.37 1.17 0.69 10.48 3.46 3.46 M 0.51 4.31 4.60 5.30 0.82 10.23 2.03 31.04 0.35 7.07 1.24 5.41 0.59 8.80 6.58 6.58 A 0.42 3.80 7.48 8.71 0.67 10.00 3.18 16.09 0.48 6.81 5.06 6.16 0.51 9.18 8.06 8.06 N 0.65 4.99 )2.63 )1.30 0.76 7.30 2.24 9.72 0.73 6.86 )2.61 1.08 0.72 9.65 1.95 1.95 MA 0.48 4.16 5.66 6.67 0.77 10.48 1.97 23.59 0.46 7.64 1.37 4.24 0.58 9.48 6.54 6.54 MN 0.63 5.06 )0.71 0.38 0.86 9.41 0.48 18.65 0.61 8.22 )4.42 0.21 0.74 10.54 1.95 1.95 AN 0.57 4.75 1.19 2.59 0.80 9.44 0.78 11.70 0.66 7.74 )1.18 1.53 0.67 10.65 3.50 3.50

Summer age 1

Autumn age 0

Autumn age 1

Published 2010. The article is a US Government work and is in the public domain in the USA.

NIGHT SAMPLING FOR YELLOW PERCH

15

perch abundance compared with the average predicted age-2 abundance of the suite of models used to establish RAH during the entire time period that the current modelling procedure has been used. Since 2000, eighteen models that used night-time catch rate (indices calculated with full sampling regimens) have been included in the suite of models used to establish RAH (Table 3) in western Lake Erie. For 14 of these models, predictions using indices that included data collected at night were lower than the mean prediction of all retained models by an average of 45%, and for all 18 models the mean is 16% lower. Thus, in almost all years since 2000, abundance predictions that included night-time CPH have resulted in lower average projected abundance of age-2 yellow perch. Discussion The diel period when bottom trawl sampling took place affected the catch rate data used in the management process for yellow perch in western Lake Erie. Catch rates for age-0 (Stapanian et al. 2009) and age-1 yellow perch (USGS Lake Erie Biological Station, unpublished data) have typically been greater at night since 1991. This study demonstrated that including night catch rates tended to improve the t of models used for projecting abundance of age-2 yellow perch.
Table 3. Projected abundance of age-2 yellow perch (millions) from regression models using indices of abundance that included night sampling and average projected abundance for all regression models used to establish recommended allowable harvest of yellow perch in western Lake Erie, 20002008 Average projected age 2 8.927 5.11 3.318 77.997 1.409 1.409 1.409 31.347 31.347 31.347 2.633 2.633 2.633 23.244 23.244 18.556 18.556 18.556

Report year 2008 2007 2006 2005 2004 2004 2004 2003 2003 2003 2002 2002 2002 2001 2001 2000 2000 2000

Index Autumn Autumn Autumn Autumn Autumn Summer Autumn Summer Autumn Autumn Summer Autumn Autumn Summer Summer Autumn Summer Summer age age age age age age age age age age age age age age age age age age 1 1 1 0 1 1 0 1 1 0 1 0 1 0 1 1 1 0

Model P, r2 <0.01, <0.01, <0.01, <0.01, <0.01, <0.01, <0.01, <0.01, <0.01, <0.01, <0.01, <0.01, <0.01, <0.01, <0.05, <0.01, <0.01, <0.05, 0.68 0.68 0.67 0.68 0.74 0.72 0.68 0.75 0.74 0.70 0.80 0.71 0.71 0.68 0.52 0.59 0.59 0.54

Projected age 2 29.945 0.145 1.523 22.1 0.489 1.149 1.902 32.174 25.72 4.4 1.863 1.338 1.957 16.129 13.146 18.178 31.336 11.907

The better tting models produce more precise estimates of abundance of yellow perch, which in turn means greater condence in RAH and a stronger basis for the use of RAH in the process of setting TAC. Each of these benets and potential causal factors of shifts in CPH over time are discussed below. Including catch rates from night sampling in index values consistently resulted in lower estimates of age-2 yellow perch abundance compared with index values that included catch rates from morning or afternoon sampling. Consistently lower estimates of age-2 sh over time would result in lower variation among years, which has the potential to reduce uctuations in RAHs and reduce population uctuations brought about by harvest as has been observed by Jonzen et al. (2001) (but see also Anderson et al. 2008). Although lower estimates are not necessarily better for harvest management, and may even be inferior in terms of commercial value if harvest is lower than the population can sustain, they reduce the risk of overshing. An example of the practical value of lower abundance estimates was the overharvest of the 2003 year class of yellow perch in western Lake Erie. Indices of abundance from bottom trawl sampling of age-0 yellow perch in 2003 and age-1 yellow perch in 2004 were among the highest in the historical time series (Yellow Perch Task Group 2005). Both ADMB model predictions, which are heavily inuenced by previous years harvests, and the average of regression-model estimates of abundance of age-2 sh resulted in one of the highest RAHs and TACs in more than two decades. The Ohio Department of Natural Resources (ODNR) responded to the apparently high abundance by increasing bag limits for recreational anglers from 25 to 30 sh per day in Ohio waters. By autumn 2007, overshing of the 2003 year class coupled with poor recruitment of the 2004 and 2005 year classes (Yellow Perch Task Group 2008) and a change in the ODNR method for allocating harvest quota to commercial and recreational sheries resulted in no harvest allocated to the commercial shery in western Lake Erie in 2008 and a reduction in recreational bag limits in Ohio waters to 25 sh per day. A review of model predictions that were used to estimate abundance of age-2 sh in 2005 (Yellow Perch Task Group 2005 and Table 3) revealed that an index evaluated in these analyses (autumn age 0), the only index in western Lake Erie to include night sampling, produced the lowest estimate of abundance of age-2 yellow perch. All other indices were calculated from primarily afternoon sampling. The analyses presented here demonstrated that abundance estimates produced using index values that include data collected at night

Published 2010. The article is a US Government work and is in the public domain in the USA.

16

P. M. KOCOVSKY ET AL.

have been below the average of all models used in estimating age-2 abundance for establishing harvest quotas in most years since this modelling procedure has been in place. Lower estimates of the abundance of the 2003 year class of yellow perch may have averted the overharvest that contributed to closure of the commercial shery in western Lake Erie in 2008. Management agencies must be aware that lower estimates of yellow perch abundance produced by index values that use data collected at night may result in a lower RAH than the population could sustain and, in turn, unrealised economic gains for the valuable commercial and recreational sheries. The risk of unrealised economic gains because of lower estimates of abundance is probably lower than the risk of overharvest due to overestimating yellow perch abundance for two reasons. First, there are no clear examples in the data set to date that lower estimates of abundance of yellow perch have resulted in unrealised catch as there are for overharvest of the 2003 year class. This does not mean that there have not been instances of unrealised catch in the past, just that there is no evidence of it, and more directly that there is no evidence of it related to sampling at night. The data series is rather short (since 2000, which is when the current Yellow Perch Task Group modelling approach began), and it is possible that RAH at some time in future will be lower than the population can sustain if night-time sampling were to become more common in western Lake Erie. Second, harvest is dictated by TAC, and decision makers use criteria in addition to RAH, such as potential economic impacts, when setting TAC. A review of ADMB predictions shows that initial ADMB estimates of age-2 yellow perch abundance (i.e. the rst time abundance of an age-2 cohort is estimated using the ADMB model) have usually been biased highly since this modelling procedure has been used. Deriso et al. (1985) acknowledged the potential for initial catch-at-age estimates to be biased (either high or low) because of the small amount of information used in the analysis the rst time a cohorts size is estimated (i.e. only one seasons harvest data). Initial estimates of abundance of age-2 yellow perch since the year 2000 were higher than the subsequent estimate by an average of 13% (Yellow Perch Task Group 2001, 2002, 2003, 2004, 2005, 2006, 2007). Since 2002, rst estimates have been higher than the subsequent estimate by 30%. The high bias of ADMB estimates in recent years may be a result of high TACs and high harvest rates that have achieved or nearly achieved TAC in most years. If the trend of high-biased ADMB estimates continues, lower estimates of abundance of age-2 yellow perch from regression models that use

night catch rate data would decrease RAH, thus potentially decreasing TAC, which would decrease the potential for overharvest (Walters & Maguire 1996). It may seem counterintuitive that higher catch rate from night sampling would produce lower estimates of abundance of age-2 yellow perch. This effect was a result of changes in not only slopes but also Y-intercepts of predictive models when night catch rate was used. For example, the slope of the autumn age-0 index using morning and night data was higher than for all other reduced sampling regimens and only marginally smaller than the slope for the full sampling regimen (Table 2). This is expected because catch rate was higher at night. However, the Y-intercept was also the lowest, and the resulting projected abundance for that reduced sampling regimen was the lowest for the autumn age-0 index. Thus, while higher slopes of regression equations produced from indices that use higher catch rate would tend to drive estimates of abundance of age-2 yellow perch higher, lower Y-intercepts for many regressions that include nighttime data oset those gains resulting in lower estimates of abundance. The results of the analyses presented here suggest that night-time sampling may improve indices of abundance of age-0 and age-1 yellow perch used in the management process for establishing RAH in western Lake Erie. The data series analysed here has provided strong relationships with ADMB estimates of age-2 yellow perch abundance in western Lake Erie in the recent past and is, therefore, an important data set for managing harvest of yellow perch in western Lake Erie. However, sampling was conducted in a comparatively small near-shore area of western Lake Erie that is not necessarily representative of the entire basin, which has offshore areas up to 12 m deep. An expanded study that includes night-time sampling throughout the western basin, and perhaps in other basins, will be needed to demonstrate whether the results presented here apply to broader areas of Lake Erie. Sampling at night presents several challenges, such as additional safety measures that must be taken, inconvenience to researchers and vessel crews and additional labour costs (e.g. overtime). Management agencies will have to weigh the additional costs and inconveniences of night-time sampling against the potential benets to management of yellow perch. Acknowledgments Assistance with data collection and vessel operations were provided by J. Bales, L. Brohl, M. Bur, W.

Published 2010. The article is a US Government work and is in the public domain in the USA.

NIGHT SAMPLING FOR YELLOW PERCH

17

Edwards, S. Galbreth, P. Gerrity, B. Ickes, A. Jones, J. Jones, B. Lilla, C. Madenjian, K. Muth, C. Muzinic, J. Myers, M. Porta, K. Powell, R. Stickel, S. Tinnerello and D. Wolfert. Vessel support was provided by T. Cherry, T. Girard, D. Hall, M. McCann, F. Notestine and P. Pepper. Constructive reviews were provided by M. Thomas, J. Schaeffer and three anonymous reviewers. Mention of a brand name does not imply endorsement by the Federal government. This article is Contribution 1554 of the USGS Great Lakes Science Center. References
Anderson N.K., Hsieh C., Sandin S.A., Hewitt R., Hollowed A., Beddington J., May R.M. & Sugihara G. (2008) Why shing magnies uctuations in sh abundance. Nature 452, 835839. Baker J.R., Peck D.V. & Sutton D.M. (1997) Environmental Monitoring and Assessment Program Surface Waters Field Operations Manual for Lakes, EPA/620/R-97/001. Washington, DC: Environmental Protection Agency, 317 pp. Casey J.M. & Myers R.A. (1998) Diel variation in trawl catchability: is it as clear as day and night? Canadian Journal of Fisheries and Aquatic Sciences 55, 23292340. Deriso R.B., Quinn T.J., II & Neal P.R. (1985) Catch-age analysis with auxiliary information. Canadian Journal of Fisheries and Aquatic Sciences 42, 815824. Folmer O. & Pennington M. (2000) A statistical evaluation of the design and precision of the shrimp trawl survey o west Greenland. Fisheries Research 49, 65178. Glass C.W. & Wardle C.S. (1989) Comparison of the reactions of sh to a trawl gear, at high and low light intensities. Fisheries Research 7, 249266. Jonzen N., Lundberg P., Cardinale M. & Arrhenius F. (2001) Variable shing mortality and the possible commercial extinction of the eastern Baltic cod. Marine Ecology Progress Series 210, 291296. Korsbrekke K. & Nakken O. (1999) Length and speciesdependent diurnal variation of catch rates in the Norwegian Barents Sea bottom-trawl surveys. ICES Journal of Marine Science 56, 284291. Larsen D.P., Thornton K.W., Urquhart N.S. & Paulsen S.P. (1994) The role of sample surveys in monitoring the condition of the Nations lakes. Environmental Monitoring and Assessment 32, 101134. Leach J.H. (1999) Lake Erie: passages revisited. In: M. Munawar, T. Edsall & I.F. Munawar (eds) State of Lake Erie (SOLE) Past, Present, and Future. Leiden, The Netherlands: Backhuys Publishers, pp. 522. McAllister M.K. & Pikitch E.K. (1997) A Bayesian approach to choosing a design for surveying shery resources: application to the eastern Bering Sea trawl survey.

Canadian Journal of Fisheries and Aquatic Sciences 54, 301311. Michalsen K., God O.R. & Ferno A. (1996) Diel variation in the catchability of gadoids and its inuence on the reliability of abundance indices. ICES Journal of Marine Science 53, 389395. Otter Research Ltd (2000) An Introduction to AD Model Builder Version 4.5 for Use in Nonlinear Modeling and Statistics. Sidney, BC: Otter Research Ltd, 187 pp. Schnute J.T. & Haigh R. (2003) A simulation model for designing groundsh trawl surveys. Canadian Journal of Fisheries and Aquatic Sciences 60, 640656. Smith S.J. & Gavaris S. (1993) Improving the precision of abundance estimates of eastern Scotian Shelf Atlantic cod from bottom trawl surveys. North American Journal of Fisheries Management 13, 3547. Stapanian M.A., Bur M.T. & Adams J.V. (2007) Temporal trends of young-of-year shes in Lake Erie and comparison of diel sampling periods. Environmental Monitoring and Assessment 129, 169178. Stapanian M.A., Kocovsky P.M. & Adams J.V. (2009) Change in diel catchability of young-of-year yellow perch in Lake Erie associated with establishment of dreissenid mussels. Freshwater Biology 54, 1593 1604. Walsh S.J. (1991) Diel variation in availability and vulnerability of sh to a survey trawl. Journal of Applied Ichthyology 7, 147159. Walters C. & Maguire J. (1996) Lessons for stock assessment from the northern cod collapse. Reviews in Fish Biology and Fisheries 6, 125137. Yellow Perch Task Group (2001) Report of the Lake Erie Yellow Perch Task Group. Report to the Great Lakes Fishery Commission. Ann Arbor, MI: Great Lakes Fishery Commission, 50 pp. Yellow Perch Task Group (2002) Report of the Lake Erie Yellow Perch Task Group. Report to the Great Lakes Fishery Commission. Ann Arbor, MI: Great Lakes Fishery Commission, 40 pp. Yellow Perch Task Group (2003) Report of the Lake Erie Yellow Perch Task Group. Report to the Great Lakes Fishery Commission. Ann Arbor, MI: Great Lakes Fishery Commission, 45 pp. Yellow Perch Task Group (2004) Report of the Lake Erie Yellow Perch Task Group. Report to the Great Lakes Fishery Commission. Ann Arbor, MI: Great Lakes Fishery Commission, 43 pp. Yellow Perch Task Group (2005) Report of the Lake Erie Yellow Perch Task Group. Report to the Great Lakes Fishery Commission. Ann Arbor, MI: Great Lakes Fishery Commission, 48 pp. Yellow Perch Task Group (2006) Report of the Lake Erie Yellow Perch Task Group. Report to the Great Lakes

Published 2010. The article is a US Government work and is in the public domain in the USA.

18

P. M. KOCOVSKY ET AL.

Fishery Commission. Ann Arbor, MI: Great Lakes Fishery Commission, 48 pp. Yellow Perch Task Group (2007) Report of the Lake Erie Yellow Perch Task Group. Report to the Great Lakes Fishery Commission. Ann Arbor, MI: Great Lakes Fishery Commission, 49 pp. Yellow Perch Task Group (2008) Report of the Lake Erie Yellow Perch Task Group. Report to the Great Lakes

Fishery Commission. Ann Arbor, MI: Great Lakes Fishery Commission, 45 pp. Yellow Perch Task Group (2009) Report of the Lake Erie Yellow Perch Task Group. Report to the Great Lakes Fishery Commission. Ann Arbor, MI: Great Lakes Fishery Commission, 47 pp.

Published 2010. The article is a US Government work and is in the public domain in the USA.

Copyright of Fisheries Management & Ecology is the property of Blackwell Publishing Limited and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

You might also like