Premium
An Analysis of Methods for Quantifying Crappie Recruitment Variability
Author(s) -
Isermann Daniel A.,
McKibbin William L.,
Willis David W.
Publication year - 2002
Publication title -
north american journal of fisheries management
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.587
H-Index - 72
eISSN - 1548-8675
pISSN - 0275-5947
DOI - 10.1577/1548-8675(2002)022<1124:aaomfq>2.0.co;2
Subject(s) - statistics , sampling (signal processing) , demography , mathematics , engineering , electrical engineering , filter (signal processing) , sociology
We used catch‐at‐age data from 122 populations of crappies Pomoxis spp. from across the midwestern and southeastern United States to compare two methods for defining recruitment variability: (1) the recruitment coefficient of determination (RCD), calculated from the weighted linear regression of log e (catch at age + 1) on age, and (2) the recruitment variability index (RVI). Ranked values of the RCD and RVI were significantly correlated, which suggests that both indices measured variability in catch‐at‐age data in a similar manner. The recruitment categories of RCD and RVI agreed 52% of the time or were within one category of each other 89% of the time. Modeling indicated that the RCD and RVI provided theoretically valid measures of recent recruitment variability but were not suitable for assessing long‐term recruitment patterns. Ranked values of RCD and RVI were not significantly associated with ranked coefficients of variation (100·SD/mean) in catch per unit effort (catch per trap‐net night) of fully recruited crappies (age 1 or age 2) observed in long‐term trap‐net sampling. Studentized residuals resulting from RCD regressions failed to represent the fluctuations in year‐class strength observed in long‐term trap‐net sampling for 15 of the 17 populations we examined. We conclude that the variability in crappie catch‐at‐age data that is not explained by annual mortality may not be entirely attributable to recruitment variability and could instead be a product of gear biases associated with fall trap‐net sampling. The insensitivity of RVI to situations in which year‐classes are weak rather than missing may limit its use as a measure of recruitment variability. Trends in recruitment detected in long‐term, annual samples might provide the best means for assessing recruitment variability in fish populations. For infrequently sampled populations, methods such as RCD, RVI, and residual analysis may offer the only options for assessing recruitment patterns.