Thirty-five experimental fisheries reveal the mechanisms of selection
Summary (2 min read)
Introduction
- Fisheries have been described as large-scale evolutionary experiments; yet such 2 “experiments” tend to be poorly replicated and therefore lack the predictive power 3 essential for designing appropriate management strategies to minimize the effects 4 of fisheries-induced selection.
- The authors study helps explain why the outcomes of fishing-induced selection reported in 18 the literature are contradictory 4.
- Disentangling these 13 factors and the different components of selection (fitness-trait correlation, trait 14 variability, fitness variability) is critical to producing flexible and relevant 15 management strategies, but does rely on repeated measurements of selection.
- 16 To date, fish have been completely removed from over 60 locations (lakes and 17 rivers); and from these, the authors analyzed lake populations where at least 15 fish were 18 captured over at least three fishing events.
Catch and length 18
- The authors assessed the selective effect of fishing by correlating fish length with catch 19 intensity.
- The authors used the same type of model to 4 assess the effect of fishing intensity on the variability of length-at-capture, which 5 was measured after each fishing event, as the standard deviation of the length of all 6 the fish remaining in the lake.
- For the sake of simplicity, the authors first assumed that fish did not grow during the study 13 period (typically 2-3 years), which would result in a conservative 14 estimate of selection.
- The authors performed the same tests (linear mixed models) on the 7 random simulations, allowing direct comparison of the observed and simulated 8 data.
- If any reproduction 21 occurred, it likely occurred during the first year following removal initiation and 22 likely produced few fish as most adults were removed quickly.
Estimating selection differentials 15
- For each lake, the authors compared the mean length of surviving fish after each fishing 16 event (1092 events) to the mean length before the same fishing event, yielding a 17 selection differential for each event in each lake.
- Using this covariance, the authors can disentangle the different components of 20 selection into the products of (1) the correlation between the trait and fitness, (2) 21 the variability (standard deviation) of fitness, and (3) the variability of the trait 1 under selection (see below) 10.
- The standard deviation of the length in the 7 initial population.
- To assess if catch intensity also affected the variability in selection differentials, the authors 17 binned catch intensity into 5% intervals and estimated, for each lake, the average 18 selection differential in each 5% interval.
- This value represents the magnitude of length change when catch 8 increases and can be understood as fishing-induced selection.
Data and statistics 1
- The data that support the findings of this study belong to the National Park Service 2 and the California Department of Fish and Wildlife, and can be accessed through 3 them upon reasonable request.
- Significance 13 tests were computed according to Kuznetsova et al. 44 using the package lmerTest.
- The importance in fishery management of leaving 5 the big ones.
Authors contributions 5
- SN, APH, AMS, and SMC 6 discussed the analyses and wrote the manuscript.
- RAK and MTB collected data and 7 provided feedback on the manuscript.
- 8 Random fishing is expected to slightly reduce variability (right-hand panel, slope of 9 variability-catch relationship = -0.03 ± 0.04), but less than non-random fishing 10 where the slope of the variability-catch relationship is expected to be ten times 11 larger (0.32 mm ± 0.07 mm/%).
- Dashed black lines represent 4 the average response using 5% and 10% growth corrections, and black dotted lines 5 represent the response with correction for reproduction after the initiation of the 6 experimental fisheries (see Methods).
Did you find this useful? Give us your feedback
Citations
50 citations
Cites result from "Thirty-five experimental fisheries ..."
...48), as do recent results from experimental fisheries (49)....
[...]
8 citations
References
272,030 citations
50,607 citations
12,305 citations
4,990 citations
665 citations
Related Papers (5)
Frequently Asked Questions (11)
Q2. How was the variability in the selection differentials affected by catch intensity?
Since there is only one value per 2 catch interval, the authors used linear models to assess how the variability in selection 3 differentials was affected by catch intensity (Figure 3b).
Q3. How did the authors assess the effect of fishing on the length of fish?
The authors then 22assessed the relationship between catch intensity and length-at-capture by means of 1 linear mixed-models; these models included length-at-capture as response variable, 2 catch intensity and species as a fixed effects, and lake ID as a random effect, allowing 3 the slope and intercept to vary among lakes.
Q4. How many fish were removed during the first year of the experiment?
If any reproduction 21 occurred, it likely occurred during the first year following removal initiation and 22likely produced few fish as most adults were removed quickly.
Q5. How did the authors measure the effect of fishing on the length of fish?
To do so, the authors applied two growth scenarios: a slow growth correction 20 assuming 5% length increase per year and a fast growth correction assuming 10% 21 length increase per year.
Q6. What is the expected value of the relationship between the length and the variability?
Random fishing 6 is not expected to alter the size structure of the population (left-hand panel, slope of 7 the length-catch relationship is 0.08 ± 1.00, not significantly different from zero).
Q7. How did the authors assess the effect of fishing on the fish?
2To test if random sampling could explain the observed selective effects of fishing - 3 estimated as the slopes of the length-catch intensity relationships (Figure 1a), and 4 variability-catch intensity relationships (Figure 1b) - the authors performed 100 simulations 5 where mortality occurred randomly; i.e., the authors used random permutations of the time-6 at-capture of each fish.
Q8. What was the weighting 16 parameter for the selection differential?
overall selection 14 was the response variable, the three components of selection were predictor 15 variables (Figure 4), and the logarithm of population size was the weighting 16 parameter.
Q9. How were the fish prevented from reaching spawning areas in the associated streams?
14When possible, fish were prevented from reaching spawning areas in the associated 15 streams by makeshift dams and/or gillnets to block access to inlet and outlet 16 streams.
Q10. How many different selection differentials were calculated for each lake?
16To assess if catch intensity also affected the variability in selection differentials, the authors 17 binned catch intensity into 5% intervals and estimated, for each lake, the average 18 selection differential in each 5% interval.
Q11. How many fish were removed after the first year of the survey?
To estimate the 1 potential effect of reproduction after the initiation of the fisheries, the authors performed a 2 sensitivity analysis by repeating the analysis after removing fish smaller than 50 3 mm that were caught after the first year of survey, i.e., 334 fish potentially born after 4 removal initiation.