The effect of hook spacing on longline catch rates / by Francisco Blaha

As I said more than once, I like reading scientific papers (having time to do it is more the issue) this one got me thinking, and again I see science confirming stuff you talk on board with people that have been doing the job longer than you: “hooks compete with each other”. And while the job was done in Halibut, I wonder how it would work on toothfish, ling and/or long tail snapper as species closer to my experience.

my strongest memories of longliners... the cursing when the mainline started to mess up

my strongest memories of longliners... the cursing when the mainline started to mess up

The paper is on the dense side for the non initiated, so I hope the authors Cole C.Monnahan and Ian J.Stewart don't mind I only quote the abstract and the discussion.

Catch per unit effort (CPUE) is a widely used index of population abundance for informing stock assessments for the purpose of estimating population status and setting fishing policies. However, for CPUE to be an unbiased index, influences that are not related to population abundance (e.g., spatial variation in effort and temporal changes in gear efficiency) must be accounted for in analyses known as CPUE standardization. In longline fisheries, one important factor that can affect CPUE is the spacing between hooks (‘spacing effect’), which influences effective effort and has largely been ignored in previous analyses. Here, we use the Pacific halibut (Hippoglossus stenolepis) long-line fishery as a case study to explore the spacing effect.

Both commercial and experimental (fishery-independent) data with hook spacing, and a survey-based CPUE series, are available for this fishery. It thus provides a unique opportunity to explore the effect of hook spacing and its effect on CPUE trends. We quantify this effect using non-parametric and parametric relationships inside a spatially-explicit (geospatial) CPUE standardization model for commercial data, and non-linear mixed-effects model for experimental data.

We found a clear non-linear spacing effect (i.e., hooks were less effective the closer they were), but accounting for space had a larger effect on CPUE trends than accounting for hook spacing. For this stock, it is likely the effect of hook spacing on CPUE was minimal due to little variation in average hook spacing over time. Regardless, historical and future trends in hook spacing can have important effects on longline CPUE standardization, highlighting the value of collecting this information. Accounting for hook spacing effects in other fisheries may improve estimates of trends in relative abundance and lead to better management.

We found clear evidence for reduction in hook fishing power (or effectiveness) at smaller spacings, supporting the hypothesis that nearby hooks compete for Pacific halibut. This implies that for CPUE analyses, the relevant unit of effort is an effective hook.

We also found that the parametric form was a reasonable approximation for this relationship. Further, the parametric model fits to both the fishery-dependent and experimental data sets were fairly consistent, demonstrating this relationship can be estimated directly from commercial data, without the need for a controlled experiment.

Estimating effective hooks in the CPUE standardization has the added benefit that the uncertainty in the spacing effect is propagated into the trends of relative abundance. Lastly, despite a clear hook spacing effect, we found limited effects on standardized CPUE trends. This was likely because although there has been a temporal shift to different gear types, on average the hook spacing has changed slightly over the time period examined. Comparisons among other regulatory areas with systematic differences in gear usage may be much more important to the interpretation of Pacific halibut trends. Further, in other stocks managed with longline CPUE with significant temporal trends, ignoring hook spacing may mischaracterize abundance trends and lead to poor management decisions.

Our results support the hypothesis that hooks compete with each other, at least under the densities observed, and conditioned on the specific foraging behavior of Pacific halibut. However, the commercial data used here only contained information on retained legal halibut and set-level characteristics, and did not include key factors that certainly affect catch rates. For example, we were not able to account for the effects of environmental factors nor halibut size structure and density. Neither we were able to account for multispecies competition, which also has important influences on longline catch rates (Rodgveller et al., 2008). Thus, we caution against a biological interpretation of our results, and against applying our estimates to other species or situations, as foraging behavior may vary widely and lead to fundamentally different relationships (Fig. 1). For instance, initial captures of sablefish do not affect subsequent captures leading to a random distribution of occupied hooks, while Pacific halibut tend to cluster (Sigler, 2000). Future lab experiments on Pacific halibut or other species, while controlling for environmental and other key factors, would provide valuable corroboration and further insights in the relationship between individual foraging behavior, hook competition, and the resulting population-level hook spacing effects.

The assessment of Pacific halibut uses CPUE that excludes snap and autoline gear due to concerns over confounding between gear type, hook spacing, and changes in density (Stewart et al., 2016). Our analysis provides a method for including all gear types in future analyses and improving the information on which management is based. Although our analysis is specific to Pacific halibut, similar analyses for other stocks assessed, at least in part, with standardized longline CPUE could use a similar approach. For instance, hook spacing for sablefish is known to be important from experiments, but is not consistently reported for commercial catches and thus cannot be directly used in the CPUE standardization (Sigler and Lunsford, 2001). Likewise, CPUE analyses for bigeye tuna account for hooks between floats and hooks per set, but the length of sets are unreported and thus the effect of hook spacing is unknown (e.g., Hoyle and Okamoto, 2011). Our results demonstrate the potential value in collecting hook spacing for commercial longline catch data, and suggest incorporating this information in the future, especially for stocks with temporal or spatial trends hook spacing over time.

Efforts to estimate fish stock status from longline CPUE trends while ignoring spatial effort have been widely criticized (e.g., see debates in Hampton et al., 2005; Myers and Worm, 2003; Walters, 2003). As a consequence, incorporating spatial strata into standardizations is commonplace (Maunder and Punt, 2004). However, these improved methods still typically ignore spatial correlation among cells, and can be sensitive to cell resolution (Ichinokawa and Brodziak, 2010; Tian et al., 2010). One promising new method for accounting for space in standardizations is hierarchical spatiotemporal models (Thorson et al., 2015). Hierarchical models have become increasingly popular tools across a wide range of applications in fisheries science (Thorson and Minto, 2014), and their application for spatiotemporal models provides a natural approach for dealing with spatial complexities when estimating fish densities. In contrast to data collected using a random design (e.g., surveys), the preferential sampling of commercial data (i.e., high density areas are targeted; see Conn et al., 2017; Diggle et al., 2010) remains an open issue when using these methods.

We did not attempt to address this issue in our simplified model, here used as a proof of concept and to investigate hook spacing effects, but note we were encouraged that our estimates closely matched a survey CPUE trend (Fig. 6). However, before using these methods for management, we suggest future studies more closely investigate the effects of preferential sampling, in addition to other factors ignored here (e.g., zero catches and anisotropy), which may have an important influence on some stocks. We expect development of these models to continue being an active area of research, and will eventually be applied widely to analyze complex spatial fisheries data.

Trends in CPUE may not accurately reflect true trends in abundance due to a wide variety of confounding factors. Accounting for all such confounding factors is thus critical for successful fisheries management, but is a difficult proposition and will be a source of continued research.

For longline gear, in particular, the spacing between hooks clearly effects the effective effort leading to observed catches. This highlights the value in collecting hook spacing data on longline sets, particularly if there is the potential for an annual trend in hook spacing as gear configuration evolves in a fishery. Fortunately, the effective effort implied by hook spacing can be estimated within a spatially-explicit CPUE standardization model fit to commercial catch data. Including this effective hook relationship will likely lead to improved trends in relative abundance, and hence better management for other species caught by longline.