Size Selectivity of Multifilament Gill Nets for Sampling Alligator Gar: Modeling the Effects on Population Metrics

Abstract

Gill nets are commonly used to sample Alligator Gar Atractosteus spatula; however, gill nets are size selective. We used five common multifilament gill-net meshes (88.9-, 101.6-, 114.3-, 127.0-, and 139.7-mm bar measure) to collect 477 gar between 1,000 and 2,300mm TL. We fit size-selectivity functions for these meshes for a net in which each mesh was fished equally, i.e., equal-effort net design (EEND), and found selectivity was best described by a binormal distribution in which the assumption of geometric similarity had been relaxed (i.e., binormal with deviations). Overall relative retention was about 30% for fish at 1,200 mm, increased to 100% for fish at 1,670 mm, then declined to 50% when fish were 2,100 mm. To reduce size selectivity, we used an optimization to create a net that minimized bias over the greatest TL range of fish sampled, i.e., weighted-effort net design (WEND). We compared the WEND to the EEND via simulation to estimate the effects of selectivity bias on population metrics. The WEND had 70% or greater relative retention for all fish between 1,200 and 2,100 mm TL. While the WEND performed slightly better with respect to the length frequency, simulations indicated that sampling Alligator Gar populations with either net design led to estimates of vital statistics for fish above 1,200 mm that were only slightly biased (e.g., survival estimates were biased by <2%). Both designs underestimated the proportion of smaller fish (i.e., <1,200 mm), which affected estimates of most vital statistics for these fish. When the goal of sampling is to estimate vital parameters for Alligator Gar < 1,200 mm TL, biologists should consider using different mesh sizes or methods. Bias adjustments for larger and older fish are probably unnecessary. Our results suggest the gill-net meshes we tested can be used to collect a representative sample of adult Alligator Gar.

Received October 29, 2015; accepted February 2, 2016 Published online May 26, 2016

Acknowledgments

We thank T. Brenden for his consultation and the ADMB code used to estimate the binormal with deviations selectivity function. Additionally, we thank Texas Parks and Wildlife Department, Inland Fisheries, District 1E for sampling assistance. Constructive comments provided by Dave Buckmeier, Nate Smith, Bob Betsill, and anonymous reviewers greatly improved this manuscript. This study was supported by Federal Aid in Sport Fish Restoration Grants F-231-R and F-221-M to the Texas Parks and Wildlife Department, Inland Fisheries Division.