Is It the Gear or the Place? Decomposing Fishing-Method Selectivity in 26,438 Taiwan Catch Records

Anglers hold strong species associations for each method: lures for predators, bottom rigs for bottom feeders, poles for small fish. All of them look true in the data. But "looks true" has two explanations: the gear genuinely selects fish, or people who fish that way go to places where those fish live. In catch records the two are indistinguishable until location is held fixed. We hold it fixed, and separate them.

Finding: Three quarters of what fishing method appears to explain about species is really anglers choosing places; gear itself accounts for a quarter. What gear reliably changes is size, not species.

Analysis date 2026-09-05 · Data 2025-01-01 to 2026-09-04 · 中文版

• H1 Gear selects species: method still explains species composition with water body held fixed. → holds, at a quarter of the marginal size
• H2 Within-angler: the same angler switching method catches different species. → holds; the gear effect is not an angler effect
• H3 Swim-layer mechanism: bottom rigs skew to bottom species, lures to surface and mid-water. → direction holds, clearest for bottom rigs
• H4 Gear selects size: same species, same water, method changes length. → holds and is robust across three specifications
• H0: the method–species association is entirely location. → rejected, though location explains most of it

We ask an information question: how much does knowing the method reduce uncertainty about the species caught? Species uncertainty is 5.97 bits. Knowing method alone removes 0.48 bits (8%). If the water body is already known, method removes only a further 0.12 bits (2%), still significantly above a null that shuffles methods within each water body (p < 0.01).

Of the method's marginal information, 25% survives controlling for location; the remaining 75% is "method chooses place". The water body alone carries 3.41 bits, seven times what method does. To catch a particular species, choose the place first and the method second.

Within-angler control. Could the gear effect be a difference between people? Among 1,121 anglers who used two or more methods (13,061 records), the same person switching method does catch different species: bias-corrected conditional mutual information 0.14 bits (p < 0.001). The gear effect is real; it is just small.

For each method, the expected catch of each species if gear were neutral is the method's catch count in each water body times that water body's species share; the index is observed over expected. An index of 2.0 means twice the local average. Cells with at least 30 observed or expected records.

Species names shown in Chinese as recorded.

The preferences follow the physics of the gear: bottom rigs take carp, whiting and eel catfish; lures take chasing predators such as cutlassfish, barracuda and cuttlefish; maebari takes structure-hugging grouper and puffers; rock fishing takes rabbitfish and pompano. Avoidances are equally sensible: lures almost never take algae-grazing rabbitfish; rock and pole fishing almost never take snakehead. This is the real quarter.

Replacing species with swim layer (surface, mid, bottom; a species constant) and repeating the observed-versus-expected comparison. This is a mechanism summary, not an independent test.

Bottom fishing is the clearest: bottom-layer species 95% against an expected 80%, surface 1% against 7%. Lures shift slightly away from the bottom, maebari and pole slightly toward it. Every direction is right, but most shifts are within five points, which again says gear selectivity exists but is modest.

Same species, same water body, same month: how big are the fish each method lands? Log length regressed on method with species, water-body and month fixed effects absorbed and angler-clustered standard errors, 23,437 measured records. Relative to lure:

* p < 0.05. The angler fixed-effects column uses only anglers who fished two or more methods.

Pole fishing lands fish 20% smaller than lure fishing, 18% on ruler-measured records and 15% within the same angler, significant in all three. Boat fishing lands fish 36% larger and shore casting about a tenth larger. Rock and bottom fishing do not differ from lure. Species by species, 20 of 61 contrasts are significant (about 3 expected by chance) with a median effect of 14%, and the direction is consistent: for all three tilapias, pole-caught fish are 12 to 27% smaller than lure-caught; for mangrove red snapper, bottom-caught fish are 26% larger.

Measured

• Mutual information of method with species: 0.48 bits marginal, 0.12 bits conditional on water body (25% retained); water body alone 3.41 bits.
• Within-angler conditional mutual information 0.14 bits (p < 0.001).
• Length: pole -19.9%, boat +36.4% relative to lure, consistent across three specifications; 20/61 per-species contrasts significant.
• Bottom-rig bottom-layer share 95% against expected 80%.

Inferred, to be read separately

• "Place" is the water-body node; microhabitat differences within a water body (bank versus deep water) are counted as gear effect, so 25% is an upper bound on gear selectivity.
• Size results describe fish that were caught and logged, not fish that encountered the gear; pole anglers might log small fish more or less readily. The ruler-only and within-angler results agree, which softens but does not remove the concern.
• Swim layer is a species constant, so the layer analysis is a re-reading of the species result rather than independent evidence.
• Without effort data we cannot say which method is more efficient; this study is about composition and size, not catch rate.

Gear selectivity is an old topic in commercial fisheries; the recreational evidence comes largely from controlled experiments or single fisheries (natural bait types, lure versus bait for Baltic cod, size selectivity in stream salmonids). Those studies measure selectivity at one site or under controlled conditions, so they never face the "method chooses place" confound. What this study adds is a measurement of that confound in free-ranging behaviour at scale: where anglers choose their own locations, three quarters of apparent method selectivity is location.

• Alós, J., Arlinghaus, R., Palmer, M., March, D. & Álvarez, I. (2009). The influence of type of natural bait on fish catches and hooking location in a mixed-species marine recreational fishery, with implications for management. Fisheries Research 97: 270–277. sourceMediterranean mixed-species recreational fishery: bait type shifts catch composition, size and hooking location; regulating bait is a simple compositional lever.
• Weltersbach, M.S., Strehlow, H.V., Ferter, K., Klefoth, T., de Graaf, M. & Dorow, M. (2018). Effect of lure and bait type on catch, size, hooking location, injury and bycatch in the western Baltic Sea recreational cod fishery. Fisheries Research 210: 121–130. sourceBaltic cod: lure versus natural bait differ in catch, size, injury and bycatch; the same gear-versus-size question as ours.
• Species-specific vulnerability to angling and its size-selectivity in sympatric stream salmonids. Canadian Journal of Fisheries and Aquatic Sciences (2021). DOIAngling vulnerability and size-selectivity differ among co-occurring stream salmonids; selectivity must be read per species, not per gear alone.

1. Data. Catch records from 2025-01-01, non-fish excluded, seven methods with 300+ records and species with 30+: 26,438 records, 125 species, 5,021 water bodies (merged to a named parent where one exists).
2. Information. Mutual information I(species; method) and I(species; water body), conditional mutual information I(species; method | water body) and I(species; method | angler). Plug-in estimates are biased upward, so each is corrected by the mean of 200 permutations that shuffle method within water body or within angler; p is the share of permutations at or above the observed value.
3. Selectivity index. Expected = Σ over water bodies of (method's catch count there × the water body's species share); index = (observed + 0.5) / (expected + 0.5).
4. Size. Log length regressed on method dummies with species, water-body and month fixed effects absorbed by alternating demeaning; angler-clustered robust standard errors. Ruler-only and angler-fixed-effects (angler additionally absorbed) as robustness. Per-species models absorb water body within species.

Read-only analysis; no production data modified. Aggregate tables for replication are available on request.

The aggregates, figures and methods here are free to read and cite; please retain the data period, sample size and analysis date. Public pages are free to read and cite page by page. Raw records are not available for download, and bulk reuse licensing for aggregated data is still under review.

We are looking for collaborators. Turning the upper bound on gear selectivity into a point estimate needs within-water-body microhabitat information or a controlled experiment. If you work on gear selectivity or recreational fisheries management, we can provide de-identified method-by-species-by-water-body aggregates. Write to support@dudefishing.net.