I agree with Freedman!

At the same time, I am also concerned about the ecosystem that stretches over 100 Km, over which you have used a bottom trawler and cleared off the bottom dwellers living in there.

Cara é by-catch da subida da rede, você pode pegar até Meka nesses arrastos durante a subida da rede, explica que devido a colmatação da rede alguns pequenos pelagicos que nao seriam acapturados normalmente entraram na rede e vieram com a amostra.

I think it depends on your research question. If you are only interested in the demersal or benthic species (for instance in relation to specific habitat/substrate use), then maybe you can exclude the pelagic fishes from your analyses but note that they were collected. If you are assessing the collection method or have more general questions, I would include all fishes. If you include the pelagic fishes in your analyses, then note that they are primary pelagic in your paper/report. You could also run analyses both including and excluding these species to ensure that, for instance, a common pelagic species is not confounding your research question about rare demersal species.

One thing to note, however, is that I think you need to be really confident that the pelagic fishes were in fact bycatch, and were not associating with the bottom. Since it sounds as though many of them were juveniles and were caught in relatively shallow water, keep in mind that there may be life-history variation in habitat use, and that therefore excluding the juveniles from the sample would be incorrect. So I think there's really no simple answer and it really depends on what your question is.

Bottom trawls as it sweeps through columnar waters do catch those varieties of fish that you have mentioned.Hence such a piece of data enhances the quality of the data collected & paper.

I agree with Freedman!

At the same time, I am also concerned about the ecosystem that stretches over 100 Km, over which you have used a bottom trawler and cleared off the bottom dwellers living in there.

I agree with Jonathan Freedman that including or excluding pelagics depends on the question you are trying to answer. However, if your question requires that you include pelagic species, you must also consider whether a bottom trawl is an adequate sampling device for your study. Bottom-trawl catchabilities vary among demersal species, of course (as well as among trawl designs and with other factors) but the catchability of a pelagic species in a particular bottom-trawl survey can additionally be highly variable over space and time -- which could introduce a lot of noise to the catch data, confusing the conclusions drawn. The problem lies in small changes in fish behaviour and/or environmental conditions that place the fish within a metre or two of the seabed, where a survey detects them, or else lift those fish above the headline height of the survey trawl. I have not seen much published discussion of this but Ian McQuinn (2009: Can J Fish Aquat Sci 66:2256-2262) argued that a much-cited "pelagic outburst" was an artifact that arose from depletion of demersal predators, allowing small-pelagics to feed close to the seabed, where they were taken by survey trawls. If he is right, a number of elegant ecological conclusions will collapse.

Ah yes, a reference. I can suggest a couple of citations related to benthic trawling in rivers. In Freedman et al. 2014 (Hydrobiologia, 727:19–33) we were using benthic trawls but did capture some pelagic fishes and noted in the Discussion that "By using an electrified benthic trawl, we were able to sample small benthic fishes that are difficult to sample using traditional methods. However, although we did also capture some non-benthic taxa, our sampling method was biased toward the

capture of benthic rather than pelagic or littoral fishes." Herzog et al. (2005, N. Am. J. Fish. Manage. 25:594–603) also captured some pelagic fishes with their benthic trawls, although they note that the benthic trawl is not efficient at capturing them. I suppose that this raises another interesting point: that if your benthic trawls are capturing, say, 90% of the benthic species but only 10% of the pelagic species, with similar differences in relative abundance, you could end up with disproportionate quality of data for benthic ("good" data) and pelagic ("bad" data).

Three issues I have not seen addressed yet are 1) trawl dimensions, 2) qualitative versus quantitative analyses, and 3) fishing up and fishing down effects. In western Lake Erie (smallest of North American Laurentian Great Lakes) we recently changed bottom trawl design from a net that fished 1 m above the bottom to ~3 m from the bottom. Our catch of clupeids increased greatly because we now sample "pelagic" habitat better. If your trawl fishes high enough to fish "pelagic" habitat, then you have justification for retaining those data. Regardless of whether your trawl fishes pelagic habitat, you caught them, hence you at least have a qualitative indication of species presence and I would retain data for that purpose alone. The possibility of confounding exists because the trawl is not deployed instantaneously on the bottom. Unless you have a Tucker trawl, which can be retained in a closed position until desired fishing depth is reached, then opened for fishing and closed again for retrieval, you might be catching the pelagic species while the trawl is being deployed to the bottom or retrieved. You could conduct "down-up" trials, in which you deploy the trawl to the bottom and retrieve it immediately to determine just how many individuals of pelagic species are caught during deployment and retrieval, then you could adjust your quantitative indices accordingly. In Lake Erie this is typically a small effect because our waters are comparatively shallow, but in your case they may be greater.

Maybe you can include them all in a summary table (with appropriate caveats) but for data analyses, I would recommend some type of criterion for inclusion (e.g. only analyze those species for which you catch AT LEAST 3 specimens at each station, or something appropriately in line with your sampling design). This prevents accidental catches from influencing data analysis. See Fossen et al. 2008 - Species composition and distribution patterns of fishes captured by longlines on the Mid-Atlantic Ridge.