The long-standing estimate of about one billion tonnes was built from net catches, and small fishes with good eyesight evade nets. Acoustic observations from a research vessel showed mesopelagic fish avoiding a trawl so effectively that catches greatly underestimated what the echosounder recorded (Kaartvedt et al., 2012). Acoustic data from a circumnavigation of the subtropical and tropical oceans then led to the suggestion that biomass could be an order of magnitude above the earlier figure (Irigoien et al., 2014).

Acoustic estimates carry their own difficulties. The echo from a scattering layer depends on which organisms are present and on the state of their gas-filled swimbladders, and gas-bearing siphonophores can return strong echoes that are easily taken for fish. Proud et al. (2019) worked through these assumptions and arrived at a plausible range for global biomass whose upper bound is several times its lower bound. A manager asked to set a sustainable catch from a stock known only to that precision has nothing to work with. The composition of the scattering layers also differs between ocean regions, so a figure obtained in one basin cannot simply be carried to another.

Animals that feed near the surface at night and respire, excrete and die at depth by day transfer carbon downwards faster than sinking particles would carry it. For the north-east Pacific, Davison et al. (2013) estimated that mesopelagic fishes account for a substantial share of the carbon exported from the surface layer. A later synthesis concluded that fishes as a whole contribute appreciably to the ocean's carbon flux but that the estimates are spread widely and rest on few measurements (Saba et al., 2021). The service is thus real in kind and uncertain in size. A fishery that removed a large part of the migrating biomass would alter it by an amount that cannot at present be predicted, and the cost would fall on everyone and not on the fishery.

Mesopelagic fishes link zooplankton to larger animals. They are eaten by tunas, billfishes, squids, seabirds, seals and toothed whales, and the feeding relationships of the deep pelagic are complex and only partly described (Drazen and Sutton, 2017). Where a fishery targets forage species near the base of a food web, the effects travel upwards to predators, some of them already the object of valuable fisheries or of conservation measures. The warning that fisheries tend to move down food webs as larger species are depleted was given a generation ago (Pauly et al., 1998), and a mesopelagic fishery would be a further step in that direction.

Fisheries on the continental slope and seamounts expanded in the late twentieth century ahead of the science. Orange roughy is the best-known case: the species proved to be far longer lived and less productive than assumed when fishing began, and stocks were reduced within a few years (Clark, 2001; Koslow et al., 2000). The lesson is not that mesopelagic fishes share the biology of orange roughy, which they do not. It is that assumptions about productivity made before the data exist have been wrong in the past, and that the error was discovered only after the damage was done.

Many myctophids mature within a year or two and are highly fecund, traits that normally allow a population to sustain a high rate of exploitation (Catul et al., 2011). This is a fair point, and it distinguishes them from slope species. It does not remove the need for surveys. The group contains hundreds of species, life histories have been described for a small minority, and those of cold and deep waters appear to live longer than tropical ones. A trawl in a scattering layer takes a mixture, and the productivity of the mixture is that of its least productive members if they are not to be depleted. Resilience is a hypothesis to be confirmed for each fished assemblage.

The demand for marine ingredients in aquaculture feed is real. Yet a fishery that turns lanternfish into meal and oil for farmed fish is an indirect route to human nutrition, and its contribution has to be weighed against what the same fish provide when left in the water as prey for wild-caught species and as carriers of carbon. Neither side of that comparison can be quantified without the baseline information this article asks for. Advocates of urgent study of the twilight zone have made the same point: the knowledge should be gathered now, before industrial pressure sets the terms (Martin et al., 2020; St. John et al., 2016).

It is sometimes said that only a fishery will pay for the ship time needed to study a remote resource. Commercial vessels can indeed collect acoustic and biological data. But data gathered by a fleet that has already invested in gear and processing plant arrive together with pressure to continue, and catch rates from schooling fish are a notoriously poor guide to abundance. Scientific surveys with public or shared funding, using commercial vessels as platforms where convenient, obtain the same information without creating a constituency for expansion before the results are in.