How Sailfish Hunt in Coordinated Groups: The Strange Hydrodynamic Engineering of Cooperative Predation

Sailfish (Istiophorus platypterus) are large pelagic predators in the billfish family, reaching three meters in length and a hundred kilograms in weight, with the distinctive elongated dorsal fin that gives the family its name. The textbook account of sailfish behavior, through about the 2000s, treated them as solitary high-speed predators that attack prey individually using their bills. The actual behavior, as documented by sustained observational work starting in the late 2000s, is substantially more elaborate. Sailfish hunt in coordinated groups of up to seventy individuals, attacking sardine schools using a turn-taking pattern coordinated by visible sail displays. The cooperation is one of the cleanest cases of vertebrate cooperation outside mammals and birds.

The Cancun observations

The sustained research on sailfish group hunting came from a team led by Jens Krause, then at the Leibniz Institute of Freshwater Ecology in Berlin, working with collaborators from Humboldt University and the Max Planck Institute for Ornithology. The team conducted underwater observations off the coast of Cancun, Mexico, where sailfish predictably aggregate to feed on migrating sardine schools.

The behavior they documented, published in a series of papers from 2014 through 2016, was substantially different from the solitary-predator account. Sailfish gathered in groups around sardine schools and attacked them in coordinated patterns. Individual sailfish would take turns striking the school, with the non-attacking fish hanging back in formation. The school was kept tightly clustered (called a "baitball") by the group's coordinated presence, preventing the sardines from dispersing.

The capture-per-attack rate was low — around 25 percent — but the cumulative capture rate across a group hunt was high. A group of forty sailfish working a baitball over an hour could consume hundreds of sardines. The same number of sailfish working independently would have produced a much smaller total capture, both because of lower per-attack success and because the school would have dispersed.

The sail display signal

The most striking detail of the Cancun observations was the signaling role of the sail itself. Sailfish raise and lower their dorsal sails dynamically during hunts. The sail is folded against the back during high-speed swimming (it produces drag when raised) and erected during the slow-cruise approach to the baitball. During attacks, the sail is fully raised, displaying its characteristic blue-black coloration with white spots and pattern.

The Krause team and others hypothesized, with experimental support from Marten 2016 follow-up work, that the sail serves three communicative functions during group hunts. First, raising the sail signals to other sailfish in the group that this individual is preparing to attack. Second, the sail display intimidates the sardines, contributing to baitball tightening. Third, the variable patterning may serve individual recognition, though this is contested.

The turn-taking pattern is the load-bearing observation for the cooperation interpretation. If sailfish attacked the baitball independently, the expected pattern would be many simultaneous attacks producing baitball dispersal. The observed pattern was one attack at a time, with the non-attacking fish maintaining position around the baitball. The implicit signaling mechanism — which sailfish attacks next, and when — must coordinate the turn-taking. The sail display is the most visible candidate signal.

The bill mechanics

The sailfish bill is not used to spear prey in the way the schoolroom image suggests. The actual attack mechanic, documented in high-speed video by the Krause team and others, is a lateral swipe. The sailfish approaches the baitball at slow speed, raises the sail, and then swipes the bill rapidly through a cluster of sardines. The swipe stuns several fish at once. The sailfish then circles back and eats the stunned fish that have not yet recovered.

The swipe rather than spear pattern explains the low per-attack capture rate. A swipe stuns multiple sardines but does not directly impale any of them. The sailfish has to recover the stunned prey before they recover their balance and rejoin the school. The success rate per swipe is around 25 percent because most stunned sardines recover before being eaten.

The bill itself is a stiff cartilaginous projection covered with small denticles (tooth-like scales). The denticles increase the chance that a swipe contacts a sardine and the friction during contact increases stun efficacy. The bill is not particularly long compared to the body, but the surface area and the swipe speed combine to produce an effective area-of-effect attack.

The cooperation puzzle

Vertebrate cooperative hunting is unusual outside of mammals (wolves, lions, orcas, chimpanzees) and birds (some corvids and raptors). Fish cooperative hunting was historically considered rare and limited to simple patterns like group herding by mackerel or tuna. The sailfish observations established a much more elaborate case.

The cooperation has the structural properties biologists look for in cooperative behavior. There is turn-taking, which requires recognition of which individual is attacking. There is signaling, via the sail display, which requires both sending and receiving capacity for the signal. There is per-individual cost (waiting one's turn means not attacking when one could) balanced against per-individual benefit (better total capture rate from coordinated baitball management).

The genetic-relatedness explanation that works for some mammal cooperation does not apply here. Sailfish form aggregations of individuals from different populations, including different cohorts and different reproductive lineages. The cooperation is between non-kin. The expected explanation is reciprocity or mutualism rather than kin selection: each sailfish benefits from the group hunt regardless of relatedness, and defecting individuals (attacking out of turn, breaking up the baitball) would reduce everyone's success including their own.

The remaining open question is the mechanism by which turn-taking is enforced. There is no obvious sanction for defectors, and individuals come and go from the aggregation over hours. The system may be stable because the immediate cost of defection (one attack with reduced capture probability) is higher than the immediate cost of cooperation (waiting one's turn for a moment), with the longer-run group-level dynamics being a consequence rather than a designed feature.

The conservation context

Sailfish are listed as Vulnerable on the IUCN Red List, primarily because of commercial and recreational fishing pressure. The behavioral observations are relevant to conservation in two ways. First, the predictable aggregation sites (like the Cancun area) are concentrations of vulnerability — protecting the aggregations protects a disproportionate share of the population. Second, the cooperative hunting behavior depends on group sizes large enough to support coordinated attacks, which means population declines below some threshold may produce a behavioral collapse that further reduces foraging success.

The applied research on sailfish behavior is in its second decade and is one of the cases where observational biology has direct conservation policy implications. The 2014-2016 Krause team papers were used to advocate for protected status for the Cancun aggregation site, which was partly granted in subsequent Mexican fisheries regulations.

Three observations and a closing thought

The first observation is that cooperative hunting is a larger inventory in the vertebrate world than the canonical mammal-and-bird framing of cognition curricula suggests. Sailfish are the most-elaborated fish example as of 2026, but similar work on other pelagic predators — particularly tuna and some species of trevally — has documented coordinated behaviors that were missed by less sustained observation. The pattern of finding cooperation when researchers look for it recurs across taxa.

The second observation is that the sustained-research-attention pattern recurs again. The basic phenomenon of sailfish aggregating to feed on sardines has been known to fishermen for centuries. The behavioral characterization required underwater video, organized expeditions, and a research team willing to spend years on a non-model species. The gap between knowing-something-exists and characterizing-the-mechanism is filled by sustained attention rather than by single decisive experiments.

The third observation is that the cooperation is more sophisticated than the canonical fish-cognition account anticipated but is not necessarily evidence of mammalian-like cognitive capacities. The turn-taking and signaling can be implemented by relatively simple behavioral algorithms running on fish-scale nervous systems. The observation reframes what fish nervous systems can do without requiring the framing that fish are smarter than previously thought. The behavioral inventory grows; the cognitive interpretation requires separate evidence.

The deeper observation is that the universe of biological behaviors is consistently larger and more elaborate than the canonical model-organism-focused curriculum prepares biologists to expect. Species occupying niches that researchers have not been sustained-attention to are likely to harbor capabilities that no current textbook anticipates. The sailfish work is one of many cases where the difference between the canonical account and the actual behavior was substantial, took specific research effort to characterize, and changed the framing for a category of organism that had been treated as well-understood.

Read more essays and technical writing at anethoth.com — a notebook on databases, distributed systems, biology, and the engineering that holds the world together.