This YouTube video, sent in by a reader, shows how a school of fish reacts to hunting behavior of blacktip sharks (Carcharhinus limbatus) off the Maldive Islands.
Notice how the fish seem to move in a coordinated fashion, almost as one, and how they tend to group behind the sharks, where they’re less liable to be nommed. Such “coordinated” group movement is not unusual in flocking animals; we’ve seen it before in the amazing behavior of flocks of starlings (see the videos here). The thing is, biologists don’t really understand what cues animals use when groups of them appear to move as one.
It hasn’t escaped my notice that the sharks seem to be driving the fish toward the wharf, perhaps to either trap them or stun them against the pilings. (Dolphins, by the way, sometimes stun prey by whacking them with their tail.)
I asked my colleague Steve Pruett-Jones to watch it and, as an animal behaviorist, send me his take. Here it is:
Animals form large groups for many reasons, from reproduction to migration to avoidance of predators. Some of the largest groups of vertebrates are seen when birds flock and fish school as an anti-predator defense.
This amazing video illustrates the apparent coordinated movement of individuals in a large school, although in fact the movement of each fish is thought to be independent. How the fish do this remains somewhat of a mystery. Obviously, vision is critical (fish don’t or can’t school after dark, and fish that have been blinded also don’t form schools) but fish also often have prominent markings on their shoulders or tails (schooling marks) which appear to serve as reference marks indicating their movement.
Other possible cues include pheromones, sound, and the sensitivity of a fish’s lateral line. Fish that have had their lateral line removed swim closer together, suggesting that the lateral line keeps fish at a minimum distance from each other; fish appear to be able to ‘feel’ when another fish comes close because the lateral line is sensitive to pressure. In contrast to the fish avoiding the sharks in this video, the movements of the sharks are clearly coordinated as it is in many predators.
By “independent” above, Steve means that the fish are not all responding to a single external cue (which may in fact be what the sharks are doing when they make their “hunting rush” in this video), but to the presence of surrounding fish. This suggests two things: first, that this “coordinated” behavior is really a response to the movement of a single individual, who sets off a wave that propagates through the group. Second, the speed of propagation seems much faster than can be explained by the sum of the reaction times of all the individuals. The fact is that we simply don’t yet understand how this type of group movement works. That seems like a simple question, but it’s a simple question that’s hard to answer.
45 thoughts on “A striking case of predator avoidance in fish”
Aren’t there computer simulations of flocking behavior that illuminate the rules by which this sort of thing can be generated? (I’m pretty sure I’ve seen work on this but have no memory of where.)
I think there are, but this is more impressive because they’re not just gently flying or swimming along — the whole group appears to be responding smoothly and quickly to a dynamically changing stimulus: the movement of the sharks.
The school of fish ‘seems’ to be responding as a single mass but if you look at them as a whole you notice that those fish further away from the sharks don’t seem to be responding to the sharks at all. There also seems to be some sort of ripple effect where the ones to respond quickest are (not surprisingly) those closest to the sharks.
It looks like the apparently coordinated movement is based on a couple of simple rules (or at least could be modeled using a couple of simple rules).
1. If a big shape moves towards you, move in the opposite direction to the big shape.
2. If a small shape moves in your close vicinity, move in the same direction that the small shape moves.
The shape of the predator may have nothing to do with it; if the fishes can respond to a pressure change via their lateral line, they can certainly sense the compression wave as a shark approaches. It’s a bit like feeling the compressed air as a train passes by, except that in water the compression wave travels much faster. I would strap an electric gizmo to a fish and send signals appearing to come from the lateral line to see if the fish swerves left or right and what size/frequency signals cause what behavior.
I doubt pheromones are involved at all; speed of dispersion and direction of dispersion in a liquid both with or without mixing would pose serious problems to the illusion of control in a school. I would go for the pressure sensors acting like a button and causing behavior dependent on how fast or how hard or how frequent you push it.
I forgot to say – I suspect vision comes in just to set the environment for the fish – perhaps they will only respond to the pressure signals when they have the optical signals to say they’re in a school.
This makes me want to put a fish into an aquarium with controlled flow, have wriggling fish-like objects along two sides, and a special pump to produce controlled pulses of water counter to the flow. I’d bet that a fish doesn’t have to see the ‘predator’ at all; it just needs to get visual (and possibly pressure) signals to make it believe it’s in a school and respond to other pressure signals as if there were a predator.
Play around with NetLogo which by now has many dozens of simulations which include flocking. On most and perhaps all you can mess with the variables and change the behavior(s) simulated. Great program at a great price (free).
It doesn’t have to be a single individual. All of the individuals close enough to sense the sharks could be responding, allowing it to spread faster.
But it still seems really smooth. I’d like to see a close-up in slow motion, to see if it’s possible to discern the transmission of the motion…
yup. and it wouldn’t be that hard to set up, either.
In fact, I would be very surprised if someone hasn’t already done that.
It reminds me of playing with a magnet and iron filings….
Great minds, and all that jazz….
That’s what it looks like to me too. In that case you don’t need to assume the fish are responding to what the other fish are doing. There’s a “field” generated by the sharks and the fish are just reacting to the field.
It reminds me of the “correlation hole” in quasiparticles. 😛
Reminds me of iron filings on a surface responding to a magnet waved about from underneath the surface
I suppose it is like Conway’s Game of Life or Boids ~ where complex & ‘purposeful’ large scale patterns emerge from a small set of local rules ~ each fish need only pay attention to events within a body length or two of itself.
GB Jones: there is work on schooling behaviour and the simple rules behind it. Check Charlotte Hemelrijk at the University of Groningen, the Netherlands
sorry, GB J_A_mes
I’ve suffered worse. 😉
I was watching sharks up in the arctic going after seals. It’s facinating to watch how the seals are tight on the sharks tail almost looking like they are the ones in pursuit of prey. Am I right that this is not about smart seals but is simply a behavior that was selected for?
Why can’t smartness be selected for?
it certainly could be, but it’s not necessary to explain the behavior in this case.
How smart do you have to be to know not to be at the pointy end of a shark?
That’s pretty cool. Reminds me of a magnet to iron filings.
You have just demonstrated the flocking effect 🙂
The biologists pondering such phenomena need to learn statistical mechanics. That’s the branch of mathematical physics that converts simple rules governing individual molecules of gas, for example, into the mass behavior of gases.
Statistical mechanical analysis has shown, for example, that putting a stop-go traffic light at the entrance to the Lincoln Tunnel would slow down individual vehicles but increase the overall throughput of the tunnel.
Thankfully, my only exposure to this hairy subject dates back over 40 years, but I vividly remember struggling with the concepts of the micro-canonical ensemble and the grand canonical ensemble (both very poorly explained, as it happened, so one had a constant feeling of wandering in fog), and some extremely tough mathematics well beyond anything I’d learned in a formal course in math.
Jerry being at the University of Chicago, a known intellectual powerhouse, can probably find a few statistical mechanicians lurking in the physics or chemistry departments if he cares to further pursue the matter.
My favorite part is when that heron or whatever it is steps in to grab itself a fish. And meanwhile, just off-camera, there are no doubt many other such scenarios playing themselves out. All of it connected and co-evolved.
What’s the story with the fishes inside the duck egg?
IMHO, groups of humans display similar behavior in response to threats as well, I think. Ever seen a group of people witnessing a violent event? They seem to back up and circle around behind the scarier guy.
Yes, yes, yes – watch footage of rioters confronted by riot police or soldiers; or, here in Japan, where stations are particularly crowded I have often seen someone out of sight of the platform start running for a train that he or she supposes to be coming, to be immediately, smoothly and, I think, almost unconsciously followed by the group he or she is part of.
That heron was just waiting for another rush attack into the shallows. Notice some of the little fish were jumping? That’s what the bird wanted to happen again.
I bet if you studied this some more you’d find the fish that were moving into position were using their lateral lines to take a minimum energy path into vortexes made by the shark’s tails.
OK, but why are the sharks schooling? I believe dolphins will come from various angles to herd prey into a tight mass and make individual rushes, which would seem likely to be more effective.
….maybe they’re just all having fun. The heron doesn’t look too sincere, either.
Um, maybe the sharks are using the “get-behind” reflex to concentrate the prey fish? Schooling makes them (sharks) into a bigger object with more “suction”. Sure would be interesting to see how this situation developed in the preceding hour or so.
it’s not just sharks that school in response to schooling prey.
tuna do this too.
I’ve seen large groups of tuna use their schooling behavior to actually herd prey fish into tighter and tighter schools.
eventually, hypoxia begins to affect the fish in the tight schools, and they start reacting more slowly.
then, it’s lunchtime for the tuna.
Orly? I didn’t know about this.
I don’t buy the notion that the changes propagate faster than the sum of individual reaction times. There’s no one measure of an animal’s reaction time. Some reactions are much quicker than others.
Consider a human pressing a button when seeing a spot on a screen, versus a human removing a hand from a hot object. One measure of reaction time will be much faster than the other, and I don’t need to specify which.
Schooling behavior in fish is no doubt quite refined, so that reactions to the movements of neighbors will be much faster than any other measure of that same fish’s reaction time.
And of course it’s entirely true that there’s at least one computer simulation that duplicates bird flocking behavior perfectly by controlling the behavior of individual birds in the simulation, and not the flock.
I find flocking/schooling behavior impressive, but not mysterious.
Nonlinear waves are the best. [/total nerd]
Speaking of crowd behavior, recently a surprisingly simple model made impressive predictions of them by using cognitive rules as opposed to physics of ordinary game simulators. I recommend to watch the simulation!
Notice how interactions propagate “faster than can be explained by the sum of the reaction times”.
The reason seems to me to be the integration of next neighbor interaction over groups of individuals. If you adjust to your neighbor and he is in turn adjusting, the subset of you will adjust faster than an individual.
I did notice that the Heron would not go further than about a foot into the water – even though there were plenty of fish just out of reach.
Do you think it was aware of the sharks, and wouldn’t risk it merely for a couple of small fish?
My money’s on sensitive detection of and reaction to pressure waves. Proving it, though…
that actually wouldn’t be too difficult a hypothesis to test.
I do recall there already having been a few studies looking at disabling lateral lines (how fish detect pressure waves) and effects on schooling behavior.
IIRC, what they found was that schooling still worked in the same general patterns, but that the fish ended up closer together than normal, suggesting that they were using pressure waves between each other to determine how far apart they should stay.
I can imagine that being too close to each other would have limiting effects on turn rates, as well as issues with localized oxygen depletion.
Tight schools do actually measurably deplete local oxygen sources quickly, causing them to tire. Some predators actually use this to their advantage.
btw, there are entire postdoc labs devoted to studying this phenomonon.
Ian Couzin’s lab at Princeton had a position offered there last year for example:
his lab site:
so if you’re interested in the subject, that’s the guy to write to.
As a fisherman with my own boat in San Diego, I’m well aware of bait fish (anchovies usually) forming into more or less of a ball (bait ball) when under attack by predators (game fish to me). I have heard or read somewhere that the outermost fish in the ball work their way toward the center, displacing those already there who then end up on the outside surface. I can’t, however, verify this observationally. I also have seen these bait balls on my fishfinder (sonar) well under the surface of the water.