How Do Fish Learn?

Matching one of the stonefly hatches, one of the most important hatches for many trout streams.

Today it might be snowing.  And the days may still be getting shorter.  But, before you know it spring will be here and the forest will come back to life. And, when it does, trout anglers will turn into entomologist and the phrase “what’s hatching” will be heard at tackle shops across the nation
 
“Match the hatch” is a common phrase among trout anglers used to describe the act of matching artificial lures to aquatic insects that are currently hatching from their juvenile into adult stages. When insects are hatching they are abundant, and so the probability of one floating downstream and being seen by a trout is fairly high.
 
But, trout are picky eaters.  They develop a search image for one or two species of insects and become hyper-focused on eating only those species to the exclusion of all other food sources. Search images are helpful because it helps fish quickly tease apart an insect from little pieces of debris or rocks that could potentially look like food as it floats downstream. It’s kind of like putting together a jigsaw puzzle.  There could be a pile of pieces in front of you that go together, but if you’ve developed a search image for edge pieces you’ll look past the other pieces for a long time. 

---

We worked at Fridley Run just north of Harrisonburg, VA. It's a small stream with populations recovering from issue related to low pH.

Search images can be helpful because it quickly allows fish to categorize floating objects as ‘food’ or ‘not food’. But, they can also be problematic because hatches don’t last for very long. Sometimes by the time a fish has developed a good search image the hatch is nearly over. If that happens, the fish has a search image for an insect that is no longer common, and the fish will wait to see that specific species float by while allowing many other insects to pass by without being consumed.
 
More problematic is that search images can take a long time to form. It’s a long trial and error process where the fish has to keep trying to eat a lot of things that look similar to the hatching insect before it hones in on the exact characteristics that make the insect look different than a piece of stick or a leaf.
 
Search images where the first thing that got me interested in studying trout. At the start it seemed a little silly and a waste of time for a trout to need a search image.  But, if you think about it, it makes a lot of sense.  If a trout’s is willing to eat everything and has a search image that is too broad, then it will spend a lot of time chasing down little sticks or, worse yet, eat something potentially toxic.  But, if their search image is too narrow, it will not eat enough to survive. It’s an interesting problem to have.
 
More interesting was that, at the time, no one had studied specifically how trout develop search images.  Yes, at some level it’s a trial and error process.  But, trout live in pools with other trout.  And, they have eyes capable of watching what those other trout do.  So, my advisor had a hunch.  Perhaps they learn search images socially.  That is, they watch other fish test out food of various shapes and sizes and that helps cut down the time it takes for a fish to develop a search image.
 
That was the theory we tested in 2008 in what still remains my favorite study I’ve ever done (a link to the publication that resulted can be found below). We started by going out and electrofishing trout from a small stream in Virginia.  But, we had a bit of a unique problem. To track search image development, we needed to not only monitor fish behavior, but be able to trace the behavior of each individual over time. In short, we needed some sort of external that had unique identification for each fish and could be seen from about 50ft away from the stream bank. 

---

Ribbon tags make it possible to document individual fish behavior during streambank observations of behavior.

The solution was ribbon tags- small pieces of plastic glued to a needed.  After the fish is anesthetized, you thread the needle under the top layer of skin, out through the other side, and then tear the needle off the plastic tag.  The plastic remains under the fish’s skin (and don’t worry, these tags fall out a few months after they are put in). So, now we had about a hundred fish in the stream swimming around with a little extra bling.
 
We then installed feeders in two pools that had a lot of fish.  The idea was to train some fish to develop a search image and then move these trained fish to new pools to see if they helped new, untrained fish develop search images. The feeders were a series of PVC pipes, a small, battery-operated toy motor, and a photocell.  Every 5 minutes the feeder would turn on, spin a little brush in the PVC tube, and out would come some mealworms.
 
The mealworms were their own story.  They came in a can and were intended to be fed to reptiles. And, relative to hatching insects, these mealworms were king size candy bar- high in fat, calories, and exactly what a starving fish wants. But, there was a problem. The worms were fairly moist, and so when they hit the water they would sink. The majority of trout diet is made up of floating insects, and so we needed these mealworms to float. So, out came the frying pan and camp stove, and we fried the mealworms to a crispy golden brown (in what may or may not have been the same skillet we cooked dinner in).
 
So, with fish tagged and feeders in place we sat and watched.  And every time the feeder went off we noted the behavior of every fish in the pool. And, for a long time, their behavior was to do nothing. Feeder goes off, worms float downstream (often right over top the trout), no one eats them. Repeat. For days. And, we did these observations from about 7am-7pm. 

---

Here we have three feeders installed in a pool we were training fish. Here's a good place to note that the entire budget for this project was $500.

But, after a few days things started to change.  It started with one brave fish finally eating the worm, but then spitting it up. There was clearly some hesitation. Then a few days later, the fish took the worm swallowed. After that the fish knew it had saddled up to a buffet and it would sit at the feeder anxiously awaiting the next round. In total, it took about 14 days for fish to develop search images.  Remember that number.

Below is a video of what this whole process looks like.  Look carefully for the tagged fish (his color code is blue-blue-blue) sitting on the far right of the screen.  He is sitting in the current waiting for the feeder (the large blue bag) to release a worm, which happens around the 50-second mark. Continue watching after the tagged fish feeds and you'll see him try to fight with another untagged fish. 

Home sweet home for two summers during my undergraduate studies. Thankfully trout learned their search images quickly, because that tent life is a little overrated after a few weeks.

After these fish were trained, we electrofished out of their home pools and moved them to new pools throughout the study area and installed feeders in these new pools. Here is where we crossed and fingers and hoped. How long would it take a naïve fish to develop a search image for mealworms if they could watch another fish that already had a search image?
 
Hours.
 
Maybe less.  We didn’t immediately start observations, but by the time we did untrained fish were already consuming mealworms. To put it another way.
 
Without social learning it takes 14 days to develop a search image.  With social learning it takes less than a day.
 
Why does that matter?  Trout basically starve during summer.  They need to be able to quickly switch their search images in order to consume the most calories possible, and social learning is a mechanism they have evolved to use to speed the process up as fast as possible.
 
Does this study help in trout conservation?  Probably not.  But, it does showcase how complex the species is socially and intellectually. When asked to give public seminars I often present on this study for several reasons.  First, because it’s so unique and different from typical fish research.  And, second, because it’s entertaining and, without fail, the audience really connects with the story. So, maybe this research does help in trout conservation.  Not with directly improving population health, but in helping from empathetic connections with what is not the most charismatic of animals.
 
And, now trout anglers can blame social learning when they don’t get a bite.


To read more about this study, click here. 

---