Thinking & Learning

When I settled on a title for this article I was thinking in terms of how fish (and trout in particular) think and learn, but it could equally be applied to those of us who seek to catch them. I have a scientific background, with a PhD in fish biology, but I’ve learnt far more through simply observing trout at close quarters over many years of fly fishing than from reading abstracts and studying graphs and statistics in scientific publications. Bob Wyatt aptly uses the term ‘voodoo’ for much of the pseudo-science promoted by anglers, but I suggest there is a fair bit of voodoo in the scientific literature too, whereupon any experienced angler would question the experimental design of some published papers, particularly those relating to hook avoidance and response to angling pressure.

I don’t intend to detail the scientific basis for the facts or assumptions I’m presenting here, or to reference the many sources. If you want to know more you will have to consult Mr Google. Mr Grubb (Thomas C. Grubb Jr. 2003: The Mind of the Trout —A Cognitive Ecology for Biologists and Anglers) is another great source of information. Instead, I want to simply state some of the key findings I’ve gleaned from reviewing recent literature and to comment on how they might influence trout to behave and respond in practical fishing situations.

Trout divide their attention between feeding and avoiding predation.

I had to start somewhere and this is a basic assumption that underpins the ‘intelligence’ of trout. They don’t wake up in the morning thinking ‘what will I do today?’ — they are attuned to respond in particular ways to specific stimuli in their environment. Evolutionary theory suggests that the principal drivers relate to survival and reproduction.

Trout feed in ways that maximise energy intake whilst minimising feeding effort.

Trout don’t know the specific energy value of an item they consume, or calculate the energy cost before swimming out to get it, but they are supremely adapted to maximise feeding efficiency whilst avoiding exposure to predators. These present constant trade-offs in relation to where, when and how trout feed.

Trout do respond to hunger state and reduce food intake as they become sated.

As anglers we know that intensive feeding bouts are often short-lived, even when food remains abundant. It’s also common knowledge that trout are more susceptible during the onset of a hatch, though this may also be related to aspects of selective feeding behaviour.

It is also important to realise that trout digest food more slowly at lower temperatures and may reduce activity and move to cooler water when not feeding or when food is limited.

Searching for several items at the same time slows the ability to detect any item.

We know that trout diet often deviates from apparent availability within the environment, and also that diet varies between individual trout in a population. Some degree of specialisation is evident, and we have all found trout feeding ‘selectively’ on one particular prey while seemingly ignoring other available food, even much bigger energy-rich items. Ignoring a grasshopper to eat a tiny caenid mayfly makes no sense in terms of maximising energy intake — or does it?

Trout form specific search images to help identify and select abundant prey.

Experimental evidence confirms that this strategy is advantageous, in terms of increasing detection rate of a particular, abundant food. At such times of selective feeding, even normally edible items that don’t conform to the search image are perhaps simply categorising as non-food, rather than identified as potential food and ignored. The fish becomes ‘blind’ to everything except some key feature, or features, of the selected prey.

On these occasions we have to accept that we need a fly that closely matches a particular prey in size and shape and general appearance, or it will not be eaten. And what would fly-fishing be without such imperatives?

Foraging trout may favour prey based on some standout feature.

Do the tiny blue wings of caenid mayflies or the bright red bellies of Tasmanian jassids stand out so strongly that trout select them purely on the basis of being easy targets? Theory suggests that such items might be so detectable that trout are compelled to eat them, even when other food is more abundant.

Occasionally we hit on a fly pattern that trout seem to take more readily than available naturals, even in a selective feeding situation. We can explain this in terms of some key feature in the fly jelling with the trout’s search image, or just making the fly stand out so it is readily identified as food. It may be a red tag, some gold flash, or a blue dun hackle — something that immediately attracts the trout’s attention. Grubb in his book refers to this as the ‘pop-out’ effect, but you might know it best as a trigger, or a supernormal releaser.

trout senses

• Trout are intimately tuned and adapted to their environment and are equipped with sensory capabilities very different from our own.

• They can detect chemical odours at incredibly low concentrations, and respond to very small changes in water temperature. They are sensitive to magnetic fields, and can detect relatively low frequency sounds.

• Trout have good colour vision, strongest in the red, green, blue and ultraviolet (invisible to us) wavelengths. They can still see and hunt effectively in very dim light, but can’t change pupil diameter in response to bright light.

• Trout can detect subtle water movements and disturbances and can position themselves in relation to stationary objects and currents by virtue of sensory pores in their lateral line system.

It makes sense that the fewer standout features included in a search image, the more efficient the rate of capture. This suggests that trout may be keyed into just one specific detail — size, shape, movement, colour, silhouette — when feeding selectively. If your fly matches that detail, or exaggerates it, you are well on the way to being the local hero.

Prey size might be an important feature in forming a search image.

More importantly, in terms of visual perception, the ability to differentiate between two stimuli (the size of two items of prey in this case) is said to be a constant fraction. According to Grubb, this is a general principle of perception known as Weber’s Law. This was new to me. The implication is that if a fish exhibits a threshold of say 25% in distinguishing between two large baitfish (i.e. a 60 mm prey compared to an 80 mm one), it might just as effectively be able to distinguish between a 6 mm nymph and an 8 mm one (25%). Put another way, trout will notice a smaller size difference among smaller prey.

We do know that size is important in flies. Often a size change up or down will mean the difference between takes and refusals. But at the small end of the scale I’ve always been sceptical, even though the Kiwis down the Mataura way will tell you how critical it is to match the local mayfly hatch down to size 16 and 18. This law says they are right, and we should be more concerned about matching the size of extremely small trout foods than about matching larger ones.

Trout exposed to drifting litter get better at avoiding it.

This relates primarily to autumn leaf fall in deciduous forests, and involves a simple form of learning, habituation, as a result of repeated exposure to non-food. It makes sense that trout must learn what not to eat in conjunction with what to eat. This dovetails with my pet theory that in areas on lakes where surface scum and litter accumulates on a regular basis, the resident trout are more likely to refuse your flies. Put simply, they have to be better at picking living food from rubbish, and so your fly will have to tick a lot more boxes to be recognised as food (which we all know it isn’t). Grubb suggests the possibility that trout might rely on a search image for non-food at such times. If they do, I fear most of my dry flies might fit their non-food criteria quite nicely.

Food detection distances are reduced for trout feeding in faster water.

Trout will move into faster riffles because more food is drifting past, but the faster current and increased visual interference in broken water means they will only detect and respond to food at close range. Our parallel angling observations confirm that the faster the water the closer you need to drift your fly to the depth and position of the feeding fish to get a response. The jury is out as to whether flies drifting further away might be detected and ignored, or not detected, but the result is the same.

Trout tend to move from faster to slower stream sections as light diminishes.

This is a function of the previous finding. As the light fades, those trout feeding in faster water can no longer do so effectively, so they move to quieter water where they can still hunt prey. If you’ve fished late evenings or at night on rivers you will know that particular pools and slower runs are the places to be, with prime fish appearing like clockwork. Any showy fly that stands out is likely to attract attention.

This gets even more interesting when we turn to lakes. We know that dawn and dusk are the best times to find trout in the shallows, though our explanation might be reduced angling pressure or reduced exposure to predation at night. But now my scientific brain is telling me that trout are seeking shallow water, and surface water, to take advantage of the remaining light to continue feeding.

On my home lake I have always known that rainbows leave the deep weedbeds and exhibit a surface cruising mode at dusk, whether there is a hatch or not, and they are invariably vulnerable to a conspicuous attractor fly at this time. Likewise I know that browns often begin to mop up beetles on the surface when it is almost too dark for me to see. Now I think of these as pure responses to light limiting feeding in deeper (and faster in the case of a stream) water.

This perhaps extends to feeding responses in dirtier water and on heavily overcast days as well.

Trout in experimental tanks may take several days to be trained to accept an unfamiliar food item.

Trout don’t immediately recognise edible items, outside their recent experience, as food. “There was a good hatch but not one trout rising” — how often do we hear such things? We have to accept that trout keyed in to a particular diet or feeding mode will not instantly respond to a sudden abundance of a novel item. A lag time can be expected before trout will switch from one abundant food source to a new one. An underlying lesson is that trout are not always feeding on the item that first seems obvious to us (beware the masking hatch).

Trout are capable of learning and remembering simple behavioural responses.

Think in terms of learning to associate food with a delivery pipe of a specific colour, or bumping a bar for a food reward. Often the value of such experiments has been limited by using groups rather than individual fish, where social learning might mask individual capacity, but at least we know fish are capable of such things.

So how good is a trout’s memory? Well, it depends. Something as important as the odour of the natal stream might be imprinted for life, whereas a simple food reward experiment may take a week to establish but be forgotten within days. If you were to sit in the middle of experimental data you might conclude that trout memory of a learned response is likely to fade within weeks if not days.

Without much evidence, it is tempting to suggest that a memory related to survivorship (cormorant attack, angling encounter) might be retained for longer than an artificial food reward experiment. That said, even I’ve been guilty of throwing gravel and swishing my rod tip to mimic hatchery pellet delivery in order to attract stockies and Atlantic salmon recently escaped from sea pens. The salmon still respond, desperate for food, weeks after they’ve escaped.

A trout’s behaviour can be influenced by the behaviour of others.

As suggested above, trout do interact in a social sense and are known to establish territories and hierarchies. Evidence for social learning is limited, but as anglers we all see likely examples. One salmon jumps and they all start jumping. One trout begins rising and others start doing the same. Likewise we have often spooked trout and seen them race off, spooking other fish that haven’t seen us.

When fishing hatches on lakes there are times when only the small (subordinate) fish seem willing, whereas on other days those prime specimens we all seek dominate the catch — “The good fish were up today!”

Trout exhibit a startle response when confronted by unfamiliar objects and movements.

This is an innate response to avoid predation: an automatic reaction to a potential, if unknown, threat. More interesting is what comes next, with salmonids generally reacting by aggregating more tightly in a school (juveniles), by reducing movement, and/or by immediately seeking shelter (deeper water or cover). These are things fly fishers witness all the time, and that we perhaps understand better than researchers in lab coats. Nigel Birt’s excellent article ‘Spooked’ (FL#33) and more recent contributions from Simon Chu and Scott Murray come to mind.

If a trout that you are stalking ‘goes stiff’, or reduces its movements to some degree, experienced anglers will tell you that it might still be catchable if you are prepared to be patient, change flies if necessary, and concentrate on exact presentation.

The literature supports this in demonstrating that salmonids may respond to a predator alert by reduced feeding (in terms of overall consumption and reaction distance). Tempt the fish with that perfect drift at the perfect depth and it might still eat your fly. And they will of course resume feeding normally if you are prepared to wait, perhaps an hour or two.

Through habituation trout may also learn, for example, to ignore movement of vegetation in the wind, or perhaps to ignore farm animals that come regularly to drink, or even to tolerate passing bushwalkers and anglers if they are constantly disturbed.

Salmonids are alarmed by traces of crushed fish skin/flesh in the water.

This likewise initiates a predator alert response that will see trout scurrying for cover. Some experienced backcountry guides in New Zealand believe that catching and handling fish (and even wading the stream), can affect the catchability of fish further downstream.

The evidence suggests that a bleeding or damaged fish might panic other nearby fish, and it reinforces the need for minimal and careful handling of trout when releasing them. It also favours upstream fishing in backcountry rivers, and may be one of several reasons why fishing behind another angling party is less than desirable.

Trout have some spatial sense or cognitive awareness of their surroundings.

This is evident in the way a trout bolts for the cover of a bank or log-jam the instant it is hooked. It doesn’t know it’s been hooked and remember how to throw the hook, but it does know its way around the available cover and escape routes.

This may also have consequences when approaching the water, whereby even if you remain crouched and motionless, a cruising fish might be alarmed by your presence on a normally bare bank. In FL#72, Scott Murray differentiates between looking like a bush and looking like a rock to blend in to the streamside environment.

A trout will stop feeding after being caught and released.

I’ve stated the obvious, but the extent to which trout are affected or ‘learn’ from being caught and released is contentious. Hunger state most likely determines how quickly a trout comes out of hiding and resumes normal feeding after a predator attack, assuming the trout has not been unduly injured or stressed. But how soon will they resume feeding? Perhaps hours, or even days later?

New Zealand research points to a day or more for a released backcountry brown trout to come out of hiding, and suggests that on remote rivers merely spooking or disturbing trout might have a similar effect.

I’m sure, like me, many readers will offer examples of individual trout they’ve caught on several consecutive visits to a stream, or trout being hooked, dropped and hooked again soon after — even flies broken off in fish and recovered only hours later, or the next day. The very fact that trout in some fisheries are caught and released frequently (the Behnke interview in FL#70 refers to Yellowstone cutthroat trout being caught on average 9.7 times in a season) suggests limited cognition on the part of the fish.

Declining catch rates have been linked to learned hook avoidance.

Now the ground really is starting to shake. The problem here is that field experiments evaluating catch and release in the wild have, along with inevitable angling variables and biases, involved excessive handling of fish. (In a recent Canadian lake study, rod-caught trout were held in bins, measured and tagged, then relocated to net pens and held overnight before release; and in a key New Zealand study captured fish were weighed, measured and tagged before release.) Excessive handling, translocation and any injury or skin damage (itself capable of alerting other, unhooked fish in the system) must be expected to extend recovery times and ultimately affect subsequent catchability.

We really don’t know the full implications of catch and release, but it seems reasonable to assume that a trout might hide and stop feeding for a day or more, and after one or several captures might avoid a particular vicinity (by feeding closer to cover for example) or avoid associated risky behaviour (surface feeding for example) for some period of time.

But it is the suggestion that trout somehow identify, remember and subsequently recognise and avoid a particular fly pattern that sounds a lot like voodoo to me. This would surely require frequent hooking with the same fly over a period of several days or weeks. But, more fundamentally, if trout have the capacity to recognise and remember a specific fly pattern, why would they eat it in the first place? Surely they would know it wasn’t the real thing?

I reckon my beetle pattern is pretty good — good enough to fool trout when feeding heavily on live beetles. So, if the negative experience of being caught and released was to lead to a trout avoiding my fly next time, wouldn’t it follow that it would have to stop feeding on beetles altogether?

I remain suspicious when it comes to talk of learned fly avoidance, particularly on lakes, but like any good scientific angler, I’m always willing to look, listen and to learn.