Risky times and risky places interact to affect prey behavior
Animals, naturally, try to avoid blundering into a predator, but avoiding predation typically comes at a cost. This cost can come in various forms and can add up to cause a reduction in growth, reproduction or survival and can have a big impact on prey populations. A clear behavioural response to risk is an increase in vigilance. Does a vigilance response during risky times depend on the risk of the place the animal is in?
Animals, naturally, try to avoid blundering into a predator, but avoiding predation typically comes at a cost. This cost can come in elevated stress hormones, increase in energy expenditure and a reduction in energy intake. For ungulates and many other species, these costs can add up to cause a reduction in growth, reproduction or survival, and thus can have a pretty big impact on the population. A clear behavioural response to risk is an increase in vigilance. When potential prey, like an ungulate, is vigilant it has its head raised above its shoulders, its eyes and ears are simultaneously focused at the same point and the animal doesn’t feed, nor chew nor ruminate (figure 1). Our lab has been looking into how risk affects different aspects of prey biology (physiology, fecundity, behaviour) and we, like many others, have found evidence for these effects.
Figure 1: Zebras display vigilance towards passing cheetahs (photo by Daan Smit).
The risks that prey face vary in time and in space at both small and large scales, but most research has examined responses to variation in risk in just one way and at just one scale. Surprisingly, we’ve never looked at interactions between the effects of short term temporal variation in risk and long term spatial variation in risk at the same time. Eighteen years ago, Lima & Bednekoff (1999) predicted that long term risk, or ‘background predation risk’, should interact with short term risk to affect the responses of prey. While Lima & Bednekoff focused their attention only on temporal variation in risk, similar logic suggests that the response of animals to immediate short term risk (‘risky times’) should depend on the long term risks associated with the place they are in (‘risky places’). Interactions between short term and long term exposure to risk have been detected in the laboratory 2,3 but we realized this hadn’t been looked at in a natural setting.
While studying the predator-prey guild of Liuwa Plain NP we realized we had a unique opportunity to study this interaction. Not only did we have a good handle on every predator species in our study area (only 1 social group of 6 lions, 22 wild dogs in 2 packs, 17 cheetahs of which 13 were regularly monitored (see figure 2) and 151 hyenas in 4 clans which were intensively monitored).
Figure 2: A cheetah mother with her 4 sub-adult offspring just before they separated.
We knew that the majority of the diet of all four predators consisted of only three prey species (wildebeest - Connochaetes taurinus, zebra - Equus quagga and oribi - Ourebia ourebi). In addition, the area is a flat, open plain, with few vegetation or geographical features that can provide cover and complicate inferences about responses to risk (figure 3 and 4).
Figure 3: Featureless plains of Liuwa Plain National Park. The black dots are wildebeests and their shadow.
Figure 4: A cheetah scans the grassland for prey.
Thus we had the opportunity to quantify long term risk by the intensity of use of an area by the actual actors who are responsible for this risk, the predators, rather than relying on landscape or vegetation characteristics as proxies of this risk.
How could we study this? We would have to quantify the behavioural responses to variation in short term risk and long term risk, and we should quantify the proportion each prey species contributed to the diet of each predator species. We quantified short term risk by searching for predators, and once a predator, or group of predators was found, we located the nearest potential prey and noted for each animals the species, behaviour, sex, age and position within the herd, every 2-10 minutes (depending on the size of the herd) for an average of 7 times per observation. We focused on the three abovementioned prey species and tried to stratify our observations so we would have comparable data for prey within 500m of the nearest predators, prey within 500m to 1km of the nearest predator and prey >1km away from the nearest predator. To not disturb the prey, and accurately assess their behaviour, we observed prey from a vehicle at a distance with the assistance of binoculars or telescope, and did not collect data from prey that directed attention to us.
To quantify long term risk we used locational data both from GPS collars and from VHF tracking; essentially, we looked at how intensively predators use specific areas within their ranges. We did this for each predator species, but also for the distinct functional groups of predators. Large carnivores can be split into coursers (like hyenas and wild dogs) who pursue their prey openly over long distances (figure 5), and stalker-sprinters (like cheetahs and lions,) who stalk their prey to launch a violent sprint.
Figure 5: A wild dog chasing a red lechwe (Kobus leche), a rare prey species in Liuwa Plain NP. This particular chase of 2 lechwes turned out to be extremely long, one was killed after 5 km, the second one was killed 7 km further (straight line distances)!
To quantify what proportion each prey species comprised in the diet of each predator we tracked collared predators for periods of up to 2 weeks, and monitored them closely during each period they were active. This led to 453 kills detected. One of the unique features of Liuwa is that we could do this work on motorbikes (except for the lions, which were monitored from a vehicle). Perhaps surprisingly, these predators are not disturbed by us being on motorbikes (figure 6-8), and we keep a distance that does not disturb the prey during the hunts.
Figure 6: Co-author Jassiel M’Soka observes a pack of wild dogs.
Figure 7: Hyenas, especially the younger ones, are very curious, yet very cautious (as are we!) (photo by Daan Smit)
Figure 8: Wild dogs at sunset. Wild dog are regularly active at night, thus we often monitored them 24/7.
All of this might sound easy in a few paragraphs but collecting the data was a huge task and cost that involved hard work over many years. While the area is flat as a pancake and as open as can be, none of these data come easily. Every year for months, huge areas are flooded. Mostly the flooding is shallow, but it’s hard to spot the deeper areas and several times motorbikes were drowned and needed major repairs. Additionally, the soil type in the area is sand, we all had to learn how the ride motorbikes without bogging down. Most of us took some falls doing that (and have the exhaust burns to prove it)! We had a huge setback when an electrical short caused a brand new Toyota Hilux (figure 9) to burn to the ground.
Figure 9: It still hurts to see this (copy right Mukula Teddy Mulenga).
Many a cold night was spend following hyaenas, wild dogs and lions on their nightly activities. Most people don’t associate Africa with cold, but in Western Zambia, at nearly 1,000m high, some nights in winter it actually drops below freezing!
After all the data was collected, and the thousands of field sheets had been entered into our database by a team of field staff and volunteers (thank you A. Chinga, F. Corry, G. Ellis, D. Hafey, V. Hoffman, T. Mukula, D. Mutanga, D. Smit, J. Tembo, J. Schietzelt, J. Melhuish, C. Dart, E. Whalen, B. Creel) we began to analyse the data. We did not expect to see a huge difference in the responses of prey to immediate threats in different places. Why would animals, in a fairly featureless area, react differently to an immediate threat in different areas? Danger is danger, right?
We were surprised to find that we were wrong, but one of the best feelings in science is to be surprised by the data, because that means you are finding something new. The model which best predicted the amount of attention animals paid to predators included both the long term intensity of use of those areas by the predators, the distance to the predator present (short term risk) and the interaction between these two. More interestingly, the very best model was the model which included the long term intensity of use of the area of the specific predator that posed the immediate threat. That is, if wildebeests encountered lions in an area often used by lions, they reacted much more strongly than if they encountered lions in an area relatively infrequently used by lions.
The results clearly suggest that the effects of long term risk could potentially mask, or be mistaken for, the effects of short term risk, and vice-versa. Unfortunately, this possibility greatly complicates the interpretation of studies which only focus on either the long term or the short term effects of risk.
Responses varied among predator-prey pairs (which can be expected, different species react differently), but not as expected with the strength of response being positively correlated to direct predation rates as found in another study 4. The strength of anti-predator responses did align with the predictions of 5 who predicted that stalking predators, like lions and cheetahs, should have stronger effects on prey behaviour than coursing predators, like wild dogs and hyenas. It is notable, however, that the effect of interactions between predator and prey identity offset this difference (e.g. wildebeest responded very strongly to wild dogs). Neither the immediate presence of a kill nor the long term distribution of kill sites had any detectable effect on vigilance, suggesting that kill locations are a weak proxy for data on the predators themselves when seeking to understand risk effects. A kill could decrease the short-term likelihood of a hunt and may be interpreted as a cue of safety.
It is becoming increasingly clear that predators can have strong effects on prey demography and dynamics through the costs of antipredator behaviour, and that these effects might cascade to affect community structure and function. For a more complete understanding of the strength of these risk effects, we still require studies that measure responses of prey to risk from complete predator guilds (we working on that!), relate the intensity of antipredator responses to the magnitude of direct predation, quantify risk at more than one spatiotemporal scale, and test for interactions in these effects. Several field studies have simultaneously examined the effects on prey behaviour of both short term and long term variation in predation risk, but data from the wild do not yet provide an empirical basis to assess whether the interaction between short term and long term risk that we observed will prove to be a general pattern, as theory suggests it should be. If so, then studies capable of describing this interaction will be critical for a proper understanding of risk effects.
We’d like to thank the Barotse Royal Establishment, the Department of National Parks and Wildlife in Zambia, African Parks for their permission and collaboration to conduct the fieldwork for this study. Funding was provided by WWF-Netherlands, African Parks Network, National Science Foundation Animal Behaviour Program (IOS- 1145749), and Painted Dog Conservation Inc..
The paper in Nature Ecology & Evolution is here: http://go.nature.com/2rvOIWX
1. Lima, S. L. & Bednekoff, P. a. Temporal Variation in Danger Drives Antipredator Behavior : The Predation Risk Allocation Hypothesis. Am. Nat. 153, 649–659 (1999).
2. Brown, G. E., Ferrari, M. C. O., Elvidge, C. K., Ramnarine, I. & Chivers, D. P. Phenotypically plastic neophobia: a response to variable predation risk. Proc. R. Soc. B 280, 20122712 (2013).
3. Joyce, B. J., Demers, E. E., Ferrari, M. C. O., Chivers, D. P. & Brown, G. E. Background Predation Risk and Learned Predator Recognition in Convict Cichlids: Does Risk Allocation Constrain Learning? Ethology 122, 841–849 (2016).
4. Relyea, R. A. The Relationship between Predation Risk and Antipredator Responses in Larval Anurans Author ( s ): Rick A . Relyea Published by : Ecological Society of America Stable URL : http://www.jstor.org/stable/2679878. Ecology 82, 541–554 (2001).
5. Preisser, E. L., Orrock, J. L. & Schmitz, O. J. Predator Hunting Mode and Habitat Domain Alter Nonconsumptive Effects in Predator–Prey Interactions. Ecology 88, 2744–2751 (2007).