We all know the feeling – sometimes there just aren’t enough hours in the day to get all our work done.  Such is the life of agoutis living in low-quality territories, who have to scrounge around the rainforest floor not only for today’s meal, but also to find seeds they can cache underground for the late rainy season when there will be even less food.  To push ourselves as deadlines approach we set the alarm extra early, and have an extra cup of coffee to keep working later.  Our latest discovery shows that hungry agoutis also stretch the hours of their day, but face much more dire consequences than a short-night’s sleep.

Our new paper published today in Animal Behavior, led by Lennart Suselbeek from Wageningen University, shows that hungry agoutis that wake up early or stay up late are much more likely to be eaten by ocelots, while those who have plenty of food sleep in and avoid the dangerous dusk and dawn periods.

activity graph

Ocelot (dashed line) and agouti (solid line) activity patterns over a typical day (time on x axis).

We used radio-collars and camera traps to measure animal activity.  The camera data showed the expected overall patterns of the population in general, with agoutis out in the day (95% of photos) and ocelots working the night-shift (78% of photos).

However, there was some overlap, especially at dusk and dawn.  Comparing the ratio of ocelot:agouti pictures at different times of day puts a number on this risk.  During the day we’d get 500 agouti photos for every one ocelot picture – this is a safe time for agoutis to roam around with a low chance of running into a predator.  At night this flips to 30 ocelots for every 1 agouti – this is the dangerous time when ocelots are on the prowl and agoutis are hiding in their dens (as we showed in another recent paper).

We also radio-tracked agoutis and used our Automated Radio Telemetry System to monitor them 24/7.  The live data stream alerted us when animals died, giving a flat-line on the activity graph, and giving us a chance to run out and try to determine the cause of death. Most of these (17/19) were confirmed as ocelot kills.  What more, we were able to pinpoint the time of day that the kill occurred, confirming that the dark hours of the night, especially dusk and dawn, are deadly for agoutis.

Radio signal of predated agouti

Radio signal of predated agouti

If the in-between periods are so important, why would an agouti get up before the sun was up, or stay up into the evening, when they must know ocelots are still active? This required us to focus in on individual agoutis, not general population-wide surveys.  First, we looked at our radio-collared animals to see when they started and ended activity based on the radio signal. Then, to get more specific, we also put a camera trap on their burrows, to see exactly when they went in and out of their bedroom. They key part of this comparison was that we knew how much food each animal had available in their home range from our tree surveys.  The results were simple but striking.

Both methods showed that agoutis in areas with less food left their burrows earlier and entered their burrows later than agoutis in food-rich areas. Hungry agoutis were much more active at twilight, and were more likely to get killed by an ocelot.

Start and end of activity by agoutis related to the food they had available to them, showing that hungry agoutis are more likely to be active when its dark.

Start and end of activity by agoutis related to the food they had available to them, showing that hungry agoutis are more likely to be active when its dark.

In the end it’s the predator that delivers the final blow, but this shows that hunger is driving them to take risks, and highlights the importance of having a good quality territory with lots of food.

So agoutis are kind of like us, busy mammals with too much to do, and not enough time in the day.  The next time your alarm seems too early, and the sky is still dark outside, consider the chances there is an ocelot outside your bedroom – maybe its ok to hit snooze one more time.

It’s the classic horror scene – a defenseless young victim sleeps as a bloodthirsty predator stalks just outside the bedroom.  This plays out nightly in the rainforest as ocelots stalk past agoutis sleeping in hollow logs or tree holes.  Our camera traps on BCI occasionally record the end result of this drama, for example, this ocelot toying with a baby agouti in the middle of the night.

From previous work we knew that ocelots eat a lot of agoutis, often catching them at night near their burrows.  However, we didn’t know if this was just random encounters, or if the ocelots were seeking out the sleeping rodents.  This question is more than simple curiosity given the importance that refuges are thought to play in how animals move around when they aren’t sleeping too.

The ‘Central Place Forager’ hypothesis suggests that prey should stick close to their refuges so they can quickly run to safety if they detect a predator sneaking up on them.  However, if predators cue in on sleeping sites, prey should avoid these areas when they aren’t actually sleeping.  Two opposite predictions  – so which is it?

To answer this question Willem-Jan Emsens led an effort to radio-track agoutis to find where they slept, then ran camera traps to monitor the agoutis as they come and go.  These cameras also recorded ocelots.  Not only did the ocelots walk by, but our videos show them actively trying to get into the agouti hide-outs.

Our camera traps recorded ocelots at agouti refuges more than 2x as often as at non-refuge sites, and showed that they hung out at agouti holes  5x longer than other sites.   Ocelots apparently could tell if the refuge was occupied or not, as they spent about a minute trying to get at agoutis in holes, but took just a few seconds to figure out that no one was home, and move on.  No agoutis were harmed by ocelots while our cameras were running, but they must have been well terrified as the cats tried to claw their way in.

So the answer is YES – ocelots do target agouti refuges, but agoutis seem safe as long as they stay tucked away out of reach.  Their bed is safe, but their bedroom (the area around the refuge) is risky.  I just hope they don’t have to get up in the middle of the night to go to the bathroom!

Based on our new paper “Prey refuges as predator hotspots: ocelot (Leopardus pardalis) attraction to agouti (Dasyprocta punctata) dens” in Acta Theriologica 2013.

An agouti can bury hundreds or thousands of seeds in the forest. This scatter-hoarding behavior is interesting to many scientists because it raises the question of how these animals can remember the locations of so many seeds. Agoutis may not be the Einstein’s of the animal world, but the ability to remember the locations of all these seeds is really impressive. The problem is, nobody really knows if these agoutis actually remember the location of their seed caches! Our new paper published in the journal Animal Behaviour sheds some light on this interesting question.

As previously discussed in this blog, we placed camera traps next Astrocaryum seeds that had been buried by agoutis to determine if they were eventually dug up by cache owners (i,e, the animal who stored it there in the first place), or cache thieves. However, these cameras also recorded a lot of other animal action in the area. When looking through the photos, I noticed that cache owners often visited their hidden seeds. It looked like the agouti cache owners were purposely re-visiting their buried seeds every few days. But why would they do that?

Could the agoutis be monitoring their seeds to determine if/when they have been stolen, or is there another reason why the agoutis would want to visit their seeds?

To investigate this question I talked to Tim Roth (now a professor at Franklin and Marshall) who is an expert on cognition and seed caching birds. Tim recently wrote a paper on long-term memory and seed caching in black-capped chickadees which gave me some great ideas. chickadee Previous studies of chickadees indicated that these small birds were able remember the locations of buried seeds for about a month, but not much longer. In those experiments, birds were allowed to cache a seed in an enclosure, and then were allowed to return after a given amount of time. After a month, the chickadees were not able to recover seeds at better than random levels. Just like in humans, the spatial memory of the chickadees degraded over time. But in the wild, these birds (as well as agoutis) often need to recover seeds that have been stored longer than a month. So, if memory degrades over time, how can scatter-hoarders remember where their seeds are over long time periods? Well, it turns out that Tim discovered a difference between those experiments and wild animals – free living birds may be able to revisit their cache locations to reinforce their spatial memory. This is important because it provides a potential mechanism for a species with limited brain power to still be able to remember the location of their seeds for longer time spans. When Tim tested this hypothesis, he found that chickadees allowed to revisit cache sites could indeed remember the location of seeds after 6 months!

Agoutis need to save their seeds until the season of lowest food availability, and so they should want to store their Astrocaryum seeds in the ground for at least 3-4 months. It would make sense if agoutis also acted like the chickadees in Tim’s experiment. If agoutis repeatedly revisit their caches to remind themselves of the locations of their caches this should lead to increased long-term memory. Laura

To test this idea using our videos, we counted how many times agoutis visited their caches and then compared this to control cameras that were placed nearby (where no caches were known to be buried). We found that cache owners were almost four times more likely to pass in front of cache cameras than control cameras, and that cache owners visited their caches about once every five days. We also noticed that cache owners acted differently when they passed the cache location, often walking directly above the seed location and sniffing the ground where the seed is. In these cases, we believe that these agoutis are reaffirming that their cached seed is indeed still buried at that location. Sometimes non-cache owners would also investigate the cache locations before stealing the seed. In contrast, cache owners purposely left their caches in the ground to save for later consumption. The behaviors we’ve seen are consistent with the idea that agouti cache owners are revisiting their seeds to reinforce their memory (and acting as a census of caches for stolen seeds). If this is indeed the case, these agoutis may be using this behavioral strategy to lengthen their memory. This is a really cool result, because if true, scatter-hoarding animals that want to increase their long-term memory abilities don’t have to evolve bigger hippocampuses (the part of the brain responsible for memory), they can simply behave differently. We expect that similar behavioral patterns may be commonly used by other scatter-hoarding animals, but nobody has tested it in the wild yet!

By: Ben Hirsch

One of the big discoveries of our project is that agoutis disperse Astrocaryum palm nuts in a complex step-wise manner: agoutis bury a seed, then dig it up and move it to another site over and over again. While collecting data in the field, and following seeds from one place to another over many weeks, we noticed that the movements didn’t seem to be completely random with respect to the surrounding trees. Astrocaryum palm trees are very conspicuous because they are covered in big spines, you have to keep an eye out for them when your walking off trail to avoid getting pricked.  When tracking down radio-tagged seeds, it almost seemed as though the seeds ended up in areas far away from other Astrocaryum trees. If this was true, this could be a really important phenomenon, because of something known as “Janzen-Connell effects”.

Vero tracking seeds

Janzen-Connell effects, first identified by Drs Janzen and Connell (of course), result in seeds falling below their parent tree being attacked by enemies while those that escape their mother’s shadow are more likely to also escape their enemies  (parasites and predators). For this reason, our previous result that step-wise dispersal led to long-distance dispersal was so important. By carrying these seeds away from their parent trees, agoutis are helping these seeds survive, and helping trees to reproduce.

Graph showing Janzen-Connel effects: the further away from a mother tree the seed moves, the greater its chance of survival.

BUT, this phenomenon comes with a big caveat. It is possible that when agoutis take a seed away from its mother tree, the seed might be buried right underneath another Astrocaryum tree, and therefore at high risk for predation or parasitism. This pattern is called contagious dispersal because if agoutis dispersed seeds in this manner, they would simply carry them from one Atrocaryum tree site to another Astrocayum tree site. If certain pests (such as insects and fungus) are more common when plants are found at high densities, seeds should survive best when not surrounded by many trees or seeds of the same species. For this reason, contagious dispersal has been hypothesized to be a major limiting factor in seed survival in some systems. (For those interested in more on the subject of distance and density dependent mortality, John Terborgh recently published a great paper in American Naturalist where he discusses this subject.)

So, if our impression that agoutis carried seeds away from adult trees is confirmed, agoutis could be dispersing seeds in a manner which would be really helpful to these trees. The question then was: how do we test this question?  Luckily, we had the data at our fingertips.  During our study we recorded the exact location of each seed cache, and we also recorded the coordinates of all Astrocaryum trees in the area. We were thus able to calculate the number of adult trees near (within 25m) our seed locations using a GIS and see if the seeds went from high to low density areas.

Path of one seed (brown oval) in relation to adult Astrocaryum trees (orange dots).

We found that the seeds indeed went from high to low density areas. But, could this have happened randomly? If a seed starts out in a high density area, no matter which way it moves, it would end up in a lower density area. To really test this question properly we created a computer model of random dispersal: we had pretend seeds moving in random directions and random distances (based on the observed dispersal kernel), and then compared the random movements to the observed movements actually made by agouti-carried seeds. What we found was that the observed seeds indeed moved to areas with fewer adult Astrocaryum trees than the random movements.

Ultimately, agoutis carried seeds to areas with 36% lower density than their original locations. This should greatly benefit the survival of Astrocaryum seeds.  But the agoutis aren’t going the extra mile just for the seeds sake, as our earlier research discusses, seeds buried in areas with fewer Astrocaryum trees are less likely to be stolen by other agoutis. So whats good for the cacher is also good for the cachee.

We think this new paper just published in Ecology Letters, and our recent PNAS paper show pretty conclusively how important scatter-hoarding rodents are to the survival of Astrocaryum trees.

By Ben Hirsch

Sometimes in science, the answer you end up with is not exactly the question you started with.  The path to discovery is not always predictable.  Researchers have to constantly evaluate what they are finding, and be ready to adjust their course when the data leads down a different path.  This is especially true in tropical ecology, where there is so much basic information yet to be learned.

Such is the case with our new paper published this week in the Proceedings of the National Academy of Sciences (PNAS).

We started tracking the fate of tropical seeds with small radio-transmitters because we thought that the predation of agoutis (the main mover of palm seeds) by ocelots (the main predator of agoutis) would leave a bunch of “orphan seeds” buried in the forest where no other agoutis would discover them.  These orphaned seeds would thus be free to germinate and grow into new palm trees.  It was a cool idea, and would show how predators affect prey, ultimately trickling down through the tropic levels to affect seed survival, forest regeneration, etc…  We had all the hypotheses, sub-hypotheses, and sub-sub-hypotheses worked out.  Now we just had to go into the jungle and prove ourselves right.

illustration of our radio tag set up

A Illustration of a buried seed with our radio tag by Patricia Kernan, NYSM.

We set out to map all the palm trees, radio-collar a bunch of agoutis, have them disperse our special radio-tagged seeds, and then wait for the ocelots to pick them off one-by-one.  Earlier research suggested that only about 1/3 of these rodents survive one year, with most falling to the island’s ocelots. If we did our part we knew we could count on the ocelots to do theirs.

This was actually a huge amount of work, we needed “our agouti” to move “our seed”, and bury it in a little hole for safe-keeping.  Camera traps told us whether one of “our agoutis” moved a particular seed, and more often than not it was an un-marked agouti, or a rat or squirrel. Initially animals just ate most of the seeds, but once they recovered from the recently-ended hungry season, they started storing seeds in scattered underground caches for later, when little fresh fruit will be available.

Agouti at seed experiment

An agouti trying to decide which seed to take next.

Finally our radio-tagged seeds were moving.  Only, and here’s where the change in the-path-to-discovery comes in, the seeds didn’t stop moving.  Once a seed was buried we figured we’d just sit and wait till it was dug up and eaten, sometime in the next few months or year.  Instead, the seeds were quickly dug-up, moved, and buried again, and again, and again.  During our first season of field-work this high rate of movement caught us off guard and the additional work of tracking down these crazy seed movements completely wore down everyone on the project.  Given the super-high rates of seed movement, we realized we needed to look for (actually, listen for radio-signals) moving seeds every single day.  Even daily checks didn’t catch all the movements because we observed some seeds actually move twice in one day.

What the heck was going on?  Why were agoutis moving seeds so often?  Some seeds were going 100’s of meters. Were agoutis shifting home-ranges and taking their seeds with them?  Or, were there thieves amongst us?

For our second field season we decided to switch tactics a bit, and investigate this new research path illuminated by the crazy seed movements. We mounted a major trapping effort to try and capture and mark as many agoutis as we could. By being able to recognize lots of animals in one area, we hoped to determine who was taking the seeds. We hid motion-sensitive cameras next to the buried seeds to see which animal’s dug the seeds up.  Our videos (example above) showed that most (84%) of seeds were being stolen by robber-agoutis.  These unscrupulous rodents weren’t just eating the buried treasure, but often moved it over to the center of their territory, where they could more easily find it during the upcoming hungry-season.  This repeated thievery resulted in seeds moving much further than you would expect from a single agouti.  Slightly more than 1/3 of seeds moved more than 100m, which is typically considered far enough to escape the competition of sibling-seeds that just drop underneath the mom-tree. One seed was cached 36 different times, traveling over 749 m back-and-forth between territories until it was 280 m from its starting point.  We made a movie illustrating this amazing amount of movement (shown below with a fun soundtrack).

Although our test of the predator-mediated seed dispersal hypothesis didn’t go off exactly as planned, our results incidentally disproved it. Even if seeds do become “orphaned” by predated agoutis, we now know that the rates of seed theft are so high that these orphaned seeds still have a good probability of being discovered. While this particular route of influence between predators-prey-trees is probably not important to forest dynamics, our other work  shows how other behavior of these agoutis is heavily influenced by the threat of predation (recent biotropica paper, and another one in the works).

This discovery of robber-rodents helping trees by moving their seeds long distances was made even more interesting by the fact that the dispersal of this particular type of tree has been a tropical enigma since Janzen and Martin published “Neotropical anachronisms: The fruits the gomphotheres ate.” In 1982.  This paper, and dozens since it, suggested that the very largest fruits and seeds found in the Neotropics must have co-evolved to be dispersed by the now-extinct Pleistocene Megafauna.  How these trees have survived the >10,000 years since megafaunal extinction has puzzled tropical ecologists for decades. These results are also important when applied to current mammalian extinctions. If tree species are able to survive due to “disperser substitution” maybe this holds a glimmer of hope for trees that are dispersed by mammals that are currently being hunting to extinction or local extirpation. Alternately, our results also show how important of a role these little agoutis can play in their ecosystems. When poaching gets so bad that they also deplete these smaller-sized mammals, the trees seeds may have no chance to survive.

Our accidental discovery of robbing rodents offers a new potential answer to this mystery, and highlights the potential rewards of following thieves down the dark and mysterious scientific path to discovery.


By Roland Kays

To some, the continuous green canopy of BCI’s rainforests looks the same across the island,

Rainforest Canopy

Rainforest Canopy

even though the forest is made up of 100’s of different tree species.  To an animal trying to make a living off seeds dropped out of these trees, however, there are the good and the bad areas.  The good neighborhoods have lots of food and the bad neighborhoods have little food.  From an agouti’s perspective, this comes down to how many palm trees are around, since palm nuts are their favorite food.

Our tree mapping already showed that there is huge variation in the number of palms in different agouti ‘neighborhoods’ across the island. In this new paper just published in the journal Biotropica, we added radio-tracking data collected both by following animals around in the forest, and by using our Automated Radio Tracking System.  We show that “rich” agoutis living in areas with palm (Astrocaryum) density had much smaller home ranges than their poorer island-mates. The reason behind this pattern is straightforward: if you have a high-quality all-you-can-eat restaurant just around the corner, why would you bother to waste your time and energy and face the risk of getting run over by a truck while going to the exact same restaurant eight blocks farther away? Although there are not too many trucks driving around on the BCI-trails, there are ocelots hunting agoutis, and the more an agouti has to run around looking for food the higher risk it has of running into an ocelot-truck.

But, agoutis live in holes in the ground or in hollow logs, not expensive houses.  These do provide refuge from ocelots, as dramatically shown in the below video.  So, if you are an agouti stuck in a bad neighborhood, why not just dig a few extra holes around your territory to give yourself more places to hide from the ocelot-trucks?  This seems like such a good idea the theory even has an official name ‘multiple-central place foraging’.  Do agoutis ‘multiple-central place forage’ to reduce ocelot predation risk in crappy neighborhoods?

Surprisingly, no, agoutis do not increase their ‘multiple-central place foraging’ in bad neighborhoods.  We tracked them down at night to see where they were sleeping, using our radio-tracking antenna to push through the thick vegetation and find their hide-outs.  Although most animals had more than one hidey-hole, there was no relationship with range size – big territories did not have more refuges.

And so we end with the classic scientific conundrum, answer one question, get a bunch of new ones.  WHY don’t agoutis make more holes in large territories?  Are refuges a limiting resource?  Do they need to import more armadillo construction workers to dig more holes?  Or maybe running away from ocelots isn’t that big of a concern for agoutis? We just don’t know, yet….

by Willem-Jan Emsens and Roland Kays

Agouti RefugeTypes

Agouti RefugeTypes

Predation has strong influences on most animal populations but is almost impossible to observe because it happens unpredictably, and only once in the life of a given potential prey species. This spring with the aid of camera traps, I was lucky enough to record a wide array of predatory behavior on Barro Colorado Island (BCI) in Panama. I was on BCI conducting noninvasive genetics research with ocelots (Leopardus pardalis) ,a mid-sized spotted cat. As part of my research I placed camera traps on ocelot latrines hoping to catch them ‘in the act’ in order to match photo records to genetic fingerprints from scat DNA. One latrine site in particular, in an open area on the end of a peninsula named Harvard, turned out to have a lot more than ocelot toilet behavior going on.

In late February spiny tailed iguanas (Ctenosaura similis) began to nest along the shoreline of Lake Gatun, and I got many photo records of these large reptiles basking in the sun out on Harvard point. I also began to get video records of stalking ocelots crouched low to the ground.  Soon enough, I got amazing footage of a male ocelot dragging off very large iguanas twice during the course of one week. While checking my cameras the day after his second kill I was lucky enough to see the ocelot in person, likely still guarding his meal in the underbrush. Many people mistakenly believe that ocelots are nocturnal, however although they frequently hunt at night, they are opportunists and hunt iguanas during the middle of the day when the cold-blooded reptiles are out basking in the sun. Around this same time another male ocelot killed several iguanas along the shore in front of the BCI labs, to the delight of the on looking scientific residents.

Not long after this other reptiles began visiting the point. My cameras frequently captured a very large crocodile (Crocodylus acutus) coming well up from the water, causing me to keep looking over my shoulder each time I scooped ocelot poop from the latrine. Did she have a nest nearby?

Next, turtles (Trachemys scripta) from the lake began to come ashore to lay their eggs in the dirt near the latrine.  Like clockwork, white-nosed coatis (Nasua narica) appeared to dig up the eggs and eat them. During the preceding weeks coatis visited every day, resulting in thousands of photo records, and leaving the point strewn with shallow holes surrounded by chewed up turtle egg shells. Then one day in late March I found a starling sequence of photos. In one photo a coati is standing directly in front of the camera, and in the next photo taken less than 2 seconds later, a crocodile is flying across the frame mouth wide open to snap! During the next few weeks I also recorded several videos of the crocodile going after coatis.

In the instances I recorded, the coatis pursued by crocodiles narrowly escaped, however not all coatis are so lucky. According to Dr. Matthew Gompper from the University of Missouri who conducted his PhD. Research on coati behavior on BCI in the early 90’s, one radio collared coati from his study was killed by a crocodile. It was witnessed by one of the forest guards, and for a month or so afterwards he could hear the collar beeping forlornly from the lake, presumably still transmitting from the stomach of the croc. Eventually he recovered part of the skull but never found the collar. According to Ben Hirsch, a postdoc on the agouti project who conducted his PhD. work on coatis in Argentina, coatis tend to like riverine ecosystems, and on BCI they may spend a large portion of time on the lake edge. Thus crocodiles might be an important source of mortality on BCI. In addition, the population of large crocodiles has been steadily increasing in the Canal Zone as a result of a prohibition on crocodile hunting put in place when Panama took control of the canal in 1999. Everyone agrees, there are more BIG crocodiles circling BCI now than any time in the last 100 years. This may be increasing the effective isolation of mammal populations on BCI, due to a greatly increased risk of mortality during swims to and from the mainland.
Interestingly however, it may be predation of crocodiles by coatis that has a strong influence on crocodile demography, not the other way around. In late April on a trip to the point to check my cameras I found a number of large, bloody, mostly eaten crocodile eggs surrounding a shallow hole. I had been walking directly over the crocodile’s nest several times per week for the last three months without even knowing it was there! Over the next few days the coatis dug up and ate every last one of the crocodile’s eggs, ruining her chances for reproduction during this nesting season. This time it was the coatis that had the last meal.

Written by Torrey Rodgers