Note: Meredith wrote this blog post, but is having internet problems in Africa, so I am posting it on her behalf.
Pole sana on the lack of recent field updates – it’s been a busy week or two and I’ve traveled halfway across Africa in the meanwhile! Sad to say, I’ve left Serengeti behind for now. I was able to set up almost all of the replacement cameras I brought down with me and completed three new rounds for my playback experiments. I then took a few days off and spent my birthday traveling around in Ethiopia, soaking in some history and culture (and eating really excellent food!). It’s nice to have a break from constant science every once in a while. I went around what is known as the “Northern Circuit” and visited the four historic cities of Gondar, Lalibella, Aksum, and Bahir Dar. I got to visit island monasteries, rock-hewn churches, the palace of the Queen of Sheba, and even made a trip to the church purported to be where the True Ark of the Covenant is kept! Have to say, the trip made me feel very “Indiana Jones”, right up until the point where I got ill from drinking the water…
After a week in Ethiopia, I flew down to Johannesburg, South Africa, to meet up with Craig and another graduate student we’re working with, Natalia. Natalia is interested in cognition and has been testing the creative problem solving and impulse control of different kinds of carnivores. We’ve spent the last few days at a reserve outside of Pretoria called Dinokeng, run by Kevin “the Lion Whisperer” Richardson. Kevin maintains a park with dozens of semi-captive lions, leopards, and hyenas which Natalia can work with for her intelligence experiments. While Natalia has been busy with her research, I’ve been putting together a rig that will enable me to examine herbivore responses to four predator species: cheetah, wild dog, lion, and hyena. Two of these predators (lion and cheetah) hunt by sneaking up on their prey, whereas the others (wild dog and hyena) rely on endurance to run prey down. I’m looking to see whether prey respond to each species of predator differently, or whether there are consistent differences in anti-predatory response by predator hunting type. I’ll be simulating predator encounters because it would be incredibly difficult to observe a sufficient number of actual encounters in the wild. As soon as I find a good internet connection, I’ll post pictures of just exactly how I plan on doing this — it’s pretty great, and I don’t want to ruin the surprise!
Just this morning, the three of us packed up all of our gear and took a small plane out of Pretoria up to South Africa’s Northern Cape province. We’ll be spending the next three to four weeks up here in the Kalahari conducting our experiments. In addition to looking at anti-predator responses, I’ll be helping to set up a NEW camera trap grid (perhaps Snapshot Serengeti will be joined by Kalahari Cameras sometime in the near future…?). Now that we’re back in action, more updates soon!
Champagne corks will be popping tonight. Snapshot Serengeti’s first peer reviewed scientific publication comes out today in Nature’s Scientific Data journal. Please give yourselves a round of applause, because we’d never have been able to do this without you.
The paper is a “data descriptor” instead of a traditional research article, meaning that we describe the detailed methods that led to the Snapshot Serengeti consensus dataset. In addition to describing all the excrutiating details of how we set the cameras in the field, we talk about the design of Snapshot Serengeti, setting retirement rules and aggregation algorithms to combine all of our answers into a single expert-quality dataset. We don’t talk about the cool ecological results just yet (those are still only published in my dissertation), but we do talk about all the cool things we hope the dataset will lead to. The dataset is publicly available here. Anyone can use it — to ask ecological questions about Serengeti species, evaluate better aggregation algorithms for citizen science research, or — we get this a lot — use the images plus consensus data to train and test better computer recognition algorithms.
Feel free to download the dataset and explore the data on your own. We’d love to hear what you find!
Getting to know the mud of Serengeti has been the major hurdle of my wet season. The mud out here comes, I have discovered, in many different treacherous flavors. Surprisingly, you can get a car through the largest puddles by plowing straight through the middle, but the edges of these small lakes (that you might, say, drive on to go *around* the seemingly-impassible body of water) are death-traps. Silty pale sand on the road is often a sign of stickiness and tall grass can hide all manner of trouble. If the sun happens to bake the right kind of crust on the road after a rainfall, solid ground can collapse right out from under your wheels: as I discovered the other evening, under you go. I had been driving out to an area of the park called Barafu when my journey ground to a halt in two feet of thick mud.
Often, with the right amount of tenacity, you call bully your way through the mud. Locking the wheels, jacking up the car, throwing spare tires and dead trees and rocks and whatever else you can find under your floundering tires sometimes works. If you’re in the right areas at the right time of day, passing tour cars often go out of their way to help you out (a favor I try and return at every possible convenience, building up karma for the next big stick). This particular situation, however, was neither the right place nor the right time. It was early evening, and I was miles and miles away from Serengeti Central. Asking for help this late and this far from our field crew would have been irresponsible and, fortunately, unnecessary — I had been heading out camping that evening to start with, and always keep the car stocked with emergency supplies (sleeping bag, snacks, and, of course, a bottle of Safari Lager).
While it was still light, I tried to make it the kilometer or two to a nearby kopje, hoping that somewhere along the route my cellphone reception would kick in and I could at least send out some coordinates for a tow the next morning. No dice: the sun quickly began to set and I had to beat a hasty retreat to the car. Making the best of an inconvenient situation, I popped the top off of a beer and crawled up onto the hood to watch as the sun drift below the horizon. I almost spilled half that beer down my front when, with a roar, a lion emerged from the grass less than two meters away from where I was sitting.
She was a lone female, a bit scrawny and mangy (I was pretty sure I recognized her from earlier on in the day). Stepping onto the road, she sauntered over to the car, roaring every few steps. Was she calling for the rest of her pride? I couldn’t see or hear any others, and the lioness circled the Land Rover only once before continuing down the road. Clutching my beer, I experienced an emotion I have felt only once before, diving with great white sharks: knowing that you’re safe (probably), in your cage or in your car, but also understanding that you’re out in a predator’s natural habitat – and that those predators know you’re there. I can only think to describe it as an overwhelming sense of respect tinged with adrenaline, followed by an aftertaste of awe and thankfulness when the predator finally passes you by.
Needless to say, as soon as the lioness was a fair distance away, I scooted quickly back inside the car and may have even rolled the windows up a bit. I could hear her roaring for the next hour or so, and based on the footprints in the mud surrounding my Land Rover the next morning, she came back once or twice to check up on me.
On a lighter note, I’d like to make a special shout-out to the six-tour car cavalcade of young university men that took time out of their safari the next morning to drag me out and feed me chocolate bars (perhaps have to get stuck more often…?)
First field update! I’ve been out in Serengeti Park for just over a week now, and I’m fairly surprised to report that things are going rather swimmingly. It was certainly my smoothest travel experience from the USA to date: no plane delays (unlike last time), no missing luggage (unlike last time), no egregiously extended stay in Arusha waiting for permits (unlike last time). To be sure, field life takes a bit of getting used to again. We’ve had spitting cobras in the bathroom, ververt monkeys breaking into my car, and little black flies are out in full force. But the Serengeti this time of year is completely worth it. Last time, my field season only encompassed the dry season, but in the current rainy period, Serengeti is a completely different place. Everything is so green it almost hurts your eyes to look at it. Up in Barafu, along the eastern edge of our camera trap grid, are herds of wildebeest, zebra, and buffalo so large you can hardly believe it. (My first time driving through a herd of buffalo several hundred strong required more courage than I’d care to admit – buffalo are big and mean and certainly warrant a healthy respect. They’ve been known to ram our field vehicles before and cause all sorts of trouble).
There’s a few projects I’m working on this year with the camera traps: first and foremost finishing up the playback experiments I began last season. Every morning for ten consecutive days, I’d head out to particular camera traps and play lion roars, simulating the short-term presence of predators in an area. We see from the camera traps that herbivores start to evacuate from these scary areas (or, “areas of artificially elevated predation risk”, to use a more scientific jargon) for not just days or hours, but periods of up to three weeks! I’m interested generally in the trade-offs that herbivores have to make between avoiding areas where predators are and still obtaining enough resources to get by. Do they only avoid areas where there’s a high chance predators will be, like lion territories, despite all the tasty forage that may be contained inside? Or is the avoidance being exhibited on a finer scale – like days to weeks, rather than months to years, like we’re seeing in this experiment? Perhaps there are some species of ungulates that don’t try and avoid predators on a spatial scale at all, but rather rely heavily on evasive and defensive behaviors when they encounter a hungry carnivore. Hopefully these continued experiments and the Snapshot data in general will help elucidate answers to some of these questions!
I’m also working on another round of habitat characterization – this time, we’re interested in the soils and vegetation that help to determine the forage quality at a particular site. Now, to tell the truth, I wasn’t originally that enthused about these particular collection tasks, but I’ve discovered that there’s incredibly satisfying about grubbing around in the mud scraping out soil samples. My inner 8-year old is feeling more rejuvenated by the day. Lion House is started to become more than a little cluttered with sample bags of dirt and grass clippings – pole sana, other field assistants, it’s for Science!
Meredith: Our brilliant team of Snapshot Serengeti undergraduate volunteers at the University of Minnesota are perhaps even more on top of the lion literature than I am! This week, we have a guest post from one such student, Clayton Mazur, describing some recent work of Dr. Packer’s on lion disease spread in Serengeti Park. This post is a synopsis of a scientific paper that can be accessed in full here.
I propose we play a word-association game. I will offer a word and you think of what comes to mind. “Africa.” Did you imagine Mt. Kilimanjaro, or the towering, lush rain-forests of the Congo? “Wildlife.” Did you envision the sprawling savannas of Tanzania, home to hundreds of thousands of migrating wildebeest? If so, I would bet that your savanna also included the enigma of Africa: the African lion. Was he a graceful figure standing upon Pride Rock looking out over his kingdom? Perhaps he was laying in the shade, his dark mane flowing in the breeze as he waits for the females to return with a kill. I would argue that elegant images such as these are what come to mind for the majority the public. The portrayals of lions in the media- from Lion King to the MGM Lion- support this notion. As elegant as these images are, reality is less than elegant for the lions living in Serengeti National Park in Tanzania, Africa.
African lions have unique social structures that help them brave the tough conditions of the savanna. Lions live in families called prides; one to two male lions rule a pride. The roles of female and male lions within a single pride are vastly different. While females hunt, raise the cubs, and reproduce, male lions defend the pride from attack by other predators such as hyenas. Living in a large family group offers lions protection, but it also comes with costs. One cost is that females need to supply a large number of individuals with enough food for survival. Female lions coordinate hunts whereby they stalk prey and then give chase, but this technique only yields about 26% success. The low success rate forces large prides to split or starve. Perhaps a more interesting difficulty of living in a large pride is the spread of disease within lion populations. As Dr. Craig Packer has found, disease prevalence is a threat to the current lion population.
A fatal disease that persists in the carnivores of Serengeti National Park is Canine Distemper Virus (CDV). CDV infects a range of carnivores, from dolphins to rodents and even some primates. The viral infection causes encephalitis, pneumonia, anorexia and eventually, death. You may be familiar with CDV if you own a dog, for many owners in the US vaccinate their dogs for CDV. In Serengeti National Park, where local villages cannot afford to vaccinate their dogs, CDV is a conservation concern for African Lions who contract CDV from domestic dogs. To try to remedy the concern, an intense vaccination regime started in 2003. Packer and colleagues attempted to characterize the progression of CDV in both domestic dogs and African Lions. Their goal was to determine if domestic dogs were responsible for the infections observed in African Lions. The team also wanted to determine if the 2003 vaccination program had any effect at reducing CDV in the domestic dogs and/or African Lions.
The scientists worked with blood plasma collected from both domestic dogs (obtained from 1992-2012) and from African lions (obtained from 1984-2012). After collecting the blood samples, the team ran serological tests to detect for the presence of the CDV virus in individual dogs or lions. Using a Bayesian model, the scientists then calculated the probability that an individual lion or dog would contract CDV in one year. The scientists also used sensitivity models to determine the extent at which domestic dogs transmit CDV to lions. From the results of these models, the scientists were able to comment on the fate of the lions with regard to CDV.
The research team drew results by interpretation of the two models. They found that CDV had persisted in the populations of both dogs and lions for more than 25 years. Outbreaks of CDV occurred in 1981, possibly in 1976, and in 1993. Not only do these results suggest a historic presence of the fatal disease in the national park, the dynamics of each outbreak of the disease was unique. The scientists found that the year in which CDV infected the most dogs differed from the year in which CDV infected the most lions. The researchers proposed that this pattern identified domestic dogs as initial vectors for CDV in the park. After a 1994 outbreak in the lion population, the dynamics of the outbreaks become more disjoint. The disjointed dynamics between the domestic dog and lion populations suggest that after 1994, infections cycled through the dog population and the lion population separately. However, the spread of CDV between dogs and lions was not eliminated after 1994.
With domestic dogs established as an initial vector of CDV, the scientists wanted to know transmission rates between domestic dogs and lions. Again, they used a sensitivity model to predict this factor. The scientists found that domestic dogs were ten times more likely to spread CDV to lions than lions were to spread CDV to domestic dogs. With such a high prevalence of CDV in the domestic dog population and the tendency for CDV to spread from dog to lion, the effects of the 2003 vaccination effort were an important factor to analyze for the conservation of the Serengeti National Park lions. The scientists first analyzed the effects of the vaccine on the domestic dog population. Before 2003, there had been very sparse vaccination in villages surrounding Serengeti National Park. As was expected, this vaccination effort did little to curb CDV infection in either lions or domestic dogs. It was not until after 2003 when all villages to the east of Serengeti National Park and all villages within 10 km to the west of Serengeti National Park vaccinated their dogs against CDV did there exist a decrease (~5%) in CDV infections.
With CDV slightly decreased in domestic dogs due to the vaccination effort, was there a similar decrease in CDV infections in the lion populations? Unfortunately, the sample size of lion serum that the scientists could obtain was not enough to comment on the updated magnitude of dog-lion CDV transmission. Overall, the scientists determined that CDV was still able to cycle in the lion population with very little reduction in the prevalence of the disease. However, not all is hopeless for the lion populations of Serengeti National Park. Dr. Packer’s team suggests that direct vaccination of lions may be more effective at preventing the disease. Additionally, the team suggests that advances in serological techniques would allow for increased accuracy when researching episodic diseases, such as CDV. Implementation of safe vaccines coupled with more accurate serological tests could minimize the effects of CDV outbreaks and ensure the health of the Serengeti lions.
As evident from the work of Dr. Packer and colleagues, there are threats to the conservation of the lion populations of Serengeti National Park. Not only do prides run the risk of individuals starving to death, splitting, and human-lion conflict, disease is another risk factor of living in a pride. Yet, these prides, these perfect families, come to mind when the public thinks about lions. Idealistic images of cubs play fighting or suckling from their mother are important for generating interest and compassion for African lions. One can be content with the image of the brave, courageous, elegant male lion standing on Pride Rock overlooking his kingdom, but one must simultaneously recognize the reality of the lion’s plight. Only then will conservation for lions be truly feasible.
Good news: thanks to funding from National Geographic, we’re heading back out to Tanzania with some new camera traps for Snapshot Serengeti!
It’s a bit short notice, but I’ll be heading back out to the field in just under two weeks to dive back into camera maintenance and data collection. I’ve been frantically ordering field equipment and gathering together all the supplies I need in Serengeti, including 50 cameras and what feels like twice my weight in rechargeable batteries. I’ll be adding new cameras back in to the grid to replace those that have been damaged or stolen, in addition to following up on some playback experiments I conducted last summer and continuing to monitor changes in the habitat around each of our camera sites. Some new data that we’ll be picking up this year include examining changes in the soil quality throughout the camera trap set-up and characterizing diversity in the plant communities in the immediate vicinity of our camera traps. Both of these factors contribute to forage quality for our ungulates and affect how appealing a particular site is for different animal species. I might even attempt to collect samples of dung (ah, the glamour of field work) from around our cameras to see whether we’re actually catching in our photos all the animals hanging out in these areas.
After a few months in Tanzania, I’ll be heading down to South Africa to conduct additional experiments in a small private reserve in the Kalahari. Look forward to updates from the field, and wish me luck!
Okay, so by now you’ve heard dozens and dozens of times that you guys produce really good data: your aggregated answers are 97% correct overall (see here and here and here). But we also know that not all images are equally easy. More specifically, not all species are equally easy. It’s a lot easier to identify a giraffe or zebra than it is to decide between an aardwolf and striped hyena.
The plot below shows the different error rates for each species. Note that error comes in two forms. You can have a “false negative” which means you miss a species given that it’s truly there. And then you can have a “false positive,” in which you report a species as being there when it really isn’t. Error is a proportion from 0 to 1.
We calculated this by comparing the consensus data to the gold standard dataset that Margaret collated last year. Note that at the bottom of the chart there are a handful of species that don’t have any values for false negatives. That’s because, for statistical reasons, we could only calculate false negative error rates from completely randomly sampled images, and those species are so rare that they didn’t appear in the gold standard dataset. But for false positives, we could randomly sample images from any consensus classification – so I gathered a bunch of images that had been identified as these rare species and checked them to calculate false positive rates.
Now, if a species has really low rates of false negatives and really low rates of false positives, then it’s one people are really good at identifying all the time. Note that in general, species have pretty low rates of both types of error. Furthermore, species with lower rates of false negatives have higher rates of false positives. There aren’t really any species with high rates of both types of error. Take rhinos, for example: folks often identify a rhino when it’s not actually there, but never miss a rhino if it is there.
Also: we see that rare species are just generally harder to identify correctly than common species. The plot below shows the same false negative and false positive error rates plotted against the total number of pictures for every species. Even though there is some noise, those lines reflect significant trends: in general, the more pictures of an animal, the more often folks get it right!
This makes intuitive sense. It’s really hard to get a good “search image” for something you never see. But also folks are especially excited to see something rare. You can see this if you search the talk pages for “rhino” or “zorilla.” Both of these have high false positive rates, meaning people say it’s a rhino or zorilla when it’s really not. Thus, most of the images that show up tagged as these really rare creatures aren’t.
But that’s okay for the science. Because recall that we can assess how confident we are in an answer based on the evenness score, fraction support, and fraction blanks. Because such critters are so rare, we want to be really sure that those IDs are right — but because those animals are so rare, and because you have high levels of agreement on the vast majority of images, it makes it really easy to review any “uncertain” image that’s been ID’d as a rare species.
Pretty cool, huh?