Two previous articles (Photo trapping or Which camera trap should we choose?) brought up a really in fashion tool to observe nature, study it and manage biodiversity: photo trapping or ‘Camera Trap’! We alluded its characteristics and functioning. However, these devises are not magical. Knowing how they work, put them in nature and then looking at the pictures is normally not enough! As any other tool, it needs to be correctly used to give real results. It can give nothing or even false results if the collection of images and its analysis is not rigorous enough.
Much more complicated than taking a selfie
Camera Traps can have a multiple usage as soon as we understand how they work. For example, I have been able to see Japanese primatologists put one anywhere in a forest, out of curiosity, to observe, after a week, what had passed by. They obtained some, rare, good surprises. However, they did not get any image of macaques, who crowded around in that area. Who could guess that we needed more to get good pictures that represent the animals or images to analyze later on or even to obtain photos from really close?
Photo trapping requires much more than a technical approach. It needs a precise idea of what we are seeking and a perfect knowing of the context and its use. Understanding of the biology of the targeted species, its environment, its social organization and its availability of resources become vital. To put it simple: we do not put it anywhere, anyhow, but taking into account the area and what we wish to obtain and analyze. Having an understanding about the area where we wish to use it, allows us to target the sectors which have a strong potential or even those which are essential for the animal we are interested in.
For instance, we could place the camera trap in tracks that are often used by the animal or by its preys. Knowing how their preys are distributed, the water sources, plants and having a map of the area are key elements. Sometimes, being aware of the places where the animal and/or its tracks have been observed can also help us to place the traps wisely. Therefore, I doubt that simple photographers could have obtained, by themselves, images of the famous and ghostly snow leopards. A third party must have given them crucial information as it is not so easy to get them just by staying some months there.
This information is particularly important for a scientific or a wildlife manager, for their aim is, above everything, to obtain data to analyze. They can use photo trapping to record animal populations, collect facts or make decisions to protect them better. Identifying human impact in their behaviors, identifying the consequences of a population management action, base this decisions in biological indicators such as density or animal organization are other examples that make essential the use of camera traps. And for every single one of these uses, this people pay special attention to details and its methodology, as well as to the data they are interested in. This is not only to obtain real information, but also to be able to repeat the experiences, compare them and make a compendium of several studies.
For them, the use of camera traps is only starting, and we do not always exploit the potential of this tool, due to a lack of knowledge. The important thing is to use it correctly, avoiding biased information. As it happens often in ecology, we cannot necessarily have all information at the area. For instance, when we are interested in the size of a population, all individuals will not be necessarily observed during the study period. Hence, we settle for a little portion of individuals, which we call it a sample and whose number is quite different of the real size of the population.
In the case of recording a population, the task of a biologist or a biostatistics expert is to obtain a representative sample size and estimate the probable size of the population. In the case of other types of studies, they are also based in samples, but here the skill lies in avoiding particular, biased samples that do not have to do with what really happens to the population. Hence, when the use of camera trapping started to spread in biology, some experts have advised to use the camera traps correctly so that the feeling of progress that these gadgets provide does not stay as a sheer feeling.
As many experts say, many programs would use camera traps based on biased samples out of convenience, as it is easy and cheap to make. And therefore, they also provide biased and false conclusions. Thus, obtaining valid inferences about populations through camera trapping requires the effort of solving two problems that could be a source of false conclusions: finding a good sampling and analyze method. If a sampling protocol helps to minimize the effects of biased sources, an analytical approach allows to get solid conclusions thanks to corrections.
Even if doing math and statistics is less attractive than putting a trap and get the photos, the use of analytic tools to get proper inferences (e.g. assessments or probability of samples) and to evaluate the risk of error have been and are in the process of improvement.
As it has been previously said, scientifics are concerned by the bad professional use of camera traps. Particularly in the case of endangered species management, as they are protected and managed by potentially false information. In order to get reliable conclusions, it is vital that the acquisition of photographic data is rigorous and effected in a well-defined sampling; then that data needs to be correctly analyzed.
For example, jaguars, Phantera once, and tigers, Panthera tigris, are present in their habitats in a low density, and their behavior does not make it easy to obtain images and information. In this case, using a camera trap requires to think about a sampling strategy adapted to the area of interest. Hence, using this tool to study or monitor de state of an animal population requires a time and effort investment even before the fieldwork starts. At least if we wish to obtain good images as well as reasonable estimations with pertinent criteria. So, firstly, we need to determine how to obtain usable images that illustrate a representative sample of the population, through which we can estimate certain criteria as less biased as possible. Secondly, we need to determine how to correctly analyze the information, so that we can compensate eventual bias and obtain pertinent data.
It’s a matter of math and analysis…
The method to obtain images must consider the analysis stage and work accordingly with the one which has been chosen (what is the use of making a good sample if there is no analysis method to draw conclusions?). There are many possible approaches to make it that we can illustrate with the case of registration of species.
Registering the representative number of a species in a given area is not as easy as it seems. Camera trapping can be used to facilitate this task. The way of initially gathering the information and the type of image that we want to obtain has to allow us to use later on one of the existing registering and analysis method. There are many of them, and not all are compatible with the camera trapping. A direct counting, for example, cannot be easily associated with photo traps: this tool is not the best way to count the maximum number of individuals as it is not easy to avoid counting the same individual several times.
Besides, methods which imply using transects are also difficult to implement with a camera trap. Transects cannot be moved and they do not allow to register individuals crossing a straight line (the transect) nor to notice individuals in the distance.
However, as long as we can individually identify the animals, camera traps can be used along with the capture, mark, recapture method (CMR), which allows to register and monitor the demography of the populations. The fact of individually identifying an animal in an image taken with a camera trap is the statistical and non-invasive equivalent of an episode of capturing and marking. These identified individuals can afterwards be photographed another time and be recognized, which allows not only to make registers but also to obtain information about their survival; and hence it allow us to model and analyze the population’s demography, understand how it evolves and make predictions. This combination between photo trapping and CMR to estimate the density of the populations is one of the firsts professional uses of camera trapping.
This method is used to estimate the density of great carnivores that present individual characteristics, such as tigers (Panthera tigris) and leopards (Panthera pardus). This combination is evolving very fast, as spatial factors are being taken into account.
Unfortunately, the majority of wild species are not so easy to individually identify by photos, so this approach can be difficult. It was necessary to look for alternative analytical approaches adapted to “not marked” species. Besides, using similar methods for a maximum of different species arrange the affaires of those who wish to do compared cases of different species or to study the biodiversity on a big scale. This could be more complicated to do when they have been registered in a different way.
When animals are not individually recognized, there are many analysis methods available. Some have proposed to recur to abundance indexes, based in detection rates obtained with camera trapping. However, this approach has been very criticized: it implies that the detection probability is constant in the space and time, and also that it remains the same for different species and individuals, which is rarely the case. Without individual identification, detection rates mix up abundance and detectability. They can be the reflex of the number and the animal behavior at the same time, but also of other factors that have an error of sampling as an origin. Good knowing of the animal (or animals) and of the place can help to reduce or prevent these problems.
The most convincing alternative when we cannot register the animals differently or when we are dealing with an imperfect detection, is the approach of model occupation. Here, more than being focused in abundance, we are interested in the probability of a place being occupied as a way of substitute. And the camera traps can just serve to document the presence of animals in a place. These models possess, like the detection index approach, hidden hypothesis. These previous assumptions have to be taken into account during the data interpretation, and they depend also on the way the camera traps have been used (sampling units, sampling occasions, etc.). Hence, the way of analyzing depends on the way the images have been obtained.
… but also a question of strategy and sampling method
When moving animals are targeted in their natural environment, the camera trapping always comes with an imperfect detection in the spatial scale. Firstly, animals can pass in front of the objective, but they can be undetected trough the relatively small detection zone of the gadget.
On the contrary, others can activate the device but generate unexploitable images (like when the animal is too near and the image only shows a big shot of a single part of its body). Secondly, animals using a larger area than the one that is covered by the cameras can even avoid passing in front of the traps.
The probability of detection and capture can be affected by several factors that are produced at two scales. This includes not only the detection zone of the camera trap, its sensitivity, the way it’s placed (height, distances, angles, etc.), but also habitat characteristics, the presence or absence of a bait or a repulsive element, the ambient temperature, the animals temperature, the timing and the duration of the sampling (not having leaved the trap long enough), the animal’s density (the rarer it is, the fewer chance we have to observe it) and its behavior. This complexity implies taking into account, carefully, the relationship between the detection systems and the underlying ecological processes.
These potential sampling problems are common and each project that involves camera traps needs to face these sampling errors, where the present animals in the area are not detected.
We have to take special attention to the spatial variability of the distribution of the targeted animals, in order to avoid putting the traps in the wrong place. Usually it is impossible to completely cover the area with photo traps. So, we have to resign ourselves to putting them in several particular places, chosen as representative samples that allow to make inferences about the sectors where no traps were placed. In principle, the problem of spatial variability can be treated by defining a target population as a sampling unit (or inference unit) and by spreading the traps following a design based in probability (random sampling for example).
Target population and associated sampling units differ according to the objective (abundance amount? distribution amount?). In short, the method to obtain data and images through camera trapping must be related to our objectives. For example, specialists that use photo traps to obtain density estimations by CMR frequently choose to put their devices in places that maximize the detection probability (routes, tracks…). However, this is not convenient when the aim is to take an interest in the occupation or richness of a species in a place. This method of obtaining data and the logic of placing the traps drive specialists to consider every single detail. Things such as the use of lure or the placement of the cameras can have important consequences to the relevance of the analysis.
Finally, we do not have to forget to consider the detection area of the camera trap before we place it in order to reduce the risk of error and failure. This allows to limit the problem of imperfect detection. And it is the same for all controllable aspects during the placement, such as the height, the distance regarding the target, the anchoring, the extirpation of vegetation and others… We should make sure that the target animal pass by the area where the camera traps have been placed, and that it goes through their detection zones and generate acceptable images, where the animals appears completely without any undesirable shots. Ecological knowledge and knowing the characteristics of the device are really useful.
The use of photo traps in ecology is still something new and its whole potential has not been achieved. But in any case, from now on you are ready to tackle the issue and try this device!