It’s about to get real up in here. I’m about to break it down for you. I’m about to throw some knowledge your way.
*Checks Facebook one more time in an instinctive attempt to procrastinate. Forces self away.*
Let’s do it!
One of These Things Is Not Like the Other…
How many are there? That is elementary, there, dear reader. One of the most basic questions to be asked by any wildlife researcher. Estimating abundance and density can help conservationists and wildlife managers with determining trends and managing populations.
Slow down! What’s the difference between abundance and density?
- Abundance is the number of animals(i.e., N = C/p), where C is the number of animals counted during the survey and p is detection probability.
- Density is the number of animals in a specific, bounded area (i.e., D = N/A), where A is the size of the area surveyed. Density (and not abundance) can be used to compare between areas.
Capture-mark-recapture (CMR) models are commonly used to estimate population abundance for closed and open populations (demographically and geographically). CMR models are models that use captures and subsequent recaptures of uniquely marked individuals to estimate detection probability and from there estimate abundance. K. Ullas Karanth pioneered the use of camera trap data to “capture” and “recapture” tigers, recognizing individuals by their unique stripes, and in that way estimated tiger abundance. Since then, researchers have estimated density for a multitude of marked animals, such as tigers, leopards, ocelots and fanaloka.
But what about fossa?
Objective 1A: Who Are You?
What? Density estimation tends to require that you are able to individually identify animals. Fossa tend to be unmarked (except for the rare numbered fossa, which we have yet to capture). Researchers have attempted to individually identify naturally unmarked species, like pumas, based on distinguishing physical characteristics (i.e., kinked tails or scars). Fossa have been individually identified before, but using baited cameras, which we don’t believe in because it’s cheating. Given that we don’t have chickens tied to our cameras, how will we individually identify the fossa we get pictures of?
How? With a little help from my friends (read: unpaid undergraduates)! I have recruited two undergraduates and we have intensively trained in the fine, ancient arts of determining what is a good mark (something permanent like a bobbed tail), staring at a fossa picture until you determine whether it is marked or not and then figuring out whether you have seen the fossa before. All three of us have created capture histories for the past four years of data. Now all that is left is to compare between the three of us. To do this, I will estimate density for all of the surveys based on each capture history. If the density estimates are similar, it can be assumed that in our case, fossa can be individually identified.
Why? This is the subobjective on which all of my other subobjectives rest. It is the foundation of my first goal: to estimate fossa density in Makira. If we cannot individually identify fossa, then we will not be able to estimate fossa density with any precision. And if I can’t estimate fossa density, I might as well go work in a Panera Bread.
Objective 1B: It’s All in the Technique
What? There are multiple ways to skin a cat. They all require sharp tools, patience, skill and an ability to handle blood and physical pain. Just as there are multiple ways to skin a cat, there are multiple ways to estimate density (you also need all of those things I listed above to estimate density). They can be broken into two different types of models: ones that estimate density directly and those that estimate abundance and then require you to estimate the effective trapping area. Furthermore, density estimation can also be broken down into how the parameters are estimated: maximum likelihood estimation and Bayesian inference. Like a particularly nerdy reincarnation of Bloods and Crips, researchers who rep maxlike and researchers that rep Bayes (as they are known on the streets) tend to argue about which one estimates harder. There is only one study to date (in the works!) that compare these techniques and approaches for estimating the density of naturally unmarked populations. So which technique estimates harder?
How? I will use the fossa capture histories created in Objective 1A to compare four types of models. Here is a handy visual.
Mark resight models are models that are similar to CMR models, except that it isn’t necessary to individually identify every single animal you capture. I will use mark resight models because we won’t be able to identify every single fossa we have pictures of. Regular mark resight models estimate abundance and require that you estimate effective trapping area. Spatial mark resight models incorporate spatial location into the modeling framework, estimating density directly. I will then compare fossa density as estimated by each of the techniques to determine which one is more precise. I will also determine the “best” density estimate by comparing how many of the modeling assumptions I violate while estimating density. So, not necessarily the “best” density estimate, as much as the “least wrong” density estimate.
Why? Because someone needs to stop the drive-by arguments that fill the air with words like “priors” and “asymptotic” and riddle ecological conferences with the bodies of those vanquished in heated debate. Because it is only natural to take a question—which is better, Bayesian or maximum likelihood?—and a dataset—fossa captures—and attempt to get at an answer. And, finally, and most importantly, because…because…well, why not?
Look, a fossa!
Now that I’ve got your attention again, let’s continue!
Objective 1C: How Many Lemurs Equal a Fossa?
What? At this point, I should have determined how fossa density at each of the sites (seven) and for each of the surveys (eleven). But, while it’s all well and good to know how many fossa there are during each survey, what about what influences them? And how has fossa density changed over the years as time has gone by?
How? I will examine the relationship between fossa density and multiple factors using linear regression. Factors to be included in analysis:
- Human activity
- Habitat fragmentation/degradation (e.g., percent edge)
- Lemur density
- Fanaloka density
- Presence of exotic species (like cats)
- Presence of small mammals
To see how fossa density has changed over the years at resurveyed sites, like Anjanaharibe (AJB) and Mangabe (MGB), I will see if there are significant differences in past and present density estimates.
Why? It is important to examine how outside factors like the abundance of another species (lemurs, fanaloka) or habitat characteristics (whether the site is intact forest or degraded) affect how many fossa there are. That way, we can recommend how to best manage fossa populations. If we see that there are fewer fossa at during surveys with a lot of human activity as compared to surveys where there are more fossa and less human activity, we can safely assume that it might be humans that are causing less fossa and give suggestions to limit human activity in the forests. By seeing how habitat fragmentation, exotic species and human activity affect fossa density, we can determine if fossa would be a good sentinel species that researchers can use to examine the condition of the greater ecosystem. Finally, by following fossa density through the years, we can potentially spot and halt a downward trend.
Objective 1D: Fossa Betseka
What? According to former lab member, current colleague and analysis whiz, Brian Gerber, 95% of the rainforest fossa population exists in the Masoala-Makira and Zahamena-Mantadia-Vohidrazana complexes. That amounts to over 4,000 fossa.
That’s a lot of freakin’ fossa.
However, this estimate is based on a density estimate from Ranomafana National Park, in southeastern Madagascar. Environmental conditions and human pressures are almost certainly different between northeastern and southeastern rainforests, likely causing differences in density. Fossa density might be higher in Makira; it might be lower. I will attempt to estimate the fossa population in Makira and the Masoala-Makira complex.
How? To estimate the fossa population in Makira and the Masoala-Makira complex will be relatively easy, especially compared to Objective 1B. I will first determine the “least worst” density technique as I described in Objective 1B. I will then take density estimates (as estimated by that technique) from the most recent camera trap survey (either AJB or MGB 2013). I will then find the lowest and highest fossa density estimate and apply it to the respective areas of Makira and the Masoala-Makira complex to obtain a range of fossa abundance.
All of that work. Just for a simple multiplication problem.
Why? Because the Masoala-Makira complex is the largest protected area complex in Madagascar, it can be thought of as the fossa’s last stand. Here is where there should be the most fossa within one continuous population. Important questions are: how many fossa are in this population? And could it be potentially viable? Whether a population is viable or not depends on whether you are thinking in the short-term or the long-term, but recent work suggests that the effective population needs to be greater than 500, and preferably in the thousands. We are not estimating the effective population with these analyses, but by estimating the regular population, we might be able to suggest that Masoala-Makira’s fossa population has the potential to be viable.
Are you still with me? It’s almost over. You’ve only glimpsed the dark analysis abyss in which I will willingly climb and attempt to navigate, all for the sake of a simple multiplication problem at the end of my struggles. Continue you on, if you dare, and prepare to shake before the horror that is the next installment, The Backup Dancers. Cue Thriller laugh.
Having trouble sleeping? Feeling exhausted and fatigued during the day? Is your productivity slipping? Try our all new, all natural, 100% effective sleeping aid. Click here!