One of my most recent challenges has been to determine the scales at which ecological processes are relevant. Given that a large component of my analysis depends on features of a site that are not local, that is to say, characteristics that occur at the landscape level, it is important to determine the proper scope. Road density, land use and population are examples of such landscape factors that must be measured within a defined spatial extent. Each of my random study sites encompasses approximately 30 square meters, but to assess landscape features, one must look at a larger area, such as 100 square meters or 5 square kilometers or whatever size best suits the question. Naturally, the choice of these spatial extents will affect the measurements of the landscape variables. To choose too small or too large of an extent would not provide substantial variation. As such an appropriate area must be determined. Ideally, however, the choice of spatial extents will reflect a meaningful ecological process. For instance, in a study monitoring fledgling survivorship of Magpies, it might make sense to choose a zone for measuring road density equivalent to the known dispersal of Magpies from the nest. Of course, values can be calculated for multiple spatial extents and unexpected scales may turn out to be the most profound variables. But either way, my focus of late has been to decide which scales are most likely to be significant by considering my different variables in an ecological context This processes relies on some intuition, both for anticipating what sorts of variables may come into play as well as the relevant spatial extents. Within the context of Japanese stiltgrass invasion, factors such as low-lying land, soil type, road density and White-tailed deer abundance are plausibly significant variables in Japanese stiltgrass dispersal. With these factors in mind, there are some intuitively appealing scales to consider. For instance, the home-range of White-tailed deer, who can disperse Japanese stiltgrass seeds. With this home range delineated, road density or land use can be given a meaningful context for measurement. By framing plausible variables in their ecological context, the analysis becomes much more powerful in that features involved in the relevant processes will be preferentially included in the model above other noise. Choosing random extents irrespective of ecological processes allows a greater chance that stray and erroneous variables might be included. As such a little leg work and forethought prior to the calculation of these values for each variable can be hugely important in determining the best model.
The Mother Matrix
Tomorrow will be the final day of field work, which is very daunting in a way. Last year I was quite relieved to be through the field season and out of the swamps. That was the pilot year, however, and only the beginning of the project. This time there is an element of finality that I find rather unsettling. Without another field season ahead, I must accept that the research I have done is what I will have to work with. My thesis and any publications depend on this data and there are no remaining opportunities to fine tune the methods or expand the sampling protocol. The data as it is will have to stand on its own, and though I feel confident in the work that has been accomplished thus far, I find myself very anxious to begin the preliminary stages of analysis.
Model validation
As the first phase of field work nears completion, my attention has been turning towards the analysis based portion of this summer’s research. From the data collected over last summer and in the first half of this summer, I am now in a position to begin building the models of invasive plant distribution and deer abundance, which are the ultimate goal of my research. The model building process, however, is often a messy affair, and will require far longer than the remainder of the summer. For now, my modelling will be mostly of the rudimentary type, just enough to prepare for the second round of field work, during which data I will collect data to be ultimately used for validating the final model. This is where some quick analysis can be helpful for saving time. For instance, a large piece of my study involves estimating the abundance of deer populations based on counts of pellet groups along transects. The relationship between observing the deer pellets and the true number of pellets, however, is not always simple. Factors such as the observer, the day of the year, the biomass cover or the understory density may all conceivably affect how many pellets one detects. As such, in order to model deer abundance based on deer pellets, other covariates must be known so that the deer population estimate can be adjusted. Obviously measuring these other variables can be quite time consuming and to measure them unnecessarily would truly be a waste of time. Yet not to measure essential covariates would essentially negate any value of the data. Some relatively simple comparison of multi-covariate distance models, however, can help to inform what covariates are important features of the model and so are worthwhile to measure.
Returning to the swamps
So, the long awaited field season is at hand and is perfectly coordinated with the hottest days of the year. Yet I do not believe it could be otherwise, especially if one is conducting field studies in the Williamsburg area. The swamps do conspire to make one sweat. But the weather is really just a nuisance, getting into the field business is when the proverbial “fun” begins. The other day, another Biology student expressed the belief that the field season is synonymous with stress. You can imagine that after planning for all the contingencies only to forget batteries for the GPS handheld and to be unable to locate study points, really does test one’s patience. With so much on the mind, it is no wonder that the little things are forgotten. My mentor seems to enjoy pointing out occurrences of Murpy’s Law, but I do not begrudge him for doing so . Trying to remain amused when faced with heat and humidity and technical bothers is the best defense against a spoiled day. I must say, hopefully without losing all credibility, that these difficulties are often what makes the work rewarding in the end.
Greetings
My name is Chris Tyson and I am conducting research this summer under my adviser Matthias Leu as part of the Applied Conservation and Ecology Research (ACER) lab in the Department of Biology.