A few weeks ago, a paper on which I am a co-author was accepted for publication in the french ecological journal Life & Environment. In this paper, we evaluate the consequences of recreative harvesting on three populations of sea urchins in the Golfe du Lion, the northwestern part of the Mediterranean sea (at the border between France and Spain).

In a nutshell, we conclude that at the most important anthropogenic pressure, populations of urchins (we focused on two edible species, Arbaxia lixula and Paracentrorus lividus) decrease in abundance, and the mean size of individuals decrease as well. This result bears important ecological consequences. Below are the boxplot of density for P. lividus at three sites, subjected to an increasing gradient of harvesting.

First, a little context about the paper. Two years ago, a small group of researchers from the south of France, backed up by an association of biodiversity watchers, started a program to evaluate biodiversity in the Mediterranean sea, and to track changes that can be introduced by an ever increasing anthropogenic pressure. This program [1] relies heavily on the participation of eco-volunteers, that can help measuring the diversity of different populations through non-invasive, non-destructive methods (a part of our paper is dedicated to the benchmarking of such a method with regard to previous results).

On with the implications of our results. As you may not know, urchins have quite an important part in the coastal ecosystem of shallow areas, in that they feed on different type of algae, and are fed upon by fishes. Additionally, the two different species we studied exploit different algal types, that can act to maintain diversity both for the algae and their consumers [2].

The main point of the paper is that an increase of harvesting at the more exploited site may ultimately lead to a loss of stock. Fishermen tend to pick the larger individuals, which are also the sexually mature one. Clearly, over-exploiting the sites can lead to a loss of reproductive individuals, triggering inbreeding depression [3]. All hope is not lost, though, as populations can be maintained if dispersing larvae can recolonize the exploited site at least as far as the local population is decimated. Should the urchins become locally extinct, that can lead to a switch in the algal communities, that will in turn impact the availability of food for other grazing organisms.

The next part is more relevant to people analyzing data, and is not focused on the ecological consequences of what we found. Part of my implication on this project was to design the statistical analysis (I was not the lucky one that got to dive to count the urchins…). One of the problems we encountered is that we needed a permutational two-way ANOVA with random effects [4]. This particular test requires a balanced design, i.e. there might be the same count of individual in each possible combination of factors. Due to a limited time to carry the whole sampling (that was a lot of work, given that only one of the authors, Anne, was actually diving), one of our sample comprised only 28 individuals (of a species at a site).

We had to work a little magic to apply the test. The obvious solution was to randomly subsample the samples of a greater size, to reduce them to 28 individuals. In order to make sure that the subsamples were comparable to the original population, we checked that their mean and variance were not significantly different. Using R, it was really easy, and I might even post the code (assuming that I find some time to clean it…). This is a good example of what R can do for you: I did not see myself doing this procedure in Statistica or JMP (I’m not even sure this is possible at all).

1. Cybelle Méditerranée
2. for many more on that, see the seminal paper by Hairston, Smith & Slobodkin in 1960
3. an important step in the extinction vortex models
4. the R code of which is made available by Pierre Legendre