So. Last week I was just west of Oslo, in Norway, for the third International Statistical Ecology Conference (as I write registration is still open). This is a core area for Methods, and there was a strong contingent of MEE editors, authors and reviewers present. This was a good opportunity to chat to them, and generally raise the profile of the journal. It’s always nice to get feedback, and also help potential authors thinking about submitting – and even one author who’s paper I had just rejected.

The weather was excellent throughout the meeting:

Norway being nice
Proof it doesn’t always rain in Norway

so, of course, we had to spend so much time inside. But what, you are wondering, did we talk about?

At the last ISEC, the big thing was clearly camera traps. But this time there was almost nothing: now spacially explicit mark-recapture is the big thing. On reflection, this is just camera trap analysis done better: the field has evolved. This was the main shift I noticed: much of the rest was a more granual evolution from what has gone before: a bit of genetics, occupancy modelling (perhaps less than last time: there was very little about species distribution models, for example), animal movement, population dynamics, etc.

Len Thomas gave the final talk, about the future of statistical ecology. He pointed out that there had been a lot about animals (mainly mammals and birds), and very little about plants (e.g. forstry). This, I’m sure, is largely historical: the first two ISEC meetings were organised by groups that mainly work on animals. But it does skew the meeting, and indeed the rest of Len’s talk was about the analysis of animals.

The other thing that struck me was that almost everyone was fitting models, and mostly being Bayesian. I’m a bit worried about this, even as a Bayesian who spends most of his time fitting models. Statistics should be about much more than this: it’s about exploring data, not modelling it. Does this mean that we have developed all the rest of exploration (visualisation, summary statistics and the rest), so we don’t need to research it? Or are we just following the trends in statistics towards fitting complex models? It was interesting, then, two plenary talks stepped away from this: Joanna Mills Flemming talked (amongst other things) about visualising the movement data being collected by the Canadian Ocean Tracking Network, and although Rachel Fewster discussed model-heavy assignment tests (i.e. using genetic data to assign individuals to populations), she stressed the importance of a graphical visualisation of the models. Although there has been a lot of work done on assignment tests, and exclusion tests, and population identification, I haven’t seen this approach before. Perhaps we need a visualisation session at the next ISEC.

Ah, the next ISEC. That will be in MarseilleMontpellier (woo, the south of France), in two years time. They already have a web page. it should be fun, so I hope to see many of you there.