2016 has been an annus horribilis in many ways, but it has also been my most productive (publication wise) year. Compared to the top, most prolific ecologists out there, my five papers* this year are perhaps pitiful, but for me this year has brought more papers and also special ones. One paper each by my now former PhD students (Drs Polaina and Lucas), a great paper that was hard to publish, one that captures my first trip to Africa, and one in which for the first time I made the entire R script available publicly. I already wrote posts about two of these papers and I am hoping my former students may do a guest posts for theirs sometime soon, so now I figure I should write an entry for the last one. You can read the peer-reviewed comprehensive version here. For a shorter, non-reviewed version keep reading.
Micky Olalla-Tarraga and I met a few years ago in a conference. Afterwards, I invited him to visit Doñana Biological Station where I was working at the time. During this visit we discussed a database describing dietary preferences for large carnivores that Eloy Revilla and I had put together and got interested in the idea of exploring phylogenetic niche conservatism (a pattern of related species retaining niche-associated traits) using these data. Niche conservatism has been primarily studied considering abiotic (Grinnellian) niche dimensions, like climatic preferences. However, we knew relatively little about conservatism in biotic dimensions also called Eltonian. So we decided to explore this question. Coincidentally, as we were starting this project two large databases describing dietary preferences for mammals in general were published (EltonTraits and MammalDIET) so we felt analyzing our more detailed but taxonomically limited database was not enough and decided to work on all three databases to see if closely related mammals tend to have more similar diets than distantly related species and that similarity was consistent with the proposed model of trait evolution: Brownian motion.
A first challenge was to decide how to measure dietary preferences, and much like we had done by deciding to analyse three databases, we did not choose one single metric but explored a whole range, from single, simple metrics describing the number of food categories consumed by each species (diet breadth) to indices that capture the number and frequency (Levin’s index) and quantitative descriptors of diet composition (for example, the proportion of the diet estimated to be composed by birds or leaves). I think this comprehensiveness made for a careful approach (we cannot really capture dietary preferences with one variable) but also resulted in lots of results that offered different conclusions. Some results supported niche conservatism others did not. And part of this variation was due to which database we considered.
This figure shows the phylogenetic signal measured as K in diet breadth for two databases and different groups of mammals (taxonomic orders). Asterisks in the PBM colum indicates the value is not consistent with niche conservatism. For some groups like bats (Chiroptera) we see that MammalDIET and EltonTraits give us different answers. And this is considering the same group of bats (some species were only described in one database but we excluded those for this analysis).
We expected different conclusions from different metrics, as these reflected different aspects of the trophic niche but surely the same metric for the same species should be consistent. After all, these databases were meant to reflect general dietary preferences at the species level, so they should tell us the “same”. The fact that results differed meant there could be a problem (see figure below). It is not a nice message to say that these large, freely available databases may not be as good as we want them to be. In a way we could have expected this, describing dietary preferences, and in general biotic niche dimensions, is difficult (that is way fewer people has tested trophic niche conservatism). But we thought differences would be small and would not affect general conclusions. Yet they did.
So Do closely related mammals have more similar diets than distantly related species? Generally yes, Are those similarities consistent with niche conservatism (expected levels of similarity)? Well, that depends on which metric and database you consider! What we can clearly conclude is that descriptors of dietary preferences need to quantitative and capture spatio-temporal variation. This is not an easy task, but no one said characterizing trophic niches would be easy. Studies describing dietary preferences are no longer popular or easy to publish but without this basic information, those large, presumably more publishable papers, are just not possible (or may end up as our saying “it depends”).
This figure shows the phylogenetic relationships and differences in diet composition for 66 large mammalian carnivores (Families Canidae, Ursidae, and Felidae) as described by the three different diet databases: MammalDIET (triangle), EltonTraits (asterisk) and our database CUFdiet (square). Composition is described by the presence (black symbol) of seven distinct food item categories: Vend (Mammal and Bird), Herptile, Fish, Invertebrate, Fruit, Nectar, and Seeds.
*Update from Jan 2017: turns out this paper has been included in the January 2017 issue, so officially I published only four papers in 2016, which matches my 2008 “record”. 2008 was probably my best professional year, I defended my PhD, secured two postdoctoral positions (which I completed one after the other) and published 4 papers (plus had another, the final chapter of my dissertation, accepted). I also got married that year (although that was largely for legal reasons, for me I already was “married” to who is still my husband).