Friday 14 June 2013

June 21: Will Cornwell - Measuring phylogenetic diversity

Next up is Will Cornwell, who I'm delighted to say will be taking up an appointment here at UNSW in just a few weeks.  He has diverse interests which include the measurement of functional and phylogenetic diversity, which will be the focus of the session he will speak in at the Symposium.  He has suggested the following introductory book chapter which we will discuss next Friday:
Vellend, Cornwell et al (2011) "Measuring phylogenetic diversity" from the book Biological diversity: frontiers in measurement and assessment, Oxford University Press.

8 comments:

  1. We liked the idea that there was simulation work in this paper - exploring the properties of different methods by carefully designed simulations. Which threw up some interesting results here.

    ReplyDelete
  2. Dan Faith has turned up to our discussion today - awesome!

    ReplyDelete
  3. Dan is not a fan of Type I approaches - one problem is that it doesn't account for redundancy. e.g. if the score for each spp is constructed in isolation of what other spp are present in a community, two closely related spp which co-occur could give a site a high overall score despite sharing lots of branches on the tree. In contrast, a Type II approach would look jointly at where co-occurring spp are in the tree and take their relatedness into account

    ReplyDelete
  4. Dan raised philosophy as an issue with phylogenetics - we often (as here) want to use a given phylo tree and assume it is the true tree, but we have no way of knowing if it actually is the true tree or not (this depends on speciation events that may have happened millions of years ago).

    ReplyDelete
  5. Speaking of uncertainty in trees, an issue that didn't get addressed in depth in this paper was how to incorporate uncertainty into measures of phylogenetic diversity?

    ReplyDelete
  6. Dan is telling us about how he sees Phylogenetic Distance (PD) as an example of a broader "pattern-process" approach to data analysis. I guess the tricky issue though is working out what the process is

    ReplyDelete
  7. Andrew says spatial scale is important - in terms of assessing whether you get phylogenetic clustering or overdispersion. (Often overdispersion at the fine scale, but not so much at broader scales)

    ReplyDelete
  8. We are looking now for take-home messages. One is use distance-based if you can (nodes loses information), type II can be argued on heuristic grounds (but simulations here actually supported type I as a surrogate for PD), ...

    ReplyDelete