Tips for Writing a Systematics DDIG Part 3: What About Preliminary Data?

You’re not going to get a DDIG without some preliminary data. There are several layers of preliminary data to consider. The first layer - showing enough to convince the reviewers that you’re capable of gathering the data that you’ve proposed to gather - is essential. Don’t try telling reviewers you’re going to sequence 10 nuclear genes if you have no published molecular phylogenetic studies and have yet to sequence a single bp for your project. A second layer involves enough data and analyses for the reviewers to determine whether the work you’ve proposed is likely to be sufficient to answer the question at hand. This is the classic chicken and egg problem with grants - you can’t get a grant if you can’t get the data and you can’t get the data if you don’t have a grant. Remember that this is a dissertation improvement grant, not a dissertation grant: you should do what you can to convince your reviewers that you’re already well on your way toward successful completion of your thesis.

Tips for Writing a Systematics DDIG Part 2: How are these things reviewed?

DDIGs are reviewed using a panel-based system similar to that used to evaluate larger proposals submitted to NSF. The DDIG panel in systematics consists of 20 or so practicing systematists drawn from a wide range of institutions (museums, research universities, liberal arts colleges) and subdisciplines (paleontology, taxonomy, biogeography). The group is supervised by the Systematics and Biodiversity Inventories cluster program officers. Several weeks prior to meeting at NSF headquarters, each proposal will be assigned to three reviewers, one of whom will be designated the primary reviewer. Each of these three reviewers is expected to read your proposal in detail and to provide written comments and a proposal evaluation (excellent to poor) prior to the time the panel convenes. Once the panel has assembled, proposals are dealt with one at a time. When a proposals name is called, the primary reviewer gives a brief overview and assessment before opening things up for discussion. Discussion is generally limited to the three previously assigned reviewers. Others on the panel are free to comment as well, but they’re generally too busy worrying about their own proposals to do so. The panel then arrives at a consensus on each proposal, which generally involves placing into one of three categories: (1) definitely fund, (2) potentially fundable, and (3) unfundable. Once review of all the proposals is completed, there may be a number of proposals in the potentially fundable column that get a second look, perhaps moving to one of the other two columns if it seems warranted with hind sight. The panel does not make final funding decisions, only recommendations.

The point of sharing this information is this: to get a DDIG you need to write a proposal that will impress a potentially diverse group of three practicing systematists.

Tips for Writing a Systematics DDIG Part 1: Organizing Your Proposal

This time of year just about every PhD candidate in systematics who doesn’t already have one is working on a proposal for one of the NSF’s lucrative Doctoral Dissertation Improvement Grants. The DDIGs are one of the smartest ideas the good folks at NSF have ever had, and represent a critical source of funding for ambitious and independent young systematists. The sad fact is that there aren’t many other grants available to graduate students that offer the type of $10,000+ windfall that can be essential to making a good thesis a great thesis. Although the program is incredibly popular, some find the application process a bit mysterious. The NSF’s formal guidelines certainly provide you with all the basics, but they’re also somewhat open ended.

How one can best prepare a competitive proposal? Although there aren’t any foolproof answers to this question, I’d like to share a few suggestions I’ve developed for my own graduate students. These suggestions, which undoubtedly reflect my own personal biases, are being made on the basis of having read previously successful (and unsuccessful) proposals and discussions with NSF reviewers who have been involved in evaluating these proposals. I’m going to kick things off in this first post with some basic advice on organizing your proposal, followed by subsequent posts on how proposals are reviewed, how best to incorporate preliminary data, how much methodological detail to include, and how to effectively discuss broader impacts.

A good proposal begins with good organization. There are lots of ways to organize a successful proposal, so how you choose to organize yours is a personal decision that requires lots of careful thought. That said, one general organizational feature that tends to characterize successful proposals is the use of a strong hypothesis testing framework. Think of this as getting back to basics: remember how your freshman biology lab reports started by outlining the specific hypotheses you tested? Doing the same here is going to help your reviewers understand exactly what you are trying to accomplish with your work, while at the same time helping you organize the remainder of your proposal.

Instead of making vague claims like “I will investigate the biogeographic history of midges”, try to make a more specific statement like “I will test the hypothesis that the distribution of midge diversity is a consequence of a vicariant event associated with the uplift of the Andean plateau.” Distilling your work into a few explicit hypotheses can feel a bit constraining when your real goal is to understand why midges are so darned diverse, but being explicit about specific hypotheses does not preclude you from following up on other interesting results that might be somewhat peripheral. You need to provide some context for your hypotheses before introducing them, but try to get to them as soon as possible; your reviewers shouldn’t be able to get past the first page of your proposal without being provided with a concise statement of the questions you intend to address. Try to restrict yourself to a manageable number of hypotheses (things get a bit out of hand when proposals try to juggle a half dozen or more hypotheses, for example). Organize the remainder of your proposal (e.g., methods, discussion, preliminary data) around the hypotheses presented on the first page of your proposal. Make sure that your work can feasibly address each of your hypotheses.

City Life - and the Evolution of Immunity

Much has been made about the very strong association between rural living and protection against allergies and asthma (for instance, in this recent study here). However, a new study (available "Early View" from Evolution) claims a strong effect of urbanization on the genetic basis of disease resistance - at least in human pre-history. In particular, the authors find evidence suggesting that the duration of urbanization strongly predicts the frequency of a TB resistance conferring genetic allele among modern human populations of known historical affinity. The allele is non-randomly distributed geographically, but the authors attempt to control for this non-independence by also analyzing their data using a partial Mantel test (a non-parametric multiple matrix regression procedure). In this test, they fit a multiple regression model with independent variables consisting of a matrix containing the differences in urbanization and a matrix containing F_ST values computed among each pair of populations. They found that the urbanization effect was still very significant in this model.

One concern raised and discussed by the authors is that the domestication and utilization of cattle (a proposed disease vector for TB) roughly coincides with the progress of urbanization in the region. They argue that we can reject this model because correlation is weaker than in the urbanization model; however, in my mind this argument falls short of persuasiveness because (as they admit) the history of cattle domestication for many of their populations is poorly known. This type of error would obviously also have the effect of depressing our perceived correlation between cattle domestication and genetic TB resistance.

Nonetheless, this is a very interesting study. If the result holds up to future scrutiny, then this will no doubt have many relevant human health implications and the study should be broadly cited.

Somebody Missed the Dover Trial...

I'm a week behind the times on this, but perhaps some of you missed out on Delaware Senatorial candidate Christine O'Donnell's most recent debate performance. This particular debate was made famous by the Republican/Tea Party candidate's ignorance of the constitution ("Can you remind me of what the [14th and 16th amendments] are?") and the fact that she was unaware of the first amendment's establishment clause ("Where in the constitution is separation of church and state?" [laughter from law school audience], "Let me just clarify, you're telling me the separation of church and state is found in the first amendment?"). However, there are also some real gems about evolution and intelligent design, including the claim that creationism and intelligent design are distinct (at the 1:15 mark). Do you think Judge John E. Jones III is available for interventions?

Testing for Trait-Dependent Molecular Evolution

Itay Mayrose & Sally Otto have just published (Molecular Biology and Evolution Advance Access) a neat new method to test the hypothesis of a discrete extrinsic cause for shifts in the rate of molecular evolution on a phylogeny.

According to this method, the authors first obtain an ultrametric phylogenetic tree for the species in their study. They then generate a set of stochastic character histories (Nielsen, 2002; Huelsenbeck et al., 2003) for the discrete character of interest. Example discrete characters might be a "life history trait, morphological feature, or habitat association" - in their empirical test they examine halophilic and freshwater Daphnia species.

Now armed with a distribution of possible character histories on their estimated phylogeny, the authors simultaneously maximize the likelihood of their sequence evolution model and a scaling factor r, a parameter that increases or suppresses the rate of molecular evolution along stochastically mapped branches in the tree. Then they average across character maps.

In an extremely clear analysis of their method, the authors show it capable of producing remarkably good estimates of r for trees with even a modest number of tips (e.g., 20-60) when the true underlying phylogeny is known without error (Figure panel A). Under these idealized circumstances, estimation of r is only slightly biased for small numbers of species - as is common for maximum likelihood methods.

The situation is slightly more complicated when an estimated phylogeny (rather than the true underlying tree and branch lengths) is used. Here, they show that estimation of r can be quite severely downwardly biased, particularly for large values of r (Figure panel B). They think that this is actually due to error in the ultrametricization of their phylogenies - since in their study they used the same data for phylogenetic inference as they do for the estimation of r. This problem is not at all ameliorated for ultrametric phylogenies obtained by Bayesian relaxed clock methods. In the end, this issue argues strongly for the simultaneous estimation of the phylogeny, the character history, and the concomitant variation in nucleotide substitution rates - something that the authors also recommend.

Leigh Van Valen (1935-2010)

Friends confirm the reports elsewhere on the web [a,b] that Leigh Van Valen died this Saturday in Chicago. He was 75.

Van Valen published on a wide variety of topics, but may be best known as the originator of the Ecological Species Concept and the Red Queen's Hypothesis. While it is difficult to summon authoritative information, the latter appeared in what may be the most successful self-published manuscript in the history of our field [1], presently cited 1402 times according to Google Scholar.

One of my personal favorites, for its vision and clarity, was his paper on clade selection [2]. It was virtually ignored until very recently, but it will likely receive a renewed look in view of recent developments [e.g., 3].

[1] Van Valen, L. 1973. A new evolutionary law. Evolutionary Theory 1:1-30.

[2] Van Valen, L. 1975. Group selection, sex, and fossils. Evolution 29:87-94.

[3] FitzJohn, R. G. 2010. Quantitative traits and diversification. Systematic Biology (in press) doi:10.1093/sysbio/syq053.

[4] Photo Source 'e&e': http://picasaweb.google.com/ecolevo.students/DirectoryPicsFaculty

Estimating Diversification Rates

A new study by Wertheim & Sanderson (Evolution; Online Accepted Articles) investigates the sensitivity of existing methods for estimating diversification rates to various types of phylogenetic error. The topic is somewhat related to the recent, provocatively titled Evolution paper by Dan Rabosky clearly showing that when the assumption of constant birth is violated, death (extinction) rates can no longer be reliably estimated from molecular phylogenies.

Especially given Rabosky's (2010) main result - that is, the high sensitivity of extinction rate estimates to certain model assumptions - the new study by Wertheim & Sanderson is particularly intriguing. Although these methods typically assume that the tree and branch lengths are known without error, Werheim & Sanderson demonstrate in their study that diversification rate estimates are not particularly sensitive to phylogenetic errors either in branch length or topology. In fact, they note in the abstract that even a "crude estimate" of the tree provides substantially more power (e.g., 1.6 x more for the conditions of their study) than, for instance, a comparable non-phylogenetic method, the widely used Slowinski-Guyer test (Slowinski & Guyer 1993; Am. Nat.).

Considered together, these two studies remind us that the robustness of a given statistical method cannot be illustrated by a broad brush. Rabosky's study shows that the estimation of extinction rates from phylogenies of extant species is quite sensitive to the underlying assumption that speciation rates are constant throughout the tree. Conversely, Werheim & Sanderson show that the estimation of speciation rate is not sensitive to the underlying assumption that the phylogenetic tree and branch lengths are known without error - and, furthermore, that even a "crude" tree will do.

Note that the figure above is from neither study - but from my 2005 paper (with Dechronization bloggers Harmon & Glor) about the sensitivity of diversification rate estimates to model parameterization. (We found it to be high.)

New Issue of Systematic Biology

The October issue of Systematic Biology is online, and there are a few really interesting articles to check out. First of all, the cover image (image stolen from Syst. Biol. website) comes from a paper by Parfrey et al. that seeks to resolve the Eukaryotic portion of the tree of life. One issue with many phylogenetic analyses at this scale is the shape of the data matrix - many characters, sometimes whole genomes, but very few taxa, typically representing "key lineages" in the tree. Parfrey et al. use sequences from a moderate number of genes (16) across many lineages (>400). They have some success in resolving the tree - in particular, they are able to place a few old enigmatic taxa in the tree - but some branches are still unresolved.

Another paper worth checking out is a short note by Folmer Bokma that represents one of the first applications of Approximate Bayesian Computation in comparative methods. If you haven't heard of ABC yet - you will. There is another remarkable aspect of the Bokma paper - let me just quote the funding section:

The author awards SEK 10,000 to the first who provides an analytical form of φ.

According to google currency converter that's about USD $1500, which would just about pay for that R. A. Fisher tattoo you've been wanting.

On another note, please watch this:

Dar-WIN

Evolution Since Darwin

I spent last night and this morning reading a number of chapters from the book Evolution Since Darwin: The First 150 Years (2010, ed. by Bell, Futuyma, Eanes, and Levinton; link). The book is the results of a symposium that was held at SUNY-Stony Brook in 2009.

I'm really impressed with this volume. Sometimes edited volumes can be a little dry - who wants to say something really new and important in a book chapter, anyway? But this book is much better than most volumes.

The book starts with an amazingly compact yet comprehensive history of the last 150 years in Evolution (Futuyma). There's a really interesting discussion of what the world would be like if Darwin had died young (Bowler), and a nice discussion of solved and unsolved problems in evolutionary genetics (Zhang). I love the whole section on Diversity and the Tree of Life, with contributions from Losos, Hillis, and Wagner, among others. And that's really just scratching the surface.

The authors have all made a deliberate attempt to tie their chapters to Darwin, following the theme of the book and associated symposium. I don't think this works that well in some of the chapters - it feels more like a distraction at times. But I do see the point of celebrating Darwin! Anyway this doesn't detract from the value of the book, which I highly recommend.