Who the #$%* is Robert Whittaker?

As I was flipping through the September issue of Ecology, I came across something rather interesting. There were two articles that addressed the importance of using multiple approaches to answer research questions. Ironically, both articles were written to address criticism from a man named Robert Whittaker. This man has his strong opinions about using multiple approaches in science.  After reading both Whittaker’s criticism and the author’s responses, I really started to dislike this Whittaker character. Not only is he not a fan of multiple approaches, he discounts hundreds of studies several times throughout both papers, and the basis for his arguments are weak to say the least. Who the #$%* is this guy?

The first of the two papers, written by Partel et. al discusses aims and approaches to the productivity-diversity relationship. The best way to analyze the relationship between productivity and diversity has been through large numbers of case studies. The controversy lies in the selection process for these case studies. The authors believe in a fairly liberal approach to study selection whereas Whittaker takes a more conservative approach. Whittaker (2010) criticises many aspects of Partel’s studies, and this paper is a response to each of Whittaker’s criticisms. This response paper fits into the scope of this blog perfectly emphasizing not only multiple approaches to answering a question, but also using multiple diversity metrics to examine productivity.

An individual ramet

The most traditional way to measure any productivity-trait relationship is per fixed area. According to Whittaker, plot size is the best measure of diversity and the size must be constant when examining field studies. The authors of the paper believe that diversity is just as effectively measured if based on fixed number of ramets, and therefore included this method in their studies. I think this measure would actually be more effective considering individual plants intrinsically vary in their size.

Whitttaker also believes that there must be an adequate surrogate for productivity—that productivity should be measured directly as the rate of carbon flux through the study species. This is rarely done in reality – I did a quick literature search and sources appear sparse. Instead, most researchers use a number of different proxies and traits to measure productivity including: biomass, temperature, precipitation, etc. The authors support this view by comparing results using different proxies to measure productivity and concluding that type of proxy has no effect on the productivity-diversity relationship. 

Studies in particular disturbance regimes are valuable information and should NOT be avoided. The authors agree with this and state that only if there are confounding effects should they be avoided. Whittaker on the other hand believes you should avoid these studies at all costs. I really don’t think Whittaker’s point of view is feasible considering all ecosystems develop under some sort of disturbance regime. Whitaker also suggests that studies of biodiversity should be limited to grassland communities less than 16m² and woodlands less than 200m². The authors cleverly point out that “this would discount most biodiversity studies ever published!”

Honestly, I do understand where Whittaker’s arguments come from. It is tough to make conclusions about data when you don’t have strict experiment selection criteria defined. However, defining more liberal, flexible criteria isn’t always bad. As long as the criteria are reasonable and necessary there is no reason to limit the approaches to answering a specific research question. These criteria should depend on the question being addressed and the scale of the question.

Noise in data->Scatter around regression line

When doing regression analyses, if you pool data from diverse studies this increases the amount of “noise” in the data i.e. the points are not tightly packed around the regression line. This concerns Whittaker, as he says this leads to false rejections of null hypotheses. This issue is something I am much too familiar with in my own research. Importantly, the slope of the line doesn’t change, only the R² value does. The authors state that a false rejection of the null hypotheses therefore does not increase unless the variables are systematically biased. This argument alone puts most of Robert Whittaker’s criticisms to rest.

The second article stressed the importance of multiple approaches when studying the productivity-species richness relationship. This article, written by Gary Mittelbach, really stood out. Right off the bat, I recognized this name from an Undergrad course I took in Lake Ecology. Many of his papers were required readings for that course, and I knew he was an advocate for multiple empirical approaches. Mittelbach started this paper with a quote from Levins’ 1966 paper from American Scientist on Modelling strategies. This paper, if you recall from my very first blog post, inspired most of the ideas behind this blog. Needless to say, I was excited to see what Whittaker had to complain about this time.

Whittaker states: “if the data are not appropriate for meta-analyses, it is invalid to proceed with one”. Well, duh! The problem is how Whittaker determines what is appropriate and what is not that differs from Mittelbach. Mittelbach says that analyzing data using this method isn’t always called for, but that all of the studies in this area that use this technique have contributed to the field as a whole. Sometimes, using data that aren’t entirely appropriate for the analyses helps you learn new, exciting things about the data or the process. Mittelbach adds that the literature is very heterogeneous with respect to study results when using meta-analyses, but that “in the end, we make progress by scrutinizing our ideas”.  

Don't be conservative!

Like Partel, Mittelbach understands the importance of defining criteria for studies but the limits don’t have to be conservative—something that Whittaker argues is unacceptable. This paper is mostly focused on Whittaker’s complaints about what satisfies the criteria to be used in meta-analyses, but Mittelbach includes a great section on alternative and multiple approaches. He knows that in cases where meta-analyses may not be appropriate, that other experimental set-ups or field studies may substitute but more importantly supplement a meta-analysis study. Mittelbach says “I see these approaches as useful alternatives...but I do not see a simple solution and would argue instead that multiple approaches are needed”. 

Mittelbach is right. Whittaker believes that there is a simple solution and that this simple solution can be found by doing very precise experiments. The feelings of Partel, Mittelbach, myself and many other scientists is the exact opposite. When you try to answer any scientific question there is no simple solution. Multiple approaches really are needed. So take that Robert Whittaker!

And in case you were wondering who the #$%* this Whittaker guy is...

http://www.geog.ox.ac.uk/staff/rwhittaker.html

 

Sources:

Partel et. al. 2010. The productivity-diversity relationship: varying aims and approaches. Ecology. 91(9):2565-2567. 

Mittelbach, G. 2010. Understanding species richness-productivity relationships: the importance of meta-analyses. Ecology. 91(9): 2540-2544.

http://qwickstep.com/search/complainer.html

http://www.youtube.com/watch?v=JqHOE1gD4WQ&feature=channel

http://www.cafepress.com/+i%27m-right-you%27re-wrong+coasters

http://mindprod.com/politics/harper.html