Posted on 4th October, 2018 by Carlo Pecoraro
Geometric morphometrics (GM) has become a standard in biological research because it combines statistical rigour and ease of interpretation. Through geometric morphometrics,
biological form is quantified, analysed and the results are expressed as easily interpretable and visually impactful shape changes.
Here we have the possibility to discuss about this fascinating field with Dr. Carmelo Fruciano (Ecole Normale Superieure, Paris (France)). Next year we will run, with Carmelo, the second edition of the “Geometric Morphometrics” Workshop (18-22 March 2019) in Berlin!
Carmelo, you have been using morphometric approaches to study a wide range of questions concerning the evolution of organismal shape and its developmental and genetic basis for quite a few years now. Could you please tell us when did you start using GM, and the people and places that have greatly influenced your path in this field?
CF: I started using geometric morphometric methods during my BSc/MSc degree in Natural Sciences back in Italy. The lab where I was carrying out my thesis research was using traditional morphometric methods and my supervisor told me something like “I’ve heard of this new method called geometric morphometrics; why don’t you have a look at it?”. I then started reading many papers and other material on this topic, and slowly using these methods on my data. As I was growing very interested in geometric morphometrics, it seemed natural to use this set of methods during my PhD, complemented by genetics/phylogeography.
It is at this point that the most important event in my development as a morphometrician occurred. At the end of my first year as a PhD student, in 2007, I visited Jim Rohlf’s lab in Stony Brook for a few months. Jim – among many other accomplishments in his career – is one of the fathers of geometric morphometrics and a reference figure in the more general field of biometry. Visiting his lab has been extremely important for me not only as a morphometrician, but more generally as a scientist and an academic. Jim is an exceptional scientist and was very patient with me, helping me better understand things which I had only understood superficially earlier. I have obviously further developed as a morphometrician/biometrician since then, but the experience in Stony Brook has been a true turning point in my career because it has changed my way of approaching morphometrics, quantitative analysis and research.
You have changed different countries for work and now you are living in Paris. Could you please describe your current position and projects?
CF: I am currently working at the Ecole Normale Superieure in Paris as a CNRS Researcher, under the supervision of Helene Morlon. The main work I am currently carrying out in Paris involves addressing the question of what are the evolutionary consequences of phenotypic integration (i.e., the fact that different parts of an organism are not independent from each other but, rather, integrated). This work involves a large amount of morphometric work/statistical analysis.
I’m also involved in multiple other projects, either from previous postdoctoral appointments, or through collaborations I have developed through the years with various research groups. Many – but not all – of these projects involve the use of geometric morphometrics, often in association with other data (e.g., genomic data), in an evolutionary context.
In general, I find very rewarding quantitatively integrating morphometrics with other data types because, while sometimes challenging, it often allows a deeper understanding of reality.
One issue that you have stressed out in your work is the presence and the extent of measurement error: why do you think it is so important to point out sources of error in geometric morphometrics? And how can these errors affect the results?
CF: I initially got interested in this topic at the beginning of my PhD. This is one of those topics that often interests researchers with limited experience in morphometrics, as I was at the time. While over time most of my attention has shifted from data gathering to downstream statistical analysis, this doesn’t mean that the problem is not there. The general tendency is to think of measurement error (e.g., variation in how specimens are placed under the camera or variation between different laser scanners) as something that decreases the “signal to noise ratio” (because artefactual variation is added to biological variation). This, in turn, means that true biological variation (e.g., variation between two populations of the same species) becomes harder to detect. However, measurement error, depending on many factors, can also result in “false positives”.
I think it is important that we understand what the sources of non-biological variation are, how large their contribution to total variation is, and what the consequences of this variation in our downstream statistical analysis are. Understanding the limitations of our work also means better weighing its strengths and its robustness, and it allows for improvement, pretending that measurement error does not exist means missing important information and placing unjustified confidence in one’s work.
Last March we run our first edition of our GM course, but you already got a very good feedback from the participants for your teaching skills and ability of sharing your experience with them in a very clear and understandable way. Can you please describe a bit your course and what are the main differences with the other training courses offered by other companies with other instructors?
CF: The course was very intense but, at the same time, very rewarding for me, because of the high level of interest and participation shown by everyone. Relative to various other geometric morphometric courses I have seen advertised, I feel this course fills a unique niche in that it caters to both beginners and intermediate users of geometric morphometric methods. That is, it is aimed at researchers who intend to use geometric morphometrics or who have started performing geometric morphometric analyses but feel they need a more structured background. This is obtained by placing an emphasis on what’s happening “under the hood” of most common geometric morphometric analyses without sacrificing the practical parts, as well as covering a few topics which are not strictly “beginner topics” (e.g., data quality control, analysis of variation in geographic space using spatially-explicit methods). Most other courses I know of, instead, either tend to cater exclusively to total beginners or, at the other end of the spectrum, tend to focus on topics and downstream analyses for very specific applications.
During the first edition of the course, we also analyse some of your own data and you are involving the participants in a methodological paper using these data. Can you tell us more about this experience and why you did decide to use such approach for the practical sessions of your course?
CF: I decided to use my own real data and involve participants in a real study for various reasons.
First of all, I feel that often example datasets used in practicals/tutorials are designed to “just work” and create at the same time a rewarding experience for the student (as they work) but also unrealistic expectations (as real-life data is often more problematic). By exposing participants to real, imperfect data, I was able to discuss with them the same issues they will encounter on their own datasets. This is also the reason why, in addition to this common dataset used by all participants, I encouraged those who had their own data to work also on their datasets.
Further, those who were interested in participating to a study led by me using the same data and willing to put in the extra effort that this involves (e.g., working on this outside of course hours and after the course end), could also participate. This has two benefits to these willing participants: a coauthorship in a paper and further exposure. This is an exposure to the practical reality of carrying out a geometric morphometric study from beginning to end, and involves interactions and discussion with me and other participants, which is not possible to “compress” in a one-week course.
Overall, this way of structuring the practical parts of this work creates a common basis that all participants share, as well as many opportunities for additional learning based on each participant’s prior experience and willingness to put in extra effort.
The experience has been very nice so far and I am currently seriously considering the possibility of offering a similar opportunity for the next edition of the course (in March 2019), perhaps this time involving participants in an empirical – rather than methodological – study.
GM is increasingly used to analyse variation in shape and discriminate among species and populations in the last years. In your opinion, how these methods are changing and how do you see this field in the near future?
CF: Geometric morphometrics is now a standard set of tools in ecology and evolution and it is applied to questions as disparate as, for instance, how shape varies among populations and species, what are the genetic bases of phenotypic traits, whether different clades evolve phenotypes at different rates, whether there is an association between phenotypes and environmental factors.
Geometric morphometrics has been and will be evolving mainly by incorporating in the workflows new or improved downstream statistical analyses. This means that refined approaches to current questions will be proposed, as well as new question will be addressed in the future using geometric morphometrics. Some of these will also involve improved integration between morphometric and genetic/genomic/ecological data.
I should also mention that geometric morphometrics is increasingly used in other fields such as anthropology, forensic science, biomedical research and archaeology. In some of these areas, geometric morphometrics is still relatively new, rarely used and has not reached the level of methodological sophistication observed in ecology and evolution. However, I believe that cross-disciplinary exchange of ideas in the future will help further developing the field in unexpected, new, and exciting directions.
Thanks Carmelo for your time. See you in Berlin!!
Strictly necessary cookies guarantee functions without which this website would not function as intended. As a result these cookies cannot be deactivated. These cookies are used exclusively by this website and are therefore first party cookies. This means that all information stored in the cookies will be returned to this website.
Functional cookies enable this website to provide you with certain functions and to store information already provided (such as registered name or language selection) in order to offer you improved and more personalized functions.
Performance cookies gather information on how a web page is used. We use them to better understand how our web pages are used in order to improve their appeal, content and functionality.
Marketing / Third Party Cookies originate from external advertising companies (among others) and are used to gather information about the websites visited by you, in order to e.g. create targeted advertising for you.