What on EARTH is morphometrics?
Indeed, what is morphometrics and why have I chosen it as a topic of interest to open up my new blog with? First of all it’s a topic which is actually new to me. I’ll be doing a morphometrics study with one of my university lecturers within the next few weeks, so it’s essential I keep the topic fresh in my mind.
Morphometrics by definition is the quantitative analysis of size, shape and form. It’s come a long way since traditional methods, now with 3D modelling and an array of specialised software. It’s an incredibly insightful method of analysis, especially in ecology and evolutionary biology. It can be categorised into traditional, geometric landmark-based, and procrustes-based.
I’ll be discussing landmark-based geometric morphometrics specifically, which effectively fills in the blanks left over from traditional or ‘linear’ morphometric methods. If I’m going to have any sort of success describing geometric morphometrics, I’ll first have to discuss its humble origins..
You’ve Got a Friend in Me, Science.
So, let’s use the loveable Disney Pixar classic ‘Toy Story’ as an analogy. All will become clear, don’t close this tab with your scepticism alone, haha.
Traditional morphometrics is the ‘Woody’ of our morphometrics based Toy Story world. During Andy’s childhood Woody provided much entertainment and fulfilment. His simple ‘good guy’ cowboy character provided a suitable role model at the time for a young boy, and at this point although Andy had Slinky, Rex and Bo-Peep…there was no need to buy another toy.
Traditional morphometrics or just Woody in this example, is a measurement of size through analysis of length, width, mass, area, angle etc. The issue with this in a scientific sense, is that size is more often than not linked or correlated with something else. There are a lack of independant variables, (i.e. just as it sounds, a variable that stands alone and is not changed by other variables). For example, how much cake you eat does not affect your age. (Unless you find a magical bakery. I feel like Mary Berry would benefit from this, she’s about 2000 years old right? Anyway…). So if you’re interested in analysing the effect that an independant variable has on something, you need it to not be linked with something else. If you want to ask a guy if he likes you, you generally don’t want his mother to be standing behind him glassy eyed and smiling an unnerving smile.
Got it? Splendid!
So despite this limitation, traditional morphometrics is still highly useful if you’re studying growth over a period of time. ‘The bone was this big then, it is this big now. It has increased in size by this much’.
So, Woody our traditional morphometrics toy helped Andy to grow into the man he is today in Toy Story 3 (soon to be 4). Think about who you are, what you’re about and your favourite childhood toy. Mine was a penguin called ‘Iceberg’, and I am now incredibly interested in polar science, the irony eh? Chances are if your favourite toy as a child was a football, you probably still like football. Or you’re Gary Lineker.
Examples of areas which benefit from traditional morphometrics can be seen with subjects such as functional morphology, say if you’re looking at the mass and relative bone or limb size of an organism. In this case, you WANT there to be dependant variables.
One study by Potter and Corneille (2008) analysed the attractiveness of faces using traditional morphometric methods, not taking into account the geometric configuration of the landmarks. In other words, it was based purely in differences in the median sizes of the features or ‘interlandmarks’ instead of the ARRANGEMENT of these features. (This is how I interpret the methods anyway). Holland (2009) discussed the limitations of this study in terms of the fact as with any traditional morphometric analysis, spatial distribution of the measurements (their geometric pattern) is not taken into account.
To Geometric Morphometrics and Beyond!
You guessed it, Buzz Lightyear is our geometric morphometrics representative. So Andy has Woody, good old Woody. His drawstring still works. ‘There’s a snake in my boot!’ HA. Still makes him laugh or at least snigger to this day. However, can he sit there with Woody years down the line cackling away at that when his mates from school come over? If he wants a ridicule free adolescence, then no.
Geometric morphometrics is based on co-ordinates of placed points on an object, rather than the simple length measurements used in our humble Woody method. However this means that if you are comparing several different samples, you need the landmarks to be the same.
Buzz Landmark to the Rescue
So hold on hold on, what are landmarks? And no, I don’t mean ‘landmarks’ as in the pyramids of Egypt.
In a geometric morphometric study, you’ll usually be comparing several different samples. To do that comparison, you’ll need to find/select common features which are present in each sample. On each sample, you place a landmark or landmark(s), point(s) or commonly referred to as ‘homologous point(s)’. These can be matched up one after the other and categorised as the ‘same point’ on each specimen/sample (Zelditch, 2004). Some scientists will argue that they are not strictly homologous, which is true. More topographically correspondent ‘characters’, which you can describe as a clear specific point.
As well as homologous landmarks, there are also ‘semi-landmarks’ which capture addition information, for example the information contained in curves.
A homologous measurement/part you might chose for example could be a mandible or ‘jawbone’ on a mammalian skull. Although geo. morphometrics can be incredibly useful in palaeontological studies, using fossils can be tricky if they are fragile or missing parts. Saying that, if one sample has something missing, you CAN estimate the landmark. But as with any scientific study, anything that’s going to increase the significance or accuracy of your results is worth using. If the samples are TOO different in terms of their PARTS, the validity of the analysis as a whole will be questionable. Why would you compare leg length between an amputee and Peter Crouch? Ludicrousy!
So back to Buzz…Andy is a little older, a little wiser and a little bit more conscious of how he appears to his friends. He wants to be cooler, he wants the latest tech. He wants the coolest toy. In comes Buzz Lightyear, with his laser, his impressive wingspan, slide back space helmet, his curly chin. The list of features goes on.
Where Woody fails is his lack of toy features, or the seeming quality of those features. Just a voice activating drawstring and a removable cowboy hat. In a landmark-based geometric morphometric study, you want to build as detailed and as accurate a description of the shape as you can with your chosen landmarks. The information that you get about the shape of the object is an entire constellation of all the points that you use.
Buzz is effectively similar to Woody in the sense that they are both toys, both used to create joy in the life our our ‘main man’ Andy. But Buzz has more to offer, Buzz has been created using the underlying PRINCIPLE of Woody but is embracing the modern world. Geometric morphometrics is the newer method which tackles the spatial geometric elements left untouched by the earlier method. If Buzz wasn’t around to save the universe, first of all – Emperor Zurg is just left to run riot and also – we’re able to quantify Woody’s dependence on praise from the other toys more accurately and to a greater extent.
Observe basic freaky Woody above. In a GOOD sample collection you’d generally take a picture of the specimen(s) from a number of different angles, not just the front angle shown in the picture. Observe the differences between traditional and geometric, as you can see, we obtain different information from each method despite studying the same shape.
Using appropriate software e.g. TPS you would then place landmark points on each picture, making sure that the points were placed in the same place on each sample. To get a good representation you’d place several points (many more than the above picture).
So say if I was sampling the variation in jaw shape between a population of living Woody dolls (a terrifying concept, I know) I’d place points along the jaw and may chose to place some on nearby features. I’ve places points on the corners of the mouth for examples sake. As these points would change depending on his expression you wouldn’t ACTUALLY place points there unless all of the dolls were smiling, and that this was a variable you wanted to analyse specifically (e.g. ‘smile shape’). But that of course, would be silly.
So, what have we learned?
Well, first of all that I’m a bit strange.Second of all, that geometric morphometrics is an increasingly useful method of statistical shape analysis and it differs from it’s traditional pedigree in both the information that it retains, and it’s uses in different scientific fields.
I’ll be continuing this post with a follow up post on more of it’s applications, morphometric software and hopefully some information about my OWN study. Woohoo!
Hope you have had an enjoying and informative read from this scientific fruitcake of a graduate. Until the next time!