Re: [MORPHMET] Re: semilandmarks in biology
And not to deflect it further, but as long as we are now speaking of automation of landmarking and its implications for transformational homology hypotheses, people may be interested in some of the perspectives and results of this paper on "fully automated" landmarks for diverse shape samples <https://www.researchgate.net/publication/269934848_A_New_Fully_Automated_Approach_for_Aligning_and_Comparing_Shapes> (e.g., claw to nail or gorilla to mouse lemur). ... here is a recent update using a gaussian process for spreading landmarks <https://arxiv.org/pdf/1807.11887.pdf> that emulates human landmarkers in many cases (see Fig. 1). ...an example of using geometric similarity <http://www.pnas.org/content/108/45/18221> to test homology hypotheses (see Fig. 3) On Mon, Nov 12, 2018 at 2:00 AM Murat Maga wrote: > This has been an interesting discussion. Hopefully it has been useful to > the newcomers to the GMM and shape analyses to better understand some of > the challenges they are likely to face. I think the issues of homology, > semi-landmarks, number of variables vs number of samples routinely > discussed here because ultimately there is no hard rule to abide by, but > realities to live with (sample sizes may not be increased) and trade-offs > to be made. I like Benedikt's argument about biological pragmatism. > > I do not want to hijack the thread and the topic, but wanted to briefly > reflect on Benedikt's comments atlas based methods. Image based analyses, > when coupled with a computationally derived anatomical atlas, do offer a > promise of automating some aspects of the acquiring information on > morphology from volumetric scans. This approach can be particularly > powerful, and appealing if one is working with a very large number of > individuals (>>100) of the same species and of similar developmental stage. > I find this approach very useful in tedious preprocessing steps > (segmentation, rigidly aligning samples to a fixed anatomical orientation > say to make standardized 3D renderings of all samples to visually assess > phenotypic variability, etc), basically in processes that can tolerate > large margin of error. Whether they can fully replace landmark based > analyses (or result in fully automated landmarking procedures), I am not > entirely sure. Basically, it boils down to the fact that there is no > independent assessment of how well the registration performed, apart from > the visual inspection of how well the template deformed into the sample (or > the other way around depending on the task). The choice of image similarity > metrics (along with many other parameters than can be tuned) can result in > different outcomes. Even in the well-chewed domain of human neuroimaging > validation of non-linear image registration remains a big issue. They > typically resort to ranking algorithms on how well they approach to the > manually segmented reference datasets. Since atlas-based landmarking is > essentially an image segmentation process, we do need to assess how well > registration simulated the human observer's landmark placement if we are to > justify using one method over another. > > While, I agree with Benedikt's comment "measure the biological effects of > interest rather than how well they simulate the behavior of manually placed > landmarks" in principal, I am not entirely sure how one can go about this > without knowing what the biological effects of interests are beforehand, > because we wouldn't know what we measured. > > M > > > -Original Message- > From: Benedikt Hallgrimsson > Sent: Thursday, November 8, 2018 11:32 AM > To: Adams, Dean [EEOBS] ; andrea cardini < > alcard...@gmail.com>; morphmet@morphometrics.org > Subject: RE: [MORPHMET] Re: semilandmarks in biology > > Dear Colleagues, > > So I’ve been wondering whether to wade into this issue.. > > There seems to be an undercurrent here of mathematics vs biology, but I > suspect that the real issue here is probably morphometric theory versus the > pragmatic compromises necessary when using morphometric tools to answer > biological questions. Others on this thread have thought (and written) > much more deeply about the interface of morphometric theory and biology > than I have, but for what it’s worth, here are my two cents on this issue. > Fundamentally, what is most important is that quantifications of morphology > capture relevant biological variation while avoiding artifacts that can > skew or mislead interpretation. That matters much more to than whether > there is real homology or not. I'm not even sure what "real homology" for > landmark coordinate data means in a biological sense, even for Type 1 > landmarks. The "identity" or homology of lan
RE: [MORPHMET] Re: semilandmarks in biology
.google.com/a/morphometrics.org/group/morphmet/> List-Unsubscribe: <mailto:googlegroups-manage+545891634474+unsubscr...@googlegroups.com>, <https://groups.google.com/a/morphometrics.org/group/morphmet/subscribe> X-MXTHUNDER-Identifier: X-MXTHUNDER-IP-Rating: 1, 140.142.234.176, Ugly c=0.165853 p=0.2 Source Normal X-MXTHUNDER-Scan-Result: 0 X-MXTHUNDER-Rules: 0-0-0-32767-c X-MXTHUNDER-Clean: Yes X-MXTHUNDER-Group: OK This has been an interesting discussion. Hopefully it has been useful to th= e newcomers to the GMM and shape analyses to better understand some of the = challenges they are likely to face. I think the issues of homology, semi-la= ndmarks, number of variables vs number of samples routinely discussed here = because ultimately there is no hard rule to abide by, but realities to live= with (sample sizes may not be increased) and trade-offs to be made. I like= Benedikt's argument about biological pragmatism.=20 I do not want to hijack the thread and the topic, but wanted to briefly ref= lect on Benedikt's comments atlas based methods. Image based analyses, when= coupled with a computationally derived anatomical atlas, do offer a promis= e of automating some aspects of the acquiring information on morphology fro= m volumetric scans. This approach can be particularly powerful, and appeali= ng if one is working with a very large number of individuals (>>100) of the= same species and of similar developmental stage. I find this approach very= useful in tedious preprocessing steps (segmentation, rigidly aligning samp= les to a fixed anatomical orientation say to make standardized 3D rendering= s of all samples to visually assess phenotypic variability, etc), basically= in processes that can tolerate large margin of error. Whether they can ful= ly replace landmark based analyses (or result in fully automated landmarkin= g procedures), I am not entirely sure. Basically, it boils down to the fact= that there is no independent assessment of how well the registration perfo= rmed, apart from the visual inspection of how well the template deformed in= to the sample (or the other way around depending on the task). The choice o= f image similarity metrics (along with many other parameters than can be tu= ned) can result in different outcomes. Even in the well-chewed domain of hu= man neuroimaging validation of non-linear image registration remains a big = issue. They typically resort to ranking algorithms on how well they approac= h to the manually segmented reference datasets. Since atlas-based landmarki= ng is essentially an image segmentation process, we do need to assess how w= ell registration simulated the human observer's landmark placement if we ar= e to justify using one method over another.=20 While, I agree with Benedikt's comment "measure the biological effects of i= nterest rather than how well they simulate the behavior of manually placed = landmarks" in principal, I am not entirely sure how one can go about this w= ithout knowing what the biological effects of interests are beforehand, bec= ause we wouldn't know what we measured. M -Original Message- From: Benedikt Hallgrimsson =20 Sent: Thursday, November 8, 2018 11:32 AM To: Adams, Dean [EEOBS] ; andrea cardini ; morphmet@morphometrics.org Subject: RE: [MORPHMET] Re: semilandmarks in biology Dear Colleagues, So I=E2=80=99ve been wondering whether to wade into this issue.. =20 There seems to be an undercurrent here of mathematics vs biology, but I sus= pect that the real issue here is probably morphometric theory versus the pr= agmatic compromises necessary when using morphometric tools to answer biolo= gical questions. Others on this thread have thought (and written) much mor= e deeply about the interface of morphometric theory and biology than I have= , but for what it=E2=80=99s worth, here are my two cents on this issue. Fu= ndamentally, what is most important is that quantifications of morphology c= apture relevant biological variation while avoiding artifacts that can skew= or mislead interpretation. That matters much more to than whether there is= real homology or not. I'm not even sure what "real homology" for landmark = coordinate data means in a biological sense, even for Type 1 landmarks. Th= e "identity" or homology of landmarks tends to become messy pretty quickly = when the underlying developmental biology is examined closely. I think Paul= O'Higgins gave a great talk once on that basic theme if I remember correct= ly. Chris Percival also did a nice analysis showing how apparently obviousl= y homologous landmarks that occur at intersections of major components of t= he face can drift in terms of the origin of the underlying tissue during de= velopment. So, I think we may sometimes get too hung up on this ideal that = the points that we place on morphological structures actually represent som= ething real. They are simply intended t
RE: [MORPHMET] Re: semilandmarks in biology
Dear Colleagues, So I’ve been wondering whether to wade into this issue.. There seems to be an undercurrent here of mathematics vs biology, but I suspect that the real issue here is probably morphometric theory versus the pragmatic compromises necessary when using morphometric tools to answer biological questions. Others on this thread have thought (and written) much more deeply about the interface of morphometric theory and biology than I have, but for what it’s worth, here are my two cents on this issue. Fundamentally, what is most important is that quantifications of morphology capture relevant biological variation while avoiding artifacts that can skew or mislead interpretation. That matters much more to than whether there is real homology or not. I'm not even sure what "real homology" for landmark coordinate data means in a biological sense, even for Type 1 landmarks. The "identity" or homology of landmarks tends to become messy pretty quickly when the underlying developmental biology is examined closely. I think Paul O'Higgins gave a great talk once on that basic theme if I remember correctly. Chris Percival also did a nice analysis showing how apparently obviously homologous landmarks that occur at intersections of major components of the face can drift in terms of the origin of the underlying tissue during development. So, I think we may sometimes get too hung up on this ideal that the points that we place on morphological structures actually represent something real. They are simply intended to quantify morphology within the context of a biological question. It's not landmarks but rather the patterns of variation that an analysis generates are the objective basis of study and those patterns are only objective within the context of a biological question. The key issue is avoiding artifacts that can influence biological interpretation. In terms of this discussion, clearly semi-landmarks present one kind of challenge where one has to be careful about artifacts. Another, perhaps more currently relevant challenge, however, is the quantification of variation in volumetric images or surfaces that have been nonlinearly registered to an atlas. In this case, one can place landmarks anywhere and recover the corresponding location in every specimen or image. That correspondence is a sort of homology and those landmarks are not slid around like semi-landmarks. However, they are not placed by an observer as distinct observations either. These kinds of points behave fairly similarly to manually placed points (albeit without measurement error and with artifacts that appear as one tries to register increasingly dissimilar shapes). However, I think that, driven by the needs of the biological questions, we are increasingly going to be using this kind of automated quantification of morphology in morphometric analyses, so we need to think carefully about how to validate such data. My own bias here is that appropriate validations address how well (and this can be defined contextually) such quantifications measure the biological effects of interest rather than how well they simulate the behavior of manually placed landmarks. I suppose this is an argument for biological pragmatism, but I hope some find this useful. Benedikt -Original Message- From: Adams, Dean [EEOBS] Sent: Wednesday, November 7, 2018 6:48 AM To: andrea cardini ; morphmet@morphometrics.org Subject: RE: [MORPHMET] Re: semilandmarks in biology Folks, I think it is important to recognize that the example in Andrea’s earlier post does not really address the validity of sliding semilandmark methods, because all of the data were simulated using isotropic error. Thus, the points called semilandmarks in that example were actually independent of one another at the outset. Yet a major reason for using semilandmark approaches is the fact that points along curves and surfaces covary precisely because they are describing those structures. Thus, this interdependence must be accounted for before shapes are compared between objects. The original literature on semilandmark methods makes this, and related issues quite clear. What that means is that evaluating semilandmark methods requires simulations where the points on curves are simulated with known input covariance based on the curve itself (difficult, but not impossible to do). But using independent error will not accomplish this. The result is that treating fixed landmarks as semilandmarks can lead to what some feel are unintended outcomes, just as treating semilandmarks as fixed points are known to do (illustrated nicely in Figs 1-4 of Gunz et al. 2005). But both are mis-applications of methods, not indictments of them. As to the other points in the thread (the number of semilandmark points, etc.), earlier posts by Jim, Philipp, and Mike have addressed these. Dean Dr. Dean C. Adams Director of Gra
RE: [MORPHMET] Re: semilandmarks in biology
Folks, I think it is important to recognize that the example in Andrea’s earlier post does not really address the validity of sliding semilandmark methods, because all of the data were simulated using isotropic error. Thus, the points called semilandmarks in that example were actually independent of one another at the outset. Yet a major reason for using semilandmark approaches is the fact that points along curves and surfaces covary precisely because they are describing those structures. Thus, this interdependence must be accounted for before shapes are compared between objects. The original literature on semilandmark methods makes this, and related issues quite clear. What that means is that evaluating semilandmark methods requires simulations where the points on curves are simulated with known input covariance based on the curve itself (difficult, but not impossible to do). But using independent error will not accomplish this. The result is that treating fixed landmarks as semilandmarks can lead to what some feel are unintended outcomes, just as treating semilandmarks as fixed points are known to do (illustrated nicely in Figs 1-4 of Gunz et al. 2005). But both are mis-applications of methods, not indictments of them. As to the other points in the thread (the number of semilandmark points, etc.), earlier posts by Jim, Philipp, and Mike have addressed these. Dean Dr. Dean C. Adams Director of Graduate Education, EEB Program Professor Department of Ecology, Evolution, and Organismal Biology Iowa State University www.public.iastate.edu/~dcadams/ phone: 515-294-3834 -Original Message- From: andrea cardini Sent: Wednesday, November 7, 2018 4:31 AM To: morphmet@morphometrics.org Subject: Re: [MORPHMET] Re: semilandmarks in biology Making cool pictures has a purpose only if both the pics and the numbers behind them are accurate. It's not an aim in itself, I hope (although this is the second time I hear that one should add as many points as needed to see a nice picture). Parsimonious explanations are, to me, much more appealing than nice pictures (as much as I like a beautiful visualization), but that might be a matter of taste. Philipp, could you clarify what "homology function" means? We're not saying that sliding creates homology, as I sometimes read in papers, are we? No doubt one does not expect anatomical regions of an organism to be independent. The open question to me is what the biological covariance is and what is the bit added by superimposing and maybe sliding. I suspect that on this there's no universal answer: it will be dependent on the study organism, the number and distribution (and type) of landmarks etc. In some studies it might not matter much, but in others may be much more relevant. Thanks all for the comments. Cheers Andrea On 06/11/2018 20:53, mitte...@univie.ac.at wrote: > Yes, it was always well known that sliding adds covariance but this is > irrelevant for most studies, especially for group mean comparisons and > shape regressions: the kind of studies for which GMM is most > efficient, as Jim noted. > If you consider the change of variance-covariance structure due to (a > small amount of) sliding as an approximately linear transformation, > then the sliding is also largely irrelevant for CVA, relative PCA, > Mahalanobis distance and the resulting group classifications, as they > are all based on the relative eigenvalues of two covariance matrices > and thus unaffected by linear transformations. In other words, in the > lack of a reasonable biological null model, the interpretation of a > single covariance structure is very difficult, but the way in which > one covariance structure deviates from another can be interpreted much easier. > > Concerning your example: The point is that there is no useful model of > "totally random data" (but see Bookstein 2015 Evol Biol). Complete > statistical independence of shape coordinates is geometrically > impossible and biologically absurd. Under which biological (null) > model can two parts of a body, especially two traits on a single > skeletal element such as the cranium, be complete uncorrelated? > > Clearly, semilandmarks are not always necessary, but making "cool > pictures" can be quite important in its own right for making good > biology, especially in exploratory settings. Isn't the visualization > one of the primary strengths of geometric morphometrics? > > It is perhaps also worth noting that one can avoid a good deal of the > additional covariance resulting from sliding. Sliding via minimizing > bending energy introduces covariance in the position of the > semilandmarks _along_ the curve/surface. In some of his analyses, Fred > Bookstein just included the coordinate perpendicular to the > curve/surface for the semilandmarks, t
Re: [MORPHMET] Re: semilandmarks in biology
ions are to some extend arbitrary (usually the case)
although still
> along a defined curve then sliding makes sense to me as it
minimizes the
> apparent differences among specimens (the sliding minimizes your
measure of
> how much specimens differ from each other or, usually, the mean
shape.
>
>
>
> _ _ _ _ _ _ _ _ _
>
> F. James Rohlf, Distinguished Prof. Emeritus
>
>
>
> Depts. of Anthropology and of Ecology & Evolution
>
>
>
>
>
> From: mitt...@univie.ac.at >
> Sent: Tuesday, November 6, 2018 9:09 AM
> To: MORPHMET >
> Subject: [MORPHMET] Re: semilandmarks in biology
>
>
>
> I agree only in part.
>
>
>
> Whether or not semilandmarks "really are needed" may be hard to say
> beforehand. If the signal is known well enough before the study,
even a
> single linear distance or distance ratio may suffice. In fact, most
> geometric morphometric studies are characterized by an
oversampling of
> (anatomical) landmarks as an exploratory strategy: it allows for
unexpected
> findings (and nice visualizations).
>
>
>
> Furthermore, there is a fundamental difference between sliding
semilandmarks
> and other outline methods, including EFA. When establishing
correspondence
> of semilandmarks across individuals, the minBE sliding algorithm
takes the
> anatomical landmarks (and their stronger biological homology)
into account,
> while standard EFA and related techniques cannot easily combine
point
> homology with curve or surface homology. Clearly, when point
homology
> exists, it should be parameterized accordingly. If smooth curves
or surfaces
> exists, they should also be parameterized, whether or not this
makes the
> analysis slightly more challenging.
>
>
>
> Anyway, different landmarks often convey different biological
signals and
> different homology criteria. For instance, Type I and Type II
landmarks
> (sensu Bookstein 1991) differ fundamentally in their notion of
homology.
> Whereas Type I landmarks are defined in terms of local anatomy or
histology,
> a Type II landmark is a purely geometric construct, which may or
may not
> coincide with notions of anatomical/developmental homology. ANY
reasonable
> morphometric analysis must be interpreted in the light of the
correspondence
> function employed, and the some holds true for semilandmarks. For
this, of
> course, one needs to understand the basic properties of sliding
landmarks,
> much as the basic properties of Procrustes alignment, etc.. For
instance,
> both the sliding algorithm and Procrustes alignment introduce
correlations
> between shape coordinates (hence their reduced degrees of
freedom). This is
> one of the reasons why I have warned for many years and in many
publications
> about the biological interpretation of raw correlations (e.g.,
summarized in
> Mitteroecker et al. 2012 Evol Biol). Interpretations in terms of
> morphological integration or modularity are even more difficult
because in
> most studies these concepts are not operationalized. They are either
> described by vague and biologically trivial narratives, or they are
> themselves defined as patterns of correlations, which is circular
and makes
> most "hypotheses" untestable.
>
>
>
> The same criticism applies to the naive interpretation of PCA
scree plots
> and derived statistics. An isotropic (circular) distribution of
shape
> coordinates corresponds to no biological model or hypothesis
whatsoever
> (e.g., Huttegger & Mitteroecker 2011, Bookstein & Mitteroecker
2014, and
> Bookstein 2015, all three in Evol Biol). Accordingly, a deviation
from
> isometry does not itself inform about integration or modularity
(in any
> reasonable biological sense).
>
> The multivariate distribution of shape coordinates, including
"dominant
> directions of variation," depend on many arbitrary factors,
including the
> spacing, superimposition, and sliding of landmarks as well as on
the number
> of landmarks relative to the number of cases. But all of this
applies to
> both anatomical landmarks and sliding semilandmarks.
>
>
>
> I don't understand how the fact that semilandmarks makes some of
t
Re: [MORPHMET] Re: semilandmarks in biology
re different > from landmarks and more is not necessarily better. There are > definitely some applications where I find them very useful but many > more where they seem to be there just to make cool pictures. > > As Mike said, we've already had this discussion. Besides different > views on what to measure and why, at that time I hadn't appreciated > the problem with p/n and the potential strength of the patterns > introduced by the covariance created by the superimposition (plus > sliding!). > > Cheers > > Andrea > > On 06/11/2018, F. James Rohlf > wrote: > > I agree with Philipp but I would like to add that the way I think about the > > justification for the sliding of semilandmarks is that if one were smart > > enough to know exactly where the most meaningful locations are along some > > curve then one should just place the points along the curve and > > computationally treat them as fixed landmarks. However, if their exact > > positions are to some extend arbitrary (usually the case) although still > > along a defined curve then sliding makes sense to me as it minimizes the > > apparent differences among specimens (the sliding minimizes your measure of > > how much specimens differ from each other or, usually, the mean shape. > > > > > > > > _ _ _ _ _ _ _ _ _ > > > > F. James Rohlf, Distinguished Prof. Emeritus > > > > > > > > Depts. of Anthropology and of Ecology & Evolution > > > > > > > > > > > > From: mitt...@univie.ac.at <> > > > Sent: Tuesday, November 6, 2018 9:09 AM > > To: MORPHMET > > > Subject: [MORPHMET] Re: semilandmarks in biology > > > > > > > > I agree only in part. > > > > > > > > Whether or not semilandmarks "really are needed" may be hard to say > > beforehand. If the signal is known well enough before the study, even a > > single linear distance or distance ratio may suffice. In fact, most > > geometric morphometric studies are characterized by an oversampling of > > (anatomical) landmarks as an exploratory strategy: it allows for unexpected > > findings (and nice visualizations). > > > > > > > > Furthermore, there is a fundamental difference between sliding > > semilandmarks > > and other outline methods, including EFA. When establishing correspondence > > of semilandmarks across individuals, the minBE sliding algorithm takes the > > anatomical landmarks (and their stronger biological homology) into account, > > while standard EFA and related techniques cannot easily combine point > > homology with curve or surface homology. Clearly, when point homology > > exists, it should be parameterized accordingly. If smooth curves or > > surfaces > > exists, they should also be parameterized, whether or not this makes the > > analysis slightly more challenging. > > > > > > > > Anyway, different landmarks often convey different biological signals and > > different homology criteria. For instance, Type I and Type II landmarks > > (sensu Bookstein 1991) differ fundamentally in their notion of homology. > > Whereas Type I landmarks are defined in terms of local anatomy or > > histology, > > a Type II landmark is a purely geometric construct, which may or may not > > coincide with notions of anatomical/developmental homology. ANY reasonable > > morphometric analysis must be interpreted in the light of the > > correspondence > > function employed, and the some holds true for semilandmarks. For this, of > > course, one needs to understand the basic properties of sliding landmarks, > > much as the basic properties of Procrustes alignment, etc.. For instance, > > both the sliding algorithm and Procrustes alignment introduce correlations > > between shape coordinates (hence their reduced degrees of freedom). This is > > one of the reasons why I have warned for many years and in many > > publications > > about the biological interpretation of raw correlations (e.g., summarized > > in > > Mitteroecker et al. 2012 Evol Biol). Interpretations in terms of > > morphological integration or modularity are even more difficult because in > > most studies these concepts are not operationalized. They are either > > described by vague and biologically trivial narratives, or they are > > themselves defined as patterns of correlations, which is circular and makes > > most "hypotheses" untestable. > > > > > > > > The same criticism a
Re: [MORPHMET] Re: semilandmarks in biology
Yes, it was always well known that sliding adds covariance but this is irrelevant for most studies, especially for group mean comparisons and shape regressions: the kind of studies for which GMM is most efficient, as Jim noted. If you consider the change of variance-covariance structure due to (a small amount of) sliding as an approximately linear transformation, then the sliding is also largely irrelevant for CVA, relative PCA, Mahalanobis distance and the resulting group classifications, as they are all based on the relative eigenvalues of two covariance matrices and thus unaffected by linear transformations. In other words, in the lack of a reasonable biological null model, the interpretation of a single covariance structure is very difficult, but the way in which one covariance structure deviates from another can be interpreted much easier. Concerning your example: The point is that there is no useful model of "totally random data" (but see Bookstein 2015 Evol Biol). Complete statistical independence of shape coordinates is geometrically impossible and biologically absurd. Under which biological (null) model can two parts of a body, especially two traits on a single skeletal element such as the cranium, be complete uncorrelated? Clearly, semilandmarks are not always necessary, but making "cool pictures" can be quite important in its own right for making good biology, especially in exploratory settings. Isn't the visualization one of the primary strengths of geometric morphometrics? It is perhaps also worth noting that one can avoid a good deal of the additional covariance resulting from sliding. Sliding via minimizing bending energy introduces covariance in the position of the semilandmarks _along_ the curve/surface. In some of his analyses, Fred Bookstein just included the coordinate perpendicular to the curve/surface for the semilandmarks, thus discarding a large part of the covariance. Note also that sliding via minimizing Procrustes distance introduces only little covariance among semilandmarks because Procrustes distance is minimized independently for each semilandmark (but the homology function implied here is biologically not so appealing). Best, Philipp Am Dienstag, 6. November 2018 18:34:51 UTC+1 schrieb alcardini: > > Yes, but doesn't that also add more covariance that wasn't there in > the first place? > Neither least squares nor minimum bending energy, that we minimize for > sliding, are biological models: they will reduce variance but will do > it in ways that are totally biologically arbitrary. > > In the examples I showed sliding led to the appearance of patterns > from totally random data and that effect was much stronger than > without sliding. > I neither advocate sliding or not sliding. Semilandmarks are different > from landmarks and more is not necessarily better. There are > definitely some applications where I find them very useful but many > more where they seem to be there just to make cool pictures. > > As Mike said, we've already had this discussion. Besides different > views on what to measure and why, at that time I hadn't appreciated > the problem with p/n and the potential strength of the patterns > introduced by the covariance created by the superimposition (plus > sliding!). > > Cheers > > Andrea > > On 06/11/2018, F. James Rohlf > > wrote: > > I agree with Philipp but I would like to add that the way I think about > the > > justification for the sliding of semilandmarks is that if one were smart > > enough to know exactly where the most meaningful locations are along > some > > curve then one should just place the points along the curve and > > computationally treat them as fixed landmarks. However, if their exact > > positions are to some extend arbitrary (usually the case) although still > > along a defined curve then sliding makes sense to me as it minimizes the > > apparent differences among specimens (the sliding minimizes your measure > of > > how much specimens differ from each other or, usually, the mean shape. > > > > > > > > _ _ _ _ _ _ _ _ _ > > > > F. James Rohlf, Distinguished Prof. Emeritus > > > > > > > > Depts. of Anthropology and of Ecology & Evolution > > > > > > > > > > > > From: mitt...@univie.ac.at > > > Sent: Tuesday, November 6, 2018 9:09 AM > > To: MORPHMET > > > Subject: [MORPHMET] Re: semilandmarks in biology > > > > > > > > I agree only in part. > > > > > > > > Whether or not semilandmarks "really are needed" may be hard to say > > beforehand. If the signal is known well enough before the
Re: [MORPHMET] Re: semilandmarks in biology
ance that wasn't there in >> the first place? >> Neither least squares nor minimum bending energy, that we minimize for >> sliding, are biological models: they will reduce variance but will do >> it in ways that are totally biologically arbitrary. >> >> In the examples I showed sliding led to the appearance of patterns >> from totally random data and that effect was much stronger than >> without sliding. >> I neither advocate sliding or not sliding. Semilandmarks are different >> from landmarks and more is not necessarily better. There are >> definitely some applications where I find them very useful but many >> more where they seem to be there just to make cool pictures. >> >> As Mike said, we've already had this discussion. Besides different >> views on what to measure and why, at that time I hadn't appreciated >> the problem with p/n and the potential strength of the patterns >> introduced by the covariance created by the superimposition (plus >> sliding!). >> >> Cheers >> >> Andrea >> >> On 06/11/2018, F. James Rohlf > <mailto:f.james.ro...@stonybrook.edu>> wrote: >>> I agree with Philipp but I would like to add that the way I think about >>> the >>> justification for the sliding of semilandmarks is that if one were smart >>> enough to know exactly where the most meaningful locations are along >>> some >>> curve then one should just place the points along the curve and >>> computationally treat them as fixed landmarks. However, if their exact >>> positions are to some extend arbitrary (usually the case) although still >>> along a defined curve then sliding makes sense to me as it minimizes the >>> apparent differences among specimens (the sliding minimizes your measure >>> of >>> how much specimens differ from each other or, usually, the mean shape. >>> >>> >>> >>> _ _ _ _ _ _ _ _ _ >>> >>> F. James Rohlf, Distinguished Prof. Emeritus >>> >>> >>> >>> Depts. of Anthropology and of Ecology & Evolution >>> >>> >>> >>> >>> >>> From: mitte...@univie.ac.at >>> Sent: Tuesday, November 6, 2018 9:09 AM >>> To: MORPHMET >>> Subject: [MORPHMET] Re: semilandmarks in biology >>> >>> >>> >>> I agree only in part. >>> >>> >>> >>> Whether or not semilandmarks "really are needed" may be hard to say >>> beforehand. If the signal is known well enough before the study, even a >>> single linear distance or distance ratio may suffice. In fact, most >>> geometric morphometric studies are characterized by an oversampling of >>> (anatomical) landmarks as an exploratory strategy: it allows for >>> unexpected >>> findings (and nice visualizations). >>> >>> >>> >>> Furthermore, there is a fundamental difference between sliding >>> semilandmarks >>> and other outline methods, including EFA. When establishing >>> correspondence >>> of semilandmarks across individuals, the minBE sliding algorithm takes >>> the >>> anatomical landmarks (and their stronger biological homology) into >>> account, >>> while standard EFA and related techniques cannot easily combine point >>> homology with curve or surface homology. Clearly, when point homology >>> exists, it should be parameterized accordingly. If smooth curves or >>> surfaces >>> exists, they should also be parameterized, whether or not this makes the >>> analysis slightly more challenging. >>> >>> >>> >>> Anyway, different landmarks often convey different biological signals >>> and >>> different homology criteria. For instance, Type I and Type II landmarks >>> (sensu Bookstein 1991) differ fundamentally in their notion of homology. >>> Whereas Type I landmarks are defined in terms of local anatomy or >>> histology, >>> a Type II landmark is a purely geometric construct, which may or may not >>> coincide with notions of anatomical/developmental homology. ANY >>> reasonable >>> morphometric analysis must be interpreted in the light of the >>> correspondence >>> function employed, and the some holds true for semilandmarks. For this, >>> of >>> course, one needs to understand the basic properties of sliding >>> landmarks, >>> much as the basic properties of Procrustes alignment, etc.. For >>> instance
Re: [MORPHMET] Re: semilandmarks in biology
Andrea, I am intrigued by your initial comment about adding covariance that was apparently absent. I tend to think of the problem from the other perspective of not accounting for covariance that should be present. As a thought experiment (that could probably be simulated, and maybe I am not correct in my thinking), I like to think of two landmark configurations that are the same in all regards except for one curve, where two groups have distinctly different curves but maybe would not be obviously distinctively different if an insufficient number of semi-landmarks (or none) were used to characterize the curve. If one were to (maybe simulate this example and) use one sparse representation of landmarks and one dense representation, perform a cross-validation classification analysis, and calculate posterior classification probabilities (let’s assume equal sample sizes and, therefore, equal prior probabilities), I would expect that the posterior probabilities of the dense landmark configuration would better assign specimens to the appropriate process that generated them (i.e., their correct groups). The posterior probabilities would be closer to 0 and 1 because of the “added covariance”, as reflected by the squared generalized Mahalanobis distances, based on the pooled within-group covariance. The added covariance would be essential for the posterior probabilities, if the sparse configurations produced similar generalized distances to group means, and therefore, similar posterior probabilities for classification. I’m not sure adding covariance is an issue. To me it simply changes the hypothetical (null) covariance structure, which Philipp mentioned should probably not be assumed to be independent (isotropic). I think your example might best highlight that a different multivariate normal distribution of residuals is to be expected with a different configuration. Cheers! Mike > On Nov 6, 2018, at 12:34 PM, alcardini wrote: > > Yes, but doesn't that also add more covariance that wasn't there in > the first place? > Neither least squares nor minimum bending energy, that we minimize for > sliding, are biological models: they will reduce variance but will do > it in ways that are totally biologically arbitrary. > > In the examples I showed sliding led to the appearance of patterns > from totally random data and that effect was much stronger than > without sliding. > I neither advocate sliding or not sliding. Semilandmarks are different > from landmarks and more is not necessarily better. There are > definitely some applications where I find them very useful but many > more where they seem to be there just to make cool pictures. > > As Mike said, we've already had this discussion. Besides different > views on what to measure and why, at that time I hadn't appreciated > the problem with p/n and the potential strength of the patterns > introduced by the covariance created by the superimposition (plus > sliding!). > > Cheers > > Andrea > > On 06/11/2018, F. James Rohlf <mailto:f.james.ro...@stonybrook.edu>> wrote: >> I agree with Philipp but I would like to add that the way I think about the >> justification for the sliding of semilandmarks is that if one were smart >> enough to know exactly where the most meaningful locations are along some >> curve then one should just place the points along the curve and >> computationally treat them as fixed landmarks. However, if their exact >> positions are to some extend arbitrary (usually the case) although still >> along a defined curve then sliding makes sense to me as it minimizes the >> apparent differences among specimens (the sliding minimizes your measure of >> how much specimens differ from each other or, usually, the mean shape. >> >> >> >> _ _ _ _ _ _ _ _ _ >> >> F. James Rohlf, Distinguished Prof. Emeritus >> >> >> >> Depts. of Anthropology and of Ecology & Evolution >> >> >> >> >> >> From: mitte...@univie.ac.at >> Sent: Tuesday, November 6, 2018 9:09 AM >> To: MORPHMET >> Subject: [MORPHMET] Re: semilandmarks in biology >> >> >> >> I agree only in part. >> >> >> >> Whether or not semilandmarks "really are needed" may be hard to say >> beforehand. If the signal is known well enough before the study, even a >> single linear distance or distance ratio may suffice. In fact, most >> geometric morphometric studies are characterized by an oversampling of >> (anatomical) landmarks as an exploratory strategy: it allows for unexpected >> findings (and nice visualizations). >> >> >> >> Furthermore, there is a fundamental difference betwee
RE: [MORPHMET] Re: semilandmarks in biology
Perhaps, but Procrustes superimposition already adds lots of covariances also. It is a bit tricky (meaning that I do not know of a good solution) to preserve the "real" covariances and distinguish them from artifacts of fitting. GM works well for testing differences among means of groups but studying covariances among shape variables is a much more difficult problem. Some ML approaches have been suggested that could minimize the covariances due to superimposition but the ones I have looked at require some very unreasonable biological assumptions about their statistical properties. Such discussions will not die until there are good solutions or else someone proves that no good solution is possible. I still have hope for some clever idea. _ _ _ _ _ _ _ _ _ F. James Rohlf, Distinguished Prof. Emeritus Depts. of Anthropology and of Ecology & Evolution -Original Message- From: alcardini Sent: Tuesday, November 6, 2018 12:35 PM To: F. James Rohlf Cc: mitte...@univie.ac.at; MORPHMET Subject: Re: [MORPHMET] Re: semilandmarks in biology Yes, but doesn't that also add more covariance that wasn't there in the first place? Neither least squares nor minimum bending energy, that we minimize for sliding, are biological models: they will reduce variance but will do it in ways that are totally biologically arbitrary. In the examples I showed sliding led to the appearance of patterns from totally random data and that effect was much stronger than without sliding. I neither advocate sliding or not sliding. Semilandmarks are different from landmarks and more is not necessarily better. There are definitely some applications where I find them very useful but many more where they seem to be there just to make cool pictures. As Mike said, we've already had this discussion. Besides different views on what to measure and why, at that time I hadn't appreciated the problem with p/n and the potential strength of the patterns introduced by the covariance created by the superimposition (plus sliding!). Cheers Andrea On 06/11/2018, F. James Rohlf wrote: > I agree with Philipp but I would like to add that the way I think > about the justification for the sliding of semilandmarks is that if > one were smart enough to know exactly where the most meaningful > locations are along some curve then one should just place the points > along the curve and computationally treat them as fixed landmarks. > However, if their exact positions are to some extend arbitrary > (usually the case) although still along a defined curve then sliding > makes sense to me as it minimizes the apparent differences among > specimens (the sliding minimizes your measure of how much specimens differ > from each other or, usually, the mean shape. > > > > _ _ _ _ _ _ _ _ _ > > F. James Rohlf, Distinguished Prof. Emeritus > > > > Depts. of Anthropology and of Ecology & Evolution > > > > > > From: mitte...@univie.ac.at > Sent: Tuesday, November 6, 2018 9:09 AM > To: MORPHMET > Subject: [MORPHMET] Re: semilandmarks in biology > > > > I agree only in part. > > > > Whether or not semilandmarks "really are needed" may be hard to say > beforehand. If the signal is known well enough before the study, even > a single linear distance or distance ratio may suffice. In fact, most > geometric morphometric studies are characterized by an oversampling of > (anatomical) landmarks as an exploratory strategy: it allows for > unexpected findings (and nice visualizations). > > > > Furthermore, there is a fundamental difference between sliding > semilandmarks and other outline methods, including EFA. When > establishing correspondence of semilandmarks across individuals, the > minBE sliding algorithm takes the anatomical landmarks (and their > stronger biological homology) into account, while standard EFA and > related techniques cannot easily combine point homology with curve or > surface homology. Clearly, when point homology exists, it should be > parameterized accordingly. If smooth curves or surfaces exists, they > should also be parameterized, whether or not this makes the analysis slightly > more challenging. > > > > Anyway, different landmarks often convey different biological signals > and different homology criteria. For instance, Type I and Type II > landmarks (sensu Bookstein 1991) differ fundamentally in their notion of > homology. > Whereas Type I landmarks are defined in terms of local anatomy or > histology, a Type II landmark is a purely geometric construct, which > may or may not coincide with notions of anatomical/developmental > homology. ANY reasonable morphometric analysis must be interpreted in > the light of the correspondence function employed, and the some h
Re: [MORPHMET] Re: semilandmarks in biology
Yes, but doesn't that also add more covariance that wasn't there in the first place? Neither least squares nor minimum bending energy, that we minimize for sliding, are biological models: they will reduce variance but will do it in ways that are totally biologically arbitrary. In the examples I showed sliding led to the appearance of patterns from totally random data and that effect was much stronger than without sliding. I neither advocate sliding or not sliding. Semilandmarks are different from landmarks and more is not necessarily better. There are definitely some applications where I find them very useful but many more where they seem to be there just to make cool pictures. As Mike said, we've already had this discussion. Besides different views on what to measure and why, at that time I hadn't appreciated the problem with p/n and the potential strength of the patterns introduced by the covariance created by the superimposition (plus sliding!). Cheers Andrea On 06/11/2018, F. James Rohlf wrote: > I agree with Philipp but I would like to add that the way I think about the > justification for the sliding of semilandmarks is that if one were smart > enough to know exactly where the most meaningful locations are along some > curve then one should just place the points along the curve and > computationally treat them as fixed landmarks. However, if their exact > positions are to some extend arbitrary (usually the case) although still > along a defined curve then sliding makes sense to me as it minimizes the > apparent differences among specimens (the sliding minimizes your measure of > how much specimens differ from each other or, usually, the mean shape. > > > > _ _ _ _ _ _ _ _ _ > > F. James Rohlf, Distinguished Prof. Emeritus > > > > Depts. of Anthropology and of Ecology & Evolution > > > > > > From: mitte...@univie.ac.at > Sent: Tuesday, November 6, 2018 9:09 AM > To: MORPHMET > Subject: [MORPHMET] Re: semilandmarks in biology > > > > I agree only in part. > > > > Whether or not semilandmarks "really are needed" may be hard to say > beforehand. If the signal is known well enough before the study, even a > single linear distance or distance ratio may suffice. In fact, most > geometric morphometric studies are characterized by an oversampling of > (anatomical) landmarks as an exploratory strategy: it allows for unexpected > findings (and nice visualizations). > > > > Furthermore, there is a fundamental difference between sliding semilandmarks > and other outline methods, including EFA. When establishing correspondence > of semilandmarks across individuals, the minBE sliding algorithm takes the > anatomical landmarks (and their stronger biological homology) into account, > while standard EFA and related techniques cannot easily combine point > homology with curve or surface homology. Clearly, when point homology > exists, it should be parameterized accordingly. If smooth curves or surfaces > exists, they should also be parameterized, whether or not this makes the > analysis slightly more challenging. > > > > Anyway, different landmarks often convey different biological signals and > different homology criteria. For instance, Type I and Type II landmarks > (sensu Bookstein 1991) differ fundamentally in their notion of homology. > Whereas Type I landmarks are defined in terms of local anatomy or histology, > a Type II landmark is a purely geometric construct, which may or may not > coincide with notions of anatomical/developmental homology. ANY reasonable > morphometric analysis must be interpreted in the light of the correspondence > function employed, and the some holds true for semilandmarks. For this, of > course, one needs to understand the basic properties of sliding landmarks, > much as the basic properties of Procrustes alignment, etc.. For instance, > both the sliding algorithm and Procrustes alignment introduce correlations > between shape coordinates (hence their reduced degrees of freedom). This is > one of the reasons why I have warned for many years and in many publications > about the biological interpretation of raw correlations (e.g., summarized in > Mitteroecker et al. 2012 Evol Biol). Interpretations in terms of > morphological integration or modularity are even more difficult because in > most studies these concepts are not operationalized. They are either > described by vague and biologically trivial narratives, or they are > themselves defined as patterns of correlations, which is circular and makes > most "hypotheses" untestable. > > > > The same criticism applies to the naive interpretation of PCA scree plots > and derived statistics. An isotropic (circular) distribution of shape > coordinates corresponds
RE: [MORPHMET] Re: semilandmarks in biology
I agree with Philipp but I would like to add that the way I think about the justification for the sliding of semilandmarks is that if one were smart enough to know exactly where the most meaningful locations are along some curve then one should just place the points along the curve and computationally treat them as fixed landmarks. However, if their exact positions are to some extend arbitrary (usually the case) although still along a defined curve then sliding makes sense to me as it minimizes the apparent differences among specimens (the sliding minimizes your measure of how much specimens differ from each other or, usually, the mean shape. _ _ _ _ _ _ _ _ _ F. James Rohlf, Distinguished Prof. Emeritus Depts. of Anthropology and of Ecology & Evolution From: mitte...@univie.ac.at Sent: Tuesday, November 6, 2018 9:09 AM To: MORPHMET Subject: [MORPHMET] Re: semilandmarks in biology I agree only in part. Whether or not semilandmarks "really are needed" may be hard to say beforehand. If the signal is known well enough before the study, even a single linear distance or distance ratio may suffice. In fact, most geometric morphometric studies are characterized by an oversampling of (anatomical) landmarks as an exploratory strategy: it allows for unexpected findings (and nice visualizations). Furthermore, there is a fundamental difference between sliding semilandmarks and other outline methods, including EFA. When establishing correspondence of semilandmarks across individuals, the minBE sliding algorithm takes the anatomical landmarks (and their stronger biological homology) into account, while standard EFA and related techniques cannot easily combine point homology with curve or surface homology. Clearly, when point homology exists, it should be parameterized accordingly. If smooth curves or surfaces exists, they should also be parameterized, whether or not this makes the analysis slightly more challenging. Anyway, different landmarks often convey different biological signals and different homology criteria. For instance, Type I and Type II landmarks (sensu Bookstein 1991) differ fundamentally in their notion of homology. Whereas Type I landmarks are defined in terms of local anatomy or histology, a Type II landmark is a purely geometric construct, which may or may not coincide with notions of anatomical/developmental homology. ANY reasonable morphometric analysis must be interpreted in the light of the correspondence function employed, and the some holds true for semilandmarks. For this, of course, one needs to understand the basic properties of sliding landmarks, much as the basic properties of Procrustes alignment, etc.. For instance, both the sliding algorithm and Procrustes alignment introduce correlations between shape coordinates (hence their reduced degrees of freedom). This is one of the reasons why I have warned for many years and in many publications about the biological interpretation of raw correlations (e.g., summarized in Mitteroecker et al. 2012 Evol Biol). Interpretations in terms of morphological integration or modularity are even more difficult because in most studies these concepts are not operationalized. They are either described by vague and biologically trivial narratives, or they are themselves defined as patterns of correlations, which is circular and makes most "hypotheses" untestable. The same criticism applies to the naive interpretation of PCA scree plots and derived statistics. An isotropic (circular) distribution of shape coordinates corresponds to no biological model or hypothesis whatsoever (e.g., Huttegger & Mitteroecker 2011, Bookstein & Mitteroecker 2014, and Bookstein 2015, all three in Evol Biol). Accordingly, a deviation from isometry does not itself inform about integration or modularity (in any reasonable biological sense). The multivariate distribution of shape coordinates, including "dominant directions of variation," depend on many arbitrary factors, including the spacing, superimposition, and sliding of landmarks as well as on the number of landmarks relative to the number of cases. But all of this applies to both anatomical landmarks and sliding semilandmarks. I don't understand how the fact that semilandmarks makes some of these issues more obvious is an argument against their use. Best, Philipp Am Dienstag, 6. November 2018 13:28:55 UTC+1 schrieb alcardini: As a biologist, for me, the question about whether or not to use semilandmarks starts with whether I really need them and what they're actually measuring. On this, among others, Klingenberg, O'Higgins and Oxnard have written some very important easy-to-read papers that everyone doing morphometrics should consider and carefully ponder. They can be found at: https://preview.tinyurl.com/semilandmarks I've included there also an older criticis
