BRIEF FILE DESCRIPTION but please READ the notes below:

mandibles_A2.nts		MAIN DATA in nts format; the zero Z coordinates has already been added to the 2D photos

mandibles.morphoj		MORPHOJ WORKED OUTS PROJECT with data (3D mandibles and photos but NO SCANS!) and most of the analyses
mandibles_A1_CENTROID_SIZE.past.txt		SIZE data for PAST
mandibles_A1_FOR_MATRIX_CORRELATION.past.txt		SHAPE data for the matrix correlation in PAST
		
mandibles_A2_mean_centered_using_regresion_FOR_CLUSTER_ANALYSIS.past.txt	MEAN CENTERED (using the regression in MorphoJ) SHAPE data for the CLUSTER ANALYSIS in PAST

A2_mandibles_size_and_shape_MorphoJ_common_superimposition.txt 		DATA FOR R: these were exported from MorphoJ after the common superimposition and also include size (and ln size, which is not used)
A2_PERMANOVA.r 		SCRIPT for performing in R the permutational ANOVA on size and mean centered shape (equivalent of the parametric ones performed in MorphoJ)


0) Approach I is abbreviated with A1; approach II with A2: HOWEVER, please notice that a few ANALYSES have TO BE DONE IN PAST: these are all those on size (use file mandibles_A1_CENTROID_SIZE.past.txt), except the ANOVA, and also the matrix correlation using shape distances (use file mandibles_A1_FOR_MATRIX_CORRELATION.past.txt) and the cluster analysis (use file mandibles_A2_mean_centered_using_regresion_FOR_CLUSTER_ANALYSIS.past.txt).

1) Just to make it faster, data in the MorphoJ project were imported together as they are in the mandibles_A2.nts file (i.e., with the zero Z coordinates for the photos). These data however can be used for both approach I and II. This is because, when data are subdivided in MorphoJ into 3D and photos for the approach I, the zero Z coordinates in the photos does not play any role in the superimposition, which is identical to the one obtained using only X,Y coordinates.

2) Read carefully the paper to know what parts of the output to skip, although this is often emphasized in the name of the branch of the project tree. For instance, in measuring the angles between PCs from 3D and photos, one should only look at the angles and SKIP ALL P VALUES.

3) The RV is part of the output of a PLS analysis in MorphoJ: this number is the only one that should be looked at in the context of the TTD analysis. Please, also notice that, instead of linking the 3D data with the photos to do the RV analysis, I exported the PC scores of the photos and reimported them as covariates in the 3D dataset. Results are identical to those one would get after linking the datasets.

4) Data are 'mean centered' using a regression on a dummy variable. Then, residuals are exported (they are equivalent to the manually mean centered data) and reimported as a new dataset for the ANOVA:
NB1: because these are shape residuals, even if MorphoJ does the ANOVA also on size, that is meaningless: never use it and ONLY LOOK AT THE SHAPE RESULTS.
NB2: again simply because of a limitation in the software, the ANOVA on size had to be performed on the original non-mean centered data (i.e., those from the common superimposition); for size this does not make any difference but, now, it is shape that it is meaningless, because of the 'depth bias' that has not yet been removed by mean centering: thus, in this analysis, don't use shape results and ONLY LOOK AT SIZE RESULTS.