Find research datasets worth reusing

Fig. 4 in A practical, step-by-step, guide to taxonomic comparisons using Procrustes geometric morphometrics and user-friendly software (part A): introduction and preliminary analyses

Fig. 4. Graphical examination of replicability in shape using the reduced 12 landmarks configuration. a. PCA scatterplot (in parentheses the variance accounted for by each PC) with convex hulls for the first (grey) and second (red) duplicate. b. Example of phenogram used to count 'sister duplicates' in the whole sample (the inset zooms in the phenogram to exemplify how duplicates 1 and 2 of each individual, e.g., number 97, a female, or number 81, a male, etc., should cluster in pairs, if ME is smaller than inter-individual differences).

Fig. 1 in A practical, step-by-step, guide to taxonomic comparisons using Procrustes geometric morphometrics and user-friendly software (part A): introduction and preliminary analyses

Fig. 1. Study flowchart for both preliminary (A) and main (B) analyses. The flowchart can be used as a reminder for the main analytical steps in a taxonomic study using GMM. To the same aim, at the end of part B, I added a checklist (Appendix B). In the flowchart, I have included the power analysis and few other analyses, which are optional (dotted lines). The power analysis is shown here connected to both the preliminary steps and the group comparisons, because it can be either prospective or retrospective. The sensitivity of results to the inclusion or exclusion of the smallest samples is also connected to both preliminary and main analyses, because it can be used at any step in the analysis. Sometimes (e.g., in the discriminant analysis (DA) of shape), if p is large relative to N and dimensionality reduction is needed, one could also assess the sensitivity of results to the inclusions of different number of PCs. I stress that, as discussed in the main text of both parts A and B, if a taxon is known to have a large SDM, which may vary in pattern depending on the species or subspecies, even preliminary analyses (such as those for ME or outlier detection) should probably be run with separate sexes.

Fig. 3. a in A practical, step-by-step, guide to taxonomic comparisons using Procrustes geometric morphometrics and user-friendly software (part A): introduction and preliminary analyses

Fig. 3. a. Final configuration, reduced to 12 landmarks after excluding low precision landmarks. b. Graphical examination of size replicability (12 landmarks configuration) using a plot of CS in the second duplicate against CS in the first duplicate.

Fig. 2. Initial configuration with 15 in A practical, step-by-step, guide to taxonomic comparisons using Procrustes geometric morphometrics and user-friendly software (part A): introduction and preliminary analyses

Fig. 2. Initial configuration with 15 landmarks (a) and analysis of absolute per-landmark imprecision (b, c). Figure 2b shows the profile plot for the summary statistics of per-landmark variance in the two digitizations. Figure 2c shows the scatter of landmarks purely due to digitization error (red landmarks mark the mean form, to which the differences between the first and second digitization were added).

Fig. A1 in A practical, step-by-step, guide to taxonomic comparisons using Procrustes geometric morphometrics and user-friendly software (part A): introduction and preliminary analyses

Fig. A1. Box and jitter plot of shape SDM estimated using randomized subsampling experiments in yellow-bellied marmots (whiskers in this figure mark the range from minimum to maximum, with no assessment of outliers, which are irrelevant in the context of this didactic example). The total female sample (F) is split in progressively smaller, mutually exclusive, random subsamples and the same is done for males (M). The first subsample of 70 individuals per sex is almost the same of the total samples; the second consists of two female and two male subsamples with 35 individuals each; the third set of subsamples is made of 17–18 individuals per sex and the fourth and fifth include only 10 or 5 individuals respectively. The vertical axis shows the Procrustes distance between means of females and means of males of each set of subsamples. The red circles show the observed mean female to male Procrustes distance in VIM, Alaskan (bro) and Olympic marmots (oly); they are added to the box and jitter plots of yellow-bellied marmot subsamples whose N is closer to the observed N is these three species.

Fig. 5 in A practical, step-by-step, guide to taxonomic comparisons using Procrustes geometric morphometrics and user-friendly software (part A): introduction and preliminary analyses

Fig. 5 (next page). Examples of ME. a. Suspected bias due to a time lag in the data collection. The scatterplots (a2-a3) are ordinations of 3D cranial shapes in a large sample of African adult men; above the ordinations (a1), the mean shape differences between green (first round of data collection, N = 377) and blue (second round, N = 161) groups are shown (magnified five times) in dorsal, side and frontal views. In the PCA (a2) there is a small amount of separation between 'green' and 'blue' on PC2 (variance explained in parentheses). In the DA (a3), the 'green-blue' separation on the horizontal axis (DF1, with, in parentheses, the between group variance explained) is almost perfect despite the fact that groups, whose differences are being maximized in this plot, are in fact the real 32 geographic populations (shown using convex hulls - see main text). b. PC1-2 scatterplots of adult 3D craniofacial shapes in a European sample of adult women (N = 351, shown in pink) and men (N = 380, in blue). The configuration is smaller and different from the one in (a). Above the scatterplots, the shape corresponding to the positive extreme of PC1 (variance explained in parentheses) is shown, in dorsal view, using displacement vectors (b1-b2). The scatterplot to the left (b3) is the full landmark configuration, whereas the one to the right (b4) has euryon removed. Euryon dominates PC1 differences (b3) in the full configuration (b1). After removing it (b2), not only there is no landmark that dominates variation on PC1 (b4), but also the separation of females and males disappears and (see main text) the shape variance accounted for by sex drops from 6% to 3%. c. PC1-2 scatterplot (variance explained in parentheses) of marmot mandibular shape using the reduced landmark configuration in real (c1) and simulated (c2) samples of hoary marmots and woodchucks. The simulated data are obtained by adding to the original shape coordinates large random gaussian noise (SD = 0.05) before Procrustes re-superimposing the data. Random noise (c2) completely obliterates the real differences (c1) and brings the F ratio and Rsq from F = 38 and Rsq = 20%, in the real data, to F = 2 and Rsq = 0.8% in the simulated ones. In both datasets, the tangent space approximation (assessed in TPSSmall) in the real data (c1) produces a correlation of one between shape distances; however, whereas the slope of the least square regression (tangent space Euclidean shape distances onto Procrustes shape distances) is one in the real data, it is 0.97 in the simulated ones, which suggests an almost problematically large amount of shape variance in this second dataset.

Medical Students use Online Study Materials more than School-Provided Resources when preparing for USMLE Step 1

<p>In this excel file contains the raw data collected from a survery sent out to students attend ULSOM and UNRSOM. The raw data was processed and analyzed in the sheet titled "graphs". </p>

Fig. 8 in Assessing the quality of biogeochemical coastal data: a step-wise procedure Abstract

Fig. 8: Percentage of assigned quality flags for a) temperature and salinity, b) nutrients and c) chlorophyll a in the LTER-MC dataset grouped by period, following the simplified QF scheme.

Fig. 4 in Assessing the quality of biogeochemical coastal data: a step-wise procedure Abstract

Fig. 4: A) Schematic representation of a possible spike (V2) evaluated based on its previous (V1) and subsequent (V3) values of the profile. B) Examples of identified spikes in nitrate profiles at low (pMC 535, -5 m), medium (●M276, -5m) and high (○MC48, -2 m) concentrations. Spikes are in orange.

Fig. 1 in Assessing the quality of biogeochemical coastal data: a step-wise procedure Abstract

Fig. 1: Geographical position of sampling stations for LTER-MC (●), Si.Di.Mar (●) and TYR (●) datasets.

Fig. 2 in Assessing the quality of biogeochemical coastal data: a step-wise procedure Abstract

Fig. 2: Frequency distribution of individual parameters in the LTER-MC dataset before applying QC (TEMP: temperature, PSAL: salinity, NTRI: nitrites, PHOS: phosphates, AMON: ammonia, NTRA: nitrates, SLCA: silicates, CHLT: chlorophyll a, S: skewness and K: kurtosis).

Fig. 9 in Assessing the quality of biogeochemical coastal data: a step-wise procedure Abstract

Fig. 9: Flags (¡ QF0, pQF2, and ÚQF6) for surface nitrates (NO3) concentrations in the Si.Di.Mar. dataset.

Fig. 6 in Assessing the quality of biogeochemical coastal data: a step-wise procedure Abstract

Fig. 6: Colour-coded flags for temperature data in monthly vertical temperature profiles in the LTER-MC dataset.

Fig. 5 in Assessing the quality of biogeochemical coastal data: a step-wise procedure Abstract

Fig. 5: Chlorophyll a data for April at 10 m depth: comparison of a standard (on the left) and an adjusted box plot (Hubert & Vandervieren, 2008) at LTER-MC.

Fig. 7 in Assessing the quality of biogeochemical coastal data: a step-wise procedure Abstract

Fig. 7: Percentages of quality flags attributed through the QC procedure to the whole LTER-MC dataset (~84,000 data-points).

FIGURE 4 in One step closer but still far from solving the puzzle - The phylogeny of marine associated mites (Acari, Oribatida, Ameronothroidea) inferred from morphological and molecular genetic data

FIGURE 4 Bayesian inference topology based on 66 morphological traits of 102 oribatid mite species. Posterior probability values are shown near nodes. Photographs of selected species are given to provide an insight into the basic morphology of each larger group. *Photograph shows Tegeocranellus knysnaensis, this species was not used for the analyses but is given here to visualize the typical habitus of Tegeocranellus species.

FIGURE 3 in One step closer but still far from solving the puzzle - The phylogeny of marine associated mites (Acari, Oribatida, Ameronothroidea) inferred from morphological and molecular genetic data

FIGURE 3 One of 14 most parsimonious trees based on 66 characters or character states of 98 ameronothroid and four terrestrial oribatid mite species. Bootstrap values are shown near nodes. Colours refer to different families and are the same as in preceding figures.

FIGURE 1 in One step closer but still far from solving the puzzle - The phylogeny of marine associated mites (Acari, Oribatida, Ameronothroidea) inferred from morphological and molecular genetic data

FIGURE 1 Bayesian inference tree of marine associated Ameronothroidea and terrestrial outgroups based on 18S sequences. Posterior probabilities>0.9 are shown near nodes; abbreviations: PRT – Portugal, DE – Germany, DR – Dominican Republic, JP – Japan, TW – Taiwan, MY – Malaysia; families are given in different colours. Photographs of selected species are given to provide an insight into the basic habitus of each larger group.

FIGURE 2 in One step closer but still far from solving the puzzle - The phylogeny of marine associated mites (Acari, Oribatida, Ameronothroidea) inferred from morphological and molecular genetic data

FIGURE 2 Bayesian topology based on the combined data set of coi, D3 and 18S sequences. Posterior probabilities>0.9 are shown near nodes; abbreviations: PRT – Portugal, DE – Germany, DR – Dominican Republic, TW – Taiwan.