trajectory.analysis: Quantify and compare shape change trajectories

In RRPP: Linear Model Evaluation with Randomized Residuals in a Permutation Procedure

Quantify and compare shape change trajectories

Description

Function estimates attributes of multivariate trajectories

Usage

trajectory.analysis(
  fit,
  fit.null = NULL,
  groups,
  traj.pts,
  pca = TRUE,
  print.progress = FALSE
)

Arguments

fit	A linear model fit using lm.rrpp.
fit.null	An alternative linear model fit to use as a null model for RRPP, if the null model of the fit is not desired. Note, if RRPP = FALSE (FRPP rather than RRPP), then the null model has only an intercept. If the null model is uncertain, using reveal.model.designs will help elucidate the inherent null model used.
groups	A factor or vector coercible to factor that defines trajectories.
traj.pts	Either a single value or a vector coercible to factor to define trajectory points. If only a single value, it is assumed that the data are already in the form, y1p1, y2p1, y3p1, ...., y2p2, y2p2, y3p2, ..., yjp1, yjp2, yjp3, ..., yjpk, for j variables comprising k trajectory points; i.e., traj.pts = k. If a factor, then a group * traj.pt factorial model is assumed, where traj.pts defines the levels for points within groups.
pca	A logical value to optionally project group:point means onto principal components (perform PCA on a covariance matrix of the means) This option only applies to factorial designs (traj.pts is a factor).
print.progress	A logical value to indicate whether a progress bar should be printed to the screen. This is helpful for long-running analyses.

Details

The function quantifies multivariate trajectories from a set of observations, and assesses variation in attributes of the trajectories via RRPP. A trajectory is defined by a sequence of points in the data space. These trajectories can be quantified for various attributes (their size, orientation, and shape), and comparisons of these attribute enable the statistical comparison of shape change trajectories (Collyer and Adams 2007; Adams and Collyer 2007; Adams and Collyer 2009; Turner et al. 2010; Collyer and Adams 2013).

This function is a modified version of pairwise, retaining the least squares (LS) means as trajectory points. Analysis starts with a lm.rrpp fit (but a procD.lm fit from geomorph can also be used). LS means are calculated using a grouping variable. Data can be trajectories, as a start(sensu Adams and Cerney 2007), or trajectories can be calculated from data using a factorial model (in which case trajectory points are defined by factor levels).

This function produces statistics that can be summarized with the summary.trajectory.analysis function. The summaries are consistent with those in the summary.pairwise function, pertaining to trajectory attributes including, magnitude difference (MD), the difference in path lengths of trajectories; trajectory correlations (TC), better thought of as angular differences between trajectory principal axes; and if trajectories have three or more points, shape difference (SD), the square root of summed squared point differences, after scaling, centering, and rotating trajectories. The SD is the "Procrustes" distance between trajectories (Adams and Collyer 2009), much the same way as the shape difference between anatomical landmark configurations in geometric morphometrics. If attribute = "TC" is chosen for the summary, then the angle type ("rad" or "deg", can be chosen for either radians and degrees, respectively, to return angles between principal axes.)

Plotting can be performed with plot.trajectory.analysis and add.trajectories. The former plots all principal component scores for the data, and allows point-by-point control of plot parameters. The later adds trajectories points and lines, with constrained control. By saving the plot.trajectory.analysis object, plotting information can be retained and advanced plotting can be performed. See examples below.

Value

An object of class "trajectory.analysis" returns a list of the following:

LS.means	LS.means from pairwise function.
trajectories	Trajectories from every permutation.
PD	Path distances of trajectories from every permutation.
MD	Magnitude differences between trajectories from every permutation.
TC	Trajectory correlations from every permutation.
SD	Trajectory shape differences from every permutation.

Author(s)

Dean Adams and Michael Collyer

References

Adams, D. C., and M. M. Cerney. 2007. Quantifying biomechanical motion using Procrustes motion analysis. J. Biomech. 40:437-444.

Adams, D. C., and M. L. Collyer. 2007. The analysis of character divergence along environmental gradients and other covariates. Evolution 61:510-515.

Adams, D. C., and M. L. Collyer. 2009. A general framework for the analysis of phenotypic trajectories in evolutionary studies. Evolution 63:1143-1154.

Collyer, M. L., and D. C. Adams. 2007. Analysis of two-state multivariate phenotypic change in ecological studies. Ecology 88:683-692.

Collyer, M. L., and D. C. Adams. 2013. Phenotypic trajectory analysis: comparison of shape change patterns in evolution and ecology. Hystrix 24: 75-83.

Collyer, M.L., D.J. Sekora, and D.C. Adams. 2015. A method for analysis of phenotypic change for phenotypes described by high-dimensional data. Heredity. 115:357-365.

Examples

### Analysis of sexual dimorphism vectors (factorial approach)
data(Pupfish)
fit <- lm.rrpp(coords ~ Pop * Sex, data = Pupfish, iter = 199)
reveal.model.designs(fit)
TA <- trajectory.analysis(fit, groups = Pupfish$Pop, 
traj.pts = Pupfish$Sex, print.progress = FALSE)

# Magnitude difference (absolute difference between path distances)
summary(TA, attribute = "MD") 

# Correlations (angles) between trajectories
summary(TA, attribute = "TC", angle.type = "deg") 

# No shape differences between vectors
summary(TA, attribute = "SD") 

# Retain results
TA.summary <- summary(TA, attribute = "MD")
TA.summary$summary.table

# Plot results
TP <- plot(TA, pch = as.numeric(Pupfish$Pop) + 20, bg = as.numeric(Pupfish$Sex),
cex = 0.7, col = "gray")
add.trajectories(TP, traj.pch = c(21, 22), start.bg = 1, end.bg = 2)
legend("topright", levels(Pupfish$Pop), pch =  c(21, 22), pt.bg = 1)

### Analysis when data are already trajectories (motion paths)

# data are planar Cartesian coordinates (x, y) across 5 points (10 variables)
data(motionpaths)
fit <- lm.rrpp(trajectories ~ groups, data = motionpaths, iter = 199)
TA <- trajectory.analysis(fit, groups = motionpaths$groups, traj.pts = 5)

# Magnitude difference (absolute difference between path distances)
summary(TA, attribute = "MD") 

# Correlations (angles) between trajectories
summary(TA, attribute = "TC", angle.type = "deg") 

# Shape differences between trajectories 
summary(TA, attribute = "SD") 

TP <- plot(TA, pch = 21, bg = as.numeric(motionpaths$groups),
cex = 0.7, col = "gray")
add.trajectories(TP, traj.pch = 21, traj.bg = 1:4)

RRPP documentation built on Aug. 16, 2023, 1:06 a.m.