| Version: | 2.1.2 |
| Date: | 2025-1-27 |
| Title: | Missing Morphometric Data Simulation and Estimation |
| Maintainer: | J. Arbour <jessica.arbour@mtsu.edu> |
| Imports: | gdata, shapes, e1071, pcaMethods, MASS, miscTools, stats, rgl, geomorph |
| Description: | Functions for simulating missing morphometric data randomly, with taxonomic bias and with anatomical bias. LOST also includes functions for estimating linear and geometric morphometric data. |
| License: | GPL-2 | GPL-3 [expanded from: GPL (≥ 2)] |
| Depends: | R (≥ 3.5.0) |
| NeedsCompilation: | no |
| Packaged: | 2025-01-27 20:44:22 UTC; fount |
| Author: | J. Arbour [aut, cre], C. Brown [aut] |
| Repository: | CRAN |
| Date/Publication: | 2025-01-27 22:20:02 UTC |
Missing morphometric data simulation and estimation
Description
LOST includes functions for simulating missing morphometric data randomly, with taxonomic bias and with anatomical bias as described by Brown et al. 2012. This package also includes functions for estimating missing morphometric data based on regression analysis and a function for checking the percentage of missing data in a matrix.
Author(s)
J. Arbour and C. Brown
Maintainer: jessica.arbour@mtsu.edu
References
Arbour, J. and Brown, C. 2014. Incomplete specimens in Geometric Morphometric Analyses. Methods in Ecology and Evolution
Brown, C., Arbour, J. and Jackson, D. 2012. Testing of the Effect of Missing Data Estimation and Distribution in Morphometric Multivariate Data Analyses. Systematic Biology 61(6):941-954.
Estimate missing landmark data
Description
This function provides several options for estimating landmark data (details of which can be found in the references below). The function first alignes the landmarks using Procrustes superimposition (align.missing). Both 2D and 3D coordinates can be accommodated.
Usage
MissingGeoMorph(x, method = "BPCA", original.scale = FALSE)
Arguments
x |
A n* l X 2 matrix (2D data only) or an l X m X n array (2D or 3D data) of coordinate data, where n is the number of specimens and l is the number of landmarks, and m is the number of dimensions. All landmarks from one specimen should be grouped together. Missing values should be given as NA |
method |
Four methods are provided for estimating missing landmark data: 1) "BPCA" - Bayesian principal component analysis, 2) "mean" - mean substitution, 3) "reg" - values are estimated based on the most strongly correlated variable available, and 4) "TPS" - thin plate spline interpolation (only available for 2D). See Arbour and Brown (2014) for a comparison of the performance of each of these methods. |
original.scale |
Rescale and translate the data back to its original size (TRUE) or leave it in the rescaled, superimposed configuration (FALSE) |
Value
Returns an n * l X 2 (or 3) matrix of coordinate data, with missing values imputed. Landmarks have been aligned and are given in the original shape space.
Author(s)
J. Arbour
References
Arbour, J. and Brown, C. 2014. Incomplete specimens in Geometric Morphometric Analyses. Methods in Ecology and Evolution 5(1):16-26.
Brown, C., Arbour, J. and Jackson, D. 2012. Testing of the Effect of Missing Data Estimation and Distribution in Morphometric Multivariate Data Analyses. Systematic Biology 61(6):941-954.
See Also
align.missing, missing.specimens
Procrustes superimposition of landmark datasets with some missing values
Description
This function carries out a generalized procrustes superimposition on all fully complete specimens and produces a consensus configuration (using "Shapes" procGPA). Each incomplete specimen is then individually rotated and aligned with the consensus configuration based on any landmarks are available (using "Shapes" procOPA). Data is returned superimposed.
Usage
align.missing(X)
Arguments
X |
An l X 2 (or 3) X n array of coordinate data, where n is the number of specimens and l is the number of landmarks. |
Value
Returns An l X 2 (or 3) X n array of coordinate data
Author(s)
J. Arbour
References
Arbour, J. and Brown, C. 2014. Incomplete specimens in Geometric Morphometric Analyses. Methods in Ecology and Evolution 5(1):16-26.
See Also
Examples
data(dacrya)
## make some specimens incomplete
dac.miss<-missing.data(dacrya,remsp=0.2,land.vec=c(1,2,3,4,5,6))
## align all specimens
dac.aligned<-align.missing(dac.miss)
Estimate missing morphometric data with a highly correlated variable
Description
Estimates missing morphometric using regression on the most highly correlated morphological variable available
Usage
best.reg(x)
Arguments
x |
A n X m matrix of morphometric data with n specimens and m variables, containing some percentage of missing values input as NA |
Value
Returns a n X m matrix containing both the original morphometric values as well as estimates for all previously missing values.
Author(s)
J. Arbour and C. Brown
References
Brown, C., Arbour, J. and Jackson, D. 2012. Testing of the Effect of Missing Data Estimation and Distribution in Morphometric Multivariate Data Analyses. Systematic Biology 61(6):941-954.
See Also
Align a bilterally symmetric landmark configuration with a plane
Description
Aligns a bilaterally symmetric landmark dataset to a specific plane by minimized the sum of squared distances of one coordinate (x, y or z). Useful for averaging bilateral landmarks or in preparation for correcting for artifacts like bending.
Usage
bilat.align(coords, land.pairs, average = TRUE, restricted = NULL)
Arguments
coords |
Either a matrix or array of landmark data with columns representing the x, y, z coordinates and rows representing landmarks. See details for how this is applied for a single vs. multiple specimens. |
land.pairs |
A 2 column matrix indicating bilaterally paired landmarks. All "left" landmarks should be in the same column (and likewise for "right landmarks") |
average |
An optional term indicating that bilaterally paired landmarks should be mirrored and averaged, leaving only one "side" and the midline landmarks. |
restricted |
A set of row numbers indicating which landmarks should be considered by "optim" when selecting the optimal rotation. Typically landmarks representing a rigid structure if some landmarks represent articulated/moveable features. |
Details
If a matrix for a single specimen's landmarks is provided this is aligned to a plane, if an array of multiple specimens is provided, these should be previously aligned with Procrustes superimposition, and the entire configuration is optimized with a single rotation applied to all specimens. SS are minimized across the third axis (coords[,3] or coords[,3,]).
Value
A matrix or array giving the rotated landmark configuration
Author(s)
J.H. Arbour
References
Arbour,J.H. In Prep. Get Unbent! R Tools for the removal of arching and bending of fish specimens in geometric morphometric shape analysis
See Also
Examples
library(rgl)
data(darters)
## align darter configuration by head landmarks (restricted)
aligned<-bilat.align(darters$coords[,,1],
darters$land.pairs,average=FALSE,darters$restricted)
plot3d(aligned, aspect=FALSE)
Simulate missing morphometric data with taxonomic bias
Description
This function simulates higher frequency of missing data points in groups that are less numerically well represented in the whole sample, relative to other group. These groups may represent taxa (as used in Brown et al., 2012), but may also represent any other group of interest (e.g. populations, trials, subsamples, etc.). From a morphometric dataset, this function first selects a number of specimens to have data points removed from at random. A vector containing the number of measurements to remove from each specimen is sorted into descending order. Specimens are then sampled without replacement with a probability relative to the sum of the entire sample sizes divided by the number of specimens its respective group. The order the specimens are sampled determines the number of data points to be removed (i.e. the first to be sampled has the most removed). A complete mathematical description may be found in Brown et al. (2012).
Usage
byclade(x, remperc , groups)
Arguments
x |
A n X m matrix of morphometric data with n specimens and m variables. Or an l X 2 or 3 X n array of geometric morphometric coordinates (2D or 3D), where l is the number of landmarks. |
remperc |
The percentage of data to be removed from the matrix, expressed as a decimal (ex: 30 percent would be entered as 0.3) |
groups |
A vector of length n specifying taxonomic group membership as integers (ex: c(1,1,2,2,3,3,...) ) |
Value
returns a matrix or array (depending on input) of morphometric data with missing variables input as 'NA'
Author(s)
J. Arbour and C. Brown
References
Brown, C., Arbour, J. and Jackson, D. 2012. Testing of the Effect of Missing Data Estimation and Distribution in Morphometric Multivariate Data Analyses. Systematic Biology 61(6):941-954.
See Also
Remove incomplete specimens from a landmark dataset
Description
This function takes a dataset containing both complete and incomplete specimens and removes all incomplete specimens.
Usage
complete.specimens(dataset, nlandmarks)
Arguments
dataset |
A n* l X 2 matrix of coordinate data, where n is the number of specimens and l is the number of landmarks. All landmarks from one specimen should be grouped together. |
nlandmarks |
The number of landmarks per specimen |
Value
Returns an c * l X 2 matrix of landmark data, where c is the number of complete specimens and l is the number of landmarks.
Author(s)
J. Arbour
References
Arbour, J. and Brown, C. In Press. Incomplete specimens in Geometric Morphometric Analyses. Methods in Ecology and Evolution
See Also
align.missing, MissingGeoMorph
Crocodile morphometrics
Description
A linear morphometric dataset featuring 23 cranial measurements from 223 specimens representing 21 crocodilian species.
Usage
data(crocs)Format
A n X m dataframe, where n is the number of specimens and m is the number of variables.
Source
http://datadryad.org/resource/doi:10.5061/dryad.m01st7p0
References
Brown, C., Arbour, J. and Jackson, D. 2012. Testing of the Effect of Missing Data Estimation and Distribution in Morphometric Multivariate Data Analyses. Systematic Biology 61(6):941-954.
See Also
obliterator, byclade,missing.data,crocs.landmarks
Coordinate data for a crocodilian reference skull
Description
Landmark data for the measurements points on a reference crocodilian skull, for use with the obliterator function
Usage
data(crocs.landmarks)Format
A 6 X m dataframe in which each column gives the start and end points for each cranial measurement in the crocs dataset, from a single reference specimen. 3D Coordinates are listed as x1, x2, y1, y2, z1, z2 in each column.
Source
Brown, C., Arbour, J. and Jackson, D. 2012. Testing of the Effect of Missing Data Estimation and Distribution in Morphometric Multivariate Data Analyses. Systematic Biology 61(6):941-954.
See Also
obliterator, byclade,missing.data,crocs
Landmark data from Guianacara dacrya
Description
Sixteen landmarks taken from the lateral profile of 73 specimens from the Essequibo and rio Branco drainages, used in the description of Guianacara dacrya
Usage
data(dacrya)Format
A 16 X 2 X 73 array of geometric morphometric coordinates
Source
Arbour, J. and Lopez-Fernandez, H. 2011. Guiancara dacrya, a new species from the rio Branco and Essequibo River drainages of the Guiana Shield (Perciformes: Cichlidae). Neotropical Ichthyology 9:87-96.
See Also
align.missing, MissingGeoMorph
Darter landmarks
Description
A 3D landmark dataset from 30 species of darter fishes (Etheostomatinae; Percidae)
Usage
data("darters")Format
The format is: List of 6 $ coords : num [1:220, 1:3, 1:30] -1.458 -0.489 -0.037 1.705 0.959 ... ..- attr(*, "dimnames")=List of 3 .. ..$ : NULL .. ..$ : NULL .. ..$ : chr [1:30] "Etheostoma_caeruleum_mtsu5_58mmsl.stl" "Ammocrypta_beanii_ummz242736_43mm.stl" "Ammocrypta_clara_ummz148570_42.23mm.stl" "Crystallaria_asprella_Ummz211889_60mmSL.stl" ... $ land.pairs:'data.frame': 101 obs. of 2 variables: ..$ left : int [1:101] 1 3 5 7 9 11 13 15 17 19 ... ..$ right: int [1:101] 2 4 6 8 10 12 14 16 18 20 ... $ sliders :'data.frame': 32 obs. of 3 variables: ..$ start: int [1:32] 22 23 24 25 26 27 28 29 31 32 ... ..$ slide: int [1:32] 23 24 25 26 27 28 29 30 32 33 ... ..$ end : int [1:32] 24 25 26 27 28 29 30 31 33 34 ... $ surface :'data.frame': 144 obs. of 1 variable: ..$ surface: int [1:144] 60 61 62 63 64 65 66 68 69 70 ... $ restricted: int [1:58] 1 2 3 4 5 6 7 8 9 10 ... $ reference : num [1:11] 22 99 180 15 16 63 176 81 178 11 ...
Details
Includes landmark coordinates (coords), a matrix indicating bilaterally paired landmarks (land.pairs), curve sliders (sliders), surface sliders (surface), rows of head landmarks (restricted) and landmarks approximating the spine/long axis (reference).
Source
Arbour,J.H. In Prep. Get Unbent! R Tools for the removal of arching and bending of fish specimens in geometric morphometric shape analysis
References
Arbour,J.H. In Prep. Get Unbent! R Tools for the removal of arching and bending of fish specimens in geometric morphometric shape analysis
See Also
unbend.spine,bilat.align,unbend.tps.poly
Examples
data(darters)
library(rgl)
plot3d(darters$coords[,,1], aspect=FALSE)
A-priori size regression for missing data estimation
Description
Estimates missing data using regression on a designated size variable. Any values of the size variable missing are estimated with the variable best correlated with size.
Usage
est.reg(x, col_indep)
Arguments
x |
A n X m matrix of morphometric data with n specimens and m variables, containing some percentage of missing values input as NA |
col_indep |
The number of the column in which the independant size variable is stored. This column will be used to estimate missing values in the other columns. |
Value
Returns a n X m matrix containing both the original morphometric values as well as estimates for all previously missing values.
Author(s)
J. Arbour and C. Brown
References
Brown, C., Arbour, J. and Jackson, D. 2012. Testing of the Effect of Missing Data Estimation and Distribution in Morphometric Multivariate Data Analyses. Systematic Biology 61(6):941-954.
See Also
Reflected Relabelling
Description
This function carries out reflected relabelling to estimate missing geometric morphometric landmarks using bilateral symmetry following Gunz et al 2009.
A set of 3D landmarks are mirrored and aligned with the original data (using procOPA from package "shapes"). Missing landmarks are interpolated from the mirrored specimen.
Usage
flipped(specimen, land.pairs, show.plot = FALSE, axis = 1)
Arguments
specimen |
An l X 3 matrix of coordinate data, where l is the number of landmarks. Some data should be missing and designated with NA. |
land.pairs |
A 2 column matrix, each row should contain row numbers (from matrix specimen) indicating bilateral pairs of landmarks. Unpaired landmarks do not need to be included. See also bilateral symmetry analyses in package "geomorph". |
show.plot |
Optionally plot the specimen using plot3d from rgl. Estimated landmarks are given in red. Defaults to FALSE. |
axis |
Which axis should be mirrored across. Default is x (1). |
Value
Returns a l X 3 matrix of landmarks.
Author(s)
J. Arbour
References
Gunz P., Mitteroecker P., Neubauer S., Weber G., Bookstein F. 2009. Principles for the virtual reconstruction of hominin crania. Journal of Human Evolution 57:48-62.
See Also
Calculate the percentage of missing morphometric data
Description
Calculates the percentage of morphometric data points that have been replaced with 'NA' by functions such as missing.data, byclade or obliterator from LOST. Used to verify the amount of missing data inputted into complete morphometric matrices.
Usage
how.many.missing(x)
Arguments
x |
A n X m matrix of morphometric data with n specimens and m variables, or a or l X 2(or 3) array of geometric morphometric data containing some percentage of missing data |
Value
Returns the percentage (as a decimal) of missing data points present in x
Author(s)
J. Arbour and C. Brown
References
Brown, C., Arbour, J. and Jackson, D. 2012. Testing of the Effect of Missing Data Estimation and Distribution in Morphometric Multivariate Data Analyses. Systematic Biology 61(6):941-954.
See Also
Randomly input missing data points
Description
Randomly replaces a set percentage of data points in a matrix of morphometric measurements with NA to simulate missing data. This is function RMD from Brown et al (2012). The amount of missing data can be chosen as an overall percentage of data (simple morphometric data) or specimens and can be constrained to a set of landmarks (for landmarks).
Usage
missing.data(x, remperc, remsp = NULL, land.vec = NULL, land.identity = NULL)
Arguments
x |
A n X m matrix of morphometric data with n specimens and m variables. Or an array of geometric morphometrics landmarks (l X m X n) |
remperc |
The percentage of data to be removed from the matrix or array, expressed as a decimal (ex: 30 percent would be entered as 0.3) |
remsp |
The percentage of specimens to be removed from the array, expressed as a decimal (ex: 30 percent would be entered as 0.3) |
land.vec |
The number of landmarks to remove per specimen in an array. This can be a single value or vector with unique or repeating values. |
land.identity |
A vector to constrain the landmarks to chose from when assigning missing data. The values correspond to row numbers in an array. |
Value
Returns a n X m matrix or l X m X n array of morphometric data with missing variables input as NA
Author(s)
J. Arbour and C. Brown
References
Brown, C., Arbour, J. and Jackson, D. 2012. Testing of the Effect of Missing Data Estimation and Distribution in Morphometric Multivariate Data Analyses. Systematic Biology 61(6):941-954.
See Also
Examples
data(dacrya)
#### remove 1 to 6 landmarks from 20% of specimens
dac.miss<-missing.data(dacrya,remsp=0.2,land.vec=c(1,2,3,4,5,6))
dac.miss
Simulate incomplete specimens
Description
Randomly selects a pre-determined number of specimens from a landmark dataset (2D or 3D) and removes some of their landmarks.
Usage
missing.specimens(dataset, nspremove, nldremove, nlandmarks)
Arguments
dataset |
A n*l X 2 (or 3) matrix of coordinate data, where n is the number of specimens and l is the number of landmarks. All landmarks from one specimen should be grouped together. |
nspremove |
The number of specimens which should have landmarks removed. |
nldremove |
The number of landmarks to remove per specimen. This may be a single value or a vector of values, none of which can be >nlandmarks. If a vector is given, for each specimen selected, the function will randomly select a value from the vector and remove that many landmarks. |
nlandmarks |
The number of landmarks per specimen |
Value
Returns an n * l X 2 (or 3) matrix with some complete and some incomplete specimens.
Author(s)
J. Arbour
References
Arbour, J. and Brown, C. 2014. Incomplete specimens in Geometric Morphometric Analyses. Methods in Ecology and Evolution 5(1):16-26.
See Also
align.missing, MissingGeoMorph
Simulate missing morphometric data with anatomical bias
Description
This function simulates the effect of proximity between measurements in morphometric data on the distribution of missing values. This attempts to replicate specimens showing regional incompleteness. From a morphometric dataset, this function selects a number of specimens to have data points removed from and a number of measurements to remove from each of these specimens based on a random distribution of missing data. For each specimen, this function randomly selects one starting data point for removal. All subsequent data points have a probability of removal that is proportional to the inverse of the distance to all previously removed data points, based on a reference set of landmarks (matrix 'distances'). For a complete mathematical description see Brown et al. (2012). See function obliteratorGM for the geometric morphometric implementation.
Usage
obliterator(x, remperc, landmarks, expo=1)
Arguments
x |
A n X m matrix of morphometric data with n specimens and m variables |
remperc |
The percentage of data to be removed from the matrix, expressed as a decimal (ex: 30 percent would be entered as 0.3) |
landmarks |
A 6 X m matrix that includes the start and end points (landmarks) for each morphometric measurement from a reference specimen (3D). The data in each column is ordered as x1,x2,y1,y2,z1,z2. See example |
expo |
An optional term for raising the denominator to an exponent, to increase or decrease the severity of the anatomical bias |
Value
Returns a n X m matrix of morphometric data with missing variables input as NA
Author(s)
J. Arbour and C. Brown
References
Brown, C., Arbour, J. and Jackson, D. 2012. Testing of the Effect of Missing Data Estimation and Distribution in Morphometric Multivariate Data Analyses. Systematic Biology 61(6):941-954.
See Also
missing.data,byclade,obliteratorGM
Simulate missing geometric morphometric landmarks with anatomical bias
Description
This is the geometric morphometric implementation of the LOST function obliterator. This attempts to replicate specimens showing regional incompleteness. For each specimen, this function randomly selects one starting data point for removal. All subsequent data points have a probability of removal that is proportional to the inverse of the distance to all previously removed data points, based on the shape of that particular specimen (this differs from the linear morphometric implementation which requires a reference set of coordinates). For a complete mathematical description see Brown et al. (2012).
Usage
obliteratorGM(x, remperc, expo=1)
Arguments
x |
A n X m matrix of morphometric data with n specimens and m variables. Or a l X 2 or 3 X n array of geometric morphometric coordinates, with l being the number of landmarks. |
remperc |
The percentage of data to be removed from the matrix, expressed as a decimal (ex: 30 percent would be entered as 0.3) |
expo |
An optional term for raising the denominator to an exponent, to increase or decrease the severity of the anatomical bias |
Value
Returns a n X m matrix of morphometric data with missing variables input as NA
Author(s)
J. Arbour and C. Brown
References
Brown, C., Arbour, J. and Jackson, D. 2012. Testing of the Effect of Missing Data Estimation and Distribution in Morphometric Multivariate Data Analyses. Systematic Biology 61(6):941-954.
See Also
missing.data,byclade,obliterator
Correct for lateral bending in fish geometric morphometric landmarks
Description
Correct for the impact of lateral bending along the spine of a fish in geometric morphometric landmarks. Fits a polynomial function along the length and width of the specimen, determines the perpendicular residuals and arc length along the polynomial and these are used as the new length and width landmarks. Landmarks are first centered and bilaterally aligned using bilat.align.
Usage
unbend.spine(coords, land.pairs, deg = 3, restricted = NULL)
Arguments
coords |
A matrix of landmark coordinate data. Columns should be coordinates, and rows landmarks. |
land.pairs |
A 2-column matrix giving the bilaterally paired landmarks. One column should be all "left" landmarks and one all "right" landmarks. |
deg |
The degrees for the polynomial function, passed to the function "poly". Typically 2 or 3. |
restricted |
A limited set of landmarks (row numbers for the coords matrix) to use for bilateral alignment. Typically those representing a rigid/fixed structure (e.g., head). Passed to bilat.align. |
Details
Resulting landmark data is in the same scale as the original landmark configuration. Can be applied over multiple specimens using for-loops or apply functions.
Value
bilat.aligned |
Provides the bilaterally aligned landmark data as a matrix |
unbent |
Provides the bilaterally aligned and unbent landmark data as a matrix |
Author(s)
J.H. Arbour
References
Arbour,J.H. In Prep. Get Unbent! R Tools for the removal of arching and bending of fish specimens in geometric morphometric shape analysis
See Also
Examples
data(darters)
library(rgl)
## bilaterally aligned using only head landmarks
lands.unbent<-unbend.spine(darters$coords[,,2],
darters$land.pairs,deg=3, restricted=darters$restricted)$unbent
plot3d(lands.unbent, aspect=FALSE)
TPS-style unbend specimens
Description
Remove dorsoventral arching effect from fish specimen landmark data. Function similar to "unbend specimens" utility in the TPS software suite. Fits a polynomial function along the length and height of the specimen, determines the perpendicular residuals and arc length along the polynomial, and these are used as the new length and width landmarks.
Usage
unbend.tps.poly(coords, reference, axes = NULL, deg = 2)
Arguments
coords |
A matrix or array of landmark coordinate data. Columns should be coordinates, and rows landmarks. Data may be 2D or 3D. If array is given, process will repeat on each slice individually. |
reference |
The rows of the matrix over which the polynomial function will be fit. Should represent the spine or other proxy for the long axis of the body. |
axes |
A vector with 2 values representing the "lateral" view of the fish. The first entry should be the "long" (anterior-posterior) axis and the second should be the vertical (dorso-central) axis. If not provided it is assumed the first column is X and the second is Y. |
deg |
The degrees for the polynomial function, passed to "poly". Typically 2 or 3 (default = 2, quadratic fit). |
Details
It is advisable to remove lateral bending with unbend.spine prior to using this function. Otherwise data should be at least bilaterally aligned to a plane (and seebilat.align) Resulting landmark data is in the same scale as the original landmark configuration. Can be applied over multiple specimens using for-loops or apply functions.
Value
Returns a matrix or array (depending on original data) of landmark data with the effect of dorso-ventral arching removed.
Author(s)
J.H. Arbour
References
Arbour,J.H. In Prep. Get Unbent! R Tools for the removal of arching and bending of fish specimens in geometric morphometric shape analysis
See Also
Examples
library(rgl)
data(darters)
## bilaterally aligned using only head landmarks
lands.unbent<-unbend.spine(darters$coords[,,3],
darters$land.pairs,deg=3, restricted=darters$restricted)$unbent
plot(lands.unbent[,c(1,3)],asp=1)
lands.unbent<-unbend.tps.poly(lands.unbent,darters$reference,axes=c(1,3),deg=3)
plot(lands.unbent[,c(1,2)],asp=1)
plot3d(lands.unbent, aspect=FALSE)