xdistance {ecodist}

Cross-distance between two datasets.

Description

Pairwise dissimilarity calculation between rows of one dataset and rows of another, for instance across different sampling periods for the same set of sites.

Usage

xdistance(x, y, method = "euclidean")

Arguments

x

A site by species or other matrix or data frame.

y

A a second site by species dataset, which must have at least the same columns.

method

This function calls distance to do the calculations, and will accept any symmetric method used there, currently: "euclidean", "bray-curtis", "manhattan", "mahalanobis" (squared Mahalanobis distance), "jaccard", "sorensen", "gower", "modgower10" (modified Gower, base 10), "modgower2" (modified Gower, base 2). Partial matching will work for selecting a method. The asymmetric "difference" method will not work for calculating cross-distances.

Details

This function will calculate rowwise dissimilarities between any pair of matrices or data frames with the same number of columns. Note that the cross-dissimilarity functions are for research purposes, and are not well-tested.

Value

A non-symmetric and possibly not square matrix of dissimilarities of class xdist, where result <- xdistance(x, y) produces a matrix with result[a, b] containing the dissimilarity between x[a, ] and y[b, ].

Author(s)

Sarah Goslee

See Also

distance, xmantel, xmgram

Examples

data(graze)

### EXAMPLE 1: Square matrices

# take two subsets of sites with different dominant grass abundances
# use cut-offs that produce equal numbers of sites
dom1 <- subset(graze, POPR > 50 & DAGL < 20) #  8 sites
dom2 <- subset(graze, POPR < 50 & DAGL > 20) #  8 sites

# first two columns are site info
dom.xd <- xdistance(dom1[, -c(1,2)], dom2[, -c(1,2)], "bray")

# environmental and spatial distances; preserve rownames
forest.xd <- xdistance(dom1[, "forestpct", drop=FALSE], 
    dom2[, "forestpct", drop=FALSE])
sitelocation.xd <- xdistance(dom1[, "sitelocation", drop=FALSE], 
    dom2[, "sitelocation", drop=FALSE])

# permutes rows and columns of full nonsymmetric matrix
xmantel(dom.xd ~ forest.xd)
xmantel(dom.xd ~ forest.xd + sitelocation.xd)

plot(xmgram(dom.xd, sitelocation.xd))


### EXAMPLE 2: Non-square matrices

# take two subsets of sites with different dominant grass abundances
# this produces a non-square matrix

dom1 <- subset(graze, POPR > 45 & DAGL < 20) # 13 sites
dom2 <- subset(graze, POPR < 45 & DAGL > 20) #  8 sites

# first two columns are site info
dom.xd <- xdistance(dom1[, -c(1,2)], dom2[, -c(1,2)], "bray")

# environmental and spatial distances; preserve rownames
forest.xd <- xdistance(dom1[, "forestpct", drop=FALSE], 
    dom2[, "forestpct", drop=FALSE])
sitelocation.xd <- xdistance(dom1[, "sitelocation", drop=FALSE], 
    dom2[, "sitelocation", drop=FALSE])

# permutes rows and columns of full nonsymmetric matrix
xmantel(dom.xd ~ forest.xd, dims=c(13, 8))
xmantel(dom.xd ~ forest.xd + sitelocation.xd, dims=c(13, 8))

plot(xmgram(dom.xd, sitelocation.xd))