nco: NCO, Nonparametric constrained ordination

In phytomosaic/ngnn: ngnn: Nonlinear Gradient Nearest Neighbors

Description

Constrained ordination based on nonparametric regression and NMS.

Usage

nco(obj, method, thresh, ...)

## S3 method for class 'nco'
summary(obj, ...)

## S3 method for class 'nco'
plot(obj, type = "points", cexn = NULL, ...)

## S3 method for class 'ncopredict'
summary(obj, ...)

Arguments

obj	object of class 'npmr' from call to npmr
method	distance measure for all ordinations
thresh	numeric threshold for stepacross dissimilarities
...	additional arguments passed to function
type	either 'points' or 'text' for plotting
cexn	expansion factor for points and text

Details

Combines existing algorithms in multivariate workflow:

NPMR + NMS = NCO

NCO (McCune and Root 2012; McCune and Root 2017) is nonmetric multidimensional scaling ordination (NMS; Kruskal 1964) of sample units based on environmentally constrained fitted values from a set of nonparametric multiplicative regressions (NPMR; McCune 2006) for each species. This implementation uses npreg from package 'np' (NPMR stage) and metaMDS from package 'vegan' (NMS stage).

Variance explained R2_enviro measures the strength of the relationship between the original community data and the ordination as constrained to the *measured* predictors. Like all constrained ordinations, NCO does not account for variation in species composition related to *unmeasured* predictors.

Value

List of class 'nco' with elements:

• scr_i = environmentally constrained scores from NCO

• NCO_model = the NCO model itself

• R2_internal = squared correlation of Dhat and Dz

• R2_enviro = squared correlation of D and Dz

• R2_partial = squared correlation of Dz and each predictor

• Axis_tau = rank correlation of each axis and predictor

References

Kruskal, J. B. 1964. Multidimensional scaling by optimizing goodness of fit to a nonmetric hypothesis. Psychometrika 29: 1-27.

McCune, B. 2006. Non-parametric habitat models with automatic interactions. Journal of Vegetation Science 17(6):819-830.

McCune, B., and H. T. Root. 2012. Nonparametric constrained ordination. 97th ESA Annual Meeting. Ecological Society of America, Portland, OR.

McCune, B., and H. T. Root. 2017. Nonparametric constrained ordination to describe community and species relationships to environment. Unpublished ms.

Examples

# set up
set.seed(978)
require(vegan)
data(varespec, varechem)
spe <- varespec ; id  <- varechem
i   <- sample(1:nrow(spe), size=floor(0.75*nrow(spe))) # sample
spe <- spe[i, ]          # in-sample species
idi <- id[i, ]           # in-sample predictors
ido <- id[-i, ]          # out-of-sample predictors
nm  <- c('Al', 'K')      # select 1 or 2 gradients of interest

# NPMR basic usage
res_npmr <- npmr(spe, idi, ido, nm, nmulti=5)
summary(res_npmr)
plot(res_npmr, pick=1:9, nm=nm)

# NCO basic usage
res_nco <- nco(res_npmr, method='bray', thresh=0.90)
summary(res_nco)

# plot the NCO gradient space
plot(res_nco)
plot(res_nco, type='text')

phytomosaic/ngnn documentation built on May 9, 2019, 5:57 a.m.