1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
|
\name{Moran.I}
\alias{Moran.I}
\title{Moran's I Autocorrelation Index}
\usage{
Moran.I(x, weight, scaled = FALSE, na.rm = FALSE,
alternative = "two.sided")
}
\arguments{
\item{x}{a numeric vector.}
\item{weight}{a matrix of weights.}
\item{scaled}{a logical indicating whether the coefficient should be
scaled so that it varies between -1 and +1 (default to
\code{FALSE}).}
\item{na.rm}{a logical indicating whether missing values should be
removed.}
\item{alternative}{a character string specifying the alternative
hypothesis that is tested against the null hypothesis of no
phylogenetic correlation; must be of one "two.sided", "less", or
"greater", or any unambiguous abbrevation of these.}
}
\description{
This function computes Moran's I autocorrelation coefficient of
\code{x} giving a matrix of weights using the method described by
Gittleman and Kot (1990).
}
\details{
The matrix \code{weight} is used as ``neighbourhood'' weights, and
Moran's I coefficient is computed using the formula:
\deqn{I = \frac{n}{S_0} \frac{\sum_{i=1}^n\sum_{j=1}^n w_{i,j}(y_i -
\overline{y})(y_j - \overline{y})}{\sum_{i=1}^n {(y_i -
\overline{y})}^2}}{\code{I = n/S0 * (sum\{i=1..n\} sum\{j=1..n\} wij(yi - ym))(yj - ym)
/ (sum\{i=1..n\} (yi - ym)^2)}}
with
\itemize{
\item \eqn{y_i}{yi} = observations
\item \eqn{w_{i,j}}{wij} = distance weight
\item \eqn{n} = number of observations
\item \eqn{S_0}{S0} = \eqn{\sum_{i=1}^n\sum_{j=1}^n wij}{\code{sum_{i=1..n} sum{j=1..n} wij}}
}
The null hypothesis of no phylogenetic correlation is tested assuming
normality of I under this null hypothesis. If the observed value
of I is significantly greater than the expected value, then the values
of \code{x} are positively autocorrelated, whereas if Iobserved <
Iexpected, this will indicate negative autocorrelation.
}
\value{
A list containing the elements:
\item{observed}{the computed Moran's I.}
\item{expected}{the expected value of I under the null hypothesis.}
\item{sd}{the standard deviation of I under the null hypothesis.}
\item{p.value}{the P-value of the test of the null hypothesis against
the alternative hypothesis specified in \code{alternative}.}
}
\references{
Gittleman, J. L. and Kot, M. (1990) Adaptation: statistics and a null
model for estimating phylogenetic effects. \emph{Systematic Zoology},
\bold{39}, 227--241.
}
\author{Julien Dutheil \email{dutheil@evolbio.mpg.de} and
Emmanuel Paradis}
\seealso{\code{\link{weight.taxo}}}
\examples{
tr <- rtree(30)
x <- rnorm(30)
## weights w[i,j] = 1/d[i,j]:
w <- 1/cophenetic(tr)
## set the diagonal w[i,i] = 0 (instead of Inf...):
diag(w) <- 0
Moran.I(x, w)
Moran.I(x, w, alt = "l")
Moran.I(x, w, alt = "g")
Moran.I(x, w, scaled = TRUE) # usualy the same
}
\keyword{models}
\keyword{regression}
|
