Is the interaction significant? A significant interaction would preclude the global
interpretation of the tests of the main factors since it would indicate that the effect
of a factor depends on the level of the other factor.
# Test the main factor ele. The factor pH and the interaction
# are assembled to form the matrix of covariables.
factor.ele.rda ele.pH.helm[, 1:2],
ele.pH.helm[, 3:8])
anova(factor.ele.rda,
permutations = how(nperm = 999),
strata = pH.fac
)
Is the factor ele significant?
# Test the main factor pH. The factor ele and the interaction
# are assembled to form the matrix of covariables.
factor.pH.rda ele.pH.helm[, 3:4],
ele.pH.helm[, c(1:2, 5:8)])
anova(factor.pH.rda,
permutations = how(nperm = 999),
strata = ele.fac
)
Is the factor pH significant?
# RDA with the significant factor ele
ele.rda.out <- rda(spe.hel[1:27, ] ~ ., as.data.frame(ele.fac))
# Triplot with "wa" sites related to factor centroids, and species
# arrows
plot(ele.rda.out,
scaling = 1,
display = "wa",
main = "Multivariate ANOVA, factor elevation - scaling 1 -
wa scores")
6.3 Redundancy Analysis (RDA)
241