Package indicspecies, which was produced as a companion package to the
De Cáceres and Legendre (2009) paper, computes various indicator species indices,
including IndVal (or, actually, the square root of IndVal). Two especially interesting
features of this package are the possibility of pooling two or more groups of a
typology in turn to look for species that may be indicators of pooled groups
(in function multipatt()) and the computation of bootstrapped confidence
intervals around indicator values (argument nboot of function strassoc()).
Let us first compute the same analysis as above, using function multipatt()
this time and taking advantage of the possibility of pooling the groups by two.
# Indval with indicspecies::multipatt() with search for indicator
# species of pooled groups
(iva2 <- multipatt(
spe,
grps,
max.order = 2,
control = how(nperm = 999)
))
In function multipatt(), the default indicator measure is func ¼ IndVal.
g. The “.g” indicates that the measure is corrected for unequal group sizes. This
corresponds to the original Dufrêne and Legendre (1997) IndVal measure and is
recommended.
The output object contains several matrices:
• comb ¼ matrix of the belonging of the sites to all possible combinations up to the
order requested in the analysis;
• str ¼ the “strength of association”, i.e., the value of the chosen index;
• A ¼ if the function is the IndVal index, value of its A component (specificity);
otherwise NULL;
• B ¼ if the function is the IndVal index, value of its B component (fidelity);
otherwise NULL;
• sign ¼ results of the best patterns, i.e., the group or combination of groups
where the statistic is the highest, and permutation test results for that result. No
correction for multiple testing is provided.
Here again, the p-values should be corrected for multiple testing:
(pval.adj2 <- p.adjust(iva2$sign$p.value))
One can also request a simpler output with the summary() function:
summary(iva2, indvalcomp = TRUE)
The summary displays the groups and combinations of groups (if any has been
requested) with significant indicator species. In each of these groups it displays the
species’ IndVal value (stat) and the p-value of the permutational test. With argument
4.11 Indicator Species
123
De Cáceres and Legendre (2009) paper, computes various indicator species indices,
including IndVal (or, actually, the square root of IndVal). Two especially interesting
features of this package are the possibility of pooling two or more groups of a
typology in turn to look for species that may be indicators of pooled groups
(in function multipatt()) and the computation of bootstrapped confidence
intervals around indicator values (argument nboot of function strassoc()).
Let us first compute the same analysis as above, using function multipatt()
this time and taking advantage of the possibility of pooling the groups by two.
# Indval with indicspecies::multipatt() with search for indicator
# species of pooled groups
(iva2 <- multipatt(
spe,
grps,
max.order = 2,
control = how(nperm = 999)
))
In function multipatt(), the default indicator measure is func ¼ IndVal.
g. The “.g” indicates that the measure is corrected for unequal group sizes. This
corresponds to the original Dufrêne and Legendre (1997) IndVal measure and is
recommended.
The output object contains several matrices:
• comb ¼ matrix of the belonging of the sites to all possible combinations up to the
order requested in the analysis;
• str ¼ the “strength of association”, i.e., the value of the chosen index;
• A ¼ if the function is the IndVal index, value of its A component (specificity);
otherwise NULL;
• B ¼ if the function is the IndVal index, value of its B component (fidelity);
otherwise NULL;
• sign ¼ results of the best patterns, i.e., the group or combination of groups
where the statistic is the highest, and permutation test results for that result. No
correction for multiple testing is provided.
Here again, the p-values should be corrected for multiple testing:
(pval.adj2 <- p.adjust(iva2$sign$p.value))
One can also request a simpler output with the summary() function:
summary(iva2, indvalcomp = TRUE)
The summary displays the groups and combinations of groups (if any has been
requested) with significant indicator species. In each of these groups it displays the
species’ IndVal value (stat) and the p-value of the permutational test. With argument
4.11 Indicator Species
123