Environmental Engineering Reference
In-Depth Information
other, and subsequent divisions can lead all the way down to individual sites or
some pre-defined stopping value. The resulting model allows the user to determine,
given the values of different environmental variables, the type of plant community
likely to occur in a site or whether the site could be suitable habitat for a species.
Using this technique, relationships sometimes emerge that are otherwise difficult to
detect using other linear or even multivariate models. For more information, see
De'ath and Fabricius (
2000
) and McCune and Grace (
2002
).
5.4.7 Mantel Test
A Mantel test is simply a method of correlating two similarity or distance matrices
with each other. It is also especially helpful when evaluating the effect of spatial
proximity on plant community similarity or the strength of plant community -
environment relationships. For example, a goal of a project may be to determine
whether plant communities respond to wetland disturbance or restoration in a
similar fashion to macroinvertebrate communities across a set of wetlands. With
the Mantel test, the similarity in response can be compared by correlating the plant
and macroinvertebrate distance matrices. A Mantel test can also be used to assess
the significance of the correlation between geographic distance and community
distance (McCune and Grace
2002
).
5.4.8
Indicator Species Development and Analysis
Although wetland plants can be used successfully as indicators of wetland health
(U.S. EPA
2002
) and wetland status (Environmental Laboratory
1987
), this section
does not focus solely on those particular applications. Indicator species analysis
(Dufrˆne and Legendre
1997
) is a mathematical technique that can determine
indicators for different groups of sites or plant communities. Therefore, it can be
used to develop wetland condition indicators, but that is not its sole purpose.
Once groups have been established either
a priori
or using the techniques described
above, indicator species analysis determines how faithful a given species is to a
particular group (whether it is always present), and how exclusive the species is to
the group (never occurring in other groups, (McCune and Grace
2002
)). A species
abundance or presence data set is input, and the output is a table of indicator values
(percent of perfect indication, with 100 % being perfect), and an associated
P
-value
based on a Monte Carlo (randomization) test with a null hypothesis of no difference
between groups. The R package labdsv (Roberts
2010
) and PC-ORD (McCune and
Mefford
2011
) both calculate indicator species values.
Indicator species analysis has been used to better describe plant community
groups (Rooney and Bayley
2011
), differentiate wetlands invaded by non-native
plant species (Johnson et al.
2010
), and associate plant species with different
