Agriculture Reference
In-Depth Information
Since the PCA showed patterns among fi elds for the FAME and soil
enzyme data, a constrained ordination technique was then used to evaluate
relationships between these data and soil physicochemical factors through
redundancy analysis (RDA). RDA combines regression and PCA and al-
lows the direct analysis of how a set of response variables is structured by
a set of explanatory variables (Borcard et al., 2011). RDA constrains ordi-
nation axes to be linear combinations of explanatory variables. Soil NH
4
+
,
NO
3
−
, and Olsen P were ln(x + 1) transformed to help correct positive
skewing and all variables were standardized prior to analysis. Forward
selection of soil physicochemical factors (i.e. 15 variables: soil C and N,
soil C:N, clay, silt, sand, pH, Olsen P, MBC, MBN, EOC, EON,
+
4
NH-N,
−
3
NO-N, and GWC) was performed independently for each set of response
variables (either enzymes or FAMEs) to derive a parsimonious set of ex-
planatory variables based on a double stopping rule of both alpha level (p
< 0.05) and adjusted R
2
(Blanchet et al., 2008). RDA was performed with
the
rda()
function in the
vegan
package.
Canonical variation partitioning (Borcard et al., 1992) was used to deter-
mine the relative importance of soil physicochemical properties and FAMEs
in explaining variation in soil enzyme activities using adjusted R
2
values to
obtain unbiased estimates (Peres-Neto et al., 2006). Soil factors were the
same as used in RDA, as identifi ed in the forward selection procedure. The
same selection procedure was used to derive a parsimonious set of indicator
FAMEs (mol percent) to explain enzyme activities. The analysis was per-
formed in R using the
varpart()
function in the
vegan
package. Signifi cance
of the fractions (i.e. explained fractions of variation accounted for by the
sum of the canonical axes) was tested by partial redundancy analyses and
permutational signifi cance tests (1000 permutations).
13.3 RESULTS
13.3.1 SOIL PROPERTIES AND C AND N POOLS
The 13 organically-managed Roma-type tomato fields had similar soil tex-
ture; measurements classified three fields as loams and ten as silt loams
(Table 2). Clay content ranged from 9.7 to 21.4% and had a coefficient of
