Excess PCR primers were removed
using QIAquick PCR purification kit (Qiagen). 3 l digested DNA was
mixed with 37 l sample loading solution (Beckman Coulter) and
fragment analysis was carried out using Beckman CEQ 8000 capillary
electrophoresis, using 640 bp standard (Beckman Coulter). To
minimise false reporting of background noise the cut-off level for
peak recognition was kept at the default of 10%. As this kind of analysis
is known to produce “shoulder peaks” around the main product
the amoA T-RFLP profiles were edited by eye. However because
many of the peaks in the 16S rDNA profiles were so close together
these shoulder peaks could not be confidently discriminated from
actual peaks so this data was left unedited. T-Align (Smith et al.,
2005) was used to assign terminal restriction fragments (TRFs) to
groups. Multivariate statistical package (Kovach computing services)
was used to carry out principal component analysis on the
T-RFLP data. In order to avoid the over-representation of changes in
rare operational taxonomic units (OTUs) and to make the data suitable
for linear analysis by PCA, the relative abundance data obtained
from the T-RFLP was transformed using a chord transformation
(Ramette, 2007). The scores of the first two principal components
were used to compare differences between the T-RFLP profiles.