126 C.D. Clegg et al. Applied Soil Ecology 14 2000 125–134
plants to more recalcitrant ligno-cellulose complexes which constitute the plants structural material. There
is variation in the composition of these components from different plant species, with bacterial selec-
tion occuring in the rhizosphere Lemanceau et al., 1995; Mahaffee and Kloepper, 1997; Siciliano et al.,
1998.
One of the difficulties in attempting to understand relationships between plant and microbial community
structure is that of description. It is relatively easy to qualify and quantify vegetation composition, whilst it
is much more difficult to analyse the composition of soil microbial communities. Traditional culture-based
measurements provide both a limited and selective representation of the total community as only 2–4 of
bacteria may be readily isolated from soil Olsen and Bakken, 1987. Phenotypic diversities of microbial
populations in soils have been profiled by measur- ing potential function using BIOLOG Garland and
Mills, 1991, phospholipid fatty acid PLFA analysis Frostegärd et al., 1996 and also fatty acid methyl
ester FAME analysis Cavigelli et al., 1995. Molec- ular techniques have also been applied to the study
of microbial populations in environmental samples. These have tended towards the application of specific
oligonucleotide probes directed at DNA Guo et al., 1997, rRNA e.g. Stahl et al., 1988; Kämpfer et al.,
1996 or to rRNA sequence analysis of microbial com- munities e.g. Giovannoni et al., 1990; Stackebrandt et
al., 1993. In a more wide ranging approach, reassoci- ation kinetics of DNA extracted from soils have sug-
gested that as many as 10,000 different bacterial types may be present in 100 g soil Torsvik et al., 1996.
Such a broad-scale approach based on the analysis of community DNA may offer a useful level of reso-
lution for plant:microbe community analyses. Com- munity DNA can be analysed by techniques which
provide different but complementary information about the overall genetic structure of the community.
These include percent guanine+cytosine G+C profiling Holben and Harris, 1995; Griffiths et al.,
1997; Clegg et al., 1998, community cross hybridis- ation Lee and Fuhrman, 1990; Griffiths et al., 1996;
Clegg et al., 1998 and reassociation kinetics Torsvik et al., 1990; Ritz et al., 1997; Clegg et al., 1998,
denaturing gradient gel electrophoresis DGGE Muyzer et al., 1993 and terminal restriction length
polymorphism Liu et al., 1997. In this paper we report on G+C profiling and
community DNA cross hybridisation to determine the genetic composition of soil microbial communities
under upland grasslands of characteristic vegetation types from three geographically distinct sites in the
UK. A central hypothesis in this study was that there is coherence between the structure of plant communi-
ties and their associated soil microbial assemblages. By comparing community structure between quadrats,
we also examined the degree of spatial variation in community structure within these grasslands.
2. Materials and Methods
2.1. Soils and sampling regimes Soil samples were taken from each of three sites
in the UK, namely Garrigill National Grid refer- ence NY35 761 387, Aber SH23 651 723 and
Sourhope NT36 850 205. Within each of the sites were three grassland types characterised by their
vascular composition as described in the National Vegetation Classification Rodwell, 1991 hereafter
referred to as unimproved typically Agrostis, Fes- tuca dominated, semi-improved Holcus dominated
and improved Lolium dominated Table 1. Soil samples were collected by removing 50 random
cores 3 cm diameter, 5 cm deep from each of three
Table 1 National vegetation classification and pH of soils
Site Grassland
National vegetation classification
a
pH Garrigill
Improved MG6
5.9 Semi-improved
U4b 6.0
Unimproved U4a
5.1 Aber
Improved MG6
5.8 Semi-improved
U4b 4.7
Unimproved U4a
4.4 Sourhope
Improved SL
6.4 Semi-improved
SH 5.2
Unimproved SAF
5.6
a
According to Rodwell 1991. Improved: MG6, SL — typi- cally Lolium, Cynosurus dominated; semi-improved: U4b, SH —
typically Holcus, dominated; unimproved: U4a, SAF — typically Agrostis, Festuca, Gallium dominated.
C.D. Clegg et al. Applied Soil Ecology 14 2000 125–134 127
replicate 5 m×5 m quadrats within each vegetation sequence. Each set of 50 cores as pooled, sieved
through a 2 mm mesh, and stored at −20
◦
C prior to use.
2.2. DNA extraction from soils Total community DNA was extracted from soils by
a direct extraction method and purified as described previously Clegg et al., 1997. Yields of extracted
DNA were determined by the diphenylamine reaction Lichtenstein and Draper, 1985. DNA purity was de-
termined after taking absorbance readings at 230, 260 and 280 nm. Values of 1.8 and 2.0 for the absorbance
ratios taken at 260:280 and 260:230 nm respectively, are indicative of pure DNA Sambrook et al., 1989.
2.3. Thermal denaturation of DNA Melting profiles of DNA samples were determined
by placing approximately 1.5 mg of DNA in 650 ml of 0.1×standard saline citrate SSC: Sambrook et
al., 1989 in a 1 cm light path quartz cuvette which was heated at 1
◦
C min
− 1
. Absorbance at 260 nm was recorded every 6 s using a Perkin Elmer Lambda 2S
spectrophotometer. Triplicate aliquots of each sample were heated thus, along with a standard solution of
E. coli DNA Sigma to check on the consistency of heating between runs. Individual melt profiles
were normalised to a scale of 0–100, whereby the largest hyperchromic shift was set at 100 the ab-
sorbance at 90
◦
C and the lowest shift set at 0 the absorbance at 45
◦
C. The profiles were param- eterised by fitting a general logistic equation to the
data between 60–90
◦
C, and transforming the resul- tant curves to G+C profiles Ritz et al., 1997. The
curve parameters can be related to microbial genetic community characteristics in terms of the G+C
distribution as follows: curve parameter ‘m’ relates to the median G+C; parameter ‘b’ gradient of the
slope indicates whether G+C bases are distributed around a narrow range i.e. the slope is steeper when
there is an abundance of DNA with a similar G+C content, or more uniformly across the entire range;
and ‘t’ relates to the skewness of the distribution, i.e. whether there is a preponderance of low G+C
over high, or vice versa. The melting profiles are de- scribed mathematically by the three curve parameters,
thus significant differences in any of the parame- ters indicates that the melting profiles are different.
The curve parameters were analysed by analysis of variance.
2.4. Cross hybridisation of soil DNA The community DNA cross-hybridisation assay ap-
plied was similar to that described previously Grif- fiths et al., 1996. Briefly, target DNA was prepared
by denaturing at 90–95
◦
C for 5–10 min followed by rapid cooling on ice. Three replicates of 500 ng of the
DNA were dot blotted onto Hybond-N nylon mem- branes Amersham International, UK. The wells in
the dot blot apparatus BioRad and DNA spots were washed with 50 ml of 5×SSC. Membranes were air
dried before DNA was crosslinked by UV irradiation and baking for 30 min at 80
◦
C. To generate the probes, 200 ng of soil microbial community DNA was ini-
tially digested with the restriction enzyme Rsa1 Gib- coBRL to provide an increased hybridisation signal
Griffiths et al., 1996. DNA probes were fluorescein labelled using a random primer labelling kit Amer-
sham International, UK. The probe generation reac- tion was incubated overnight at 30
◦
C and yield was determined according to the manufacturers instruc-
tions. Hybridisation reactions were carried out using the hybridisation buffer recommended by the manu-
facturer Amersham International, UK. Membranes were allowed to prehybridise for 1 h at 60
◦
C. The hybridisation buffer was then removed and replaced
with fresh hybridisation buffer at 0.3 ml cm
− 2
mem- brane. Heat denatured labelled probe was added to
the hybridisation buffer at 10 ng ml
− 1
and allowed to hybridise at 60
◦
C for 20–24 h. Membranes were washed and a luminescent signal generated accord-
ing to manufacturers instructions. Intensity of the hy- bridisation signals on the membranes was determined
by measuring light output from each dot using a lu- minometer Dynotech, California. Similarity indices,
S, of the DNA samples were calculated as previously described Lee and Fuhrman, 1990; Griffiths et al.,
1996. Briefly, S=R
s
− R
b
R
c
− R
b
×100, where R
s
is the signal from the sample target spot, R
c
is the signal from the control spot i.e. same DNA as probe
and R
b
is the background signal from bank spots. Mean values and 95 confidence intervals were cal-
128 C.D. Clegg et al. Applied Soil Ecology 14 2000 125–134
culated using a bootstrap procedure Griffiths et al., 1996. If the target and probe DNA have the same
genetic composition then the reciprocal cross of the target-probe hybridisation will give the same S value.
If one DNA sample is more diverse i.e. has a greater range of genetic sequence types than the other, then
S is not equal and the pair of values are asymmetric Lee and Fuhrman, 1990; Griffiths et al., 1996. The
probe giving the highest S value is the more diverse of the two samples and the ‘true’ degree of similar-
ity is denoted by the lower of the two hybridisation values Lee and Fuhrman, 1990; Ritz and Griffiths,
1994.
The assay was carried out using two classes of DNA sample. One class involved the cross hybridis-
ation of DNA extracted from the individual quadrats within each vegetation type at each site; this tested for
the degree of similarity in community DNA between quadrats within vegetation types. A second class tested
Table 2 Yields and purities of DNA extracted from grassland soils
Site Grassland
Quadrat Yield mg g
− 1
dry soil Abs. ratio 260:280
Abs. ratio 260:230 Garrigill
Improved 1
6.30 1.52
1.76 MG6
2 13.6
1.32 2.02
3 35.7
1.35 2.21
Semi-improved 1
35.7 1.31
1.85 U4b
2 17.1
1.48 2.02
3 21.2
1.46 2.08
Unimproved 1
29.0 1.38
1.96 U4a
2 33.6
1.35 2.13
3 47.5
1.42 2.01
Aber Improved
1 19.4
1.71 1.60
MG6 2
27.3 1.66
1.50 3
23.8 1.85
1.74 Semi-improved
1 32.8
1.61 1.29
U4b 2
26.9 1.38
1.11 3
14.7 1.58
1.16 Unimproved
1 44.1
1.25 0.96
U4a 2
34.9 1.51
1.12 3
45.9 1.66
1.28 Sourhope
Improved 1
20.23 1.98
1.78 MG6
2 19.68
1.96 1.78
3 30.10
1.74 1.50
Semi-improved 1
57.42 1.53
1.19 U4b
2 69.82
1.31 1.06
3 25.50
1.66 1.20
Unimproved 1
14.96 1.74
1.44 U4a
2 19.58
1.58 1.58
3 49.84
1.60 1.23
for similarity between vegetation types within sites, and was based on aliquots of DNA produced by pool-
ing DNA from each of the quadrats within a vegeta- tion type.
2.5. Statistics Results of the various analyses were considered sta-