Uploaded by Aisyah Salma Nurfahima

(Q1) Carpenter2012 Article BiodiversityOfSoilMacrofaunaIn

advertisement
Biodivers Conserv (2012) 21:3385–3410
DOI 10.1007/s10531-012-0369-0
ORIGINAL PAPER
Biodiversity of soil macrofauna in the New Forest:
a benchmark study across a national park landscape
Daniel Carpenter • Peter M. Hammond • Emma Sherlock
Angela Lidgett • Kerry Leigh • Paul Eggleton
•
Received: 4 May 2012 / Accepted: 11 September 2012 / Published online: 18 October 2012
Springer Science+Business Media B.V. 2012
Abstract The New Forest National Park is a hotspot for biodiversity in the UK. A long
history of grazing by ponies in the New Forest landscape has created a diverse mosaic of
habitats that are of international significance. We investigated patterns of species diversity
and composition across the New Forest landscape by sampling soil, leaf litter and ground
macrofauna from woodland, grassland and heathland plots across the entire landscape. We
used a spatially replicated design of hand sorted soil pits, Winkler extraction of leaf litter,
and pitfall traps. We concentrated on diversity patterns of the following target groups:
Coleoptera, Formicidae, Isopoda, Chilopoda, Diplopoda, Opiliones and Lumbricidae. The
most striking difference in species assemblages is between wooded and open areas.
Woodlands are the most diverse habitats and have a distinct assemblage, largely due to
those leaf litter invertebrate species which are only present under a closed canopy. Open
areas are less diverse, with diversity particularly low in the wet grasslands. However, the
open areas do have a distinct fauna, especially in the wet and dry heaths which are home to
a number of rare species, particularly of Formicidae. We discuss the potential conservation
problems facing the New Forest and how these might affect soil macrofauna biodiversity in
the future and conclude that climate change; over-grazing; and land use changes represent
the largest threats. Although a relatively stable landscape which benefits from protection
under UK law, changes in grazing intensity or management practices in the New Forest,
particularly for some of the habitats of European importance (e.g. wet heathlands), could
negatively affect soil macrofauna biodiversity. Climate change may also exacerbate biodiversity decline as a result of increased grazing intensity or changes in management.
Electronic supplementary material The online version of this article (doi:10.1007/s10531-012-0369-0)
contains supplementary material, which is available to authorized users.
D. Carpenter (&) P. M. Hammond A. Lidgett K. Leigh P. Eggleton
Soil Biodiversity Group, Life Sciences Department, Natural History Museum, London SW7 5BD, UK
e-mail: d.carpenter@nhm.ac.uk
E. Sherlock
Aquatic Invertebrates Division, Life Sciences Department, Natural History Museum,
London SW7 5BD, UK
123
3386
Biodivers Conserv (2012) 21:3385–3410
Keywords Conservation threats Landscape ecology National parks Terrestrial
macrofauna Soil biodiversity
Introduction
Society’s ability to respond to threats to the natural world is entirely dependent on our
knowledge of how that world is structured. Unfortunately, that knowledge is patchy,
particularly of biodiversity information relevant to conservation and management of entire
landscapes. This deficit is particularly apparent when considering smaller body-sized
taxonomic groups that are species-rich and ecologically important but relatively poorly
known taxonomically. Non-herbivorous invertebrates are a particularly good example
where this knowledge shortfall exists (Stewart and New 2007).
Increasingly, landscapes are becoming the scale at which conservation management
occurs, particularly when considering the effects of human activities, such as fragmentation
(McIntyre and Hobbs 1999). Studies of terrestrial invertebrates at the landscape scale are
relatively rare, concentrating, for example, on hedgerows (Maudsley 2000), heathlands
(Webb 1989), tropical grasslands (Chambers and Samways 1998), and agricultural landscapes (Smith et al. 2008). Complex semi-natural mosaic landscapes are particularly poorly
represented.
The landscape level that has become especially relevant for conservation practice within
the UK is the national park. In the UK a national park is a geographically demarked
protected area containing one or more habitats considered to be of conservation importance. Within such a park biodiversity surveys tend to be extensive but patchy and qualitative (e.g. see papers in Newton 2010a). Often such studies are inventories of areas
known to be of high conservation value, and of taxonomic groups that are charismatic
(e.g. mammals), easily identifiable (e.g. flowering plants) or both (e.g. birds, butterflies).
The studies, although of great value individually, are inadequate for characterising the
overall biodiversity of an entire landscape, and are particularly unsuited to measurements
of hyperdiverse and/or functionally dominant organisms.
Here we document biodiversity patterns of the soil macrofauna across a UK national
park, and assess the ecological and conservation value of the landscape and its components. We examine representative plots across the entire landscape of the New Forest,
southern England, and explore the environmental factors that influence assemblage composition. We use three sampling methods (pitfall traps, Winkler bag extraction of leaf litter
and soil pits) to survey six representative habitats, ranging from ancient woodland to dry
heathland. We examine the patterns not only for their intrinsic importance as representative
of arguably the most biodiverse terrestrial landscape in the UK, but also as a benchmark for
assessing the potential threats to the landscape, its habitats and their macrofauna, through
possible sources of environmental change in the decades to come.
Methods
The New Forest National Park
The New Forest National Park was established in 2005 and covers an area of approximately
56,000 ha. It is an important area for nature conservation as it has the largest area of lowland
heath in southern England (JNCC 2011) and the largest area of pasture woodland in north west
123
Biodivers Conserv (2012) 21:3385–3410
3387
Europe (JNCC 2011). As a result it has been designated a Special Area for Conservation
(SAC) under the European instrument Natura 2000 (European Comission 2012). The New
Forest is a well-established biodiversity hotspot for lichens (Sanderson 2010), fungi (Newton
2010b) and bryophytes (Stern 2010). The national park is an IUCN-designated category V
protected area (‘Protected Landscape’) (Chape et al. 2005). The extraordinary diversity of the
New Forest is in part due to the presence of a grazing livestock, maintained on the forest by
common rights. So important are grazing animals for maintaining the unique landscape of the
New Forest that they have been termed ‘forest architects’ (Tubbs 1997). Their role in the
shaping and maintaining of the forest landscape is unquestionable and vital.
Project design
We defined the entire New Forest National Park perimeter as the landscape within which
we undertook our surveys (Fig. 1). We surveyed 40 one hectare plots (Table 1; for full
details see supplementary Table 1)) across the landscape, split into core and periphery
plots. The core area was divided into five distinct parcels (Fig. 1) and within each parcel
we marked out one 1 hectare plot within each of six habitats (Table 2). The periphery plots
consist of five ancient woodland and five wet heathland plots.
The main synthetic environmental unit of analysis was the habitat, defined using the
Forestry Commission vegetation survey mapping definitions (see below). This gave us six
habitats (ancient woodland, plantation woodland, dry grassland, wet grassland, dry
heathland, wet heathland, see Fig. 2), and two general cover classes—open and wooded.
We also distinguished between core plots (all habitats) and peripheral (ancient woodland
and wet heathland only) plots (Figs. 1, 2).
The specific spatial arrangement of plots and habitats chosen here sought to avoid the
confounding spatial effect of different habitats being sampled in different parts of a large
landscape (Fig. 3). The New Forest is unusually well suited to removing these effects at the plot
selection stage as it is made up of a mosaic of different habitats distributed across the entire
landscape. To take advantage of this we surveyed all six habitats in each of five independent
parcels all within 15 km of each other within the central catchment. This allowed us to partition
out the pure environmental and spatial associations (Borcard et al. 1992) and greatly reduces the
possibility of finding any confounding ‘environmental 9 spatial’ interaction terms.
Data from each sampling method (i.e. each pit, leaf litter and pitfall point sub-samples)
was pooled per plot and used as the smallest unit of statistical analysis in this study, as
employed and discussed in Eggleton et al. (2009). This was necessary to avoid pseudoreplication within plots. In addition, for the soil pits (see below) the numbers per pit
were often too low to allow us to use them as independent data points anyway and so a
more complex split plot design analysis was not possible.
Environmental variables
A survey of the vegetation was made in each of our plots. Sampling was based on 1 m2
quadrats from which the macrofauna were sampled (see below). For each quadrat a record
was made of the plant species present. In the woodlands, these comprised separate records for
species in the canopy, the understorey (up to 2 m tall) and the ground layer, while in the other
habitats only the ground layer species were recorded. Accurate identification was possible for
trees, shrubs and conspicuous forbs but not for grasses, mosses or inconspicuous forbs.
A range of abiotic variables were recorded. Soil temperature was measured using a bi-metal
soil thermometer. Soil moisture was measured using an SM200 probe and an HH2 meter
123
3388
Biodivers Conserv (2012) 21:3385–3410
Fig. 1 Simplified map of New Forest National Park showing the positions of plots. Polygons show the
sampling parcels within the core area, with points outside that area being peripheral plots
123
Biodivers Conserv (2012) 21:3385–3410
3389
Table 1 Details of the forty 1 hectare plots used in this study
Site name
Label
Habitat type
OSGB grid ref
Plain Heath woodland
PHW
Ancient woodland (p)
SZ 21100 99600
Whitten Bottom
WBJ
Wet heathland (p)
SU 20300 00900
Shaves Wood
SWW
Ancient woodland (p)
SU 28800 11900
Rushpole Wood
RWJ
Wet heathland (p)
SU 30400 10100
Tantany Wood
TTW
Ancient woodland (p)
SU 36700 04100
Yew Tree Heath
YTJ
Wet heathland (p)
SU 36300 07100
Bramshaw Wood
BSW
Ancient woodland (p)
SU 26154 16591
Half Moon Common
HMJ
Wet heathland (p)
SU 29700 17400
Red Shoot Wood
RSW
Ancient woodland (p)
SU 19100 08700
Coopers Hill
CHJ
Wet heathland (p)
SU 20500 15200
Whitley Wood
WWW
Ancient woodland
SU 29983 05683
Balmer Lawn
BLD
Dry grassland
SU 30690 03347
Ober Heath
OHH
Dry heathland
SU 28500 04000
New Park Plantation
NPI
Plantation woodland
SU 29296 05408
Balmer Lawn
BLL
Wet grassland
SU 30692 03623
Ober Heath
OHJ
Wet heathland
SU 27903 03535
Berry Wood
BWW
Ancient woodland
SU 21488 05152
Clay Hill
CHD
Dry grassland
SU 23971 02473
Goatspen Plain
GPH
Dry heathland
SU 22238 02226
South Oakley inclosure
SOI
Plantation woodland
SU 22216 04655
Burley Lawn
BUL
Wet grassland
SU 22962 03599
Burley Lawn
BLJ
Wet heathland
SU 23173 03274
Mark Ash wood
MAW
Ancient woodland
SU 24461 07632
Acres Down
ADD
Dry grassland
SU 26600 09100
Pilmore Gate Heath
PGH
Dry heathland
SU 26784 09423
Highland Water inclosure
HWI
Plantation woodland
SU 24558 08473
Millyford Bridge
MBL
Wet grassland
SU 267 078
White Moor
WMJ
Wet heathland
SU 27742 07831
Anses Wood
ANW
Ancient woodland
SU 230 124
Janesmoor Plain
JPD
Dry grassland
SU 24588 12805
Ocknell Plain
OPH
Dry heathland
SU 23401 11498
South Bentley Inclosure
SBI
Plantation woodland
SU 23411 12892
Brook Common
BCL
Wet grassland
SU 26340 14103
Ocknell Plain
OPJ
Wet heath
SU 23103 11578
Hincheslea Wood
HLW
Ancient woodland
SU 27200 00800
Long Slade Bottom
LSD
Dry grassland
SU 26400 00700
Widden Bottom
WBH
Dry heathland
SZ 28700 98800
Set Thomas inclosure
STI
Plantation woodland
SZ 26200 99600
South Weirs
SWL
Wet grassland
SU 28600 01500
Long Slade Bottom
LSJ
Wet heathland
SU 28300 00000
p ‘peripheral’, plots all other plots are ‘core’ plots
123
3390
Biodivers Conserv (2012) 21:3385–3410
Table 2 Summary of environmental conditions in each habitat type, based on existing records and data
collected during the project
Habitat
Age
(*years)
Mean
fragment
size (ha)
Vegetation
Litter
Soils
pH
Ancient
woodland
(core and
periphery)
C103
117
Quercus spp., Fagus
sylvatica, Ilex
aquifolium, Betula
pendula , ground flora
generally sparse, some
Hedera helix. More base
rich woodlands on brown
earth soils have
Crataegus monogyna,
Euphorbia
amygdaloides, Acer
campestre, with Viola
canina, Anenome
nemorosa and Oxalis
acetosella in the herb
layer
Present
Predominantly gleyed clay
soils: stagnogleys
common, gleyed acid
brown earths less
common. A few
ungleyed acid brown
earths and podsols
Variable
(low
neutral)
4.7–6.2
Plantation
woodland
102
276
Planted or sown Quercus
spp., self-seeded Fagus
sylvatica, Ilex aquifolium
understorey, occasional
Castanea sativa and
Pinus sylvestris
Present
As Ancient woodland
Variable
(low to
neutral)
5.2–6.2
Dry
grassland
101
33
Dominated by Agrostis
curtisii grass, with
Calluna vulgaris, Ulex
europaeus. Also
prostrate herbs such as
Potentilla erecta,
Gallium saxatile,
Polygala serpyllifolia
Absent
Predominantly ungleyed
brown earths
Mildly
acidic to
neutral
5.5–6.8
Wet
grassland
101
22
Dominated by Molinia
caerulea and Cirsium
dissectum, with Juncus
species
Absent
Predominantly gleyed
brown earths
Mildly
acidic
5.5–6.1
Dry
heathland
101
23
Calluna vulgaris
dominated, with
Deschampsia flexuosa
grass, Pteridium
aquilinum, Potentilla
erecta and Gallium
saxatile
Sparse
Poorly developed rankers
and freely-draining
podsols
Acidic
5.1–5.8
Wet
heathland
(core and
periphery)
101
18
Erica heathers
predominate, with purple
moor grass (Eriophorum
sp.) and Sphagnum moss
forming hummocks.
Drosera rotundifolia
present in wet hollows
Absent
Peats and stagnohumic
gleys
Acidic to
neutral
4.3–6.2
(Delta-T Devices Ltd., Cambridge, UK) and recorded as percentage water filled pore space.
From each 1 m2 quadrat a soil sample was taken and pH (in water) was determined in the
laboratory using a pHenomenal pH 1000L meter (VWR International Ltd., Lutterworth, UK).
Litter composition in each of the woodland plots was determined. One hundred leaves
were collected from each quadrat in the woodland plots and the tree/s from which the
123
Biodivers Conserv (2012) 21:3385–3410
3391
Fig. 2 Organogram showing the hierarchical structure of sampling across the plots
leaves came were identified to give proportions of different leaf species in the litter. These
one hundred leaves were then air dried to constant mass. The mass of ten leaves from each
species was recorded and an estimate for the total mass of each leaf species in each quadrat
calculated from the mean of those ten leaves.
Soil taxonomy was determined in the field from six randomly located samples per site
(work undertaken in January 2011). These samples were collected using an auger, to the
depth of the parent material. Soil colour, soil texture, major soil features (gleying, podzolic
horizons etc…) and the depth of each horizon was recorded. Soils were classified using the
simplified key in Trudghill (1989, p. 344). Details of the equivalent taxonomy for the
USDA and WRB classifications are available as supplementary Table 2.
We calculated the habitat fragment size of each of our plots using mapping data supplied by the Forestry Commission. We used ArcMap10 (ESRI, Redlands, USA) to identify
the habitat fragments in which each of our plots was located, taking into consideration
continuous and contiguous habitat and used an area calculation procedure to generate
fragment sizes for each plot.
123
3392
Biodivers Conserv (2012) 21:3385–3410
Fig. 3 Map showing extent of different habitat types across the New Forest National Park. a ancient
woodlands, b plantation woodlands, c dry heath, d wet heath, e wet grassland, f dry grassland
Habitat descriptions
Ancient woodlands are dominated by Fagus sylvatica (beech), Quercus robur (pedunculate
oak) and Quercus petraea (sessile oak), with significant Ilex aquifolium (holly) in the understorey. In more base-rich woodlands, Acer campestre (field maple), Crataegus monogyna
123
Biodivers Conserv (2012) 21:3385–3410
3393
(hawthorn) and Taxus baccata (yew) are common, the latter on thin soils. The herb layer is
species poor, with Pteridium aquilinum (bracken) and Rubus fruticosus (bramble) the dominant
species. Plantation woodlands were clear-felled and either planted or sown with Quercus robur,
but there are occasional Quercus petraea plantations. Fagus sylvatica, Betula pendula (silver
birch) and Pinus sylvestris (Scot’s pine) have self-sown in many of the plantations. Ilex
aquifolium is again significant in the understorey. The herb layer is often less diverse in the
plantations. Dry grasslands are neutral to acidic grassland dominated by Agrostis curtisii, with
Calluna vulgaris and Ulex europaeus common shrub species. This plant community is resistant
to grazing. Wet grasslands are ground water-fed grasslands dominated by Molinia caerulea and
Cirsium dissectum, with Juncus spp. as a significant part of the community. This community is
also maintained by grazing. Dry heaths are Calluna vulgaris-dominated heaths on acidic sandy
or gravely soils with Deschampsia flexuosa common. Burning and grazing regimes create
variation in vegetation structure. Wet heaths are dominated by Erica species, with Sphagnum
moss and Eriophorum grass common, the latter forming hummock structures.
Sampling methods
Three sampling methods were used for macrofauna: hand sorting of soil pits, Winkler bag
extraction of leaf litter, and pitfall trapping. Soil pits and leaf litter quadrats were located at
distances from the centre of each one hectare plot using randomly generated compass
bearings and radii (in metres). Six 1 m2 quadrats of leaf litter were sieved per plot. The
sieved litter was hung in Winkler bags for 3 days (see Krell et al. 2005 for a discussion of
this method) and invertebrates preserved in 80 % Industrial Methylated Spirits (IMS). Six
soil pits (25 9 25 9 10 cm deep) were dug per plot and invertebrates hand sorted into
80 % IMS. Each pit was located in the centre of a 1 m2 litter quadrat. Six pitfall traps
(9 cm diameter, 14.5 cm deep) were located in a line 2 m apart through the centre of the
plot. These were a third filled with water, with anti-freeze used as a preservative and
washing-up liquid added to reduce the water’s surface tension. The pitfall traps were in
position for a week and the collected invertebrates transferred into 80 % IMS.
Soil and litter sampling was conducted in May 2010, while pitfall trapping was undertaken
in September 2011. The sampling was undertaken in two field periods due to time constraints.
We identified specimens in the following major clades to species: Coleoptera; Formicidae;
Opiliones; Chilopoda; Diplopoda; Isopoda and Lumbricidae. Analyses were conducted on the
whole dataset, pooling all data per sampling method per plot from the target groups. This gives the
most complete model for total biodiversity within the studied sampling methods and avoids very
complicated sub-analyses of individual clades, many of which have only a handful of species.
Analyses
Variation of environmental variables within habitat
We undertook an initial standardised principal components analysis (PCA) of environmental
variables (within CANOCO, (Leps and Smilauer 2003)) to show the validity of the habitat classification and exactly how collateral environmental variables varied within and between habitats.
Gamma-diversity
We estimated, from all sampling strata (see below), the total number of species found
across the national park simply by combining the data from each sampling method and
123
3394
Biodivers Conserv (2012) 21:3385–3410
calculating how many species from each taxonomic group were sampled in total. We also
made estimates of the total extrapolated number of species from the target groups in the
park using the first-order jack-knife (using specpool in vegan (Oksanen et al. 2011) with
the index set to ‘‘jack1’’; these analyses and all subsequent ones, unless otherwise stated,
were undertaken in R [R Development Core Team 2011]).
The jack-knife (and related re-sampling metrics) is based on the assumption that there is
a species pool (i.e. that the community is closed and that there is a fixed pool size that can
be estimated). The approach also assumes that we can accurately identify species which are
found in only a single plot (i.e. many of which will be singleton ‘tourists’ which may
normally be found in another habitat stratum). Both assumptions are hard to justify and
therefore the estimates must be treated with extreme caution. All indices that extrapolate
species numbers are biased for open communities (Colwell and Coddington 1994). Also
these estimates are only for the individual compartments of habitats that the methods
sample within, they do not give estimates for other compartments that are more patchily
distributed but which may contain additional rare and/or cryptic species (e.g. saproxylic
beetles).
Alpha-diversity
Two measures of alpha diversity (i.e. defined here as estimates of species diversity within
each hectare plot) were employed in each plot and for each sampling methods: species
density (as calculated by specnumber in vegan) and a measure of evenness (Pielou’s index,
Pielou 1966)). Pielou’s index was calculated as J = H/log S, where H = the Shannon
diversity value for a plot and S the total number of species in the plot.
We did not compute alpha-diversity comparisons between sampling methods directly,
as sampling effort is clearly not directly comparable between them.
Beta-diversity (species turnover)
We used the vegdist program in vegan package, with method = ‘‘bray’’ and binary =
‘‘FALSE’’ to generate a Bray-Curtis dissimilarity matrix as the basis for calculating
turnover between and within plots and habitats. The only exception to this was for the
combined dataset, which is a binary matrix, where we set binary = ‘‘TRUE’’ to find
indices for the presence/absence data.
We tested the significance of compositional difference using adonis (a permutational
MANOVA-type method which tests for overall compositional differences between habitats) and betadisper (testing for within habitat heterogeneity differences). Data combined
across sampling methods were visualised using a cluster analysis derived from the distance
matrices (using the agnes command in the package cluster). Both adonis and betadispers
generate Principal Co-ordinates (PCoA) axis scores from the distance matrices. Post hoc
Tukey HSD tests were used to examine pairwise differences between axis 1 and 2 of the
PCoA ordinations to explore in more detail those adonis results that were statistically
significant overall.
We identified ‘indicator species’ for habitats and clusters of habitats using the indicspecies package in R (De Caceres and Legendre 2009) using the multipatt command (with
func = ‘‘IndVal.g’’, nperm = 999, all habitat combinations considered). The combined
binary data from all sampling methods were used for this analysis with the main associations of species and sampling method made post hoc from the results. ‘Indicator species’
are those which have a IndVal.g association index which is significant at P \ 0.05.
123
Biodivers Conserv (2012) 21:3385–3410
3395
Composition
The beta diversity analyses (described above) test only the association of the habitats with
assemblage composition (i.e. the adonis results presented below). Ordinations were used to
add an extra layer of information: to examine the influence of other, directly measured
environmental variables. These variables may vary within habitats, or cut across the habitat
classification. These analyses were intended to give a clearer indication of the relationship
between habitats, environmental variables and species distributions than is possible using
the PCoAs.
Initial DCA analyses were used to establish the size of latent environmental gradients
(Leps and Smilauer 2003). We then used linear methods (Principal Components Analysis,
PCA) for gradient lengths less than 2 and unimodal methods (Correspondence Analysis,
CA) for gradient lengths greater than 2, as appropriate, to visualise the multivariate patterns in the data. Initial spatial analyses were undertaken to confirm the size of spatial
effects, but these were only significant in one trivial case (see results below) and so the
spatial variables did not need to be included (‘partialed out’) as co-variables. Environmental variables were passively mapped on to the unconstrained analyses. We used this
indirect approach to avoid producing models with just one or two significant environmental
variables, which make very strong assumptions about the influence of the measured
variables and therefore often produce biologically implausible, and therefore potentially
misleading, ordination results.
Results
Habitat structure and environmental correlates
A Monte Carlo permutation test of the association of the habitats (i.e. with each plot within
a habitat treated as a replicate) with the directly measured environmental variables showed
a strongly significant association, both overall (F = 4.9, P \ 0.01, 999 permutations) and
for individual habitat types (for all habitat types, k [ 0.4, P \ 0.01, 999 perms.). The
habitats differed in vegetation and abiotic variables (patterns summarised in Table 2 from
the standardised PCA results), but there are significant sources of within habitat variation
and variation that cut across the habitat classification (see Fig. 4).
Between-habitat variation is associated with an open/woodland split on axis 1, and a
grassland/heathland split on axis 2. The differences in environmental conditions between
plots within the broad habitat categories are small except for variations in the ancient
woodland and dry grassland plots. The primary source of these differences within habitats
is variation in soil pH. This is seen most clearly in the spread of dry grassland plots in the
PCA biplot (Fig. 4), with plots with high axis 2 values having higher soil pHs. Three
woodland sites also have high axis 2 values, and although they are not so clearly high soil
pH sites, they do have characteristically relatively calcicolous plants (Euphorbia amygdaloides, Crataegus monogyna, Acer campestre and Taxus baccata) growing in them. A
secondary trend is a decrease in soil moisture with increased axis 2 scores.
Gamma diversity
We sampled 21,163 individuals of macrofauna across all plots and sampling methods,
11,120 [52 %] of which were in the target taxonomic groups [87 % of totals in soil pits,
123
3396
Biodivers Conserv (2012) 21:3385–3410
LSD
1.0
Lotus
Trifoliu
Hiera
CHD
Tarax
Ranunucu
pH
Polygonu
Plantago
Bellis
A.campes
Crataegu
Bearth
Oxalis
temp
Euphor
WWW
Anenome
GBEarth
RSW
Hedera
logFrag
Taxus
Fagus
Calluna
Sphag
Erica
Pteridiu
Quercus
Sgley
Ilex
litt-dep
-0.6
moisture
-0.6
1.0
Fig. 4 Standardised PCA biplot of environmental variables showing plots (circles), continuous environmental variables (arrows) and categorical variables (black diamonds). Habitat circles are as follows: filled
circles—core habitats, open circles—peripheral habitats; green—ancient woodlands, brown—plantation
woodlands, red—dry grasslands, orange—wet grasslands, dark blue—dry heathlands, light blue—wet
heathlands. Categorical abbreviations as follows: Trifoliu—Trifolium repens; Lotus—Lotus corniculatus;
Hiera—Hieracium agg; Tarax—Taraxacum officinale; Ranuncu—Ranunculus repens; Plantago—Plantago
lanceolata; Bellis—Bellis perennis; Polygonu—Polygonum aviculare; Calluna—Calluna vulgaris; Sphag—
Sphagnum moss; Erica—Erica spp.; Pteridiu—Pteridium aquilinum; Ilex—Ilex aquifolium; Fagus—Fagus
sylvatica; Quercus—Quercus spp.; Hedera—Hedera helix; Oxalis—Oxalis acetosella; Anemone—Anemone
nemorosa; A. campes—Acer campestre; Crataegu—Crataegus monogyna; Euphor—Euphorbia amygdaloides; Bearth—brown earth; GBEarth—gleyed brown earth; Sgley—stagnogley. Continuous variable
abbreviations as follows: temp—temperature; litt-dep—litter depth; logFrag—fragment area (log scale).
Plot abbreviations are as in Table 1
62 % of totals in pitfall traps, 40 % of totals in litter]. In all cases the two major non-target
groups were Diptera (particularly Tipulidae in the soil pits in open areas) and spiders. We
identified 250 species across the entire landscape, comprising 182 beetle, 21 centipede, 13
harvestmen, 12 ant, 11 earthworm, seven millipede and four woodlice species.
123
Biodivers Conserv (2012) 21:3385–3410
3397
The first order jack-knife estimated 120 ± 10 [se] species in soil, 175 ± 18 species in
litter and 154 ± 11 species in pitfall traps. The total species pool can also be estimated
using the pooled-sampling method data, giving us an overall estimate of 350 ± 17 species.
Alpha diversity
The wooded habitats had higher species density in the pitfall trap and soil pits than the
open habitats, although this was not always statistically significant (Fig. 5b, d). The wet
grassland habitat had the lowest species density within the pitfall traps, and this difference
was statistically significant for pairwise comparisons between wet grasslands and core
ancient woodlands and plantation woodlands, and marginally significant between wet
grasslands and peripheral ancient woodlands (Tukey HSD test, see Fig. 5a).
Within the litter there was a significant difference in species density between the core
woodlands and the other two woodland types (Tukey HSD test, Fig. 5c).
The combined sampling methods data shows an overall significant difference in species
density between the wooded plots and the open plots, and no significant differences within
a
b
50
20
60
a
b
15
40
a,b
b,c
b,c
10
20
c
c
a,b
c
c
a,b
a,d
5
10
c
AC
AP PW
DG WG
DH
WC WP
AC
AP PW
DG
a,b
WG
DH
WC WP
d
12
45
50
14
c
a
10
a
8
35
40
Species density (per plot)
30
a,b
a,b
a
b
a
a
6
30
a
a
4
a
20
2
25
b
a
AC
AP
PW
AC
AP
PW
DG WG DH
WC
WP
Habitats
Fig. 5 Boxplots of species density per habitat type for a combined data, b soil pits, c leaf litter and d pitfall
traps. AC—core ancient woods; AP—peripheral ancient woods; PW—plantation woods; DG—dry
grassland; WG—wet grassland; DH dry heath; WC—core wet heath; WP—peripheral wet heath. Boxes
with the same letter are not significantly different in a Tukey HSD test following a one-way ANOVA with
habitat as a factor
123
3398
Biodivers Conserv (2012) 21:3385–3410
those broad categories (Fig. 5a). The core ancient woodlands had the highest species
density of all the plots and this was significantly higher than the peripheral ancient
woodlands, but not significantly higher than the core plantations (Fig. 5a).
There was no significant difference in evenness (Pielou’s J) between habitats for any of
the sampling methods (analysis results not shown).
There was no relationship between fragment size and species density when habitat
differences were taken into account within a split plot design (i.e. the within plot effect
[F = 0.59, P = 0.45] was not significant, but the between habitat effect was (F = 9.4,
P = 0.02]). Open habitats have smaller fragments, so we were unable to separate species/
area and habitat effects.
Beta diversity (turnover)
RWJ
BCL
WBL
CHD
BLJ
* * *
*
*
*
BUL
LSD
*
BLL
BLD
JPD
OPJ
MAW
SOI
HWI
SWW
OIW
0.3
*
TTW
SBI
RSW
BSW
STI
BWW
HLW
WWW
NPI
GPH
PGH
OPH
OHH
WBH
YTJ
ADD
CHJ
OHJ
WBJ
LSJ
WMJ
0.6
PHW
*
0.4
0.5
Height
0.7
HMJ
SWL
0.8
0.9
There were high, and significantly different, levels of turnover between sampling methods
and between habitats, as expected, with the turnover between sampling methods being
higher than the turnover between habitats (adonis analysis of dissimilarity matrix, sampling method factor, F = 20.2, P 0.0001, habitat factor for combined dataset, F = 3.2,
P \ 0.0001). There is almost no overlap in species between the sampling methods, except
for a few very generalist species (e.g. the Isopoda species). Results from the litter samples
are much more homogeneous in composition than from the other sampling methods, but
leaf litter is obviously only well developed in wooded habitats.
The combined dataset with sampling methods pooled showed the largest differences in
composition between habitats (Table 3) and it is that analysis that we concentrate our
attention on predominantly. There was a significant overall difference (adonis analysis,
habitat factor, F = 4.01, P \ 0.0001) in habitat composition. The cluster diagram (Fig. 6)
shows large compositional differences, with a highly depauperate wet grassland site (SWL)
splitting off first and then three main clusters which correspond broadly to the three main
habitat types (i.e. woodland, grassland, heathland). HMJ appears to be another anomalous
*
Fig. 6 Cluster diagram showing the structure of the compositional data across the entire landscape. The
height indicates the Bray-Curtis dissimilarity between branches either side of nodes of the cluster tree. Plot
abbreviations as in Table 1
123
Biodivers Conserv (2012) 21:3385–3410
3399
Table 3 Mean turnover between habitat types, using data from the pooled methods dataset
Ancient
(periph.),
%
Ancient
(core)
Ancient
(core), %
Plantation,
%
Wet
grassland,
%
Dry
grassland,
%
Wet heath
(periph.), %
67a
Plantation
67
71
Wet grass
13
3
Dry grass
13
5
9
67
Wet heath
(periph.)
27
18
23
34
34
Wet heath
(core)
17
11
13
38
57
46
Dry heath
29
24
30
15
31
39
a
Wet heath
(core), %
9
43
Turnover calculated as (1-Bray-Curtis dissimilarity x 100)
site that cannot easily be classified with either the heathland or the grassland, and three
sites (ADD, OPJ and RWJ) appear to be misclassified according to their macrofauna
composition. Among the woodlands the PHW site is most divergent, being an almost pure
holly stand. The largest overall difference between habitats is between the core ancient
woodlands and the wet grasslands which only had a 3 % similarity in composition
(Table 3).
Considering each sampling method separately, the pitfall traps show an overall difference in composition across the plots (adonis analysis, F = 3.1, P \ 0.0001), with the
pattern being very similar to the pooled results.
There was a just significant overall difference in composition between habitats in the
leaf litter (adonis analysis, F = 1.5, P = 0.019), reflected in a marginally significant
difference between the core ancient woodland plots and the plantation plots on axis 2 of the
PCoA (Tukey HSD post hoc test, P = 0.06).
The soil pit data shows considerable overlap between habitats, although there is an
overall difference (adonis analysis, habitat factor, F = 2.23, P \ 0.0001), with axis 1 of
the PCoA showing a significant difference between the wooded and open plots (F = 11.47,
P 0.0001), but with no differences in subsequent axes.
Results for the combined dataset show a strong preponderance of indicator species in
the woodland habitats for all three sampling methods (Table 4)
In none of the sampling methods were there significant differences in within-habitat
heterogeneity (as estimated by betadivers) across the six habitats (not differentiating edge
from core habitats), suggesting comparable degrees of variation in beta diversity within
habitats.
Ordinations
Initial spatial analyses found almost no significant spatial structure in the soil and pitfall
trap data. There was slight, but significant, spatial structure in the litter data (k = 0.35,
P = 0.02), caused by a single woodland (PHW), which is in the far south of the national
park and has a very different vegetation structure.
123
3400
Biodivers Conserv (2012) 21:3385–3410
Table 4 Indicator species for each habitat using the indicspecies package in R
Habitat(s)
Strata
Species
Dry Grass
Pitfall traps
Longitarsus pratensis, Calathus fuscipes,
Calathus melanocephalus
Wet Grass
Soil pits
Octolasion tyrtaeum
Wet ? Dry Grass
Pitfall traps
Lamyctes emarginatus
Grass ? Dry Heath
Soil pits, pitfall traps
Lasius niger
Heath ? Wet Grass
Pitfall traps
Phalangium opilio
Wood ? Wet Heath
Pitfall traps
Carabus granulatus
Plantation
Leaf litter
Moycta clientula, Xantholinus gallicus,
Stenamma debile, Othius punctulatus
All
Wood
Wood ? Plantation
Leaf litter
Carabus granulatus, Trichoniscus pusillus,
Agriotes acuminatus, Notiophilus rufipes
Soil pits
Geophilus flavus
Leaf litter
Acalles ptinoides, Geostiba circellarius,
Barypeithes araneiformis, Agriotes
pallidulus, Cylindroiulus punctatus,
Cephennium gallicum, Caenopsis waltoni,
Habrocerus capillaricornis, Notiophilus
biguttatus, Orchestes fagi, Stenichnus
collaris, Cryptocephalus pusillus,
Euophyrum confine, Lophopilio palpinalis,
Dalopius marginatus, Mocyta fungi,
Mycetoporus lepidus
Carabus violaceus
Cryptops hortensis, Athous haemorrhoidalis,
Pitfall traps
Anoplotrupes stercorosus, Oligolophus
tridens, Oniscus asellus, Nebria
brevicollis, Lithobius variegatus, Philoscia
muscorum, Philonthus decorus,
Polydesmus denticulatus,
Soil pits ? leaf litter
All
Wood ? Plantation ? Dry
Heath
Wood ? Plantation ? Heath
Soil pits
Nalassus laevioctostriatus,
Soil pits ? leaf litter
Geophilus truncorum, Geophilus easoni
Pitfall traps
Pterostichus madidus
All
Porcellio scaber
Pitfall traps ? leaf litter
Pitfall traps
Abax parallelepipedus
Ocypus olens
All species included which have an IndVal.g association index which is significant at P \ 0.05
The presence of plots with soil pit samples that had no identified individuals of the
target groups meant that the soil pit dataset had to be analysed using PCA, in order for
samples not to be excluded altogether from the analysis. The soil pit ordinations (Fig. 7)
show a moderately strong woodland vs. open canopy split on axis 1, with the dry
123
Biodivers Conserv (2012) 21:3385–3410
3401
WWW
1.0
A.campes
Crataegu
Euphor
Oxalis
Anenome
cGEOfla
lALLchl
RSW
STI
lLUMrub
logFrag
litt-dep
GBEarth
Hedera
cGEOtru
Ilex
NALlae
Quercus
Fagus
Bellis
Sgley
cGEOeas
Calluna
-0.4
Erica
-1.0
Sphag
moisture
1.0
Fig. 7 Standardised PCA biplot of soil pit data showing plots (circles), continuous environmental variables
(dashed arrows), categorical variables (black diamonds) and species (solid black arrows). Habitat circles are
as follows: filled circles—core habitats, open circles—peripheral habitats; green—ancient woodlands,
brown—plantation woodlands, red—dry grasslands, orange—wet grasslands, dark blue—dry heathlands,
light blue—wet heathlands. Categorical abbreviations as follows: Oxalis—Oxalis acetosella; Anemone—
Anemone nemorosa; A.campes—Acer campestre; Crataegu—Crataegus monogyna; Euphor—Euphorbia
amygdaloides; Ilex—Ilex aquifolium; Fagus—Fagus sylvatica; Quercus—Quercus spp.; Hedera—Hedera
helix; Calluna—Calluna vulgaris; Sphag—Sphagnum moss; Erica—Erica spp.; Sgley—stagnogley;
GBEarth—gleyed brown earth. Continuous variable abbreviations as follows: temp—temperature; littdep—litter depth; logFrag—fragment area (log scale). Species abbreviations as follows: cGEOfla—
Geophilus flavus; cGEOtru—Geophilus truncorum; cGEOeas—Geophilus easoni; lALLchl—Allolobophora
chlorotica; lLUMrub—Lumbricus rubellus; NALlae—Nalassus laevioctostriatus
heathlands somewhat intermediate between the wooded and non-wooded plots. Axis 2, in
contrast, has a split between brown earth (gleyed and non-gleyed) and non-brown earth soil
sites. The woodland and dry heathland plots (low values on axis 1) had characteristic soilinhabiting centipedes (Geophilus truncorum, Geophilus easoni) and larvae of the tenebrionid beetle, Nalassus laevioctostriatus. The brown earth plots had soil-inhabiting
earthworms (Allolobophora chlorotica and Lumbricus rubellus) and a centipede (Geophilus flavus). Within the woodland sites (WWW, RSW, STI) the characteristic earthworms
were found in plots with broadly calcicolous plants.
Litter ordinations revealed no significant compositional differences between the three
habitat types that have leaf litter but two clear outlier ancient woodland plots—WWW and
PHW with distinct macrofauna assemblages (Fig. 8).
Pitfall trap ordinations (Fig. 9) show a split between woodland and non-woodland along
axis 1, with the woodland sites being uniform in composition and with a distinct set of
123
3402
Biodivers Conserv (2012) 21:3385–3410
1.0
PHW
holly_L
ASAcur
CoreD
Lmass
BEMlam
BAEvar
TREobt
Ranker
PTRstr
a.o_peri
Ilex
Fagus
Pterid
ACApti
Hedera
Anenome
WWW
Euphor
Oxalis
GBEarth
birch_L
BEMman STOpum
Crataegu
BRAfos cCRYpar Viola
A.camp
Rubus
mPOLcor
-0.6
litt-dep
beech_L
-0.2
0.8
Fig. 8 Standardised PCA biplot of leaf litter data showing plots (circles), continuous environmental
variables (dashed arrows), categorical variables (black diamonds) and species (red triangles). Habitat
circles are as follows: filled circles—core habitats, open circles—peripheral habitats; green—ancient
woodlands, brown—plantation woodlands. Categorical abbreviations as follows: Pteridiu—Pteridium
aquilinum; Ilex—Ilex aquifolium; Fagus—Fagus sylvatica; Hedera—Hedera helix; Rubus—Rubus fruticosus agg.; Oxalis—Oxalis acetosella; Anemone—Anemone nemorosa; A.campes—Acer campestre;
Crataegu—Crataegus monogyna; Euphor—Euphorbia amygdaloides; Viola—Viola riviniana; GBEarth—
gleyed brown earth; a.o_peri—peripheral ancient woods. Continuous variable abbreviations as follows: littdep—litter depth; Lmass—litter mass; holly_L—holly litter; beech_L—beech litter; birch_L—birch litter;
CoreD—distance from the centre; ASAcur—Asaphidion curtipes; BEMlam—Bembidion lampros; BSEvar—Baeocrara variolosa; TREobt—Trechus obtusus; PTRstr—Pterostichus strenuous; ACApti—Acalles;
BEMman—Bembidion; STOpum—Stomis pumicatus; BRAfos—Bra; cCRYpar—Cryptops; mPOLcor—
Polydesmus coriacious
woodland species (e.g. Abax parallelepipedus, Pterostichus madidus, Carabus violaceus,
Anoplotrupes stercorosus). Axis 2 separates two dry grassland plots (CHD, LSD) from the
bulk of the open plots at low axis values and one wet heathland plot (WMJ) at high axis
values. The dry grassland plots have characteristic beetle species: Carabidae (Calathus
fuscipes and C. melanocephalus) and Chrysomelidae (Longitarsus pratensis and Longitarus succineus). The wet heathland plot (WMJ) has populations of the ant, Formica piceus.
123
Biodivers Conserv (2012) 21:3385–3410
3403
1.0
WMJ
fFORpic
moisture
Drosera
temp
oPHAopi
Erica
oOLItri ABApar
GEOste Pteridiu
PTEmad Sgley
Calluna
fLASnig
fMYRrug
Hedera Quercus
Ilex
Fagus
CARvio
litt-dep
Polygonu
NEBbre
Ranunucu
Bearth
logFrag
CALmel
Tarax
LONpra
CALfus
Hierac
Plantago
Trifoliu
LONsuc
-1.0
Lotus
-0.8
0.8
Fig. 9 Standardised PCA biplot of pitfall trap data showing plots (circles), continuous environmental
variables (arrows), categorical variables (black squares) and species (red triangles). Habitat circles are as
follows: filled circles—core habitats, open circles—peripheral habitats; green—ancient woodlands,
brown—plantation woodlands, red—dry grasslands, orange—wet grasslands, dark blue—dry heathlands,
light blue—wet heathlands. Categorical abbreviations as follows: Pteridiu—Pteridium aquilinum; Ilex—Ilex
aquifolium; Quercus—Quercus spp.; Fagus—Fagus sylvatica; Hedera—Hedera helix; Trifoliu—Trifolium
repens; Lotus—Lotus corniculatus; Hiera—Hieracium agg; Tarax—Taraxacum officianlis; Ranuncu—
Ranunculus repens; Plantago—Plantago lanceolata; Polygonu—Polygonum sp.; Calluna—Calluna vulgaris; Erica—Erica spp.; Drosera—Drosera rotundifolia; Bearth—brown earth; Sgley—stagnogley; Continuous variable abbreviations as follows: temp—temperature; litt-dep—litter depth; logFrag—fragment area
(log scale). Species abbreviations as follows: fFORpic—Formica picea; fLASnig—Lasius niger;
fMYRrug—Myrmica ruginodis; oPHAopi—Phalangium opilio; oOLItri—Oligolophus tridens; GEOste—
Anoplotrupes stercorarius; PTEmad—Pterostichus madidus; CARvio—Carabus violaceus; ABApar—Abax
parallelepipedus; NEBbre—Nebria brevicollis; CALmel—Calathus melanocephalus; CALfus—Calathus
fuscipes; LONpra—Longitarsus pratensis; LONsuc—Longitarsus succineus
The combined dataset (Fig. 10) showed less distinct patterns than the other datasets. In
particular, the heath plots cluster close to the woodland plots on axis 1, and the wet
grassland plots are widely spread across axis 2.
The ordinations produce a very similar pattern of the sample positions except for the
litter and pitfall trap ordinations which showed significantly different multivariate patterns
(Mantel tests of plot positions on axis 1 and 2, see Table 5).
123
Biodivers Conserv (2012) 21:3385–3410
1.0
3404
PTEver
Deschamp
oMITmor
Molinia
WHp
CHEhor
lLUMrub
oOLItri
fMYRrug
ACApti dCYLpun
Sphag WG
cLAMema
Pteridiu
ACRspp Fagus
fLASnig
Calluna
temp
Potentil
AGRpal Quercus
Sgley
oPHAopi
Ilex
LONpra
PHImus
Plantago
litt-dep
TRIpus ONIase
Tarax
Polygonu
logFrag
CALfus Ranunucu
iTRIpus GEOcir
Hawks
LIOmic
LONlur
XANlin
Lotus
-1.0
Trifoliu
-0.8
0.8
Fig. 10 Standardised PCA biplot of the combined data showing plots (circles), continuous environmental
variables (arrows), categorical variables (black diamonds) and species (red triangles). Habitat circles are as
follows: filled circles—core habitats, open circles—peripheral habitats; green—ancient woodlands,
brown—plantation woodlands, red—dry grasslands, orange—wet grasslands, dark blue—dry heathlands,
light blue—wet heathlands. Categorical abbreviations as follows: Pteridiu—Pteridium aquilinum; Ilex—Ilex
aquifolium; Quercus—Quercus spp.; Fagus—Fagus sylvatica; Trifoliu—Trifolium repens; Lotus—Lotus
corniculatus; Hiera—Hieracium agg; Tarax—Taraxacum officinale; Ranuncu—Ranunculus repens; Plantago—Plantago lanceolata; Polygonu—Polygonum sp.; Calluna—Calluna vulgaris; Potentil—Potentilla
erecta; Sphag—Sphagnum moss; Molinia—Molinia caerula; Deschamp—Deschampsia flexuosa; Sgley—
stagnogley; WHp—peripheral wet heath; WG—wet grassland. Continuous variable abbreviations as
follows: temp—temperature; litt-dep—litter depth; logFrag—fragment area (log scale). Species abbreviations as follows: fLASnig—Lasius niger; fMYRrug—Myrmica ruginodis; oPHAopi—Phalangium opilio;
oOLItri—Oligolophus tridens; oMITmor—Mitopus morio; lLUMrub—Lumbricus rubellus; iTRIpus—
Trichoniscus pusillus; PHImus—Philoscia muscorum; dCYLpun—Cylindriulius punctatus; cLAMema—
Lamyctes emarginatus
Discussion
Patterns of soil macrofauna diversity in the New Forest landscape
These analyses point clearly to the large differences in species density and composition
between the wooded and open areas of the New Forest. This difference is found in both the
123
Biodivers Conserv (2012) 21:3385–3410
3405
Table 5 Mantel tests comparing ordination results between sampling methods, environmental variables
and the pooled-methods dataset
Combined
Env. var.
Combined
Pits
Litter
0.51***
Soil
Litter
Ground
0.59***
0.55*
0.58***
0.98***
0.8***
0.49***
0.83***
0.56***
0.47
ns
Based on Euclidean distance between plot centroids on first two axes of ordination plots. Key to superscripts: ns not significant, * P = 0.01–0.05, ** P = 0.001–0.009, *** P B 0.001
pitfall trap and leaf litter data, and is obviously strongly augmented by leaf litter being
found only in the wooded plots.
Within the three woodland habitats there is very little species turnover, suggesting a
broad species pool from which the assemblage for each plot is being drawn. The lower
species density of the peripheral woodlands may be a geometric effect, whereby species
ranges overlap in the centre of the landscape so producing a peak in species richness
(Colwell et al. 2004). Woodlands on the periphery also have fewer surrounding woodlands
to act as source populations to allow re-colonisation of species from the meta-community
after local extinctions. Being at the edge of the New Forest landscape, they have more
farmland at their boundaries than woodlands in the core of the landscape, which are
surrounded by other woods. The core woodlands, by contrast, have a larger area of
woodland from which species can re-colonise after local extinction events. This effect is
unlikely to apply to other habitat types as they are far more fragmented than the woodland
habitats.
Whitley Wood (WWW) has a rather different species composition from other woodlands, particularly in the litter. The reasons for this are complex. High habitat heterogeneity
is probably partially responsible, as WWW has a mixture of relatively calcicolous and noncalcicolous vegetation types within the hectare plot. It also has gleyed acid brown earth
soils. Red Shoot Wood (RSW) and Set Thorns Inclosure (STI) share some of these
characteristics (relatively calcicolous vegetation and brown earth soils) and these two
factors seem to indicate a different assemblage in these woodland habitats. This is primarily a difference in the earthworm assemblages; the three woodlands have a more
abundant and diverse earthworm assemblage than the other woodlands in our dataset.
Among litter samples, Whitley Wood (WWW) and Pigsty Hat (PHW) have significantly
different compositions than the other woodlands. The CA analyses (Fig. 8) shows that the
species found predominantly in Whitley Wood are those more commonly found in wet
places (e.g. Bembidion mannerheimi, Stomis pumicatus). Whitley Wood is patchily seasonally waterlogged, as indicated by the gleyed soils, suggesting that waterlogging creates
niches for species which are not present in other woodland sites. Pigsty Hat (PHW) is an
oak wood with significant Ilex aquifolium understorey. This is a type of woodland found in
small fragments across the New Forest (Tubbs 2001), and is more like scrub vegetation
than typical woodland. These woodlands are characterised by a high proportion of Ilex
aquifolium in the understorey, very little ground vegetation and gravely soils. Pigsty Hat
(PHW) is the only woodland of this type among our plots. It undergoes regular disturbance
through the management of the Ilex aquifolium in the understorey and generally has a more
open canopy than other woodland plots. The assemblage in this woodland suggests that it is
much more like an open habitat (grassland or heath) than the other woodlands and the CA
123
3406
Biodivers Conserv (2012) 21:3385–3410
(Fig. 8) shows that species that occur in this plot are more common in open habitats than in
woodlands (e.g. Bembidion lampros, Trechus obtusus).
Heaths, especially dry heaths, have species compositions that are intermediate between
woodlands and other open habitats in our dataset. IndVal analysis shows that some species
(especially from ground samples) are present in both heath and woodland habitats (e.g.
Abax parallelepipedus, Ocypus olens). Open habitats are subject to more repeated disturbance than wooded habitats, with heaths the most constantly disturbed of the plots
studied here. Heaths are burnt on an approximately 20 year rotation, creating a mosaic of
patches of different ages. Species in heaths must therefore be resistant to this disturbance
or be able to rapidly re-colonise.
Open habitats, particularly dry heaths, are much less buffered than wooded habitats and
so have greater extremes of temperature and soil moisture. High summer temperatures
allow some thermophilic species to become established in New Forest heaths, in particular
ant (Formicidae) species which are at the edge of their range in the UK. Dry heath species
include Temnothorax nylanderi, T. interruptus, and Tapinoma erraticum which all have
strongly southern distributions. Pilmore Gate Heath (PGH) and Goat Pen Plain (GPH) have
the highest number of these species. Formica fusca is also found in dry heaths and is
widespread in southern Britain, but it is probably not really a ‘thermophilic’ species.
Outside the wooded habitats, the wet heathlands have the largest number of locally
restricted species. Formica picea is a marsh/wet heathland specialist, found in the New
Forest, Dorset and South Wales and was recorded from the Cooper’s Hill (CHJ) and White
Moor (WMJ) wet heath sites. Other wet heath specialists are Dendrobaena hortensis
(Oligochaeta), Apporectodea limnicola (Oligochaeta), Pterostichus versicolor (Coleoptera;
Carabidae), Pterostichus diligens (Coleoptera; Carabidae), Carabus granulatus (Coleoptarea; Carabidae) and Lochmaea suturalis (Coleoptera; Chrysomelidae). The assemblage in
heaths is depauperate when compared with wooded areas, but does possess some species
particularly well adapted to the relative extremes of heath microclimatic conditions.
Wet and dry grasslands have very similar species assemblages, but the classification of
grasslands is not distinct, with some being intermediate between habitats (e.g. due to heath
to grassland conversion as discussed below). Two sites, however,—Clay Hill (CHD) and
Longslade Bottom (LSD)—have very different assemblages from the other sites, particularly in the pitfall trap samples (Fig. 9), with distinctive Longitarsus and Calathus
species.
Grasslands in the New Forest are heavily grazed by both cattle and ponies which results
in a very short sward, low architectural complexity and eutrophication from deposited
dung. Wet grassland has the lowest diversity among our plots (particularly in the pitfall
trap data, Fig. 5b). We suggest that the high input of pony dung in grasslands has led to the
formation of fertile brown earth soils with large earthworm populations (Curry 1987),
particularly in CHD and LSD.
The habitat definitions used in this study were broadly supported by the data, with much
of the ‘noise’ in the data due to the very broad classifications of habitats in the original
Forestry Commission data and the changes in nature and extent of some of the sites since
the original surveys. This applied particularly to the wet grassland habitat, which was very
variable across the plots, and often had a very low species density of macrofauna. Wet
grassland appeared to be generally the most degraded of the habitats, particularly when
close to roads or human habitations, as in the South Weirs (SWL) and Millyford Bridge
(MBL) plots.
Some of our results cut across the habitat classification. In particular, the distribution of
earthworms across the plots is strongly influenced by the nature of the soils (Fig. 7).
123
Biodivers Conserv (2012) 21:3385–3410
3407
Regardless of habitat there is a higher species density and abundance of earthworms in soil
pits in higher pH (gleyed) brown earth soils (at high ax2 values in the PCA plots). These
plots are either woodland sites, with relatively calcicolous plant species (e.g. Crataegus
monogyna, Euphorbia amygdaloides, Acer campestre, Taxus baccata) or wet grassland
plots very strongly grazed by ponies and cattle. Within the wet grasslands it is not surprising to find that herbivore grazing and associated soil enrichment encourages the
presence of earthworms found in fertile soils (particularly Allolobophora chlorotica and
Aporrectodea caliginosa (Curry 1987)). It is perhaps more surprising to find such a
diversity of earthworms in deciduous woodland, but these results agree with those of a
more detailed study from Whitley Wood (WWW) (Eggleton et al., 2009) and appear to be
due to soil properties that differ substantially from more typical New Forest ancient
woodlands (Fig. 7).
The soils of New Forest woodlands have been previously discussed by a number of
authors and a confusing picture has arisen. Flower (1980) suggests that acid brown earths
are the original forest soils and that their presence is evidence of continual forest cover.
However, in our dataset only three of the ten ancient woodlands have brown earth soils:
Redshoot Wood (RSW), Whitely Wood (WWW) and Berry Wood (BWW). Dimbleby and
Gill (1955) postulate that woodlands that have podzols have re-colonised from heathland,
particularly on gravel soils. However, we only found podzols in one wood (BWW, which
also has brown earth soils in some patches) and so it appears that none of our woodlands,
except possibly parts of BWW, have regenerated from heathland.
Our findings are static snapshots of soil macrofauna biodiversity within a narrow time
frame (the data is drawn from sampling only in May and September 2010) but they are
drawn from a set of habitats that occupy different positions in the dynamic history of the
Forest (Fig. 2). The most striking differences across the habitats are probably due to their
positions in this dynamic history: ancient woodlands are very old and stable; many of the
open habitats are young or subject to periodic disturbance (Table 2).
Compositional differences between sampling methods are over-estimated between
pitfall trap samples and the others due to the sampling being undertaken at different times.
Pitfall trap samples were taken in September while litter and soil samples were taken in
May. It is likely that a temporal shift in species assemblages would have occurred between
May and September, particularly in open habitats, which could account for a small but
significant proportion of the differences in composition between the pitfall trap samples
and the other samples. Additionally pitfall traps collect over long periods of time and so
also collect nocturnal species, which might not be present in leaf litter or soil pit samples.
Our estimates of beta diversity, along with our estimates of the size of species pools are
strongly biased by our snapshot approach to the sampling. We think it very likely that
seasonal turnover will be much greater in the open plots than in the wooded ones, because
conditions within the woodlands are far more buffered than in the open areas. This implies
that if we sampled the habitats across an entire year we would add substantially more
species to our inventories in the open areas than in the wooded ones. Initial unpublished
data from a study using the same study plots comparing beetles from pitfall trap sampling
in September 2010 and May 2011 show this effect very clearly (Churchill 2011, unpublished Masters thesis).
Conservation of soil macrofauna biodiversity in the New Forest
The general conservation picture is of a highly complex mosaic of habitats across the
landscape, differing greatly in their sizes, microclimates, stability and ages. The habitats
123
3408
Biodivers Conserv (2012) 21:3385–3410
generally co-vary as follows: the least fragmented, climatically buffered habitats with the
greatest long-term stability are wooded and have a high alpha diversity, but relatively low
turnover between plots. In contrast, the most fragmented, climatically variable habitats,
undergoing the most short-term disturbance are open areas, particularly heaths. These open
areas have low alpha diversity, but potentially high between-plot turnover, due to the
dynamic succession of niches that are created by temporal heterogeneity (e.g. management
practice and seasonality). All these factors make responding to conservation threats at the
landscape level a potentially very challenging problem.
We identify three main threats to soil macrofauna biodiversity in the New Forest:
(1) changes in grazing pressure; (2) changes in management practice and (3) climate
change, as well as the complex interaction of these three factors.
Grazing
The presence of grazing animals is vital for maintaining the diversity of macrofauna across
the New Forest landscape. The open habitats are maintained by grazing and its reduction in
these open habitats risks succession to woodland. The increased input of organic matter
caused by increased grazing leads to soil eutrophication. In addition, colonisation of the
soil by earthworms increases incorporation of dung into the soil which will lead to
increased soil fertility and a probable corresponding decline in soil macrofauna diversity.
High grazing pressure also reduces the ability of seedlings to establish in woodlands. As
veteran trees die and fall over, few new trees are able to grow in their place. Over time,
woodland becomes more open and the canopy collapses. In order to maintain the current
area of open habitat, sustainable levels of grazing animals are required to prevent succession to woodland, while at the same time preventing the conversion of other habitats
(e.g. heath to grassland). Lower grazing pressure would allow more seedlings to establish
and allow woodland to regenerate.
Management
There is a long history of the persistence of heaths in the New Forest that owe their
existence to grazing animals (Tubbs 2001). Heaths require management to prevent the
ageing and decline of woody shrubs (Calluna vulgaris, Erica spp., Ulex spp.), to encourage
shrub re-growth as food for livestock and to prevent encroachment by scrub and trees
(Tubbs 2001). Historically, burning has been the main management technique used to clear
old woody growth of shrub species. More recently, cutting by machinery has been used for
a variety of reasons as an alternative technique for clearing old woody shrubs. A number of
studies have compared these two methods and their impact on biodiversity (e.g. Bullock
and Pakeman 1997; Sedlakova and Chytry 1999; Barker et al. 2004) but it remains unclear
which method is more suited for conservation purposes, particularly the effect on belowground biodiversity. Changes in the management practices of New Forest heaths could see
a decline in species density and diversity, particularly for the invertebrates which are the
focus of this study, and further work is required to investigate these management practices.
The plantation woodlands are in a state of flux because they are no longer managed for
timber due to a decline in the demand for oak. The inclosures face the possible threat of
timber extraction if oak timber is once again in demand. Felling of trees and associated
works would place a huge pressure on inclosure woods and lead to a rapid decline in soil
macrofauna biodiversity. There is no specific protection in place for the inclosure woodlands, except that which exists for all woodlands in the UK. The plantation woodlands have
123
Biodivers Conserv (2012) 21:3385–3410
3409
similar macrofauna species assemblages to the ancient woods (but not for other groups e.g.
lichens (Wolsley and Sanderson, unpublished data)) and form large continuous woodland
areas in the New Forest. Any timber extraction in these plantations would severely fragment woodland cover and could lead to a decline in connectivity with woodland species
pools, which aid re-colonisation of ancient woods.
Climate change
Wet heathland is a habitat of European importance and those in the New Forest are at risk
of drying out due to increased summer temperatures. The effect of drying out of wet heaths
can already be seen in sites such as Balmer Lawn. Tubbs (2001) states that c. 50 years ago
Balmer Lawn was a wet heath, with tussocks formed of Molinia caerulea, Sphagnum spp.
and Erica spp. Grazing of M. caerulea allowed silt to infiltrate the tussock causing them to
dry out, which in turn led to a decline in Erica spp. and Sphagnum spp. These tussocks are
now present as remnant grassy hummocks, forming a distinct microtopography on Balmer
Lawn and other sites. This conversion from wet heath to wet grassland with hummocks can
be seen in progress at a number of wet heath sites in our study. Changes in climate which
lead to drier summers and increased drying of wet sites will lead to still greater conversion
of wet heath to wet grassland habitats, particularly if grazing pressure increases. This
conversion to a wet grassland habitat is likely to lead to a decrease in macrofauna diversity
as soil fertility increases via the colonisation of earthworms and the increased incorporation of organic matter. Drying out of the wet heaths will also reduce the diversity of species
adapted to these waterlogged conditions.
In summary, the New Forest is a relatively stable landscape that enjoys protection under
a range of UK laws (e.g. national park status, SAC designation, SSSI designations). The
future of this internationally important landscape looks secure. However, changes in
grazing intensity or habitat management could have a negative impact on soil macrofauna
biodiversity across the landscape. Any future changes in stocking density or management
practices should be subject to rigorous ecological risk assessment, coupled with ongoing
monitoring to evaluate the impact of any changes on soil macrofauna. Increased summer
drought and increased temperatures as a result of changing climate could exacerbate any
potential changes in management practices.
Acknowledgments This work forms part of the ‘New Forest Quantitative Inventory’ and was funded by a
grant from the NHM’s Annual Fund and the NHM Entomology Department’s Departmental Investment
Fund. Thanks to David T. Jones, Barbara Smith, Jo Smith and two anonymous reviewers for comments on
the manuscript. Thanks also to the following volunteers for helping with field work and for sample sorting:
Keiron Brown, Kate Harrington, Cassius Morrison, Simon Powell, Samantha Ho, Judi Allette, Laura
McCoy, Matthew Dickinson, Benjamin Lawrence, Nicky Nicoll, Anna Platoni, Heather Mikhail, Salma
Mustafa, Ian Sosney and Gerardo Mazzetta.
References
Barker CG, Power SA, Bell JNB, Orme CDL (2004) Effects of habitat management on heathland response
to atmospheric nitrogen deposition. Biol Conserv 120:41–52
Borcard D, Legendre P, Drapeau P (1992) Partialling out the spatial component of ecological variation.
Ecology 73:1045–1055
Bullock JM, Pakeman RJ (1997) Grazing of lowland heath in England: management methods and their
effects on heathland vegetation. Biol Conserv 79:1–13
123
3410
Biodivers Conserv (2012) 21:3385–3410
Chambers BQ, Samways MJ (1998) Grasshopper response to a 40-year experimental burning and mowing
regime, with recommendations for invertebrate conservation management. Biodivers Conserv
7:985–1012
Chape S, Harrison J, Spalding M, Lysenko I (2005) Measuring the extent and effectiveness of protected areas
as an indicator for meeting global biodiversity targets. Philos Trans R Soc B Biol Sci 360:443–455
Colwell RK, Coddington JA (1994) Estimating terrestrial biodiversity through extrapolation. Philos Trans R
Soc B Biol Sci 345:101–118
Colwell RK, Rahbeck C, Gotelli NJ (2004) The mid-domain effect and species richness patterns: what have
we learned so far? Am Nat 163:E1–E23
Curry JP (1987) The invertebrate fauna of grassland and its influence on productivity. III. Effects on soil
fertility and plant growth. Grass Forage Sci 42:325–341
De Caceres M, Legendre P (2009) Associations between species and groups of sites: indices and statistical
inference. Ecology 90:3566–3574
Dimbleby GW, Gill JM (1955) The occurrence of podzols under woodlands in the New Forest. Forestry
28:95–106
Eggleton P, Inward K, Smith J, Jones DT, Sherlock E (2009) A six year study of earthworm (Lumbricidae)
populations in pasture woodland in southern England shows their responses to soil temperature and soil
moisture. Soil Biol Biochem 41:1857–1865
European Commission (2012) Natura 2000 network http://ec.europa.eu/environment/nature/natura2000/
index_en.htm. Accessed 12 March 2012
Flower P (1980) The management history and structure of unenclosed woods in the New Forest, Hampshire.
J Biogeog 7:311–328
JNCC (Joint Nature Conservation Council) (2011) http://jncc.defra.gov.uk/page-4. Accessed 08 December
2011
Krell F-T, Chung AYC, DeBoise E, Eggleton P, Giusti A, Inward K, Krell-Westerwalbesloh S (2005)
Quantitative extraction of macro-invertebrates from temperate and tropical leaf litter and soil: efficiency and time-dependent taxonomic biases of the Winkler extraction. Pedobiologia 49:175–186
Leps J, Smilauer P (2003) Multivariate analysis of ecological data using CANOCO. Cambridge University
Press, Cambridge
Maudsley MJ (2000) A review of the ecology and conservation of hedgerow invertebrates in Britain.
J Environ Manag 60:65–76
McIntyre S, Hobbs R (1999) A framework for conceptualizing human effects on landscapes and its relevance to management and research models. Conserv Biol 13:1282–1292
Newton AC (ed) (2010a) Biodiversity in the New Forest. Pisces Publications, Newbury
Newton AC (2010b) Fungi. In: Newton AC (ed) Biodiversity in the New Forest. Pisces Publications,
Newbury
Oksanen J, Guillaume BF, Kindt R, Legendre P, O’Hara RB, Simpson GL, Solymos P, Stevens MHH,
Wagner H (2011). Vegan: community ecology package. R package version 1.17–12. http://CRAN.
R-project.org/package=vegan
Pielou E (1966) The measurement of diversity in different types of biological collections. J Theoretical Biol
13:131–144
Sanderson NA (2010) Lichens. In: Newton AC (ed) Biodiversity in the New Forest. Pisces Publications,
Newbury
Sedlakova I, Chytry M (1999) Regeneration patterns in Central European dry heathland: effects of burning,
sod-cutting and cutting. Plant Ecol 143:77–87
Smith J, Potts S, Eggleton P (2008) The value of sown grass margins for enhancing soil macrofaunal
biodiversity in arable systems. Agric Ecosyst Environ 127:119–125
Stern R (2010) Bryophytes. In: Newton AC (ed) Biodiversity in the New Forest. Pisces Publications,
Newbury
Stewart AJA, New TR (2007) Insect conservation in temperate biomes: issues, progress and prospects. In:
Stewart AJA, New TR, Lewis OT (eds) Insect conservation biology: The 22nd symposium of the royal
entomological society. CABI, Wallingford
R Development Core Team (2011) R: a language and environment for statistical computing. R Foundation
for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL:http://www.R-project.org/
Trudghill S (1989) Soil types. A field identification guide. Field Stud 7:337–363
Tubbs CR (1997) The ecology of pastoralism in the New Forest. British Wildl 9:7–16
Tubbs CR (2001) The New Forest: history ecology and conservation, 2nd edn. New Forest Ninth Centenary
Trust, Lyndhurst
Webb NR (1989) Studies on the invertebrate fauna of fragmented heathland in Dorset, UK, and the
implications for conservation. Biol Conserv 47:153–165
123
Download