SEARCH

SEARCH BY CITATION

Keywords:

  • Caledonian pine forest;
  • invertebrate conservation;
  • land management;
  • local and regional processes;
  • soil biodiversity

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • 1
    Land-use intensification strongly influences biodiversity by altering habitat heterogeneity, the distribution of habitat types and their extent. This study explores these effects within mixed semi-natural/agricultural mosaic habitats in Scotland, examining the effect of land-use intensification on the soil macrofauna at point (m2), landscape (km2) and regional (> 1 km2) scales.
  • 2
    The soil macrofauna in six 1-km2 sampling areas (land-use units; LUU) were sampled using a combined hand-sorting and Winkler bag extraction technique. Within each LUU, 16 1-m2 samples were taken in each of 2 successive years. Each LUU had a mixture of land-use types, representing an agricultural intensification gradient.
  • 3
    The following hypotheses were tested: (i) the study area sustains a number of distinct habitats as defined by soil macrofaunal composition; (ii) a greater number of restricted range species are found in semi-natural habitats; (iii) local (point) species density is related to habitat type; (iv) overall levels of species richness per habitat at regional scales are related to species-area effects; and (v) landscape-level species density is correlated with habitat heterogeneity.
  • 4
    Initial analysis revealed five distinct habitat types: Caledonian forest (semi-natural pine forest), closed canopy woodland (pine plantation and broadleaved woodland), riparian habitats (wet woodland and grassland), pasture (improved grassland) and arable (crop fields).
  • 5
    As hypothesized, the Caledonian habitat contained a greater number of restricted-range species than the other habitats. However, conifer plantations contained more restricted range species than expected, given their anthropogenic origin. Species density per m2 was most strongly affected by habitat type. At the regional level, the size of the species pool was correlated with the size of habitat areas. There were more species overall in LUU with greater habitat heterogeneity.
  • 6
    Synthesis and applications. Caledonian pine forests have high species densities and contain species of conservation value. Mixed conifer plantations also appear to have a surprisingly high invertebrate conservation value. In contrast, intensively managed agricultural habitats have low species densities and conservation value. Generally, mixed land-use areas have higher species densities than single land-use areas. This emphasizes the need for careful management of forest systems within the matrix of agricultural habitats to maximize landscape diversity.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Agricultural intensification has profoundly affected European landscapes over the last few thousand years, predominantly by converting large areas of natural woodland ecosystems into farmland (Matson et al. 1997), leading to shifts in the range of species and local extinctions (Myers & Knoll 2001). The loss of species will inevitably have consequences for diversity, community structure and ecosystem processes (Vazquez & Simberloff 2003). No component of ecosystems is potentially more important, both ecologically and economically, than soils and their associated biodiversity (Giller et al. 1997).

Soils contain a highly diverse community of organisms with a range of ecological functions (Giller et al. 1997). Soil processes rely on this community to provide them with vital services, including nutrient cycling (Lavelle 1996), soil structure stabilization (Giller et al. 1997), pest control and soil detoxification (Altieri 1999). Land-use intensification affects the quality and quantity of the soil communities and such soil modifications may be very long-lived (Dupouey et al. 2002). Biodiversity changes therefore have the potential to alter greatly the biological regulation of decomposition and nutrient availability in the soil (Matson et al. 1997; Altieri 1999). Consequently these biological functions commonly have to be compensated for in intensive agriculture by the use of fertilizers and mechanical tillage (Giller et al. 1997).

Within Scotland, c. 80% of the land area is classified as agricultural with only c. 16% forested. Of this forested area only c. 4% is classified as native woodland (Haines-Young et al. 2000), of which Caledonian pine forest represents one of the most important natural habitats (Summers, Moss & Halliwell 1994). The recent increased concern in environmental issues makes it particularly pressing to document the exact status of elements of biodiversity under different land-use regimes.

Within each habitat there are a number of ecological factors that influence the size and structure of species assemblages. Two of the most important of these are habitat quality and habitat area. Broadly, we can imagine that where habitat quality (e.g. plant productivity) is high then point species density is high, as there are abundant resources that can be partitioned among a large number of species (Whittaker, Willis & Field 2001). This relationship, however, may be humped, with diversity low at very high productivities, because of increased dominance of highly competitive species. (Mittelbach et al. 2001).

Habitat area, in contrast, works to reduce overall species richness in smaller habitat fragments (Storch, Izling & Gaston 2003). This species–area effect is well established (Cam et al. 2002). Each habitat is therefore likely to have its own scale- and resource-dependent dynamics, and the habitats as a whole will feed into the general landscape according to both these dynamics and the mix of habitats making up the landscape. However, most human-influenced landscapes are a mosaic, such that a number of habitats will be present across the landscape. There is a generally positive relationship between landscape species richness and habitat heterogeneity (Tews et al. 2004).

Unravelling this complex mix of factors is clearly a key requirement for the management and conservation of biodiversity in mixed land-use areas. We examined these factors in a serious of habitat mosaics in north-east Scotland, examining assemblages of soil macrofauna at point (1 m2), landscape (1 km2) and regional (> 1 km2) scales.

We tested the following specific hypotheses. (i) There are, within the landscape, a number of habitat types that can be clearly defined by the relationship between vegetation type and macrofaunal composition. We expect these habitats to coincide with some of the obvious land uses in the area (e.g. pine woodland, arable, pasture). (ii) We expect more restricted range species within semi-natural habitats than within highly disturbed ones. (iii) Point (local) macrofauna species density will be influenced by habitat type, as habitat types will differ in habitat quality. (iv) Regional (e.g. > 1 km2) species richness within habitats will be influenced by the overall size of the habitat within the region. (v) Within a landscape, species richness will be positively correlated with habitat heterogeneity.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Six 1-km2 squares (land-use units; LUU), representing a land-use gradient of increasing intensity and frequency of disturbance (Table 1; see Appendix S1), were set up in Aberdeenshire, Scotland: a region of varying altitude, with upland Scots pine forests Pinus sylvestris Linnaeus along with lowland deciduous forests and farmed land. These LUU were all within 10 km of the most distant neighbouring LUU. Each LUU was selected using a pre-defined set of criteria corresponding to its predominant land use. Sampling at each LUU was on a regular 4 × 4 grid, giving 16 sampling points each separated by 200 m from orthogonally neighbouring points and the edge of the 1 km2. All data were gathered at these sampling points. Sampling was conducted in May 2001 and May 2002, with the same sampling points being used on each occasion (i.e. there were 32 samples in total per LUU). These LUU (Table 1) formed a standardized series for the European Union (EU) BIOASSESS (Framework V) programme run throughout Europe. At each sampling point soil macrofauna (invertebrates visible to the naked eye, excluding mites and collembola) were collected using two methods.

Table 1.  Descriptions of the LUU across the entire sampling area
LUULand usesDominant speciesHumus typeNameDecimal latitude/longitude
  1. Authorities: Juniperus communis Linnaeus; Calluna vulgaris Linnaeus; Vaccinium myrtilus Linnaeus; Alnus glutinosa (Linnaeus); Pinus contorta Douglas ex London; Betula pendula Roth; Sorbus aucuparia Linnaeus; Acer pseudoplatanus Linnaeus; Prunus padus Linnaeus.

1Semi-natural pine woodland (open canopy)Scots pine Pinus sylvestris, juniper Juniperus communis, heather Calluna vulgaris, bilberry Vaccinium myrtilis, Sphagnum spp.Podzolic morGlen Tanar57·019057 N, 02·859627 E
2Conifer plantation (closed canopy)Scots pine, larch Larix spp., lodgepole pine Pinus contorta, sitka and other spruces Picea spp.MorBalfour Wood57·047952 N, 02·752446 E
3Mixed land use dominated by woodland (mixture of plantation, mixed regeneration woodland and permanent pasture)Plantation: silver birch Betula pendula, Scots pine, larch. Woodland: Scots pine, birch, rowan Sorbus aucuparia, cherry Prunus spp. Pasture: untilled, unfertilized, cattle and horse grazedMull and mor mixtureWoodend Farm57·015483 N, 02·690494 E
4Mixed land use, not dominated by one vegetation type (mosaic of pasture, arable, plantations and native deciduous woodland)Pasture: improved. Arable: barley. Plantations: Scots pine, larch, spruces. Woodland: birch, sycamore Acer pseudoplatanusPredominantly mullsCandycraig Farm57·068030 N, 02·872091 E
5Mixed land use dominated by pasture (mixture of pasture, arable, small plantation)Pasture: rotational. Arable: barley. Plantation: sitka spruce.Predominantly mullsTillyfruskie Farm57·030556 N, 02·810504 E
6Mixed land use dominated by agriculture (mixture of arable, pasture, and a small riparian woodland)Arable: barley, turnips. Riparian woodland: alder Alnus glutinosa, silver birch, rowan, bird cherry Prunus padusPredominantly mullsCraskins Farm57·143960 N, 02·810504 E

hand sorting

A steel frame quadrat (25 × 25 × 15 cm deep) was forced into the ground and the litter and soil were excavated and hand sorted on trays for macrofauna. When the base of the quadrat (frame) had been reached, 1·5 L of a 0·02% formalin solution were added to the exposed soil surface and all macrofauna emerging in a 10-min period were collected.

winkler method

A 1-m2 sample of litter and the top layer of soil (2–3 cm) were collected into bags at each sampling point. At the woodland sites the larger parts of the litter were excluded using a 1-cm2 wire mesh sieve. At the pasture sites the turf was dug up so that soil and root material could be separated from the grass. To ensure minimal disturbance to the crop at the arable sites, an equivalent area of soil was collected from between crop rows. Invertebrates were extracted from the sieved litter and soil using Winkler bags (supplied by F. H. Winkler,Vienna, Austria), with the samples being hung for 4 days.

Samples for the two methods were taken within 1 m of each other. Sample data from quadrats and Winkler bags were pooled for analytical purposes and standardized to give per m2 values.

All sampled macrofauna specimens were identified to named species, except larval forms, adult Hemiptera and adult non-ant Hymenoptera, none of which could be consistently identified to species. All the identification work was undertaken at The Natural History Museum, London, UK.

environmental variables

Continuous variables

We measured environmental variables at each sampling point that were likely to influence invertebrate assemblage structure. We determined canopy density (%) using a spherical densiometer (Forest Densiometers, Bartlesville, OK). Soil pH and soil temperature (°C) were measured using a Hanna HI 9024 pH meter (Hanna Instruments Ltd, Bedfordshire, UK); three measurements were taken and averaged to give a mean value for each sampling point. Soil moisture (m−3 m−3) was determined using a Delta-T theta probe type ML2x (Delta-T Inc., Cambridge, UK) and theta meter HH1 and calibrated using the humus type (mull, moder, mor or peat). The depth of the O horizon (cm) was measured in all of the soil pits.

Nominal variables

We classified the environment at each sampling point using a series of nested binomial (presence/absence) variables (Table 2) based on the centre of each 1 m2. These were broadly primary land use, secondary major subsets of primary land use, tertiary major vegetation types (tree species, grazing animals, pasture and agricultural types), understorey vegetation, ground level vegetation and soil type. Each nominal environmental variable enters into each analysis as a separate variable, although some variables (but not all) are logically nested within each other (e.g. birch is nested within deciduous which is nested within woodland).

Table 2.  Nominal land-use variables used in the analyses. Presence of any of the features within the 1-m2 area (including overhanging trees) counted as a ‘presence’ datum. ‘Riparian’ was defined as being within 5 m of running or standing water
PrimarySecondaryTertiaryUnderstoreyGround levelSoil type
WoodlandConiferPlantationHeatherAlgaeMor
RiparianDeciduousScots pineBrackenBilberryModer
PastureGrazedLodgepoleJuniperSphagnumMull
ArableUngrazedLarchBroomGrassGley
SitkaGorseCloverPeat
BirchNettlesMoss 
RowanThistleBog 
Beech Barley 
Sycamore Bare (turnips) 
Alder   
Sheep   
Cows   
Set aside   
Silage   

statistical analysis of habitat structure

The habitat analyses were conducted using a three-step process. (i) Multivariate methods were used to assess (a) habitat heterogeneity and (b) the relationship between environmental variables and species composition, to reveal coherent clusters, and so test hypothesis i. (ii) Each sampling point was classified into a habitat derived from (i) that reflected coherent invertebrate units compositionally. (iii) This habitat classification was used as a framework for examining species distribution (hypothesis ii), local species density (hypothesis iii), regional species density (hypothesis iv) and habitat heterogeneity (hypothesis v) patterns within and between the habitats and LUU.

Data from both years were entered into analyses simultaneously as m2 samples. Species abundances were log(x + 1) transformed for all statistical analyses. Neither species density nor abundance (across all taxa) were significantly different in the 2 years, so year was not used as a covariable in univariate analyses.

The environmental variables were initially ordinated using correspondence analysis (Leps & Smilauer 2003), with continuous variables treated as vectors (rather than as centroids). Heterogeneity of each LUU (LUU heterogeneity) was calculated by the average Euclidean distance between sample points in ordination space within each LUU.

We used partial canonical correspondence analysis (pCCA) to examine the relationship between species composition per sample and the environmental variables. (Year was treated as a covariable in the CCA analyses because although there were some significant compositional differences between the years the study was not intended to examine annual differences.)

We tested the statistical significance of the influence of each environmental variable using Monte Carlo permutation tests (canoco version 4·5; ter Braak & Smilauer 2003) with 999 random draws for each test. Variables were tested both marginally and conditionally. Marginal tests are where each variable is tested separately without reference to the other variables. Conditional tests are where the most influential variable (as indicated by marginal tests) is tested first then placed into the covariables, and then the next most influential is tested and also placed in the covariables, and so on, until no more variation remains. In subsequent ordination diagrams conditionally significant variables are plotted actively (i.e. they are used in the analysis) while variables that are only marginally significant are plotted passively (they are not used in the analysis but placed in the ordination diagrams after the analysis is completed).

habitat classification (hypothesis i)

From the resulting ordination diagram showing only conditionally significant environmental variables, we derived habitat groups forming clear clusters of samples in ordination space. Canonical covariates analysis (CVA) was used to determine the best fitting classification. Initially the classification was estimated by eye from the ordination diagram, subsequently alternative classifications were run through CVA until the habitat grouping with the best overall fit was uncovered, and we then used these groupings as habitat types.

These habitat types were then used in subsequent analysis of species density patterns. This classification step was essential because the original kilometre squares were not homogeneous and so could not be treated as distinct habitat types. Alternative methods using a priori within-LUU habitat classifications would be potentially arbitrary and would risk missing the main environmental factors structuring the invertebrate assemblage. In addition, areas that had apparently uncharacteristic vegetational structure within habitats (e.g. cleared grassy areas within closed canopy forest) were commonly found in ordination analyses to group within the wider habitat grouping (e.g. closed canopy forest) rather than with their apparent habitats (e.g. pasture). Inclusion of those samples into their ‘apparent’ habitat types would therefore introduce statistical noise into the data set.

distribution patterns (hypothesis ii)

Species were classified as (i) widespread across Britain, (ii) showing a clear northern British distributional bias or (iii) conservationally notable, as defined by Hyman & Parsons (1992). Distributional data were obtained from Eason (1964), Hopkin (1991), Hillyard & Sankey (1989), Sims & Gerard 1999), Morris (1990, 1997, 2002), Luff (1998), Johnson (1993), Brendell (1975), P. M. Hammond (personal observation), and Harvey, Nellist & Tefler (2002). Categories (ii) and (iii) together are subsequently referred to as restricted range species. We examined proportions of distributional categories per habitat and whether LUU heterogeneity was correlated with the number of restricted range species per km2.

species density patterns (hypotheses iiiv)

We examined environmental correlates of per m2 species density using an ancova (using statistica, StatSoft Inc. 1998). Independent variables were LUU heterogeneity (covariate) and habitat type (factor). Total species density per LUU was also derived and correlated with LUU heterogeneity. Species accumulation curves per habitat were constructed using the Biodiversity Pro program (McAleece 1997). Two sorts of curves were produced: (i) species richness per number of individuals (rarefaction curves) and (ii) species richness per number of m2 samples. Rarefied species richness per habitat type was also calculated using Biodiversity Pro.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

The pCCA ordination (with only conditionally significant environmental variables included) showed a clear structure, with high eigenvalues for the first two axes (Fig. 1; see Appendix S2). There was a clear axis 1 gradient of decreasing soil pH and soil temperature (Fig. 1). Axis 1 therefore separated the woodland point samples from arable and pasture, with the riparian sites between these two extremes. The arable and pasture sites clearly had very similar environmental conditions: high soil pH (c. 6–7), high soil moisture and temperature, low canopy cover, shallow O horizon and predominantly mull soils. In contrast, the woodland sites all had lower soil moisture, lower pH (c. 3–5) and lower temperature, moderate to high canopy cover, deeper O horizons and moder or mor soils. The riparian samples were intermediate in soil conditions. The relatively close proximity of the grass centroid to the riparian one was artefactual, as a range of sites across the entire axis 1 gradient had grass ground cover.

image

Figure 1. Scatter diagram showing pCCA ordination of conditionally significant environmental variables (bold italic text, crosses, solid line arrows) with marginally significant environmental variables (roman text, asterisks), plotted passively. Variables as in Table 2. Symbols indicate centroids and arrows indicate vectors. Eigenvalues for the first two axes are 0·720 and 0·421, respectively, and species–environment correlations are high (0·980, 0·954).

Download figure to PowerPoint

Axis 2 broadly separated the semi-natural Scots pine woodland sites from the other sites (Vanbergen et al. 2005), partly on soil type and partly on canopy cover. The Scots pine woodlands were on mor (and peat) soils with deep O horizons. The remaining woodlands (including Scots pine plantations) had a varied mix of tree species but all had high canopy cover, either naturally (as with the native deciduous trees) or artificially (as with the conifer plantations). They had shallower O horizons and a greater proportion of moder and mull soils.

Ordination including only conditionally significant environmental variables (Fig. 1) gave five broad non-overlapping clusters of samples (Fig. 2). This supported the hypothesis that the study area could be split into discrete habitat types (hypothesis i).

image

Figure 2. Biplot showing envelopes for the five habitats and environmental variables (nominal variables, crosses, not labelled but as in Fig. 1). Envelopes are drawn around non-overlapping sample points from a CCA biplot (samples × environmental variables). Note that the ‘wood’ (i.e. deciduous woodland) environmental centroid sits between the Caledonian and closed canopy habitat envelopes.

Download figure to PowerPoint

The first cluster was Caledonian pine forest samples with a heather understorey, low soil pH and low canopy cover. These had open Scots pine canopies, juniper understorey and bilberry ground cover in patches, and they were on predominantly mor humus soils, with low pH, moisture and temperature. The second cluster consisted of closed canopy samples with a bracken understorey, low pH and a high canopy cover of birch and assorted coniferous trees. A major subset of this cluster consisted of managed Scots pine plantation with dense canopy cover and generally without bracken understorey. The third cluster consisted of riparian samples, with intermediate soil pH, relatively high soil moisture content and canopy cover. This cluster included samples under beech (Fagus sylvestris Linnaeus) (single sample) and rowan, as well as samples with a broom (Bromus scoparius Linnaeus) understorey and clover at ground level. This was a rather heterogeneous class. The fourth cluster consisted of pasture samples, with high soil pH and low canopy cover. This cluster included grazed and ungrazed samples, as well as silage samples, which were intermediate in composition between arable and pasture plots. The final cluster consisted of arable samples with high soil pH and low canopy cover and included a single sample under set aside that was intermediate in composition between silage and arable.

Subsequently we treated these five main clusters of samples (defined by their macrofaunal composition) as habitats, as they consisted of clearly defined land-use types separated by the CCA using the invertebrate species data (as predicted in hypothesis i). These habitat groupings depended only on species compositional data; overall species densities did not affect them. They are subsequently referred to as the Caledonian, closed (forest habitats), riparian (forest/pasture intermediates), pasture and arable (non-forest habitats) habitats, in rank order of position along axis 1 of the pCCA. There were very low compositional similarities between the forest and non-forest habitats (Table 3; there was only a c. 1% similarity in species representation between the arable and Caledonian habitats). The number of habitats within LUU ranged from one to three (Table 4).

Table 3.  Percentage similarities between habitats (Jaccard metric based on presence/absence data)
HabitatClosedRiparianPastureArableAverage similarity
Caledonian261011< 1 9
Closed 2724   721
Riparian  33   1622
Pasture       2022
Arable    11
Table 4.  LUU heterogeneity and numbers of point samples taken in each LUU for each habitat class. Boxed numbers are modes for each LUU. Note that sampling is uneven because of the original sampling design that was based on mixed-habitat kilometre squares
LUUHeterogeneityArablePastureRiparianClosedCaledonian
10·713 0 0 0 230
20·266 0 0 032 0
30·917 0 6 618 2
40·877 01610 4 0
50·1451019 3 0 0
60·2671713 2 0 0

We identified 338 species from our samples. Beetles, earthworms and spiders were found in all habitats. Centipedes, harvestmen, ants and millipedes were not sampled in the arable habitat. Woodlice were not sampled in the pasture or arable habitats. Earwigs were only sampled in the pasture habitat. Individual groups bear further discussion.

beetles

We sampled 178 species of beetles across the six LUU (see Appendix S3). These represented 71% of the total species sampled across our target groups. Of these, 60% were Staphylinidae. Beetles were much more evenly distributed across habitats than the other groups. In addition, beetle species were often rare in our samples: singletons represented 53% of the beetle species.

As with the other invertebrate groups, most beetle species are widespread throughout Britain but there are some that are more limited in geographical range and/or habitat. These include a number of nationally scarce (as listed in Hyman & Parsons 1992) beetles (Table 5). The distribution of these species across habitats was not even: half of the 16 species were found only in densely canopied conifer plantation within the closed habitat. Furthermore two beetle species that were found in conifer plantations are apparently relatively recent colonizers of Britain: Laricobious erichsoni Rosenhauer (Derodontidae) and Atomaria turgida Erichson (Cryptophagidae) (P. M. Hammond, personal observation).

Table 5.  Beetle species with restricted or recently expanding ranges. R, recent colonizers. Other abbreviations as in Hyman & Parsons (1992): N, nationally scarce; IK, insufficiently known; (p), pine plantation; (s), silage
SpeciesFamilyHabitatStatus
Leistus rufomarginatus CarabidaeClosed (p)R
Pterostichus oblongopunctatus CarabidaeClosed (p)N
Longitarsus suturalis ChrysomelidaePasture (s)N
Mantura rustica ChrysomelidaeArableN
Atomaria hislopi CryptophagidaeClosedIK
Atomaria ornate CryptophagidaeClosed (p)IK
Atomaria turgida CryptophagidaeClosed (p)R
Otiorhyncus scaber CurculionidaeClosed (p)N
Laricobius erichsoni DerodontidaeCaledonianR
Paraphotistus impressus ElateridaeClosed (p)N
Euryptilium saxonicum PtilidaeClosed (p)N
Aphodius nemoralis ScarabaeidaeClosed (p)N
Philhygra britteni StaphylinidaePasture (s)N
Philhygra scotica StaphylinidaeCaledonianN
Euryporus picipes StaphylinidaeCaledonianIK
Mycetoporus despectus StaphylinidaeRiparianN
Omalium rugatum StaphylinidaeClosed (p)N
Proteinus crenulatus StaphylinidaeClosed (p)N

spiders

Spiders were the second most species rich group, with 75 species (22% of the total). Most of the spiders found are known from a wide range of habitats (Harvey, Nellist & Tefler 2002), especially those Linyphiidae species (making up 16% of the spider species sampled here) that are ‘aeronauts’ and thus can balloon across whole landscapes (e.g. Erigone spp.) Although there were no notable species among the spiders, there was a relatively high proportion of ‘northern restricted’ species (22% of spider species).

earthworms

Earthworms were mostly sampled in non-woodland habitats on mull soils, with only Dendrobaena octaedra (Savingy), an acid-tolerant species (Sims & Gerard 1999), extending into forest areas on mor or moder soils. Other species were predominantly associated with pasture, with only Allolobophora chlorotica (Savingy) better represented in the arable habitat. None of the earthworm species sampled were of conservation importance; all were very common and widespread soil invertebrates in Britain.

other invertebrates

Most of these species (millipedes, centipedes, woodlice, opiliones and earwigs) were found within one of the three woodland habitats, with millipedes especially well represented in the coniferous woodlands. Some of these species are normally associated with a wider range of habitats and it may be that at warmer times of the year they might have been sampled more widely across the habitats. None of these invertebrate species had a restricted northern range.

restricted range species

The proportion of restricted range species within habitats increased proportionally along axis 1 of the pCCA (logistic regression χ2 = 26·5, P < 0·00001) but not along axis 2 (χ2 = 0·39, P > 0·5). The Caledonian habitat contained the highest proportion of restricted range species and arable the lowest proportion, thereby supporting hypothesis (ii). This implieds, given the relationship between the measured environmental variables and the pCCA axis 1, that the conservation value of soil invertebrates within habitats increased as they became more acidic and when they were on more infertile soils (i.e. in undisturbed areas).

Within LUU there was no significant correlation between LUU heterogeneity and absolute numbers of restricted range species (r2 = 0·395, P > 0·1) or between total species richness per km2 and number of restricted range species (r2 = 0·43, P > 0·09). Additionally, there was no correlation between the proportion of restricted range species and LUU heterogeneity (logistic regression χ2 = 0·851, P= 0·37). Therefore, proportions of restricted range species seemed dependent only on habitat type and not on the habitat heterogeneity of the landscape that they were in.

factors affecting local, landscape and regional species densities

As expected, habitat type explained the largest amount of variation in the per m2 species density data (ancova, F= 7·91, P < 0·01; % variance explained = 16%) (supporting hypothesis iii), with LUU heterogeneity explaining a small but still significant amount of the remaining variation (F = 5·87, P < 0·05; % variance explained = 4%). The main difference between the habitats was between the non-forested habitats and the riparian/forested habitats (Fig. 3). LUU heterogeneity was also significantly correlated with total LUU species density (r2 = 0·672, P < 0·03; Fig. 4), as predicted by hypothesis (v). The presence of multiple habitats with differing species compositions appeared to be sufficient explanation of the larger number of species in more heterogeneous landscape mosaics.

image

Figure 3. Species density (a) and abundance density (b) of invertebrates in the five habitats. Point, mean; box, standard error; whiskers, standard deviation. Samples with the same letter do not differ significantly from each other in an anova.

Download figure to PowerPoint

image

Figure 4. Plot of LUU heterogeneity against LUU species density. Inset shows species densities for the six LUU in order LUU1 to LUU6, left to right.

Download figure to PowerPoint

The five main habitats showed large differences in local (per m2) species density and abundance (Fig. 3). The forested habitats generally had higher mean abundance and species densities. Species accumulation curves showed distinct responses to increases in sample sizes (individuals and samples) (Fig. 5). The riparian sites accumulated species most quickly, even though they only had intermediate abundance densities. The pasture and arable habitats also accumulated species quickly per individual sampled, but in the case of the arable abundance density was so low that the rate of accumulation of species per sample was comparatively slow (i.e. a very large number of samples would have to be taken to reach a high species richness). Despite its high mean abundance densities, the Caledonian habitat accumulated species the least rapidly with individuals sampled.

image

Figure 5. Species accumulation curves by (a) individuals and (b) m2 point samples. Plots constructed using Biodiversity Pro (McAleece 1997).

Download figure to PowerPoint

These results were broadly consistent with hypothesis (iv), that the size of regional species pools is related to the area of the habitat regionally. Caledonian pine forest is poorly represented in Scotland (Table 6) and is therefore likely to have a relatively small species pool. Therefore, although each individual Caledonian habitat m2 could sustain large numbers of individuals, the regional species pool appeared to be too small to give asymptotic species richnesses that were as high as the other habitats. In contrast the arable habitat had very low numbers of individuals per sample but the local species pool was drawn from a regional pool covering a large area of lowland Britain and northern Europe that is presumably very species rich. The other habitats were intermediate in local abundance density and land cover area, except for the riparian habitat which seemed to have a higher regional richness than would be expected from its very small land cover area. This may have been a predominantly ecotone effect, because the riparian habitat had both pasture (grass) and woodland (waterside trees) elements.

Table 6.  Land cover data for Great Britain (GB) (Haines-Young et al. 2001). Rarefied species richness estimates for each habitat type derived using Biodiversity Pro (McAleece 1997)
HabitatGB area (thousand ha)GB percentage areaRarefied species richness
Pasture (improved grassland)5482   23·759·11
Arable (including horticultural)5249   22·749·79
Riparian 850    3·758·64
Closed (broadleaved, mixed, conifer)2845   12·244·87
Caledonian (semi-natural pine forest)  16< 0·137·47

Mean species density per m2 was also probably linked to the presence or absence of soil horizons from the profile. The forested habitats had a horizon (the leaf litter layer) that was presumably reduced (pasture) or almost absent (arable) in the other two habitats.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

These results suggest that diversity patterns in Scottish matrix landscapes are inherently scale-dependent (Table 7). Point (m2) local species density and composition are linked predominantly to habitat type (hypotheses i and iii) and, to a lesser degree, larger scale habitat heterogeneity (hypothesis v). Landscape (km2) species density is linked to habitat (LUU) heterogeneity (hypothesis v). Regional (> 1 km2 up to Scottish- or UK-wide) species richness may be linked to species–area effects (hypothesis iv).

Table 7.  Summary of results
HabitatAbundance density (m−2)Species density (m−2)Rate of species accumulationRegional pool areaConservation value
CaledonianHighHighLowLowHigh
ClosedHighHighModerateHighModerate
RiparianHighHighModerateLow/moderateModerate
PastureLowModerateHighHighLow
ArableVery lowLowHighHighVery low

Although it was unsurprising that habitat type was the key factor explaining species compositional difference in our point data, it was striking to find an almost completely different set of species in the Caledonian forest than from the arable land (Vanbergen et al. 2005), particularly given the small geographical area sampled. The key environmental gradients associated with the compositional differences were soil pH and canopy cover; pH is a well established correlate of differential soil processes (Dubbin 2001) and here it can be broadly generalized into earthworm-dominated mull soils and earthworm-poor mor soils. This difference is partly because of inherent soil properties (upland acid soils in contrast to lowland alkaline soils) and partly because of human clearance of land, where agricultural intensification leads to more alkaline, more fertile, soils. Canopy cover is associated with differences in ground cover for the woodland sites, with bracken and/or bare litter at ground level in pine plantations and mature deciduous woodland (open habitat), and heather and moss in the Caledonian habitat.

The semi-natural pine forests (Caledonian) held the highest abundance densities in our data set, as well as having the highest proportion of northern-biased species (hypothesis ii). However, the regional species pools seemed to be small, and the habitat must be considered not only of high conservation value but potentially threatened by future environmental changes because of its restricted extent. In contrast, the other habitats are all more widespread across Britain, form more of a compositional continuum, and are of lower apparent conservation value.

The conifer plantation sites (all within the closed habitat) contain a high proportion of notable species, but none of these have high abundances. The habitat may, however, have been undervalued as a conservation refuge for invertebrates that require dense canopies with little understorey vegetation. The enormous increase in coniferous plantations in the 1970s and 1980s may also have provided a suitable habitat for beetle colonization from the continent (e.g. the cryptophagid Atomaria turgida). Other studies broadly support this finding that conifer plantations can contribute significantly to invertebrate biodiversity (Humphrey et al. 1999; Ozanne et al. 2000). Furthermore, a parallel survey of the carabid fauna along this land-use gradient showed that species richness was higher in the plantation site (LUU2) than in the semi-natural site (LUU1) and that the plantation supported a different species assemblage associated with closed canopy forest (Vanbergen et al. 2005).

The other closed canopy forest types sampled here (predominantly birch within the closed and riparian habitats) have broadly similar species compositions to the plantation sites. Wetter deciduous woodland, however, tend to have more open canopies and groups within the riparian habitat.

The upland acid soil environments have particularly high invertebrate conservation value. In contrast, heterogeneity in lowlands clearly increases km2 (LUU) species richness, but predominantly through increased numbers of commonly distributed species (see above). This effect may therefore be the result of increased numbers of ‘tourists’ rather than ‘residents’. However, in all cases predominantly non-arable landscapes (with low pH and less fertile soils) have higher conservation value than predominantly arable landscapes.

management recommendations

This study and others show that the Caledonian pinewood habitat offers an important refuge for species with restricted distributions (Vanbergen et al. 2005), and understanding the regeneration dynamics of old-growth Caledonian pinewood forests is crucial to the future management and conservation of this habitat. These data (Beaumont et al. 1994; Summers, Moss & Halliwell 1994; Vanbergen et al. 2005) show that such habitats are a key conservation priority at both local and regional scales. These areas are now afforded a degree of protection under UK and EU directives but, given their limited cover and the likelihood of future climate change, they are possibly both too isolated from each other and in total too small to offer a sustainable, long-term refuge for rarer species. This current isolation of Caledonian pinewood fragments is important given the variety of ways in which these forests have been managed. For example, the suppression of fire over the last 200 years means that a naturally occurring disturbance that promotes regeneration and an uneven age structure has been excluded. However, until the natural forested areas are of sufficient size, the use of fire for management must be limited as the risk to small fragments may be too great (Nixon & Clifford 1994). Furthermore, historical selective felling of mature trees has led to an even aged structure that has significantly altered the natural structure of the forests (Nixon & Clifford 1994). To extend the area and/or connectivity of Caledonian pinewood would ensure the persistence of species of restricted distribution (Peterken, Baldock & Hampson 1995) by facilitating the use of natural processes (e.g. fire) in habitat management and by species–area effects.

This study, and others (Vanbergen et al. 2005), has shown that conifer plantations with a dense closed canopy can support elements of forest-adapted communities. Therefore these commercial forests potentially have an important role in biodiversity management. However, it is important to note that old-growth forest often supports whole assemblages (e.g. the specialist dead-wood insect community) that, while not completely absent from younger stands, do require an age-structured succession of dead wood to breed and sustain viable populations (Welch 1986). In the longer term, woodland structure should diversify as commercial wood extraction proceeds, but within the context of the current management policy that promotes biodiversity in production forests (Scottish Forest Strategy, Forestry Commission 2000).

Our data suggest that a mixed-use management approach to the landscape can maximize soil invertebrate diversity, with even relatively small areas of habitat (e.g. riparian woodland) adding disproportionately to landscape species density. We recommend that plantations should be situated close to semi-natural stands and open areas, to increase overall α-diversity and allow better persistence of both forest (natural and plantation) and open communities in a landscape matrix (Buse & Good 1993; Petit & Usher 1998; Ings & Hartley 1999).

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Thanks to Ben Woodcock, Carolyn Dawson, Liz Wickens and Chantal Beaudoin for assistance with fieldwork, and Kelly Jackson for assistance with logistics and ant identifications. Thanks also to Mick Marquiss, Allan Watt and three anonymous referees for comments on an earlier draft and to the landowners for their permission to carry out this work. This work was funded in part by the BIOASSESS project funded by the European Union (contract no. EVK2-CT-1999-00041).

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Altieri, M.A. (1999) The ecological role of biodiversity in agroecosystems. Agriculture, Ecosystems and Environment, 74, 1931.
  • Beaumont, D., Dugan, D., Evans, G. & Taylor, S. (1994) Deer management and tree regeneration in the RSPB reserve at Abernethy forest. Our Pinewood Heritage Conference Proceedings 1994 (ed. J.R. Aldous), pp. 186195. Forestry Commission, RSPB, Scottish Natural Heritage, Inverness, UK.
  • ter Braak, C.J.F. & Smilauer, P. (2003) canoco for Windows, Version 4.51. Wageningen, The Netherlands.
  • Brendell, M.J.D. (1975) Handbook for the Identification of British Insects, Coleoptera. Tenebrionidae, Volume 10. Royal Entomological Society, London, UK.
  • Buse, A. & Good, J.E.G. (1993) The effect of conifer forest design and management on abundance and diversity of rove beetles (Coleoptera: Staphylinidae): implications for conservation. Biological Conservation, 64, 6776.
  • Cam, E., Nichols, J.D., Hines, J.E., Sauer, J.R., Alpizar-Jara, R. & Flather, C.H. (2002) Disentangling sampling and ecological explanations underlying species–area relationships. Ecology, 83, 11181130.
  • Dubbin, W. (2001) Soils. The Natural History Museum, London, UK.
  • Dupouey, J.L., Dambrine, E., Laffite, J.D. & Moares, C. (2002) Irreversible impact of past land use on forest soils and biodiversity. Ecology, 83, 29782984.
  • Eason, E.H. (1964) Centipedes of the British Isles. Frederick Warne & Co. Ltd, London, UK.
  • Forestry Commission (2000) Scottish Forestry Strategy. Forestry Commission. http://www.forestry.gov.uk .
  • Giller, K.E., Beare, M.H., Lavelle, P., Izac, A.-M.N. & Swift, M.J. (1997) Agricultural intensification, soil biodiversity and agroecosystem function. Applied Soil Ecology, 6, 316.
  • Haines-Young, R.H., Barr, C.J., Black, H.I.J., Briggs, D.J., Bunce, R.G.H., Clarke, R.T., Cooper, A., Dawson, F.H., Firbank, L.G., Fuller, R.M., Furse, M.T., Gillespie, M.K., Hill, R., Hornumg, M., Howard, D.C., McCann, T., Morecroft, M.D., Petit, S., Sier, A.R.J., Smart, S.M., Smith, G.M., Stott, A.P., Stuart, R.C. & Watins, J.W. (2000) Accounting for Nature: Assessing Habitats in the UK Countryside. Centre for Ecology and Hydrology, Huntingdon, UK.
  • Harvey, P.R., Nellist, D.R. & Tefler, M.G. (2002) Provisional Atlas of British Spiders (Arachnida, Araneae), Volumes 1 and 2. Biological Records Centre, Huntingdon, UK.
  • Hillyard, P.D. & Sankey, J.H.P. (1989) Harvestmen: Keys and Notes for the Identification of the Species, 2nd edn. EJ Brill, Leiden, the Netherlands.
  • Hopkin, S.P. (1991) A key to the woodlice of Britain and Ireland. Field Studies, 7, 599650.
  • Humphrey, J.W., Hawes, C., Peace, A.J., Ferris-Kaan, R. & Jukes, M.R. (1999) Relationships between insect diversity and habitat characteristics in plantation forests. Forest Ecology and Management, 113, 1121.
  • Hyman, P.S. & Parsons, M.S. (1992) A Review of the Scarce and Threatened Coleoptera of Great Britain. UK Joint Nature Conservation Committee, Peterborough, UK.
  • Ings, T.C. & Hartley, S.E. (1999) The effect of habitat structure on carabid communities during the regeneration of a native Scottish forest. Forest Ecology and Management, 119, 123136.
  • Johnson, C. (1993) Provisional Atlas of the Cryptophagidae-Atomariinae (Coleoptera) of Britain and Ireland. Biological Records Centre, Huntingdon, UK.
  • Lavelle, P. (1996) Diversity of soil fauna and ecosystem function. Biology International, 33, 316.
  • Leps, J. & Smilauer, P. (2003) Multivariate Analysis of Ecological Data Using canoco. Cambridge University Press, Cambridge, UK.
  • Luff, M.L. (1998) Provisional Atlas of the Ground Beetles (Coleoptera, Carabidae) of Britain. Biological Records Centre, Huntingdon, UK.
  • McAleece, N. (1997) Biodiversity Pro Software. The Natural History Museum, London, UK.
  • Matson, P.A., Parton, W.J., Power, A.G. & Swift, M.J. (1997) Agricultural intensification and ecosystem properties. Science, 277, 504509.
  • Mittelbach, G.G., Steiner, C.F., Scheiner, S.M., Gross, K.L., Reynolds, H.L., Waide, R.B., Willig, M.R., Dodson, S.I. & Gough, L. (2001) What is the observed relationship between species richness and productivity? Ecology, 82, 23812396.
  • Morris, M.G. (1990) Handbooks for the Identification of British Insects, Coleoptera. Orthocerous Weevils, Volume 16. Royal Entomological Society, London, UK.
  • Morris, M.G. (1997) Handbooks for the Identification of British Insects, Broad-nosed Weevils. Royal Entomological Society, London, UK.
  • Morris, M.G. (2002) Handbooks for the Identification of British Insects, True Weevils (Part I). Royal Entomological Society, London, UK.
  • Myers, N. & Knoll, A.H. (2001) The biotic crisis and the future of evolution. Proceedings of the National Academy of Sciences of the United States of America, 98, 53895392.
  • Nixon, C.J. & Clifford, T. (1994) The age and structure of native pinewood remnants. Our Pinewood Heritage Conference Proceedings 1994 (ed. J.R. Aldous), pp. 177195. Forestry Commission, RSPB, Scottish Natural Heritage, Inverness, UK.
  • Ozanne, C.M.P., Speight, M.R., Hambler, C. & Evans, H.F. (2000) Isolated trees and forestry patches: patterns in canopy arthropod abundance and diversity in Pinus sylvestris (Scots pine). Forest Ecology and Management, 137, 5363.
  • Peterken, G.F., Baldock, D. & Hampson, A. (1995) A forest habitat network for Scotland. Scottish Natural Heritage Research. Survey and Monitoring Report No. 44. Scottish Natural Heritage, Battleby, UK.
  • Petit, S. & Usher, M.B. (1998) Biodiversity in agricultural landscapes: the ground beetle communities of woody uncultivated habitats. Biodiversity and Conservation, 7, 15491561.
  • Sims, R.W. & Gerard, B.M. (1999) Earthworms. Field Studies Council, Shrewsbury, UK.
  • StatSoft Inc. (1998) STATISTICA for Windows (Computer Program Manual). Tulsa, Oklahoma, USA.
  • Storch, D., Izling, A.L. & Gaston, K.J. (2003) Geometry of the species–area relationship in central European birds: testing the mechanism. Journal of Animal Ecology, 72, 509519.
  • Summers, R.W., Moss, R. & Halliwell, E.C. (1994) Scotland's native pinewoods: the requirements of birds and animals. Our Pinewood Heritage Conference Proceedings 1994 (ed. J.R. Aldous), pp. 222241. Forestry Commission, RSPB, Scottish Natural Heritage, Inverness, UK.
  • Tews, J., Brose, U., Grimm, V., Tielbörger, K., Wichmann, M.C., Schwager, M. & Jeltsch, F. (2004) Animal species diversity driven by habitat heterogeneity/diversity: the importance of keystone structures. Journal of Biogeography, 31, 7992.
  • Vanbergen, A.J., Woodcock, B.A., Watt, A.D. & Niemela, J. (2005) Effect of land-use heterogeneity on carabid communities at the landscape scale. Ecography, 28, 316.
  • Vazquez, D.P. & Simberloff, D. (2003) Changes in interaction biodiversity induced by an introduced ungulate. Ecology Letters, 6, 10771083.
  • Welch, R.C. (1986) What do we know about insects in Scottish woods?. Trees and Wildlife in the Scottish Uplands (ed. D. Jenkins), pp. 2738. ITE Symposium No. 17. Institute of Terrestrial Ecology, Abbots Ripton, UK.
  • Whittaker, R.J., Willis, K.J. & Field, R. (2001) Scale and species richness: towards a general, hierarchical theory of species diversity. Journal of Biogeography, 28, 453470.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Appendix S1. Location of land use sampling grids in Scotland. Appendix S2. Eigenvalues for pCCA Monte Carlo tests. Appendix S3. List of species in major groups, with authorities and broad distributional patterns

FilenameFormatSizeDescription
JPE_1090_sm_Appendix1.doc892KSupporting info item
JPE_1090_sm_Appendix2.doc47KSupporting info item
JPE_1090_sm_Appendix3.doc527KSupporting info item
JPE_1090_sm.txt0KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.