Ecology, 95(5), 2014, pp.

1213–1224
Ó 2014 by the Ecological Society of America

Drivers of carabid functional diversity: abiotic environment,

plant functional traits, or plant functional diversity?
ROBIN J. PAKEMAN1 AND JENNI A. STOCKAN
James Hutton Institute, Craigiebuckler, Aberdeen AB15 8QH United Kingdom

Abstract. Understanding how community assembly is controlled by the balance of abiotic

drivers (environment or management) and biotic drivers (community composition of other
groups) is important in predicting the response of ecosystems to environmental change. If
there are strong links between plant assemblage structure and carabid beetle functional traits
and functional diversity, then it is possible to predict the impact of environmental change
propagating through different functional and trophic groups. Vegetation and pitfall trap
beetle surveys were carried out across twenty four sites contrasting in land use, and hence
productivity and disturbance regime. Plant functional traits were very successful at explaining
the distribution of carabid functional traits across the habitats studied. Key carabid response
traits appeared to be body length and wing type. Carabid functional richness was significantly
smaller than expected, indicating strong environmental filtering, modulated by management,
soil characteristics, and by plant response traits. Carabid functional divergence was negatively
related to plant functional evenness, while carabid functional evenness was positively
correlated to plant functional evenness and richness. The study shows that there are clear trait
linkages between the plant and the carabid assemblage that act not only through the mean
traits displayed, but also via their distribution in trait space; powerful evidence that both the
mean and variance of traits in one trophic group structure the assemblage of another.
Key words: Coleoptera; functional divergence; functional evenness; functional richness; ground beetle;
response traits.

INTRODUCTION niche differentiation. Using a range of different indices

One of the most long-running pursuits in ecology is of functional diversity can help shed light on the
the study of community assembly processes—that is dominant processes (Mouchet et al. 2010).
how communities arise to display only part of the local Many studies have shown how the mean values for
species pool (Mason and de Bello 2013). Functional plant functional traits are controlled by the environment
traits have played a key role in understanding the (e.g., Dı́az et al. 1998, Violle et al. 2007); these are so
linkages between species and the environmental drivers called ‘‘response traits’’ (Lavorel and Garnier 2002).
present (particularly for plants) and have become the With respect to carabid beetles (Coleoptera, Carabidae),
focus of testing for the effects of community assembly Ribera et al. (2001) demonstrated a strong link between
processes at either the individual trait level (e.g., Weiher life-history traits and the main environmental gradient
et al. 1998) or across a number of traits using an index of (a combined land use intensity/temperature gradient).
functional diversity (e.g., Mason et al. 2013). Evidence Gerisch (2011) demonstrated that carabid beetle assem-
for niche complementarity (increased fitness arises from blages in flood plains were governed by the interaction
trait dissimilarity between co-occurring species), is between disturbance, hydrology, and the traits of the
assumed when functional diversity (the variance in trait species. However, more recent studies have also
values present in a community) is higher than expected addressed how community assembly differs along
(divergence; Mason and de Bello 2013) and environ- abiotic gradients of drivers such as productivity and
mental filtering is assumed when functional diversity is disturbance. Disturbances such as fire reduce the range
less than expected (convergence; Cornwell et al. 2006).
of trait values (functional richness) expressed in plant
However, Mayfield and Levine (2010) suggest that
communities (Pausas and Verdu 2008), as does increased
competition is also a potential driver of community
intensity of agricultural management (Pakeman 2011a).
assembly and that in some circumstances coexistence
Contrasting patterns have been found for community
might stem from similar competitive abilities rather than
assembly along productivity gradients; there was a
greater range (functional richness) in leaf traits on
Manuscript received 6 June 2013; revised 24 October 2013; wetter, more productive soils in Californian chaparral
accepted 28 October 2013. Corresponding Editor: N. J. B.
Kraft. (Cornwell and Ackerly 2009), while functional richness
1 E-mail: robin.pakeman@hutton.ac.uk declined along productivity gradients in a UK agricul-
1213
1214 ROBIN J. PAKEMAN AND JENNI A. STOCKAN Ecology, Vol. 95, No. 5

tural area (Pakeman 2011a) and in Mediterranean tional diversity of carabid beetles is correlated to plant
rangelands (Bernard-Verdier et al. 2012). community or functional diversity to explore potential
Biotic interactions can also drive community assem- links between plant and carabid community assembly
bly, either through competition/facilitation within a and (5) that the assembly of carabid communities is
trophic level, or through the provision of resources (e.g., affected by disturbance, edaphic factors, and the
Brooks et al. 2012) or habitat structure (Dennis et al. functional traits of the vegetation.
1998) across trophic levels. For instance, in arable
MATERIALS AND METHODS
systems small, seed-feeding, spring-breeding carabids
were associated with autumn-germinating, small-seeded Field sites
weeds while larger, autumn-breeding omnivorous cara- This study formed part of a larger study investigating
bids were associated with spring germinating, larger- how land use influences biodiversity and ecosystem
seeded weeds (Brooks et al. 2012). Plant community function. The study was centered in a system with a
composition directly controlled the taxonomic and range of land uses and intensities within a small area (5
functional composition of ants in an arid steppe 3 7 km) on the west coast of Scotland centered on 57.318
(Frenette-Dussault et al. 2013) and grasshoppers in a N, 5.668 W. Sites shared the same geology and climate.
forest : grassland mosaic (van der Plas et al. 2012), while Twenty-four of the sites were sampled for carabid
abiotic drivers operated via their control of plant beetles, with land management covering arable, fallow
composition. In these examples, the traits of one group areas cropped for hay, winter grazed rough grassland
(plants) were clearly mediating the community compo- and tall herb communities, unimproved silage fields (see
sition of another via trait linkages (Lavorel and Garnier Plate 1), unimproved pasture, and moorland (Appendix:
2002). Determining the strength of this trait linkage Table A1). Sites ranged in size from 0.5 ha to .100 ha,
between plants and carabid beetles in comparison to but all sampling was restricted to areas not exceeding 2
other potential drivers of carabid community composi- ha. This excluded the wetland and woodland sites
tion (such as soil or management) is the first aim of this referred to in previous papers (Pakeman 2011a, b,
paper. Pakeman et al. 2011) because of the difficulty of using
If there is a strong linkage between plant traits and pitfall traps in wetland sites and the substantially
carabid beetle traits, it is possible that this trait linkage different beetle fauna of woodlands.
may be strong enough to impact community assembly
processes (Mason et al. 2005) alongside the impact it Sampling
may have on the mean traits present (Pakeman 2011a). Vegetation surveys were carried out in July 2007.
As yet there is little information on what drives Relative abundance (the proportion of aboveground
community assembly processes in carabid communities. biomass) of all higher plant species, bryophytes and
Schirmel et al. (2012) showed that functional dispersion litter was estimated by eye in between four and seven,
(a weighted version of functional richness) was lowest on randomly placed, 1 3 1 m quadrats per site.
the youngest successional stages in a sand dune, Carabid beetle sampling was carried out on the same
suggesting that environmental filtering is strongest in area using pitfall traps, consisting of plastic cups with an
more disturbed habitats. However, they also showed upper diameter of 7 cm and a height of 9.5 cm. The traps
that there was little correlation between other functional were partially filled with commercial antifreeze (ethylene
diversity measures (functional divergence and functional glycol), covered with 25-mm mesh to reduce both
evenness) and the same gradient of disturbance. It is interference by livestock and entry by small mammals
possible that the plant assemblage may play a role in the (Downie et al. 2000), and surmounted with three-legged
assembly of the carabid community, either because a key roofs to prevent rainwater flooding the traps. Trapping
driver of carabid abundance is vegetation structure or took place over two eight-week periods from 7 June to
through trophic cascades with the plant traits affecting 19 July and from 23 August to 4 October within which
the abundance and type of herbivorous prey of the traps were emptied at two-week intervals. The gap was
carabids. The second aim of the paper is, therefore, to necessary to accommodate hay and cereal harvesting
assess what drives carabid beetle community assembly and the associated use of machinery. Ten traps per site
using functional diversity metrics as proxies for assessing were positioned linearly in two rows of five and spaced 5
assembly. m apart.
The two aims of the paper were assessed through
addressing the following hypotheses: (1) there are clear Trait data
linkages between the traits of the carabid beetle Plant trait data were assembled from two main
assemblage (response traits) and traits of the plant sources (Table 1), BiolFlor (Klotz et al. 2002) and
assemblage that are in turn controlled by the environ- LEDA (Kleyer et al. 2008), and was restricted to the
ment or management, (2) this linkage is independent of eight response traits identified for this system that linked
other drivers such as management or edaphic factors, management and soil conditions with the vegetation
and (3) that good response traits of the carabid beetle (Pakeman 2011b). These are from the same region and
assemblage can be identified. Moreover, (4) the func- the variable traits are relatively robust to be used in
May 2014 CONTROLS OF CARABID FUNCTIONAL DIVERSITY 1215

different situations (Cordlandwehr et al. 2013, Kazakou TABLE 1. Plant and carabid beetle traits used in the analysis
et al. 2014). Where gaps were present (;2% of cases) with source and coding information.
information was supplemented with data from floras
Taxa and trait Attributes Units/coding
(Clapham et al. 1987, Stace 2010) and by averaging data
from congeners. Canopy structure and life form Plant
(renamed bud height) were coded to reduce the number Bud height1 geophyte, therophyte 0
hemicryptophyte 0.333
of attribute columns in the analysis. chamaephyte 0.667
Life history and other traits of carabid beetles were phanerophyte 1
determined from several sources (Lindroth 1985, 1986, Canopy structure1 rosette 0
hemirosette 0.5
Luff 1998, 2007, Traugott 1998, Ribera et al. 1999, erosulate 1
Forsythe 2000, Duff 2012). Where gaps were present Leaf dry matter content2 mg/g
personal observation was used to supplement the data Leafing period1 summer green 0/1
log(canopy height)2 m
(Table 1; Appendix: Table A2 for species information). log(leaf size) 2
mm2
Where there was a conflict in the literature concerning a Vegetative spread1 rhizome 0/1
particular attribute, the most recent or local source was Terminal velocity2 m/s
used (local taking preference). Size (body length) was Carabid
calculated as the average value documented across all Size (body length) mm
sources. Wing morphology brachypterous 0/1
macropterous 0/1
dimorphic 0/1
Environmental data apterous 0/1
Basic soil parameters were characterized on one Diet omnivore 0/1
predator 0/1
bulked sample per site made up of 10 5 cm diameter phytophagous 0/1
cores (0–10 cm in depth) taken from random positions Life cycle annual 0/1
across the site. Measures included soil pH, moisture loss, biennial 0/1
Activity period nocturnal 0/1
loss on ignition, soil total nitrogen and carbon (and a diurnal 0/1
C:N ratio), extractable Ca, K, Mg, and P, soil texture, Breeding period spring 0/1
and soil nitrogen release (see Pakeman [2011b] for autumn 0/1
further details). Soil texture measurements were used to Moisture preference3  1–5
estimate field capacity using the method of Saxton and Sources: 1, BiolFlor (Klotz et al. 2002); 2, LEDA (Kleyer et
Willey (2006). In total 14 soil parameters were included al. 2008); 3, Forsythe (2000).
  Levels of moisture preference are 1, very dry areas usually
in the analysis. Disturbance was quantified using metrics associated with porous soils such as sand: heaths, quarries,
developed from Kühner and Kleyer (2008); frequency of coastal regions, dry moorland; 2, fairly dry areas, well drained
disturbance, biomass loss at disturbance, belowground and dry soils; 3, areas of intermediate moisture content; 4, moist
disturbance, start of disturbance, presence of summer areas, river and stream banks, wet heathland, soils and debris
that retain moisture; 5, very wet areas, fens, bogs, marshes, mud
and/or winter grazing, and vegetation height (see Pake- banks, stream and river edges, wet moorland.
man [2011b] for further details). A total of seven
variables described site management. potential for over-fitting, each category was subject to
an initial ordination using PCA, with the axes used in
Statistical analysis
the variance partitioning selected after stepwise forward
A preliminary assessment of the potential linkage selection in a redundancy analysis (RDA; Økland and
between the plant and the carabid community across the Eilertsen 1994). Significance was tested by 1000 Monte
24 sites was done using a Procrustes analysis to assess Carlo unrestricted permutations. As a check of the
the correlation between the respective species ordina- potential for there to be direct species to species
tions. The initial species ordinations were both principal linkages, plant species were also used as a substitute
components analyses (PCA) carried out on Hellinger- for plant response traits, with the data reduction being
transformed data (Legendre and Gallagher 2001). The carried out with a PCA on Hellinger-transformed data.
significance of the correlation was assessed against 1000 This analysis was carried out in vegan (Oksanen et al.
Monte Carlo permutations. The ordinations and Pro- 2011) in R.
crustes analysis were carried out using vegan (Oksanen There are a number of ways of linking three tables of
et al. 2011) in R (R Development Core Team 2010). environmental information (in this case, a table of
The potential for plant traits to contribute to the carabid activity-density, carabid traits, and a table of
explanation of patterns in carabid species distribution environmental drivers [Kleyer et al. 2012]). Here a
and in the distribution of carabid traits (community parsimonious list of beetle response traits was generated
weighted means, i.e., the abundance weighted mean of using the iterative RLQ method of Bernhardt-Römer-
the community) was investigated with variance parti- mann et al. (2008), but with a stricter stopping criterion
tioning (Borcard et al. 1992). Three categories of (Pakeman 2011b). The analysis was carried out sepa-
explanatory variables were used: soil variables, manage- rately to identify response traits for soil characteristics,
ment variables, and plant response traits. To reduce the management and plant response traits, with these
1216 ROBIN J. PAKEMAN AND JENNI A. STOCKAN Ecology, Vol. 95, No. 5

PLATE 1. (Left) Nebria brevicollis one of the two most common carabid beetle species trapped during the study. (Right)
Unimproved silage fields at Plockton, Scotland. These are cut for hay and winter grazed. Photo credits: J. A. Stockan.

restricted to the variables identified in the Redundancy from expected for each functional diversity index was
Analysis above. RLQ was carried out in ade4 (Dray and measured using the standardized effect size, ES ([FDobs
Dufour 2007) in R. – mean FDran]/[SD FDran], where obs stands for
The functional diversity indices functional divergence, observed and ran stands for randomized [Gotelli and
functional evenness and functional richness were calcu- McCabe 2002]) and as a significance test, it was assessed
lated following the method and R scripts of Villéger et whether the site fell in the 5% of most extreme values of
al. (2008) for the beetle and the plant community. These the simulated assemblages (i.e., 2.5% at either end of the
indices were chosen as they are relatively orthogonal distribution). Differences between the ES values of the
(Mouchet et al. 2010) and measure different aspects of sites and zero were tested using a one-sample t test, with
functional diversity, namely the spread of abundant trait a value below zero meaning functional diversity was
values in trait space, the evenness of distribution of lower than expected given random assembly and thus
abundance and the overall spread of traits in the indicated environmental filtering.
community (Villéger et al. 2008). As the trait data Correlations between the individual carabid function-
contained both qualitative and quantitative traits, an al diversity indices and also with plant functional
initial matrix of Gower distances was first calculated and
diversity indices were calculated. Correlations (Pearson)
then subjected to a principal coordinates analysis (PCoA
were also calculated between the beetle functional
[Villéger et al. 2008, Laliberté and Legendre 2010]) in
diversity indices and the main axes of variation (from
ade4 (Dray and Dufour 2007). The first four axes were
PCA) in soil characteristics, management and plant
taken from the PCoA and they represented 79.5% of the
response traits identified from the RDA. Finally, where
variance. The calculated values of functional diversity
a significant correlation was present, a stepwise-forward
were compared with values from 999 simulated assem-
multiple regression was carried out to assess if drivers
blages (produced by the R function sample). These
simulated assemblages were constructed using all the were complementary or overlapping in their contribu-
species recorded across the sites, with species drawn tion to explain variation in carabid functional diversity.
randomly from the list without replacement. Species Forward selection was used because of the high number
were then allocated an abundance from the list of of possible predictors (15) relative to the number of sites
abundances of the species at that site by matching the (24).
order of draw with the rank order of abundance. Given
RESULTS
that the sites are within a small area and dispersal
limitation is unlikely to constrain species occurrences Over the sampling period, a total of 3781 carabids
then this method of null model generation with equal were caught and identified from 32 species (see Plate 1).
chances of selection for all species is appropriate (Mason This compared to 125 plant species across the same sites.
et al. 2013). This type of simulated assemblage The two species ordinations, carabid beetles and plants,
construction is a matrix swap randomization (Manly were significantly correlated with r ¼ 0.53 and P ¼ 0.002.
1995) and it maintained the species richness and the This indicated that there was a potential link between
pattern of abundance within each site. The deviation the vegetation and the carabid beetle assemblage.
May 2014 CONTROLS OF CARABID FUNCTIONAL DIVERSITY 1217

FIG. 1. Ordination biplots from principal components analysis of (a) management variables, (b) soil variables, (c) plant
response traits, and (d) plant species abundances, where solid diamonds represent sites, open squares represent categorical factors,
and open circles represent the 35 most abundant species. Abbreviations are (a) BgrdDist, belowground disturbance; BiomLoss,
biomass loss; DistFreq, disturbance frequency; StrtDist, start of disturbance; Sumgraz, summer grazing; VegHt, vegetation height;
and Wingraz, winter grazing; (b) C:N, carbon to nitrogen ratio; FldCap, field capacity; LOI, loss on ignition; and MoistLos,
moisture loss; (c) BudHt, bud height; LDMC, leaf dry matter content; Lfsumgrn, leafing period, summer green; logCanHt, log of
canopy height; logLfsze, log of leaf size; Rhizome, vegetative spread by rhizome; Structure, canopy structure; and Termvelo, seed
terminal velocity; (d) Ac, Agrostis capillaris; Ae, Arrhenatherum elatius; Ao, Anthoxanthum odoratum; As, Agrostis stolonifera; Ca,
Chenopodium album; Cc, Crepis capillaris; Ccr, Cynosurus cristatus; Ch, Carex hostiana; Cn, Centaurea nigra; Cp, Carex pallescens;
Cv, Calluna vulgaris; Ea, Eriophorum angustifolium; Et, Erica tetralix; Ev, Eriophorum vaginatum; Fo, Festuca ovina; Fr, F. rubra;
Fv, F. vivipara; Gs, Galium saxatile; Hl, Holcus lanatus; Je, Juncus effusus; Lp, Lolium perenne; Mc, Molinia caerulea; Pa, Potentilla
anserina; Pl, Plantago lanceloata; Pm, Persicaria maculata; Pp, Phleum pratense; Rm, Rhinanthus minor; Rr, Ranunculus repens; Sa,
Spergula arvensis; Sp, Sagina procumbens; Ss, Stachys sylvatica; St, Solanum tuberosum; Tc, Trichophorum cespitosum; Tr, Trifolium
repens; and Vo, Valeriana officinalis.

The initial forward selection of environmental param- explained ¼ 0.655) and axis 1 from the plant species PCA
eters for the variance partitioning of the carabid beetle (Fig. 1d; P ¼ 0.003, variance explained ¼ 0.435). A
traits resulted in selection of axes 1 and 3 from the smaller number of axes were selected as significant in
management PCA (Fig. 1a; P ¼ 0.003, variance explained explaining the distribution of species; axis 2 from the soil
¼ 0.467), axes 1 and 2 from the soil PCA (Fig. 1b; P ¼ PCA (P ¼ 0.043, variance explained ¼ 0.088), axis 2 from
0.003, variance explained ¼ 0.432), axes 1, 2, and 3 from the management PCA (P ¼ 0.033, variance explained ¼
the plant response trait PCA (Fig. 1c; P ¼ 0.001, variance 0.125), axis 1 and 2 from the plant response traits PCA (P
1218 ROBIN J. PAKEMAN AND JENNI A. STOCKAN Ecology, Vol. 95, No. 5

FIG. 2. Variance partitioning of beetle traits by soil, management and either (a) plant response traits or (b) ordination axes of
plant species composition, and beetle species composition by soil, management, and either (c) plant response traits or (d) ordination
axes of plant species composition. Numbers represent the proportion of variance explained. The significance of the relationships is
shown by asterisks.
* P , 0.05; ** P , 0.01; *** P , 0.001; ns indicates not significant.

¼ 0.047, variance explained ¼ 0.178) and axis 1 from the soil on plant species distribution. However, in this
plant species composition PCA (P ¼ 0.049, variance partitioning, only management explained a substantial
explained ¼ 0.086). proportion, 12%, of the variation independently.
The variance partitioning showed that plant response Soil, management, and plant response traits explained
traits were more successful than plant species composi- only 17% of the variation in carabid beetle species
tion in explaining the variation in both carabid beetle distribution, though in this case only 7% was shared
traits and species composition (Fig. 2a and c vs. b and (Fig. 2c). Each group independently explained a similar
d). What was also evident was that the carabid beetle small proportion of the variation (3–5%). Substituting
traits were substantially better explained by manage- plant species composition for plant response traits
ment and soil characteristics than the species composi- reduced the variation explained slightly (down to 16%;
tion (Fig. 2a and b vs. c and d). Fig. 2d), but as for the partitioning of carabid beetle
Soil, management, and plant response traits explained traits involving plant species composition, there was a
62% of the variation in carabid beetle traits (Fig. 2a). A shift in variance explained from the plant to manage-
high proportion of this variation was shared (51%) ment and a reduction in the shared variation explained.
between the three groups of variables. However, the The RLQ analysis showed a significant correlation
largest proportion of variation explained singly by the between carabid beetle traits and soil, management and
three groups of variables was by the plant response traits plant response traits (Table 2). The different environ-
at 11%, substantially higher than the 4% explained by mental predictors yielded different numbers of selected
management or the 2% explained by soil alone. Soil, response traits; soil three, management six, and plant
management, and plant species composition explained response traits four (Fig. 3, Table 2). However, the
51% of the variation in carabid beetle traits (Fig. 2b). correlation L scores were relatively low (0.53–0.58 out of
Again a substantial proportion was shared variation 2.00) compared to a similar analysis of plant traits (1.12–
(41%), again reflecting the influence of management and 1.35 [Pakeman 2011b]), though significant (P between
May 2014 CONTROLS OF CARABID FUNCTIONAL DIVERSITY 1219

TABLE 2. The order of selection of the carabid response traits from the RLQ analysis of the link
between carabid beetle traits and the three groups of environmental drivers (soil, management,
and plant response traits).

Driver
Order of selection Soil Management Plant response traits
1,2 apterous, predator brachypterous, apterous size, apterous
3 size nocturnal phytophagous
4 size nocturnal
5 annual
6 biennial
CorrL 0.538 0.526 0.575
p 0.035 0.037 0.017
Note: CorrL represents the correlation L score from the RLQ (maximum 2) and p is the
probability from the permutation test of the RLQ (n ¼ 1000 permutations).

0.017 and 0.037). Two carabid beetle response traits (Table 4). Carabid beetle functional divergence and its
were selected in all the analyses: size and wing effect size were negatively correlated with plant func-
morphology (apterous). The community-weighted mean tional evenness and its effect size; as plant functional
of size was positively correlated to axis 1 of the evenness increased, then carabid beetle functional
management PCA (r ¼ 0.559, P , 0.01; Fig. 1a), higher divergence declined. The effect size of carabid beetle
values correspond to less and later disturbed sites. Wing functional evenness was positively correlated to plant
morphology was strongly related to axis 2, with functional evenness and richness and their respective
macropterous species associated with ploughed sites (r effect sizes; at high plant evenness and richness, carabid
¼ 0.642, P , 0.001) and apterous species with grassland, evenness was close to or exceeded expectations.
especially nutrient poor, lightly managed ones. Activity Carabid beetle species richness was negatively corre-
period (nocturnal) was selected for both management lated to axis 1 of the management PCA (Table 5, Fig.
and plant traits, while diet (predator/phytophagous) was 1a) and positively to axis 3; richness was a feature of
selected for both soil and plant traits. sites with greater disturbance intensity and frequency, as
Of the 24 sites, two had functional divergence (FDiv) well as taller, winter grazed sites. The effect size for
scores in the lowest 2.5% of the randomly drawn functional richness was also negatively correlated to axis
communities and three had functional richness (FRic) 1: at high values for this axis (less disturbed sites) it was
scores in the lowest 2.5%. However, two sites had further reduced below zero. The first axis from the soil
functional evenness (FEve) scores in the highest 2.5% of PCA (Fig. 1b) was similarly negatively correlated to ES
the simulated communities. This compared to the plant FRic, where this axis represented a transition from
communities where one site was in the lowest 2.5% for nutrient rich sites at negative values to more nutrient
FDiv and two sites were in the lowest 2.5% for FRic.
poor sites at positive ones. There was also a negative
Beetle FEve was significantly higher than plant FEve
correlation between this axis and ES FEve, though in
(Table 3), while carabid beetle FRic was substantially
this case ES FEve was positive at low values (fertile sites)
lower than that of the plants when expressed as a
proportion of the functional richness of all sites
combined. This latter pattern was associated with a
standardized effect size (ES) for FRic being significantly
lower than zero for the carabid beetles; the only effect
size to differ from zero. Plant species richness was higher
than carabid beetle species richness across the 24 sites.
Carabid beetle functional richness was positively
correlated to carabid beetle species richness (Table 4).
However, species richness was also positively correlated
to ES FRic; at high species richness FRic was close to
expectations, but at low species richness FRic was
substantially lower than expected. Consequently FRic
and ES FRic were highly positively correlated. There
were also high correlations between FDiv and ES FDiv
and FEve and ES FEve, as sites with a high FDiv or FIG. 3. Effects of increasing the number of traits in the
FEve were characterized by higher than expected RLQ analysis on the best correlation of L scores with the
different sets of environmental descriptors: diamonds, soil
functional diversity or evenness. variables; squares, management variables; triangles, plant
There were a number of correlations between the response traits. Arrows highlight the chosen number of traits
plant and the carabid beetle functional diversity indices for each analysis.
1220 ROBIN J. PAKEMAN AND JENNI A. STOCKAN Ecology, Vol. 95, No. 5

TABLE 3. Species richness and functional diversity values for the 24 sites.

Parameter Carabids Plants t P

Species richness 10.79 25.29 9.24 ,0.001
FDiv 0.796 0.793 0.07 ns
ES FDiv 0.035 0.032 0.11 ns
FEve 0.545 0.461 3.67 0.001
ES FEve 0.021 0.015 1.67 ns
FRic 0.147 0.359 4.67 ,0.001
ES FRic 0.443*** 0.079 5.15 ,0.001
Notes: FDiv is functional divergence, FEve is functional evenness, FRic is functional richness,
and ES is the standardized effect size for each measure. FRic is expressed as a proportion of the
maximum FRic for that taxa across all sites combined. The results of a paired t test for each
parameter are shown as their respective t and P values (ns, not significant).
*** Represents a significant difference from zero for standard effect sizes at P , 0.001.

and negative at high values for this axis (in fertile sites). 0.278), suggesting these drivers are complementary in
ES FRic was positively correlated to axis 2 from the their effects.
response traits ordination (Fig. 1c)—at low scores on
DISCUSSION
this axis (tall, erosulate) ES FRic was more negative
than at high scores. Carabid beetle species richness was The aims of this study were to assess the strength of
positively associated with the first axis of the plant trait linkages between plants and carabid beetles in
species ordination (Fig. 1d); higher richness was comparison to other potential drivers of carabid
associated with the open arable sites to the right and community composition (such as soil or management)
lower species richness with the wet moorland sites to the and to assess what drives carabid beetle community
assembly. The initial redundancy analyses clearly
right, with grassland areas intermediate. Finally, carabid
showed that the traits of the carabid beetle assemblage
beetle ES FRic was more negative on sites with low axis
were correlated to a number of soil and management
1 scores; sites characterized by moorland vegetation.
drivers. However, the redundancy analyses also showed
Multiple regression showed that generally the most
that plant traits (and to a lesser extent plant species
parsimonious model included only the driver with the composition) were biotic drivers of assembly. This
highest correlation coefficient (Tables 4 and 5), i.e., confirms hypothesis 1, that there is a clear linkage
plant FEve for carabid FDiv and ES FDiv and soil PC between the traits of the plant and the carabid
axis 1 for carabid ES FRic, suggesting that for these assemblages.
proxies of community assembly the drivers were largely The variance partitioning (Fig. 2) revealed that plant
overlapping in effect. However, for carabid ES FEve, response traits were more successful than plant species
both plant ES FRic (P ¼ 0.037) and soil PC axis 1 (P ¼ composition in independently predicting the pattern of
0.046) combined to give the best model ([carabid ES carabid traits across the sites. There was a considerable
FEve] ¼ 0.046 þ 0.312[plant ES FRic] – 0.048SPC1; R 2 ¼ proportion of shared variation between plant traits,

TABLE 4. Correlations between beetle functional diversity metrics and between carabid beetle and plant functional diversity
metrics.

Carabid beetle
Species richness FDiv ES FDiv FEve ES FEve FRic ES FRic
Carabid
FDiv 0.217
ES FDiv 0.209 0.798***
FEve 0.147 0.010 0.031
ES FEve 0.248 0.058 0.074 0.809***
FRic 0.767*** 0.303 0.292 0.004 0.299
ES FRic 0.555** 0.002 0.008 0.226 0.367 0.815***
Plant
Species richness 0.353 0.108 0.092 0.232 0.258 0.168 0.207
FDiv 0.027 0.085 0.089 0.126 0.205 0.174 0.251
ES FDiv 0.047 0.080 0.081 0.112 0.214 0.174 0.250
FEve 0.084 0.583** 0.594** 0.402 0.421* 0.063 0.387
ES FEve 0.173 0.453* 0.467* 0.386 0.430* 0.080 0.355
FRic 0.249 0.151 0.135 0.373 0.404* 0.089 0.100
ES FRic 0.164 0.142 0.135 0.396 0.447* 0.027 0.041
* P , 0.05; ** P , 0.01; *** P , 0.001.
May 2014 CONTROLS OF CARABID FUNCTIONAL DIVERSITY 1221

TABLE 5. Correlations between carabid beetle functional diversity metrics and the major axes from ordinations of management
(MPC) and soil variables (SPC), plant response traits (RPC), and plant species abundances (PPC).

Species richness FDiv ES FDiv FEve ES FEve FRic ES FRic

MPC1 0.450* 0.248 0.266 0.049 0.194 0.333 0.471*
MPC3 0.464* 0.305 0.304 0.127 0.095 0.252 0.003
SPC1 0.247 0.325 0.322 0.267 0.430* 0.294 0.527**
SPC2 0.309 0.110 0.099 0.328 0.154 0.095 0.214
RPC1 0.313 0.307 0.322 0.095 0.172 0.145 0.251
RPC2 0.326 0.170 0.162 0.066 0.133 0.337 0.469*
RPC3 0.013 0.069 0.070 0.095 0.084 0.203 0.132
PPC1 0.413* 0.327 0.123 0.072 0.060 0.303 0.455*
* P , 0.05; ** P , 0.01.

management and soil variables, which was expected and environmental severity control carabid assemblage.
given the correlation between soil fertility and manage- In their study, species tended to be smaller on more
ment (the best soils tend to be managed differently from intensively managed sites and these sites contained a
the poorer ones) and the dependence of the plant higher proportion of macropterous individuals. They
assemblage on management and soil conditions (Pake- suggested that disturbance selected for smaller, r-
man 2011b, Pakeman et al. 2011). However, the largest selected species or that recolonization was limited for
independent proportion of the variance explained was often flightless, larger carabids. Both of these findings
by the plant traits alone (11%), and individually the agree with this study and with patterns in other systems
plant traits explained almost the same amount of where disturbance is associated with smaller species,
variation as the three groups together (61% vs. 62%). including temporary wetlands in arable management
This confirms hypothesis 2, that there is a control of (Brose 2003), seasonally flooded grasslands (Gerisch
beetle traits by plant traits that is independent of the 2011) and forests (Barbaro and van Halder 2009), but it
measured management or edaphic factors, though it appears not to be a universal pattern (cf. woodland;
does not exclude the possibility that unmeasured Barton et al. 2011).
environmental drivers are indirectly controlling carabid Both activity period and diet were selected in some of
community composition. Substantial variation was also the RLQ analyses. However, these patterns appear not
explained jointly by soil, management, and plant species to be replicated in the literature. For instance, activity
composition, but in this part of the analysis plant species period did not show a relationship to management in the
composition had no independent explanatory power. study of Ribera et al. (2001). However, no other authors
This control of invertebrate traits by plant traits is have attempted to link plant traits and carabid beetle
similar to the findings of Gobbi et al. (2010) for beetles traits in this way. Moretti et al. (2013) show similar
in a deglaciated area, Brooks et al. (2012) for beetles in strong linkages between plant traits and those of
arable fields and van der Plas et al. (2012) for grasshoppers, particularly for leaf traits as grasshoppers
grasshoppers in a forest : grassland mosaic. However, are phytophagous. Overall, it appears that these results
for ants in an arid steppe, taxonomic composition was a do indicate that there is a strong linkage between the
stronger control than functional composition (Frenette- traits of the two assemblages, confirming the ideas of
Dussault et al. 2013). Lavorel and Garnier (2002).
In contrast, soil factors, management, and either plant Carabid functional diversity was not associated with
traits or species composition performed much less well plant species richness, in contrast to other studies where
in explaining carabid species distributions across the there was a clear correlation (Brose 2003). There was no
sites. The proportion of variation explained dropped simple correlation between plant and carabid functional
from 62% to 17% for the analyses involving plant traits richness, suggesting that the plant assemblage is not
and from 51% to 15% for the analyses involving plant simply creating niche space for the carabid assemblage.
species composition. This demonstrates the utility of This contrasts to the situation where grassland height
predicting the response of carabid communities through and leaf area trait ranges were negatively and positively,
their traits rather than using their species identity respectively, correlated to the range in wing size in
(Ribera et al. 2001, McGill et al. 2006, Gobbi et al. grasshoppers (van der Plas et al. 2012). Other compar-
2010). isons of this kind are lacking. However, there were
The iterative RLQ analysis was successful at identi- strong correlations between plant functional evenness
fying the most useful traits in predicting the response of and carabid functional divergence and evenness. Thus at
the carabid assemblage to differences between the sites high plant functional evenness there was low carabid
(confirming hypothesis 3). Of these size (body length) functional divergence (indicating dominant species lie in
and wing morphology were consistent across the three a small region of trait space and hence environmental
drivers. This is consistent with the main finding of filtering) but a small departure of carabid evenness from
Ribera et al. (2001); the intensity of land management expected (indicating an even spread of species activity-
1222 ROBIN J. PAKEMAN AND JENNI A. STOCKAN Ecology, Vol. 95, No. 5

density in trait space). This indicates there is some 2009), and mammals (Flynn et al. 2009). Functional
degree of filtering when plant functional divergence is divergence and evenness have been measured in few
high but that the spacing between species remains close studies for carabids (or other taxa), though Gerisch et al.
to expected. There was also a strong linkage between (2012) showed they decreased with disturbance. In
plant functional richness and carabid functional even- contrast functional evenness increases with disturbance
ness indicating that beetle functional evenness diverged for plants (Pakeman 2011a). Here, only functional
less from expected in more functionally rich plant evenness showed a significant relationship with the
assemblages. This offers some support to hypothesis 4, environment; it was reduced at high fertility but was
that there is a correlation between the functional not impacted by disturbance. However, this paper is the
diversity of the carabids and that of the plants. first one dealing with carabids that uses a null model
However, the correlation between the expected values comparison made to assess the expected patterns of
of functional richness and plant traits suggest that there functional diversity measures, and hence the interpreta-
is some control of carabid functional diversity by the tion of the others may be simplistic (Mason et al. 2013).
vegetation and that they are not just responding to the In conclusion, it appears that there is a strong linkage
same environmental drivers. Potentially a more even and between plant functional traits and the functional traits
a more rich functional diversity of the plant community of the beetle assemblage. This is despite the fact that
produces an environment suitable for a more even only a small proportion of the beetle fauna was classified
distribution of carabid species in trait space. It is also as phytophagous (,4.0% overall numbers), with the
clear that disturbance and soil characteristics have a majority predatory, so it is not due to food resources but
major role in controlling carabid functional diversity. is rather likely to be a structural effect (Brose 2003).
Overall it appeared that the functional diversity of the Given this distribution of diet preferences in the carabids
carabids was more restricted by environmental filtering present top-down effects are unlikely to be acting here;
than that of the plants (Table 3). top-down effects have been shown for grasshoppers in
The carabid assemblages were richer in the more one system (Moretti et al. 2013) but not another (van der
disturbed sites in contrast to other studies, which Plas et al. 2012). Soil and management drivers had
showed that a lack of disturbance was associated with limited independent impact on beetle traits, but there
species richness in wetlands (Bettacchioli et al. 2012). was a strong indirect linkage through their control of the
Functional richness was similarly affected by distur- traits of the plant assemblage (Pakeman 2011b). The
bance; at less disturbed sites, the functional richness was main traits these drivers acted on was size and wing
much lower than expected, as it was on sites with more morphology, but diet, activity period and life cycle were
fertile soils and taller vegetation. Similarly, functional also important traits. Functional diversity indices
evenness was lower than expected on more fertile sites. revealed that beetle functional diversity was more
This indicates that the assembly of carabid assemblages restricted that plant diversity, and was reduced at sites
is affected by disturbance, edaphic factors and the traits with lower disturbance, richer soils, and taller plants.
present in the vegetation, thus confirming hypothesis 5, There was also evidence that functional diversity is
but that the strongest driver was plant functional correlated in a number of ways between the two groups,
diversity, or for carabid ES FRic, a combination of with potential links between plant functional richness
plant functional diversity and edaphic factors. This and evenness and carabid functional divergence
suggests that environmental filtering of the carabid and evenness.
community is stronger in less disturbed or more fertile ACKNOWLEDGMENTS
sites with taller vegetation. The pattern is opposite to I would like to thank Iain Turnbull of the National Trust for
that shown by Schirmel et al. (2012) who showed that Scotland for all his help in arranging access and the many
environmental filtering was stronger early in dune crofters and the grazing clerks of Balmacara, Drumbuie,
succession (more disturbed sites). It may be that, within Duirinish, and Plockton for their help in our work. Rob
Brooker, Antonia Eastwood, Roger Cummins, and Jelle van
a landscape with a long history of cultivation and other
Rijmenant all helped with the fieldwork. Many thanks to the
human-driven land uses, assembly processes have editor and the two reviewers whose comments helped improve
different emphases than in a successional sequence the clarity of this paper.
where niches appear to increase in number as the
LITERATURE CITED
vegetation develops. The pattern seen here is also
opposite to that in field margins where functional Barbaro, L., and I. van Halder. 2009. Linking bird, carabid
beetle and butterfly life-history traits to habitat fragmenta-
diversity (measured as Rao’s entropy [Rao 1982]) tion in mosaic landscapes. Ecography 32:321–333.
increases with less disturbance (Woodcock et al. 2010); Barragán, F., C. E. Moreno, F. Escobar, G. Halffter, and D.
Rao’s entropy is positively correlated with FDiv, and to Vavarrete. 2011. Negative impacts of human land use on
a lesser extent FRic (Mouchet et al. 2010). It is also dung beetle functional diversity. PLoS ONE 6:e17976.
opposite to the impact of increased land use intensity on Barton, P. S., H. Gibb, A. D. Manning, D. B. Lindenmayer,
and S. A. Cunningham. 2011. Morphological traits as
dung beetle functional richness (Barragán et al. 2011) predictors of diet and microhabitat use in a diverse beetle
and on other taxa including plants (Laliberté et al. 2010, assemblage. Biological Journal of the Linnean Society 102:
Pakeman 2011a), birds (Fischer et al. 2007, Flynn et al. 301–310.
May 2014 CONTROLS OF CARABID FUNCTIONAL DIVERSITY 1223

Bernard-Verdier, M., et al. 2012. Community assembly along a Gerisch, M. 2011. Habitat disturbance and hydrological
soil depth gradient: contrasting patterns of plant trait parameters determine the body size and reproductive strategy
convergence and divergence in a Mediterranean rangeland. of alluvial ground beetles. ZooKeys 100:353–370.
Journal of Ecology 100:1422–1433. Gerisch, M., V. Agostinelli, K. Henle, and F. Dziock. 2012.
Bernhardt-Römermann, M., C. Römermann, R. Nuske, A. More species, but all do the same: contrasting effects of flood
Parth, S. Klotz, W. Schmidt, and J. Stadler. 2008. On the disturbance on ground beetle functional and species diversity.
identification of the most suitable traits for plant functional Oikos 121:508–515.
trait analyses. Oikos 117:1533–1541. Gobbi, M., et al. 2010. Plant adaptive responses during primary
Bettacchioli, G., M. Taormini, F. Bernini, and M. Migliorini. succession are associated with functional adaptations in
2012. Disturbance regimes in a wetland remnant: implica- ground beetles on deglaciated terrain. Community Ecology
tions for trait-displacements and shifts in the assemblage 11:223–231.
structure of carabid beetles (Coleoptera: Carabidae). Journal Gotelli, N. J., and D. J. McCabe. 2002. Species co-occurrence: a
of Insect Conservation 16:249–261. meta-analysis of J. M. Diamond’s assembly rules. Ecology
Borcard, D., P. Legendre, and P. Drapeau. 1992. Partialling out 83:2091–2096.
the spatial component of ecological variation. Ecology 73: Kazakou, E., C. Violle, C. Roumet, M.-L. Navas, D. Vile, J.
1045–1055. Kattge, E. Garnier. 2014. Are trait-based species ranking
Brooks, D. R., J. Storkey, S. J. Clark, L. G. Firbank, S. Petit, consistent across data sets and spatial scale? Journal of
and I. P. Woiwod. 2012. Trophic links between functional Vegetation Science 25:235–247
groups of arable plants and beetles at a national scale. Kleyer, M., et al. 2008. The LEDA traitbase: a database of life-
Journal of Applied Ecology 81:4–13. history traits of the Northwest European flora. Journal of
Brose, U. 2003. Bottom-up control of carabid beetle commu- Ecology 96:1266–1274.
nities in early successional wetlands: mediated by vegetation Kleyer, M., S. Dray, F. de Bello, J. Lepš, R. J. Pakeman, B.
structure of plant diversity. Oecologia 135:407–413. Strauss, W. Thuiller, and S. Lavorel. 2012. Assessing species
Clapham, A. R., T. G. Tutin, and D. M. Moore. 1987. Flora of and community functional responses to environmental
the British Isles. Third edition. Cambridge University Press, gradients: which multivariate methods? Journal of Vegeta-
Cambridge, UK. tion Science 23:805–821.
Cordlandwehr, V., et al. 2013. Do plant traits retrieved from a Klotz, S., I. Kühn, and W. Durka. 2002. BIOLFLOR: Eine
database accurately predict on-site measurements? Journal of Datenbank mit biologisch-ökologischen Merkmalen zur
Ecology 101:662–670. Flora von Deutschland. Bundesamt für Naturschutz, Bonn-
Cornwell, W. K., and D. D. Ackerly. 2009. Community Bad Godesberg, Germany.
assembly and shifts in plant trait distributions across an
Kühner, A., and M. Kleyer. 2008. A parsimonious combination
environmental gradient in coastal California. Ecological
of functional traits predicting plant responses to disturbance
Monographs 79:109–126.
and soil fertility. Journal of Vegetation Science 19:681–692.
Cornwell, W. K., D. W. Schwik, and D. D. Ackerly. 2006. A
Laliberté, E., and P. Legendre. 2010. A distance-based
trait-based test for habitat filtering: convex hull volume.
framework for measuring functional diversity from multiple
Ecology 87:1465–1471.
traits. Ecology 91:299–305.
Dennis, P., M. R. Young, and I. J. Gordon. 1998. Distribution
Laliberté, E., et al. 2010. Land-use intensification reduces
and abundance of small insects and arachnids in relation to
functional redundancy and response diversity in plant
structural heterogeneity of grazed, indigenous grasslands.
communities. Ecology Letters 13:76–86.
Ecological Entomology 23:253–264.
Dı́az, S., M. Cabido, and F. Casanoves. 1998. Plant functional Lavorel, S., and E. Garnier. 2002. Predicting changes in
traits and environmental filters at a regional scale. Journal of community composition and ecosystem functioning from
Vegetation Science 9:113–122. plant traits: revisiting the Holy Grail. Functional Ecology 16:
Downie, I. S., I. Ribera, D. I. McCracken, W. L. Wilson, G. N. 545–556.
Foster, A. Waterhouse, V. J. Abernethy, and K. J. Murphy. Legendre, P., and E. D. Gallagher. 2001. Ecologically
2000. Modelling populations of Erigone atra and E. meaningful transformations for ordination of species data.
dentipalpis (Araneae: Linyphiidae) across an agricultural Oecologia 129:271–280.
gradient in Scotland. Agriculture, Ecosystems and Environ- Lindroth, C. H. 1985. The Carabidae (Coleoptera) of Fenno-
ment 80:15–28. scandia and Denmark. Volume 15, part 1. Fauna Entomo-
Dray, S., and A. B. Dufour. 2007. The ade4 package: logica Scandinavica, Leiden, Kopenhagen, Denmark.
implementing the duality diagram for ecologists. Journal of Lindroth, C. H. 1986. The Carabidae (Coleoptera) of Fenno-
Statistical Software 22:1–20. scandia and Denmark. Volume 15, part 2. Fauna Entomo-
Duff, A. G. 2012. Beetles of Britain and Ireland, volume 1: logica Scandinavica, Leiden, Kopenhagen, Denmark.
Sphaeridae to Silphidae. AG Duff Publishing, London, UK. Luff, M. 1998. Provisional atlas of the ground beetles
Fischer, J., D. B. Lindenmayer, S. P. Blomberg, R. Montague- (Coleoptera, Carabidae) of Britain. Biological Records
Drake, A. Felton, and J. A. Stein. 2007. Functional richness Centre, Huntingdon, UK.
and relative resilience of bird communities in regions with Luff, M. 2007. The Carabidae (ground beetles) of Britain and
different land use intensities. Ecosystems 10:964–974. Ireland. Field Studies Council, Telford, UK.
Flynn, D. F. B., M. Gogol-Prokurat, T. Nogeire, N. Molinari, Manly, B. J. F. 1995. A note on the analysis of species co-
B. Trautman Richers, B. B. Lin, N. Simpson, M. M. occurrences. Ecology 76:1109–1115.
Mayfield, and F. DeClerck. 2009. Loss of functional diversity Mason, N. W. H., and F. de Bello. 2013. Functional diversity: a
under land use intensification across multiple taxa. Ecology tool for answering challenging ecological questions. Journal
Letters 12:22–33. of Vegetation Science 24:777–780.
Forsythe, T. G. 2000. Ground beetles. Richmond Publishing Mason, N., F. de Bello, D. Mouillot, S. Pavoine, and S. Dray.
Company, Slough, UK. 2013. A guide for using functional diversity indices to reveal
Frenette-Dussault, C., B. Shipley, and Y. Hingrat. 2013. changes in assembly processes along ecological gradients.
Linking plant and insect traits to understand multitrophic Journal of Vegetation Science 24:794–806.
community structure in arid steppes. Functional Ecology 27: Mason, N. W. H., D. Mouillot, W. G. Lee, and J. B. Wilson.
786–792. 2005. Functional richness, functional evenness and functional
1224 ROBIN J. PAKEMAN AND JENNI A. STOCKAN Ecology, Vol. 95, No. 5

divergence: the primary components of functional diversity. Ribera, I., S. Dolédec, I. S. Downie, and G. N. Foster. 2001.
Oikos 111:112–118. Effect of land disturbance and stress on species traits of
Mayfield, M. M., and J. M. Levine. 2010. Opposing effects of ground beetle assemblages. Ecology 82:1112–1129.
competitive exclusion on the phylogenetic structure of Ribera, I., G. N. Foster, I. S. Downie, D. I. McCracken, and
communities. Ecology Letters 13:1085–1093. V. J. Abernethy. 1999. A comparative study of the
McGill, B. J., B. J. Enquist, E. Weiher, and M. Westoby. 2006. morphology and life traits of Scottish ground beetles
Rebuilding community ecology from functional traits. (Coleoptera, Carabidae). Annales Zoologici Fennici 36:21–
Trends in Ecology and Evolution 21:178–185. 37.
Moretti, M. et al. 2013. Linking traits between plants and Saxton, K. E., and P. H. Willey. 2006. The SPAW model for
invertebrate herbivores to track functional effects of land-use agricultural field and pond hydrologic simulation. Pages
changes. Journal of Vegetation Science 24:949–962. 401– 435 in V. P. Singh and D. K. Frevert, editors.
Mouchet, M. A., S. Villéger, N. W. H. Mason, and D. Mathematical modeling of watershed hydrology. CRC Press,
Mouillot. 2010. Functional diversity measures: an overview Boca Raton, Florida, USA.
of their redundancy and their ability to discriminate Schirmel, J., I. Blindow, and S. Buchholz. 2012. Life-history
community assembly rules. Functional Ecology 24:867– trait and functional diversity patterns of ground beetles and
876. spiders along a coastal heathland successional gradient. Basic
Økland, R. H., and O. Eilertsen. 1994. Canonical correspon- and Applied Ecology 13:606–614.
dence analysis with variation partitioning: some comments Stace, C. 2010. New flora of the British Isles. Third edition.
and an application. Journal of Vegetation Science 5:117– Cambridge University Press, Cambridge, UK.
Traugott, M. 1998. Larval and adult species composition,
126.
phenology and life cycles of carabid beetles (Coleoptera:
Oksanen, J., F. G. Blanchet, R. Kindt, P. Legendre, R. B.
Carabidae) in an organic potato field. European Journal of
O’Hara, G. L. Simpson, P. Solymos, M. H. H. Stevens, and
Soil Biology 34:189–197.
H. Wagner. 2011. vegan: community ecology package. R
van der Plas, F., T. M. Anderson, and H. Olff. 2012. Trait
package version 1.17-8. http://CRAN.R-project.org/
similarity patterns within grass and grasshopper communi-
package¼vegan ties: multitrophic community assembly at work. Ecology 93:
Pakeman, R. J. 2011a. Functional diversity indices reveal the 836–846.
impacts of land use intensification on plant community Villéger, S., N. W. H. Mason, and D. Mouillot. 2008. New
assembly. Journal of Ecology 99:1143–1151. multidimensional functional diversity indices for a multifac-
Pakeman, R. J. 2011b. Multivariate identification of plant eted framework in functional ecology. Ecology 89:2290–
functional response and effect traits in an agricultural 2301.
landscape. Ecology 92:1353–1365. Violle, C., M.-L. Navas, D. Vile, E. Kazakou, C. Fortunel, I.
Pakeman, R. J., J. J. Lennon, and R. W. Brooker. 2011. Trait Hummel, and E. Garnier. 2007. Let the concept of trait be
assembly in plant assemblages and its modulation by functional! Oikos 116:882–892.
productivity and disturbance. Oecologia 167:209–218. Weiher, E., G. D. P. Clarke, and P. A. Keddy. 1998.
Pausas, J. G., and M. Verdu. 2008. Fire reduces morphospace Community assembly rules, morphological dispersion, and
occupation in plant communities. Ecology 89:2181–2186. the coexistence of plant species. Oikos 81:309–322.
R Development Core Team. 2010. R: a language and Woodcock, B. A., J. Redhead, A. J. Vanbergen, L. Hulmes, S.
environment for statistical computing. R Foundation for Hulmes, J. Peyton, M. Nowakowski, R. F. Pywell, and M. S.
Statistical Computing, Vienna, Austria. http://www. Heard. 2010. Impact of habitat type and landscape structure
R-project.org/ on biomass, species richness and functional diversity of
Rao, C. R. 1982. Diversity and dissimilarity coefficients: a ground beetles. Agriculture, Ecosystems and Environment
unified approach. Theoretical Population Biology 21:24–43. 139:181–186.

SUPPLEMENTAL MATERIAL
Appendix
Site locations and brief management details and carabid beetle present in the study with their traits used in the analysis
(Ecological Archives E095-104-A1).