The potential of parasitoid Hymenoptera as bioindicators of arthropod diversity in agricultural grasslands

Authors

  • Annette Anderson,

    Corresponding author
    1. UCD School of Agriculture, Food Science and Veterinary Medicine, Agriculture and Food Science Centre, University College Dublin, Belfield, Dublin 4, Ireland
      Correspondence author. E-mail: annette.anderson@ucd.ie
    Search for more papers by this author
  • Stephen McCormack,

    1. UCD School of Agriculture, Food Science and Veterinary Medicine, Agriculture and Food Science Centre, University College Dublin, Belfield, Dublin 4, Ireland
    Search for more papers by this author
  • Alvin Helden,

    1. UCD School of Agriculture, Food Science and Veterinary Medicine, Agriculture and Food Science Centre, University College Dublin, Belfield, Dublin 4, Ireland
    2. Animal and Environmental Research Group, Department of Life Sciences, Anglia Ruskin University, East Road, Cambridge, CB1 1PT, UK
    Search for more papers by this author
  • Helen Sheridan,

    1. UCD School of Agriculture, Food Science and Veterinary Medicine, Agriculture and Food Science Centre, University College Dublin, Belfield, Dublin 4, Ireland
    Search for more papers by this author
  • Anne Kinsella,

    1. Rural Economy Research Centre, Teagasc, Mellows Campus, Athenry, Co. Galway, Ireland
    Search for more papers by this author
  • Gordon Purvis

    1. UCD School of Agriculture, Food Science and Veterinary Medicine, Agriculture and Food Science Centre, University College Dublin, Belfield, Dublin 4, Ireland
    Search for more papers by this author

Correspondence author. E-mail: annette.anderson@ucd.ie

Summary

1. As measuring biodiversity in its entirety is impractical, there is a need for bioindicators. This study tested the hypothesis that parasitoid Hymenoptera are potential bioindicators that provide a useful means to assess the wider biodiversity of arthropod populations in agro-ecosystems. There are a range of theoretical arguments to support such a claim, including the high trophic position of these taxa within the arthropod communities in which they occur, and the unique nature of their biological relationships with the majority of terrestrial arthropod groups.

2. A survey of 48 commercial farms was conducted and Generalized Linear Models used to investigate relationships between six taxa—parasitoid Hymenoptera, Coleoptera, Hemiptera, Diptera, Araneae and plants (species richness and sward height)—in agricultural grasslands. As well as relationships between these groups, the relationship of each individual group to the overall biodiversity of all other arthropod groups was explored.

3. Both abundance (r2 = 0·58) and taxon richness (r2 = 0·54) of parasitoid Hymenoptera had stronger relationships with overall arthropod taxon richness than any other arthropod group investigated. Parasitoid abundance also had a positive relationship with species richness of Coleoptera (r2 = 0·23) and Hemiptera (r2 = 0·47).

4. An historical data set demonstrated how the relationship between parasitoid abundance and overall arthropod taxon richness changes over the growing season. July, when the relationship was strongest, is potentially the most useful time to sample.

5. For use in routine monitoring, it is important that an effort be made to understand the seasonal influence on the relationship in the context being studied. Equal sampling effort must be made for all sites being compared and sites should be sampled as close together in the season as is possible.

6.Synthesis and applications. We show that, within agricultural grasslands, both the abundance and taxon richness of parasitoid Hymenoptera are more closely related with overall arthropod diversity than any other arthropod group investigated. The use of parasitoid abundance provides a simple and practicable monitoring tool for tracking change in wider arthropod diversity in agro-ecosystems.

Introduction

The use of the term biodiversity has become popular and widespread in the last 20 years. It is used for all aspects of biological diversity including taxon richness, genetic variation and ecosystem complexity (Magurran 2004). Maintaining biodiversity has been linked with both ecosystem functioning and stability and the processes involved in sustaining ecosystem functions (Tilman et al. 2001; Tilman, Reich & Knops 2006). Measuring biodiversity in its entirety is impractical because of the numbers of species involved and the effort that would be required for their collective identification. There is therefore a need to find appropriate bioindicators or surrogate taxa that can be monitored on the basis of their being representative of wider biodiversity and/or used for conservation planning (Duelli & Obrist 2003a; Favreau et al. 2006; Lewandowski, Noss & Parsons 2010). A biological indicator can reflect the abiotic or biotic state of an environment; represent the impact of environmental change; or be indicative of wider diversity within an area (McGeoch 1998).

There can be no single indicator for all aspects of biodiversity in all contexts (McGeoch 1998). The issue is often guided by the availability of taxonomic expertise and resources (Duelli & Obrist 2003b). However, an essential first step in selecting useful biodiversity indicators is to identify taxa whose incidence best correlates with overall taxon richness in a particular context (Sauberer et al. 2004). In order for bioindicators to be used to their fullest advantage, it is also necessary to understand the ecological relationships between the chosen indicator group(s) and wider community structure, as well as the particular ecological influences they reflect (Paoletti 1999). Invertebrates have potential as bioindicators as they are small and often mobile, have short generation times and are sensitive to local conditions such as temperature and moisture changes (Samways, McGeoch & New 2010). However, after an extensive review of the literature on surrogate and indicator taxa, Favreau et al. (2006) found that, due to the diversity of studies involved, general conclusions on the success of surrogate species approaches in conservation could not be drawn.

The Convention on Biological Diversity (CBD) included an agreement to integrate biodiversity policy into all economic sectors (UNEP 1992). Clearing of land for agriculture has resulted in a decline of biodiversity (Benton et al. 2002; Green et al. 2005). Following the signing of the CBD, the European Commission (2000, 2001) published a Biodiversity Action Plan (BAP) for Agriculture [COM (2001)162 vol. III] as part of a strategy to halt the global decline in biodiversity by 2010. Whilst this ambitious aim has not been met, the need to halt biodiversity loss remains (Butchart et al. 2010).

Agriculture accounts for about 62% of total land use in Ireland [DAFF, (Department of Agriculture, Fisheries and Food) 2008]. Due to the intensification of farming methods, there has been a drastic change in the Irish landscape since the middle of the last century, and a widely perceived decline in biodiversity similar to that experienced across much of western Europe. It has been suggested that between 1970 and 2000, species populations in European farmland declined by an average of 23% (de Heer, Kapos & ten Brink 2005). Considering the relatively large proportion of farmed land in the Irish landscape, amelioration of the negative impacts of agricultural practices leading to increased biodiversity within farmland, could have a positive influence on biodiversity throughout the wider countryside.

Parasitoid Hymenoptera comprise one of the most diverse groups of arthropods (Gauld & Bolton 1988; Quicke 1997), are an important functional component of agricultural ecosystems (Altieri, Cure & Garcia 1993; Marino & Landis 2000), and are known to be sensitive to habitat fragmentation and environmental changes (Kruess & Tscharntke 1994; Siemann 1998; Lewis & Whitfield 1999; Komonen et al. 2000; Fraser, Dytham & Mayhew 2008; Maeto et al. 2009). An initial survey of 10 agricultural grassland sites in the southeast of Ireland gave an indication that parasitoid Hymenoptera, identified to family level, have potential as bioindicators of other terrestrial arthropods (Anderson et al. 2005; Table S1, Supporting Information). The aim of the current study was to test the hypothesis that, because of their close ecological relationships with practically all other insect groups (Gauld & Bolton 1988; Quicke 1997), the parasitoid Hymenoptera have good theoretical potential as functionally significant bioindicators within agro-ecosystems (LaSalle & Gauld 1993). This hypothesis is tested within the context of agriculturally managed grasslands by:

  • 1Investigating relationships between the taxon richness of six taxa – parasitoid Hymenoptera (genus and family), Coleoptera (species), Hemiptera (species), Diptera (family), Araneae (species) and plants (species richness and sward height), and the total taxon richness of all other arthropod groups (i.e. when parasitoids were the explanatory variable, the response variable was the total taxon richness of Coleoptera + Hemiptera + Diptera + Araneae).
  • 2Exploring the relationship between the abundance and taxon richness of each of the five arthropod taxa and taxon richness of each individual arthropod group (e.g. when parasitoids were the explanatory variable, the response variable was Coleoptera, Hemiptera, Diptera or Araneae).

Materials and methods

Site selection

Fifty commercial farms were chosen for an extended survey of grassland arthropod populations in the relatively intensively farmed East and south-eastern counties of Ireland (Carlow, Cork, Kilkenny, Laois, Meath, Waterford, Wexford and Wicklow) (see Table S2, Supporting Information for site locations). The Irish National Farm Survey maintains a nationally representative data base of Irish farm statistics by regular collation of farm management data from a sample of over 1,000 farms (Connolly, Kinsella & Quinlan 2004). A random subsample of 50 predominantly grassland farms, proportionately stratified by selected counties and the main livestock farm enterprise (dairying, beef cattle or sheep) were selected from this data base; see Anderson et al. (2008a) for further details of the selected farms. As two of the farms could only be sampled under less than optimal wet conditions, they were excluded from the analyses, leaving a total of 48 farms in the study.

Arthropod sampling

Arthropods were sampled from one randomly chosen grassland field on each farm in July 2005, using a Vortis Insect Suction Sampler (Burkard Manufacturing Co. Ltd, Rickmansworth, Hertfordshire, UK) (Arnold 1994), which has been found to be effective for sampling arthropods in agricultural grassland vegetation (Brook et al. 2008). Twenty samples were taken from each field, with each sample comprising six randomly positioned suction samples (each of 10 s duration). On intensive, rotationally grazed pastures, these samples were taken at the mid-point between grazing periods, when the sward was between 7 and 14 cm in height. The total area sampled per field was 2·4 m2. Catches were preserved in 70% ethanol prior to sorting and identification. Five major arthropod groups dominated the collected samples. Araneae, Hemiptera and Coleoptera were identified to species level (see Anderson et al. 2008a; Helden, Anderson & Purvis 2008a,b respectively, for details of the taxonomic literature used). Diptera were identified to family level. The parasitoid Hymenoptera (hereafter referred to as parasitoids) were initially identified to family level, and subsequently to genus level using the literature listed by Anderson et al. (2008b).

Vegetation sampling

Plant species richness and dry-weight were estimated within sampled swards on each of the 48 farm sites. Data were collected from 50 randomly located circular quadrats (3 dm2) per pasture, using the dry-weight-rank (DWR) method (‘t Mannetje & Haydock 1963) with yield correction (Jones & Hargreaves 1979). This method was developed to provide a rapid and accurate estimate of the botanical composition of heterogeneous agricultural grassland swards on a dry-weight basis, without requiring the clipping and hand sorting of constituent species (‘t Mannetje & Haydock 1963; Neuteboom, Lantinga & Struik 1998). The DWR method assumes that when a large number of samples are taken, the dominant species will on average account for 0·702, the second for 0·211 and the third for 0·087 of total herbage mass. All other species were recorded on a presence–absence basis. Species were identified according to Stace (1997). Further details on the DWR method of botanical assessment are provided in ‘t Mannetje & Haydock (1963) and Neuteboom, Lantinga & Struik (1998). Mean vegetation height was estimated within each sampled sward by recording height measurements at 50 randomly selected locations using a Filips Folding Plate Pasture Meter (http://www.jenquip.co.nz).

Data analysis

Taxon accumulation and singleton (number of taxa represented by a single individual) curves, indicating the rate at which new taxa were collected over the 48 sites, were created using the programme EstimateS version 7·5·0 (Colwell 2001). Chao 1 estimates of taxon richness were also calculated using EstimateS version 7·5·0 (Colwell 2001).

Using pooled data from individual sites as replicates (= 48), the relationships between the taxon richness of each major arthropod group in turn, and the total taxon richness of all other arthropod groups (excluding the group being evaluated) were determined using Generalized Linear Model (GLM) analyses in the R statistical package; version 2·11·1 (R Development Core Team 2010). The models included sampling date as a covariate and had a Poisson error structure. In addition, the relationships between individual arthropod groups were investigated. When the data were overdispersed, a quasi–Poisson error structure was specified. Values of r2 were calculated using the R statistical package; version 2·11·1 (R Development Core Team 2010).

The relationship between the abundance of each arthropod group, and the total taxon richness of other arthropod groups was similarly quantified.

The relationships between arthropod taxon richness and plant species richness or mean sward height respectively were investigated using GLM as described previously.

Because of the multiple comparisons it was necessary to perform Bonferroni adjustments of the P-value (Sauberer et al. 2004). In testing relationships between taxon richness (= 31) or abundance (= 26) of individual groups and overall arthropod richness P-values were adjusted to 0·0016 (0·05/31) and 0·0019 (0·05/26), respectively.

Historical data set

Available data

Seasonal variation in the relationship between total parasitoid abundance and the total taxon richness of all other arthropod groups was further investigated using a data set collected in 1976–1977 (Purvis & Curry 1981). Weather conditions varied in these years, with 1976 being an exceptionally dry summer and 1977 being a relatively wet summer (Purvis & Curry 1981). Taxonomic literature available at the time precluded parasitoid identification. However, this historical data set did allow temporal variation in the parasitoid abundance relationship to be investigated. Samples were collected on a monthly basis from replicated field plots using a D-Vac suction sampler (Master Fan Corporation, Los Angeles, California; Dietrick 1961) (see Purvis & Curry (1981) for further details). For the purpose of the current study, total catches were pooled across 10 individual sampling points in each of 10 experimental plots in each sampling month (May, June, July and August) and year (1976 and 1977). The total area of each pooled monthly sample (= 10) was 9 m2.

Data analysis

Generalized linear mixed modelling using the lmer function from the lme4 package in R version 2·11·1 (R Development Core Team 2010) was used to investigate the relationship between the total abundance of parasitoids and the taxon richness of other arthropods. A maximal model with a Poisson error structure, including parasitoid abundance and sampling month as fixed effects, and year of sampling as a random effect, was fitted. Likelihood ratio tests (χ2) were used to determine significant model parameters. As the 2 years differed in weather conditions, model outputs were plotted separately.

In order to determine the optimum month in which to quantify the relationship between parasitoid abundance and overall arthropod taxon richness, individual GLM with Poisson error structure were used for each sampling month and r2 values calculated for each model as described above.

Results

Arthropods

A total of 95,910 adult arthropods, representing 431 taxa were collected from a total pooled sample area of 120 m2 from the 48 sites. This included 36 Araneae species, 179 Coleoptera species, 63 Hemiptera species, 121 parasitoid genera and 32 families of Diptera. Taxon accumulation curves showed that, with the exception of Araneae spp. and Diptera families, new taxa continued to be found in sampling the 48 sites (Fig. 1a). However, for all groups except the Coleoptera, the number of singletons collected reached an asymptote (Fig. 1b).

Figure 1.

 Curves showing (a) accumulation of arthropod taxa and (b) number of singletons collected from a survey of 48 commercial farms in southeast Ireland.

With the exception of Coleoptera and Araneae, the taxon richness of each group was significantly related to the total taxon richness of all other groups (Table 1a). The significant relationships between parasitoid genera (Fig. 2a), parasitoid families (Fig. 2b) and Hemiptera species, and overall arthropod richness were maintained even after Bonferroni correction (P ≤ 0·0016). Date was significant in 45% of the 31 models (Table S3, Supporting Information), including all those that were significant after Bonferroni correction, suggesting that relationships vary with the growing season. Parasitoid genera richness had the greatest r2 value (0·538) for the relationship with total other arthropod taxon richness (Table 1a; Fig. 2). Full details of the GLM analyses can be found in Table S3, Supporting Information.

Table 1.   Summary of GLM’s describing relationships between (a) group taxon richness or (b) abundance of individual arthropod groups, and sward characteristics, as bioindicators for other taxa
Response variable
Explanatory variable:Total (other) arthropod richnessParasitoid (genera)Coleoptera spp.Hemiptera spp.Diptera familiesAraneae spp.
(a) Group taxon richness
Parasitoid (genera)0·538 (<0·0001)0·391 (0·024)0·430 (<0·0001)0·147 (0·186)0·310 (0·084)
Parasitoid (families)0·452 (0·0002)0·371 (0·108)0·284 (<0·0001)0·081 (0·332)0·317 (0·085)
Coleoptera spp.0·165 (0·085)0·179 (0·020)0·062 (0·209)0·075 (0·354)0·172 (0·844)
Hemiptera spp.0·470 (<0·0001)0·462 (<0·0001)0·364 (0·139)0·036 (0·522)0·177 (0·615)
Diptera families0·429 (0·002)0·243 (0·0005)0·415 (0·016)0·122 (0·018)0·181 (0·629)
Araneae spp.0·282 (0·117)0·257 (0·0001)0·352 (0·626)0·048 (0·353)0·017 (0·696)
Plant spp.0·306 (0·111)
Response variable
Explanatory variableTotal other taxon richnessParasitoid (genera)Coleoptera spp.Hemiptera spp.Diptera familiesAraneae spp.
  1. The r2 values are shown with P-value in parentheses. Significant relationships after Bonferroni correction (P ≤ 0.0016 for Table 1a and 0.0019 for Table 1b) are shown in bold. (Details of the significance of sampling date in the models are provided in Table S3, Supporting Information)

(b) Abundance
Parasitoid0·586 (<0·0001)0·500 (0·0001)0·224 (0·0007)0·332 (0·047)0·170 (0·837)
Coleoptera0·218 (0·012)0·230 (0·0008)0·079 (0·101)0·072 (0·361)0·182 (0·624)
Hemiptera0·374 (0·023)0·110 (0·030)0·473 (0·0001)0·036 (0·522)0·225 (0·314)
Diptera0·300 (0·688)0·132 (0·218)0·353 (0·584)0·041 (0·939)0·206 (0·435)
Araneae0·284 (0·154)0·128 (0·374)0·399 (0·032)0·044 (0·600)0·043 (0·486)
Sward height0·327 (0·048)
Figure 2.

 Significant relationships between (a) parasitoid genera richness, (b) parasitoid family richness, and (c) parasitoid abundance and overall taxon richness of all other arthropods collected from 48 commercial farms in southeast Ireland. All relationships remain significant following Bonferroni correction for multiple comparisons.

The total abundances of Coleoptera, Hemiptera and parasitoids (Fig. 2c) were also significantly related to the total taxon richness of all other groups (Table 1b), with date also being significant in the latter two (Table S3, Supporting Information). However after Bonferroni adjustment, only parasitoid abundance remained a significant predictor of the total taxon richness of all other groups (Table 1b). Overall, the abundance and genera richness of parasitoids (r2 = 0·586 and 0·538, respectively) provided the best predictors of total arthropod taxon richness, with the species richness of Hemiptera providing the ‘next best’ relationship with overall arthropod taxon richness (r2 of 0·470) (Table 1). Over 20% of the total 48·52% deviance explained by the Hemiptera model was accounted for by date (Table S3, Supporting Information). In contrast, only about 14% of the total 55·41% deviance explained by the parasitoid taxon richness model was accounted for by date. In the parasitoid abundance model, this figure was approximately 10% (Table S3, Supporting Information). The relationship between parasitoid abundance and overall Chao 1 estimate of taxon richness approached an asymptote at approximately 6,000 parasitoid individuals (Fig. 3).

Figure 3.

 The relationship between the number of parasitoid individuals collected and a Chao 1 estimate of taxon richness of all other arthropods collected from the 48 site survey.

The richness of parasitoid genera and families were also good predictors of Hemiptera species richness. Conversely, Hemiptera species, Araneae species and Diptera family richness were good predictors of parasitoid genera richness (Table 1a; Fig. S4, Supporting Information).

In analyzing relationships between individual group abundance and the taxon richness of other groups, parasitoid abundance was a good predictor of both Coleoptera and Hemiptera species richness. Significant relationships were also found between Coleoptera abundance and parasitoid genera richness and between Hemiptera abundance and Coleoptera species richness (Table 1b; Fig. S5, Supporting Information).

Plants

A total of 32 species of plant were identified from the swards on the 48 surveyed farms and Lolium perenne was the dominant species (Table S6, Supporting Information). No significant relationship was found between plant species richness and arthropod taxon richness (Table 1a). Mean sward height had a marginally significant relationship (P = 0·048) with total arthropod taxon richness, but was no longer significant following Bonferroni adjustment (Table 1b).

Historical data set

The data from the 1970s showed a similarly strong relationship between parasitoid abundance and other arthropod taxon richness (Table 2; Fig. 4), although seasonality had an influence on the relationship (Fig. 4). The abundance of both parasitoids and other arthropod taxa was substantially greater in 1977, the wetter year (Purvis & Curry 1980, 1981), although seasonal trends in the relationship between overall arthropod richness and parasitoid abundance were similar in both years. Earlier in the season, when the total abundance of parasitoids was relatively low, the slope of the line was relatively steep. Towards the end of the season, however, parasitoid abundance reaches a peak and the slope begins to flatten (Fig. 4). It is important not to extrapolate beyond the data, as the relationship likely reaches an asymptote as the seasonal diversity of arthropod taxa reaches a late summer peak (Purvis & Curry 1980). The relationship between parasitoid abundance and other arthropod taxon richness was particularly strong in May for both sampling years and in July 1977 with r2 values of 0·873, 0·800 and 0·872 respectively (Table 3).

Table 2.   Summary of likelihood ratio tests, following a generalized linear mixed model (lmer) investigating the relationship between parasitoid abundance and total taxon richness of other arthropods collected in the historical data set
Parameter testedχ2P-value
  1. Parasitoid abundance and sampling month were included as fixed effects and year of sampling as a random effect. For the likelihood ratio tests, year was the null model and other parameters and interactions (shown by *) were systematically added to subsequent models.

Sampling month73·40<0·001
Parasitoid abundance89·05<0·001
Year*month13·850·003
Year*Parasitoid abundance16·21<0·001
Parasitoid abundance*month77·97<0·001
Figure 4.

 Model prediction of the relationship between the abundance of parasitoids and the total taxon richness of all other arthropods (excluding parasitoids) collected from agricultural grassland swards over four seasons in (a) 1976 and (b) 1977, respectively. Fitted lines indicate the relationships in May (×), June (+), July (triangles) and August (circles) respectively.

Table 3.   Summary of linear models investigating relationships between parasitoid abundance and other taxon richness in individual sampling months from the historical data set
Sampling monthr2Model P-value
May 19760·873<0·001
May 19770·800<0·001
June 19760·0520·625
June 19770·5050·0075
July 19760·5000·0647
July 19770·872<0·001
August 19760·551<0·001
August 19770·676<0·001

Discussion

We recommend that parasitoid Hymenoptera can be used as an indicator group of arthropod taxon richness in agricultural grasslands. Abundance, family and genera richness of parasitoids were all found to have positive significant relationships with overall arthropod taxon richness. The strongest observed predictor of overall arthropod taxon richness was parasitoid abundance. With the exception of a relationship between Hemiptera abundance and Coleoptera species richness, all significant relationships between individual groups involved the parasitoids in some way. Our studies therefore support the hypothesis that, of the groups investigated, parasitoid Hymenoptera have the greatest potential as indicators of arthropod taxon richness in agricultural grasslands.

Many invertebrate groups have been shown to be good ecological indicators of environmental change (Buchs 2003; Maleque, Maeto & Ishii 2009). In addition, many groups, often more easily identified than parasitoids, have been used as indicators or surrogate taxa of wider biodiversity (Asteraki, Hanks & Clements 1995; Butterfield et al. 1995; Favreau et al. 2006; Scott, Oxford & Selden 2006; Lewandowski, Noss & Parsons 2010). However, Lawton et al. (1998) investigated correlations between birds, butterflies, flying beetles, canopy beetles, canopy ants, leaf litter ants, termites and soil nematodes and found that no one group was a good indicator of overall taxon richness. Studies of less intensively managed semi-natural and natural habitats, have shown a bottom-up influence in which botanical and vegetation structural diversity reflect diversity at higher trophic levels (Lawton & Schroder 1977; Crisp, Kickinson & Gibbs 1998; Siemann 1998). Similarly, the diversity of hedgerow invertebrates was found to be strongly related to plant diversity (Bowden & Dean 1977) and plants have also been shown to be good indicators of arthropod taxon richness (Asteraki et al. 2004; Sauberer et al. 2004). Invertebrates also responded positively to plant architecture in field margins of intensively managed grassland (Woodcock et al. 2007, 2009). Fraser et al. (2009) and Fraser, Dytham & Mayhew (2007) found that vegetation of woodlands in agricultural landscapes can be used as a surrogate indicator of the species richness of one group of parasitoids, the pimpline Hymenoptera, for conservation purposes.

However, as managed grasslands have a simplified botanical diversity, it was not surprising that in the current study, plant species richness was a poor predictor of arthropod diversity. Vegetation structure, however, as measured by mean sward height did have a marginal relationship with overall arthropod taxon richness, but this was not as strong as the relationships with parasitoid abundance or taxon richness. Our data suggest that the trend towards vegetation simplification in agro-ecosystems may make taxa at higher levels in the trophic pyramid more sensitive to, and therefore a better bioindicator of, changes in the ecosystem. A similar observation was noted by Purvis et al. (2009) regarding the apparently greater sensitivity of parasitic ‘Cuckoo’ bumblebees (Psithyrus spp.) to changing agricultural intensity in comparison to non-parasitic Bombus spp. In this context, hymenopteran parasitoids would be expected to be good indicators of arthropod diversity as they are amongst the most speciose taxa in the world, parasitise virtually all other arthropod groups (Askew & Shaw 1986; Hassell 1986; Altieri, Cure & Garcia 1993; Hawkins 1993; LaSalle 1993; Quicke 1997) and play a key ecological role in biological control (Altieri, Cure & Garcia 1993). Accordingly, parasitoid species richness of the family Braconidae has been found to be useful for monitoring the recovery of forest biodiversity in Indonesian plantation stands of East Kalimantan forests (Maeto et al. 2009).

The strong relationship between total parasitoid abundance and overall arthropod richness is potentially the easiest and most practical relationship to use as a simple monitoring tool, and its legitimacy was supported by our analysis of the historical data set. This observation has particular significance, as the reliable identification of parasitoids, even to family level, requires considerable expertise and time, making it logistically difficult to integrate into a realistic framework of widespread monitoring. The sampling and quantification of parasitoid abundance in arthropod communities, however, is a relatively straightforward and practicable option for routine monitoring that eliminates the need for taxonomic expertise. The relationship between simple parasitoid abundance and wider arthropod diversity needs to be used with a degree of caution. As can be inferred from Fig. 3, the abundance relationship has the form of a species accumulation curve, suggesting that, as more sampling is undertaken, greater numbers of both individual parasitoids and other arthropod taxa are recovered. As more sampling is undertaken, the relationship effectively reaches an asymptote as the numbers of other arthropod taxa reach an upper limit, whilst the numbers of parasitoids continue to increase with further sampling. In the current study, the accumulation of arthropod taxa started to level off when approximately 6,000 parasitoid individuals had been collected. This relationship is influenced by seasonal changes in both arthropod taxon richness and parasitoid abundance. Peak abundance and taxon richness of arthropod populations in agricultural grasslands occurs in mid-late summer (July–August) (Purvis & Curry 1980, 1981).

Our study of 48 commercial farm sites was undertaken over a period of approximately 5 weeks in July/August, and the resulting model using parasitoid abundance as a predictor of other arthropod taxon richness showed that sampling date explained only 10% of the model deviance. However, as Fig. 4 shows, both the number of taxa and individuals, and the relationship between them (which is strongest in May and July), is likely to vary much more over a whole growing season. The slope of the relationship changes considerably between these months, as sampling in May collects relatively low density, early season generations, whilst sampling in July collects seasonally increased populations. Therefore, July would probably be the best time to sample. It seems plausible that the parasitoid abundance relationship may hold in other agro-ecosystems with constrained botanical diversity, such as arable crops, making parasitoids a more widely useful bioindicator group for agro-ecosystems. However, the relationship needs to be tested in other crop habitats.

The practical difficulties of using parasitoid taxon richness as a bioindicator of the changing diversity of wider arthropod populations when combined with increasing knowledge of parasitoid–host relationships, make their use as bioindicators both a significant challenge, and a strategy with great potential advantage. By using knowledge of host group affinities, a greater insight into the underlying drivers of environmental and/or management change may be gained by identifying the parasitoid taxa, and by inference and association with their host group, the wider fauna being affected. Theoretically, parasitoids could be used to document effects on cryptic biodiversity, including plant mining and gall-forming host groups that would be difficult to document by any other means. Further examination of our current and additional data sets is ongoing to demonstrate this additional potential.

Conclusion

We have shown that overall parasitoid abundance can be used as a practical indicator of wider arthropod biodiversity. Including information on the identification of parasitoids provides more biological information but requires substantial expertise and, in this study, did not add to the usefulness of parasitoid abundance as an indicator of arthropod biodiversity. For use in routine monitoring, it is important that an effort be made to understand the seasonal influence on the relationship in the context being studied, and that, subsequently, equal sampling effort is made for all sites being compared; sites should also be sampled as close together in season as possible. Agri-environment schemes are compulsory under European Union regulations, but are typically implemented with little if any monitoring. A simple assessment of parasitoid count in arthropod communities could provide a useful base level assessment. This will hopefully stimulate an interest in the development and practical application of knowledge of this fascinating and ecologically meaningful bioindicator group, in terms of both management and conservation.

Acknowledgements

This project was funded by the EPA through the ERTDI (2000–2006) programme as part of the National Development Plan. We thank Tim Carnus for statistical advice, along with Rónan Gleeson, Julie Melling and Yasmine Lovic for assistance with arthropod collection and sorting; the Teagasc sites and farmers for their cooperation and farm access; Dr. Gavin Broad, Dr. Andrew Polaszek and Dr. John Noyes (Natural History Museum, London), and Hannes Bauer (Natural History Museum, Bern) for assistance with verifying and determining parasitoid identifications; Dr Jim O’Connor for access to references and the collection at the National Museum of Ireland (Natural History); and three anonymous reviewers for valuable comments on an earlier version of this manuscript.

Ancillary