Introduction
The use of avermectin veterinary drugs in livestock farming systems and the potential consequences for non-target dung insects has been generating environmental concern for the last two decades (e.g. Wall & Strong, Reference Wall and Strong1987; McCracken, Reference McCracken1993; Hutton & Giller, Reference Hutton and Giller2003; review in Floate et al., Reference Floate, Wardhaugh, Boxall and Sherratt2005). Since the 1980s, avermectin (such as ivermectin and doramectin) and milbemycin (moxidectin) products have been on the animal health market for the control of internal (e.g. lung worm, stomach worm) and external (e.g. ticks, mites) parasites. Following treatment of an animal with an avermectin, active residues of the drug and its metabolites are excreted in the faeces (Campbell, Reference Campbell1985; Halley et al., Reference Halley, Nessel, Lu and Campbell1989), and subsequent exposure to avermectin residues can be harmful for the survival and development of various dung insect species (e.g. Wardhaugh & Rodriguez-Menendez, Reference Wardhaugh and Rodriguez-Menendez1988; Krüger & Scholtz, Reference Krüger and Scholtz1997; Dadour et al., Reference Dadour, Cook and Hennessy2000). Highest concentrations of residues are excreted within several days of treatment, but residues can be excreted for up to several months post-treatment (Herd et al., Reference Herd, Sams and Ashcraft1996; Toutain et al., Reference Toutain, Upson, Terhune and McKenzie1997; Floate, Reference Floate1998; Suarez et al., Reference Suarez, Lifschitz, Sallovitz and Lanusse2003; Floate et al., Reference Floate, Bouchard, Holroyd, Poulin and Wellicome2008).
The overall objective of the current study was to assess the abundance and dispersal of dung beetles (Coleoptera: Scarabaeidae) in response to use of avermectin treatments on pastured cattle. This was achieved by both a field-scale study on livestock farms in southwest Scotland and by a series of four smaller-scale colonisation trials. To our knowledge, no other field-scale studies have been performed in temperate climates although such studies have been performed in South Africa (Krüger & Scholtz, Reference Krüger and Scholtz1998a,Reference Krüger and Scholtzb; Kryger et al., Reference Kryger, Deschodt and Scholtz2005). The smaller-scale studies were performed to identify factors that could potentially be important in influencing field-scale dispersal, such as avermectin treatment, dung moisture content and resource availability (Barth, Reference Barth1993; Wratten & Forbes, Reference Wratten and Forbes1995; Finn et al., Reference Finn, Gittings and Giller1998) in order to better understand and complement the results from the field-scale study.
Methods
Study sites and species
The field-scale study was carried out on seven dairy farms and one beef and sheep farm in Ayrshire, southwest Scotland between late April and mid-July in both 2002 and 2003. This sampling period was selected because treated livestock are grazing fields at this time, and it encompasses the main dung insect breeding season. Prior to sampling, a questionnaire survey of livestock farmers in southwest Scotland ascertained that doramectin and ivermectin (both administered as a pour-on) were the most commonly used avermectins in that region (Webb, Reference Webb2004). Consequently, study farms using these treatments were selected to be representative of the treatment strategies in the wider area.
Over the two years, sampling was conducted in 14 fields grazed by cattle treated with doramectin (Dectomax™ ; n=12 fields) or ivermectin (Ivomec®, Noromectin®; n=2 fields) applied topically at the manufacturer's recommended dose rate of 500 μg per kg body weight. A further 12 fields were grazed by cattle that had received no endectocide treatment during the study period and for at least six months prior. A subset of the fields (six ‘untreated’ and six ‘treated’) was sampled in both years while all of the other fields were sampled during only one of the two study years. For a map showing the spatial distribution and Ordnance Survey grid reference of study fields, see fig. S1 and table S1 in Supplementary Material, respectively.
Fig. S1. Maps showing study location in southwest Scotland and spatial distribution of fields sampled during 2002 and 2003. •, fields grazed by avermectin-treated cattle; ○, fields grazed by untreated cattle.
Table S1. Grid references of fields sampled in the wider field study during 2002 and 2003.
The treated fields in this study were grazed by young livestock, which, because they are more susceptible to parasite infections than older animals, were dosed with an anthelmintic at turnout and thereafter to afford them protection throughout the grazing season. These cattle in treated fields were grazed permanently in the same field for the duration of the season. The untreated fields were grazed by dairy cows, which remain untreated due to acquired parasite immunity and avoidance of drug residues in the milk supply. The untreated dairy cattle were grazed rotationally through several fields on the farms.
On farms in the study area, cattle are typically housed over winter before being turned out to grass in April or May for approximately six months. The time of turnout and the number of months spent out at grass is dependent on weather and the growth and quality of grass. For example, cattle may be periodically returned to housing during particularly wet periods. In two-thirds of the treated fields, cattle were treated at turnout, i.e. within four days before or after turnout in late April or early May, followed by a second dose some eight weeks later. Cattle in the remaining treated fields were dosed within 14 days following turnout and given a second dose some eight weeks later. This variation in timings of treatments reflects the reality of doing studies incorporating a range of commercial farms, but such variation was accounted for within the statistical analyses (see ‘Method: Data analyses’).
We limited our study to dung beetles in the genus Aphodius, which are the most specious genus in temperate climates (e.g. Hutton & Giller, Reference Hutton and Giller2003). Among Aphodius, there are two larval feeding strategies (Gittings & Giller, Reference Gittings and Giller1997): (i) coprophagous larvae that feed exclusively on dung (guild 1); and (ii) saprophagous larvae that feed on plant roots and decaying vegetation (guild 2). Hence, larvae of guild 1 Aphodius will have greater exposure to avermectins than guild 2 larvae; and guild 1 adults could be expected to be relatively more selective when colonising dung, in terms of its potential suitability for larval development, than guild 2 adults. Thus, Aphodius were split into the two guilds according to larval feeding strategy prior to data analyses.
Environmental and management variables
We took into account many of the environmental variables known to be important for Aphodius populations, e.g. weather (Finn et al., Reference Finn, Gittings and Giller1998), seasonal factors (Hanski, Reference Hanski1980a; Gittings & Giller, Reference Gittings and Giller1997), soil type (Ryan et al., Reference Ryan, Clare and Cole1978) and agricultural management practices (Hutton & Giller, Reference Hutton and Giller2003).
Climate data were obtained from the weather station at the Scottish Agricultural College, Auchincruive, which was situated within 1–16 km of the study fields. Total rainfall (mm), sunshine hours and the mean maximum and minimum temperatures were calculated for each individual trapping period. Aspect, altitude, size of field, field boundary type (fence, gappy hedge, well-established hedge, woodland), surrounding land use and avermectin use were recorded for each field. Sward height was measured ten times in the area around the traps, using the direct method (Stewart et al., Reference Stewart, Bourn and Thomas2001), and mean sward height was calculated for each trapping period.
The availability of fresh dung within a study field in each trapping period was gauged using a ‘dung index’. This was the product of the number of days that cattle grazed a field in a particular trap period and the number of cattle, corrected for field size, to give an index of the number of dung pats that would be deposited per unit area per trapping period. An ‘avermectin index’ was calculated to estimate the proportion of land within a 400 m radius of the centre of the study field that was grazed by avermectin-treated livestock. The 400 m radius was selected because it encompasses the maximum area travelled by many dung-breeding insects (Smith & Wall, Reference Smith and Wall1998; Roslin, Reference Roslin2000). Similarly, the ‘pasture index’ estimated the percentage area of grazed pasture within a 400 m radius around the study field to reflect potential availability of livestock dung in surrounding fields.
Larval development of some guild 2 Aphodius species takes place in the soil below dung pats (Gittings & Giller, Reference Gittings and Giller1997), and soil type can influence dung beetle abundance (Ryan et al., Reference Ryan, Clare and Cole1978). To compare general soil quality between fields, six soil cores (10 cm length×6.5 cm diameter) were taken from the area around the traps in each study field and sent for laboratory analysis to determine soil pH, moisture and organic matter. Cores were taken once in January 2003 and once in August 2003, with the former taken to reflect 2002 study fields and the latter taken to reflect 2003 study fields. The temporal difference in sampling meant that comparisons of soil characteristics between years were treated with caution; however, comparisons were made between fields for each year.
An index of management intensity was calculated for each study field. Management information was collected through interview of farmers who were asked about management practices in the fields: sward type, age (time since re-seeding), soil disturbance, cutting regime, grazing intensity, inorganic and organic fertiliser input and herbicide use. A score between 0 and 3 was assigned to each factor and summed to give an overall management intensity score (MIS) between 0–24, with 24 being the most intensively managed (Downie et al., Reference Downie, Ribera, McCracken, Wilson, Foster, Waterhouse, Abernethy and Murphy2000).
Dung beetle sampling – the field-scale study
Adult dung beetles were sampled using dung-baited pitfall traps. Traps were 1-l plastic containers (11.5 cm diameter) sunk flush with the ground and filled with approximately 3 cm depth of 70% monopropylene glycol (to act as a killing agent and preservative). Wire mesh was secured over the traps to support dung baits. Baits were formed using a hemispherical mould (6 cm diameter and 3 cm depth) and were placed on the centre of the mesh. Dung was collected on one occasion in both 2002 and 2003 from grass silage-fed housed cattle at each study farm in April, prior to their receiving any avermectin treatment. Dung from each farm was mixed thoroughly and stored in a sealed container at 4°C. It was observed that dung remained attractive to dung insects after being kept in cold storage for up to six months (Webb, Reference Webb2004). In order to maintain good biosecurity (reflecting concerns following the outbreak of Foot and Mouth Disease in 2001), dung from different farms was kept in separate containers and used as bait only on farms from where it had been collected. Eight pitfall traps were set in two grids of four in each field, with traps spaced approximately 8 m apart within each grid. Grids were set in a central position at each end of the field, away from margins to counteract possible edge effects. Traps were emptied and re-baited every 7–10 days in 2002 and every 14 days in 2003, thus giving a maximum of nine and six collections per field over the sampling season in 2002 and 2003, respectively. In each year, trapping took place in the same fields for the duration of each sampling season. These sampling intervals were selected as the best compromise between sampling a sufficient number of fields and the required sampling effort. Variability in the duration of trapping periods, between five and 20 days, did not affect the composition of species trapped in dung-baited pitfall traps (Webb, Reference Webb2004). On each sampling date, two traps from each of the two grids in each study field were selected, and the individuals of each Aphodius species were counted and summed across those four traps.
Colonisation trials
Four separate trials were conducted in an area of grassland, on the farm at the Scottish Agricultural College, which was ungrazed by livestock but within approximately 200 m of grazing sheep and cattle. Dung beetles were sampled using dung-baited pitfall traps of the design described above. For all experiments, dung was collected from housed Holstein-Friesian cattle fed on grass silage. Dung was collected from the same cattle five days prior to, and two days after, treatment with a doramectin pour-on at the recommended dose of 500 μg per kg bodyweight. The two-day post-treatment collection was selected because relatively high concentrations of residues are excreted in dung within this period following treatment (Toutain et al., Reference Toutain, Upson, Terhune and McKenzie1997); and, in addition, it enabled dung to be collected from housed animals before they were turned out to grass. Dung was mixed thoroughly to homogenise it and was stored in a sealed container at 4°C until use. The specific details for each colonisation trial are given below.
Trial 1
The aim of this trial was to determine whether Aphodius dung beetles exhibited a preference to, or an avoidance of, dung from doramectin-treated cattle. In the first phase of the trial, traps were set in two grids of 3×2, with traps within each grid spaced 1.5 m apart (fig. S2). The two grids were established approximately 70 m apart. All six traps in one grid were baited with dung from untreated cattle, and the six in the other grid were baited with dung from treated animals. In May 2003, traps were set and the contents collected after five days. To exclude any bias in the location of traps, each grid of traps was then re-baited with the alternate dung type, and the trap contents were collected after a further five days. After checking for no significant differences between the two trapping periods, collections from the two trap periods were pooled to increase sample size, and thus data from twelve untreated traps and twelve treated traps were analysed.
Fig. S2. Diagram showing the spatial location of pitfall traps baited with untreated dung (○) and treated dung (
) for trial 1.
Trial 2
The experimental set-up was as described for trial 1 with the exception that, within each grid of six traps, traps were baited alternately with ‘treated’ and ‘untreated’ dung to determine whether Aphodius made a choice between the two dung types at a scale of 1.5 m (fig. S3). In June 2003, traps were set and emptied after five days and then re-baited with the same dung type and collected after a further five days. Trap collections from the two trap periods were pooled, as for trial 1, prior to analyses.
Fig. S3. Diagram showing the spatial location of pitfall traps baited with untreated dung (○) and treated dung () for trial 2.
Trial 3
The aim of this trial was to assess the combined influence of moisture content and doramectin treatment on dung colonisation by Aphodius. Dung was homogenised and water was mixed into one lot of treated and untreated dung to increase moisture content. Water was added until the dung reached a consistency that was extremely moist but still could be supported on the mesh of the pitfall trap. Three samples (approximately 20 g each) of each of the four dung types were taken and dried in an oven to constant weight to determine moisture content. Eight replicate traps baited with each of the four dung types were set out in an 8×4 grid, with traps spaced approximately 5 m apart, and each of the four dung types were used in a repeated alternating sequence (fig. S4). Traps were collected in May 2004 after six days exposure.
Fig. S4. Diagram of layout of traps baited with ‘treated dry’ (TD), ‘treated wet’ (TW), ‘untreated dry’ (UD) and ‘untreated wet’ (UW) dung for trial 3.
Trial 4
The availability of suitable dung in the vicinity of pitfall traps could influence the number of Aphodius trapped. This trial investigated whether the number of beetles caught in pitfall traps baited with dung was affected by the presence of fresh dung pats within 1 m of those traps. Two 4×2 grids of eight traps were set up and all traps were baited with untreated dung. The two grids were approximately 70 m apart, and individual traps within a grid were spaced at 1.5 m (fig. S5). In order to simulate a field that was permanently stocked with cattle, six ‘artificial’ pats of approximately 20 cm diameter were formed from the collected untreated dung and placed within 1 m of the traps at one grid. At the other grid, the baited pitfall traps had no dung surrounding them (to simulate the intermittent periods where no dung would be deposited in a rotationally grazed field). Traps were set in May 2004 and collected after seven days.
Fig. S5. Diagram of trap layout in trial 4 with pitfall traps baited with untreated dung and surrounded either by simulated cow pats or no cow pats.
Data analyses
Potential effects of avermectins on Aphodius population size at different times of the season were investigated using general linearised mixed models in SAS (SAS Institute, 2001). We used the GLIMMIX macro with a Poisson error, log link function and a Satterthwaite correction to derive the degrees of freedom for the denominator. Abundance data from the field samples were corrected to a standard period of ten days to allow comparisons between trapping periods of different duration. ‘Field’ and ‘farm’ were included as random factors, and sampling date was included as a repeated measure. The autoregression of order 1 covariance structure was used because it assumes that measures further apart in time are less correlated than measures closer together (Littell et al., Reference Littell, Milliken, Stroup and Wolfinger1996). The Akaike's Information Criterion was used to support selection of the best covariance structure and model (Littell et al., Reference Littell, Milliken, Stroup and Wolfinger1996). Ordination was used to simplify and reduce the number of habitat parameters introduced into the model (Fox, Reference Fox2004; Rushton et al., Reference Rushton, Ormerod and Kerby2004). Altitude, aspect, field boundary type and ‘pasture index’ data were ordinated using detrended correspondence analysis to generate a ‘habitat’ score based on the first two axes scores for each site and, hence, reflecting the degree of similarity between sites with regard to these characteristics. All environmental and management variables (table 1) were tested independently for significance at the P<0.05 level. Those variables that were significant were then added to a multivariate model using a step-wise procedure, whereby variables were retained in the model if they were significant and any variables that were no longer significant were removed. To consider non-linear relationships, the quadratics of all continuous variables and interactions of interest were tested and retained in the model if significant. Only the final models are presented in this paper.
Table 1. List of environmental and management variables included in mixed model analyses of Aphodius abundance data. The recorded ranges of continuous variables across ‘treated’ and ‘untreated’ fields are given in parentheses. Note that the negative number of days in the ‘days post-turnout’ category refers to the fact that some sampling occurred prior to cattle being turned out to pasture.
The Mann-Whitney test was used to compare Aphodius abundance in trials 1, 2 and 4. Data were collated across two trap periods in those trials; therefore, patterns of significance were checked for consistency across the two trap periods to rule out variation due to the time of trapping or trap location. For trial 3, data were square-root transformed to normalise and then analysed using a two-way analysis of variance to assess the significance of treatment, moisture level and the interaction between the two.
Results
Dung beetle abundance in grazed pastures
Guild 1 species
Over the two study years for guild 1 species, 1552 and 3534 beetles were collected in fields grazed by untreated and treated cattle, respectively. Guild 1 species comprised Aphodius depressus (Kugelann) (80%), A. rufipes (L.) (13%) plus A. ater (Degeer), A. fimetarius (L.), A. fossor (L.), A. lapponum Gyllenhal, A. pusillus (Herbst), and A. rufus (Moll) (7% combined).
The April to July sampling period in 2003 was warmer (mean maximum temperature 16.1°C) and drier (230.5 mm total rainfall) than in 2002 (14.3°C and 312 mm, respectively). More guild 1 beetles were recorded in 2002 than in 2003 and in fields grazed by treated cattle versus untreated cattle (fig. 1, table 2). The abundance of guild 1 beetles peaked in late April and early May, then subsequently declined. There was a significant non-linear relationship between rainfall and guild 1 abundance. Total rainfall in individual trapping periods ranged from 0.2–67.4 mm, and numbers of guild 1 Aphodius were lowest in trapping periods with 30–42 mm of rainfall.
Fig. 1. Mean number (±SE) of guild 1 Aphodius beetles trapped in treated and untreated fields in each sample month in 2002 and 2003 (□, untreated 2002; 
, treated 2002; ▪, untreated 2003; 
, treated 2003).
Table 2. Generalised linear mixed model (see methods) describing the variation in Aphodius guild 1 abundance in fields containing avermectin-treated and untreated cattle sampled from April to July in 2002 and 2003. All interactions were non-significant.
Guild 2 species
Of the two species belonging to guild 2 (Aphodius prodromus and A. sphacelatus), 931 individuals were trapped in untreated fields and 5843 individuals were trapped in fields grazed by treated cattle. Aphodius prodromus was trapped more frequently (77.8% of the total catch of guild 2 species) than A. sphacelatus (22.2%). Guild 2 beetles were trapped in higher numbers in 2002, had their highest abundance in April and May and then decreased continuously as the sampling season progressed (fig. 2, table 3). There was a significant interaction between year and avermectin treatment. In 2002, guild 2 individuals were recorded more often in treated than in untreated fields (F1, 43.9=7.27, P=0.01), whereas in 2003 there was no difference between treated and untreated fields (F1, 22.9=0.97, P=0.34).
Fig. 2. Mean number (±SE) of guild 2 Aphodius beetles trapped in treated and untreated fields in each sample month in 2002 and 2003 (□, untreated 2002; , treated 2002; ▪, untreated 2003; , treated 2003).
Table 3. Generalised linear mixed model (see methods) describing the variation in Aphodius guild 2 abundance in fields containing avermectin-treated and untreated cattle sampled from April to July in 2002 and 2003. All interactions were non-significant.
Although dung density was not significant in either of the models, it is notable that trapping periods in 2003 had significantly higher dung density indices (mean±SE, 51.7±2.8, n=120 trapping periods) than those in 2002 (32.5±1.7, n=123 trapping periods; F1, 217=11.21, P=0.001). There was no significant difference in dung density between fields containing untreated or avermectin-treated cattle in 2002 (F1, 14.2=0.2, P=0.66) or in 2003 (F1, 18.8=1.6, P=0.22).
The variables ‘farm’ and ‘field’, which were included as random effects in both guild 1 and guild 2 models were not significant in either. This indicates that there was no significant level of variation in dung beetle abundance between individual farms or individual fields.
Colonisation trials
The abundance of each dung beetle guild caught in traps in trials 1 and 2 are summarised (table 4). When traps were 70 m apart, untreated dung attracted significantly more guild 1 (Mann-Whitney, n=24, P<0.001) and guild 2 individuals (Mann Whitney, n=24, P<0.0001) than treated dung. In traps spaced 1.5 m apart, significantly more guild 2 beetles were attracted to untreated dung (Mann Whitney, n=24, P<0.001), but this difference was not significant for guild 1 beetles (Mann Whitney, n=24, P>0.05).
Table 4. Mean abundance (±SE) of Aphodius beetles in pitfall traps: baited with dung from treated and untreated cattle at 70 m apart (trial 1) and 1.5 m apart (trial 2); baited with dung of different moisture levels from treated and untreated cattle (trial 3); baited with untreated dung and surrounded either by dung pats (present) or no dung pats (absent) (trial 4).
Numbers of Aphodius in traps baited with dung of varying treatment and moisture level were compared in trial 3 (table 4). ‘Moist’ dung had a higher moisture content (88.8+0.2%) than ‘dry’ dung (85.2+0.3%; F1, 8=248.97, P<0.001) irrespective of avermectin treatment (F1, 8=4.83, P=0.06; table 4 for percentage moisture). For Aphodius guild 1, significantly more individuals were attracted to traps baited with dung from untreated cattle than to dung from doramectin-treated cattle (F1, 28=6.58, P=0.02) but the effect of dung moisture was not significant (F1, 28=0.68, P=0.42). However, the interaction between moisture and treatment was close to significance (F1, 28=4.05, P=0.054) as more beetles were attracted to ‘dry’ untreated dung than to any other dung type. This difference was driven by the colonisation behaviour of A. ater, which was more attracted to traps baited with untreated ‘dry’ dung than to all other dung types (Kruskal Wallis test: H=8.67, df=3, P=0.03). Guild 2 Aphodius were more attracted to untreated dung than to treated dung (F1, 28=18.27, P<0.001), regardless of moisture level (F1, 28=0.31, P=0.58).
In trial 4, traps with dung pats present trapped similar numbers of guild 1 beetles as traps where dung pats were absent (Mann Whitney, n=16, P=0.67). There was a non-significant trend towards higher numbers of guild 2 Aphodius in traps where dung pats were absent than in those where dung pats were present (Mann Whitney, n=16, P=0.07; table 4).
Discussion
Colonisation trials – relative attractiveness of dung
These trials indicated that Aphodius beetles belonging to each guild preferred to colonise dung from untreated cattle rather than doramectin-treated cattle. Guild 2 beetles were relatively more attracted to untreated dung regardless of moisture content, while guild 1 species may select dung on the basis of both moisture content and treatment.
Our results have shown that Aphodius dung beetles can discriminate between dung from untreated cattle and cattle treated with a doramectin pour-on at a spatial scale of at least 70 m. This preference for dung from untreated cattle was evident even when dung type was alternated between trapping grids; and, therefore, the observation was not a consequence of trap location. The ability to distinguish between dung from treated and untreated cattle at such a scale has implications for the movements of beetles between natural cattle-grazed pastures. Dung beetles tend to remain in the pasture in which they emerge rather than dispersing large distances (Roslin, Reference Roslin2000). Therefore, Aphodius beetles are unlikely to disperse further than to adjacent pastures provided there is a suitable dung resource to colonise nearby. Our results suggest that, where fields grazed by untreated cattle surround a pasture grazed by avermectin-treated cattle, emergent beetles from the ‘treated’ pasture may disperse to colonise dung in the adjacent ‘untreated’ fields. However, if a pasture of treated cattle is surrounded by fields also grazed by treated cattle or by land with no dung resource, e.g. arable cropping, then the insects may have little choice but to colonise dung with avermectin residues.
Our study also indicates that dung beetles can select between dung from untreated cattle and dung from doramectin-treated cattle at a spatial scale of 1.5 m, whereby higher numbers of guild 2 beetles occurred in traps baited with untreated dung. Previous research has shown that Aphodius beetles can distinguish between dung from untreated and avermectin-treated cattle at 3.5 m (Holter et al., Reference Holter, Sommer, Grønvold and Madsen1993a) and 3 m (Floate, Reference Floate2007), and our findings indicate that they can do this at an even finer scale.
Other studies on the attractiveness of dung from avermectin-treated cattle to insects have yielded mixed results. As in our research, some studies have found untreated dung to be more attractive to Aphodius beetles than dung containing ivermectin residues (Holter et al., Reference Holter, Sommer and Grønvold1993b), while others have shown that dung beetles preferred dung from avermectin-treated livestock than from untreated livestock (Wardhaugh & Mahon, Reference Wardhaugh and Mahon1991; Floate, Reference Floate1998). Floate (Reference Floate2007) found seven cases where Aphodius were more attracted to dung from doramectin-treated animals compared to two cases where Aphodius preferred untreated dung. These conflicting observations cannot be attributed to differences in study methodologies. It has been proposed that differences in attraction could be due to changes in cattle diet (Barth, Reference Barth1993; Floate, Reference Floate1998). However, this can be excluded as a contributory factor in our study because cattle were fed on the same silage diet throughout the dung collection process. Differences in the attractiveness of dung could be due to a change in dung quality caused by avermectin treatment rather than simply an avoidance of, or attraction to, the avermectin compound. For example, Wratten & Forbes (Reference Wratten and Forbes1995) suggested that avermectin treatment may result in dung with lower moisture content due to the alleviation of worm-induced diarrhoea and Wardhaugh & Mahon (Reference Wardhaugh and Mahon1991) suggested that changes in the gut flora of livestock following avermectin treatment could alter dung quality. Our study indicated that moisture was not critical for the attraction of guild 2 species, as there was an overwhelming preference by A. prodromus and A. sphacelatus for untreated dung regardless of moisture content. As these two species oviposit in soil (Gittings & Giller, Reference Gittings and Giller1997), they do not have the same constraint as dung-ovipositing species in terms of eggs drowning in very moist dung. Both moisture and treatment were important for guild 1 beetles, whereby more individuals were attracted to drier untreated dung than to the drier treated dung. This difference was mainly due to the colonisation behaviour of A. ater, a species known to favour dry dung for oviposition (Hirschberger & Degro, Reference Hirschberger and Degro1996; Gittings & Giller, Reference Gittings and Giller1997, Reference Gittings and Giller1998).
Hence, our colonisation trials suggest that doramectin treatment indirectly reduced the suitability of dung for Aphodius beetles. The mechanism by which avermectins alter the attractiveness of dung remains unclear. We speculate that the quality of dung as a feeding resource is reduced following treatment. Adult beetles feed on the energy-rich bacteria that are abundant in fresh dung (Hirschberger, Reference Hirschberger1999); therefore, a significant reduction or change in the bacterial assemblage could potentially make the dung less nutritious and less desirable. Alternatively, the attractiveness of dung to beetles may be lower due to reduced activity of pre-colonising species. Larvae of dung-breeding flies aerate dung during their development, thus making it more suitable for beetle colonisation (e.g. Suarez et al., Reference Suarez, Lifschitz, Sallovitz and Lanusse2003). Avermectin residues in dung can kill or impair fly larvae (e.g. Fincher, Reference Fincher1992; Floate et al., Reference Floate, Spooner and Colwell2001) and Scathophaga stercoraria L. (Diptera: Scathophagidae) larvae may undergo developmental stress in dung from treated cattle (e.g. Strong & James, Reference Strong and James1993; Webb et al., Reference Webb, Beaumont, Nager and McCracken2007). Therefore, dung fly larval activity within individual dung pats may be impeded.
The results from trial 4 indicated that abundance of dung beetles in traps did not differ with surrounding dung density. However, there was a non-significant trend for more guild 2 beetles in traps without alternative dung pats present, suggesting a possible ‘dilution effect’ whereby fewer individuals were trapped when there were greater quantities of suitable dung nearby.
It is notable that there was large variation in numbers of beetles trapped in the colonisation trials where mean abundance (±SE) ranged from 0.92±0.45 to 842±105 individuals. The reason for the observed higher abundance of Aphodius in trial 4 compared to the other three trials is unclear, as it could not be attributed to location of traps, sample year or season. It is possible that, because a greater quantity of dung (i.e. the artificially formed pats) was present around the pitfall traps, relatively more dung beetles were attracted into the vicinity of the traps. Results from the colonisation trials are discussed below in the context of our wider field study on avermectin effects on Aphodius populations.
Field-scale study
These results show that fields containing avermectin-treated livestock had significantly higher numbers of Aphodius dung beetles recorded in baited pitfall traps than fields containing untreated livestock. This difference in abundance could not be attributed to habitat characteristics of the field, such as aspect, boundary type, sward height or dung density, or wider landscape factors, such as availability of grazed pasture. We propose that increased trapping of Aphodius in fields grazed by treated livestock could have been due to differences between pastures in terms of avermectin treatment and/or grazing regime.
Avermectin treatment
The presence of avermectin residues in dung, in a pasture situation, could have affected Aphodius abundance via the ‘attraction/repellency effect’ discussed above. All baits on the pitfall traps were formed from dung from untreated cattle. Hence, inflated numbers of beetles may have occurred in traps in treated fields if beetles were avoiding the naturally occurring dung from avermectin-treated cattle in those fields.
Livestock grazing regime
Average dung density did not differ significantly between treated and untreated fields and was not a significant factor in the models. However, the ‘pattern’ of dung deposition differed between treated and untreated fields, whereby the former were grazed permanently by cattle while the latter were grazed rotationally by cattle at a relatively higher stocking density. As dung beetles may follow the movements of cattle around pastures (Finn et al., Reference Finn, Gittings and Giller1998), it cannot be ruled out that the lower abundance of Aphodius in untreated fields in this study was due to beetles migrating from a field as cattle were periodically rotated. Conversely, we might expect fewer beetles to occur in traps in permanently grazed fields because a constant fresh supply of natural dung would be available for beetles to colonise. However, if the freshly deposited dung in those fields was unattractive then one could expect more beetles to be attracted to the untreated dung on traps in these fields. This would explain our observation of more beetles being trapped in treated fields.
Variation due to environmental factors
Abundance of Aphodius changed significantly between the two sample years. It is not uncommon for Aphodius abundance to vary naturally between years (Finn et al., Reference Finn, Gittings and Giller1999). Duration of trapping period varied between years in this study because, for logistical reasons, trapping periods ranged from 7–10 days in 2002 but were increased to 14 days in 2003. Therefore, we must be cautious when comparing abundance between years.
The model shows that the relationship between guild 1 Aphodius and rainfall was not straightforward, as insect activity was highest during trapping periods with close to no rainfall and in trapping periods with most rainfall. Heightened beetle activity during low rainfall could be explained by a negative correlation between rainfall and sunshine, as beetle activity is positively correlated with radiant energy (Lobo et al., Reference Lobo, Lumaret and Jay-Robert1998). Moreover, insects may have been more active during periods of low rainfall because heavy rain can impair activity (Finn et al., Reference Finn, Gittings and Giller1998). Conversely, rainy weather and a lack of sunshine can optimise pat colonisation by preventing formation of a hard impenetrable crust on dung. It was noted that fields sampled in 2002 became extremely waterlogged because of high rainfall in June and July (personal observation). The dung beetles that occurred in high numbers in the spring of 2002 could have been expected to breed successfully in dung at that time. However, disintegration of pats and flooding could have impaired the subsequent development of larvae throughout the period of high rainfall in that year.
Seasonality was significant for guild 1 beetles although there was no distinct seasonal pattern presumably because the two most common species have phenologies that ‘balanced’ each other. The most abundant species of that guild, A. depressus, peaked in abundance in May 2002, while A. rufipes was typically most abundant in June and July (White, Reference White1960; Holter, Reference Holter1979; Gittings & Giller, Reference Gittings and Giller1997). Seasonality was an important factor for the guild 2 species, A. prodromus and A. sphacelatus, which were trapped in highest numbers in April and May. This seasonal pattern is typical of these two species (Wassmer, Reference Wassmer1994; Gittings & Giller, Reference Gittings and Giller1997).
Management recommendations
Dung beetles' avoidance of dung from doramectin-treated cattle in a pasture environment could minimise any negative effects associated with exposure to residues if alternative avermectin-free dung is available for them to colonise. Thus, any management practice that maximises the availability of doramectin- and ivermectin-free dung in the landscape could be beneficial for dung insects. On dairy farms, adjacent fields often hold the same livestock cohort, i.e. younger treated cattle are often grazed together in a cluster of adjacent fields, while untreated milking cows are rotationally grazed through fields that are often situated together, close to the milking parlour (Webb, Reference Webb2004). Where possible, young stock undergoing avermectin treatment should be grazed in fields beside untreated milking cows to enable dung beetles to disperse between fields to locate the more attractive dung from untreated cattle. This could be achieved either at an individual farm level or by taking into account the locations of treated and untreated animals on neighbouring farms. Livestock managers should also consider (consulting a veterinary surgeon as necessary) whether all animals require to be dosed or whether older animals have acquired immunity. It is recommended that the precautionary principle be retained in areas managed for nature conservation, especially where there are likely to be populations of dung invertebrates or dependent wildlife of high conservation value. In such areas, a wormer from an alternative anthelmintic class (e.g. benzimidazoles) or a less toxic avermectin, such as moxidectin (Floate et al., Reference Floate, Spooner and Colwell2001, Reference Floate, Colwell and Fox2002) should be used where possible. If a doramectin or ivermectin product must be used, then altering the time of treatment so that it is outside the main dung insect breeding period could be beneficial, i.e. restricting use to autumn or when livestock are housed.
Acknowledgements
This research was carried out as part of a PhD studentship funded by the Royal Society for the Protection of Birds and the Scottish Agricultural College Dr D. MacLagan Trust Fund. The Scottish Agricultural College receives financial support from the Scottish Government Rural and Environment Research and Analysis Directorate (RERAD).
Supplementary material
The online table and figures can be viewed at http://journals.cambridge.org/ber
