Earthworms are an important food resource for a large number of mammals and one of the most largely used method to assess their contribution to mammals' diet is the search for chaetae in faecal samples. Nevertheless, the observation and the identification of these bristles often require properly trained operators and suitable instruments, since their detection can be difficult and time-consuming. We tested the “standard” method for chaetae detection by washing 1046 scats – belonging to four mammal species of the Western Italian Alps (Vulpes vulpes, Martes spp., Lutra lutra, Sus scrofa) – through three sieves differing in mesh size (1.00, 0.54 and 0.21 mm). Mesh size significantly affected the detection of earthworm chaetae, suggesting that the use of three sieves should be recommended in faecal analyses. Moreover, both earthworm size and mean number of chaetae per individual should be assessed in each study area prior to diet analysis.
Introduction
Diet analysis is a major research issue in mammal ecology, especially for predators (Pineda-Munoz & Alroy 2014). This kind of research mainly relies on the examination of stomach contents and faecal samples, the second being a non-invasive technique which does not imply the death of target individuals (Balestrieri et al. 2011). Although widely used, faecal analysis is not free from methodological issues (reviewed by Reynolds & Aebischer 1991), which may lead to either incomplete or unreliable results. In particular, the detection and quantification of some food items can be particularly arduous, especially when undigested remains are very small (< 1 mm), transparent and dispersed within each faecal sample, as in the case of earthworm chaetae.
Earthworms represent a major proportion of the animal biomass of temperate ecosystems (Bouché 1982), and thus play a key role in the diet of several species, including many vertebrates (Granval & Muys 1995). On the basis of the frequency of occurrence (FOC) of earthworms in their diet, earthworm consumers have been classified as occasional (FOC <10 %), regular (10 % < FOC < 50 %) or primary (FOC > 50 %) predators (Granval & Muys 1995). Following this classification, the European badger (Meles meles) is considered a primary earthworm consumer (Skoog 1970, Kruuk & De Kock 1981, Kruuk & Parish 1981, Lucherini & Crema 1995, Balestrieri et al. 2004), and one of the most important mammalian predators of these invertebrates (Granval & Muys 1995), together with shrews (Soricidae) and moles (Talpa europea) (Macdonald 1983). Among European carnivores, many authors have highlighted the consumption of earthworms also by the red fox (Vulpes vulpes) (Jefferies 1974, MacDonald 1980, Cavani 1991, Lucherini & Crema 1994, Balestrieri et al. 2011), and occasionally, least weasel (Mustela nivalis), pine marten (Martes martes) and raccoon (Procyon lotor) (Macdonald 1983, Granval & Muys 1995). In addition, earthworms are an important food resource for wild boars (Sus scrofa; Baubet et al. 2003, 2004).
Since earthworm tissues usually undergo a complete digestion, their detection in faecal samples is mainly related to the observation of gizzard rings (Bradbury 1977) and chaetae (Kruuk & Parish 1981, Wroot 1985). While gizzard rings are fragile, being therefore susceptible to damage during storing processes (e.g. drying or freezing; Wroot 1985), chaetae appear to be a more suitable indicator to assess earthworm contribution to mammal diets. Indeed, these bristles are composed by a protein-chitin complex, hardened by a quinine tanning (Dennel 1949, Morgan 2001), allowing a strong resistance to the digestion process and bacterial degradation. Chaetae (also called setae) are earthworms' locomotion structures and protrude from spots placed laterally and ventrally on every segment of the worm, except the first and the last ones (Sims & Gerard 1985, Edwards & Bohlen 1996, Morgan 2001). Typically, chaetae are f-shaped, with a median nodule representing the insertion of retractor muscles (Sims & Gerard 1985, Edwards & Bohlen 1996) (Fig. 1a-c and 1e). In the Lumbricidae, typically there are eight sigmoid chaetae per segment (Edwards & Bohlen 1996). Similar structures are also involved in earthworm copulation process: genital chaetae are situated in the region of the genital pores (the 26th in Lumbricus terrestris; Hegner 1933), are longer and more slender than the ordinary ones and show a hook-like process in their distal end (Sims & Gerard 1985, Edwards & Bohlen 1996, Morgan 2001) (Fig. 1d).
Since the early 1980s, for assessing the occurrence and volumetric contribution of earthworms in the diets of mammal carnivores, most reports (e.g. Ciampalini & Lovari 1985, Pigozzi 1991, Fedriani et al. 1998, Goszczyński et al. 2000, Fischer et al. 2005, Mysłajek et al. 2016) have made reference to the pioneering study on badger food habits in Scotland by Kruuk & Parish (1981). Briefly, this method entails that each scat is washed through a sieve with a mesh size of 0.5 mm, catching the rinsing water in a beaker. Then the solid fraction is subsampled by pipette, after it has been allowed to settle, and examined using a stereoscopic microscope.
Although this method has subsequently been the object of some criticism (Reynolds & Aebischer 1991), to our knowledge laboratory procedures have no longer been tested. Therefore, in this paper we aimed to verify the effectiveness of sieves with different mesh sizes in retaining earthworm chaetae, and provide some methodological recommendations for improving the precise assessment of the contribution of earthworms to carnivore diets.
Material and Methods
This study was performed using faecal samples belonging to mammal species from the northwestern Italian Alps: red fox, pine- and stone marten (Martes spp.), Eurasian otter (Lutra lutra) and wild boar (Sus scrofa). Scats were identified to species-level in the field, based on their morphology and size (Davison et al. 2002, Monterroso et al. 2013, Laguardia et al. 2015), and, eventually, the occurrence of footprints (Prugh & Ritland 2005). Uncertain samples were discarded.
The study area coincided with the Gran Paradiso National Park (710.4 km2; NW Italy). From December 2013 to July 2015, scats were collected in a large range of Alpine habitats, between 900 and 2300 m above sea level. Eurasian otter spraints were collected while monitoring the movements of an individual escaped from an otter centre (Ferrari et al. 2017).
In laboratory, each faecal sample was washed using three sieves with progressively smaller mesh sizes: 1.00, 0.54 and 0.21 mm respectively. The resulting three fractions were analysed separately, recording earthworm chaetae presence/absence in each of them. Each fraction was partitioned, in as many Petri dishes as necessary to investigate each sub-sample accurately, and diluted in a thin water layer. Chaetae were sorted from other food remains by means of a stereo-microscope (Leica EZ4D) with magnification from 8× to 35× and multiple light sources (above, below and backwards). In addition, bristles were photographed using the software “Leica Application Suite 3.0.0 – LAS EZ”, interfacing with the stereomicroscope.
Results were expressed as absolute percent frequency of occurrence [% OCC: (number of occurrence of earthworm chaetae/total number of faeces) × 100], a parameter largely used in literature to estimate the importance of food categories in mammals' diet. To test whether mesh size affects the detection of earthworm chaetae in faecal analysis, we ran a Generalised Linear Mixed Model (GLMM), with the presence/absence of chaetae in each fraction as dependent variable, and mesh size as fixed effect (categorical with three levels: 0.21, 0.54 and 1.00 mm). To incorporate the dependency among samples belonging to a same species, the univocal code of each sample and species were included in the model as nested random effects (Bolker et al. 2009). GLMM was performed using lme4 (Bates et al. 2015) package for R 3.2.2 (R Core Team 2015).
Fig. 2.
Stereo-microscope magnification (35×) demonstrating the effectiveness of 0.21 mm sieves in holding earthworm chaetae.

Table 1.
Percent frequency of occurrence (% OCC) of earthworm chaetae in the three fractions obtained by washing each faecal sample (N = number of samples analyzed) through three sieves with progressively smaller mesh sizes (1.00, 0.54, 0.21 mm).

Results
Overall, 1046 samples were collected and analyzed, 969 belonging to the red fox, 31 to Martes sp., 23 to the Eurasian otter and 23 to the wild boar. Earthworm chaetae occurred in the faeces of all investigated mammals: red fox (43.96 %), wild boar (43.48 %), Martes sp. (12.90 %) and Eurasian otter (4.35 %) (Table 1).
In our study area, chaetae ranged between 0.5 and 2.5 mm in length and from 0.05 to 0.1 mm in width (Figs. 1, 2). Small hair fragments and the detached bristles of Rosa sp. seeds, both appearing vitreous and sometimes amber-coloured, were identified as a potential source of mistakes in chaetae detection, being similar especially to the genital ones. However, hairs and rose bristles are softer than earthworm setae, thus they can easily be recognized by bending them with a tweezer. Moreover, little crystal splinters may appear similar to chaetae, but their occurrence in faecal samples is rare and their shape is usually irregular.
Mesh size significantly affected the recording of earthworm chaetae in faecal samples. In particular, the 0.54 and the 1.00 mm sieves had a negative effect on bristles detection with respect to the 0.21 mm one (binomial GLMM; 0.54 mm: z = –18.69, P < 0.001; 1.00 mm: z = –18.27, P < 0.001).
Discussion
Effective laboratory procedures can minimize the uncertainty of scat-based diet investigations by allowing the reliable assessment of the contribution of each prey type (Reynolds & Aebischer 1991), a goal which may be of outermost importance, especially when the target prey is a key resource, affecting predators' Darwinian fitness components (Avery et al. 1993) or macronutrient balance (Raubenheimer & Simpson 1997). Moreover, as properly stressed by Reynolds & Aebischer (1991), variation in laboratory procedures are likely to hinder the reliable comparison of results between dietary studies.
We demonstrated that the use of three sieves with different mesh sizes allows a proper sample filtration and the effective detection of earthworms chaetae. In our study area, the best instrument to avoid chaetae loss during the washing process is a sieve with meshes not larger than 0.21 mm. Most diagnostic undigested remains – namely hairs, seeds, feathers and fragments of insects' exoskeleton – are retained by the 1 mm sieve, while the use of a third sieve with an intermediate mesh-size (0.3–0.5 mm) allows to lower the volume of the chaetae-rich sediment of the smallest fraction. Our results suggest that the percentage of chaetae retained by the two upper sieves is negligible and probably proportional to the number of earthworms eaten. Nonetheless, by analysing earthworms from lowland areas of northern Italy, it has been reported that, on average, 31 % of chaetae is lost through washing and sieving (Rosso 1991). Such an error should be taken into account whenever volumes or biomasses have to be assessed.
The detection of earthworm chaetae in the smallest fraction is best performed by diluting the sediment in a known quantity of water. Rather than examining the whole mixture, it is less time consuming to place a subsample in a Petri dish (Kruuk & Parish 1981, Reynolds & Aebischer 1991). We suggest removal by a Beral pipette 3-5 0.5 ml sub-samples after homogenization of the mixture. To avoid double-counting, each subsample may be further separated in several (6-10) drops, which can be examined separately. The average number of chaetae must then be multiplied by the total volume of the mixture to estimate the total number of bristles within the faecal sample.
The conversion of the total number of chaetae to earthworm biomass is a delicate task. With reference to Kruuk (1989), most studies considered 1000 the mean number of chaetae per earthworm. Such a number is valid for as large earthworms as Lumbricus terrestris, which has, on average, 134.7 segments and 0.8 mm long chaetae (Wroot 1985), while the average wet weight is 2700-3000 mg (Carley 1978). However, smaller species are expected to have less segments and smaller chaetae than L. terrestris (Wroot 1985). As an example, in northern Italy the mean number of chaetae per earthworm has been assessed to be 400, corresponding to an average of ∼ 50 segments (Pedrazzini 1990), while earthworm mean weight is 602 mg (Zenato 2010). This also entails that the use of a single sieve with a mesh size of 0.5 mm (Kruuk & Parish 1981) will increase the risk of chaetae loss, enhancing the probability of underestimating earthworm contribution to the diet of the target species wherever (as in NW Italy; Balestrieri et al. 2004, 2009) earthworms smaller than L. terrestris are usually preyed on. As an example, we suspect that the not-thought-out application of Kruuk and Parish's method may explain the recorded variation in earthworm importance in badger diet throughout the Alps (Balestrieri et al. 2009), where small epigeic species form the bulk of earthworm biomass (Omodeo 1952, 1962).
We argue that the assessment of the contribution of earthworms to the diet of mammals, especially those species, such as badger (Kowalczyk et al. 2003), for which worms represent a key resource, would greatly benefit by the prior estimation of both the mean size and number of chaetae of a sample of earthworms representative of the study area. This may be easily achieved by digesting earthworms using a concentrated hydrochloric acid solution (Pedrazzini 1990).
Acknowledgements
The research was funded by the Gran Paradiso National Park; we are especially grateful to Bruno Bassano for logistic and scientific support, while Caterina Ferrari gave us the opportunity of analysing Eurasian otter spraints. We also thank Carla Martinez Ucha and Yolanda Merchan Fernandez, who provided a fundamental help in fieldwork and laboratory analyses, and Elena Piano, for her help with statistical analyses.