Soil Biology & Biochemistry 35 (2003) 313–322 www.elsevier.com/locate/soilbio Feeding activity of the earthworm Eisenia andrei in artificial soil Tjalling Jager*, Roel H.L.J. Fleuren, Willem Roelofs, Arthur C. de Groot Laboratory for Ecotoxicology, National Institute of Public Health and the Environment (RIVM), P.O. Box 1, NL-3720 BA Bilthoven, The Netherlands Received 3 June 2002; received in revised form 19 September 2002; accepted 29 September 2002 Abstract Quantitative information on the feeding activity of earthworms is scarce but this information is valuable in many eco(toxico)logical studies. In this study, the feeding activity of the compost worm Eisenia andrei is examined in artificial soil (OECD medium), with and without a high-quality food source (cow manure), and at two temperatures (10 and 20 8C). Methods are provided to estimate the most important parameters: gut load, selection of organic matter (OM), digestion efficiency, compaction, gut retention time, and fraction of manure in the diet. Lanthanides (Lu and Tm) were successfully used as inert markers in soil and manure, and we applied Bayesian statistics to analyse the data and fully capture the compounded uncertainty in the parameter estimates. Results show that the compost worm does not feed on soil indiscriminately but is able to select an OM-enriched diet from apparently homogeneous OECD medium. When manure is present on the soil surface, approximately three-quarters of the diet still consists of soil particles. The gut load of the worms was approximately 10% (dwt gut/wwt empty worm), varying little with the treatments. Unfortunately, the digestion efficiency could only be reliably estimated at 20 8C, and was approximately 40%. Temperature clearly affected feeding as a 108 temperature decrease nearly doubled the gut retention time (from 2.9 to 5.5 h), which corresponds to a two-fold decrease in feeding rate. The present data may be used to interpret toxicity and accumulation studies with E. andrei in OECD medium. However, care must be taken, as it seems possible that feeding is influenced by the size of the worm and subtle differences in experimental set-up. q 2003 Elsevier Science Ltd. All rights reserved. Keywords: Earthworms; Eisenia andrei; Feeding activity; Digestion; Gut load; Gut retention time 1. Introduction Earthworms play a vital role in many soil ecosystems, but quantitative information on their feeding habits is scarce. Nevertheless, this information is particularly valuable for studying earthworm energy budgets (Bolton and Phillipson, 1976), nutrient cycling in ecosystems (Whalen et al., 1999), dung removal from pastures (Hendriksen, 1991), and the uptake of chemicals by earthworms from their food (Jager, 2003). Because much of the earthworm’s activity is located below the surface, it is difficult to observe and quantify feeding without disturbing the organisms. It may seem that earthworms just feed indiscriminately on soil particles at a fixed rate, but a few examples illustrate that the situation is more complex. Firstly, earthworms are able to select a particular fraction from the soil matrix that is more * Corresponding author. Present address: Department of Theoretical Biology, Vrije Universiteit, FALW, De Boelelaan 1085, NL-1081 HV Amsterdam, The Netherlands. Tel.: þ 31-20-444-7134; fax: þ 31-20444-7123. E-mail address: [email protected] (T. Jager). organic than the bulk soil (Bolton and Phillipson, 1976), moreover, the exact preference seems to be species specific (Piearce, 1978). Secondly, many earthworm species prefer high-quality food sources like leaf litter or manure. However, even when such a preferred food source is present, the diet will still consist of an appreciable amount of mineral soil (Barley, 1958; Hendriksen, 1991), possibly because of the mechanical grinding action of the mineral particles (Schulmann and Tiunov, 1999). As a third example, the retention time of materials in the gut may depend on whether the animal is feeding or making new burrows (Barley, 1958; Parle, 1963). The set of parameters necessary to describe the feeding activity depends on the subsequent use of the data. The following set is needed to describe feeding in an extended bioaccumulation model for chemicals (Jager, 2003): gut load, digestion efficiency, selection of organic matter (OM) from soil, weight decrease of the gut contents due to digestion and absorption of food (compaction), gut retention time, and the fraction of a specific food source in the diet. Digestion efficiency and 0038-0717/03/$ - see front matter q 2003 Elsevier Science Ltd. All rights reserved. PII: S 0 0 3 8 - 0 7 1 7 ( 0 2 ) 0 0 2 8 2 - 1 314 T. Jager et al. / Soil Biology & Biochemistry 35 (2003) 313–322 OM selection may be determined from measuring carbon in the crop contents and faeces (Morgan and Morgan, 1992), gut load by weighing the faeces produced on starvation (Hartenstein et al., 1981). A dynamic approach is, however, needed to determine the retention time. Simple methods relate the defecation rate to the weight of the gut contents (Barley, 1958; Bolton and Phillipson, 1976) or follow faecal production (egestion) of recognisable ingested material (Curry and Bolger, 1984; Hartenstein et al., 1981). Others apply inert markers; i.e. materials that are not assimilated by earthworms but can be easily measured like chromic oxide (Hendriksen, 1991). In this study, rare earth metals (lanthanides) are proposed. Lanthanides have no known biological function and have already been successfully applied as inert markers in studies with ruminants (Ellis, 1968) and fish (Austreng et al., 2000). They can be added to the soil or a food source in solution (e.g. as chloride salts), thereby allowing homogeneous distribution and likely preventing irritation (which could lead to shorter retention times; Parle, 1963). These compounds are easily measured by ICP-MS in low concentrations, and have very high sorption coefficients in soils (Jones, 1997) and sediments (Weltje et al., 2002), making them unavailable for uptake into the earthworm’s tissues (Helmke et al., 1979). In the present study, several methods are combined to obtain a complete set of parameters for the compost worm (Eisenia andrei) in artificial soil (OECD, 1984). The feeding activity is quantified in the presence and absence of a food source (ground cow manure), and at two temperatures (10 and 20 8C). Although this system is not really representative for soil ecosystems, it may act as a model system. Artificial soil and ground cow manure are highly homogeneous and the combination with E. andrei is recommended for routine testing of chemicals (OECD, 1984). If quantification of the feeding habits is not possible in this simplified system, it will certainly fail for more field-relevant situations. 2. Material and methods The symbols that are used in this study are explained in Table 1, along with their units. 2.1. Exposure media and acclimatisation of earthworms Artificial soil (OECD medium) was prepared according to OECD (1984). The water content was brought to 40% (l water/kg dry medium). After wetting the soil, it was stored in closed plastic containers at 5 8C for 4 – 6 weeks, prior to the experiments. After storage, the pH (KCl) of the OECD medium was 5.0. Cow manure was oven dried, ground (0.5 mm), and brought to 100% water content (l water/kg dry manure). The fraction OM (Fom) in the OECD medium was 11% and in cow manure 52% (determined as described in Section 2.2). Sub-adult earthworms (E. andrei, weighing between 200 and 300 mg), were taken from mass cultures at our laboratory. First, the animals were allowed to evacuate their gut contents by keeping them on moist filter paper for 24 h at 20 8C. Subsequently, the animals were transferred to plastic containers with 175 g of unspiked OECD medium. Three animals were used per container, and the containers were placed at 10 and 20 8C, covered by a black plastic pot to minimise disturbance. For the animals that were fed with manure, ground cow manure was wetted and added several times (ad libitum conditions) in a small hole at the soil surface. The animals were left to acclimatise for approximately 1 week under these conditions after which they were used for the experiments described below. 2.2. Determination of carbon in the crop and the anterior gut Per treatment, six containers were used (three animals each). After 1 week under the test conditions, animals were killed by brief immersion in hot tap water. Crop contents (representative for ingesta) as well as gut contents from the posterior part of the gut (last cm of the gut, representing egesta) were removed by dissection and weighed separately into aluminium cups for carbon analysis. Crop contents of three animals were combined in one cup; contents from the posterior part of the gut were analysed individually when possible (values were averaged afterwards). Additionally, samples of soil and manure were taken. All samples were freeze dried before carbon analysis (elemental analyser by Fisons Instruments Model EA 1108, Rodana, Italy). Carbon levels were translated to Fom by multiplication with 1.7. The digestion efficiency of OM after gut passage (Fdig) may be directly calculated from the Fom in ingesta and egesta Fdig ¼ Fom-ingesta 2 Fom-egesta Fom-ingesta ð1Þ This equation is sufficiently accurate as long as the digestion efficiency is not too high and the Fom of the ingesta is low. Otherwise, digestion and absorption of OM will result in a significant weight decrease of the gut contents (compaction), thereby affecting Fom of the egesta (which are effectively concentrated). This effect is usually ignored in feeding studies, and, therefore, we need to derive a new equation for digestion, accounting for compaction. If we assume that the weight decrease of the gut contents during gut passage is only due to digestion of OM, we obtain the following relation for the compaction factor (Fcom) Fcom ¼ ingestion rate ðkg=hÞ 1 ¼ egestion rate ðkg=hÞ 1 2 Fdig Fom-ingesta ð2Þ The validity of this relationship between digestion and compaction was verified using the data of Dickschen and Topp (1987) who followed feeding of Lumbricus rubellus T. Jager et al. / Soil Biology & Biochemistry 35 (2003) 313–322 on a diet of leaf litter in the absence of soil medium. Although this study provides an excellent way to check our assumptions, the use of a litter-only diet leads to very high compaction rates, irrelevant for field situations. When (low OM) soil is part of the diet, compaction is likely to be (much) less important (Bolton and Phillipson, 1976; Piearce, 1972). When compaction is significant, the measured Fom in the egesta should satisfy the following equation Fom-egesta ¼ Fcom ð1 2 Fdig ÞFom-ingesta ð3Þ This equation could also be verified on the basis of the data from Dickschen and Topp (1987). Eqs. (2) and (3) can be rearranged to free Fdig; leading to an estimate of digestion on the basis of the measured OM content in ingesta and egesta, and accounting for the effects of compaction ! Fom-egesta 1 Fdig ¼ 1 2 ð4Þ Fom-ingesta 1 2 Fom-egesta The measured carbon content in the crop can also be used to derive a selectivity factor (Fsel) for OM. This factor is defined as the ratio of the organic carbon content in the crop and the soil, and gives an indication of the ability of earthworms to select an enriched fraction from the soil matrix. 2.3. Weight of the faeces on starvation Per treatment, four containers were used (three animals each). After 1 week of acclimatisation, the animals were placed in a petri dish with wet filter paper to evacuate their gut contents (24 h at 20 8C). Animals as well as their faeces were weighed (faeces were collected after drying the entire petri dish overnight at 80 8C). The animals were transferred to clean filter paper and left for another 24 h. Again, worms and faeces were weighed. An estimate of the growth rate could be made as the worms were also weighed before the acclimatisation. Under constant environmental conditions, the weight of the gut contents appears to be a fixed fraction of the body weight (Bolton and Phillipson, 1976; Hartenstein et al., 1981; Curry and Bolger, 1984). The weight fraction of faeces (Fege) was taken as the total dry weight of gut contents evacuated in 48 h, divided by the worm wet weight after 24 h. The body weight after 24 h was used, as the worms seemed to lose more weight between 24 and 48 h than could be attributed to gut contents alone (presumably water loss from the tissue). Furthermore, the fraction of the gut contents remaining in the worm after 24 h starvation (Frem) was calculated from the faecal weights. The fact that compaction of the gut contents may occur during gut passage also has consequences for the measurements of the gut load. Weighing the faeces provides us with an estimate of the gut load after compaction. When compaction of the gut contents is not negligible, the faecal 315 weight (Fege) cannot be related directly to feeding rates. Instead, the weight of the gut contents before compaction (Fing) is required, which can be calculated as the product of Fege and Fcom. The relative weight of the gut contents present in a living worm (Fgut) is calculated as the average of Fing and Fege. Body growth is not problematic for this assessment, as long as the gut load remains a constant fraction of the body weight. 2.4. Experiments with inert markers Pilot experiments showed that E. andrei does not appreciably absorb lanthanides from OECD medium. This is true for the naturally present lanthanides as well as spiked salts of lutetium (Lu) and thulium (Tm). Lu and Tm were selected as these have very low background concentrations in OECD medium (approx. 40 mg kg21 dwt). Animals taken directly from the culture may behave differently in the initial phase following transfer to OECD medium. Therefore, all animals were pre-treated on this medium (Section 2.1) before transfer to medium spiked with lanthanides. Per treatment, 11 containers were used (three animals each). Lu and Tm were purchased as hydrated chloride salts (purity 99.9%) from Alfa Aesar (Karlsruhe, Germany). Chemicals were dissolved in tap water and added to the dry medium (Lu to soil, Tm to cow manure) to achieve nominal concentrations of 15 mg kg21 dwt. The medium was immediately mixed with a high-speed blender for several minutes. The spiked OECD medium was stored in closed plastic containers at 5 8C for 4 – 6 weeks, prior to the experiments to achieve equilibrium in the soil system. The spiked cow manure was stored at 5 8C for 3 days only, to minimise fungal and bacterial growth. After the acclimatisation period, the animals were transferred to similar containers, containing OECD medium spiked with Lu. The soils had been moved to test temperatures 1 day before the animals were introduced, and for the treatments with food, 3 gwwt of Tm-spiked cow manure was added to the soil surface. After an exposure period (0, 1, 2, 3, 5, 7, 9, 15, 24 and 48 h), the animals were recollected, rinsed in tap water, weighed and frozen at 2 20 8C (including their gut contents). Four samples of soil and three of manure were taken to check the spiked concentrations. An additional container with earthworms was used to check whether the worms assimilated the chemicals. These worms were exposed for 48 h under the test conditions, and allowed to evacuate their gut contents for 48 h on moist filter paper at 20 8C (the filter paper was changed after 24 h). Before analysis, the frozen earthworms, soil and manure samples were freeze dried for 2 days. An acid digestion of the samples was performed as described earlier (Janssen et al., 1997a,b), including procedural blanks for contamination in the digestion procedure, and for the proper functioning of the microwave oven. In order to check the digestion procedure, the certified 316 T. Jager et al. / Soil Biology & Biochemistry 35 (2003) 313–322 reference material NIST 2710 ‘Montana soil’ (Garthersburg, MD, US) was used. Because Lu and Tm were not certified in this material, Cd was used as a target compound for the digestion efficiency (under the assumption that this compound is representative for the spiked Lu and Tm). The recovery for cadmium was within the certified range indicating that the digestion was complete. The digests were analysed for Lu and Tm on an ICP-MS (Hewlett Packard 4500plus, Avondale, PA, US). Standard addition experiments in the different matrices digests showed excellent recovery of these elements (100 ^ 5%) indicating that interferences did not occur during final analysis. Limits of determination were 15 and 3 mg kg21 for Lu and Tm, respectively. Table 1 Summary of the parameters used in this study (symbols, explanation and units) 2.5. Models used to analyse experiments with inert markers Fgut After the earthworm is transferred to the spiked soil, the concentration of Lu in the worm will linearly increase until the entire gut is filled with Lu-spiked medium (assuming perfect ‘plug-flow’ conditions). While the animal is replacing clean soil with spiked soil in its gut, the total concentration in the worm (including gut contents) is given by Fing Cw ðtÞ ¼ Qf t Fsel Cs Ww Fsolids þ Fgut Ww ð5Þ Symbols Explanation Unit Cf Cs Cw Concentration of marker in food Concentration of marker in soil Concentration of marker in total worm (incl. gut contents) Compaction factor of gut contents during gut passage Digestion efficiency of OM during gut passage Egested faeces (after compaction), fraction of empty worm wwt Fraction of specific food source in diet Gut weight, fraction of empty worm wwt Ingested weight (gut weight fraction before compaction) Fraction OM Fraction of total gut contents, remaining after 24 h starvation Factor by which ingesta contain more OM or Lu than bulk soil Fraction solids in empty worm (wet-dry weight ratio) Time needed before worm starts to feed on surface Retention time of material in the gut Feeding rate Worm wet weight mg kg21 dwt mg kg21 dwt mg kg21 dwt Fcom Fdig Fege Ffood Fom Frem Fsel Fsolids Tlag Tret where t is the exposure time (other symbols explained in Table 1). Because the worm may select specific fractions of the soil, a selectivity factor for Lu is included (Fsel). The wwt/dwt ratio of the worm (Fsolids) enters the equation because Fgut is expressed on the basis of worm wet weight while Cw is measured on a dry weight basis. The Fsolids was determined by comparing the weight of empty worms before and after freeze-drying. The feeding rate (Qf) can be written as a function of the gut weight before compaction (Fing Ww) and the retention time of the gut contents (Tret) as Qf ¼ Fing Ww Tret ð6Þ After replacing Eq. (6) in Eq. (5), the full concentration equation becomes Fing t Cw ðtÞ ¼ F C min ;1 ð7Þ Tret Fsolids þ Fgut sel s The construction at the right-hand side of the equation (with the minimum of t/Tret and 1) ensures that the uptake is linear until the retention time is reached, after which no further increase of the total concentration in the worm takes place. Growth is not problematic for this assessment, as long as the gut load remains a constant fraction of the body weight (Cw is weight-based, just as Fing and Fgut). In two of the four treatments, Tm-spiked manure is given at the soil surface. The earthworms will feed on Qf Ww (wt fraction) (wt fraction) kgdwt kg21 wwt (wt fraction) kgdwt kg21 wwt kgdwt kg21 wwt (wt fraction) (wt fraction) (wt fraction) kgdwt kg21 wwt h h kgdwt h21 kgwwt the preferred food source, but will also include the Luspiked soil in their diet. We, therefore, have two uptake curves for these treatments, which need to be analysed simultaneously. Two equations similar to Eq. (7) are needed, but we need to add a parameter that divides the ingestion into food and soil (Ffood). For the marker in soil, this implies multiplication of the right-hand side of Eq. (7) with a factor 1 2 Ffood. The Tm equation is given by Eq. (7), multiplied with Ffood and lacking a selectivity factor (the worm is not expected to select for Tm-enriched fractions in its food). We observed that, although the Lu concentration immediately increases after transfer to spiked medium, the Tm concentration only starts to increase after several hours. This behaviour is probably caused by the stress of handling, particularly affecting feeding from the soil surface, where the food is located. The equation for Tm, therefore, also includes a lag time (Tlag). Eq. (7) contains many parameters, which implies that the measured marker concentrations against the time are not sufficient to identify them all. However, we have prior information on several of the parameters: Cs and Cf were measured, as well as Fsolids. Knowledge about Fing and Fgut is available from the determination of Fege and Fcom T. Jager et al. / Soil Biology & Biochemistry 35 (2003) 313–322 (Section 2.3). These measurements are however, also uncertain, and we want to include this uncertainty in the estimation of the other parameters. Furthermore, when food is present, several parameters are common to the models for Lu and Tm: Ffood is part of both model equations, as well as the retention time of the gut contents. Moreover, the retention time in the experiments with added manure is likely to be the same as on OECD medium only, as retention time seems to be independent of food type (Hartenstein et al., 1981). A Bayesian statistical framework provides a straightforward approach for the analysis of these data. In a Bayesian framework, parameters are characterised by a probability distribution (representing the ‘degree of believe’ about the parameter’s value). Before analysing new data, the a priori knowledge about a parameter is summarised in a so-called prior distribution. This prior distribution is multiplied by the parameter information contained in the data (the likelihood), to obtain the posterior distribution from which statistical inference can be made. The measurements from the previous experiments (Section 2.2 and 2.3) are considered prior information for the analysis of the marker data; the parameters that are truly unknown receive uniform (noninformative) prior distributions (Table 2). For a lucid introduction to Bayesian methods in ecology, see Ellison (1996). When the data are normally distributed, the likelihood of the parameters is related to the sum-of-squares (SSQ), resulting from the fit of the model to the data (Box and Tiao, 1992). Here, we expect multiplicative errors and, therefore, chose to log-transform the data before Table 2 List of parameters included in the Bayesian analysis and their prior distributions (symbols explained in Table 1) Symbols Prior information General parameters Cs Measured: 13.8 (s.e. 1.06, n ¼ 4) mg kg21 dwt Cf Measured: 11.6 (s.e. 0.272, 21 n ¼ 3) mg kgdwt Fsolids Measured: 0.14 (s.e. 0.0013, n ¼ 9) OECD-soil only See Table 3 (temperature dependent) Fcom See Table 3 (temperature dependent) Fege Fsel Non-informative Tret Non-informative Manure added to surface See Table 3 (temperature dependent) Fcom See Table 3 (temperature dependent) Fege Fsel Non-informative Tret Posterior from OECD (temperature dependent) Ffood Non-informative Tlag Non-informative Distribution Student-t Student-t None (constant) None (constant) Student-t Uniform Uniform None (constant) Student-t Uniform Custom Uniform Uniform 317 calculating the SSQ (thus assuming a log-normal distribution of the data). The resulting posterior probability of the total set of m parameters (u1…um), given the observed n data points, is calculated by pðu1 …um ldataÞ / ½SSQðu1 …um ; dataÞ2n=2 m Y pðui Þ ð8Þ i¼1 where p(ui) stands for the prior probability distribution of the parameter ui. Multiplication of the likelihood (a function of the parameters and the data) with the prior knowledge thus gives us the joint posterior distribution of the parameters. We can now search for the parameter values that maximise the posterior probability (simplex search). The model of Eq. (7) is highly non-linear because of the thresholds times (Tret and Tlag), hampering the estimation of confidence intervals (Klepper and Bedaux, 1997). Therefore, Eq. (8) was evaluated numerically on a regular parameter grid. From this joint distribution, probability distributions of the individual parameters were determined, which were summarised with a 90% posterior probability interval. For the studies with food, the posterior distributions of Tret from the experiments without food at 10 and 20 8C were used as prior information. The entire analysis was programmed in MatLab 6.1; a list of all prior distributions is given in Table 2. 3. Results 3.1. Selection, digestion and compaction The dissection and subsequent determination of carbon in crop contents and posterior gut provides estimates for selection (Fsel, ratio of Fom in crop and soil), digestion (Fdig, Eq. (4)), and compaction (Fcom, Eq. (2)), summarised in Table 3. Most samples showed a clear difference between the amount of carbon in crops and posterior gut. The data at 20 8C are quite consistent: when no food is present, the digestion of OM is approximately 35% but when the worms are fed manure, digestion was higher (46%). This difference was, however, not significant because of the large variation in the data. At 10 8C the data are less consistent, firstly because there are less data points (several samples were lost in the analysis), and secondly, several of the data lead to unrealistic estimates and appear to be outliers. The current study shows compaction Eq. (2) to be appreciable, and diet specific: 9% for OECD medium and 18% when fed manure (at 20 8C). On the basis of the carbon content in the crop, the worms appear to select a fraction of the soil that is more than two times as organic as the bulk soil. When manure is present on the soil surface, the Fom in the ingesta results from the soil as well as the manure, and, therefore, no estimation of selectivity is possible. 318 T. Jager et al. / Soil Biology & Biochemistry 35 (2003) 313–322 Table 3 Summary of the parameters (symbols explained in Table 1) and the resulting values for the different treatments (n.a. ¼ not applicable). Each value is derived from n pooled samples of three worms Symbols Fsel Fdig Fcom Fege Frem r2adja Fsel Tret Ffood Tlag a OECD 10 8C OECD 20 8C Manure 10 8C Selection, digestion and compaction (mean with s.e. or individual data) n¼3 n¼6 n¼3 2.3, 3.4, 1.7 2.1 (0.053) n.a. 0.35, 0.51, 0.048 0.35 (0.033) 20.18, 0.24, 0.30 1.08, 1.20, 1.01 1.09 (0.011) 0.96, 1.09, 1.08 Gut load (mean with s.e.) n¼4 n¼4 n¼4 0.096 (0.0043) 0.12 (0.012) 0.082 (0.0042) 0.074 (0.012) 0.056 (0.021) 0.21 (0.031) Inert markers (best estimate with 90% probability region) n ¼ 11 n ¼ 11 n ¼ 20 0.99 0.99 0.88/99 1.6 (1.3–2.2) 1.7 (1.4–2.3) 0.94 (0.57– 1.4) 5.6 (4.5–7.5) 2.9 (2.5–3.3) 5.4 (4.3–6.5) n.a. n.a. 0.35 (0.28– 0.42) n.a. n.a. 2.3 (2.0–2.5) Manure 20 8C n¼5 n.a. 0.46 (0.067) 1.18 (0.038) n¼4 0.098 (0.010) 0.13 (0.025) n ¼ 22 0.94/0.93 1.0 (0.72–1.6) 2.9 (2.4–3.1) 0.22 (0.16–0.31) 1.0 (0.96–1.5) Adjusted r 2 of the model fit after log-transformation, for manure treatment separate r 2 values for Lu and Tm, respectively. 3.2. Gut load On OECD medium, the growth over the acclimatisation period was 1 and 4% (20 and 10 8C, respectively), and on manure 35 and 16% (20 and 10 8C, respectively). The egested dry weight of faeces was approximately 10% of the fresh body weight of the worm (Fege, Table 3). There is no significant difference between the treatments, although there is a tendency for a slight decrease with temperature and with the addition of manure. The fraction of the gut contents remaining in the worm after 24 h starvation (Frem) appears to be related to the treatment. As can be seen from Table 3, the remaining fraction is larger when the worms are fed manure, and larger at lower test temperatures (the only Fig. 1. Concentration of lanthanides in the earthworms (including gut contents) against time. Filled symbols are Lu from soil, open symbols Tm from manure, the line is the highest probability fit. Triangles are the concentrations in worms after 48 h starvation. T. Jager et al. / Soil Biology & Biochemistry 35 (2003) 313–322 significant differences are between Frem of manure at 10 8C versus both OECD-only treatments). 3.3. Inert markers Concentrations of the lanthanides in the worms (including gut contents) are shown in Fig. 1, together with the fits of Eq. (7). The model fits are acceptable (adjusted r 2 . 0.88 after log-transformation), although the variation in the data from worms that were fed manure is clearly higher than in soil only. The lanthanides appear to function well as inert markers, judging from the fact that the concentration in the worms does not increase further after the retention time. After 48 h exposure and 48 h starvation on filter paper, the concentration of Lu in the worm samples is 4– 11% of the concentration in worms with gut contents, and for Tm this was even lower (1 – 3%). The model parameters are quite accurately identified in the Bayesian fitting procedure; the results are shown in Table 3 as the highest-probability estimate with 90% probability intervals. The earthworms are clearly selecting an Lu-enriched diet from OECD medium, independent of the test temperature (Fsel is 1.6 and 1.7, somewhat less than for OM). A clear lag time can be observed before the worms start feeding on the surface, which is larger at 10 8C (2.3 vs. 1.2 h). The selectivity factor (Fsel) from soil is lower when manure is present than in the OECD-only experiment. The worms are thus selecting less Lu-enriched soil fractions when a preferred food source is available. Despite the preference for manure, this matrix makes up less than half of the total diet: 0.22 at 20 8C, and 0.34 at 10 8C (probability intervals overlap). The retention time of the gut contents (Tret) is approximately two times higher at 10 than at 20 8C (approximately 5.5 vs. 2.9 h). The posterior distribution of retention time for the OECD-only experiments was used as prior information for the experiments with manure (Table 2). This was helpful, as feeding activity was more variable when a food source was present in the system (the data from the manure-fed worms did not provide clear information on the retention time). As a result, the resulting posterior distribution for retention time from the manure data was very similar to those of the OECD-only experiments (Table 3). 4. Discussion 4.1. Selection, digestion and compaction The dissection of E. andrei turned out to be rather difficult because of the small size of the tested sub-adults. Furthermore, not all animals had material in the last part of their gut, and some of the crop contents could not be properly removed. An additional difficulty was the loss of samples at 10 8C and the apparent outliers. For this reason, 319 the discussion will focus on the results at 20 8C. In general, the variation in the data is large, precluding conclusions on differences between the treatments. This variation partly reflects inter-individual variation in feeding pattern and physiology, but is also caused by the form of Eq. (3). An increase in digestion does not lead to a proportional decrease of Fom in egesta because digestion also increases compaction (Eq. (2)), especially when Fom of the ingesta is high. Small measurement errors can therefore, be magnified in the estimates of digestion and compaction. The observed digestion efficiencies around 40% are much higher than values predicted earlier for E. andrei (Jager, 2003). They are also much higher than values reported for the geophageous Aporrectodea rosea (2%; Bolton and Phillipson, 1976), and are more in line with values given for A. longa and L. terrestris (30 – 40%, Morgan and Morgan, 1992). These species are larger and have a much higher gut retention time (20 h for L. terrestris, Parle, 1963). Compaction is often ignored in earthworms feeding on soil (Piearce, 1972), but was demonstrated to be up to a factor of 4 when worms were kept on a litter-only diet (Dickschen and Topp, 1987). In the present study, compaction was less extreme, though still considerable (9% in OECD medium, 18% with manure). One should note that the estimates of digestion efficiency and compaction are not independent, as discussed above. The measured carbon content in the crop suggests selectivity for OM of a factor of 2.1 on OECD medium. However, it is not clear whether the carbon measured in the crop and in the gut is derived from the diet only; it may include secretions from the organism itself (from the calciferous glands, mucus and digestive enzymes), or contamination of the samples with coelomic fluid. The exact origin of the carbon in the gut requires further elaboration, but for now we will assume that the main contribution is from the diet. 4.2. Gut load Weighing the faecal production proved to be very simple to perform on a routine basis although it is likely to underestimate the faecal output, as it is difficult to collect all of the faeces from the filter paper. Here, the faeces produced over 48 h are taken as the total output, but even then, the worm may not be completely clean. In most bioaccumulation studies, 24 h is taken to evacuate the gut contents. However, the data in Table 3 show that at least 6 –21% of the gut contents remains after this period, and worse, that this fraction may depend on the treatment. It is for this reason that several authors propose methods with a longer duration (Denneman, 1994; Pokarzhevskii et al., 2000), which decreases the bias from remaining gut contents but may also lead to substantial elimination of the chemical from the tissues. Stafford and McGrath (1986) proposed to correct for remaining soil in the worm by measuring the acid-insoluble residue in the worm and comparing it to 320 T. Jager et al. / Soil Biology & Biochemistry 35 (2003) 313–322 the soil. However, the validity of this approach is questionable, given the fact that the earthworm is selecting a particular fraction from the total soil (which may differ in insoluble residues and in chemical content from the bulk soil). The relative egested weight (Fege) is quite constant around 10% (dwt/wwt worm, Table 3). This fraction seems to be lower when food is present, but this difference may also be caused by an increased compaction in the gut (at least at 20 8C, Table 3). Hartenstein et al. (1981) found a much higher effect of temperature and addition of manure on E. fetida which, given the close relation of this species with E. andrei, is remarkable. The reasons remain unclear, but may be related to differences between the studies in the quality of the soil and manure. The addition of manure allows the worms to grow considerably, but OECD soil still contains sufficient nutrition to sustain them throughout the experiment. The total amount of OM digested when fed manure can be calculated from the parameters and is 1.4 times larger than on OECD soil only (at 20 8C). However, the quality of the OM may also have led to the differences in growth. 4.3. Inert markers The model fits in Fig. 1 clearly indicate the usefulness of the lanthanides Lu and Tm as markers of the feeding activity, and support the assumption of plug flow conditions in the gut. Nevertheless, some Lu and, to a lesser degree, Tm is remaining in the worm after 48 h of starvation. It is possible that this represents soil particles still left in the gut, or sorption to the outer skin of the worm, but some assimilation cannot be ruled out. Compared to simpler methods, the use of markers like Lu and Tm has the advantage that these compounds are unlikely to irritate the worm (and thus influence retention), provide a dynamic picture of feeding (and are thereby able to test the plug-flow assumption, and show lag times), and allow quantification of feeding from multiple sources (here from soil and from manure). The Bayesian framework allows accurate parameter estimations, but the computations cannot be done by standard statistical packages, and are quite time consuming. However, this framework lends itself readily to the use of existing (uncertain) information in the analysis of new data (Ellison, 1996). The selection of Lu from soil is less than estimated on the basis of the carbon measured in the crop contents (Fsel, Table 3). Even though the variation in both values is large enough to ignore this difference, it is conceivable that Lu is also partly sorbed to clay minerals and is, therefore, not entirely representative for OM selection (although the affinity for kaolinite clay is not very high at this pH; Coppin et al., 2002). However, if not all of the carbon in the crop is derived from the soil (Section 4.1), these two figures may be more consistent. Using a marker that is more specifically bound to OM may resolve this matter (Kukkonen and Landrum, 1995). The retention times (Tret) observed, are quite comparable to the estimates made by Hartenstein et al. (1981), also showing a clear effect of temperature. As the gut load hardly depends on temperature, the feeding rate (Eq. (6)) of the worms will be nearly halved at 10 8C, compared to 20 8C. The manure data do not seriously affect the estimate of the retention time, although there is a slight shift to lower values. Care must be taken in applying the estimates for retention time directly to other studies, as it seems likely that retention is directly related to the length of the gut. A large difference between adults and juveniles was indeed found in a polychaete (Ahrens et al., 2001), in contrast, Hartenstein et al. (1981) showed no influence of size in E. fetida. The data collected by Bolton and Phillipson (1976) show slightly higher retention times for adult A. rosea, compared to juveniles, and suggest that temperature influences gut retention for juvenile worms more than for adults. The data for the manure-fed worms are clearly more variable than for the worms kept in OECD medium only (Fig. 1). It is likely that the worms do not feed continuously on a mix of soil and manure but alternate between feeding on soil and manure (as indicated by the co-varying data of Lu and Tm in Fig. 1). The worms also use the soil in a different way when manure is present as the selection factor is reduced to around unity. Possibly, the worms now feed on soil indiscriminately as a better source of nutrition is available, and soil only serves to add mineral particles to the ingesta (Schulmann and Tiunov, 1999). Even though the worms were acclimatised to a situation with manure on the surface, it still took them several hours to start feeding after transfer to the spiked media (but ingestion of soil started immediately). This lag time (Tlag) was nearly twice as long at 10 8C. Despite the high variation, the fraction of manure in the diet (Ffood) is quite accurately fixed at 0.22 at 20 8C and 0.34 at 10 8C. These figures are, however, not consistent with the measured Fom in the crop contents (data not shown). If we can assume that no selection of OM from soil occurs, the Fom data in the crop indicate a fraction of manure in the diet of approximately 0.5. This discrepancy could point at the presence of carbon in the crop, derived from various secretions by the worm. On the other hand, the carbon content in the crop and in the faeces were determined after 1 week exposure to manure, whereas the Tm-spiked manure was only a few days old and included a small amount of chloride (the counter-ion in the metal salt). Possibly, the chloride and the age of the manure play a role in its quality for the worm and may thus influence the diet composition. Hendriksen (1991) observed variation in gut load and retention time with dung age for L. festivus. These observations stress the sensitivity of the feeding process to all kinds of factors. Care should thus be taken to perform T. Jager et al. / Soil Biology & Biochemistry 35 (2003) 313–322 the estimation of digestion and gut load, and the marker experiments under the exact same conditions. 5. Conclusions In this study, the feeding activity of the compost worm E. andrei is examined, and methods are provided to estimate the physiological parameters. Gut load was derived from weighing the faecal output; selection, digestion and compaction from measuring carbon in the gastro-intestinal tract. Lanthanides (Lu and Tm) are successfully used as inert markers and provide estimates of the gut retention time and the fraction of manure in the diet. The methods and models applied in this study may also be used for experiments with more field-relevant worm-soil combinations in ecological studies. This study clearly shows that the compost worm does not feed on soil indiscriminately but is able to select an OM-enriched diet from apparently homogeneous OECD medium. Even when manure is present, a large part of the diet still consists of soil particles. The gut load is not significantly affected by the presence of manure or temperature, but the gut retention time nearly doubles by a 108 temperature decrease. The present data may be used to aid the interpretation of routine studies with E. andrei in OECD medium (e.g. how chemical exposure via ingestion is affected by temperature and providing manure). Especially, these data can be used to parameterise bioaccumulation models that include the feeding process. However, care must be taken in using these data, as feeding activity may be influenced by subtle differences in experimental set-up (e.g. the age of the manure used as feed). Acknowledgements The authors would like to thank the department of inorganic analytical chemistry at the RIVM (especially Rob Ritsema and Carlo Strien) for the measurements of the lanthanides and carbon. Furthermore, we are grateful to Lennart Weltje, Willie Peijnenburg, Kees van Leeuwen, Joop Hermens, and two anonymous reviewers for valuable comments on drafts of this paper. References Ahrens, M.J., Hertz, J., Lamoureux, E.M., Lopez, G.R., McElroy, A.E., Brownawell, B.J., 2001. The effect of body size on digestive chemistry and absorption efficiencies of food and sediment-bound organic contaminants in Nereis succinea (Polychaeta). Journal of Experimental Marine Biology and Ecology 263, 185 –209. Austreng, E., Storebakken, T., Thomassen, M.S., Refstie, S., Thomassen, Y., 2000. Evaluation of selected trivalent metal oxides as inert markers 321 used to estimate apparent digestibility in salmonids. Aquaculture 188, 65– 78. Barley, K.P., 1958. The influence of earthworms on soil fertility II. Consumption of soil and OM by the earthworm Allolobophora caliginosa (Savigny). Australian Journal of Agricultural Research 10, 179–185. Bolton, P.J., Phillipson, J., 1976. Burrowing, feeding, egestion and energy budgets of Allolobophora rosea (Savigny) (Lumbricidae). Oecologia 23, 225–245. Box, G.E.P., Tiao, G.C., 1992. Bayesian Inference in Statistical Analysis, Wiley, New York. Coppin, F., Berger, G., Bauer, A., Castet, S., Loubet, M., 2002. Sorption of lanthanides on smectite and kaolinite. Chemical Geology 182, 57 –68. Curry, J.P., Bolger, T., 1984. Growth, reproduction and litter and soil consumption by Lumbricus terrestris L. in reclaimed peat. Soil Biology & Biochemistry 16, 253 –257. Denneman, W.D., 1994. Extended water starvation; an improved method for sample preparation of lumbricidae in ecotoxicological studies. Fresenius Journal of Analytical Chemistry 348, 684 –687. Dickschen, F., Topp, W., 1987. Feeding activities and assimilation efficiencies of Lumbricus rubellus (Lumbricidae) on a plant-only diet. Pedobiologia 30, 31–37. Ellis, W.C., 1968. Dysprosium as an indigestible marker and its determination by radioactivation analysis. Journal of Agricultural and Food Chemistry 16, 220– 224. Ellison, A.M., 1996. An introduction to Bayesian inference for ecological research and environmental decision-making. Ecological Applications 6, 1036–1046. Hartenstein, F., Hartenstein, E., Hartenstein, R., 1981. Gut load and transit time in the earthworm Eisenia foetida. Pedobiologia 22, 5–20. Helmke, P.A., Robarge, W.P., Korotev, R.L., Schomberg, P.J., 1979. Effects of soil-applied sewage sludge on concentrations of elements in earthworms. Journal of Environmental Quality 8, 322– 327. Hendriksen, N.B., 1991. Gut load and food-retention time in the earthworms Lumbricus festivus and L. castaneus: a field study. Biology and Fertility of Soils 11, 170–173. Jager, T., 2003. Modelling ingestion as an exposure route for organic chemicals in earthworms (Oligochaeta). Ecotoxicology in press. Janssen, R.P.T., Peijnenburg, W.J.G.M., Posthuma, L., Van den Hoop, M.A.G.T., 1997a. Equilibrium partitioning of heavy metals in Dutch field soils. I. Relationships between metal partition coefficients and soil characteristics. Environmental Toxicology and Chemistry 16, 2470–2478. Janssen, R.P.T., Posthuma, L., Baerselman, R., Den Hollander, H.A., Van Veen, R.P.M., Peijnenburg, W.J.G.M., 1997b. Equilibrium partitioning of heavy metals in Dutch field soils. II. Prediction of metal accumulation in earthworms. Environmental Toxicology and Chemistry 16, 2479–2488. Jones, D.L., 1997. Trivalent metal (Cr,Y, Rh, La, Pr, Gd) sorption in two acid soils and its consequences for bioremediation. European Journal of Soil Science 48, 697 –702. Klepper, O., Bedaux, J.J.M., 1997. Non-linear parameter estimation for toxicological threshold models. Ecological Modelling 102, 315–324. Kukkonen, J., Landrum, P.F., 1995. Measuring assimilation efficiencies for sediment-bound PAH and PCB congeners by benthic organisms. Aquatic Toxicology 32, 75–92. Morgan, J.E., Morgan, A.J., 1992. Heavy metal concentrations in the tissues, ingesta and faeces of ecophysiologically different earthworm species. Soil Biology & Biochemistry 24, 1691–1697. OECD, 1984. Guideline for Testing of Chemicals no. 207. Earthworm, Acute Toxicity Tests, Organization for Economic Cooperation and Development, Paris, France. Parle, J.N., 1963. Micro-organisms in the intestines of earthworms. Journal of General Microbiology 31, 1–11. Piearce, T.G., 1972. The calcium relations of selected Lumbricidae. Journal of Animal Ecology 41, 167–188. Piearce, T.G., 1978. Gut contents of some lumbricid earthworms. Pedobiologica 18, 153 –157. 322 T. Jager et al. / Soil Biology & Biochemistry 35 (2003) 313–322 Pokarzhevskii, A.D., Van Straalen, N.M., Semenov, A.M., 2000. Agar as a medium for removing soil from earthworm guts. Soil Biology & Biochemistry 32, 1315–1317. Schulmann, O.P., Tiunov, A.V., 1999. Leaf litter fragmentation by the earthworm Lumbricus terrestris L. Pedobiologia 43, 453–458. Stafford, E.A., McGrath, S.P., 1986. The use of acid insoluble residue to correct for the presence of soil-derived metals in the gut of earthworms used as bio-indicator organisms. Environmental Pollution 42, 233–246. Weltje, L., Heidenreich, H., Zhu, W., Wolterbeek, H.T., Korhammer, S., De Goeij, J.J.M., Markert, B., 2002. Lanthanide concentrations in freshwater plants and molluscs, related to those in surface water, pore water and sediment. A case study in The Netherlands. The Science of the Total Environment 286, 191– 214. Whalen, J.K., Paustian, K.H., Parmelee, R.W., 1999. Simulation of growth and flux of carbon and nitrogen through earthworms. Pedobiologia 43, 537 –546.
© Copyright 2026 Paperzz