Archived at

Estimating the incidence of entomopathogenic nematodes in soil by the use of bait insects

Otto Nielsen1 (e-mail: on[a]

Ib Michael Skovgaard2 (e-mail: ims[a]

Holger Philipsen1(e-mail: hp[a] (corresponding author)

1Zoology Section, Department of Ecology. 2Department of Chemistry and Biometry.

The Royal Veterinary and Agricultural University

Thorvaldsensvej 40, 3, DK-1871 Frederiksberg C, Denmark.





Soil samples were stored and baited in different ways in order to develop a bait method that could estimate the level of steinernematid nematodes in soil in a time efficient way. The method was based on incidence (proportion of samples with a given species of Steinernema). Tenebrio molitor (Coleoptera) larvae were used as bait and it was concluded that: 1) Repeated baiting in a later period improved the bait result; 2) Extension of the bait periods from one to two weeks was not necessary; 3) The precision of the estimated incidence did not improve noteworthy by taking more than 60-80 samples; 4) The minimum time required to analyse one soil sample was 6 minutes.

Key words

Baiting, sampling, entomopathogenic nematodes, Steinernema affine, S. feltiae, Tenebrio molitor


Nematodes belonging to the genera Steinernema (Filipjev) and Heterorhabditis (Poinar) are known as entomopathogenic nematodes. These nematodes have received much attention due to their potential use as biological control agents of insects (Gaugler, 2002; Ehlers, 2001). The nematodes are found in soil as infective juveniles and are dependent on the presence of an insect host in order to complete their life cycle. The infective juveniles enter the host through natural openings (Heterorhabditis and Steinernema) and through the cuticle (Heterorhabditis). The nematodes are remarkable as they live in a mutualistic relationship with insect pathogenic bacteria (Burnell & Stock, 2000; Boemare et al., 1997). The bacteria are released when the nematodes reach the insect hemocoel and the host is killed within few days by the combined action of toxins produced by nematodes and bacteria.

In order to describe nematode populations in soil, methods are required for the estimation of their population size. Traditionally, in agricultural nematology, this is based on numerous methods of extraction from soil (flotation, decanting and sieving, elutriation, centrifugal flotation) or methods where nematodes actively leave the soil (Baermann funnel and its modifications) (Verschoor & de Goede, 2000; Seinhorst, 1988; Hooper, 1986). Some of these methods have also been applied for the study of entomopathogenic nematodes (Sturhan, 1995; Curran & Heng 1992; Saunders & All, 1982) but mostly, entomopathogenic nematodes are detected by placing nematode-susceptible insect larvae in soil samples for a period of time. In this way, nematodes present in the soil can infect and kill the larvae and subsequently, nematodes can be isolated from the insect cadaver (e.g. Fan and Hominick, 1991). The method was described for the first time by Bedding & Akhurst (1975) and is referred to as the live-bait method (baiting) or the Galleria trap method because Galleria mellonella (Lepidoptera: Pyralidae) larvae are often used as bait. The collection of nematodes from dead bait larvae is achieved by placing the larvae on moist filter paper surrounded by water (water trap / White trap). Infective juveniles that leave the cadaver will then be trapped in the water. Alternatively, the dead larvae are dissolved in pepsin (Mauléon et al., 1993) or dissected in order to detect the nematodes (Bednarek, 1998; Köppenhöfer et al., 1998; Curran & Heng, 1992).

The bait method is widely used because it can be performed without special equipment and with limited nematological experience. Further, it includes the ability of the nematodes to infect and recycle in an insect host (nematode quality). The disadvantage of the method is, however, that it depends on the infectivity of the nematodes, which may vary over time (phased infectivity) (Fairbairn et al., 2000; Bohan & Hominick, 1997a; 1997b; 1995; 1996) and that it excludes nematodes that are unable to recycle in the bait insect. Further, susceptibility of the bait insect to different nematode species may vary and the bait method may not directly reflect the actual presence of the observed species.

Generally, two types of estimates can be obtained on the basis of the above described methods. These are either the number of nematodes per soil unit (density) or the proportion of samples with nematodes (incidence). Both estimates can be analysed statistically. The reproducibility of the chosen method is primarily a matter of analysing a representative part of the soil from a given area. In order to enable the analysis of larger and thus more representative parts of a given area, soil samples are often pooled and followed by the baiting of smaller fractions (sub-samples) of the proper mixed bulk sample.

The amount of soil that can be handled in a given time is probably one of the main practical limitations when nematode populations have to be estimated. The problem can be reduced by storage of the soil at low temperature until a period where soil analysis is more convenient. However, storage of soil samples may affect entomopathogenic nematode survival and infectivity - and thus the bait result - and it has been shown that prior storage in water can increase the infectivity of laboratory produced nematodes (Griffin, 1996).

In the present study, the aim was to test a bait method based on incidence. The basic idea was to take many samples and analyse them separately in a relatively easy way. The obtained estimate of nematode incidence should reflect the likelihood of an insect getting infected in a given area (agricultural fields in the present study). The method should be able to handle the problems related to phased infectivity and thus each sample was baited for a total of four weeks. In addition, the effects of a range of other parameters were studied. These included the influence of crop, storage and soil moisture.

The experiments required large numbers of bait larvae and Tenebrio molitor (Coleoptera) was chosen as this insect is relatively easy to culture. It has previously been demonstrated that T. molitor is susceptible to the dominating entomopathogenic nematodes in agricultural soil (Nielsen & Philipsen, 2004a; Husberg et al., 1988). The results of the study are used to recommend sampling design and bait procedure.

Materials and methods

The study was based on soil samples that were taken at the sites Snubbekorsgård (KVL) and Årslev (DJF) in 2001 and 2002. The two sites represent a sandy and a clay soil type, respectively. The total number of samples that were used in the analysis were 5,300 and included 2,529 isolates of Steinernema feltiae Filipjev and 126 isolates of S. affine Bovien (Table 1).

Snubbekorsgård is an experimental farm and part of the Royal Veterinary and Agricultural University, Denmark. The soil samples were taken within an area of 70 x 100 metres in a larger field. The sampled area was divided into plots for the purpose of other studies and was grown with a range of different crops (barley, chicory, clover, grass, oil seed rape and wheat). The soil was a sandy loam with the following characteristics in the upper 25 cm: 10 % clay, 5 % silt and 85 % sand.

Årslev is another experimental farm and part of the Danish Institute of Agricultural Sciences, Department of Fruit, Vegetables and Food Sciences. The soil here was an Agrudalf soil which in the upper 25 cm layer consisted of approximately 15 % clay, 27 % silt and 55 % sand (Thorup-Kristensen, 2001). The soil samples were taken in six 1-hectare fields within an organic cropping system. The fields were grown with a range of different crops (alfalfa, barley, cabbage, carrot, clover, leek, onion, and pea) in a fixed rotation scheme (see Thorup-Kristensen (1999) for details). At DJF, the entomopathogenic nematode S. feltiae had been inoculated to parts of the fields in May 2001 as part of another study. The nematodes studied were thus a combination of naturally occurring and released nematodes.

An estimate of nematode incidence was generally obtained by collecting 5 replicates of 25 soil samples (soil core of 4.5 cm diameter x 15 cm depth) within a given crop or field. The 25 soil samples were taken at random in small plots. At DJF all plots were circular with a diameter of 3 m (appr.7 m2). Plot centres within the same field were distanced by 10 m or more. The circular plots were also used at KVL. In addition, 20 rectangular plots of 3.5 x 36 m (126 m2) were studied here. These were placed side by side in a randomised block design that represented four different treatments (crops etc.) of five replicates each.

The soil samples were sampled by pressing a cylindrical iron pipe (sharpened in the lower end) into the soil. Removal of the soil from the soil sampler was achieved next time the pipe was pressed into the soil. In order to avoid the transfer of nematodes from one sample to the other, an amount of soil (approximately 4-6 cm core from the edge of the following sample position) was discarded. All soil samples were stored individually in small plastic bags (poly-ethylene, 4 l, Sækko, Denmark) and kept at 5° C until further analysis.

The water content (w/w) of the soil was estimated for each field and sampling date. This was based on pooled soil from 6-10 samples. The soil was dried at 80° C until its weight remained constant (after 2-4 days).

The reproducibility of the bait method was initially tested by sampling the same areas twice. At KVL, 125 soil samples (5 x 25) were taken in five 126 m2 plots (within a total area of 2.800 m2) April, 3 and April, 20 and at DJF, 50 (2 x 25) soil samples were taken in two 7 m2 plots April, 27 and May, 17. The biological activity in the soil was assumed to be limited in this period due to low soil temperatures. The soil was baited as described below (storage and handling was similar for both sampling dates but spaced in time).

In order to prepare the soil for baiting, soil from each soil sample was mixed in the bag in which the soil sample was stored by turning and moving the bag and crushing larger soil parts by hand. Hereafter, a plastic jar (8.5/7.2 cm upper/lower diameter x 9 cm height (Grathwool, Denmark)) was filled approximately half full with part of the soil (approximately 200 ml). The remaining soil in the bag (50-60 % of the soil sample) was again stored at 5° C until repeated baiting in a later period (3-7 weeks after the first bait period). The soil in each jar was further prepared for baiting by removing larger root parts and stones and by gently splitting larger soil clumps with a blond knife. The volume of soil was adjusted by removing excess soil to obtain a half filled jar and finally the jar was closed with a lid. Baiting was initiated the same day or stored at 5° C for a maximum period of one week.

Baiting was initiated by adding one T. molitor larva obtained from a local producer to each jar. Older larvae were avoided to reduce the likelihood of pupation during the bait period. The bait larvae were randomised prior to use. The jars were placed in the dark at 18-24° C. After one week (6-8 days) all larvae in the jars were checked. Dead larvae were collected and replaced by a new larva and live larvae were left in the jars. After another week of incubation (13-15 days in total), all jars were checked again and dead larvae were collected.

Dead bait larvae were placed individually on water traps to collect nematodes emerging from the cadavers. The water traps were constructed by placing a plastic lid with a piece of filter paper on top in a 5-cm-Petri dish containing approximately 2 ml of tap water. The Petri dishes were observed for up to four weeks and additional tap water was supplied if necessary. If entomopathogenic nematodes appeared in the water, the water traps were stored at 5° C for later identification of the nematodes. The nematodes stayed alive in the water traps for several months when stored under these conditions.

Identification of nematodes was based on morphology of heat fixed specimens of infective juveniles from the water traps. Two species (S. affine and S. feltiae) were present at the two sites and these could easily be distinguished. The general characteristics that were used for identification was body size and curvature and shape of the tail (spine/reflection is normally visible in the tail of S. affine). Generally, large quantities of nematodes were available so that identification could be based on several specimens per isolate. In order to check species identity, 14 isolates were cultured and examined by PCR-RFLP according to the procedure described by Reid et al. (1997). The restriction enzymes Alu I, Dde I, Hha I and Hinf I were used.

The above described method could result in up to four nematode isolates from one soil sample. The abbreviations shown in Table 1 (A, B, C, D) will be used to designate when the isolates were obtained. A refers to the first week in the first bait period, B refers to the second week in the first bait period, C refers to the first week in the second bait period and D refers to the second week in the second bait period. Depending on whether an nematode isolate was obtained or not A, B, C and D obtain the value 0 (no isolate) or 1 (isolate) in the further analysis. Further, AC refers to the combined result from the first week in each period, AB refers to the combined result from the first period and ABCD refers to the combined result of all four bait larvae. Also AC, AB and ABCD can either be 0 (no isolate) or 1 (one or more isolates).
The first step in the analysis was to test whether work could be saved by reducing the full bait version (ABCD) to AC, AB or A. This was considered valid if AC, AB or A, respectively, was approximately proportional to ABCD and independent of experimental factors. The experimental factors included sampling period (autumn 2001, spring 2002, and autumn 2002), site (DJF, KVL), crop (13 different), soil moisture (8.3-18.2%) and storage (2-20 weeks (see below)). The tests were made by a logistic linear model using the GENMOD procedure (link=logit, (binary distribution)) provided by SAS (Statistical Analysis System) for each reduced version of the bait procedure. However, only the situations where ABCD=1 could be used, as proportions cannot be calculated when ABCD=0. Further A, AB and AC would always be 0 if ABCD is 0.

Secondly, the relationship between sample numbers and the precision of the estimated incidence was studied. For this purpose, the variance (V) of the incidence based on m plots and n samples per plot was decomposed as

V = Vplot/m + Vsamp/mn = Vplot/m + p(1-p)/mn,

Where Vplot denotes the variance of incidence between plots, Vsamp denotes the (binomial) sampling variance and p denotes the mean incidence. For any given mean incidence we can compute the relationship between m, n and V. Like the sampling variance, the plot variance is expected to depend on p, and we estimated Vplot by regressing the sample plot variance on p(1-p) to obtain a so-called overdispersion factor, F= Vplot/(p(1-p)), under various circumstances. A graph of the standard error, Ö V, versus n, using the expression V= p(1-p)(F+ 1/n)/m, thus displays the gain in precision with increasing sample number.

Further, the effect of storage on nematode activity was analysed separately in a regression analysis (GLM procedure in SAS). Storage was included as a relative measure: Storage = (Storage2-Storage1)/Storage1 where Storage1 and Storage2 is the period from soil sampling to the first and second bait period, respectively. Storage was then related to the differences in nematode activity (NA) in the two bait periods: NA=(CD-AB)/ABCD where AB, CD and ABCD are expressed as incidences (proportion of positive samples based on AB, CD and ABCD, respectively, and calculated for each set of 25 soil samples).

Finally, the time requirements for the bait procedure was judged on the basis of experience with work planning (7.000-8.000 samples have been analysed over the last years) and rough estimates are given for the time needed to sample and analyse soil with different characteristics.


The reproducibility of the bait method was tested at both sites before the method was used routinely and almost identical incidences were obtained in two independent sampling events (48 and 44 % ,respectively, at KVL and 36 and 35 % ,respectively, at DJF) . It was thus concluded that 125 soil samples would give reproducible results within larger areas (2.800 m2) and that 50 soil samples from two 7 m2 plots also would give reproducible results. A number of 50 samples would, however, hardly enable the statistical studies that were applied in the present work and for that reason, 125 samples were generally sampled.

The two sites differed in their composition of nematodes and how these were obtained by the bait technique (Table 1). At DJF S. feltiae and S. affine were isolated; S. feltiae being the dominating species. At KVL nearly all isolates were S. feltiae. At DJF more isolates were obtained in the second (C+D) than in the first bait period (A+B) whereas equal number of isolates were obtained in the two bait periods at KVL. In most cases the number of isolates were lower in the second (B or D) than in the first week (A or C) of a bait period. Exceptions were seen at DJF in the autumn of 2001 for both nematode species and for S. affine at DJF in the autumn of 2002.

Although many isolates were obtained in the second week, their contribution to the incidence was limited as the likelihood of getting additional positive samples in an extra week of baiting was generally low. This is obvious when incidences based on A and C is compared to incidences based on ABCD (Figure 1a-b-c). Most points on the graphs are close to the 1:1 line and AC-based incidences would generally have had values of 80-95 % of the ABCD-based incidences.

The consequence of excluding the second bait period would for S. feltiae at KVL have been incidences that were 60-80 % of the ABCD-based incidences (Figure 1d). At DJF the incidences for both nematodes species would have been reduced to around 50 % (Figure 1e-f). A further reduction to an A-based incidence was also tested and this would have resulted in much reduced incidences for both species at DJF (Figure 1h-i) and a reduction to 40-60 % for S. feltiae at KVL (Figure 1g).

A statistical approach to the above reductions in the bait procedure is presented in Table 2. Here it was tested whether the proportions of nematodes obtained in a reduced bait version was influenced by experimental factors. For AC it was found that increasing soil moisture significantly reduced the proportion of positive samples isolated in the first week and further, there was an interaction between sampling period and crop. Similarly, a number of experimental factors were found to influence the proportion of positive samples for AB and A (Table 2). The results of the statistical analysis have to be used with caution as several experimental factors were unbalanced and hence partly confounded. For instance, some crops were exclusively grown at either DJF or KVL and the soil was generally moister at DJF than at KVL. For that reason the statistical analysis was also applied to the two sites separately. The results showed that the proportion of positive samples varied for all reductions of the bait procedure (A, AB, AC) and that this variation was statistically significant for one or more experimental factors (data not shown).

In the GENMOD analysis, only observations where ABCD=1 could be included. In the excluded observations both A, AB, AC and ABCD is 0, which means that the calculated incidences would be more similar than the GENMOD procedure indicates. It is obvious from Figure 1 that A- and AB-based incidences in a very variable degree reflect ABCD-based incidences whereas AC-based incidences relatively well describe the results obtained by the full bait procedure. For that reason, the relationship between AC and ABCD was further analysed by linear regression (Table 3). The relationship was found to be significantly different for the two sites but for mid-sized incidences the reductions in AC-based incidences would be similar for the two sites as the two regression lines are crossing.

In the second step of the statistical analysis of the data, the relationship between number of samples and precision of the estimated incidence was studied for the AC-based method. The overdispersion factor, F, was estimated to F=1/31 at KVL and F=1/35 at DJF. This means, that with around 30-35 samples per plot, the sampling variation is reduced to the same size as the plot variation, and many more samples per plot would be rather wasteful compared to adding more plots. This is a rough average, however, ignoring some variation in the overdispersion due to variation in incidence, but it does show that sampling variation seems to be the more important variation compared to plot variation. With F=1/30 the standard error of the incidence from five plots is shown in Figure 2 as a function of number of samples per plot for incidences of 0.1, 0.3 and 0.5.

In order to specifically analyse the effect of storage of the samples on nematode activity, the relative difference in activity in the two bait periods were related to the time of storage. By regression analyses, it was found that nematode activity was independent of the length of storage between the first and the second bait period (data not shown). The time requirements for the bait procedure presented above can be divided into a number of activities (Table 4). The most important factor is of course the number of soil samples that have to be taken and analysed. But also the characteristics of the soil will influence the amount of hours that have to be used. Here we have distinguished between a "best case" situation where the soil is very easy to handle and bait larvae mortality is low and a "worst case" situation where the soil is difficult to handle because of roots and clumps and where bait larvae mortality also is high. The total number of hours to analyse 500 soil samples was calculated to 64 man hours in the best case and 136 man hours in the worst. The amount of working hours that can be saved by excluding the second bait period within each period is also shown in Table 4. In the best case, the soil can be analysed in 47 hours and in the worst case in 83 hours. The highest relative reduction is thus in the worst case situation. This is because work can be saved that normally would have been used for the handling of large numbers of dead bait larvae and identification of nematodes. Further, bait larvae mortality due to other factors (entomopathogenic fungi and bacteria, etc.) is normally highest in the second bait week.



The objective of the present study was to develop a method that could estimate nematode occurrence in a given area in a time efficient way. For that reason a bait method based on nematode incidence was used. The advantage of such a method is that each sample can be analysed relatively fast and that information about the spatial occurrence of the nematodes is obtained. The actual density of nematodes in the field is not estimated but there are at least two arguments for neglecting this parameter.

One argument is that nematode density can be difficult to estimate precisely due to the patchy distribution of the nematodes. This may, however, be overcome by mixing soil samples and subsequent analysis of sub-samples but then information about the spatial occurrence is lost and the result will depend on the accuracy of soil mixing.

Another argument is that nematode density may change within short periods as a few successful infections of insect larvae can propagate a small number of nematodes into thousands (Nielsen & Philipsen, 2004b). The generation time in small larvae - where only one generation of nematodes is produced – can further be very short. An example is second stage Delia radicum larvae where the generation time is 4-6 days (Nielsen, 2003). The density of nematodes may thus be misleading as this can change within a short time. In contrast, a method based on incidence is less sensitive to nematode propagation but incidence will still increase over time (longer terms) if the propagated nematodes are able to survive and disperse. An incidence-based method thus gives a more stable picture of the nematode occurrence in an area and will be relevant for seasonal estimates and for monitoring purposes.

Bait conditions can be varied in many ways and the main aim of the present work was to study the effect of multiple bait periods. Previous work has stated that repeated baiting may be necessary in order to avoid that isolates are overlooked (Stuart & Gaugler, 1994; Griffin et al., 1991; Hominick & Briscoe, 1990) and that this probably is due to phased infectivity (Fairbairn et al., 2000; Bohan & Hominick 1997a; 1996). The present study showed that many positive samples can be misjudged if baiting is restricted to one period of just one week and that the degree of bias is site dependent (Figure 1g-i). If different sites have to be compared it is thus not advisable to use an A-based incidence. Also an AB-based incidence was site dependent and only the AC-based incidence gave similar reductions for the two sites. Further, there was a reasonably high correlation between AC and ABCD-based incidences (Table 3) although, the proportions of positive AC-samples depended on a few experimental factors (Table 2). An important factor here was soil moisture and this can be avoided by sampling in periods with similar soil moisture. Alternatively dry samples could be moistened in the laboratory and wet samples could dry in the bait containers prior to baiting. The significant effect of interaction of sampling period and crop for AC based incidences cannot be overcome by experimental means. A closer study of the GENMOD analysis, however, reveals that the significant effects depended on the results from a single crop. Thus, in most cases AC-based incidences were proportional to ABCD-incidences.

It has not previously been reported whether bait larvae susceptibility varies over time. The data from KVL, however, indicate that similar results can be obtained with different batches of bait larvae, as A and C gave results that were very similar to B and D. On the other hand there were some divergences in the A-based incidences for KVL (Figure 1g) which also advocate for two independent bait periods (AC) to reduce the effect of an odd batch of bait larvae.

It can be speculated why the second bait period gave many more isolates than the first bait period at DJF (Table 1; Figure 1e-f). A reasonable explanation could be that the soil structure is damaged at the time of sampling as a consequence of the higher clay and water content. These two factors in combination with a low root content sometimes resulted in samples that tended to plasticity. In such samples, nematodes probably would get caught with reduced ability to infect the bait insect. In most cases, however, the samples were only moderately affected and there is a possibility that handling of the soil prior to the first bait and subsequent storage at 5° C re-established a soil structure that gave the nematodes better conditions for infection of the bait larvae in the second period.

Another conclusion that could be drawn from the data was that different lengths of storage from the first to the second bait period did not change the number of obtained isolates. This is in accordance with an earlier study where 85 large soil samples (2-4 l) were collected in September and sub-samples (200 ml) of these were baited in early October, late November and late January (Nielsen, 2001). In this study, the number of steinernematid isolates increased from the first bait (12 isolates) to the second (21 isolates) and remained the same in the third bait (20 isolates). Studies with other types of nematodes and storage similarly showed that samples could be stored for a few months without affecting the number of isolated nematodes (Seinhorst, 1988; Barker et al., 1969).

Larvae of the wax moth, G. mellonella are commonly used as bait larvae. This was not possible in the present study but since the above conclusion were a result of other factors than the bait insect (site, soil moisture etc.), it can be expected that the same conclusion would have been obtained with G. mellonella larvae. In general, G. mellonella is assumed to be more susceptible than T. molitor but a direct comparison on a reasonable number of soil samples has - to the author’s knowledge - not been conducted.

The advantage of using T. molitor larvae is that they are very easily found in the samples. They are either situated on the top or can be found quickly by turning the half filled jar. It is also an advantage to use only one larva in each jar as searching can be stopped as soon as the larva is observed and if mortality is low, 500 jars can be checked in two hours by two persons (Table 4).

The present bait technique gave incidences that ranged from 0 to 100% in a plot and two different soil types were studied (sandy and clay soil). Thus, the conclusions from our study are probably general for most combinations of soil type and nematode incidence.


We wish to thank DARCOF (Danish Agricultural Research Center of Organic Farming) for financial support. The technical assistance of Hanna Hansen and Anne Anttila is highly appreciated.







Barker , K. R., Nusbaum, C. J. & Nelson, L. A. (1969). Effects of storage, temperature and extraction procedure on recovery of plant parasitic nematodes from field soils. Journal of Nematology, 1, 240-247.

Bedding, R. A. & Akhurst, R. J. (1975). A simple technique for the detection of insect rhabditid nematodes in soil. Nematologica 21, 109-110.

Bednarek, A. (1998). The agricultural system, as a complex factor, effects the population of entomopathogenic nematodes (Rhabditida: Steinernematidae) in the soil. IOBC Bulletin 21 (4), 155-160.

Boemare, N., Givaudan, A., Brehelin, M. & Laumond, C. (1997). Symbiosis and pathogenicity of nematode-bacterium complexes - review article. Symbiosis 22, 21-45.

Bohan, D. A. & Hominick, W. M. (1995). Examination of the Steinernema feltiae (Site 76 strain) infection interaction with the Galleria mellonella host, using an infection model. Parasitology 111, 6-17.

Bohan, D. A. & Hominick, W. M. (1996). Investigations on the presence of an infectious proportion amongst populations of Steinernema feltiae (Site 76 strain) infective stage. Parasitology 112, 113-118.

Bohan, D. A. & Hominick, W. M. (1997a). Long-term dynamics of infectiousness within the infective-stage pool of the entomopathogenic nematode Steinernema feltiae (Site 76 strain) Filipjev. Parasitology 114, 301-308.

Bohan, D. A. & Hominick, W. M. (1997b). Sex and the dynamics of infection in the entomopathogenic nematode Steinernema feltiae Filipjev. Journal of Helminthology 71, 97-201.

Burnell, A. M. & Stock, S. P. (2000). Heterorhabditis, Steinernema and their bacterial symbionts -lethal pathogens of insects. Nematology 2, 31-42.

Curran, J. & Heng, J. (1992). Comparison of three methods for estimating the number of entomopathogenic nematodes present in soil samples. Journal of Nematology 24, 170-176.

Ehlers, R.-U. (2001). Mass production of entomopathogenic nematodes for plant

protection. Applied Microbiology & Biotechnology 56, 623-633.

Fan, X. & Hominick, W. M. (1991). Efficiency of the Galleria (Wax Moth) baiting

technique for recovering infective stages of entomopathogenic rhabditids

(Steinernematidae and Heterorhabditidae) from sand and soil. Revue de Nématologie

14, 381-387.

Fairbairn, J. P., Fenton, A., Norman, R. A. & Hudson, P. J. (2000). Re-assessing the infection strategies of the entomopathogenic nematode Steinernema feltiae (Rhabditidae; Steinernematidae). Parasitology 121, 211-216.

Gaugler, R. (ed.) 2002. Entomopathogenic Nematology. CABI Publisher, New

York. 388 pp.

Griffin, C. T. (1996). Effects of prior storage conditions on the infectivity of Heterorhabditis sp. (Nematoda: Heterorhabditdae). Fundamental and Applied Nematology 19, 95-102.

Griffin, C. T., Moore, J. F. & Downes, M. J. (1991). Occurrence of

insect-parasitic nematodes (Steinernematidae, Heterorhabditidae) in the Republic of

Ireland. Nematologica 37, 92-100.

Hominick, W. M. & Briscoe, B. R. (1990). Survey of 15 sites over 28 months for

entomopathogenic nematodes (Rhabditida: Steinernematidae). Parasitology

100, 289-294.

Hooper 1986. Extraction of free-living stages from soil. In: Southey, J.F. (ed.)

Laboratory methods for work with plant and soil nematodes, Ministry of Agriculture,

Fisheries and Food, HMSO, London. Pp. 5-30.

Husberg, G. B., Vanninen, I. & Hokkanen, H. (1988). Insect pathogenic fungi and nematodes in fields in Finland. Vaxtskyddsnotiser 52, 38-42.

Koppenhöfer, A.M., Campbell, J.F., Kaya, H. K. & Gaugler, R. (1998).

Estimation of entomopathogenic nematode population density in soil by correlation

between bait insect mortality and nematode penetration. Fundamental and Applied

Nematology 21, 95-102.

Mauleon, H., Briand, S., Laumond, C. & Bonifassi, E. (1993). Use of digestive enzymes to study parasitism of insect larvae by Steinernema and Heterorhabditis. Fundamental and Applied Nematology 16,185-186.

Nielsen, O. (2001). Interactions of entomopathogenic nematodes and insects from

cruciferous crops. PhD-thesis. The Royal Veterinary and Agricultural University,

Frederiksberg, Denmark.

Nielsen, O. (2003). Susceptibility of Delia radicum to steinernematid nematodes. Biocontrol 48, 431-446.

Nielsen, O. & Philipsen, H. (2004a). Occurrence of Steinernema species in

cabbage fields and the effect of inoculated S. feltiae on Delia radicum and its

parasitoids. Agricultural and Forest Entomology 6, 25-30.

Nielsen, O. & Philipsen, H. (2004b). Recycling of entomopathogenic nematodes

in Delia radicum and in other insects from cruciferous crops. BioControl 49, 285-294.

Reid, A. P., Hominick, W. M. & Briscoe, B. R. (1997). Molecular taxonomy and

phylogeny of entomopathogenic nematode species (rhabditida, Steinernematidae) by

RFLP analysis of the ITS region of the ribosomal DNA repeat unit. Systematic

Parasitology 37, 187-193.

Saunders, M. C. & All, J. N. (1982). Laboratory extraction methods and field

detection of entomophilic rhabditoid nematodes from soil. Environmental Entomology

7, 605-607.

Seinhorst, J. W. (1988). The estimation of densities of nematodes populations in

soil and plants. Växtskyddsrapporter, Jordbruk 51, (107 pp). Swedish Agricultural

University, Uppsala, Sweden.

Stuart, R. J. & Gaugler, R. (1994). Patchiness in populations of

entomopathogenic nematodes. Journal of Invertebrate Pathology 64, 39-45.

Sturhan, D. (1995). Studies on the sympatric occurrence of entomopathogenic

nematodes. Nachrichtenblatt des Deutschen Pflanzenschutzdienstes 47, 54.

Thorup-Kristensen, K. (2001). Are differences in root growth of nitrogen catch crops important for their ability to reduce soil nitrate-N content, and how can this be measured ? Plant and Soil 230, 185-195.

Thorup-Kristensen, K. (1999). An organic vegetable crop rotation system aimed at self-sufficiency in nitrogen. In: Olesen, J. E., Eltun, R., Gooding, M. J., Jensen, E. S., Köpke, U. (Eds.). Designing and testing crop rotations for organic farming. DARCOF report 1, 133-140.

Verschoor, B. C. & de Goede, R. G. M. (2000). The nematode extraction efficiency of the Oostenbrink elutriator-cottonwool filter method with special reference to nematode body size and life strategy. Nematology 2000, 325-342.

Figure 1. Incidence (percentage of positive samples) was calculated on the basis of either one week of baiting in two periods (AC) (a-c), two weeks of baiting in one period (AB) (d-f) or one week of baiting in one period (A) (g-i) and compared to two weeks of baiting in two periods (ABCD). Each point was estimated on the basis of 75-200 cores (typically 125 cores) and by reducing ABCD to AC, AB and A, respectively.

Figure 2. The relationship between number of samples and the standard error of incidence. The calculations are based on an overdispersion factor (F) of 1/30 for three different levels of incidence.