The burrowing characteristics of 3 common earthworm species were studied using X-ray computed tomography (CT) scanning in large cylinders (24.1 cm diam.) packed with topsoil (0-25 cm) and subsoil (25-50 cm) to representative field bulk density values and sown with ryegrass. Replicated cylinders (n = 3), kept under constant moisture and temperature conditions, were inoculated with mature species of Lumbricus rubellus, Aporrectodea caliginosa, or Octolasion cyaneum earthworms at rates similar to their population density in the field. A non-inoculated, unreplicated control was also included. The number, biomass, and activity of the 3 species were then examined. CT scans were taken every 5-10 cm through the soil cylinders 20, 40, and 60 days after inoculation to measure burrow parameters. Mean pore area was greatest for O. cyaneum and L. rubellus, and least for A. caliginosa. Porosity produced by both L. rubellus and A. caliginosa declined with depth. L. rubellus was most active in the top 5 cm, whereas A. caliginosa was most active in the top 10 cm. O. cyaneum created its burrows relatively uniformly throughout the top 20 cm. No species created significant porosity below 20 cm. The greatest amount of porosity and number of pores were created in the cylinders inoculated with A. caliginosa. However, porosity created per earthworm was least for A. caliginosa and L. rubellus and greatest for O. cyaneum. Porosity per biomass was least for L. rubellus and greatest for A. caliginosa and O. cyaneum. A. caliginosa created mainly temporary burrows, with 72-85% of its burrows backfilled between scans. L. rubellus burrows lasted longer (56-64% backfilled between successive scans) and hydraulic conductivity measurements suggested that L. rubellus burrows were surface-connected and more continuous than those created by A. caliginosa. It appears that, of the 3 species studied, L. rubellus has the most beneficial effects on the measured soil physical properties.

Additional keywords: Aporrectodea caliginosa, earthworm burrows, hydraulic conductivity, Lumbricus rubellus, Octolasion cyaneum, porosity.

Introduction

In agricultural soils in New Zealand, earthworm populations consist of introduced European lumbricid species and are dominated by Aporrectodea caliginosa, with smaller numbers of both Lumbricus rubellus and Octolasion cyaneum (Springett 1992; Fraser et al. 1996). The burrowing of lumbricid earthworms in general increases the volume and continuity of soil macropores, soil structural stability, and consequently the movement of water and air. However, information on the burrowing characteristics of individual species is needed so that management practices can be identified that will encourage their specific beneficial effects.

 In New Zealand studies, earthworm burrowing has been assessed through direct observation along the walls of soil containers (Springett 1983; Springett and Gray 1998), destructive sampling of large soil containers for image analysis (Francis and Fraser 1998), or in the field using minirhizotrons (Springett and Gray 1997). However, as earthworm burrowing behaviour changes on encountering container walls or observation tubes (Joschko et al. 1991), the burrows observed by Springett (1983) and Springett and Gray (1997) may not be typical of those created in the field. The large containers used by Francis and Fraser (1998) reduced edge effects on burrowing behaviour, but the destructive sampling of the containers prevented an assessment of the longevity of earthworm burrows.

X-ray CT scanning of large-diameter soil cylinders offers an alternative method for obtaining information on the burrowing characteristics of earthworms (Jegou et al. 1999). As this method is non-destructive, repeat measurements can be made and the use of large cylinders minimises edge effects. The objectives of this study were to: (i) assess the burrowing characteristics of 3 earthworm species (under artificial conditions) through measurement of 2-D porosity using X-ray CT scanning, (ii) estimate the extent of burrow backfilling between sequential scans, and (iii) estimate the continuity of earthworm burrows with depth through hydraulic conductivity measurements.

Materials and methods

Experimental design and management

In July 1997, samples of topsoil (0-25 cm) and subsoil (25-50 cm) were taken from an intensively cropped Wakanui silt loam soil (Udic Dystochrept; USDA Soil Taxonomy or Immature Pallic Soil; Hewitt 1993) from the Crop and Food Research Farm at Lincoln, New Zealand (43 [degrees] 38'S, 172 [degrees] 30'E). The soil was sieved through an 8-mm-diameter sieve and then hand sorted to remove any resident earthworms and eggs.

Ten cylinders were constructed from PVC pipe, with an external diameter of 25 cm and an internal diameter of 24.1 cm. The cylinders consisted of a topsoil (35 cm long) and a subsoil (25 cm long) section that were bolted firmly together. The base and top plates had flattened edges to ensure the same orientation was used for successive CT scans. The internal surfaces of the cylinders were coated with silica sand, stuck on with PVA glue, to discourage worms preferentially burrowing down the sides of the cylinders (Francis and Fraser 1998).

Soil was weighed (corrected for moisture content) into the amounts required to fill 5-cm layers of the cylinders to a bulk density of 1.2 g/[cm.sup.3] for the topsoil sections and 1.3 g/[cm.sup.3] for the subsoil sections. These bulk densities are typical values for this soil in the field (Watt and Burgham 1992). Equal amounts of dried, ground lucerne and ryegrass were mixed together and this herbage was mixed with topsoil at a rate of 8 g/kg soil and with subsoil at a rate of 2 g/kg soil, to provide a food source for the earthworms (Springett 1983). The subsoil sections of the cylinders were then packed using a manual press with soil in 5-cm-layer increments (to ensure even packing) to a total depth of 25 cm. The topsoil sections of the cylinders were then attached, and the topsoil packed in a similar way. The total depth of the topsoil was 25 cm, resulting in a 10-cm gap at the top of the cylinder to prevent escape of earthworms. Ryegrass (cv. Tama) was then sown at a rate equivalent to 15 kg/ha just under the soil surface and the soil volumetric moisture content adjusted to 34%, which corresponded to 95% of field capacity (Francis et al. 1999). The cylinders were transferred to a growth cabinet and initially kept at 20 [degrees] C and illuminated for 12 h per day to encourage grass germination and growth. The cylinders were weighed daily throughout the experiment, with water applied using a hand-held sprayer to bring the soil moisture content to 34%. Grass was trimmed with scissors to 10 cm above ground level when growth exceeded the height of the cylinder.

Sixteen days after sowing, the cylinders were inoculated with mature Aporrectodea caliginosa, Lumbricus rubellus, or Octolasion cyaneum earthworms at rates similar to the population density of that species in the field (Springett 1983; Fraser et al. 1996) (Table 1). After the earthworms were introduced, the growth cabinet temperature was adjusted to 16 [degrees] C and day length was adjusted to 14 h (Springett 1983). The moisture and temperature conditions used in this experiment are close to the optimal conditions for encouraging activity and survival of the 3 earthworm species (El-Duweini and Ghabbour 1968; Lee 1985). A nil earthworm control treatment was included in the experiment. All treatments were replicated 3 times, except the control, which consisted of 1 replicate only.

CT scanning and image analysis

A test of this method was made using an additional cylinder that had been packed with topsoil and inoculated with A. caliginosa earthworms. This cylinder was placed on the table of the CT scanner so that the X-ray beam would pass through the cylinder in a horizontal plane at approximately 5 cm below the soil surface. The locating light in the CT scanner was then turned on to produce a thin (0.5 mm) light beam around the outside of the cylinder. The position of this light beam was marked on the cylinder using a thin (0.5 mm) pen. An additional mark was made on the cylinder wall where it met with the line that ran down the centre of the scanner table. A CT scan was taken at this position and then the cylinder was removed from the scanning table. This cylinder was then immediately replaced on the scanning table so that the locating light was superimposed on the previously drawn pen line. A further CT scan was then taken at this position to test the precision with which the cylinder was replaced. This procedure was repeated at a different sampling depth.

Using this method, CT scans were taken of all the cylinders at 20, 40, and 60 days after earthworm inoculation. The control, A. caliginosa, and O. cyaneum cylinders were scanned in horizontal planes at depths of 5, 10, 15, 20, 30, and 40 cm from the soil surface, while the L. rubellus treatment cylinders were scanned at 5, 10, 15, and 20 cm depths. Scans below 5 cm depth were automatically positioned using the scanner's electronic controls.

The cylinders were scanned on a Technicare Deltascan 2020-G machine (operating at settings of 120 kV, 75 mA and duration of 4 s) at Lincoln University, Canterbury. Explanations of the principles and underlying theory of CT scanning have been given by others (e.g. Hainsworth and Aylmore 1983; Heijs et al. 1995). The cylinders were scanned in a 25-cm-diameter field to provide a 2-D image with a slice thickness of 2 mm. Each image corresponded to a 512 by 512 pixel matrix, where the pixel size was 0.49 x 0.49 [mm.sup.2]. Each pixel was characterised by an X-ray attenuation value that was converted to a Hounsfield Unit (HU). X-ray attenuation is influenced by both soil bulk density and moisture content. In this study, CT scans were always taken at the same soil moisture content. Consequently, the relationship between bulk density and Hounsfield Unit was linear, with a scale defined by -1000 for air and zero for water (Crestana et al. 1985). Using the part of the Hounsfield scale defined by a window width setting of 2048 and a window centre setting of 0, images were converted into a greylevel scale (from 0 = black to 255 = white). In the resulting greyscale images, black corresponded to soil pores and grey areas corresponded to soil solids.

Greylevel histograms of the images showed 2 well-separated peaks that corresponded to soil pores (mean greylevel approx. 20) and the soil matrix (mean greylevel approx. 200). These greyscale images were analysed using a standard image analysis system (videoPro 32, Leading Edge Pty, Australia), in which images were binarised into either soil pores or soil matrix based on a greylevel threshold value (Capowiez et al. 1998). Pixels with greylevel values that were less than half the mean greylevel of the soil matrix (i.e. <100) were designated as pores; the remaining pixels were designated as soil matrix (Anderson et al. 1990). Due to the averaging effect within pixel boundaries, the minimum object size that could be detected was about twice the pixel dimension (i.e. about 1 mm diam. in this case) (Warner et al. 1989; Capowiez et al. 1998). At each scanned depth, number, total area, and mean area of the pores were determined. The mean pore diameter was calculated as the equivalent spherical diameter of the mean pore area. The total pore area was also expressed on a per earthworm and a per biomass basis, using the number and biomass of inoculated earthworms. 

Images obtained at the first and second scan times were compared by electronically overlaying images (following image rotation if required to align images) to calculate the area of pores that was common for successive scans and to assess the extent of burrow backfilling of the different species. These comparisons were only made for the topsoil due to the very low burrowing activity in the subsoil of all the test species. Similar comparisons were made between the second and third scan times.

Hydraulic conductivity measurements and earthworm recovery

After the last scan (60 days after earthworm inoculation) the cylinders were separated into their topsoil and subsoil sections. The lower face of each topsoil section and both faces of each subsoil section were `picked' with a sharp needle and loose soil was removed using a vacuum cleaner to unblock any macropores that may have been smeared during the separation of the cylinders (Cameron et al. 1990). Disc permeameters (Perroux and White 1988) were used to measure the unsaturated (water supply potential of-25 mm) and saturated (supply potential of +25 mm) hydraulic conductivities of the separated topsoil and subsoil sections (Francis and Fraser 1998). Following the hydraulic conductivity measurements, the species, number, fresh weight, and age classification of the earthworms in the soil cylinders were determined by hand sorting.

Statistical analyses

The porosity data have a complex structure as (i) there was an unequal number of replicates for the main treatment (species, control), (ii) not all depths were measured for all treatments, and (iii) depths and scans both constituted repeated measures. Thus, REML methods of analysis (Patterson and Thompson 1971) were used to allow for less biased testing of the unbalanced treatment sets. Mean pore area and number of pores were square-root transformed before analysis to make the variance more homogeneous across treatments. For each of the porosity variables, various random and correlation structures were tested (Welham and Cullis 1999), and the best structure was chosen. In this process, the parts of the experimental structure (e.g. cylinders) that contributed significantly to the random variability of the data were assessed. The existence of any correlations from depth to depth (and scan to scan), and whether these correlations followed particular patterns between depths that became successively further apart, were examined. Wald statistics (GENSTAT 1997) were then used to assess which of the treatment, depth, scan, and their interactions were significantly (P < 0.05) affecting the data. For this analysis, the calculation of the exact l.s.d, is time consuming and only slightly more accurate than the approximate l.s.d. that is based on a t-value of 2. Consequently, approximate l.s.d. bars are presented in the figures. In general, there was either no correlation, or a uniform correlation, between depths (or scans). Thus, it was appropriate to treat depth as a standard split-plot treatment.

Data for comparisons between scans are balanced as depths below 20 cm were excluded. This is because no species did a significant amount of burrowing below 20 cm. Examination of data showed that the nesting of scan dates could be ignored, allowing examination using a simple factorial ANOVA. Earthworm numbers and weights, porosity per earthworm, and porosity per biomass (both summed over all sample depths) and hydraulic conductivity data were examined with ANOVA.

Results

Earthworm recovery

The experimental conditions resulted in good survival rates of all earthworm species (Table 1). Although the biomass of introduced mature earthworms in each cylinder was similar for all species, the recovered earthworm biomass was significantly less for L. rubellus than for A. caliginosa or O. cyaneum (Table 1). In addition, the weights of the individual earthworms increased for the A. caliginosa and O. cyaneum species, but declined for the L. rubellus species. In all the cylinders, the recovered mature earthworms were the same species that had been inoculated. All the cylinders (including the control) were contaminated with very low and similar numbers of immature species of A. caliginosa (data not shown). Averaged across all cylinders, the biomass of theseimmature A. caliginosa species (0.9 g/cylinder or 0.15 g/earthworm) was much less than for the mature inoculated species.

CT method testing

Greylevel images and porosity results for the repeated scanning of the test cylinder (Fig. 1, Table 2) showed that cylinders could be relocated between successive scans with good precision. When repeat images were electronically overlaid and subtracted (i.e. Scan 2 - Scan 1), the difference between images was very small, representing only about 2% of the total porosity in the image.

[FIGURE 1 OMITTED]

Soil porosity

For the control, soil porosity (expressed as a percentage of the scanned horizontal area in soil cylinders) remained very low and approximately constant with both time and depth (Fig. 2). For all the inoculated species, soil porosity increased with time. At all scan times, the porosity created in the topsoil by both A. caliginosa and L. rubellus declined significantly with depth, with the rate of decline faster for L. rubellus. In contrast, the porosity created by O. cyaneum was relatively uniform with depth in the topsoil. No species created a significant amount of porosity at 30 or 40 cm depth at any scan time. Except for the first scan time, the porosity at 5 cm depth was similar for A. caliginosa and L. rubellus. At all scan times, A. caliginosa produced more porosity than L. rubellus at 10-20 cm depth. A. caliginosa and L. rubellus produced more porosity than O. cyaneum in the top 10 cm at all scan times. O. cyaneum produced more porosity than the other 2 species at 20 cm depth at the second and third scan times.

[FIGURE 2 OMITTED]

For the inoculated species, similar results were obtained for soil porosity and the number of pores [presented as [square root of (number of pores)] in Fig. 3]. In the control, the number of pores did not show any clear trend with depth, but there was a trend toward an increase in pore numbers with time. At all scan times, A. caliginosa produced a greater number of pores than L. rubellus or O. cyaneum to 15 and 20 cm, respectively.

[FIGURE 3 OMITTED]

The mean pore area results, presented as [square root of (mean pore area)], are averaged over all 3 scan times (Fig. 4), as the mean pore area at each scan was similar for all depths and treatments. Mean pore area in the control did not vary significantly with depth. Mean pore area declined rapidly and significantly with depth for L. rubellus. In contrast, mean pore area was relatively constant throughout the top 20 cm for both A. caliginosa and O. cyaneum, with significantly larger pores for O. cyaneum. At 5 cm depth, mean pore area was similar for O. cyaneum and L. rubellus, both of which were greater than A. caliginosa. Mean pore area for O. cyaneum remained greatest to 20 cm depth. At 30 and 40 cm depth, mean pore areas for A. caliginosa and O. cyaneum were not significantly different from that for the control.

[FIGURE 4 OMITTED]

The mean amount of porosity per inoculated earthworm and per inoculated biomass (summed over all measured depths) was calculated for each species (Fig. 5). A. caliginosa and L. rubellus created the least porosity per earthworm and O. cyaneum created the most porosity per earthworm. A. caliginosa and O. cyaneum created a similar amount of porosity per biomass at all scan times, with significantly less created by L. rubellus.

The extent of backfilling of earthworm burrows in the topsoil did not significantly vary with depth for any species (data not shown). The area of backfilled pores averaged over all measured depths in the topsoil is presented as a percentage of the pore area present at the initial scans (Fig. 6). The extent of pore backfilling was high (>50%) for all species, with a trend towards more backfilling for A. caliginosa than for the other species. The only significant difference was between A. caliginosa and L. rubellus between the first and second scans.

Hydraulic conductivity

Unsaturated hydraulic conductivity (supply potential -25 mm) was unaffected by earthworm species in either the topsoil or the subsoil (Table 3). In the topsoil, saturated hydraulic conductivity was affected by earthworm inoculation, with significantly greater values for L. rubellus than for the other species. Saturated hydraulic conductivity was unaffected by earthworm species in the subsoil. Saturated hydraulic conductivity was significantly greater in the topsoil than in the subsoil for both L. rubellus and A. caliginosa.

Discussion

The conditions used in this experiment were conducive to the survival of most of the inoculated earthworms. All 3 species were active in all the periods between scans, although the activity of L. rubellus decreased by the final scan. This is supported by the low recovery of L. rubellus biomass at the end of the experiment. As environmental conditions remained unchanged during the experiment, it is likely that the reduced activity of L. rubellus with time was due to a decline in the availability of fresh organic material, which it consumes in preference to mineral soil (Piearce 1972). In contrast with the other species, L. rubellus did not burrow to any significant extent to access the organic material that was present in the soil below 10 cm. Different results for L. rubellus could possibly have been obtained if different experimental conditions (e.g. a greater amount of fresh organic material mixed with the soil) had been used.

In contrast to our previous study (Francis and Fraser 1998), contamination of the cylinders with pre-existing earthworms was very low. This was partly due to the low earthworm population in this intensively cropped soil and partly due to efficient removal of earthworms and their eggs during sieving. As a result, the control cylinder had very few pores created by earthworm burrowing. The pores that were present in the control cylinder were small and probably a combination of packing pores and pores created by root growth. The porosity in the other cylinders was therefore mainly the result of the burrowing activity of the mature, inoculated earthworms.

Results from the test cylinder showed that the method used to relocate cylinders at successive scan times and the electronic subtraction of overlaid images produced only small errors (1-2%). These errors were much smaller than the differences between scan times or between earthworm species.

In previous CT studies of earthworms, burrows were often observed to have been made preferentially along the cylinder walls (Joschko et al. 1991, 1993). However, burrowing along the cylinder walls was limited in our experiment (Fig. 7), partly because of the large diameter of our cylinders and partly because of the fine sand adhering to the cylinder walls.

Burrow sizes and burrowing behaviour clearly differed between the earthworm species. At their depth of maximum activity, mean pore diameters were about 4.5-5.5 mm for O. cyaneum and L. rubellus and about 4.0 mm for A. caliginosa. Similar differences in the mean pore diameter created by these earthworm species have been reported by other workers (Springett 1983; Joschko et al. 1991; Francis and Fraser 1998), and are to be expected from the mean weights of the inoculated earthworms (Table 1). The depth of maximum pore production for L. rubellus was 5 cm, with very little activity below 10 cm (Fig. 2). A. caliginosa created most of its burrows in the top 10 cm, with little activity below 15 cm. In contrast, O. cyaneum created its burrows relatively uniformly throughout the top 20 cm. Very few pores were created below 20 cm in this experiment by any of the species, probably due to the favourable temperature and moisture conditions that were maintained in the topsoil. The lower bulk density and greater food source in the topsoil compared with the subsoil could also have contributed to the lack of burrowing in the subsoil. This contrasts with other studies where burrows of both A. caliginosa and O. cyaneum have been reported to 45 cm depth, presumably to avoid unfavourable conditions closer to the soil surface (Pitkanen and Nuutinen 1997; Francis and Fraser 1998).

The pores produced by the earthworms in this experiment occupied 2-12% of the total soil area in the topsoil (Fig. 2). Similar porosities and numbers of pores due to earthworm burrowing have been reported for other studies both of undisturbed field soils (Munyankusi et al. 1994; Pitkanen and Nuutinen 1997) and 3-6 months after the inoculation of repacked soil cores with comparable earthworm populations (Springett 1983; Munyankusi et al. 1994; Francis and Fraser 1998; Langmaack et al. 1999).

Porosity created by earthworms was greater in the cylinders inoculated with A. caliginosa than in those inoculated with either L. rubellus or O. cyaneum. This was due to the greater rate of inoculation for A. caliginosa earthworms than the other species. However, the mean amount of porosity created by individual earthworms declined in the order O. cyaneum > L. rubellus = A. caliginosa (Fig. 5a), mainly in relation to earthworm size. Expressing porosity per inoculated biomass removes the effect of different rates of earthworm inoculation and the different size of individual earthworms. L. rubellus had the lowest porosity per inoculated biomass, suggesting that it had a lower burrowing activity than either A. caliginosa or O. cyaneum. The relative activity of L. rubellus may be even lower than it initially appears since the extent of burrow backfilling was least for L. rubellus. The lower rate of activity of L. rubellus could be a result of the apparently unfavourable experimental conditions that also led to its reduced survival rate compared with A. caliginosa or O. cyaneum.

[FIGURE 5 OMITTED]

Representative images from O. cyaneum at 10 cm depth at the first (Fig. 7a) and second scan times (Fig. 7b) illustrate the backfilling of earthworm burrows. Between the scan times some pores were backfilled and some new pores were created. Other workers have also observed backfilling of both A. caliginosa and O. cyaneum burrows (Joschko et al. 1991, 1993), resulting in a low level of burrow continuity (Munyankusi et al. 1994; Langmaack et al. 1999). The extent of burrow backfilling by A. caliginosa in our experiment (72-85%) was similar to that reported using a direct observation technique over 68 days (Hirth et al. 1996). As well as a change in porosity, some areas of the soil matrix became whiter in colour, indicating an increase in X-ray attenuation. This change in the colour of the soil matrix appears to be due to earthworm burrowing, as the colour for the control between these scan times remained unchanged (Fig. 7c, d). X-ray attenuation probably increased because the water content of the freshly deposited earthworm casts in backfilled burrows was higher that the water content of the surrounding soil matrix (Marinissen and Dexter 1990).

[FIGIRE 7 OMITTED]

We measured hydraulic conductivity in this experiment to assess the continuity of earthworm burrows. During the measurement of unsaturated hydraulic conductivity, pores greater than 1.2 mm equivalent spherical diameter were excluded from participating in water flow. As the mean diameter of pores for all species was >1.2 mm, unsaturated hydraulic conductivity was not increased in either the topsoil or the subsoil by earthworm inoculation. However, under saturated conditions, when all pores could potentially participate in water flow, hydraulic conductivity in the topsoil was significantly affected by earthworm inoculation. Earthworms did not affect saturated hydraulic conductivity in the subsoil, which was to be expected from the low level of earthworm burrowing at this depth (Fig. 2). In the topsoil, saturated hydraulic conductivity was not clearly related to soil porosity. Saturated hydraulic conductivity in the topsoil was greatest for L. rubellus, but soil porosity in the topsoil was greatest for A. caliginosa. These results suggest that the pores created by L. rubellus were surface-connected and more continuous than those created by A. caliginosa. The relatively low saturated hydraulic conductivity for O. cyaneum in the topsoil was probably due to the limited number of its burrows that were likely to be open at the soil surface (Francis and Fraser 1998).

This study has clearly shown the contrasting burrowing characteristics of L. rubellus, A. caliginosa, and O. cyaneum. The results we obtained in sieved soil are likely to be similar to results that would have been obtained in the topsoil of cultivated paddocks, as the soil bulk density and aggregate size are similar in both cases. In contrast, burrowing may be easier in sieved than undisturbed subsoil, but this is unlikely to have significantly affected our results as the extent of burrowing in the subsoil was small in all cases. The species varied in the size and number of pores produced, the depth of burrowing, and the extent of burrow backfilling. L. rubellus seems to make a particularly important contribution to improving soil physical conditions as its burrows appear to be connected to the soil surface and are the most permanent, but limited to the top 10 cm, for the 3 species studied. In New Zealand, L. rubellus are present in greatest numbers under pasture (Fraser et al. 1996) and this may partly explain the improvement in soil physical conditions that are observed when soils are converted from arable to pastoral production (Francis et al. 1999). Nevertheless, low numbers of L. rubellus can be found in soils that have been cropped for less than 3 years (Fraser et al. 1996). It is possible that changing the management practices for cropped soils to increase fresh soil organic matter inputs to the surface soil (e.g. direct drilling, retaining postharvest crop residues, growing green manure crops) may help to sustain a population of L. rubellus under cropping for longer. This needs further investigation.

Table 1. The mean number and fresh weight biomass of mature earthworms inoculated in the soil cylinders at the beginning of the experiment and  mature earthworms recovered at the end of the experiment.

                                Inoculated earthworms

Species               No./      Biomass/     Biomass/                    
                       cylinder    cylinder      worm (B)                                
                                         (AB) (g)        (g)  

L. rubellus          18       15 (-2.7)    0.8 (-0.19) 

A. caliginosa       36      13 (-2.59)   0.4 (-1.00) 

O. cyaneum          9      12 (-2.45)   1.3 (-0.25) 

l.s.d. (P = 0.05;                (-0.06)       (-0.06)  
  d.f. = 6)                                     

                                 Recovered earthworms  

Species              No./    Recovery     Biomass/         Biomass/                    
                          cylin-      (%)           cylinder           worm                     
                           der                          (AB) (g)          (B) (g)   

L. rubellus          13.7      75.9      6.9 (-1.93)    0.51 (-0.0679) 

A. caliginosa       34.7      96.3     16.7 (-2.81)    0.48 (-0.734) 

O. cyaneum          7.3      81.5     14.3 (-2.66)    1.97 (-0.679) 

l.s.d. (P = 0.05;    3.8       24.6          (-0.493)        (-0.2886) 
  d.f. = 6)  

(A) After gut voidance in water for 24 h.  

(B) Inoculated and recovered weights of earthworms were natural log
transformed before analysis. Back-transformed means are given, with log means and l.s.d. values in brackets.

Table 2. Soil porosity (%) for repeated CT scans of the test cylinder and following electronic overlaying of images                       

                                  5 cm depth   10 cm depth  

Scan 1                             5.74         6.54 

Scan 2                            5.73         6.61 

Scan 2 - Scan 1            0.11         0.14 

 
Table 3. Hydraulic conductivities (mm/h) in the topsoil and subsoil 

sections of the cylinders, measured at the end of the experiment  

Species                 Unsaturated         Saturated                       
                        Topsoil   Subsoil   Topsoil   Subsoil  

L. rubellus        82         78       594        53 

A. caliginosa    59          71      345         51 

O. cyaneum     72          75       178        96 

Control            64         102       107        67  

l.s.d. (P = 0.05)      46.0 (A)            198.5 (A)                          

                              65.0 (B)            280.8 (B)                          

                              33.2 (C)            208.9 (C)  

(A) For comparisons between 2 species, within topsoil or subsoil
(d.f. [approximately equal to] 10).  

(B) For comparisons between control and any species, within topsoil  or subsoil (d.f. [approximately equal to] 10).  

(C) For comparisons between topsoil and subsoil for the same species
(d.f. = 5).

Acknowledgments

We thank Kathryn White for expert help in the running of this experiment, Dr Mark Young and Nigel Jay (Lincoln University) for assistance with CT scanning, and Dr Graeme Coles for assistance with image analysis. This work was carried out as part of the Foundation for Research, Science and Technology contract number C02613.

References

Anderson SH, Peyton RL, Gantzer CJ (1990) Evaluation of contructed and natural soil macropores using X-ray computed tomography. Geoderma 46, 13-29.

Cameron KC, Harrison DF, Smith NP, McLay CDA (1990) A method to prevent edge-flow in undisturbed soil cores and lysimeters. Australian Journal of Soil Research 28, 879-886.

Capowiez Y, Pierret A, Daniel O, Monestiez P, Kretzschmar A (1998) 3D skeleton reconstructions of natural earthworm burrow systems using CAT scan images of soil cores. Biology and Fertility of Soils 27, 51-59.

Crestana S, Mascarenhas S, Pozzi-Mucelli RS (1985) Static and dynamic three-dimensional studies of water in soil using computed tomographic scanning. Soil Science 140, 326-332.

El-Duweini AK, Ghabbour SI (1968) Nephridial systems and water balance of three Oligochaeta genera. Oikos 19, 61-70.

Francis GS, Fraser PM (1998) The effects of three earthworm species on soil macroporosity and hydraulic conductivity. Applied Soil Ecology 10, 11-19.

Francis GS, Tabley FJ, White KM (1999) Restorative crops for the amelioration of degraded soil conditions in New Zealand. Australian Journal of Soil Research 37, 1017-1034.

Fraser PM, Williams PH, Haynes RJ (1996) Earthworm species, population size and biomass under different cropping systems across the Canterbury Plains, New Zealand. Applied Soil Ecology 3, 49-57.

GENSTAT (1997) `GENSTAT 5 Release 4.1 Reference Manual Supplement.' (NAG: Oxford)

Hainsworth JM, Aylmore LAG (1983) The use of computer-assisted tomography to determine spatial distribution of soil water content. Australian Journal of Soil Research 21, 435-443.

Heijs AWJ, de Lange J, Schoute JFT, Bouma J (1995) Computed tomography as a tool for non-destructive analysis of flow patterns in macroporous clay soils. Geoderma 64, 183-196.

Hewitt AE (1993) `New Zealand soil classification.' (Manaaki-Whenua Press: Lincoln, NZ)

Hirth JR, McKenzie BM, Tisdall JM (1996) Volume density of earthworm burrows in compacted cores of soil as estimated by directed and indirect methods. Biology and Fertility of Soils 21, 171-176.

Jegou D, Hallaire V, Cluzeau D, Trohen P (1999) Characterization of the burrow system of the earthworms Lurnbricus terrestris and Aporrectodea giardi using X-ray computed tomography and image analysis. Biology and Fertility of Soils 29, 314-318.

Joschko M, Graff O, Muller PC, Kotzke K, Lindner P, Pretschner DP, Larink O (1991) A non-destructive method for the morphological assessment of earthworm burrow systems in three dimensions by X-ray computed tomography. Biology and Fertility of Soils 11, 88-92.

Joschko M, Muller PC, Kotzke K, Dohring W, Larink O (1993) Earthworm burrow system development assessed by means of X-ray computed tomography. Geoderma 56, 209-221.

Langmaack M, Schrader S, Rapp-Bernhardt U, Kotzke K (1999) Quantitative analysis of earthworm burrow systems with respect to biological soil-structure regeneration after soil compaction. Biology and Fertility of Soils 28, 219-229.

Lee KE (1985) `Earthworms: Their ecology and relationships with soils and land use.' (Academic Press: Sydney)

Marinissen JCY, Dexter AR (1990) Mechanisms of stabilization of earthworm casts and artificial casts. Biology and Fertility of Soils 9, 163-167.

Munyankusi E, Gupta SC, Moncrief JF, Berry EC (1994) Earthworm macropores and preferential transport in a long-term manure applied Typic Hapludalf. Journal of Environmental Quality 23, 773-784.

Patterson HD, Thompson R (1971) Recovery of inter-block information when block sizes are unequal. Biometrika 58, 545-554.

Perroux KM, White I (1988) Designs for disc permeameters. Soil Science Society of America Journal 52, 1205-1215.

Piearce TG (1972) The calcium relations of selected Lumbricidae. Journal of Animal Ecology 41, 167-188.

Pitkanen J, Nuutinen V (1997) Distribution and abundance of burrows formed by Lumbricus terrestris L. and Aporrectodea caliginosa Sav in the soil profile. Soil Biology and Biochemistry 29, 463-467.

Springett J, Gray R (1997) The interaction between plant roots and earthworm burrows in pasture. Soil Biology and Biochemistry 29, 621-625.

Springett JA (1983) Effect of five species of earthworm on some soil properties, dournal of Applied Ecology 20, 865-872.

Springett JA (1992) Distribution of lumbricid earthworms in New Zealand. Soil Biology and Biochemistry 24, 1377-1381.

Springett JA, Gray RAJ (1998) Burrowing behaviour of the New Zealand indigenous earthworm Octochaetus multiporus (Megascolecidae: Oligochaeta). New Zealand Journal of Ecology 22, 95-97.

Warner GS, Nieber JL, Moore ID, Geise RA (1989) Characterizing macropores in soil by computed tomography. Soil Science Society of America Journal 53, 653-660.

Watt JPC, Burgham SJ (1992) Physical properties of eight soils of the Lincoln area, Canterbury. DSIR Land Resources Technical Record.

Welham S, Cullis B (1999) Analysis of correlated data using REML. In `Proceedings of Australian GENSTAT Conference, GENSTAT 99'. Melbourne, 1999.

Manuscript received 5 June 2000, accepted 12 April 2001

G. S. Francis, F. J. Tabley, R. C. Butler, and P. M. Fraser

New Zealand Institute for Crop and Food Research Limited, Private Bag 4704, Christchurch, New Zealand.