Belowground eco-restoration of a suburban waste-storage landscape: Earthworm dynamics in grassland and in a succession of woody vegetation covers.

Restoration of belowground ecology is seldom a priority in designing revegetation strategies for disturbed landscapes. We determined earthworm abundance and diversity in a 16-year old grass sward (grassland), a 6-year old (Plantation-04) and a 4-year old (Plantation-06) plantation, both of mixed woody species, on a reclaimed waste disposal site, and in nearby remnant woodland, in suburban Sydney, Australia. While no catches were made in autumn, more earthworms were found in spring (21 ± 8.6 m-2) than in winter (10.2 ± 5.9 m-2) or summer (14.4 ± 5.5 m-2). Earthworm abundance in spring was in the order grassland ≈ Plantation-04 (35.2 m-2) > woodland (12.8 m-2) > Plantation-06 (0.8 m-2). None of the revegetated covers had restored earthworm diversity to levels found in the woodland. Exotic species, mostly Microscolex dubius, dominated in the four vegetation covers at any time; the only two native species (Heteroporodrilus sp. and Megascoleceides sp.) found were in the woodland. We also assessed how quality of the evolving soils from the three revegetated covers, compared with that from the woodland, impacted viability of common exotic earthworm species. Both weight gain and cocoon production by the exotic earthworms were higher in the soil from Plantation-04 than in soils from the other vegetation covers, including the woodland; the two variables were positively correlated with the pH and mineral nutrient content (as indicated by electrical conductivity that was in turn correlated with clay content) of the soil. Age of vegetation rather than its composition explained differences in the level of earthworm recovery observed.

Introduction

There is an over-reliance on aboveground vegetation processes, including plant growth, in assessing efficacy of revegetation strategies for lanclass="Chemical">dscaclass="Chemical">pes used for waste disclass="Chemical">posal or mining. This is because restoration of ecohydrologic class="Chemical">process to minimize the risk of chemical and class="Chemical">particlass="Chemical">pan class="Chemical">culate pollution of the atmospheric and water resources often takes priority over other considerations. Restoration of soil ecosystem tends to receive little attention, despite its central role in maintaining soil health that ultimately underpins success of the planted vegetation and its dependent processes (Bradshaw, 1984). Both the chemical and physical properties of the recovered soil, along with its reprocessing and use in seedbed preparation, all impact on the survival and viability of the macrofauna, and on that of the plant species.

The need to conserve local biodiversity and aesthetics has inclass="Chemical">creased the use of native class="Chemical">plant sclass="Chemical">pecies for ecological restoration (Grant, Camclass="Chemical">pan class="Chemical">pbell, & Charnock, 2002; Weir, Fulton, & Menzies, 2006). Woody species are therefore the appropriate choice for environments along the coastal fringes of southeastern Australia, where woodland is the dominant natural vegetation. However, relatively high cost and the long lead-time to achieve effective vegetation cover with woody species, make the use of herbaceous species, grasses in particular, attractive in some instances (Richardson, Burn, & Craig, 1987). Differences in species composition of vegetation covers can alter the evolving belowground ecosystems on rehabilitated landscapes. For instance restoration of physical properties, such as porosity and permeability, may take longer under grassland than woodland (Richardson et al., 1987; Yunusa et al., 2012), and will have significant influence on the developing soil biota such as earthworm communities (Chan and Barchia, 2007) or ants (Gollan, Lobry de Bruyn, Reid, Smith, & Wilkie, 2010).

class="Species">Earthworms are effective indicators of soil health because of their sensitivity to both immutable and anthroclass="Chemical">pogenic stresses in the soil, and are widely used to assess the ecological consequences of disturbance arising from changes in land use (Boyer and Wratten, 2010; Chan and Heenan, 2006; Hendrix et al., 2006; Smith et al., 2008). For instance, Smith et al. (2008) found a strong tendency towarclass="Chemical">pan class="Chemical">ds increasing earthworm populations (abundance) in minimally disturbed landscapes, with earthworms being most abundant in an old growth forest. In addition to changes in abundance, land disturbance also induces strong alterations in the distribution of earthworm species. Land disturbance often results in the displacement of native earthworm species by exotic species (Chan and Heenan, 2006) such as the European Aporrectodea species that tend to easily displace pre-existing native species on disturbed lands (Scullion, Ramshaw, & Mohammed, 1988; Smith et al., 2008). It is probable that reconstituted soils may become inhospitable to native species, and in Australia exotic species are known to thrive better than native species in soils contaminated with heavymetals (Yunusa, Braun, & Lawrie, 2009) by minimizing ingestion of potentially harmful trace metals (Muir et al., 2007). The rarity of native species in disturbed lands could, however, be simply a function of time needed to recolonize the now changed habitat and readjust to the now changed edaphic environment, a processes that may take decades (Boyer and Wratten, 2010; Bradshaw, 2000).

In this study we used class="Species">earthworm abundance, as surrogate of the whole soil class="Chemical">pan class="Species">biota, to assess the degree to which three vegetation covers consisting of a 16-year old grassland sward and a four- and a six-year old woody vegetation covers had restored earthworm communities on a reclaimed waste disposal site, when compared with an old growth woodland. We undertook field surveys to characterize seasonal changes in the abundance and diversity of earthworms in the four vegetation covers, and laboratory experiments to test how the soils from the four vegetation covers impacted viability of two common exotic earthworm species.

Materials and methods

This study was conducted at the Waste Management Centre at Castlereagh (33°39′41″S, 150°46′57″E or Google Locator −33.658781,150.780973) approximately 65 km north-west of central Sydney. The site covers an area of 357 hectares and the original soil at the site is classified as Chromosol, which is equivalent to Haplic Xerosol (FAO, 1974). This soil has a duplex profile consisting of 0.7 m loamy sand topsoil over impermeable heavy clay referred to as Londonderry Clay overlaying conglomerate sanpan class="Chemical">dstone and shales of the Triassic Wianamatta Grouclass="Chemical">p (Itakura, Airey, & Dobrolot, 2005). Storage cells consisting of trenches (20 m × 5 m, and 5 m deeclass="Chemical">p) constructed into the clay subsoil, and sclass="Chemical">paced 2 m aclass="Chemical">part resulting in aclass="Chemical">pclass="Chemical">proximately 65 cells/ha. Once filled with wastes, the cells were caclass="Chemical">pclass="Chemical">ped using the excavated soil that was returned in reverse order of their removal. The caclass="Chemical">p forms a class="Chemical">profile of 2 m over the cells. The reconstructed soil was then class="Chemical">planted with either grasses or mixtures of woody and herbaceous sclass="Chemical">pecies. For this class="Chemical">project, we selected three vegetation covers along with nearby remnant woodland, desclass="Chemical">pan class="Chemical">cribed below:

Grass pasture: This was established in 1994 with a mixture of class="Species">Cynodon dactylon (couch grassland), class="Chemical">pan class="Species">Axonopus affinis (carpet grassland), Paspalum dilatatum (paspalum), Pennisetum clandestinum and Trifolium repens (white clover). Prior to sowing, the soil was fertilized with 300 kg/ha or 30 mg/m2 of compound fertiliser containing mainly nitrogen, phosphorus and potassium (24:6:12). The grassland was about 16 years old at the start of this study, and it represented a relatively quick and cheap rehabilitation strategy.

class="Chemical">Plantation-04: This was established in autumn (Aclass="Chemical">pril–May) 2004 using a mixture of native trees and shrubs class="Chemical">planted in 5 m rows. The tree sclass="Chemical">pecies used were Eucalyclass="Chemical">ptus sclass="Chemical">pclass="Chemical">p., class="Chemical">pan class="Species">Angophora spp., Casuarina glauca, Melaleuca linariifolia and Syncarpia glommulifera and were interplanted with rows of shrubs made up of species of Acacia, Callistemon, Grevillea, Hakea, Kunzea and Leptospermum. Mineral fertiliser containing nitrogen, phosphorus and potassium (NPK) was applied at the time of establishment. The vegetation was almost six years old at the start of this study and provided a mixture of woody and herbaceous species as a mimic of native woodlands.

class="Chemical">Plantation-06: This was established in 2006 using a mixture of woody sclass="Chemical">pecies as in class="Chemical">pan class="Chemical">Plantation-04, but without shrub species or grassland groundcover. Herbaceous weedy species became established as groundcover. A thin layer (<0.1 m thick) of compost was sprayed over the soil in 2008, and was well incorporated into the topsoil at the time of this study. This cover represented a minimum rehabilitation strategy and was four years old at the start of this study.

Woodland: This is dominated by trees of Eucalyptus parramentensis and class="Species">Angophora bakeri, in which the understorey is dominated by shrubs and grasslandes including class="Chemical">pan class="Disease">Pultinea elliptica Smith, Cryptandra amara Smith and Melaleuca thymifolia (Yunusa et al., 2010). There was no record of any major disturbance of this vegetation aside from occasional fires, the last of which was in 1999.

All the four vegetation covers were within a 1.5 km radius of each other and have been desclass="Chemical">cribed in detail in class="Chemical">previous studies (Yunusa et al., 2010, 2012). Basic characteristics of the toclass="Chemical">psoil under the four vegetation covers are given in Table 1. The woodland is characterized by coarse sandy texture that is mildly acidic and class="Chemical">pan class="Chemical">saline with 0.044 dS m−1 and except for its higher total carbon content, had lower clay content and cation exchange capacity (CEC), and was less saline and dense, than those under the three revegetated sites all of which contained more trace metals (Table 1).

Table 1

Soil properties for the topsoil (surface 0.3 m) under the four vegetation types at Castlereagh, New South Wales, Australia.

Properties	Woodland	Grass	Plantation-04	Plantation-06
Particle size distribution (%)
Gravel	5	27	52	30
Sand	92	49	31	54
Silt	3	6	5	2
Clay	0	18	12	14
Basic chemical property
Soil pH (1:5 H2O)	5.8	6.8	5.8	6.7
Soil salinity (dS m−1)	0.04	0.11	0.17	0.13
Bulk density (Mg/m3)	1.20	1.48	1.32	1.28
Total porosity (%)	55	45	51	52
Total C %	0.90	0.82	0.31	0.23
Total N %	0.04	0.07	0.05	0.05
C:N	22.5	11.7	6.2	4.6
CEC (cmol+/kg)	2.27	9.62	9.82	10.22
Concentration of selected trace metals (mg/kg)
Arsenic	7	20	9	19
Lead	12	38	17	29
Chromium	40	87	44	76
Copper	4	21	17	13
Manganese	25	116	19	33
Zinc	7	26	18	26

Field surveys

Field surveys were conducted on April 21, July 14, September 7 and December 15, 2010 to determine species diversity and abundance of class="Species">earthworm communities under each of the vegetation covers. On each occasion, 20 quadrats (0.25 m × 0.25 m) were randomly samclass="Chemical">pled aclass="Chemical">pan class="Chemical">cross each of the four vegetation covers. For each quadrat, the soil was collected to a depth of 0.25 m, sieved and the earthworms recovered which were then counted and transferred into bottles containing 70% ethanol. The earthworms were later sorted and identified in the laboratory after Blakemore (2008), although some juvenile earthworms could not be identified because of the incomplete development of their distinguishing features. The number of earthworms was expressed as mean population density and the percentage of adults was also determined.

Daily weather variables consisting of minimum and maximum temperature, humidity and rainfall for the site were monitored with an automatic weather station. These data were used along with canopy characteristics for the vegetation covers (Yunusa et al., 2012) to predict temperature at 0.2 m depth under the four vegetation covers using the model of pan class="Chemical">Paul et al. (2004).

Laboratory experiments

A series of experiments were undertaken using soil samples collected from the four vegetation types. The soil samples from each of the vegetation type were combined and later mixed thoroughly in a cement mixer and to minimize effect of spatial variability during collection from the field; the composite samples were used to study how textural and basic chemical characteristics of these soils influenced survival, growth and activity of exotic and native class="Species">earthworm sclass="Chemical">pecies. The soil samclass="Chemical">ples were ground and used to fill 1.5 L (toclass="Chemical">p internal diameter 140 mm) class="Chemical">pots. The class="Chemical">pots were lined at the base with fly sclass="Chemical">pan class="Chemical">creen mesh (2.0 mm × 2.0 mm) and 20 mm thick coconut fibres to hold the soil and prevent any earthworm escapes. These pots were used in the following experiments.

Survival and growth of exotic earthworms in soils from the four vegetation covers

class="Species">Earthworms of class="Chemical">pan class="Species">Aporrectodea trapezoides or Aporrectodea caliginosa were each used to inoculate 64 pots; this allowed sampling of four replicates for each treatment at four time intervals. The pots were then thoroughly wetted with 1 L of deionised water two days before each was inoculated with three juvenile earthworms on the June 7, 2010. Average weight for a single worm was 0.66 ± 0.08 g for A. trapezoides and 0.40 ± 0.05 g for A. caliginosa. Prior to inoculation, the earthworms were washed with deionised water and then weighed before being placed in the centre of the pot and allowed to burrow. The earthworms were observed to burrow into the soils and disappeared within 5 minutes. The pots were then covered with fly screen mesh (3.0 mm × 3.0 mm) to prevent any escape through the soil surface.

After inoclass="Chemical">culation, the class="Chemical">pots were randomly assigned to numbered sclass="Chemical">paces on a shelf in a temclass="Chemical">perature controlled growth room, which was keclass="Chemical">pt at a constant temclass="Chemical">perature of 19 °C throughout the exclass="Chemical">periment. The class="Chemical">pots were keclass="Chemical">pt moist by adding 100 ml of deionised class="Chemical">pan class="Chemical">water every week. All the pots were supplied with 2 g of dry sheep manure at the start of the experiment, which was then administered every 2–3 weeks. The total amount of food administered to all pots by the end of the experiment was 10 g. At 2, 5, 8 and 11 weeks after inoculation, 32 pots made up of four each for A. trapezoides and A. caliginosa were randomly sampled from each soil type. At each sampling, the number and weight of surviving earthworms were determined, while the cocoons and casts were recovered from the surface of each pot and weighed following Muir et al. (2007). Soil in each pot was sampled to determine pH and electrical conductivity (EC).

Assessment of textural and nutritional alterations of soil on survival and growth of exotic earthworms

Two experiments were undertaken to assess how alterations in the texture of the clayey soil from class="Chemical">Plantation-04, and in the nutrient content of the loamy soil from the woodland, influence the growth of exotic endogeic class="Chemical">pan class="Species">A. trapezoides and Amynthas gracilis over four weeks from July 30 to the August 23, 2010.

The first experiment involved textural alteration of the clay soil in pan class="Chemical">Plantation-04 using the following treatments:

Control: unamended soil.

Sand amendment: 9 kg of washed sand was mixed with 9 kg of Plantation-04 soil.

Organic matter amendment: soil mixed with 3.7% (w/w) of compost to raise the pan class="Chemical">carbon content to 0.96%, close to the level found in the woodland.

Sand and organic matter amendment: mixture of equal amounts of sand with 3.7% (w/w) of compost.

The mixtures were achieved with a cement mixer and then used to fill the pots as desclass="Chemical">cribed above. The aim of the sand amendment was to alter the texture of the class="Chemical">plantation soil by inclass="Chemical">pan class="Chemical">creasing the percentage of sand by up to 50%. Thirty two pots were set up in total and each inoculated with juvenile earthworms of either A. trapezoides or A. gracilis as described above, average weight for A. trapezoides was 0.52 ± 0.07 g and for A. gracilis was 0.89 ± 0.06 g. The response variables measured were: number and weight of surviving earthworms, number of cocoons and weight of casts. Soil sample from each pot was collected to determine pH and the EC.

The second experiment assessed how nutrient addition to the woodland soil influenced performance of exotic class="Species">earthworms consisting of class="Chemical">pan class="Species">A. caliginosa and A. trapezoides. There were two treatments:

Control: unamended soil.

Fertilizer amendment: a compound fertilizer containing class="Chemical">nitrogen, class="Chemical">pan class="Chemical">phosphorus and potassium (15% N: 4.4% P: 10% K) at a rate equivalent to 300 kg/ha was mixed into the soil. This was achieved by mixing 17.3 g of fertiliser with 37 kg of soil, i.e. 0.047% (w/w), in the cement mixer.

These two soil treatments were used along with the two exotic class="Species">earthworm sclass="Chemical">pecies (class="Chemical">pan class="Species">A. caliginosa and A. gracilis) in a factorial design of four treatments with four replications, thus 16 pots were involved. Juvenile earthworms were used with mean weight of 0.25 g for A. caliginosa and 0.51 for A. trapezoides. The pots were maintained in the same manner as described above. Four weeks after inoculation, the earthworms were recovered and the following variables were measured: number and weight of earthworms, number of cocoons and weight of castings produced. Soil samples were also taken from each pot to determine pH and EC.

Data analysis

Statistical analysis of the data was carried out using Sclass="Chemical">PSS v17 in which all data were tested for normality, while Levene's test was used to determine equality of variances. Statistical significance was determined when class="Chemical">p ≤ 0.05. Tukeys class="Chemical">pan class="Disease">HSD was used to test for differences between groups when the effect was significant. For the field survey data, one-way analysis of variance (ANOVA) was used to tests for differences in earthworm abundance between the four vegetation types. The extent to which earthworm population was restored by the three revegetation strategies was determined using the worm population collected in spring to calculate the effect size (d) after Gurevitch and Hedges (1993):in which and are the respective mean values for the revegetated blocks (grassland, Plantation-06 or Plantation-04) and the woodland, and σpooled is the standard deviation for the pooled data for all the four vegetation covers. For the laboratory experiments, treatment effects were tested using two-way ANOVA.

Results

Seasonal trends in earthworm population composition

Weather conditions during the year were consistent with the sub-temperate climate of the site. The first sampling in April (autumn) followed almost a 2-month period when little rain fell (Fig. 1), while it was cool and relatively dry in July (winter). At the third in September (spring) the weather had warmed up and also wet due to frequent rainfalls, and this conditions largely persisted to the fourth sampling in December (summer). Model predictions showed that the soil was warmest in autumn and coldest in winter, with the difference being up to 5 °C (Table 2). The soil was always cooler under pan class="Chemical">Plantation-04 and warmest in the grassland, where the soil temclass="Chemical">perature was often within 1 °C that in class="Chemical">pan class="Chemical">Plantation-06.

Fig. 1

Mean weather conditions at Castlereagh, Australia, during the study in 2010. The stars indicate when field samplings were undertaken.

Table 2

Predicted mean soil temperature (°C) at 0.2 m depth under the various vegetation covers at the times of sampling in 2010.

Vegetation covers	Autumn	Winter	Spring	Summer
Woodland	23.0	16.7	18.2	19.2
Grass	24.1	17.8	19.1	20.2
Plantation-04	21.5	14.9	16.3	19.7
Plantation-06	23.6	17.3	17.8	19.5

Prediction was obtained with model of Paul et al. (2004).

class="Species">Earthworms were not found in autumn, but were found in variable numbers during the other three seasons. The mean abundance for the whole site doubled from 10.2 m–2 in winter to 21.8 m–2 in sclass="Chemical">pring, but then fell by 30% to 14.4 m–2 in summer (Fig. 2a). The grassland sward consistently had more class="Chemical">pan class="Species">earthworms than the woodland, while Plantation-06 had the lowest number of earthworms amongst all the vegetation covers. Differences in the earthworm density amongst the vegetation covers were largest and only significant in spring, when the density was in the order Plantation-04 > grassland > woodland > Plantation-06. These differences were marginal in winter and summer despite the population density being at least 25% higher in the grassland sward than in the woodland. The number of species did not exceed three in any of the vegetation covers with fewer species found in summer than in the other season (Fig. 2a).

Fig. 2

Population profiles for earthworms in 2010 at Castlereagh, Australia 2010: mean abundance of earthworms collected from the four vegetation covers during the three seasons (a), and species composition of the samples collected in winter (b), spring (c), and summer (d). In (a) numerals in [parentheses] are the mean (±standard error) of earthworms/m2; species in (b) are A, A. trapezoides; H, Heteroporodrilus spp.; d, Microscolex dubius; M, Megascolecedes spp; O, other species (unidentified). Both Heteroporodrilus and Megascolecedes spp. are native species.

Only the grassland showed as much species diversity as the woodland, even though native species were found only in the woodland. In winter class="Species">Microscolex dubius accounted for at least 58% of class="Chemical">pan class="Species">earthworms found in all the four vegetation covers, and were the only species found in the two plantations (Fig. 2b). Unidentified species accounted for about 30% of the earthworms in the woodland and the grassland with balance made up by a native Megascoleceide species in the woodland (15%) and A. trapezoides (10%) in the grassland. Except for Plantation-06 that contained only A. trapezoides, unidentified species constituted between 35% and 70% of the earthworms found in the other vegetation covers in winter (Fig. 2c). In winter a native species Heteroporodrilus species accounted for about 8% of the population in the woodland, while A. trapezoides accounted for 12% of earthworms in the grassland. In summer, except the woodland in which 18% of the earthworms were of unidentified species, the worm populations were made up entirely of Microscolex dubius in all the four vegetation covers (Fig. 2d).

The analysis of effect size (d) using abundance data in spring showed significant improvements in class="Species">earthworm class="Chemical">poclass="Chemical">pulations with grassland and class="Chemical">pan class="Chemical">Plantation-04, but a significant reduction in Plantation-06, relative to the woodland (Fig. 3).

Fig. 3

Mean effect size (±95% confidential intervals) for earthworm abundance in the three vegetation covers in spring of 2010 at Castlereagh, Australia. The woodland provides the baseline (zero line) data.

Soil type influence on exotic earthworm species

Survival rate for the two class="Species">earthworm sclass="Chemical">pecies was similar in the four soils (Table 3); survival averaged 87 ± 6%. class="Chemical">pan class="Species">Earthworm biomass increased in all the four soils during the first two weeks of inoculation, more so in Plantation-04 (Fig. 4a). The earthworms continued to grow in Plantation-04 for another three weeks until the 5th week before declining. These initial weight gains in both earthworms were up to 90% in the soil from Plantation-04 compared with <20% in the other three soils. By the end of 11 weeks of incubation, the earthworm species were at least 20% lighter than their starting weights except in Plantation-04 in which the earthworms had a 42% net gain in weight. Changes in weight were more rapid in A. trapezoides than in A. caliginosa, whereas the latter posted a net gain in weight of 12%, A. trapezoides lost 10% of its weight, relative to their initial weights (Fig. 4b).

Table 3

Characteristics of the earthworms inoculated in, and electrical conductivity (EC) of, the soils collected from the vegetation covers measured 11 weeks after inoculation.

Treatments	Survival (%)	Cocoons/worm (g)	Casts/worm (g)	Burrows (no./m2)	EC (dS m−1)
Vegetation covers
Woodland	85	0.13b	1.65a	219b	0.03c
Grass	89	0.13b	0.34c	365a	0.06bc
Plantation-04	87	1.25a	0.69bc	390a	0.13a
Plantation-06	83	0.00c	0.19c	244b	0.09b
SE	6.9	0.22	0.30	28.9	0.01
Species effects
A. caliginosa	81	0.06b	0.82	349a	0.08
A. trapezoides	90	0.69a	0.62	260b	0.10
SE	25.3	0.27	0.21	20.4	0.02

Means followed with different letter(s) are statistically different at p < 0.05.

Fig. 4

Trends observed in the changes in the mean weights of earthworms due to differences in the vegetation covers (a) and worm species (b). The capped bars are the least significance difference at p ≤ 0.05. The mean weight for individual earthworm at the beginning of the experiment was 0.41 ± 0.07 g for A. caliginosa and 0.66 ± 0.04 g for A. trapezoides.

More cocoons were produced by class="Species">earthworms in the soil from class="Chemical">pan class="Chemical">Plantation-04 than in soils from either grassland or woodland, while none was produced in soil from Plantation-06, after 11 weeks of inoculation (Table 3). During the same period, the earthworms in woodland soil produced the most casts, which was more than twice that produced in soil from Plantation-04, and several factors larger (>5) than cast produced in soils from the grassland or Plantation-06 (>8). Many more burrows were created in the soil from Plantation-04 or from grassland than in the soil from either Plantation-06 or woodland. A. trapezoides produced more cocoons and burrows than A. calignosa during the 11 weeks of the trial (Table 3).

At the end of the 11th week, the pH was similar in soils from grassland and class="Chemical">Plantation-04 (6.4 ± 0.2) comclass="Chemical">pared with soils from woodland or class="Chemical">pan class="Chemical">Plantation-06 (6.0 ± 0.1). Also the soil EC was higher under the two plantations than under the grassland and the woodland (Table 3). The mean weight of earthworms was significantly correlated with pH (r = 0.91), EC (r = 0.93), clay content (r = 0.75), sand (r = −0.99) and phosphorus content (r = −0.53), of the soil. As a result of these correlations, influences of variable EC and sand content on earthworms were explored further.

Soil textural and nutritional influence on exotic earthworms

Amending the clay soil from class="Chemical">Plantation-04 with sand and/or organic matter did not change the class="Chemical">pH, which averaged 5.6 ± 0.1 (data not class="Chemical">presented), but addition of organic matter significantly inclass="Chemical">pan class="Chemical">creased salinity (Table 4). The earthworms lost weight in unamended soil over the 4-week period, but gained weight in the amended soils. A. trapezoides lost more weight in unamended soil, and gained proportionally less weight in the amended soil, than A. gracilis. While no casts were produced by earthworms in the unamended soil, those in amended soil each produced about 1.0 g of cast over the 4-week period. Amendment of the clayey soil from Plantation-04 with sand reduced the bulk density (Table 4).

Table 4

Responses in selected soil properties and in the growth of earthworm species to amendments of soils from the woodland and Plantation-04 measured four weeks after inoculation.

Soil amendments	Electrical conductivity (dS m−1)	Bulk density (Mg/m3)	Initial weight/final weighta
			A. gracilis	A. trapezoides
Textural alterations of Plantation-04 soil
Control	0.15b	1.33a	0.90b	0.63b
Sand	0.15b	1.02b	4.03a	1.23a
Organic matter	0.21a	1.27ab	3.58a	1.23a
Sand + organic matter	0.22a	1.11b	3.56a	1.27a
SE	0.01	0.12	0.71	0.15
Nutritional alterations of woodland soil
Control	0.03	1.22	0.88b	0.99b
Fertiliser	0.07	1.23	0.99a	1.14a
SE	0.02	0.18	0.07	0.03

Means followed with different letter(s) are statistically different at p < 0.05.

Intial weights were 0.90 g for A. gracilis and 0.25 g for A. trapezoides.

Application of fertilizer to the loamy sand from the woodland reduced pH to 5.0 ± 0.1 from 5.4 ± 0.1 in the untreated soil; it also inclass="Chemical">creased salinity (Table 4). While class="Chemical">pan class="Species">A. gracilis did not lose weight in amended soil, it lost 12% of its mass in the untreated soil. By contrast, A. trapezoides maintained its weight in the unfertilized soil but gained 14% in weight with fertilizer application.

Discussion

Seasonal dynamics in earthworm populations

Seasonal abundance and diversity in class="Species">earthworm class="Chemical">poclass="Chemical">pulations were reflections of weather conditions and characteristics of the vegetation covers. class="Chemical">pan class="Species">Earthworm catches are sensitive to moisture and temperature conditions of the soil. The absence of catches in autumn was consistent with the relatively high soil temperature (>20 °C) at this time (Table 2), which was outside the optimum range of 10–15 °C for releasing the earthworms from dormancy and thrive (Baker and Whitby, 2003; Lee, 1985). The autumn in 2010 was also relatively dry with rainfall events being few and far between (Fig. 1) and this condition was only slightly alleviated by winter (June–August), which along with exceptionally low minimum air temperature, must have constrained earthworm hatches at this time. By spring however, the confluence of low temperatures and several rainfall events provided cool and moist soil conditions resulting in most earthworms being found at this time (Fig. 2a). These favourable conditions largely persisted into early summer in December, such that there was a mean decline of only 30% in earthworm abundance compared with spring.

Both the grassland and class="Chemical">Plantation-04 consistently maintained class="Chemical">parity or out-class="Chemical">performed the woodland in terms of class="Chemical">pan class="Species">earthworm abundance, but this could not be associated with soil fertility indices presented in Table 1. The soil C was lower in Plantation-6 than in either grassland or the woodland, while the woodland had lower N and CEC than all the other vegetation covers. Worm abundance was not even consistent with the differences in the C:N ratios of the soil that determines digestibility of surface litter as sources of nutrients (Ndegwa and Thompson, 2000). It was only in the woodland soil that the C:N of 22.5 (Table 1) was close to the optimum of 25 for earthworm viability and function; while the values for the soils in the grassland (11.7) and Plantation-04 (6.2) were far from ideal. This inconsistency between fertility indices and earthworm abundance across the four vegetation covers, however, was consistent with the earlier finding by Staaf (1987) that abundance and species composition of earthworms were not explained by the concentrations of lignin, N, or P or their combinations in the litter.

Differences amongst the vegetation covers in the class="Species">earthworm abundance were also not consistent with the trenclass="Chemical">pan class="Chemical">ds in the pH or salinity of the soil. Neither could they be associated with concentrations of heavy metals in the soils since high concentrations of Cr, Cu, Ni, Pb and Zn that are known to be injurious to earthworms (Muir et al., 2007; Spurgeon and Hopkin, 1999) were higher in the three revegetated covers than in the woodland (Table 1). For instance, the concentration of Zn, which is known to be especially toxic to earthworms (Muir et al., 2007; Spurgeon, Svendsen, Rimmer, Hopkin, & Weeks, 2000), was larger by a factor of 3–4 in the grassland soil than in the soils of the two plantations and the woodland.

The most likely explanation for differences in the abundance of class="Species">earthworms was the age of the vegetation covers. Since disruclass="Chemical">ption of the habitat is the class="Chemical">primary threat to class="Chemical">pan class="Species">earthworm viability, it is logical that the length of post rehabilitation period has direct bearing on the degree of recovery in the earthworm community. While the grassland and Plantation-04 exceeded woodland in restoring earthworm numbers, Plantation-06 had yet to attain this status due to its juvenile age (Fig. 3). This was consistent with an observation by Scullion et al. (1988) that a minimum of 5 years was needed for earthworms to recover on rehabilitated mine site. Without any significant perturbation no measurable difference in earthworm populations were found between grassland, woodland and forest in the savannah (Moïse, Seydou, Kolo, & Mamadou, 2011). However, species diversity was still poor in the plantations and grassland neither of which contained native earthworm species (Fig. 2b–d). The dominance of M. dubius and A. trapezoides, especially the former, in all the vegetation covers at this site was consistent with both species being widely distributed on cultivated landscapes in Australia, where they have largely displaced the native species (Chan and Barchia, 2007; Chan and Heenan, 2006; Hendrix et al., 2006), and on contaminated lands elsewhere (Spurgeon and Hopkin, 1999).

Vegetation type and earthworm viability

The absence of clear correlation between soil properties and class="Species">earthworm class="Chemical">poclass="Chemical">pulation amongst the four vegetation covers could be the result of large variability in soil class="Chemical">proclass="Chemical">perties commonly encountered in the field. Such variability is often not aclass="Chemical">pclass="Chemical">parent in soil class="Chemical">proclass="Chemical">perties obtained from bulked samclass="Chemical">ples such as those class="Chemical">presented in Table 1, but could be large enough to class="Chemical">pan class="Chemical">create discrete niches that are conducive for various macrofauna, including earthworms. Bulked samples, however, are still appropriate for testing how earthworm growth and reproduction potential respond to chemical and nutritive properties of individual soil type. It was therefore possible to ascribe the significant weight gains and retention by earthworms in the soil from Plantation-04 (Fig. 4a) to the higher fertility of this soil in which P and EC were higher than in the other soils (Table 1); these were also reflected in the differences in EC amongst the potted soils from the four vegetation covers following 11 weeks of inoculation (Table 3). Indeed we found significant correlation between weight gain with EC (r = 0.81) as with pH (r = 0.83). Although land preparation and establishment procedures were largely similar for both plantations and the grassland, substantial amounts of nutrients would have leached out over the past 16 years from the grassland sward; while Plantation-06 received less preparation, including fertilization, at establishment, and also lacked topsoil for several years afterwards (Yunusa et al., 2010).

Reasons for the subsequent loss in weight by the class="Species">earthworms were not quite clear. It is class="Chemical">possible, however, that key nutrients that were initially readily available were quickly exhausted, but less raclass="Chemical">pidly in the more fertile soil from class="Chemical">pan class="Chemical">Plantation-04. The large amounts of casts produced in the other three soils (Table 3) suggests greater foraging for nutrients that is often associated with increased burrowing, but known to impair weight gain (Bottinelli, Henry-des-Tureaux, Hallaire, Mathieu, & Benard, 2010). This was also consistent with the positive correlation between the mean weight of the earthworms and EC (r = 0.93) as a measure of soluble mineral nutrients in the soil, and with the weight gained by the earthworms when the woodland soil was amended with fertilizer or organic matter or both (Table 4). However, the subsequent weight loss could also be the result of the earthworms attaining maturity and onset of reproductive phase and cast formation. This critical phase could have been attained earlier in A. trapezoides than in A. caliginosa, consistent with the earlier onset of weight loss in the former (Fig. 4b). Thus A. trapezoides produced 10 times more cocoons than A. caliginosa (Table 3), with the former producing more than its initial mean weight in cocoons during the 11-week period. A response that could be associated with preference of A. trapezoides for compost and animal manure (used as food source in this study) compared with A. gracilis natural preference for decaying plant matter (Barois, 1992). Also, increases in earthworm weight in amended clayey soil could partly be due to the reductions in the bulk density as found in an earlier study (Hendrix et al., 2006; Reinecke, 1983).

In the present study pan class="Chemical">Plantation-04 effectively restored the class="Chemical">pan class="Species">earthworm population to preclearance levels. This vegetation system was also found in an earlier study to be quite effective in restoring hydrologic function to the site (Yunusa et al., 2010, 2012) and will therefore be preferred over the grassland cover for rehabilitating this and similar sites.

Conclusions

We initiated this study with the expectation that the choice of vegetation cover in rehabilitating waste disposal or mining sites would produce differential outcomes in terms of soil class="Species">biota. Our data, however, suggest that the differences in soil class="Chemical">proclass="Chemical">perties class="Chemical">pan class="Chemical">created by the contrasting vegetation covers were not large enough to impact earthworm abundance. Recovery of species diversity in the reclaimed vegetation covers was still well below that found in the old growth woodland. The grassland was devoid of native species even after 16 years suggesting that much longer time is needed for the return of these species to such environment. All the earthworms recovered in the woodland were endogeic, even the native species, and thus belong to the same functional group as the exotic species that dominated in the grassland and the plantations. Therefore the functions of earthworms in maintaining soil structure and nutrient cycling (Bottinelli et al., 2010; Lee, 1985) would have been largely restored to preclearance levels within six years of revegetation (Table 1); this process would not be any slower in grassland that had even a more rapid establishment than woody species. Ecological restoration of disrupted landscapes having duplex soil, such as used in this study, can be enhanced by mixing fine particles from the mineral clayey subsoil with the coarse particles from the loamy topsoil that is relatively high in organic carbon to promote earthworm establishment and growth, and general belowground biota.