Printer Friendly


Abstract. The consequences of permanent loss of species or species groups from plant communities are poorly understood, although there is increasing evidence that individual species effects are important in modifying ecosystem properties. We conducted a field experiment in a New Zealand perennial grassland ecosystem, creating artificial vegetation gaps and imposing manipulation treatments on the reestablishing vegetation. Treatments consisted of continual removal of different subsets or "functional groups" of the flora. We monitored vegetation and soil biotic and chemical properties over a 3-yr period. Plant competitive effects were clear: removal of the [C.sub.3] grass Lolium perenne L. enhanced vegetative cover, biomass, and species richness of both the [C.sub.4] grass and dicotyledonous weed functional groups and had either positive or negative effects on the legume Trifolium repens L., depending on season. Treatments significantly affected total plant cover and biomass; in particular, [C.sub.4] grass removal reduced total plant biomass in summer, because no other species had appropriate phenology. Removal of [C.sub.3] grasses reduced total root biomass and drastically enhanced overall shoot-to-root biomass ratios. Aboveground net primary productivity (NPP) was not strongly affected by any treatment, indicating strong compensatory effects between different functional components of the flora.

Removing all plants often negatively affected three further trophic levels of the decomposer functional food web: microflora, microbe-feeding nematodes, and predaceous nematodes. However, as long as plants were present, we did not find strong effects of removal treatments, NPP, or plant biomass on these trophic groupings, which instead were most closely related to spatial variation in soil chemical properties across all trophic levels, soil N in particular. Larger decomposer organisms, i.e., Collembola and earthworms, were unresponsive to any factor other than removal of all plants, which reduced their populations. We also considered five functional components of the soil biota at finer taxonomic levels: three decomposer components (microflora, microbe-feeding nematodes, predaceous nematodes) and two herbivore groups (nematodes and arthropods). Taxa within these five groups responded to removal treatments, indicating that plant community composition has multitrophic effects at higher levels of taxonomic resolution. The principal ordination axes summarizing community-level data for different trophic groups in the soil food web were related to each other in several instances, but the plant ordination axes were only significantly related to those of the soil microfloral community. There were time lag effects, with ordination axes of soil-associated herbivorous arthropods and microbial-feeding nematodes being related to ordination axes representing plant community structure at earlier measurement dates. Taxonomic diversity of some soil organism groups was linked to plant removals or to plant diversity. For herbivorous arthropods, removal of [C.sub.4] grasses enhanced diversity; there were negative correlations between plant and arthropod diversity, presumably because of negative influences of [C.sub.4] species in the most diverse treatments. There was evidence of lag relationships between diversity of plants and that of the three decomposer groups, indicating multitrophic effects of altering plant diversity.

Relatively small effects of plant removal on the decomposer food web were also apparent in soil processes regulated by this food web. Decomposition rates of substrates added to soils showed no relationship with treatment, and rates of [CO.sub.2] evolution from the soil were only adversely affected when all plants were removed. Few plant functional-group effects on soil nutrient dynamics were identified. Although some treatments affected temporal variability (and thus stability) of soil biotic properties (particularly [CO.sub.2] release) throughout the experiment, there was no evidence of destabilizing effects of plant removals.

Our data provide evidence that permanent exclusion of plant species from the species pool can have important consequences for overall vegetation composition in addition to the direct effects of vegetation removal, and various potential effects on both the above- and belowground subsystems. The nature of many of these effects is driven by which plant species are lost from the system, which depends on the various attributes or traits of these species.

Key words: biodiversity; community structure; competition; decomposition; ecosystem properties; functional groups; plant traits; removal experiments; soil food web; stability.


A major issue in ecological research involves the effects of individual species and groups of species on community-level and ecosystem-level attributes (Hobbie 1992, Lawton 1994). It is becoming increasingly appreciated that invasions of ecosystems by new species have profound effects on ecosystem function (Vitousek and Walker 1989, Vitousek et al. 1997). In addition, anthropogenic activity is causing rapid loss of species from many of Earth's ecosystems (Ehrlich and Ehrlich 1981, E. O. Wilson 1988), raising the issue of how loss of species alters community- and ecosystem-level properties and processes (Schulze and Mooney 1993). From an applied perspective, managed, production-oriented systems have frequently (although not always) involved maintenance of biological communities with an artificially low plant and herbivore species diversity, and recent focus has been on the ecological consequences of deliberate addition of more species (with corresponding shifts in species composition) in both agricultural (Vandermeer 1988, Giller et al. 1997) and forested (Wormald 1992) landscapes.

When a plant species is lost from a given community, often there are important consequences for the remaining floristic components of that community, especially if that species is potentially dominant (Gross 1980, Fowler 1981, Hils and Vanket 1982, Aarsson and Epp 1990). The positive effects of the loss of one plant species on the other plant species often stem from the competitive effects of that species (J. B. Wilson 1988), although, in some instances, loss of a species can also have negative consequences for the remaining flora (Calloway 1995). Studies of the loss of a dominant species from an ecosystem (e.g., Abdul-Fatih and Bazzaz 1979, Armesto and Pickett 1985, Gurevitch and Unnasch 1989, Wardle and Barker 1997) provide indirect evidence that dominant species may suppress overall diversity by preventing establishment of other species. Removal of selected species can have negative effects on total plant biomass and productivity, especially if species have separate niches and demonstrate complementary resource use. When niche overlap is high and competition for resources is intense, loss of a given species can be largely compensated for by the other species, resulting in overall biomass and productivity being less sensitive to species removal (Lawton and Brown 1993, Hooper and Vitousek 1997, Hooper 1998).

Although most studies involving manipulations of plant communities have focused on floristic changes following removal of a subset of the flora, a more complete picture of the effects of losses of plant species in an ecosystem can be gained by simultaneously considering the response of other trophic levels, e.g., by utilizing a food web-based approach (Wiegert and Owen 1971, Schoener 1989, Pastor et al. 1993, Wardle 1995, Wootton et al. 1996). In particular, the decomposer food web is a critical component of most ecosystems because it is directly responsible for regulating key ecosystem processes such as decomposition and nutrient mineralization. Formal trophic dynamic models using prey-dependent predator-prey interactions (e.g., Rosenzweig 1971, Oksanen et al. 1981) and studies presenting evidence for the existence of trophic cascades (Carpenter and Kitchell 1988) indicate that biomasses of some trophic levels are likely to respond to alteration of basal resources (e.g., through changes in net primary production due to removal of a plant species) whereas others are not because their biomasses are regulated by higher trophic levels. There is recent evidence from soil food webs that the biomass of different trophic levels responds rather uniformly to the amounts of basal resources added (Mikola and Setala 1998b), although little is known about how changes in the composition of basal resources (e.g., due to changes in plant community structure) affect higher trophic levels in soil food webs. Further, at finer levels of taxonomic resolution, shifts in species composition and diversity at the basal (plant) trophic level have the potential to induce corresponding changes in higher trophic levels of soil food webs (Griffiths et al. 1992, J. Zak et al. 1994), although this issue remains largely unexplored.

Understanding soil food web structure at the level of functional groups provides clues to a better understanding of ecosystem function (Wardle et al. 1998b) because of the role of the structure and dynamics of soil food webs in influencing decomposition and mineralization processes (Bengtsson et al. 1996, Mikola and Setala 1998a), and, ultimately, ecosystem productivity (Ingham et al. 1985, Setala 1995). However, at finer levels of taxonomic resolution, little is known about whether changes in taxonomic composition of organisms within trophic levels are important at the ecosystem level, and the available evidence is ambiguous (Andren et al. 1995, Mikola and Setala 1998c). Further, there is indirect evidence of important feedbacks existing between the structure of plant and soil communities (e.g., Bever 1994, Bever et al. 1997). Effects of removal of plant species on decomposer food webs therefore have the potential to influence ecosystem properties by altering those processes that soil food webs regulate.

In the present investigation, we sought to determine the effects of plant composition on community- and ecosystem-level attributes in a perennial grassland in New Zealand, both aboveground and belowground. We achieved this by performing experimental removals of different plant functional groups in small field plots over a 3-yr period. Specifically, our intention was to test each of the following four hypotheses: (1) removal of subsets of the existing flora results in increased diversity, biomass, and productivity of the remaining flora, but the total (overall) biomass and productivity is reduced because of partially compensatory interactions between floristic components; (2) modification of the producer trophic level has important effects on some trophic levels, but not others, because of the interplay of top-down and bottom-up forces in regulating soil food webs; (3) shifts in plant community-level attributes such as composition and diversity are important determinants of composition and diversity at higher trophic levels in soil food webs; and (4) removal of plant functional groups, representing permanent exclusion of plant species from the species pool, has important consequences for ecosystem-level processes and properties such as primary production, biomass, decomposer activity, nutrient levels, and ecosystem stability. The ultimate goal of our study was to determine the general consequences of removal of floristic components, and the resultant shifts in composition, in order to understand how species losses are manifested at both the community and ecosystem levels.


System and approach

Over the past few centuries, ~75% of New Zealand's forest vegetation cover has been cleared, and much of this land is currently under perennial grassland, which is primarily managed for livestock production. In the Waikato district of New Zealand's North Island, the majority of grazed land is utilized for managing dairy cattle. Here, the grasslands are dominated by introduced perennial herbaceous species, mainly the [C.sub.3] grass Lolium perenne L. and the legume Trifolium repens L., with significant levels of winter-annual [C.sub.3] grasses (principally Poa annua L.), summer-annual [C.sub.4] grasses, and several short-lived, dicotyledonous weed species.

In grazed perennial grasslands in New Zealand, plant regeneration is mainly concentrated in disturbed or denuded patches or gaps resulting from disturbance such as cattle feeding and treading, summer droughting, and invertebrate infestations; these gaps can be up to 20 cm in diameter (Panetta and Wardle 1992). Establishment of plant seedlings usually occurs in these gaps, and it is the processes and biotic interactions occurring in these gaps that ultimately contribute to the species composition of the pasture (Panetta and Wardle 1992). This mechanism is widespread throughout temperate perennial grasslands, including both Europe (Bullock et al. 1994, 1995) and North America (Platt and Weiss 1977, Goldberg and Werner 1983, Kotanen 1997). The net result is a mosaic of patches of different ages and varying species composition.

We used vegetation gaps as our experimental unit to investigate how removals of plant functional groups affect the ecological properties of grasslands at the spatial scale of pasture gaps. We created artificial, devegetated gaps and allowed the vegetation to recolonize over time, with the imposition of treatments consisting of continual removal of different plant functional groups. This approach is conceptually based on the "removal experiment" approach that is often used to study plant competition (Harper 1977, Aarssen and Epp 1990, Herbert et al. 1997).

Site and experimental setup

The experimental site was a perennial grassland, dominated by herbaceous plant species of mainly European origin, and grazed by dairy cattle, near Hamilton in the Waikato district (37 [degrees] 45' S, 175 [degrees] 15' E). It was grazed over an approximately monthly rotation to a height of 4-6 cm, typical of many grazing systems in New Zealand. Consequently, competitive interactions above ground are probably much less important than those below ground (see J. B. Wilson 1988). The climate is warm temperate, and the site has a long-term mean annual rainfall of 1204 mm/yr (Fig. 1). The soil present is a Horotiu silt loam (Vitric Hapludand) and at the start of the experiment had 13.0% C, 0.06% N, and a pH of 5.1 at 0-7.5 cm depth.


The experimental site occupied an area of 0.5 ha, which was subdivided into 10 blocks of 0.05 ha each that served as the units of replication. Within each block, six 0.8 x 1.2 m "plots" were set up. Within each plot, six subplots (hereafter referred to as "gaps") were set up; these gaps were circular, 20 cm in diameter, and arranged in a 3 x 2 grid in the plot so that all gaps were [is greater than or equal to] 20 cm apart. The size of gap that we selected has been previously shown to be appropriate for investigating biotic interactions in model plant communities with similar species compositions and soils (Wardle and Barker 1997). Permanent wooden pegs were placed in the corners of each plot; a portable steel frame was constructed that fitted over the entire plot (and around the pegs) and contained six 20 cm diameter circles matching exactly the location of the six gaps in the plot. This frame enabled relocation of each gap with considerable accuracy. In addition to these gaps, one further gap was established in each replicate block for destructive harvesting to provide baseline (t = 0) data.

The gaps were created by killing all of the vegetation within them. This was achieved by spraying each gap separately with two nonresidual herbicides, glyphosate (formulated as Round-up) at 2 kg ai/ha, and tribenuron (formulated as Granstar) at 30 g ai/ha, on 27 January 1994. To prevent the herbicide from affecting vegetation outside each gap, a large plastic funnel with a bottom inside diameter of 20 cm was pressed to the ground when the herbicide was applied to that gap. All dead vegetation and emerging seedlings were removed regularly from each gap until the start of the experiment at 1 March 1994.

The six gaps within each plot were randomly assigned to six different treatments. These treatments involved removal of different plant functional groups, and were maintained by regular (2-4 weekly) manual hand-weeding of unwanted seedlings as they emerged during the entire course of the experiment. The treatments were:

1) Removal of all plants. This was intended to serve as a baseline treatment for assessing soil biological and chemical properties.

2) Removal of all [C.sub.4] grasses (occupying 0-57% of total plant biomass, depending on season and year). Over 96% of the [C.sub.4] grass biomass consisted of summer-annual species, namely Panicum dichotomiflorum Michx. and Digitaria sanguinalis (L.) Scop.

3) Removal of all [C.sub.3] annual grasses (occupying 0-49% of total biomass). Practically all of the biomass consisted of the winter-annual P. annua.

4) Removal of all [C.sub.3] grasses (occupying 22-53% of total biomass). Over 98% of the biomass consists of P. annua and L. perenne. This treatment was intended to correspond to treatment (2) by enabling comparison of [C.sub.3] and [C.sub.4] grass removal effects (cf. Wedin and Tilman 1993, Campbell et al. 1996).

5) Removal of dicotyledonous weeds (occupying 5-25% of total biomass). These are essentially short-lived, rapidly growing ruderal species, and represent the majority of plant species in these grasslands. The split between monocotyledonous and dicotyledonous plant species is recognized as a fundamental functional split because of the clear separation of niches and strategies of the two sets of species (Wilson and Roxburgh 1994, Wardle and Barker 1997).

6) Removal of no plants.

All plant species in the study site and encountered during the investigation are listed in Table 1. It is assumed for the purposes of this study that most of the effects observed in each gap (both aboveground and belowground) can be attributed to the plant species occupying these gaps. In this context, edge effects of the surrounding vegetation are assumed to be of minor importance because the plants are usually maintained at a height of [is less than] 6 cm, and because root biomass in the plant-free gaps was very low compared to that in the others (see Results). Effectively, the plant-free gaps operated as "blanks" against which other treatment effects could be assessed.
TABLE 1. List of plant species encountered during the experimental
period, their functional groupings, and families to which they
belong (in parentheses). Bold type denotes the dominant species in
functional groups.

     Functional group                        Species

[C.sub.4] grasses             Digitaria sanguinalis (L.) Scop.,
                                Panicum dichotomiflorum Michx.,
                                Paspalum dilatatum Poir., P.
                                distichum L. (all Gramineae)
[C.sub.3] annual grasses      Poa trivialis L., Poa annua L.
                                (all Gramineae)
[C.sub.3] perennial grasses   Bromus wildenowii Kunth, Dactylis
                                glomerata L., Lolium perenne L.
                                (all Gramineae)
Clovers                       Trifolium repens L. (Leguminosae)
Dicotyledonous weeds          Achillea millefolium L. (Asteraceae),
                                Amaranthus hybridus L. (Amaranth-
                                eraceae), Anagallis arvensis L.
                                (Primulaceae), Aphanes sp.
                                (Rosaceae), Bellis perennis L.
                                (Asteraceae), Capsella bursa-
                                pastoris (L.). Med. (Brassicaceae),
                                Cerastium glomeratum Thiull.
                                (Caryophyllaceae), Chenopodium
                                album agg. (Chenopodiaceae), Conyza
                                canadensis (L.) Cronq. (Aste-
                                raceae), Coronopus didymus (L.) Sm.
                                (Brassicaceae), Crepis capillaris
                                (L.) Wallr. (Asteraceae), Daucus
                                carota L. (Umbelliferae), Geranium
                                molle L. (Geraniaceae), Gnaphalium
                                spicatum Lam. (Asteraceae),
                                Hypochaeris radicata L. (Aste-
                                raceae), Leontodon taraxacoides
                                (Vill.) Merat (Asteraceae), Modiola
                                caroliniana (L.) G. Don (Malva-
                                ceae), Montia sp. (Portulacaceae),
                                Oxalis sp. (Oxalidaceae), Plantago
                                lanceolata L. (Plantaginaceae), P.
                                major L. (Plantaginaceae), Poly-
                                gonum aviculare agg. (Polygon-
                                aceae), Portulaca oleracea L.
                                (Portulacaceae), Prunella vulgaris
                                L. (Labiatae), Ranunculus repens L.
                                (Ranunculaceae), Rumex obtusifolius
                                L. (Polygonaceae), R. pulcher L.
                                (Polygonaceae), Sagina procumbens
                                L. (Caryophyllaceae), Sonchus asper
                                (L.) Hill (Asteraceae), Spergula
                                arvensis L. (Caryophyllaceae),
                                Stellaria media (L.) rill.
                                (Caryophyllaceae), Taraxacum
                                officinale Weber (Asteraceae),
                                Veronica arvensis L.

Over the course of the experiment, the study area was subjected to periodic light grazing (with a mean stocking rate of 2.2 cows/ha) and a grazing rotation length of 1 mo.

Vegetation assessment and harvest

In addition to the t = 0 harvest on 1 March 1994, destructive harvesting took place on six separate occasions (14 September 1994, 7 March 1995, 19 September 1995, 5 March 1996, 17 September 1996, and 5 March 1997); these dates were selected because summer-annual plant biomass at the study area is maximal in early March and winter-annual biomass is maximal in mid-September. For each destructive harvest, all of the gaps from one randomly selected plot in each block were selected and processed. For each of those gaps harvested at the end of the experiment (on 5 March 1997), the vegetation composition was nondestructively assessed monthly during the entire course of the experiment. Cattle were excluded from the entire experiment for [is greater than or equal to] 1 wk prior to each assessment date. Measurements on each gap were performed by using point quadrat analysis (Goodall, 1952, Jonasson 1983, as described by Wardle et al. 1995a) to determine the total ground cover of each component plant species present. We used a pointing frame with points 2 cm apart, and took measurements in a grid pattern so that 50 points were measured for each gap at each sampling; we recorded the total number of intercepts for each species. For each gap subjected to the all-removal treatment and harvested on 5 March 1997, all emerging seedlings weeded from the gap during the entire experiment were identified to species level and counted.

Exactly 2 wk prior to each destructive harvest, the vegetation present in each gap was trimmed to 1 cm height, and cattle were then excluded until harvest. On the harvest date (by which time the vegetation had grown to a height of 4-6 cm), the vegetation was again trimmed to 1 cm height and the clipped material was collected to enable determination of net aboveground production over the 2-wk period (Wardle et al. 1994). The remaining vegetation was then trimmed at ground level; the sum of this material and that clipped for productivity assessment for each gap represented the total standing aboveground plant biomass at harvest (i.e., following two weeks' growth since trimming). All clipped material (i.e., both 0-1 cm and [is greater than] 1 cm height) was hand-sorted in the laboratory into the component species, and was oven-dried (80 [degrees] C, 24 h) for productivity and biomass determinations of each species.

After vegetation removal, two intact soil cores (each 2.5 cm diameter x 7.5 cm depth) were collected from each gap for soil microarthropod determinations. Then the soil in the entire gap was sampled to 7.5 cm depth using a 20 cm diameter corer; this soil was immediately taken to the laboratory and processed.

Belowground assessments

The intact cores sampled for microarthropod measurements were placed into a Tullgren invertebrate extractor (Merchant and Crossley 1970, as described by Wardle et al. 1995b), and were left for ~4 d, after which the arthropods in the collection fluid were counted and classified into feeding groups. The total soil sample from each gap (i.e., that making up the 20 cm diameter core) was carefully hand-sorted within 24 h of collection, and all larger invertebrates (mainly insects and earthworms) present were enumerated and identified to species level. All live roots were then removed from the sample and hand-sorted into three root diameter classes, i.e., [is less than] 1 mm, 1-3 mm, and [is greater than]3 mm, and oven-dried for mass determination.

The soil was then subsampled and 250 mL of soil was used for nematode determination by using a tray variant of the Baermann method (Southey 1986, Yeates et al. 1993b). Total nematodes were counted under a stereomicroscope at 40x and then were killed and fixed with double-strength FA 4:1 (100 mL 40% formaldehyde: 10 mL glacial acetic acid: 390 mL distilled water). Subsequently, an average of 126 specimens from each sample was identified to nominal genus or family level and placed into six functional groupings occupying three trophic levels, following Yeates et al. (1993a, b). These trophic levels (with component functional groups in brackets) are: microbefeeders (bacterial feeders and fungal feeders); predators (top predators and "omnivores," i.e., predators that feed at more than one trophic level), and herbivores (plant parasites and plant associates).

The remaining soil was sieved (mesh size 4 mm) for soil microbial and chemical measurements. Microbial basal respiration was determined as described by Wardle et al. (1993). This involved adjusting 15 g (dry mass) soil to 55% moisture content (on a dry mass basis) by either gradual air-drying or rewetting with a fine mist, placing it in 169-mL airtight incubation vessels, and incubating at 22 [degrees] C. [CO.sub.2]-C evolution between 1 h and 4 h of incubation was then determined by injecting 1-mL subsamples of headspace gas into an infrared gas analyzer. Substrate-induced respiration (SIR; a relative measure of active microbial biomass) was determined using the approach of Anderson and Domsch (1978), as modified by West and Sparling (1986) and Wardle et al. (1993). This was performed as for basal respiration, but with an amendment of 6000 [micro]g glucose/g soil at the beginning of the incubation. The ratio of basal respiration to SIR was used as a relative measure of the microbial metabolic quotient, inversely related to microbial efficiency (Anderson and Domsch 1985). This quotient is based upon Odum's theory of ecosystem succession (Odum 1969), and is elevated in relatively disturbed or stressed situations in which the microbial biomass allocates most of the available C resources to respiration (C loss) rather than to biomass growth and maintenance. For obtaining relative measures of active bacterial and fungal biomass, the selective inhibition technique of Anderson and Domsch (1975), as modified by West (1986) and Wardle et al. (1993), was used. This was performed as for the SIR measurements, except that when the glucose was added, 10 000 [micro]g/g of the bacterial inhibitor streptomycin sulphate was also added for bacterial assessment, and 15000 [micro]g/g of the fungal inhibitor cycloheximide was added for fungal assessment.

Relative total microbial biomass was determined by the fumigation-incubation method of Jenkinson and Powlson (1976), as described by Wardle et al. (1993). Briefly, this involved fumigating (with chloroform) one 15 g (dry mass) subsample amended to 55% moisture content, and leaving another subsample (16.67 g dry mass, 55% moisture) unfumigated, reinoculating the fumigated sample with 1.67 g dry mass soil, and incubating both subsamples for 10 d at 22 [degrees] C in 500-mL airtight containers, each with a vial of 20 mL 1.0 mol/L NaOH. Total [CO.sub.2]-C released in each container was assessed by titrating this NaOH against 0.5 mol/L HCl. The difference in [CO.sub.2]-C release between the fumigated and nonfumigated soil is assumed to be proportionally related to the total (chloroform-susceptible) microbial biomass (Jenkinson and Powlson 1976).

For each sample collected on the final sampling date (5 March 1997), the microbial community structure was characterized by determining the phospholipid fatty acid (PLFA) composition. Lipids were extracted from soil using the method of Bligh and Dyer (1959), as modified by White et al. (1979) and described by Bardgett et al. (1996). Briefly, 1.5 g (fresh mass) subsamples were repeatedly extracted in a one-phase mixture of chloroform, methanol, and citrate buffer (1:2: 0.8 by volume). Extracts were split into two phases by adding chloroform and buffer; the lower lipid-containing phase was transferred to a test tube, dried under a stream of nitrogen, and stored at -20 [degrees] C. Lipids were dissolved in chloroform and fractionated on glass columns containing activated silicic acid. The neutral lipids and glycolipids were eluted with chloroform and acetone, respectively, and the phospholipids were eluted with methanol. The phospholipid fraction was dried under a stream of nitrogen and stored at -20 [degrees] C until preparation of fatty acid methyl esters. Samples were then dissolved in a methanol-toluene mixture and subjected to a mild alkaline methanolysis (Dowling et al. 1986). Resulting fatty acid methyl esters were analyzed and quantified using a Hewlett Packard 5890 II gas chromatograph (GC) equipped with a 5972A mass-selective detector (MDSII; Hewlett Packard, San Diego, California, USA), together with appropriate standards. Fatty acid nomenclature used was as described by Frostegard et al. (1993a, b). The abundance of the fatty acid 18:2[Omega]6 was used as a measure of the phospholipids derived from fungi, because it is found only in eukaryotes and is abundant in a range of fungal species (Federle 1986). The other main microbial lipids identified are generally specific to subsets of the bacterial component of the soil microflora. The ratio of phospholipid 18:2 [Omega] 6 to the sum of bacterial phospholipids was used as a relative measure of fungal to bacterial biomass (Bardgett et al. 1996).

For each of the soil samples collected on 5 March 1997, decomposition potential was also measured, essentially as described by Wardle et al. (1998a). A subsample of soil was amended to 55% moisture content (dry mass basis) and was used to fill a 9 cm diameter petri dish, which was then sealed and left to equilibrate at 22 [degrees] C for 3 d. After this time, a pre-weighed strip of cellulose filter paper (~0.2 g) was inserted into the soil in each dish; the dish was again resealed and incubated at 22 [degrees] C for 4 wk. After this time, the filter paper was removed and cleaned, and the oven-dry dry mass (80 [degrees] C, 2 h) remaining was determined.

For each soil sample collected at each date, the following chemical analyses were determined: soil pH; total C concentration through loss on ignition; total N through Kjeldahl analysis; total P as described by Jackson (1958); bicarbonate-extractable P by the Olsen P method; and ammonium and nitrate contents through Technicon auto-analysis following extraction with 2 mol/L KCl. In addition, water-soluble C concentrations were determined as described by Wardle and Ghani (1995); this involved extracting 10 g (dry mass) soil in 20 mL water and centrifuging this for 20 min, followed by filtration, dichromate oxidation, and titration with [Fe.sup.2+].

Data analysis

Responses of plant and soil variables to the plant removal treatments were analysed by using ANOVA, following appropriate transformations; least significant difference (LSD) and standard error of the mean (SEM) values were derived from these for expressing data variability and comparisons of treatments. Temporal variability of selected vegetation and soil properties across the six harvest dates (incorporating variation due to both inter-year climate differences and season) was determined by calculating coefficient of variation (CV) values across times (cf. Tilman 1996) for each treatment x block combination, with the 10 blocks serving as the units of replication. Spatial variability was calculated as the CV across replicate blocks for each treatment x sampling time combination, with sampling time as the unit of replication; this was permissible because different, randomly selected plots were harvested at different dates. Correlation analyses were used to test for relationships between soil organism variables and plant and soil variables (usually within sampling times), and, in most cases, the plant-free gaps (all-plant removal treatment) were excluded because inclusion of zero values for plant-associated variables for these gaps would have violated assumptions about normality of data. Community-level data (i.e., matrices of taxa x gaps) for each of six groups (plants, microbes [PLFA data], microbe-feeding nematodes, predaceous nematodes, herbivorous nematodes, and herbivorous arthropods) were summarized by the use of the ordination technique detrended correspondence analysis (Hill and Gauch 1980), but by using detrending by polynomials, rather than by segments, to enhance robustness (Ter Braak 1987). Ordinations were performed by the CANOCO package, version 3.15 (Ter Braak 1987, 1988), and the principal ordination axes were related to treatments by ANOVA and to the ordination axes of the other trophic levels through correlation analysis.


Vegetation dynamics

In the devegetated gaps, seedlings emerged continuously over the 3-yr period (Fig. 2). Seedlings of four [C.sub.4] grass species appeared during the experiment, with most of those in the 1994-1995 summer consisting of Panicum dichotomiflorum and, in the subsequent summers, of Digitaria sanguinalis. Large numbers of perennial [C.sub.3] grass seedlings (mostly Lolium perenne with traces of Bromus wildenowii Kunth and Dactylis glomerata L.) emerged in the first few months, with very few appearing thereafter. Appreciable numbers of annual [C.sub.3] grass seedlings (Poa annua with traces of Poa trivialis L.) emerged each winter. Dicotyledonous weed seedlings emerged in most months, with maximal values during the winter; 33 species were identified, with no clearly dominant species and with the four most abundant species (Fig. 2) collectively accounting for less than half of the seedlings present. Low levels of Trifolium repens emerged throughout the experiment. It is assumed that these seedling measurements represent an approximate measure of the potentially available pool of seedlings available for recruitment into gaps during the course of the experiment.


The [C.sub.4] grasses responded strongly to the various plant removal treatments, especially in the first and third years (Fig. 3, Table 2). The strongest effects resulted from removal of the [C.sub.3] grasses (L. perenne and P. annua in combination) on P. dichotomiflorum cover, total [C.sub.4] grass cover, and [C.sub.4] grass species richness during the 1994-1995 and 1996-1997 summers. These effects appear to be caused by L. perenne, because removal of [C.sub.3] annual grasses generally did not have much effect. There were no consistent treatment effects on D. sanguinalis.

TABLE 2. Results of ANOVA testing for significance of treatment
(species removal) effects on various ground cover and species
richness parameters, corresponding to the data depicted in
Figs. 3-7.

                                                        Year 1

 Plant group        Dependent variable       df       F       P

[C.sub.4]        P. dichotomiflorum cover   3, 27    4.31    0.013
  grasses        D. sanguinalis cover       3, 27    1.14    0.352
                 Total cover                3, 27    4.73    0.008
                 Total species richness     3, 27    6.05    0.003
[C.sub.3]        L. perenne cover           3, 27    1.49    0.239
  grasses        P. annua cover             2, 18    0.47    0.636
Dicotyledonous   Total cover                3, 27    5.18    0.006
  weeds          Total species richness     3, 27    3.20    0.039
Clovers          T. repens cover            4, 36    6.93   <0.001
All plants       Total cover                4, 36    7.26   <0.001
                 Total species richness     4, 36   16.68   <0.001

                                               Year 2

 Plant group        Dependent variable       F       P

[C.sub.4]        P. dichotomiflorum cover   1.13    0.353
  grasses        D. sanguinalis cover       1.78    0.178
                 Total cover                1.73    0.183
                 Total species richness     0.97    0.420
[C.sub.3]        L. perenne cover           0.08    0.969
  grasses        P. annua cover             0.56    0.582
Dicotyledonous   Total cover                8.18    0.001
  weeds          Total species richness     4.72    0.009
Clovers          T. repens cover            0.53    0.717
All plants       Total cover                2.83    0.039
                 Total species richness     6.89   <0.001

                                                Year 3

 Plant group        Dependent variable        F       P

[C.sub.4]        P. dichotomiflorum cover    3.06    0.045
  grasses        D. sanguinalis cover        0.98    0.418
                 Total cover                 5.43    0.005
                 Total species richness      5.41    0.005
[C.sub.3]        L. perenne cover            0.84    0.482
  grasses        P. annua cover              1.15    0.339
Dicotyledonous   Total cover                 2.95    0.051
  weeds          Total species richness      5.79    0.003
Clovers          T. repens cover             0.49    0.741
All plants       Total cover                 6.21    0.001
                 Total species richness     20.70   <0.001

Notes: Analyses are performed separately for each of the three
years, with each data value consisting of cover or species richness
values for each gap, averaged over all of the measurement events
for that year, and with the 10 replicate blocks serving as the
units of replication.

None of the plant removal treatments had significant effects on L. perenne cover when data were analyzed on an annual basis (Table 2), but there were ephemeral significant effects for some sampling dates over the first 12 mo (Fig. 4), with the highest L. perenne cover sometimes resulting from removal of dicotyledonous weeds and the lowest cover usually occurring in the plots from which [C.sub.4] grasses had been removed. There were no detectable treatment responses for P. annua (Table 2, Fig. 4), but there were important inter-year differences, with a much greater cover by P. annua in the first year than in the other two.


None of the dicotyledonous weed species present in our study showed a statistically significant response to the treatments that we imposed (P [is greater than] 0.20 for each species when data were pooled for each year). When total dicotyledonous plant biomass was considered, however, there were important treatment effects (Table 2, Fig. 5), which were due to the very strong positive response to the [C.sub.3] grass removal treatment, and were presumably caused by the removal of competition from L. perenne. These effects of L. perenne reduced not only the total cover of dicotyledonous weeds, but also their species richness, particularly during the cooler months.


When the cover of T. repens was analyzed on an annual basis, there were significant treatment effects only in the first year (Table 2), with greatest cover in the plots in which all [C.sub.3] grasses had been removed (Fig. 6). However, in the subsequent two years, there were also shorter term significant effects of the [C.sub.3] grass removal treatments, with both enhancement of T. repens by L. perenne removal in the cooler months, and reduction of T. repens by L. perenne removal in the warmer months (particularly during the summer of 1995-1996). Dynamics of T. repens cover in the [C.sub.3] grass removal treatment were clearly out of synchrony with those in the other four treatments, which generally did not differ much from each other.


The removal treatments had statistically significant effects on total plant cover (Table 2), but these effects were typically small (Fig. 7); the greatest negative effects were caused by the removal of all [C.sub.3] grasses (occupying 22-53% of the total cover in the control treatment). Total cover in the treatment with all of the plant functional groups present generally did not differ much from the [C.sub.3] annual grass removal, [C.sub.4] grass removal, and dicotyledonous weed removal treatments. However, in the late summer (February-March) of each of the three years (i.e., when [C.sub.4] grass biomass is normally maximal), there was a brief period in which plant cover was least for the [C.sub.4] grass removal treatments. There were large increases in total plant cover, reflecting high productivity, in the first three months of the study. Total species richness differed significantly between treatments (Fig. 7), with the lowest richness occurring in the gaps in which the dicotyledonous weeds (i.e., by far the most species-rich functional group) had been removed.


The plant aboveground biomass and productivity data showed strong seasonality (Fig. 8). For the [C.sub.4] grasses, both variables were generally greatest for the [C.sub.3] grass removal treatments each summer, whereas the [C.sub.3] grasses did not show consistent responses to any treatments. For the majority of sample dates, dicotyledonous weed biomass and productivity were greatest when [C.sub.3] grasses had been removed. For T. repens, biomass and productivity were usually greatest in the [C.sub.3] removal treatment each September, but T. repens biomass was generally less in this treatment than in the other treatments each March. There were no consistent effects of removal treatments on total aboveground biomass or net primary productivity, but there were some important inter-year differences. Production of [C.sub.3] perennial and annual grasses and biomass of annual [C.sub.3] grasses were highest in the first year, with declining amounts in the second and third years.


Total root biomass was consistently less in the [C.sub.3] removal treatments than in the other treatments (Fig. 9) and, consequently, the shoot-to-root mass ratio was greater in this treatment than in the others. The shoot-to-root ratio was often less in the [C.sub.4] grass removal treatments than in the other treatments, particularly for the March samplings. Root biomass in the gaps with all plants removed was usually [is less than] 10% of that in the other treatments.


The species removal treatments resulted in a corresponding reduction in various measures of species richness calculated for the biomass data, but only for some sampling occasions (Fig. 10). Removal of dicotyledonous weeds resulted in reduced species richness relative to the other treatments throughout the experiment; during the March samplings, the same was also true of the [C.sub.4] grass removal treatment. When diversity was assessed on a "functional group" basis (Fig. 10), the

[C.sub.3] grass removal treatment resulted in the lowest diversity, but the other removal treatments induced only relatively small reductions in diversity.


Removal of plant functional groups did not consistently alter temporal and spatial variability of either plant biomass or productivity measures (Table 3). In relation to temporal variability (encompassing both inter-year and inter-seasonal variability across all six sampling occasions for each replicate block), total plant biomass varied least in the [C.sub.3] annual grass removal treatment, indicating that removal of the [C.sub.3] winter-annual P. annua from the species pool had stabilizing effects. Spatial variability of both aboveground biomass and primary productivity over the 10 replicate blocks was significantly greater in the [C.sub.3] grass removal treatment than in at least some of the other treatments.
TABLE 3. Temporal and spatial variability for plant biomass and
production variables.

                                                Treatment by
                                                 removal of:

                                            [C.sub.4]   perennial
    Measurement        Response variable     grasses     grasses

Temporal variability   Total aboveground
  ([dagger])             biomass              0.402       0.282
                       NPP([sections])        0.598       0.450
                       Total root biomass     0.432       0.524
Spatial variability    Total aboveground
                         biomass              0.322       0.318
  ([double dagger])    NPP([sections])        0.427       0.418
                       Total root biomass     0.390       0.465

                                                Treatment by
                                                removal of:

                                            [C.sub.3]   Dicoty-
                                             annual     ledonous
    Measurement        Response variable     grasses     weeds

Temporal variability   Total aboveground
  ([dagger])             biomass              0.393      0.370
                       NPP([sections])        0.524      0.536
                       Total root biomass     0.522      0.465
Spatial variability    Total aboveground
                         biomass              0.467      0.354
  ([double dagger])    NPP([sections])        0.553      0.361
                       Total root biomass     0.484      0.459

                                                 Treatment by
                                                 removal of:

    Measurement        Response variable    Nothing   ([parallel])

Temporal variability   Total aboveground
  ([dagger])             biomass             0.359        0.094
                       NPP([sections])       0.493        0.159
                       Total root biomass    0.419        0.166
Spatial variability    Total aboveground
                         biomass             0.386        0.115
  ([double dagger])    NPP([sections])       0.414        0.153
                       Total root biomass    0.380        0.162

([dagger]) Expressed as the coefficient of variation (CV = mean/SD)
across sampling dates for each treatment for each replicate block.

([double dagger]) Expressed as the cv across blocks for each
treatment for each sampling time.

([sections]) Aboveground net primary productivity.

([parallel] The least significant difference at P = 0.05 after

Soil abiotic properties

Although the soil chemical characteristics that we measured generally showed both significant temporal and spatial variability, there was little evidence of any soil chemical response, even when all of the vegetation was removed (Table 4). The only treatment response that was significant at P = 0.05 was for total soil P, which over the course of the study was less in the [C.sub.3] grass removal treatment (mean P content over experiment = 0.182%) and the all-plant removal treatment (mean P content = 0.183%) than in the other treatments (range = 0.190-0.193 %). In addition, treatment effects on both water-soluble C and total N concentrations were marginally nonsignificant at P = 0.05, with the [C.sub.3] grass removal and all-plant removal treatments yielding lower values than the other treatments. For most of the soil chemical properties, there were important differences between the 10 replicate blocks (Table 4), as reflected by the relatively high values of the spatial cv (mean/sD) across blocks for some of the variables at the start of the experiment, i.e., soil C (6.7%), soluble C (23.5%), soil N (17.7%), mineral N (63.8%), soil P (19.2%), bicarbonate-extractable P (38.1%), and pH (30.1%). Soil moisture content differed significantly between treatments ([F.sub.2, 314] = 7.24, P [is less than] 0.001), because the plant-free treatment had a greater moisture content (mean across all dates = 42.9%, dry mass basis) than did the other treatments, which did not differ significantly from each other (mean for remaining treatments = 40.4% moisture).
TABLE 4. Summary of ANOVAs for soil chemical properties over the
entire experimental period, testing for plant removal treatment,
and sampling time and blocking effects.

                                 Plant removal
                                  effect (PL)

      Soil property         [F.sub.5, 314]     P

Total soil C                     1.17        0.323
Water-soluble C                  1.88        0.096
Total N                          1.87        0.098
Mineral N                        0.62        0.684
Total P                          2.36        0.040
Bicarbonate-extractable P        1.48        0.196
Soil pH                          1.30        0.262

                                Time effect (T)

      Soil property         [F.sub.5, 314]     P

Total soil C                    57.37        <0.001
Water-soluble C                 145.67       <0.001
Total N                         28.06        <0.001
Mineral N                       17.10        <0.001
Total P                         36.54        <0.001
Bicarbonate-extractable P       14.79        <0.001
Soil pH                         13.69        <0.001

                               PL x T interaction

      Soil property         [F.sub.25, 314]     P

Total soil C                     0.42         0.994
Water-soluble C                  0.69         0.871
Total N                          0.54         0.966
Mineral N                        1.33         0.139
Total P                          0.53         0.971
Bicarbonate-extractable P        0.57         0.976
Soil pH                          1.05         0.399

                               Blocking effect

      Soil property         [F.sub.9, 314]     P

Total soil C                    27.05        <0.001
Water-soluble C                  1.79         0.070
Total N                         29.60        <0.001
Mineral N                        1.67         0.094
Total P                         40.62        <0.001
Bicarbonate-extractable P       15.31        <0.001
Soil pH                          4.34        <0.001

Notes: Time effects could be tested in these analyses without risk
of problems associated with pseudoreplication (cf. Hurlbert 1984)
because different, randomly selected plots were harvested at
different sampling times. All data analyses of soil chemical
components are calculated on a concentration per unit soil mass

Soil microbial biomass and activity

Those variables based on soil microbial biomass and activity showed statistically significant treatment responses during most of the experimental period, particularly from the second year onward (Fig. 11). This was mostly due to the effects of the all-vegetation removal treatment, which caused detectable and consistent reductions in SIR (active biomass), flush of [CO.sub.2] after fumigation (total biomass), the SIR to organic C ratio, and basal respiration. The microbial metabolic quotient, inversely related to microbial efficiency, was weakly, although significantly, enhanced by the all-vegetation removal treatment. Both the bacterial and fungal components of the active biomass (i.e., those measured by selective inhibition) followed the same pattern as SIR (data not presented), although removal of all plants caused a slight (but significant) shift from bacteria toward fungi over the final two years of the experiment (Fig. 11). The other treatments did not differ much, although many of the biomass-related properties were slightly enhanced in the nonremoval treatment relative to the others at the end of the experiment (Fig. 11). There were some important overall temporal trends, with the bacterial to fungal ratio generally increasing in all treatments throughout the study. Decomposition of added cellulose in the soils collected on 5 March 1997 did not show a significant response to the removal treatments ([F.sub.5,41] = 1.96, P = 0.105; data not presented).


Our results show that there was no apparent relationship between microbial biomass (or related variables), and plant biomass or NPP (Table 5), meaning that altering these plant variables did not lead to a corresponding alteration of biomass in the next highest trophic level (primary saprophytes) or the processes it regulates (soil [CO.sub.2] evolution, decomposition) over the course of the experiment. Instead the strongest correlates were soil chemical properties such as soil moisture, C, N, and pH.
TABLE 5. Pearson's correlation coefficient values between selected
soil microbial properties and plant mass, diversity, and soil
abiotic properties for the final winter (17 September 1996) and
summer (5 March 1997) sampling dates.

                                          Plant variables

                          Sampling                       ground
  Microbial variable        date      NPP ([sections])   biomass

SIR([dagger])           17 Sep 1996        0.111         -0.136
                         5 Mar 1997        0.077          0.034

BR([double dagger])     17 Sep 1996        0.067         -0.025
                         5 Mar 1997        0.276          0.267

Bacteria: fungi         17 Sep 1996       -0.063         -0.110
                         5 Mar 1997       -0.052         -0.001

BR: SIR                 17 Sep 1996       -0.106          0.099
                         5 Mar 1997        0.179          0.180

Fumigation [CO.sub.2]   19 Sep 1996       -0.285(*)      -0.135
  flush                  5 Mar 1997        0.181          0.283(*)

SIR: Total C            17 Sep 1996        0.159         -0.175
                         5 Mar 1997        0.164         -0.014

Decomposition rate of    5 Mar 1997       -0.082         -0.136

                                             Plant variables

                                           M to D
                          Sampling        biomass         Root
  Microbial variable        date      ratio([parallel]   biomass

SIR([dagger])           17 Sep 1996        0.066          0.039
                         5 Mar 1997        0.137          0.108

BR([double dagger])     17 Sep 1996        0.233          0.190
                         5 Mar 1997        0.143          0.148

Bacteria: fungi         17 Sep 1996        0.260          0.109
                         5 Mar 1997       -0.152          0.230

BR: SIR                 17 Sep 1996        0.056          0.071
                         5 Mar 1997       -0.056          0.017

Fumigation [CO.sub.2]   19 Sep 1996       -0.070          0.027
  flush                  5 Mar 1997        0.024          0.138

SIR: Total C            17 Sep 1996        0.091         -0.021
                         5 Mar 1997        0.093          0.094

Decomposition rate of    5 Mar 1997        0.081         -0.084

                                        Plant variables

                                        diversity index

                          Sampling              Functional
  Microbial variable        date      Species     groups

SIR([dagger])           17 Sep 1996   -0.015      -0.083
                         5 Mar 1997   -0.131      -0.049

BR([double dagger])     17 Sep 1996   -0.166       0.053
                         5 Mar 1997    0.050      -0.081

Bacteria: fungi         17 Sep 1996   -0.113      -0.131
                         5 Mar 1997    0.270       0.142

BR: SIR                 17 Sep 1996   -0.049       0.127
                         5 Mar 1997    0.185       0.023

Fumigation [CO.sub.2]   19 Sep 1996    0.220      -0.142
  flush                  5 Mar 1997    0.106       0.030

SIR: Total C            17 Sep 1996   -0.047      -0.076
                         5 Mar 1997    0.038       0.016

Decomposition rate of    5 Mar 1997   -0.220      -0.095

                                      Soil chemical variables
  Microbial variable        date      Soil moisture     Soil C

SIR([dagger])           17 Sep 1996    0.532(***)      0.172
                         5 Mar 1997    0.409(**)       0.344(*)

BR([double dagger])     17 Sep 1996    0.372(**)       0.279(*)
                         5 Mar 1997    0.239           0.393(**)

Bacteria: fungi         17 Sep 1996   -0.356(*)       -0.058
                         5 Mar 1997   -0.143          -0.14

BR: SIR                 17 Sep 1996   -0.205           0.022
                         5 Mar 1997   -0.233           0.135

Fumigation [CO.sub.2]   19 Sep 1996    0.234           0.213
  flush                  5 Mar 1997    0.372(**)       0.306(*)

SIR: Total C            17 Sep 1996    0.286(*)       -0.335(*)
                         5 Mar 1997    0.108          -0.511(***)

Decomposition rate of    5 Mar 1997   -0.074           0.126

                                      Soil chemical variables
  Microbial variable        date        Soil N        Soil pH

SIR([dagger])           17 Sep 1996    0.409(**)     0.485(***)
                         5 Mar 1997    0.249         0.647(***)

BR([double dagger])     17 Sep 1996    0.389(**)     0.354(*)
                         5 Mar 1997    0.259         0.393(***)

Bacteria: fungi         17 Sep 1996    0.061         0.026
                         5 Mar 1997   -0.144        -0.043

BR: SIR                 17 Sep 1996   -0.095        -0.217
                         5 Mar 1997    0.028        -0.368(**)

Fumigation [CO.sub.2]   19 Sep 1996    0.480(***)    0.374(**)
  flush                  5 Mar 1997    0.24          0.671(***)

SIR: Total C            17 Sep 1996    0.129         0.378(**)
                         5 Mar 1997   -0.227         0.423(**)

Decomposition rate of    5 Mar 1997    0.244        -0.143

Notes: The non-plant-removal gaps are not included in the analysis.
Correlation coefficients that differ significantly from zero
(n = 50) are indicated by asterisks: (*) P < 0.05, (**) P < 0.01,
(***) P < 0.001.

([dagger]) Substrate-induced respiration.

([double dagger]) Basal respiration.

([sections]) Net aboveground primary productivity.

([parallel]) M, monocot; D, dicot.

The vegetation removal treatments did not affect the temporal variability of either SIR or the fumigation [CO.sub.2] flush (Table 6). However, there were important treatment effects on the temporal variability of basal respiration, with the CV values for the [C.sub.3] annual grass removal treatment being lower than for some of the other treatments. Temporal variability of SIR, fumigation CO2 flush, and basal respiration did not show detectable relationships with either plant biomass or diversity variables, other than a positive relationship between the cv of basal respiration and the cv of plant biomass (Table 7). However, temporal variability of basal respiration was negatively related to soil C and N concentrations and positively related to temporal variability of soil N. Spatial variability of soil microbial properties showed detectable treatment responses (Table 6), with removal of annual [C.sub.3] grasses enhancing variability relative to most other treatments, and with the no-plant removal treatment showing less variability than the other treatments for basal respiration.
TABLE 6. Temporal and spatial variability for microbial biomass
and activity variables.

                                               Treatment by
                                               removal of:

                                             All     [C.sub.4]
Measurement            Response variable    plants    grasses

Temporal variability   SIR([sections])      0.294      0.301
  ([dagger])           Fumigation
                         [CO.sub.2] flush   0.153      0.196
                       Basal respiration    0.338      0.405
Spatial variability    SIR([sections])      0.193      0.212
  ([double dagger])    Fumigation
                         [CO.sub.2] flush   0.131      0.147
                       Basal respiration    0.217      0.198

                                                 Treatment by
                                                 removal of:

                                            [C.sub.3]      All
                                             annual     [C.sub.3]
Measurement            Response variable     grasses     grasses

Temporal variability   SIR([sections])        0.316       0.318
  ([dagger])           Fumigation
                         [CO.sub.2] flush     0.210       0.219
                       Basal respiration      0.319       0.406
Spatial variability    SIR([sections])        0.272       0.239
  ([double dagger])    Fumigation
                         [CO.sub.2] flush     0.201       0.146
                       Basal respiration      0.202       0.190

                                               Treatment by
                                               removal of:

                                            ledonous     No
Measurement            Response variable     weeds     plants

Temporal variability   SIR([sections])       0.291     0.272
  ([dagger])           Fumigation
                         [CO.sub.2] flush    0.199     0.181
                       Basal respiration     0.346     0.364
Spatial variability    SIR([sections])       0.218     0.233
  ([double dagger])    Fumigation
                         [CO.sub.2] flush    0.147     0.141
                       Basal respiration     0.209     0.159

Measurement            Response variable     ([parallel])

Temporal variability   SIR([sections])          0.065
  ([dagger])           Fumigation
                         [CO.sub.2] flush       0.047
                       Basal respiration        0.082
Spatial variability    SIR([sections])          0.047
  ([double dagger])    Fumigation
                         [CO.sub.2] flush       0.045
                       Basal respiration        0.045

([dagger]) Expressed as the coefficient of variation (CV = mean/sD)
across sampling dates for each treatment for each replicate block.

([double dagger]) Expressed as the cv across blocks for each
treatment for each sampling time.

([sections]) Substrate-induced respiration.

([parallel]) Least significant difference at P = 0.05 after ANOVA.
TABLE 7. Pearson's correlation coefficients between temporal
variability of microbial biomass or activity measurements
(determined as coefficients of variation across six
sampling dates) and various plant and soil properties,
across the 50 replicate block X treatment combinations
(plant-free treatment gaps are not included in analysis).

                              CV for microbial variable

Soil or plant variable   SIR([double dagger])   [CO.sub.2] flush

Mean NPP([dagger])              0.032                 0.247
CV of NPP                       0.202                -0.128
Mean plant biomass             -0.024                 0.138
CV of mean plant bio-
  mass                          0.166                 0.226
Plant species richness         -0.021                 0.034
Shannon-Weiner di-
  versity index (plant
  species)                      0.160                 0.141
Shannon-Weiner di-
  versity index (plant
  functional groups)            0.078                -0.009
Mean soil C                     0.048                 0.108
CV of mean soil C              -0.245                -0.148
Mean soil N                    -0.002                 0.149
CV of mean soil N               0.138                 0.012

                         CV for microbial

Soil or plant variable    BR([sections])

Mean NPP([dagger])          -0.090
CV of NPP                    0.000
Mean plant biomass           0.043
CV of mean plant bio-
  mass                       0.378(**)
Plant species richness      -0.018
Shannon-Weiner di-
  versity index (plant
  species)                   0.096
Shannon-Weiner di-
  versity index (plant
  functional groups)         0.127
Mean soil C                 -0.481(***)
CV of mean soil C           -0.031
Mean soil N                 -0.404(**)
CV of mean soil N            0.536(***)

Note: Correlation coefficients that differ significantly from
zero are indicated by asterisks: (*) P < 0.05, (**) P < 0.01,
(***) P < 0.001.

([dagger]) Net primary productivity.

([double dagger]) Substrate-induced respiration.

([sections]) Basal respiration.

Microbial community composition

When considered separately, the principal phospholipid fatty acids (PLFAs) generally responded to treatments in a similar manner to that found for the microbial biomass data (Fig. 12), with the concentration of each PLFA usually being significantly less in the all-plant removal treatment than in some of the other treatments. There were also important differences in PLFA composition (and, hence, in microbial community structure) between the five treatments with plants present (Fig. 12). The treatments did not have a significant effect on the total amounts of bacterial PLFAs ([F.sub.5,42] = 1.36, P = 0.260) or the total microbial PLFAs ([F.sub.5,42] = 1.42, P = 0.236), although the levels were least in the all-plant removal treatment. The ratio of bacterial to fungal PLFAs was also not significantly related to treatment ([F.sub.5,42] = 1.18, P = 0.545).


The Shannon-Weiner diversity index for the microbial PLFAs ranged from 2.57 to 2.63 across the six treatments, with treatment effects being nonsignificant ([F.sub.5,42] = 1.18, P = 0.336). Further, when all the treatments except the all-plant removal treatments were considered, PLFA diversity was not significantly correlated with plant species diversity (Shannon-Weiner index), either at the same sampling date (Fig. 13) or at earlier vegetation measurement dates (Fig. 14). However, PLFA diversity did show some relatively weak negative associations with plant functional group diversity (Shannon-Weiner index) at earlier measurement dates (Fig. 14). With regard to the ordination analyses of the PLFA data, the main axis was not significantly related to treatment ([F.sub.5,42] = 1.79, P = 0.136), but the second axis showed a marginally significant treatment response ([F.sub.5,42] = 2.56, P = 0.041), with the all-plants-removed or only [C.sub.4]-grasses-removed treatments being separated from the other treatments (data not presented). Further, there were highly significant correlations between the PLFA and plant community ordination results (Fig. 13), suggesting detectable effects of plant community structure on microbial community structure. There were no lag effects of plant community structure on microbial community structure, with the PLFA ordination data being most closely related to that of the plant community measured at the same time (Fig. 14).


Soil nematode functional and trophic groups

The treatments that we imposed had significant effects on various components of the nematode fauna, but these were intermittent and only detectable for some of the sampling dates (Fig. 15). Generally, the all-plant removal treatment resulted in lower numbers of bacterial-feeding nematodes than did the other treatments; during the second half of the experiment, there was also an apparent reduction of these nematodes in the [C.sub.3] grass removal treatment. Toward the end of the study, the nonremoval treatment also supported more bacterial-feeding nematodes than did most of the other treatments. Fungal-feeding nematodes were less responsive to treatments, although there were important treatment effects for the 7 March 1995 sampling, with the all-plant removal treatment and (to a lesser extent) the [C.sub.4] grass removal treatment supporting fewer nematodes than did the other treatments. For the final two years, the numbers of top predatory nematodes were consistently less in the all-plant removal treatment than in most of the others; the [C.sub.3] grass removal treatments often resulted in more nematodes than the [C.sub.3] annual grass and [C.sub.4] grass removal treatment, indicative of beneficial effects of removing L. perenne. Omnivorous nematodes were less abundant in the all-plant removal treatment than in most of the other treatments from 7 March 1995 onward, although there were also some negative effects of removing all [C.sub.3] grasses. In summary, for those functional groups involved in the decomposer food web, it is apparent that the strongest decrease in biomass was obtained either by removing all plants, or through removing all [C.sub.3] grasses. Plant pathogens and plant associates showed less obvious trends, and although differences between treatments sometimes emerged, these were generally sporadic. There were important temporal differences between treatments, with top predatory nematodes showing maximal population density later in the study and the other functional groups generally showing maximal densities at earlier samplings.


Microbe-feeding nematodes showed the strongest correlations with plant properties early in the experiment, and nematode-microbial correlations tended to be stronger for the winter samplings (Table 8). Microbe-feeding nematodes were also consistently related to soil chemical properties, particularly soil N. Top predatory nematodes were negatively correlated with microbe-feeding nematode numbers at the end of the experiment, and also showed consistent negative relationships with soil N (Table 8).
TABLE 8. Pearson's correlation coefficients between nematode
populations in the microbe-feeder and predator categories of the
decomposer food web, and the properties of lower trophic levels.

                                 Net primary
                               productivity vs.

                       MFN([double dagger])   PN([sections])

Trophic levels

Sampling date                1 vs. 3             1 vs. 4
  14 September
    1994                     0.140                0.184
   7 March 1995              0.291(*)            -0.090
  19 September
    1995                     0.362(**)            0.020
   5 March 1996              0.109               -0.175
  17 September
    1996                     0.086               -0.065
   5 March 1997              0.016                0.109

                           Total plant
                           biomass vs.

                          MFN        PN

Trophic levels

Sampling date           1 vs. 3    1 vs. 4
  14 September
    1994               -0.067       0.246
   7 March 1995         0.292(*)   -0.221
  19 September
    1995                0.213       0.084
   5 March 1996        -0.152       0.021
  17 September
    1996               -0.071       0.030
   5 March 1997         0.147       0.005

                              Soil N vs.

                          MFN           PN

Trophic levels

Sampling date           1 vs. 3       1 vs. 4
  14 September
    1994               0.306(*)     -0.389(**)
   7 March 1995        0.444(**)    -0.330(*)
  19 September
    1995               0.465(***)   -0.413(**)
   5 March 1996        0.361(**)    -0.199
  17 September
    1996               0.330(*)     -0.464(***)
   5 March 1997        0.016        -0.070

                               SIR vs.

                          MFN          PN       MFN vs. PN

Trophic levels

Sampling date           2 vs. 3      2 vs. 4      3 vs. 4
  14 September
    1994                0.426(**)    -0.298(*)    0.025
   7 March 1995         0.083        -0.191      -0.140
  19 September
    1995                0.505(***)   -0.175      -0.305(*)
   5 March 1996         0.144        -0.087      -0.260
  17 September
    1996               -0.014        -0.155      -0.477(***)
   5 March 1997         0.072         0.128      -0.481(***)

Note: Correlation coefficients that differ significantly from 0
(at N = 50 gaps; plant-free gaps not included in the analysis)
are indicated by asterisks: (*) P < 0.05, (**) P < 0.01,
(***) P < 0.001.

([dagger]) Numbering for trophic levels: 1, resource base (sensu
Wardle 1995); 2, primary consumer; 3, secondary consumer; 4,
tertiary consumer. Relationships between trophic levels 1 and 2
are shown in Table 4.

([double dagger]) Microbe-feeding nematodes.

([sections]) Predaceous nematodes (top predators and omnivores).

Temporal variability of nematode populations in each of the nematode functional groups was not significantly related to treatment (data not presented). Further, temporal variability of these groups did not show significant correlations with plant biomass or productivity, or their temporal variability (data not presented). In addition, temporal variability of those functional groups dependent on the microbial energy channels was not correlated with any of the microbial variables or their temporal variability. However, the cv of the populations of top predatory nematodes was significantly correlated with both soil N content (r = 0.474, P [is less than] 0.001) and the cv of soil N content (r = -0.389, P = 0.007). Similarly, there were few consistent treatment effects on the spatial variability of soil nematodes, although the spatial cv for the fungal-feeding and plant-parasitic nematodes was significantly greater in the dicotyledonous weed removal treatment than in some of the other treatments (data not presented).

Soil nematode community composition

At finer taxonomic levels of resolution, the effects of plant functional group removal on nematodes were much clearer (Fig. 16). Over the entire experimental period, 15 taxa showed distinct responses to treatments, and there were clear differences between the six treatments with regard to the overall community composition of nematodes (Fig. 16). These community-level effects were matched by a reduction in Shannon-Weiner diversity indices calculated for the microbe-feeding nematodes in the [C.sub.3] grass removal treatment, but this was not the case for diversity indices for herbivorous or predaceous nematodes or for nematode functional groups (Table 9). Further, nematode diversity indices were not related to those of plants or microbes (PLFA data) at the end of the experiment (Fig. 13). However, there were apparent lag effects, with taxonomic diversity indices for microbe-feeding nematodes on 5 March 1997 being most closely related to plant functional group diversity indices (point quadrat analysis data) of the same gaps 3 mo earlier, and with predaceous nematode diversity indices showing marginally significant negative relationships with plant species diversity indices at some earlier sampling dates (Fig. 14). No lag effects were apparent with regard to herbivorous nematode diversity indices. It is therefore apparent that diversity indices calculated for nematodes are at best only weakly related to those of lower trophic levels.

TABLE 9. Shannon-Weiner diversity index for taxa (mainly genera) in
three nematode trophic groups, and for the nematode functional
group diversity (all nematodes were classified into the six
functional groups shown in Fig. 15), in response to vegetation
removal treatments, averaged over the entire experimental period.

                                    Treatment by removal of:

 Nature of                         All     [C.sub.4]    annual
 diversity     Trophic grouping   plants    grasses     grasses

Diversity of   microbe feeders     1.66      1.61        1.69
  taxa         herbivores          1.43      1.53        1.43
               predators           0.65      0.83        0.67
Diversity of
  groups       all                 1.70      1.70        1.66

                                      Treatment by removal of:

 Nature of                        All [C.sub.3]   ledonous     No
 diversity     Trophic grouping      grasses       weeds     plants

Diversity of   microbe feeders        1.46          1.62      1.72
  taxa         herbivores             1.46          1.45      1.51
               predators              0.73          0.73      0.77
Diversity of
  groups       all                    1.71          1.71      1.69

 Nature of
 diversity     Trophic grouping   [LSD.sub.0.05]([dagger])

Diversity of   microbe feeders              0.16
  taxa         herbivores                   0.18
               predators                    0.20
Diversity of
  groups       all                          0.10

([dagger]) Least significant difference at P = 0.05, derived from
ANOVA testing for treatment, time, treatment X time, and blocking

The taxon-specific responses of microbe-feeding nematodes to treatments were also reflected in the ordination analyses; ordination scores differed significantly between treatments for at least one of the two main axes for all sampling dates except the first one (Table 10). In particular, there were often large differences between the all-plant removal treatment, or the [C.sub.3]-grass removal treatment, and most of the others. In contrast, ordination analyses of the herbivorous nematode trophic level showed no consistent treatment effects (data not presented). There was no close relationship between the ordination axes of any of the nematode trophic levels and those of plants, but ordination axes for predaceous nematodes were strongly significantly correlated with those of the two next lowest trophic levels (Fig. 13). Further, lag effects were apparent, with the ordination axes of the microbe-feeding and predaceous nematodes being most closely correlated with those of the plant community (point quadrat analysis data) 1 mo earlier and 2 mo earlier, respectively (Fig. 14).
TABLE 10. Average detrended correspondence analysis (DCA) values
(rank-transformed) for the two axes explaining the greatest
proportion of variation following ordination of the microbe-feeding
nematode community data for each sampling date.

                Ordination   explained
Sampling date      axis         (%)

14 Sep 1994         I          18.7
                    II         15.1
 7 Mar 1995         I          19.3
                    II         12.8
19 Sep 1995         I          16.0
                    II         12.8
 5 Mar 1996         I          18.2
                    II         12.7
17 Mar 1996         I          19.1
                    II         11.9
 5 Mar 1997         I          15.8
                    II         14.9

                               Treatment by removal of:

Sampling date      axis       All plants      C4 grasses

14 Sep 1994         I        [27.1.sup.a]    [30.1.sup.a]
                    II       [27.3.sup.a]    [34.4.sup.a]
 7 Mar 1995         I        [22.8.sup.bc]   [37.7.sup.a]
                    II       [22.3.sup.b]    [37.4.sup.ab]
19 Sep 1995         I        [22.6.sup.b]    [29.3.sup.ab]
                    II       [29.9.sup.a]    [25.9.sup.a]
 5 Mar 1996         I        [35.5.sup.a]    [31.1.sup.a]
                    II       [28.8.sup.ab]   [38.5.sup.a]
17 Mar 1996         I        [24.5.sup.b]    [32.9.sup.ab]
                    II       [30.4.sup.ab]   [29.4.sup.ab]
 5 Mar 1997         I        [38.6.sup.a]    [30.5.sup.ab]
                    II       [34.5.sup.a]    [31.1.sup.a]

                               Treatment by removal of:

                Ordination     C3 annual        All C3
Sampling date      axis         grasses         grasses

14 Sep 1994         I        [28.4.sup.a]    [34.3.sup.a]
                    II       [37.5.sup.a]    [27.6.sup.a]
 7 Mar 1995         I        [38.5.sup.a]    [20.7.sup.c]
                    II       [29.2.sup.ab]   [39.6.sup.a]
19 Sep 1995         I        [29.3.sup.ab]   [34.9.sup.ab]
                    II       [30.1.sup.a]    [30.1.sup.a]
 5 Mar 1996         I        [27.3.sup.a]    [30.2.sup.a]
                    II       [20.3.sup.b]    [31.5.sup.ab]
17 Mar 1996         I        [27.6.sup.b]    [25.0.sup.b]
                    II       [29.4.sup.ab]   [31.3.sup.ab]
 5 Mar 1997         I        [28.5.sup.ab]   [30.1.sup.ab]
                    II       [25.3.sup.a]    [30.5.sup.a]

                               Treatment by removal of:

                Ordination     ledonous
Sampling date      axis          weeds         No plants

14 Sep 1994         I        [32.9.sup.a]    [30.3.sup.a]
                    II       [25.6.sup.a]    [30.7.sup.a]
 7 Mar 1995         I        [34.0.sup.ab]   []
                    II       [32.4.sup.ab]   [22.3.sup.b]
19 Sep 1995         I        [37.3.sup.a]    [29.8.sup.ab]
                    II       [32.1.sup.a]    [35.1.sup.a]
 5 Mar 1996         I        [32.3.sup.a]    [26.5.sup.a]
                    II       [34.5.sup.a]    [29.7.sup.ab]
17 Mar 1996         I        [31.8.sup.ab]   [41.4.sup.a]
                    II       [42.2.sup.a]    [20.4.sup.b]
 5 Mar 1997         I        [26.2.sup.ab]   [25.2.sup.b]
                    II       [33.6.sup.a]    [24.6.sup.a]

Note: Numbers followed by the same superscript letter within each
row are not significantly different at P < 0.05 (least significant
difference test following ANOVA on rank-transformed data).

Soil-associated mesofauna and macrofauna

The mesofaunal components of the decomposer food web were not strongly influenced by plant species removal effects. Removal of all plants resulted in reduced populations of Collembola relative to the other treatments, but there were no clear or consistent differences among the other five treatments (Fig. 17). Mite populations were low and sporadic, and treatment effects were nonsignificant over the experimental period for both saprophagous/fungivorous oribatid mites ([F.sub.5, 314] = 1.25, P = 0.284) and predatory mites ([F.sub.5, 314] = 0.38, P = 0.864). Further, none of these mesofaunal groups showed statistically significant relationships with any of the plant or soil variables when correlation analysis was used (data not presented).


The earthworm (Lumbricidae) fauna consisted entirely of two endogeic species, i.e., Aporrectodea caliginosa (Savigny) and Lumbricus rubellus Hoffmeister. Although there were significant overall effects of treatments on both species, the only treatment that had a consistent influence was the all-plant removal treatment, which generally reduced populations of both species throughout most of the experimental period (Fig. 18). Correlation analysis revealed no detectable relationship between earthworm populations and plant biomass or productivity (data not presented), and earthworms were not related to plant species composition as long as plants were present. A. caliginosa sometimes showed significant relationships with soil P, pH, and N, but these relationships were not consistent over time, and even when significant, correlation coefficients were usually below 0.40 (data not presented).


In contrast, the most abundant soil-associated herbivorous arthropod species present all showed clear relationships with treatments (Fig. 19). Floresianus sordidus Hustache (Curculionidae) only occurred in the March samplings each year. Larval populations were significantly reduced by all treatments except the [C.sub.3] annual grass, whereas adults were only significantly reduced by the all-plant removal treatment. Listroderis difficilis Germain (Curculionidae) was present only in the 14 September 1994 sampling, where it appeared in reasonable numbers; this species was significantly reduced in the [C.sub.4] removal and all-plant removal treatments. Sitona lepidis Gyllenhal (Curculionodae) occurred only in the 17 September 1996 sampling, consistent with its very recent accidental introduction to New Zealand and subsequent rapid spread over the previous few months. Both adults and larvae were inhibited in the all-plant removal treatment and (for adults) in the [C.sub.3] and the [C.sub.4] grass removal treatments. Naupactus leucoloma Boheman (Curculionidae) was abundant at all sampling dates. Larval populations were significantly elevated in the [C.sub.4] grass removal treatment and suppressed in the all-plant removal treatment relative to the no-plant removal treatment; significant differences were also apparent between the different removal treatments. Costelytra zealandica (White) (Scarabaeidae), which was only present in the 7 March 1997 sampling, showed only relatively weak treatment responses. Populations of Heteronychus arator (F.) (Scar abaeidae), which were only present each March, were enhanced in the [C.sub.4] grass and dicotyledonous weed removal treatments relative to the nonremoval treatment.


Correlation analyses revealed few significant relationships between populations of herbivorous arthropods and plant biomass or productivity. Further, ordination analyses of the herbivorous arthropod community structure at each sampling date revealed little evidence of treatment effects (data not presented), and there was no clear relationship with the ordination analyses of the plant community structure of the same gaps (Fig. 13). However, there was an important lag effect, with the main arthropod ordination axis for the 5 March 1997 data being most strongly related to the ordination axes of the plant community (point quadrat analysis data) for the same gaps 2 mo earlier (Fig. 14).

Species richness of the herbivorous arthropod fauna over the whole experiment was much less in the all-plant removal treatment than in all of the other treatments, which did not differ significantly from each other (Fig. 20). However, removal of [C.sub.4] grasses significantly enhanced the Shannon-Weiner diversity index relative to the nonremoval treatment. Further, there were significant negative relationships between Shannon-Weiner diversity indices for plants and herbivorous arthropods throughout the experiment (e.g., Fig. 13). There was also a negative lag effect of plant functional group diversity (point quadrat analysis data) and herbivorous arthropod data, with the strongest relationships occurring between arthropod diversity at 5 March 1997 and plant functional diversity 3 mo earlier (Fig. 14).



Our study has shown that removal of plant functional groups can have important effects on the composition of the remainder of the flora, and that this can influence vegetation dynamics, biomass, productivity, and diversity. These responses of the aboveground subsystem have the potential to induce corresponding responses in the belowground subsystem, influencing soil food webs, community composition of soil organisms and their diversity, and, ultimately, ecosystem properties and processes. We now discuss each of these issues in turn.

Vegetation responses

When a major component of the flora is removed from the species pool, competition theory predicts that other species should benefit and, therefore, at least partially compensate for the lost production and biomass of the removed species. The degree of compensatory response detected should be related to the degree of niche overlap between the removed and remaining species, especially in relation to their utilization of limiting resources (Hooper and Vitousek 1997). In our study, such compensatory effects were often apparent, consistent with the hypotheses that we proposed. The strongest treatment responses by most components of the flora were to the removal of all [C.sub.3] grasses (occupying 22-53% of total biomass). Because removal of [C.sub.3] annual grasses (occupying 0-49% of biomass) alone did not exert much effect, we therefore conclude that [C.sub.3] perennial grasses, consisting almost entirely of Lolium perenne, were responsible for many of the treatment effects that we observed, consistent with previous investigations (Wardle et al. 1995a, Campbell et al. 1996). The data for the L. perenne x Trifolium repens interaction were, however, less consistent; there were periods during the winter when T. repens growth was strongly stimulated by L. perenne removal, and some periods in the summer when it was strongly inhibited by L. perenne removal (especially in early 1996). This summer stimulation is inconsistent with studies indicating generally suppressive effects of Lolium species on T. repens (e.g., Mann and Barnes 1953, Harris 1987), whereas the suppressive effect of L. perenne removal in summer partially supports the observation that this species is less suppressive of T. repens under drier conditions (Thomas 1984).

Our data are consistent with the view that species are not equal in terms of their ecosystem-level implications (Allen and Forman 1976, Goldberg 1987, Power et al. 1996). L. perenne clearly exerted a disproportionate effect on the other components of the flora, meaning that the ecophysiological traits of this species presumably conferred some competitive advantage (see Abdul-Fatih and Bazzaz 1979, Austin and Smith 1989, Huston 1994, Wardle et al. 1998a). One such trait may be the relative allocation of mass to aboveground and belowground tissues; L. perenne clearly had a far lower shoot to root ratio than did most of the other species; removal of all [C.sub.3] grasses resulted in a highly significant (and for the 7 March 1995 sampling, very large) enhancement of the total shoot mass to root mass ratio in the gaps. This means that L. perenne probably has very different strategies for nutrient acquisition than do plants of the other functional groups present in our study (Chapin 1980, Ingestad and ,[Angstrom] gren 1988).

The compensatory effects between different components of the flora infer that removal of functional groups does not generally cause large effects with regard to total plant cover, biomass, or productivity. However, intermittent effects are apparent. For example, total aboveground plant biomass was much greater on 7 March 1995, when [C.sub.3] grasses were removed, and this is clearly due to the very high biomass of [C.sub.4] species present in the [C.sub.3] grass removal gaps. This is because removing L. perenne enables greater [C.sub.4] grass seedling establishment, inducing greater [C.sub.4] grass growth during periods in the summer when other species are suppressed by moisture limitation. In contrast, removal of [C.sub.4] grasses caused reduced total aboveground biomass on 5 March 1996, and temporarily reduced total cover in the late-summer period of each year. This is attributable to the inability of the other species present to occupy part of the niche left vacant by removing [C.sub.4] grasses during the summer; the absence of [C.sub.4] species simply results in a higher incidence of bare ground. In a similar vein, removal of L. perenne results in reduced total root biomass relative to the other treatments, simply because no other species is capable of occupying as much of the soil volume as is L. perenne.

There were important patterns of temporal variation throughout the study, with rapid initial increases in cover, NPP, and biomass (reflecting colonization and initial succession on the bare ground), followed by a decline after a few months in most treatments, probably due, in part, to unusually dry conditions during December 1994. Plant biomass and cover increased again over the third year of the study as conditions became more moist. It is unclear whether the system had reached an asymptotic phase with regard to these plant properties by the end of the study (or whether a longer time period would be necessary for this to occur), although there were no clear successional trends during the final two years of the study. Exclusion of plant functional groups from the species pool did not significantly alter temporal variability of NPP or plant biomass, and thus did not affect stability over the experimental period (with the exception of reduced variability in the [C.sub.3] annual grass removal treatment). Our study was not of sufficient duration to separate the effects of the two components of this temporal variability, i.e., inter-year and inter-seasonal (March vs. September) variability, but despite this, our data still provide little evidence that excluding components of the total flora has destabilizing influences (cf. Tilman 1996). However, our study does provide clear evidence that removal of L. perenne enhanced spatial variability of both biomass and NPP across replicate blocks. There were large differences between replicate blocks throughout the study and it is apparent that the presence of L. perenne was able to at least partially reduce this variation and enhance uniformity across plots, at least at the spatial scale that we considered.

Our results supported our hypotheses, in that removal of L. perenne also enhanced the species richness of the dicotyledonous weeds and, in the early summer period, that of the [C.sub.4] grasses. This means that at the within-gap scale, some plant species aree simply excluded by competition from L. perenne. This is consistent with earlier work pointing to the ability of some plant species to reduce species richness of the remainder of the species pool (e.g., Gurevitch and Unnasch 1989, Ten Harkel and van der Muelen 1996, Collins et al. 1998; but see Hils and Vankat 1982, Wardle and Barker 1997). Our results suggest that compensatory effects, such as we observed for plant cover, NPP, and biomass, also occur with regard to plant species richness, in that removal of a subset of the flora can be offset by a corresponding increase in the species richness of the remainder of the flora. This helps to explain why removal of significant components of the flora often did not result in a drop in overall species richness.

Decomposer food web responses

We found that removing all plants from gaps over a 3-yr period (i.e., resulting in zero NPP and plant biomass) had adverse effects relative to the other treatments for all three consumer trophic levels, indicative of strong bottom-up control when extreme differences in NPP and plant biomass occurred. This is consistent with the study of Mikola and Setala (1998b), which found that adding basal resources (which simulate increased root NPP) can simultaneously increase biomass in adjacent trophic levels, and is therefore partially inconsistent with the predictions of both basic trophic-dynamic models (e.g., Rosenzweig 1971, Oksanen et al. 1981) and the hypotheses that we formulated. However, as long as plants were present, there were few consistent relationships between these plant variables and any of, the decomposer food chain components. Further, temporal patterns in NPP were not matched by corresponding shifts in microbes or predatory nematodes, although patterns of decline in NPP over much of the study did correspond to a decline in microbe-feeding nematodes. Generally, however, our results suggest that as long as plants are present, the components of the soil food chain that we considered are not strongly driven by NPP in the manner hypothesized (and sometimes shown) for aboveground food chains (e.g., Oksanen et al. 1981, Power 1992, Van de Koppel et al. 1996). Our failure to detect a consistent effect of NPP on the measured soil food web components is partially consistent with previous studies that have found relationships between NPP and microbial biomass to be either positive (D. Zak et al. 1994) or negative (Wardle et al. 1995a), and those that have shown uncertain relationships between NPP and numbers of microbe-feeding nematodes (Yeates and Coleman 1982, Yeates 1987).

We identify three possible reasons as to why our hypotheses were not supported and why soil food web properties may be only weakly related to plant variables such as NPP and biomass. Firstly, soil food webs, unlike aboveground food webs, are controlled by longterm effects of litter input, so the effects of shifts in these plant variables may be manifested over much longer time scales than the 3-yr duration of our study. In this context, it is apparent that relatively stable soil chemical variables, especially soil C and N levels, were much more strongly correlated with soil food web properties than were plant treatment effects, when all gaps containing plants were considered. In particular, soil N appeared to work as a bottom-up control, affecting all components of the food chain, although these effects were frequently negative for the predatory nematodes. This latter effect is consistent with the negative association that sometimes occurs between predatory and bacterial-feeding nematodes (Wardle et al. 1995b). Our results suggest that effects of spatial variability on soil properties, usually determined mainly by historical vegetation properties over the past few centuries (Tate 1992), are more likely to determine soil food web components than are contemporary patterns of plant biomass or productivity. Secondly, with regard to the decomposer organisms, the quality of organic matter entering the soil may be more important than the amounts of material added (Wardle and Lavelle 1997). In this light, there were detectable effects of plant community composition on components of the decomposer food chain that we considered; in particular, both the [C.sub.3] and [C.sub.4] grass removal treatments influenced the microbefeeding and top predatory nematode populations. Thirdly, the complex interplay of top-down and bottom-up forces that structure and stabilize decomposer food webs (Wardle 1995, De Ruiter et al. 1995) is likely to buffer food web components against shifts in NPP and plant biomass and community structure. The soil microbial biomass, which is the component that is most likely to respond to shifts in NPP, has previously been shown to be unresponsive to such changes, simply because it was grazed by microbe-feeding nematodes (Yeates et al. 1997). Similarly, the complex nature of predator-prey cycles that can occur between bacterial-feeding and top predatory nematodes (Wardle et al. 1995b, Yeates and Wardle 1996) could reasonably be expected to reduce the likelihood of detecting associations between these components at fixed sampling times.

The larger soil decomposer animals, such as earthworms, Collembola, and mites, were highly unresponsive to plant variables. Although populations of Collembola and earthworms were often severely reduced where plants were entirely excluded, as long as plants were present, there was little evidence that NPP, plant biomass, or plant species composition had important effects on these organisms. Thus, even large differences in the quality and quantity of resource input, such as that which occurred across treatments in our study, do not emerge as significant population determinants for the larger, generalist saprophagous organisms present in our system.

Soil organisms: responses of community structure and diversity patterns

Although whole-trophic-level responses to the manipulation of plant community structure were generally weak, at finer levels of taxonomic resolution, there were strong multitrophic responses to the manipulation treatments, consistent with the hypotheses that we proposed. With the PLFA data, most fatty acids of microbial origin showed responses to treatments, although the principal response was to the removal of all plants. These results are generally consistent with earlier studies predicting that vegetation composition can operate as a determinant of microbial community structure (J. Zak et al. 1994), presumably because of the patterns of specificity of different microbial species for resources of different quality (Robinson et al. 1994). These effects were also apparent at higher trophic levels; for example, half of the microbe-feeding nematode taxa depicted in Fig. 16 showed clear preferences for treatments occupied by some plant species over those occupied by others, indicating that plant community structure affects microbe-feeding nematode communities by altering the types of microbes that are present. There are few previous studies demonstrating such effects, although Freckman and Ettema (1993) and Wasilewska (1995) provide data suggesting that some bacterial-feeding nematode taxa are responsive to plant species differences. Plant community effects may also have the potential to affect taxonomic composition of nematodes in the fourth trophic level; mononchid and dorylaimid nematodes both showed clear treatment preferences. The ordination data indicate that the structure of microbial, microbe-feeding nematode, and predaceous nematode communities are linked to plant community structure, although there are apparent time lags for the latter two groups, reflective of the generation times of those nematodes. In this light, it is apparent from Fig. 14 (top right subgraph) that the main ordination axis summarizing the community structure of the primary detrital consumers (microbes) was most closely linked to that summarizing plant community structure measured at the same date (i.e., 5 March 1997). The main axis for the secondary consumers (microbe-feeding nematodes) was most closely related to that for the plant community measured 1 mo earlier; and the main axis for the tertiary consumers (predacious nematodes) was best related to that for the plant community measured 2 mo earlier. Therefore, our data show that removal of plants from an ecosystem can have detectable effects on the community structure of decomposer organisms, which may be manifested over several trophic levels.

Curiously, herbivorous nematodes were less closely linked to plant community structure than were those nematodes involved in the decomposer food web, although four taxa did show significant preferences for some treatments over others. One particularly surprising result is that there were appreciable numbers of herbivorous nematodes in the plant-free gaps, even after three years. The ability of plant-parasitic nematodes to survive for lengthy periods in the absence of host plants has been recorded previously (e.g., Harrison and Hooper 1963, McKendry 1987), and may be due to survival strategies based on anhydrobiosis, at least in dry conditions (Demeure and Freckman 1981), or because these nematodes could have survived on other resources present in the soil, such as nonliving plant material (see Eriksson 1974, Verdejo-Lucas and Pinochet 1992) or soluble soil C (see Nicholas 1962).

Strong patterns of plant specificity were demonstrated by the main species of herbivorous arthropods present. Populations of the commonest curculionid species were generally smaller under the dicotyledonous weed removal treatment than the nonremoval treatments, consistent with literature suggesting that many of these species prefer dicotyledonous rather than monocotyledonous species (Parker and Berry 1950, May 1966, 1993, Lanterni and Marvaldi 1995). The beneficial effects of removal of [C.sub.4] grasses for Naupactus leucoloma are consistent with the inferior resource quality provided by [C.sub.4] species, such as low nitrogen levels and high levels of recalcitrant structural carbohydrates (Campbell et al. 1996, Wardle et al. 1998b). Further, the higher levels of Heteronychus arator under some of the treatments allowing dense grass cover is in agreement with the known preferences of this species for grass roots (Watson and Wrenn 1980, Watson and Marsden 1981). The ordination results also demonstrate detectable linkages between the community structures of plants and herbivorous arthropods, although these are complicated by a time lag due to the generation times of the arthropod species present (Fig. 14).

A related issue is how plant diversity may affect the diversity of other groups of organisms. We would expect that, with the enhanced habitat heterogeneity present in a more diverse plant community, a wider range of niches would be provided, facilitating a greater diversity of consumer organisms. This has often been demonstrated for aboveground consumers (MacArthur 1965, Pimm 1991, Huston 1994), but only occasionally for belowground herbivores (House 1989) and decomposer organisms (Anderson 1978, Barker and Mayhill 1999, Sulkava and Huhta 1998). Contrary to our hypotheses, we not only failed to find such a relationship, but also actually detected a negative association between plant diversity (Shannon-Weiner index) and that of some of the other groups that we considered. This was especially apparent for the herbivorous arthropods, probably because reducing plant diversity through removing the [C.sub.4] grasses also improved the overall resource quality, enabling a greater frequency of some of the arthropod fauna to occur. In other words, a highly diverse plant community consisting of a proportion of species of low palatability may potentially support a less diverse herbivorous arthropod fauna than would a less diverse community consisting mainly of palatable species. A similar (although weaker) pattern is apparent with regard to the microflora (PLFA data) and the predatory nematodes (at least as revealed by the lag data in Fig 14); the reasons are less clear, but this result also appears to be reflective of the effects of resource quality manifesting itself through several trophic levels. The only negative treatment effect on the diversity of any of the five consumer groups that we assessed was the effect of [C.sub.3] grass exclusion on microbe-feeding nematode diversity, but this appears to be a specific response to L. perenne removal rather than to a reduction of grassland diversity. Therefore, there is little evidence from our study to support the view that reducing diversity in one trophic level is necessarily matched by a corresponding loss of diversity in other trophic levels. Rather, the response of consumer diversity to plant community attributes is more likely to reflect the traits or characteristics of the plant species present.

Ecosystem properties

Several recent experimental studies have attempted to manipulate organism diversity and have interpreted results in terms of the effects of loss of species from ecosystems on ecosystem function (Naeem et al. 1994, Tilman et al. 1996), ecosystem stability (Tilman 1996, McGrady-Steed et al. 1997), and variability (Naeem and Li 1997). Although these studies have concluded that loss of species can have predictable adverse consequences for ecosystem properties, the interpretations of several of these findings remain controversial (see Aarssen 1997, Gamier et al. 1997, Grime 1997, Huston 1997, Wardle 1998a). Although our study did not directly test for the effects of varying diversity, it does directly address the issue of whether permanent exclusion of species has important effects on ecosystem properties and processes at the spatial scale that we are considering. There is no evidence from our study to support our hypothesis that permanent loss of subsets of the flora has consistent unidirectional or negative consequences for ecosystem properties, either aboveground (e.g., NPP, standing biomass, plant cover) or belowground (e.g., decomposition rates, soil [CO.sub.2] release, nutrient dynamics, biomasses of those organisms that carry out decomposition-related processes). Further, there are no clear, unidirectional effects of functional group removal on either the stability (temporal variability) or spatial heterogeneity ("ecosystem reliability"; Naeem and Li 1997) of the properties that we considered. Part of the reason we did not detect such effects may be because our system seems to be largely buffered against the effects of species removal; exclusion of a subset of the flora may be compensated for by increased production and biomass of other components (see Hooper 1998).

It is also apparent that some ecosystem properties responded more strongly to treatments than did others. Aboveground responses were generally greater than those below ground; exclusion of some components of the flora often enhanced aboveground biomass and production of the remaining components, especially when treatments involved removal of L. perenne. Levels of soil organisms were adversely affected by removal of all plants, but as long as plants were present, vegetation composition did not greatly alter components of the decomposer subsystem, at least at the functional group level of resolution. Because decomposer-related processes are known to be regulated by the magnitude of the active soil microbial biomass (Beare et al. 1991), the biomass and populations of soil animals that catalyze its turnover (Visser 1985), and the structure of the decomposer food web (De Angelis 1992, Setala 1995, Bengtsson et al. 1996), it is perhaps not surprising that ecosystem properties determined by the soil biota were less consistently correlated with treatments than were many of the aboveground response variables. Soil chemical properties such as C and N concentrations were especially insensitive to treatment effects, presumably because they reflect much longer term changes than we could consider during the course of this study (see Tate 1992), and possibly because of the buffering effect of the relatively high soil organic matter levels characteristic of our study site (see De Angelis 1992, Wardle 1998b). Given that the microbial biomass and the fauna that feed upon it are often closely linked to soil C and N levels (Jenkinson and Ladd 1981), the soil biota also would have been buffered to some extent against shifts in vegetation composition.

Our study demonstrates that the various responses of ecosystem- and community-level properties to permanent exclusion of functional groups were ultimately driven by traits of the component plant species. This concurs with recent findings pointing to the role of vegetation composition and the significance of dominant plant species in driving how ecosystems function (Hooper and Vitousek 1997, 1998, Tilman et al. 1997, Wardle et al. 1997, Grime 1998). Ultimately, the exclusion of a given plant species, or a group of plant species, has the potential to affect a wide range of community- and ecosystem-level attributes, both above- and belowground. The nature of these effects is likely to be governed mainly by the specific attributes or traits of those species that are lost.


For technical support for the various components of this study, we owe many thanks to N. Bell, M. Dexter, A. Firth, J. Gow, T. James, F. Neville, B. Ryburn, and M. Vojvodic-Vukovic. Many thanks also to B. Campbell and A. Rahman for support and helpful discussions, and to P. Hunt and T. Pearson for preparing the illustrations. P. Bellingham, D. Coomes, J. Mikola, and H, Setala made numerous helpful comments on an early draft, and M. Huston and two anonymous referees provided very thorough reviews of the submitted version that greatly enhanced the clarity and focus of the manuscript. This work was supported by a Non-Specific Output Funding vote from AgResearch, and by The N.Z. Foundation for Science, Research and Technology.


Aarssen, L. W. 1997. High productivity in grassland ecosystems: effected by species diversity or productive species? Oikos 80:183-184.

Aarssen, L. W., and G. A. Epp. 1990. Neighbor manipulations in natural vegetation: a review. Journal of Vegetation Science 1:13-30.

Abdul-Fatih, H. A., and F. A. Bazzaz. 1979. The biology of Ambrosia trifida L. 1. Influence of species removal on the organization of the plant community. New Phytologist 83: 813-816.

Allen, E. B., and R. T. Forman. 1976. Plant species and old-field community structure and stability. Ecology 57:1233-1243.

Anderson, J. M. 1978. Inter- and intra-habitat relationships between woodland Cryptostigmata species diversity and the diversity of soil and litter microhabitats. Oecologia 32: 341-348.

Anderson, J. P. E., and K. H. Domsch. 1975. Measurement of bacterial and fungal contributions to respiration of selected agricultural and forest soils. Canadian Journal of Microbiology 21:314-322.

Anderson, J. P. E., and K. H. Domsch. 1978. A physiologically active method for the quantitative measurement of microbial biomass in soils. Soil Biology and Biochemistry 10:215-221.

Anderson, T.-H., and K. H. Domsch. 1985. Determination of ecophysiological maintenance requirements of microorganisms in a dormant state. Biology and Fertility of Soils 1:81-89.

Andren, O., M. Clarholm, and J. Bengtsson. 1995. Biodiversity and species redundancy among litter decomposers. Pages 141-151 in H. A. Collins, G. P. Robertson, and M. J. Klug, editors. The significance and regulation of soil biodiversity. Kluywer, Dordrecht, The Netherlands.

Armesto, J. J., and S. T. A. Pickett. 1985. Experiments on disturbance in old-field plant communities: impacts on species richness and abundance. Ecology 66:230-240.

Austin, M. P., and T. M. Smith. 1989. A new model for the continuum concept. Vegetatio 83:35-47.

Bardgett, R. D., P. J. Hobbs, and A. Frostegard. 1996. Changes in the structure of soil microbial communities following reductions in the intensity of management of an upland grassland. Biology and Fertility of Soils 22:261-264.

Barker, G. M., and P. C. Mayhill. 1999. Patterns of diversity and habitat relationships in terrestrial mollusc communities in the Pukeamaru Ecological District, northeastern New Zealand. Journal of Biogeography, in press.

Beare, M. H., C. L. Neely, D. C. Coleman, and W. L. Hatgrove. 1991. Characterization of a substrate-induced respiration method for measuring fungal, bacterial and total microbial biomass on plant residues. Agriculture, Ecosystems and Environment 34:65-73.

Bengtsson, J., H. Setala, and D. W. Zheng. 1996. Food webs and nutrient cycling in soil: data, theory and implications. Pages 30-38 in G. A. Polis and K. O. Winemiller, editors. Food webs: interactions of patterns and dynamics. Chapman and Hall, New York, New York, USA.

Bever, J. D. 1994. Feedback between plants and their soil communities in an old-field community. Ecology 75:1965-1977.

Bever, J. D., K. M. Westover, and J. Antonovics. 1997. Incorporating the soil community into plant population dynamics: the utility of the feedback approach. Journal of Ecology 85:561-573.

Bligh, E. G., and W. J. Dyer. 1959. A rapid method of total lipid extraction and purification. Canadian Journal of Biochemistry and Physiology 37:911-917.

Bullock, J. M., B. Clear Hill, M. P. Dale, and J. Silvertown. 1994. An experimental study on the effects of sheep grazing on vegetation change in a species-poor grassland and the role of seedling recruitment into gaps. Journal of Applied Ecology 31:493-507.

Bullock, J. M., B. Clear Hill, J. Silvertown, and M. Sutton. 1995. Gap colonization as a source of grassland community change: effects of gap size and grazing on the rate and mode of colonization by different species. Oikos 72:273-282.

Calloway, R. M. 1995. Positive interactions among plants. Botanical Reviews 61:306-349.

Campbell, B. D., D. A. Wardle, P. W. Woods, T. R. O. Field, D. Y. Williamson, and G. M. Barker. 1996. Ecology of subtropical grasses in temperate pastures: an overview. Proceedings of the New Zealand Grasslands Association 57: 189-197.

Carpenter, S. R., and J. F. Kitchell. 1988. Consumer control of lake productivity. BioScience 38:764-769.

Chapin, F. S., III. 1980. The mineral nutrition of wild plants. Annual Review of Ecology and Systematics 11:233-260.

Collins S. L., A. K. Knapp, J. M. Briggs, J. M. Blair, and E. M. Steinauer. 1998. Modulation of diversity by grazing and mowing in native tallgrass prairie. Science 280:745-747.

De Angelis, D. L. 1992. Dynamics of nutrient cycling and food webs. Chapman and Hall, London, UK.

De Ruiter, P. C., A.-M. Neutel, and J. C. Moore. 1995. Energetics, patterns of interaction strengths and stability in real ecosystems. Science 269:1257-1260.

Demeure, Y., and D. W. Freckman. 1981. Recent advances in the study of anhydrobiotic nematodes. Pages 205-226 in B. M. Zuckerman and R. A. Rohde, editors. Plant parasitic nematodes. Volume III. Academic Press, New York, New York, USA.

Dowling, N. J. E., F. Widdel, and D. C. White. 1986. Phospholipid ester-linked fatty acid biomarkers of acetate-oxidizing sulphate-reducers and other sulphide-forming bacteria. Journal of General Microbiology 132:1815-1825.

Ehrlich, P. R., and A. H. Ehrlich. 1981. Extinction: the causes and consequences of the disappearance of species. Random House, New York, New York, USA.

Eriksson, K. B. 1974. Interspecific variation in Ditylenchus dipsaci. I. Compatability tests with races. Nematologica 20: 147-162.

Federle, T. W. 1986. Microbial distribution in soils: new techniques. Pages 493-498 in F. Megusar and M. Gantar, editors. Perspectives in microbial ecology. Slovene Society for Microbiology, Ljubljana, Slovenia.

Fowler, N. L. 1981. Competition and coexistence in a North Carolina grassland. II. The effects of the experimental removal of species. Journal of Ecology 69:849-854.

Freckman, D. W., and C. H. Ettema. 1993. Assessing nematode communities in agroecosystems of varying human intervention. Agriculture, Ecosystems and Environment 45: 239-261.

Frostegard, A., E. Baath, and A. Tunlid. 1993a. Shifts in the structure of soil microbial communities in limed forests as revealed by phospholipid fatty acid analysis. Soil Biology and Biochemistry 25:723-730.

Frostegard, A., A. Tunlid, and E. Baath. 1993b. Phospholipid fatty acid composition, and activity of microbial communities from two soil types experimentally exposed to different heavy metals. Applied and Environmental Microbiology 59:3605-3617.

Garnier, E., M.-L. Navas, M. P. Austin, J. M. Lilley, and R. M. Gifford. 1997. A problem for biodiversity-productivity studies: how to compare the productivity of multispecific plant mixtures to that of monocultures. Acta (Ecologica 18: 657-670.

Giller, K. E., M. H. Beare, P. Lavelle, A.-M. N. Izac, and M. J. Swift. 1997. Agricultural intensification, soil biodiversity and ecosystem function. Applied Soil Ecology 6:3-16.

Goldberg, D. E. 1987. Neighborhood competition in an old-field plant community. Ecology 68:1211-1233.

Goldberg, D. E., and P. A. Werner. 1983. The effects of size of opening in vegetation and litter cover on seedling establishment of goldenrods (Solidago spp.). Oecologia 60: 149-155.

Goodall, D. W. 1952. Some considerations in the use of point quadrats for the analysis of vegetation. Australian Journal of Science Research (Series B) 5:1-41.

Griffiths, B. S., R. Welschen, J. J. C. M. Arendonk, and H. Lambers. 1992. The effects of nitrate-nitrogen on bacteria and the bacterial-feeding fauna of different grass species. Oecologia 91:253-259.

Grime, J. P. 1997. Biodiversity and ecosystem function: the debate deepens. Science 277:1260-1261.

--. 1998. Benefits of plant diversity to ecosystems: immediate, filter and founder effects. Journal of Ecology 86: 902-910.

Gross, K.L. 1980. Colonization of Verbascum thapsus (Mullein) in an old field in Michigan: the effects of vegetation. Journal of Ecology 68:919-927.

Gurevitch, J., and R. S. Unnasch. 1989. Experimental removal of a dominant species at two levels of soil fertility. Canadian Journal of Botany 67:3470-3477.

Harper, J. L. 1977. Population biology of plants. Academic Press, London, UK.

Harris, W. 1987. Population dynamics and competition. Pages 203-298 in M. J. Baker and W. M. Williams, editors. White clover. CAB International, Wallingford, UK.

Harrison, B. D., and D. J. Hooper. 1963. Longevity of Longidorus elongatus (de Man) and other nematodes in soil kept in polythene bags. Nematologica 9:158-160.

Herben, T., F. Krahulec, V. Hadincova, and S. Pechackova 1997. Fine-scale species interactions of clonal plants in a mountain grassland: a removal experiment. Oikos 78:299-310.

Hill, M. O., and H. Gauch. 1980. Detrended correspondence analysis: an improved ordination technique. Vegetatio 42: 47-58.

Hils, M. H., and J. L. Vanket. 1982. Species removals from a first-year old-field plant community. Ecology 63:705-711.

Hobbie, S. E. 1992. Effects of plant species on nutrient cycling. Trends in Ecology and Evolution 7:336-339.

Hooper, D. U. 1998. The role of complementarity and competition in ecosystem responses to variation in plant diversity. Ecology 79:704-719.

Hooper, D. U., and P. M. Vitousek. 1997. The effects of plant composition and diversity on ecosystem processes. Science 277:1302-1305.

Hooper, D. U., and P. M. Vitousek. 1998. Effects of plant composition and diversity on nutrient cycling. Ecological Monographs 68:121-149.

House, G. J. 1989. Soil arthropods from weeds and crop roots of an agroecosystem in a wheat-soybean rotation: impact of herbicides and tillage. Agriculture, Ecosystems and Environment 25:233-244.

Hurlbert, S. H. 1984. Pseudoreplication and the design of ecological field experiments. Ecological Monographs 54: 187-211.

Huston, M. A. 1994. Biological diversity. The coexistence of species on changing landscapes. Cambridge University Press, Cambridge, UK.

--. 1997. Hidden treatments in ecological experiments: re-evaluating the ecosystem function of biodiversity. Oecologia 110:449-460.

Ingestad, T., and G. I. Agren. 1988. Nitrogen uptake and allocation at steady state nutrition. Physiologia Plantarum 72:450-459.

Ingham, R. E., J. A. Trofymow, E. R. Ingham, and D. C. Coleman. 1985. Interactions of bacteria, fungi, and their nematode grazers on nutrient cycling and plant growth. Ecological Monographs 55:119-140.

Jackson, M. L. 1958. Soil chemical analysis. Constable, London, UK.

Jenkinson, D. S., and J. N. Ladd. 1981. Microbial biomass in soil: measurement and turnover. Pages 415-471 in E. A. Paul and J. N. Ladd, editors. Soil biochemistry. Volume 5. Dekker, New York, New York, USA.

Jenkinson, D. S., and D. S. Powlson. 1976. The effects of biocidal treatment on metabolism in soil. V. A method for measuring soil biomass. Soil Biology and Biochemistry 8: 209-213.

Jonasson, S. 1983. The point intercept method for non-destructive estimation of biomass. Phytocoenologia 11:385-388.

Kotanen, P. M. 1997. Effects of gap area and shape on recolonization by grassland plants with differing reproductive strategies. Canadian Journal of Botany 75:352-362.

Lanterni, A. A., and A. E. Marvaldi. 1995. Graphognathus Buchanan, a new synonymn of Naupactus Dejean, and systematics of the N. leucoloma species group (Coleoptera: Curculionidae). Coleopterist's Bulletin 49:206-228.

Lawton, J. H. 1994. What do species do in ecosystems? Oikos 71:367-374.

Lawton J. H., and V. K. Brown. 1993. Redundancy in ecosystems. Pages 255-270 in E.-D. Schulze and H. A. Mooney, editors. Biodiversity and ecosystem function. Springer-Verlag, Berlin, Germany.

MacArthur, R. H. 1965. Patterns of species diversity. Biological Reviews 40:510-533.

Mann, H. H., and T. W. Barnes. 1953. The mutual effect of ryegrass and clover when grown together. Annals of Applied Biology 40:566-572.

May, B. M. 1966. Identification of the immature forms of some common soil-inhabiting weevils, with notes on their biology. New Zealand Journal of Agricultural Research 9: 286-316.

--. 1993. Larvae of Curculionoidea (Insecta: Coleoptera): a systematic overview. Fauna of New Zealand Number 28. Manaaki Whenua Press, Lincoln, New Zealand.

McGrady-Steed, J., P. M. Harris, and P. Morin. 1997. Biodiversity regulates ecosystem predictability. Nature 370: 162-165.

McKendry, M. V. 1987. Control strategies in high value crops. Pages 329-349 in R. H. Brown and B. R. Kerry, editors. Principles and practice of nematode control in crops. Academic Press, Sydney, Australia.

Merchant, V. A., and D. A. Crossley. 1970. An inexpensive, high efficiency Tullgren extractor for soil microarthropods. Journal of the Georgia Entomological Society 5:84-87.

Mikola, J., and H. Setala. 1998a. No evidence of trophic cascades in an experimental microbial-based soil food web. Ecology 79:153-164.

Mikola, J., and H. Setala. 1998b. Productivity and trophic-level biomasses in a microbial-based soil food web. Oikos 82:158-168.

Mikola, J., and H. Setala. 1998c. Relating species diversity to ecosystem functioning: mechanistic backgrounds and an experimental approach with a decomposer food web. Oikos 83:180-194.

Naeem, S., and S. Li. 1997. Biodiversity enhances ecosystem reliability. Nature 390:507-509.

Naeem, S., L. J. Thompson, S. P. Lawler, J. H. Lawton, and R. M. Woodfin. 1994. Declining biodiversity can alter the performance of ecosystems. Nature 368:734-737.

Nicholas, W. L. 1962. A study of a species of Acrobeloides (Cephalobidae) in nematode culture. Nematologica 8:99109. Odum, E. R 1969. The strategy of ecosystem development. Science 164:262-270.

Oksanen, L., S. D. Fretwell, J. Arruda, and R Niemala. 1981. Exploitation ecosystems in gradients of primary productivity. American Naturalist 118:240-261.

Panetta, F. D., and D. A. Wardle. 1992. Gap size and regeneration in a New Zealand dairy pasture. Australian Journal of Ecology 17:169-175.

Parker, H. L., and P. A. Berry. 1950. Vegetable weevils and their natural enemies in Argentina and Uruguay. U. S. Department of Agriculture, Washington, D.C., USA.

Pastor, J., B. Dewey, R. J. Naiman, P. F. McInnes, and Y. Cohan. 1993. Moose browsing and soil fertility in the boreal forests of Isle Royale National Park. Ecology 74: 467-480.

Pimm, S. L. 1991. The balance of nature. Ecological issues in the conservation of species and communities. University of Chicago Press, Chicago, Illinois, USA.

Platt, W. J., and I. M. Weiss. 1977. Resource partitioning and competition within a guild of fugitive prairie plants. American Naturalist 111:479-513.

Power, M. E. 1992. Top-down and bottom-up forces in food webs: do plants have primacy? Ecology 73:733-746.

Power, M. E., D. Tilman, J. A. Estes, B. A. Menge, W. J. Bond, L. S. Mills, G. Daily, J. C. Castilla, J. Lubchenco, and R. T. Paine. 1996. Challenges in the quest for keystones. BioScience 46:609-620.

Robinson, C. H., J. Dighton, J. C. Frankland, and J. D. Roberts. 1994. Fungal communities on decaying wheat straw of differing resource qualities. Soil Biology and Biochemistry 26:1053-1058.

Rosenzweig, M. L. 1971. Paradox of enrichment: destabilization of exploitation ecosystems in ecological time. Science 171:385-387.

Schoener, T. W. 1989. Food webs from the small to the large. Ecology 70:1559-1589.

Schulze, E.-D., and H. A. Mooney, editors. 1993. Biodiversity and ecosystem function. Springer-Verlag Berlin, Germany.

Setala, H. 1995. Growth of birch and pine seedlings in relation to grazing by soil fauna on ectomycorrhizal fungi. Ecology 76:1844-1851.

Southey, J. F., editor. 1986. Laboratory methods for work with plant and soil nematodes. Her Majesty's Stationery Office, London, UK.

Sulkava, R, and V. Huhta. 1998. Habitat patchiness affects decomposition and faunal diversity: a microcosm experiment on forest floor. Oecologia 116:390-396.

Tate, K. R. 1992. Assessment, based on a climosequence, of soils in tussock grassland, of carbon storage and release in response to global warming. Journal of Soil Science 43: 697-707.

Ten Harkel, M. J., and F. van der Muelen. 1996. Impact of grazing and atmospheric nitrogen deposition on the vegetation of dry coastal dune grasslands. Journal of Vegetation Science 7:445-452.

Ter Braak, C. J. F. 1987. Ordination. Pages 91-173 in R. M. Jongman, C. J. F. Ter Braak, and O. F. R. van Tongeren, editors. Data analysis in community and landscape ecology. Pudoc, Wageningen, The Netherlands.

--. 1988. CANOCO--a FORTRAN program for canonical community ordination by [partial] [detrended] [canonical] correspondence analysis, principal components analysis and redundancy analysis. Agricultural Mathematics Group, Wageningen, The Netherlands.

Thomas, H. 1984. Effects of drought on growth and competitive ability of perennial ryegrass and white clover. Journal of Applied Ecology 21:591-602.

Tilman, D. 1996. Biodiversity: population versus ecosystem stability. Ecology 77:350-363.

Tilman, D., J. Knops, D. Wedin, R Reich, M. Ritchie, and E. Siemann. 1997. The influence of functional diversity and composition on ecosystem processes. Science 277:1300-1302.

Tilman, D., D. Wedin, and J. Knops. 1996. Productivity and sustainability influenced by biodiversity in grassland ecosystems. Nature 379:718-720.

Van de Koppel, J., J. Huisman, R. van de Wal, and H. Olff. 1996. Patterns of productivity along a productivity gradient: an empirical and theoretical investigation. Ecology 77:736-745.

Vandermeer, J. H. 1988. The ecology of intercropping. Cambridge University Press, New York, New York, USA.

Verdejo-Lucas, S., and J. Pinochet. 1992. Population densities of five migratory endoparasitic nematodes in carrot disc cultures. Journal of Nematology 24:96-98.

Visser, S. 1985. Role of the soil invertebrates in determining the composition of soil microbial communities. Pages 297-317 in A. H. Fitter, editor. Ecological interactions in soil. Blackwell Scientific Publications, Oxford, UK.

Vitousek, R M., C. M. D'Antonio, L. L. Loope, M. Rejmanek, and R. Westbrooks. 1997. Introduced species: a significant component of human-induced global change. New Zealand Journal of Ecology 21:1-16.

Vitousek, P. M., and L. R. Walker. 1989. Biological invasion by Myrica faya in Hawai'i: plant demography, nitrogen fixation, ecosystem effects. Ecological Monographs 59: 247-265.

Wardle, D. A. 1995. Impacts of disturbance on detritus food webs in agro-ecosystems of contrasting tillage and weed management practices. Advances in Ecological Research 26:105-185.

--. 1998a. A more reliable design for biodiversity study? Nature 394:30.

--. 1998b. Controls of temporal variability of the soil microbial biomass: a global scale synthesis. Soil Biology and Biochemistry 30:1627-1637.

Wardle, D. A., and G. M. Barker. 1997. Competition and herbivory in establishing grassland communities: implications for plant biomass, species diversity and soil microbial activity. Oikos 80:470-480.

Wardle, D. A., G. M. Barker, K. I. Bonner, and K. S. Nicholson. 1998a. Can comparative approaches based on plant ecophysiological traits predict the nature of biotic interactions and plant species effects in ecosystems? Journal of Ecology 86:405-420.

Wardle, D. A., and A. Ghani. 1995. Why is the strength of relationships between pairs of methods for estimating the soil microbial biomass often so variable? Soil Biology and Biochemistry 27:821-828.

Wardle, D. A., and P. Lavelle. 1997. Linkages between soil biota, plant litter quality and decomposition. Pages 107-124 in G. Gadisch and K. E. Giller, editors. Driven by nature. Plant litter quality and decomposition. CAB International, Wallingford, UK.

Wardle, D. A., K. S. Nicholson, M. Ahmed, and A. Rahman. 1994. Interference effects of the invasive plant Carduus nutans L. against the nitrogen fixation ability of Trifolium repens L. Plant and Soil 163:287-297.

Wardle, D. A., K. S. Nicholson, M. Ahmed, and A. Rahman. 1995a. Influence of pasture forage species on seedling emergence growth and development of Carduus nutans. Journal of Applied Ecology 32:225-233.

Wardle, D. A., H. A. Verhoef, and M. Clarholm. 1998b. Trophic relationships in the soil microfood-web: predicting the responses to a changing global environment. Global Change Biology 4:713-728.

Wardle, D. A., G. W. Yeates, R. N. Watson, and K. S. Nicholson. 1993. Response of the soil microbial biomass and plant litter decomposition to weed management strategies in maize and asparagus cropping systems. Soil Biology and Biochemistry 25:857-868.

Wardle, D. A., G. W. Yeates, R. N. Watson, and K. S. Nicholson. 1995b. Development of the decomposer food-web, trophic relationships, and ecosystem properties during a three-year primary succession in sawdust. Oikos 73:155-166.

Wardle, D. A., O. Zackrisson, G. Hornberg, and C. Gallet. 1997. The influence of island area on ecosystem properties. Science 277:1296-1299.

Wasilewska, L. 1995. Differences in development of soil nematode communities in single- and multi-species grass experimental treatments. Applied Soil Ecology 2:53-64.

Watson, R. N., and R. S. Marsden. 1981. Effects of adult black beetle (Coleoptera: Scarabaeidae) feeding on two grass species. Proceedings of the Third Australasian Conference on Grassland Invertebrate Ecology: 107-115.

Watson, R. N., and N. R. Wrenn. 1980. An association between Paspalum dilatatum and black beetle in pasture. Proceedings of the New Zealand Grassland Association 41: 96-104.

Wedin, D., and D. Tilman. 1993. Competition among grasses along a nitrogen gradient: initial conditions and mechanisms of competition. Ecological Monographs 63:199-229.

West, A. W. 1986. Improvements to the selective respiratory inhibition technique to measure eukaryote: prokaryote ratios in soil. Journal of Microbiological Methods 5:125-138.

West, A. W., and G. P. Sparling. 1986. Modifications to the substrate-induced respiration method to permit measurement of microbial biomass in soils of differing water content. Journal of Microbiological Methods 5:177-189.

White, D. C., W. M. Davis, J. S. Nikels, J. C. King, and R. J. Bobbie. 1979. Determination of the sedimentary microbial biomass by extractable lipid phosphate. Oecologia 40:51-62.

Wiegert, R. G., and D. F. Owen. 1971. Trophic structure, available resources and population density in terrestrial vs. aquatic ecosystems. Journal of Theoretical Biology 30:69-81.

Wilson, E. O., editor. 1988. Biodiversity. Macmillan, New York, New York, USA.

Wilson, J. B. 1988. Shoot competition and root competition. Journal of Applied Ecology 25:279-296.

Wilson, J. B., and S. H. Roxburgh. 1994. A demonstration of guild-based assembly rules for a plant community, and demonstration of intrinsic guilds. Journal of Ecology 83: 801-807.

Wootton, J. T., M. S. Parker, and M. E. Power. 1996. Effects of disturbance on river food webs. Science 273:1558-1561.

Wormald, T. J. 1992. Mixed and pure forest plantations in the tropics and subtropics. FAO (Food and Agriculture Organization), Rome, Italy.

Yeates, G. W. 1987. How plants affect nematodes. Advances in Ecological Research 17:61-113.

Yeates, G. W., T. Bongers, R. G. M. de Goede, D. W. Freckman, and S. S. Georgiva. 1993a. Feeding habits in nematode families--an outline for soil ecologists. Journal of Nematology 25:315-331.

Yeates, G. W., and D. C. Coleman. 1982. Role of nematodes in decomposition. Pages 55-80 in D. W. Freckman, editor. Nematodes in soil ecosystems. University of Texas Press, Austin, Texas, USA.

Yeates, G. W., K. R. Tate, and P. C. D. Newton. 1997. Response of the fauna of a grassland soil to doubling of atmospheric carbon dioxide level. Biology and Fertility of Soils 25:307-315.

Yeates, G. W., and D. A. Wardle. 1996. Nematodes as predators and prey: relationships to biological control and soil processes. Pedobiologia 40:43-50.

Yeates, G. W., D. A. Wardle, and R. N. Watson. 1993b. Relationships between nematodes, soil microbial biomass and weed management strategies in maize and asparagus cropping systems. Soil Biology and Biochemistry 25:869-876.

Zak, D. R., D. Tilman, R. R. Parmenter, C. W. Rice, F. M. Fisher, J. Vose, D. Milchunas, and C. W. Martin. 1994. Plant production and soil microorganisms in late-succes sional ecosystems: a continental-scale study. Ecology 75: 2333-2347.

Zak, J. C., M. R. Willig, D. L. Moorhead, and H. G. Wildman. 1994. Functional diversity of microbial communities: a quantitative approach. Soil Biology and Biochemistry 26: 1101-1108.


(1) AgResearch, Ruakura Agricultural Research Centre, Private Bag 3125, Hamilton, New Zealand

(2) Landcare Research, Private Bag 11052, Palmerston North, New Zealand

(3) School of Biological Sciences, University of Manchester, Oxford Road, Manchester, M13 9PT, UK

Manuscript received 10 October 1997; revised 1 October 1998; accepted 15 October 1998.

(4) Present address: Landcare Research, P.O. Box 69, Lincoln 8152, New Zealand. E-mail:

(5) Present address: Landcare Research, Private Bag 3127, Hamilton, New Zealand.

(6) Present address: Department of Biological Science, Institute of Natural and Environmental Sciences, University of Lancaster, Lancaster LA1 4YQ, UK.
COPYRIGHT 1999 Ecological Society of America
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1999 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Wardle, David A.; Bonner, Karen I.; Barker, Gary M.; Yeates, Gregor W.; Nicholson, Kathryn S.; Bardg
Publication:Ecological Monographs
Geographic Code:1USA
Date:Nov 1, 1999

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters