Are ectomycorrhizas more abundant than arbuscular mycorrhizas in tropical heath forests?

Authors

  • B. Moyersoen,

    Corresponding author
    1. Biology Department, Universiti Brunei Darussalam, Tungku Link BE1410, Brunei Darussalam;
    2. Current address: Université de Liège, Département de Botanique, Sart Tilman B22, 4000 Liège, Belgique;
      Author for correspondence: B. Moyersoen Tel/Fax: +32 43837487 Email:bmoyersoen@hotmail.com
    Search for more papers by this author
  • P. Becker,

    1. Biology Department, Universiti Brunei Darussalam, Tungku Link BE1410, Brunei Darussalam;
    2. Current address: PO Box 367, Bunker, MO 63629, USA;
    Search for more papers by this author
  • I. J. Alexander

    1. University of Aberdeen, Department of Plant and Soil Science, Cruickshank Building, Aberdeen AB24 3UU, UK
    Search for more papers by this author

Author for correspondence: B. Moyersoen Tel/Fax: +32 43837487 Email:bmoyersoen@hotmail.com

Summary

  •  Tropical heath forests have accumulations of slowly decomposing organic matter at the soil surface. To test the hypothesis that ectomycorrhizas are more abundant than arbuscular mycorrhizas under such conditions, we compared the abundance of ectomycorrhizal (EcM) tree species, and the relative proportions of EcM and arbuscular mycorrhizal (AM) roots, in heath forest and lowland rain forest.
  •  The mycorrhizal status of trees in two heath forest and two lowland (mixed dipterocarp) forest plots in Brunei Darussalam was determined by literature and field survey. Fine-root density, proportion of EcM roots, and fractional colonization of EcM and AM roots were measured in monoliths from organic and mineral soil.
  •  There was no difference in the relative basal area abundance (10–41%) of EcM trees, the proportion of EcM roots in monoliths (8–46%), or fractional colonization (90%) of EcM roots, between the two forest types. However, fractional colonization of AM roots was higher (54%) in heath forest than in mixed dipterocarp forest (27%).
  •  Our data do not support the hypotheses that ectomycorrhizas are more abundant in, or determine the floristic composition of, tropical heath forests.

Introduction

Tropical heath forests (HF) grow on acidic, sandy podzolic soils of the Neotropics and Borneo where slowly decomposing organic matter accumulates at the soil surface. They are known locally as caatinga in Venezuela, campina or campinarana in Brazil, wallaba in the Guianas and kerangas in Borneo. Neo-and Palaeotropical HFs differ in floristics, but share structural and physiognomic characteristics distinguishing them from nearby forests developed on highly weathered, clayey kaolisols rich in sesquioxides (Whitmore, 1990). The distinctive characteristics of HF vegetation have been attributed to nutrient deficiency (Specht, 1979), water stress (Whitmore, 1984), an interaction between the two (Bongers et al., 1985; Medina et al., 1990) or H+ toxicity (Proctor, 1999). HF ecosystems in the Neotropics and Borneo are similar in having a thick humus layer, low rates of organic matter decomposition, and N-limited plant growth (Herrera, 1979; Medina & Cuevas, 1989; Coomes, 1997; Moran et al., 2000; Turner et al., 2000).

Most tropical tree species associate with arbuscular mycorrhizas (AM), which improve uptake of inorganic P (Alexander, 1989b). However, rich communities of ectomycorrhizal (EcM) fungi have been described from fruiting body surveys in Amazonian campinarana (Singer & Araujo, 1979), prompting the suggestion that EcM associations might offer particular advantages over HF soils (Singer & Araujo, 1979; Janos, 1980, 1985). Kubitzki (1989) also suggested that the floristic differences between Amazonian forests growing on extremely nutrient-poor white sands, and those on kaolisols, were due to the selective advantage of EcM. This might arise in part from the ability of EcM fungi to access organic sources of phosphorus and nitrogen in soils where mineralization rates are not sufficient to supply plant uptake requirements (Smith & Read, 1997). Singer & Araujo (1979) further suggested that the activity of EcM fungi in campinarana forests might reduce the activity of decomposer fungi and encourage the accumulation of organic matter at the soil surface – the so-called Gadgil effect (Gadgil & Gadgil, 1971, 1975). A comparison of the mycorrhizal status of vegetation between HF and neighbouring mixed forests on kaolisols would be an essential initial test for the importance of EcM as a strategy of HF vegetation to cope with the peculiar soil conditions of HF.

Recent studies have revealed that the relationship between soil characteristics and the relative abundance of EcM and AM tree species in tropical rain forests is complex. In Venezuelan caatinga most of the dominant tree species were AM, but a few EcM tree species were locally codominant in the same area (Moyersoen, 1993). There were some EcM tree species in adjacent mixed forests on kaolisols (Moyersoen, 1993), but the dominant species were AM (Cáceres, 1989). Béreau et al. (1997) did not find EcM on the most abundant tree species in HF in French Guyana, but Henkel (1999) did observe EcM on several species of Dicymbe on Guyana’s border with Brazil. These Dicymbe species are in the characteristically ectomycorrhizal caesalpinioid tribe Amherstiae (Alexander, 1989a) and are co- or monodominants on podzols, white sands and sandy kaolisols over large areas in Guyana (Richards, 1996). No detailed studies on the mycorrhizal status of trees in HF in Borneo have been published, but the EcM families Dipterocarpaceae and Fagaceae are important components of the flora, as are members of Myrtaceae, another family known to contain EcM genera (e.g. Eucalyptus, Tristania). A survey of mycorrhizal infection would clarify the situation and allow comparisons with Neotropical HF.

When EcM trees are present in tropical rain forests, they usually comprise a great proportion of total basal area (e.g. Newbery et al., 1988). Connell & Lowman (1989) suggested several mechanisms that might lead to monodominance of EcM trees in tropical rain forests, based on the supposed advantages of specific host–EcM fungus associations on nutrient-poor soils. However a number of authors (Alexander, 1989a; Torti et al., 1997; Torti & Coley, 1999) have pointed out that the capacity of tree taxa to associate with EcM is not necessary for monodominance in tropical forest. In fact, in most cases where tropical rain forests have a high abundance of EcM trees, AM tree species are either codominant in the canopy (e.g. Newbery et al., 1988; Alexander, 1989a; Moyersoen, 1993) or are an important component of the understorey vegetation (Alexander, 1989a; Hart et al., 1989).

A possible explanation for the co-occurrence of EcM and AM tree species is that different mycorrhizal types occupy different soil niches. Greater colonization of organic soil horizons by EcM relative to AM might be expected because EcM are considered to be advantageous in highly organic soil derived from recalcitrant litter (Northup et al., 1995). However, Moyersoen et al. (1998) found that in west African rain forest EcM and AM root colonization was similar in all soil layers (litter, organic and mineral). On the other hand, the presence in a soil volume of one mycorrhizal type was negatively correlated with the presence of the other type, suggesting competition between mycorrhizal types.

If the conventional hypothesis about the selective advantage of EcM in nutrient-poor podzolic soils is true, we expect EcM to be more abundant in HF than in forests on kaolisols. To test this hypothesis, we compared the relative abundance of EcM and AM trees between HF and neighbouring mixed dipterocarp forests (MDF) on kaolisols in Brunei Darussalam, Borneo. Because both EcM and AM tree species occurred in both HF and MDF we were also able to test the hypotheses that: (1) EcM roots preferentially colonize organic horizons; and (2) there is a competitive relationship between EcM and AM roots in organic horizons.

Materials and Methods

The study areas

Three HF plots and three MDF plots were selected for this study. Badas (4°34′ 05′ N, 114°24′ 53′ E, elevation 11–16 m) and Sawat (4°34′ 37′ N, 114°30′ 11′ E, elevation 11–23 m) HF plots are described in Davies & Becker (1996). They are on Pleistocene terraces with humic podzols. Bukit Patoi HF plot (4°45′ N, 115°10′ E, elevation approximately 250 m) is located on top of a sandstone plateau of the Belait formation. Its soils were described by Brünig (1974) as shallow, sandy, grey-white podzols or shallow humus podzols with frequent sandstone boulders. Andulau MDF plot (4°39′ 26′ N, 114°30′ 57′ E, elevation 37–59 m) was also described by Davies & Becker (1996), and soils and vegetation in the same area were extensively studied by Ashton (1964). Andulau soils are derived from interlaminated clay and sandstone of the Liang formation; they grade from yellow podzols on the hills to somewhat hydromorphic alluvial soils in the valley. Peradayan MDF plot (4°45′ N, 115°10′ E, elevation approximately 200 m) is located between 180 and 210 m up the northern footpath to Bukit Patoi. The soils, derived from the Belait formation, are mainly clayey. Aarhus MDF plot (4°32′ N, 115°10′ E, elevation approximately 250 m) was described by Poulsen et al. (1996). The soils derived from the Setap Shale formation are clayey.

Average annual rainfall was 3000 mm at Badas, Sawat and Andulau plots (Davies & Becker, 1996), and > 4000 mm at Bukit Patoi, Peradayan, and Aarhus plots (Becker, 1992; Poulsen et al., 1996). At all six sites there were fewer rainy periods February–March and July–August (Becker, 1992; Cooper, 1992).

Estimation of EcM and nonEcM basal area in Badas, Sawat, Andulau and Aarhus plots

On-site determination of the mycorrhizal status of all tree taxa is time consuming in tropical rain forests with high diversity, as in our study plots. Information on mycorrhizal status therefore came from observation of selected key taxa (to be described later) and from published surveys in tropical areas (e.g. deAlwis & Abeynayake, 1980; St John, 1980; Norani, 1983; Lee, 1990; Smits, 1994; Reddell et al., 1996), as well as the large dataset compiled by Newman & Reddell (1987). Trees belonging to genera and/or families not typically associated with EcM, as previously reported in the literature, were considered to be nonEcM. The mycorrhizal status of several key tree species was determined by observation after root tracing. Individuals (5–10 cm d.b.h.) were selected at Badas and Sawat using the floristic data base of Davies & Becker (1996), and roots were traced from the bases of 1–4 trees in February and November 1998. Roots were scanned for EcM with a dissecting microscope and EcM status was confirmed by the presence of a Hartig net in root cross-sections. AM status was determined in cleared and stained roots using methods described below.

Data on tree species composition were available for two HF (Badas & Sawat) and two MDF (Andulau & Aarhus) plots from surveys of trees ≥ 5 cm d.b.h. in Badas, Sawat and Andulau plots (Davies & Becker, 1996) and ≥ 10 cm in Aarhus plot (Poulsen et al., 1996).

Root sampling in monoliths

Between 22 February and 23 March 1999, roots were sampled to determine the relative abundance of EcM and nonEcM roots, and the proportion of root tips or length colonized by EcM and AM (so-called fractional colonization), respectively, in different soil horizons in the three HF and three MDF plots. To avoid the possible influence of soil water drainage on mycorrhizal colonization (Lodge, 1989), roots were sampled only in well-drained soil, not in valley bottoms. Roots were sampled in monoliths, from a defined volume of soil in litter (variable depth), organic soil (just below litter, 5 cm depth) and mineral soil (just below organic horizon, 5 or 10 cm depth, depending on root density). In MDF plots with a thin organic layer, ‘organic’ soil samples included a mixture of organic and organically enriched mineral soil. The use of a square metal frame (15 cm each side) with sharp edges, as well as knife and secateurs, minimized soil disturbance. Organic soil samples were stored in plastic boxes of a similar size to the frame to allow for subsampling in the laboratory. Litter and mineral soil samples were stored in polythene bags. Three replicates per soil layer and per plot were taken at random locations. Each replicate from each soil layer was taken in a different location, giving a total number of nine completely independent monoliths per plot.

In each monolith where litter and organic soil samples were taken, litter and organic soil depth were recorded. A subsample of fresh soil was taken from each soil (organic and mineral) core to measure pH in a slurry using distilled water.

Fine root sorting

Roots were carefully washed from the soil, then sorted by size and mycorrhizal type. For organic soil samples, a subsample of 125 cm3 was taken from one corner of the monolith. Soil sample volumes from which roots were separated for the two remaining soil layers varied between 75 and 2250 cm3. All fine roots (generally < 2 mm diameter and apparently not lignified) were separated and divided into EcM and nonEcM groups using a dissecting microscope.

Dipterocarp roots presented distinctive features, including thin diameter, absence of hairs, pale to hyaline colour when young, light to dark brown when older (Lee et al., 1997), and were classified as EcM. Fine roots from common dipterocarp genera traced in a preliminary survey were used as reference material. Fine roots belonging to other EcM plant families were identified by the presence of a fungal sheath. NonEcM roots were defined as fine roots lacking a hyphal sheath. A less frequent category comprised unidentified roots, where EcM colonization assessment was difficult using the dissecting microscope. EcM and nonEcM status of this last category was determined after root staining (to be described later).

Total lengths of EcM and nonEcM roots in each sample were estimated using Tennant’s (1975) modified line intersect method. Root length per sample volume was used to estimate root density. D. wt of EcM (Bukit Patoi and Peradayan plots not included) and nonEcM roots was obtained (see below) and specific EcM and nonEcM root length calculated from the ratio between root length and d. wt.

Scoring of mycorrhizas

Separated EcM and nonEcM root samples were cut into approximately 0.5 cm-long pieces, which were mixed thoroughly and spread on a grid in a Petri dish. To score EcM fractional colonization, a subsample of EcM root pieces was taken in randomly selected squares (either 0.5 or 1 cm2 depending on total root length) to provide a total of c. 250 root tips. Percent of tips colonized by EcM was measured following Hatch’s (1937), method using a dissecting microscope. After scoring, EcM roots were oven dried at 65°C for 3 d before d. wt measurement.

Although we cannot exclude the possibility of AM colonization of EcM roots (Moyersoen & Fitter, 1999), this study focused on AM fractional colonization of nonEcM roots. A subsample of approximately 1 m of nonEcM root pieces was randomly selected as before. The f. wt of this subsample and of remaining roots was measured. The remaining roots were then oven dried at 65°C for 3 d before d. wt measurement. Root subsamples were cleared and stained using a phenol-free modification of Phillips & Hayman (1970) method. Arbuscular mycorrhizas were scored using the magnified intersections method (McGonigle et al., 1990) by inspecting intersections between the microscope eyepiece cross hair and roots at 200 × magnification. Arbuscules, vesicles and hyphal coils were classified as AM. One to three slides were screened for AM, with a range of 3–25 microscopic fields screened per slide.

Statistical analysis

Litter depth, organic soil depth, and pH of organic and mineral soil in the two forest types, and EcM and nonEcM specific root length were compared by t-tests. The effects of forest types, plots within forest types and soil layers on total root density, proportion of EcM in total root density, and EcM and AM fine root colonization were tested in a General Linear Model (Systat vers. 7.0, SPSS, Inc), using a nested crossed design. The relationships between EcM and AM fractional colonization and between EcM and nonEcM root density in the different soil layers were tested by Pearson correlation analyses. To normalize distributions, data on proportion of EcM in total root density and mycorrhizal fractional colonization were arcsine-transformed before further statistical analysis. Throughout this paper, quantitative results are reported as mean ± SE of untransformed data.

Results

Characteristics of litter and organic and mineral soil in monoliths

Litter depth in nine monoliths where litter was sampled was similar in HF and MDF (Table 1). In contrast, the organic layer was more than twice as thick in HF as in MDF (Table 1). The depth of the organic layer at Bukit Patoi was influenced by the presence of rocks, and these data were therefore not included in Table 1. The pHs of organic and mineral soils were similar in the two forest types.

Table 1.  Litter and organic soil layer depth in monoliths and pH in soil cores from two (organic soil layer depth in HF) or three plots in heath forest (HF) and mixed dipterocarp forest (MDF). Values are means (± SE) and probabilities are for t-tests comparing forest types
 HFMDFP*
  • *

    N = 9 for all samples eξcept organic soil depth in HF (n = 6).

Depth (cm)   
Litter 1.4 ± 0.2 0.9 ± 0.20.115
Organic soil26.6 ± 211.2 ± 2.50.001
pH in water slurry   
Organic soil 3.6 ± 0.09 3.8 ± 0.090.062
pH mineral soil 4.0 ± 0.05 4.1 ± 0.040.162

EcM and nonEcM tree abundance

Most EcM tree species belonged to Dipterocarpaceae and Fagaceae. Of the four species and three genera of Myrtaceae sampled in HF plots, EcM were observed only on Tristania beccarii (Table 2). The important undertorey tree species Eugenia bankensis and E. muelleri, as well as Whiteodendron moultonianum were associated only with AM.

Table 2.  Mycorrhizal status of selected species in HF plots, determined by clearing and staining. Species were selected on the basis of their contribution to total tree basal area (ba), their uncertain/unknown mycorrhizal status in the literature and the availability of small individuals for root tracing. Determinations are based on material from one to four individuals from one or more plots
TaxaEcMAM%ba Badas%ba Sawat
Araucariaceae    
Agathis borneensis Warb.NoYes 64.5  3.4
Burseraceae    
Canarium caudatum King.NoYes  0.9  0.1
Clusiaceae    
Calophyllum ferugineum Ridl.NoYes  0.1  2.0
Dipterocarpaceae    
Various spp.YesNo  8.4 33.2
Fagaceae    
Lithocarpus nieuwenhuisiiYesNo  0.2< 0.1
(v. Seem.) A. Camus    
L. pusillus SoepadmoYesNo  0.1< 0.1
Fabaceae    
Copaifera palustris (Sym.) de WitNoYes< 0.1  3.0
Dialium patens BakerNoYes  0.7  0.2
Sindora leiocarpa Backer eξ de WitNoYes  0.4  0.6
Icacinaceae    
Stemonurus umbellatus Becc.NoYes  0.7  1.5
Lauraceae    
Alseodaphne insignis GambleNoNo< 0.1  0.1
Cinnamomum politum Miq.NoNo  0.1  0.1
Meliaceae    
Aglaia glabrata Teijsm. et Binn.NoYes  1.0  0.2
Myrtaceae    
E. bankensis (Hassk.) Back.NoYes  4.9  0.1
Eugenia muelleri Miq.NoYes  1.1  0.8
Tristania beccarii RidleyYesYes  0.1  2.2
Whiteodendron moultonianum    
(W. W. Smith) van SteenisNoYes  0  1.8
Podocarpaceae    
Podocarpus cf. polystachyus    
R. Br. ex Endl.NoYes< 0.1  0
Rosaceae    
Prunus arborea (Bl.) Kalkm.NoYes< 0.1< 0.1
Sapindaceae    
Nephelium lappaceum Linn.NoYes  1.4  0
Sapotaceae    
Isonandra lanceolata WightNoYes  0.9  0
Payena obscura BurckNoYes< 0.1< 0.1
Sterculiaceae    
Heritieria albiflora (Ridl.) Kosterm.NoYes  1.4  0.6

The proportion of total basal area contributed by EcM trees (c. 32–41%) was similar in the two MDF plots (Andulau & Aarhus) and in the Sawat HF plot, but the proportion (c. 10%) was lower in the Badas HF plot (Table 3). Species of Fagaceae and especially Dipterocarpaceae accounted for over 70% of EcM basal area in both HF and MDF (Table 3). Rosaceae, Sapindaceae and Sapotaceae were considered nonEcM, despite reports of EcM in these families (e.g. Trappe, 1962; Singer, 1978; Read & Haselwandter, 1981) because no EcM could be observed after tracing of selected species in Badas and Sawat plots (Table 2).

Table 3.  EcM basal area (ba) as a percentage of total ba, percentage dipterocarp ba in total EcM ba, and (N = 9) EcM roots as a percentage of total fine root density. Data averaged across litter, organic and mineral soil in two HF plots and two MDF plots
Forest typePlot% EcM ba* in total ba% dipterocarp ba* in total EcM ba% EcM root density in total root density
  1. *Lower EcM ba values comprise all Dipterocarpaceae, all Fagaceae and Tristania (Myrtaceae). Upper EcM ba values also include Ulmaceae, Tiliaceae, Rubiaceae, Juglandaceae and Celastraceae. Individuals from these families were not checked by us, but there are records of EcM in these families (Trappe, 1962; Smits, 1994).

HFBadas10–1271–89 8 ± 6.3
 Sawat37–4181–9042 ± 8.4
MDFAndulau33–4178–9829 ± 9.7
 Aarhus32–3390–9546 ± 12.4

EcM and nonEcM root abundance

For any given soil layer, root density was similar in HF and MDF, and total root density was highest in the organic layer in both forest types (Fig. 1, Table 4). Roots from organic layers were sampled at the same depth in both forest types. However, the organic layer was thicker in HF plots, so roots from mineral layers were sampled at greater depths in HF. Total root density varied among plots and was greatest in Andulau MDF (Table 4). The only significant interaction was between soil layer and plots within forest types (Table 4).

Figure 1.

Mean (± SE, n = 9) total (solid square), nonEcM (shaded square) and EcM (open square) fine root densities in litter, organic (0 to −5 cm depth) and mineral soil (0–5 or 0–10 cm depth below organic horizon) in HF and MDF.

Table 4.  Effects of forest types (HF, MDF), plots within forest types, soil layer (litter, organic (0 to −5 cm), mineral (variable depth)) and interactions between these parameters on total fine root density, proportion of EcM fine roots in total fine root density and EcM and AM fractional colonization, based on ANOVAS with nested crossed design. Probability values corresponding to statistically significant effects (P < 0.05) are shown in bold typeface
Total root density  EcM roots proportion EcM colonization AM colonization 
Forest type F1,4 = 0.11P = 0.761 F1,4 = 1.19P = 0.337 F1,4 = 2.34P = 0.201 F1,4 = 14.03P = 0.020
Plots within forest typeF4,36 = 7.20P = 0.000F4,36 = 3.25P = 0.022F4,27 = 0.71P = 0.593F4,36 = 1.25P = 0.308
Soil layerF2,36 = 6.32P = 0.000F2,36 = 0.27P = 0.765F2,27 = 0.33P = 0.720F2,36 = 1.46P = 0.246
Soil layer × forest typeF2,36 = 0.05P = 0.951F2,36 = 0.426P = 0.656F2,27 = 1.14P = 0.334F2,36 = 0.39P = 0.680
Soil layer × plots within forest typeF8,36 = 7.86P = 0.000F8,36 = 0.68P = 0.707F8,27 = 2.11P = 0.07F8,36 = 0.64P = 0.741

The average percentage of EcM roots in total fine root density was 32 ± 6.2% (n = 27) in HF and 48 ± 7.1% (n = 27) in MDF. The only statistically significant difference in this parameter was observed among plots within forest types (Tables 3 and 4). This plot effect occurred because the AM species Agathis borneensis accounted for 64% of stand basal area at Badas (Davies & Becker, 1996). When Badas was excluded from the comparison between HF and MDF, similar proportions of EcM in total fine root density were observed among plots (F3,30 = 1.57, P = 0.217). The proportion of EcM roots in total fine root density was the same in all soil layers and there was no interaction between the proportion of EcM roots and forest types or plots within forest types (Table 4).

There was a negative relationship between EcM and nonEcM fine root density in samples from both organic (r8 = −0.76, P = 0.028) and mineral soil (r9 = −0.73, P = 0.025) layers in which both root types occurred in MDF plots (Fig. 2). This effect was not present in the litter of this forest type, nor in the three soil layers of HF plots.

Figure 2.

The relationship between the density of nonEcM fine roots and EcM fine roots in litter (open triangle), organic (solid square) and mineral (solid circle) soil, from HF (a) and MDF (b).

Specific root length (root length per unit root weight) of EcM roots (37 ± 4.8 cm g−1) was significantly greater than in those of nonEcM roots (27.7 ± 2.69 cm g−1)(t = −2.87, P = 0.006).

Mycorrhizal fractional colonization in EcM and nonEcM roots

EcM fractional colonization of EcM roots was high (88.9 ± 2.49%, n = 46) and similar in HF and MDF (Table 4). Neither plots within forest types, nor soil layers, affected EcM colonization and there were no significant interactions.

Of 2781 microscope fields screened for AM in the nonEcM or unidentified root category, only 13 contained EcM. Of these, five belonged to an unidentified subsample subsequently reallocated to the EcM root category. These results confirmed that most roots classified as nonEcM were indeed uninfected by EcM.

AM fractional colonization of nonEcM root length was 40 ± 3.8% (n = 54). In contrast with EcM colonization, AM fractional colonization was significantly higher in HF (54 ± 4.5%, n = 27) than MDF (27 ± 4.8%, n = 27). No trend in AM colonization was observed either among plots within forests or among soil layers, and there were no interactions (Table 4).

In soil monoliths containing both EcM and nonEcM fine roots, grouped across forest types, there was no correlation between the fractional colonization of EcM (on EcM roots) and the fractional colonization of AM (on nonEcM roots).

Discussion

Abundance of EcM in HF and mixed forests on kaolisols

This study demonstrated that EcM trees do not dominate Bornean HF, reinforcing similar findings in Neotropical HF (Moyersoen, 1993; Béreau et al., 1997). In one case, EcM basal area in HF (Badas) was substantially lower than that in nearby MDF (Andulau) on more clayey and less organic soils. In both forest types, both EcM and nonEcM root density increased in organic layers, but the proportion of EcM roots was similar in litter, organic and mineral soil layers. Fractional colonization of EcM roots was also similar in both forest types and was not affected by soil layers. The hypotheses that EcM are more abundant in HF than in MDF, and that EcM roots are relatively more abundant in organic soil layers in these forests, are therefore not supported by our data. Our observations on the vertical distribution of EcM agree with those of Moyersoen et al. (1998) in the west African rain forest.

We measured fine root densities and fractional colonization of EcM roots because they are estimates of the amount of EcM fine roots per unit soil volume and of the extent of EcM colonization, respectively, but we made no attempt to compare the EcM communities in the two forest types. EcM fungal species differ in morphological (e.g. amount of extramatrical hyphae) and physiological features (e.g. enzymatic capabilities) and this has important functional implications. It is quite possible that the communities of EcM fungi differed between the forest types.

Only a limited number of mycorrhizal surveys in Neotropical HF were available when Janos (1980, 1985) and Kubitzki (1989) suggested that EcM trees dominate on tropical podzols. Of these, the influential paper of Singer & Araujo (1979) was based on fruit body surveys, with only limited observations on EcM root colonization. Recent studies have demonstrated a lack of correlation between the number of fruiting bodies and EcM root colonization (Taylor & Alexander, 1989; Gardes & Bruns, 1996). St John & Uhl (1983) reported EcM on Eperua, an important genus of the Amazonian caatinga. This genus has since been intensively surveyed in several countries in the Neotropics (Moyersoen, 1993; Béreau & Garbaye, 1994, IJ Alexander, unpublished) and found only to form AM.

A number of EcM tree taxa grew in both HF and MDF. Most dipterocarp genera are common to both HF and MDF (Davies & Becker, 1996; Poulsen et al., 1996), although habitat specificity is expected at species level (Ashton, 1964). Similarly, the EcM genera Neea and Guapira (Nyctaginaceae) occur in both Amazonian HF (Moyersoen, 1993; Coomes & Grubb, 1996), mixed forests (St John, 1980) and savannas on white sands. The less frequent EcM trees Aldina kunhardtiana (Caesalpiniaceae) and Coccoloba excelsa (Polygonaceae) in Amazonian Caatinga also occur in mixed forests on kaolisols. The EcM tree genus Dicymbe, although common in HF, also occurs on a wide range of soil types in Guyana (Richards, 1996). In summary therefore, EcM trees occur in a variety of tropical forest ecosystems, and on a wide range of soils.

Abundance of AM in HF and mixed forests on kaolisols

This study demonstrates the importance of AM in tropical HF. A striking characteristic of Badas HF plot was the dominance of A. borneensis. Roots of this species were heavily colonized by AM as previously reported for Agathis australis (Morrison & English, 1967).

A number of other important HF tree species were AM, including Eugenia, an important understorey genus in Bornean HF. This is consistent with the status of Eugenia in Amazonian rain forest (St John, 1980). Eugenia is in Myrtaceae, which includes several EcM genera such as Tristania (Alexander & Högberg, 1986), Backhousia, Lophostemon and Syzygium (Reddell et al., 1996). Myrtaceae also includes dual EcM/AM genera like Eucalyptus (Lapeyrie & Chilvers, 1985), Ixora and Syzygium (Reddell et al., 1996). It is possible that Eugenia and Whiteodendron form EcM in other situations.

It is unlikely that a significant proportion of the roots scored as EcM was also colonized by AM. In the root survey (Table 2) only Tristania showed dual colonization and most EcM trees were dipterocarps (Table 3), which generally form only EcM (Lee, 1990).

AM colonization of nonEcM roots was greater in HF than MDF, but there was no difference in AM colonization among soil layers. Average AM colonization (using scoring methods comparable to those in this study) was also greater in an Amazonian HF than in the nearby mixed forest near San Carlos de Río Negro (HF: 89%, n = 22 tree species, mixed forest: 49%, n = 5 tree species) (Cáceres, 1989; Moyersoen, 1993). Klironomos (1995) reported an increase in AM colonization of Acer saccharum roots in moder-humus on podzols relative to mull-humus on brunisols and luvisols in Canada. Not only was overall colonization greater, but there was a significant increase in proportion of intracellular hyphal coils and vesicles. We did not compare the proportion of those structures between forest types in our study, but the abundance of coils and vesicles in the roots of some species in HF was striking. Clearly the functional role and taxonomic diversity of AM, and the significance of the apparent predominance of Paris-type (Smith & Smith, 1997) infection (characterized by extensive development of intracellular coiled hyphae) in the acidic mor-type organic layers of tropical HF merits further investigation.

The relationship between EcM and AM

There was a negative relationship between the density of EcM and nonEcM roots in the same volume of both the organic and mineral soil layers in MDF. This relationship was not found in the litter. However, fractional EcM colonization (of EcM roots) was unrelated to fractional AM colonization (of nonEcM roots) in the same soil volume. A negative relationship between EcM and AM fractional colonization in mixed root samples from both litter and soil layers was previously found in Korup National Park, west Africa (Moyersoen et al., 1998). This probably reflected a negative relationship between the density of EcM and nonEcM roots. Therefore, in both west African and Bornean rain forests, when a volume of soil was exploited by EcM roots it tended not to be exploited by nonEcM roots. Whether this relationship is determined by the distribution of tree hosts or by root competition still has to be tested, but it is possible that when EcM tree roots colonize soil layers they exclude competing roots. In the Borneo study, EcM fine roots had greater specific root length than nonEcM fine roots. Finely branched root systems with frequent emanating hyphae and strands might explore soil volume more effectively than nonEcM fine roots. Our third hypothesis, that there is a competitive relationship between EcM and AM roots, therefore gains some support from our data.

This negative relationship between EcM and nonEcM roots was not observed in HF, although EcM abundance and fractional colonization (but possibly not EcM fungal species) were similar between the two forest types. On the other hand, AM colonization of nonEcM roots was greater in HF. Hodge et al. (2000) demonstrated, in a pot experiment, that AM colonization can enhance root proliferation. Perhaps high AM fractional colonization makes AM roots more competitive.

Conclusions

For the past 20 yr, mycorrhizal studies in tropical forests on podzols have focused on potentially EcM trees and largely ignored AM species. This study clearly demonstrated that HF soil conditions enhance AM colonization of nonEcM roots. Béreau et al. (1997) have already pointed out the importance of the Guyana region as a reservoir of AM fungal genotypes adapted to extremely poor soil conditions. By demonstrating the importance of AM in Bornean HF, our study showed that this is probably the case in many tropical HFs. Further work on the ecological significance of AM associations in podzols is necessary to better understand the diversity of functions of these associations.

The presumed importance of EcM trees in HF has been used as circumstantial evidence of the important ecological role of EcM in nutrient-poor organic forest soils in the tropics. This study, as well as recent mycorrhizal surveys in the Neotropics, does not support the notion that host EcM status is a factor determining floristic composition of HF.

Acknowledgements

We are grateful to the Forestry Department of Brunei Darussalam for permission to work in forest reserves, to C. Maycock for collaborations at Kuala Belalong Field Studies Centre, Temburong District, Brunei to E. I. Newman for making available the original data used by Newman & Reddell (1987) and to A. H. Fitter for valuable comments on the manuscript. The work was funded by a Postdoctoral Fellowship from Universiti Brunei Darussalam (BM), research grants from Universiti Brunei Darussalam (PB) and the New Phytologist Trust (IJA).

Ancillary