Research

Controls on foliar nutrient and aluminium concentrations in a tropical tree flora: phylogeny, soil chemistry and interactions among elements Faizah Metali1, Kamariah Abu Salim1, Kushan Tennakoon1 and David F. R. P. Burslem2 1

Environmental and Life Sciences Group, Faculty of Science, Universiti Brunei Darussalam, Jalan Tungku Link, Bandar Seri Begawan BE1410, Brunei Darussalam; 2School of Biological

Sciences, University of Aberdeen, Cruickshank Building, St Machar Drive, Aberdeen AB24 3UU, UK

Summary Author for correspondence: Faizah Metali Tel: +673 2463001 Email: [email protected] Received: 10 April 2014 Accepted: 16 July 2014

New Phytologist (2014) doi: 10.1111/nph.12987

Key words: aluminium (Al) accumulators, foliar elemental and Al concentrations, heath forests, mixed dipterocarp forests, phylogeny, soil nutrient concentrations, tropical rain forests.

 Foliar elemental concentrations are predictors of life-history variation and contribute to spatial patterns in biogeochemical cycling. We examined the contributions of habitat association, local soil environment, and elemental interactions to variation in foliar elemental concentrations in tropical trees using methods that account for phylogeny.  We sampled top-soils and leaves of 58 tropical trees in heath forest (HF) on nutrient-poor sand and mixed dipterocarp forest (MDF) on nutrient-rich clay soils. A phylogenetic generalized least squares method was used to determine how foliar nutrient and aluminium (Al) concentrations varied in response to habitat distribution, soil chemistry and other elemental concentrations.  Foliar nitrogen (N) and Al concentrations were greater for specialists of MDF than for specialists of HF, while foliar calcium (Ca) concentrations showed the opposite trend. Foliar magnesium (Mg) concentrations were lower for generalists than for MDF specialists. Foliar element concentrations were correlated with fine-scale variation in soil chemistry in phylogenetically controlled analyses across species, but there was limited within-species plasticity in foliar elemental concentrations. Among Al accumulators, foliar Al concentration was positively associated with foliar Ca and Mg concentrations, and negatively associated with foliar phosphorus (P) concentrations.  The Al-accumulation trait and relationships between foliar elemental and Al concentrations may contribute to species habitat partitioning and ecosystem-level differences in biogeochemical cycles.

Introduction Plants display a wide range of elemental concentrations in leaves (e.g. Peace & Macdonald, 1981; Thompson et al., 1997; Broadley et al., 2003, 2004; Watanabe et al., 2007). For nitrogen (N) and phosphorus (P), the ranges of variation in foliar concentrations in a global data set of 2548 species were c. 32- and 75-fold, respectively (Wright et al., 2004), while the ranges of variation for foliar concentrations of potassium (K), calcium (Ca), magnesium (Mg) and aluminium (Al) were 108-, 764-, 36- and 9443fold, respectively, in a global data set of 1044 species (Metali et al., 2012). The foliar concentrations of nutrients such as N and P are important traits that determine rates of photosynthesis and plant growth (Grime et al., 1997; Lawrence, 2001), and hence define a component of a species’ ecological strategy (Reich et al., 1997; Wright et al., 2004; Poorter & Bongers, 2006). In addition, differential tissue element concentrations may be associated with specific habitats, or with tolerance of nonessential elements or polluted soils (Baker & Brooks, 1989; Ma et al., Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

2001; Olivares et al., 2009; Viani et al., 2014). However, in order to evaluate the functional role of foliar elemental concentrations in defining plants’ ecological strategies or species distributions, it is necessary to determine the contributions of potential physiological, environmental and phylogenetic constraints, which are rarely independent (Grime et al., 1997; Thompson et al., 1997; Poorter, 1999; Aerts & Chapin, 2000; Jansen et al., 2000a,b). Previous studies have suggested that foliar nutrient and Al concentrations are strongly associated with both habitat and phylogeny but have not explicitly accounted for phylogenetic community structure to determine effects related to habitat for plants from a common species pool (Thompson et al., 1997; Jansen et al., 2002; Watanabe et al., 2007; White et al., 2012). A phylogenetic signal has been detected for foliar concentrations of nutrient elements (Thompson et al., 1997; Broadley et al., 2004; Kerkhoff et al., 2006), silicon (Si) (Hodson et al., 2005), and metals including Al, cadmium (Cd), chromium (Cr), manganese (Mn), lead (Pb) and zinc (Zn) (Jansen et al., 2000a, 2002; Broadley et al., 2003, 2007; Fernando et al., 2007; Metali et al., 2012). New Phytologist (2014) 1 www.newphytologist.com

New Phytologist

2 Research

The existence of a phylogenetic signal for these traits implies that physiological processes such as element accumulation and ion transport may differ consistently among species from different clades (Broadley et al., 2001). It is reasonable to propose that these physiological differences may contribute to contrasting patterns of habitat association among taxa, although this has not been explored. Tree distributions and community composition in tropical forests are associated with variation in edaphic factors, in particular nutrients, water and topography (Ashton, 1964; Baillie et al., 1987; Davies et al., 2005; John et al., 2007). The ultisols and oxisols that underlie a high proportion of evergreen lowland tropical forests are often leached and strongly acidic (Vitousek & Sanford, 1986). These conditions potentially give rise to soils with high concentrations of Al3+ ions in soil solution (Whitmore, 1984; Jansen et al., 2002). The most acidic soils in Southeast Asian rain forests are the podsolized white sands that occur beneath heath (‘kerangas’) forest (Proctor et al., 1983a; Whitmore, 1984), which is inherently highly infertile, strongly leached and possesses low concentrations of cations, including Al (Proctor, 1999). However, the effects of differences in soil solution Al and cation concentrations on plant tissue chemistry and species habitat associations are poorly understood. In this study, we tested the hypothesis that foliar Al concentrations are greater for plants growing on soils with higher concentrations of total and exchangeable Al, after accounting for phylogenetic nonindependence among the species represented in our sample. The effects of Al on nutrient uptake by roots, and the relationships among foliar Al and nutrient concentrations, are variable among plants because of differences in the physiological mechanisms underlying ion uptake (Masunaga et al., 1998a; White, 2001). In many Al-intolerant and crop plants, growth limitation by Al toxicity may be related to antagonistic effects of Al on the uptake of other nutrients (Marschner, 2012). Al interferes with the uptake, transport and utilization of several macronutrients (P, K, Ca and Mg), which may induce negative relationships between foliar concentrations of Al and these elements (Foy et al., 1978; Godbold et al., 1988). Negative relationships between foliar concentrations of Al and P, K, Ca and Mg in comparisons across tropical plant species provide evidence that these physiological mechanisms may contribute to variation in foliar elemental concentrations within communities (Osaki et al., 1998, 2003). In Al accumulators, by contrast, the concentration of Al in leaves correlates positively with that of several other elements. Foliar concentrations of Al and Ca, Mg and Zn are positively correlated for Al accumulators of the nutrient-poor Brazilian Cerrad~ao (Haridasan, 1982), while foliar concentrations of Al and Ca, Mg, P, S and Si are positively correlated for Al accumulators in Indonesian tropical rain forest (Masunaga et al., 1998a). Some species adapted to acidic tropical soils respond to Al toxicity by increasing growth rate resulting from the Al-induced stimulation of P uptake (Watanabe & Osaki, 2001, 2002). These findings suggest that interactions with other elements may contribute to variation in foliar Al concentrations and the chemical composition of leaves of species occupying different soil habitats. New Phytologist (2014) www.newphytologist.com

In this study, we determined how habitat associations and local soil conditions affect foliar nutrient and Al concentrations in a sample of 58 tropical tree species growing in two distinct forest types on contrasting soils in Brunei Darussalam. We addressed the following specific questions using methods that controlled for phylogenetic nonindependence within the sample of species. (1) Do foliar concentrations of nutrients (N, P, K, Ca and Mg) and Al differ among specialists of mixed dipterocarp forest (MDF), specialists of heath forest (HF), and generalists? (2) Does variance in foliar concentrations of nutrients and Al within and between species relate to local variation in soil pH and nutrient availability? (3) Are there consistent differences between Al accumulators and non-Al accumulators in the concentrations of nutrient elements? (4) Is the foliar concentration of Al correlated with that of the macronutrient elements N, P, K, Ca or Mg within either Al accumulators or non-Al accumulators?

Materials and Methods Study sites Three 0.96-ha permanent plots representing two different lowland tropical rain forest types in Brunei Darussalam, northwest Borneo, were selected for sampling of foliar elemental concentrations in relation to soil properties. The forest types were MDF at Andulau Forest Reserve (4°390 N, 114°300 E; elevation 37–59 m) and HF, locally referred to as ‘kerangas’, in the Badas (4°340 N, 114°240 E; elevation 11–16 m) and Bukit Sawat Forest Reserves (4°340 N, 114°300 E; elevation 11–23 m). The plots were established in 1996 by identifying and permanently marking all stems ≥ 5 cm diameter at breast height (dbh) on 24 subplots of 20 9 20 m. The soils underlying the Andulau (MDF) are sandy haplic acrisol (clay) with dystric fluvisols (alluvium), whereas the HF soils are albic arenosols, with gleyic podsols in Badas and humic podsols in Sawat (Davies & Becker, 1996). A detailed description of the study sites and an account of their structure and species composition are provided by Metali et al. (2012) and Davies & Becker (1996), respectively. Sampling A total of 322 leaf samples were collected from individuals ≥ 5 cm dbh of 58 species (31 genera in 18 families) which were tagged and mapped on the three plots (Supporting Information Table S1). The criteria for species selection, number of individual trees sampled per species and sample preparation are provided by Metali et al. (2012). A total of 576 soil samples were collected in April–May 2008 using a stratified random sampling strategy. Soil cores (0–15 cm depth) were collected from four randomly selected locations in each 20 9 20 m subplot (n = 24 subplots) of each 0.96-ha plot (n = 3 plots) and samples from 0–5 and 5–15 cm depths were separated. The pH (soil : distilled water in the ratio 1 : 2) of the fresh soil samples was measured in the laboratory using a portable pH meter (Hanna Instruments Ltd, Bedfordshire, UK) on the Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist day of sampling. The samples were then air-dried, ground and passed through a 2.0-mm sieve and transferred to Aberdeen University, Aberdeen, UK, where they were stored at room temperature before chemical analysis. One soil core from each subplot was chosen at random for analysis, yielding a total of 144 samples (24 subplots 9 2 depths 9 3 plots). Chemical analysis Dried leaf and soil samples were ground, digested (c. 0.1 g for the leaf samples and c. 0.2 g for the soil samples) in 4.5 ml of a 1.2 : 1 mixture of 98% H2SO4 and 30% H2O2 and analysed for total N, P, K, Al and Ca (Allen et al., 1989). In addition, acidextractable K and available P concentrations in soils were determined following extraction with 2.5% v/v acetic acid, while exchangeable Al, Ca and Mg concentrations were determined following extraction with 1 M KCl (Allen et al., 1989). Total N and P concentrations in the acid digests were determined colorimetrically using flow injector auto-analysers (FIA, Skalar, Breda, the Netherlands and FIA, Tecator FIAstar 5010 Analyzer, Croydon, UK, respectively). K concentrations were determined using flame emission spectrophotometry (AAnalyst 100; Perkin Elmer, Norwalk, CT, USA) and Al, Ca and Mg concentrations were measured using atomic absorption spectrophotometry after diluting the acid-digested samples with LaCl3 (H2SO4 : LaCl3 in the ratio of 1 : 1). Total C concentrations in soil samples were measured using an automated Fison NA1500 NCS Analyser mass spectrophotometer (Elemental Microanalysis Ltd, Okehampton, UK) on 0.2-g samples. Phylogenetic and statistical analyses All analyses were conducted in R 2.9.2 (R Development Core Team, 2009). During model development, the data were explored to confirm the normality of residuals and homogeneity of variances, and, where necessary, data were log10-transformed. Associations of foliar nutrient and Al concentrations with habitat classifications of tropical forest trees A conservative randomization test based on DeWalt et al. (2006) was used to classify each tree species as a specialist of either MDF or HF or as a generalist (indicating no significant association with either habitat). Species representation on the plots at Badas or Sawat was considered to reflect their association with the HF habitat, and their abundance on the Andulau plot was considered to represent their association with MDF. These tests involved randomly shuffling the habitats assigned to each of the 72 subplots on the three plots, while maintaining the total number of HF and MDF subplots constant (48 and 24, respectively). For each of 1000 randomizations, the observed relative density of each species was compared with its expected relative density generated by randomly shuffling the habitats. The observed relative density for each randomization was the average across all 24 subplots of MDF, or 48 subplots of HF, of the proportion of stems of all species belonging to the target species on that subplot. A species Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

Research 3

was found to be statistically significantly associated with a habitat if its relative density was > 97.5% of the expected relative densities (two-tailed test with a = 0.05). When the observed density was greater or less than the expected density for > 975 of the 1000 randomizations, the species was positively or negatively associated with a habitat, respectively. This analysis was performed initially for the 42 tree species represented by at least six stems on the plots, as this is the minimum sample size that would be required to perform goodness-of-fit v2 tests. Thirty-four species were positively and/or negatively associated with forest habitat type, while the rest (eight species) were classified as generalists. When the randomization test was then repeated for all 58 species, the remaining 16 species with less than six stems on the plots were classified as generalists. However, 10 of these 16 species could be assigned to one of the habitats using published information (Table S2), which suggests that the randomization test for tree species with less than six stems on the plots could not detect the habitat association because of a lack of power. These published sources suggested that one species is confined to HF (Andrews, 2002), six species are confined to MDF (Wong & Madani, 1995; Maschwitz et al., 1996; Ng, 2002; Ashton, 2004; de Wilde & Dufyjes, 2007) and three species occur in both forest types (Soepadmo et al., 1996; Middleton, 2004; Whitmore, 2008). We assumed that the remaining six species are generalists. We used a phylogenetic comparative approach to determine the association between mean species’ foliar Al concentrations and habitat association. The phylogenetic tree was constructed by the method described in Metali et al. (2012). A phylogenetic generalized least squares (gls) model with an Ornstein–Uhlenbeck (OU) model of trait evolution was used to estimate the differences in foliar Al concentrations among species classified as specialists of either MDF or HF or as generalists while accounting for the phylogenetic structure of the species in the three classes (Martins & Hansen, 1997; Paradis, 2006). The OU model assumes that the trait is constrained to evolve similarly to the evolution of traits under stabilizing selection, and it was implemented using the corMartins phylogenetic correlation structure in the APE package (Paradis et al., 2004) of R 2.9.2 (R Development Core Team, 2009). An analysis of variance was used to determine whether there were significance differences in mean species-level foliar elemental concentrations across the three habitat association classes based on maximum likelihood estimation of model parameters. Pair-wise differences in elemental concentrations among habitat association classes were determined using post hoc Tukey’s honest significant difference tests. Associations of foliar nutrient and Al concentrations with soil chemistry The values of soil pH and nutrient concentrations for the 0–5 cm and 5–15 cm depth fractions were averaged for each 20 9 20 m subplot and mean values used to derive a principal component analysis (PCA) of top-soil chemistry across the three plots (a matrix of 72 subplots 9 13 variables). The principal component New Phytologist (2014) www.newphytologist.com

New Phytologist

4 Research

(PC) axes that collectively accounted for 90% of the total variation were identified following Crawley (2007) and the loadings of the soil chemistry variables extracted. The scores (or eigenvectors) for each subplot (n = 72 subplots) along the first six PC axes were used to compute indices defining each species’ representation along the soil chemistry PC axes in terms of the subplots where foliar samples were collected. This index was computed as the average of the PC scores of the subplots, where foliar samples were taken, weighted by the number of samples per species taken per subplot. These indices were used as explanatory variables in phylogenetic gls models to explain among-species variation in concentrations of foliar nutrient and Al concentrations while accounting for the phylogenetic structure of the species sample (Martins & Hansen, 1997; Paradis, 2006). Interactions among explanatory variables were not included in the full model. Backwards selection based on likelihood ratio tests was then used for model simplification to select the subset of the explanatory variables best explaining each response variable (Zuur et al., 2009b). The final model was refitted using restricted maximum likelihood estimation to determine the significance of each explanatory variable. The direct effects of variation in soil nutrients on the respective foliar nutrient concentrations within species were examined by fitting linear regression models for foliar nutrient concentrations as a function of the subplot score along soil PC axis 1 for foliar N, P and Mg concentrations, soil PC axis 2 for foliar Al and K concentrations, and soil PC axis 3 for foliar Ca concentrations. Soil PC axis 2, for example, was chosen for foliar Al and K concentrations because this axis is most strongly associated with total and exchangeable soil Al and K concentrations (see the Results section). A threshold of n ≥ 6 values of foliar nutrient concentrations (equivalent to n ≥ 6 stems) was used to select species for analysis, and this criterion was met for 24 species. Associations of foliar Al with nutrient concentrations The relationship between the foliar concentrations of Al and other nutrients was also determined using a phylogenetic gls method. Foliar concentrations of N, P, K, Ca and Mg were explanatory variables in the maximal model and a corMartins phylogenetic correlation structure was applied. Separate models were derived for all species (n = 58 species), Al accumulators (n = 16 species), defined as species with mean foliar Al concentrations ≥ 2 mg Al g1, and non-Al accumulators (n = 42 species). The classification of species according to the Alaccumulation trait was based on a cluster model analysis (see Metali et al., 2012). Model simplification and the significance of the fixed explanatory variables were determined (Zuur et al., 2009b). Before the backwards selection process, foliar nutrient concentrations were log10-transformed and checked for collinearity (Zuur et al., 2009a,b). When the correlation coefficients between pairs of covariates were ≥ 0.7 they were sequentially removed and Akaike Information Criterion (AIC) values were used to determine which covariate should be included in the maximal model. New Phytologist (2014) www.newphytologist.com

Differences in foliar nutrient concentrations between Al accumulators and non-Al accumulators We used phylogenetic gls models with an OU model of trait evolution to determine the significance of the difference in foliar Al and nutrient concentrations between Al accumulators and nonAl accumulators (see Metali et al., 2012 for Al accumulation status of trees) while accounting for phylogeny (Martins & Hansen, 1997; Paradis, 2006). Estimation was by maximum likelihood and analysis of variance was used for significance testing.

Results Associations of foliar nutrient and Al concentrations with habitat classifications of tropical forest trees In the randomization tests of 42 species with ≥ six stems across all three plots, 34 species (81.0%) showed at least one significant association with either MDF or HF (Table S2). Thirteen species were positively associated with HF and negatively associated with MDF, 21 species were positively associated with MDF and negatively associated with HF, and eight species were not associated with either habitat and were classified as generalists. By contrast, the randomization tests using 58 species across all three plots and published information found that 41 species (70.7%) showed at least one significant association with either MDF or HF (Table S2). Fourteen species were positively associated with HF and negatively associated with MDF, 27 species were positively associated with MDF and negatively associated with HF, and 17 species were not associated with either habitat and were classified as generalists. Habitat classification influenced mean foliar concentrations of Al, N, Ca and Mg in phylogenetically corrected analyses of both samples of species (P < 0.05 in all cases; Table 1) and foliar K concentrations in the core sample of 42 more abundant species. Foliar Al and N concentrations were greater for specialists of MDF than for specialists of HF, while foliar Ca concentration showed the opposite trend (P < 0.05; Figs 1, 2). Generalists had values of foliar Al, N and Ca that were intermediate between those of

Table 1 F and P-values from analyses of variance with phylogenetic correction on foliar aluminium (Al) and nutrient concentrations among habitat categories (specialists of mixed dipterocarp forest and heath forest and generalists) for the 42 species with ≥ six stems across all three plots and all 58 species sampled in Brunei Darussalam 42 species

58 species

Explanatory variable (foliar nutrients)

F2,39

P

F2,55

P

Foliar Al Foliar N Foliar P Foliar K Foliar Ca Foliar Mg

4.29 8.48 0.40 4.05 4.82 5.62

0.021 0.001 0.675 0.025 0.013 0.007

3.88 15.39 0.34 0.48 4.63 3.55

0.027 < 0.001 0.713 0.621 0.014 0.035

P-values in bold are significant (P < 0.05). Ca, calcium; K, potassium; Mg, magnesium; N, nitrogen; P, phosphorus. Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist

–0.5

0.0

ab

log10 foliar N concentration

0.5

b

MDF

HF

0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5

1.0

1.5

(a) Foliar Al

a

–1.0

log10 foliar Al concentration

Research 5

Generalist

MDF

–0.4

log10 foliar K concentration

Generalist

0.7 0.8 0.9 1.0 1.1 1.2 1.3

0.2 0.0

HF

–0.2

a

–0.6

log10 foliar P concentration

a

ab b

MDF

0.6

0.8

1.0

ab

0.2

0.4

b

0.0

log10 foliar Mg concentration

1.0 0.8 0.6 0.4 0.2

log10 foliar Ca concentration

0.0

MDF

HF

Generalist

Habitat classification

specialists of either MDF or HF. Foliar Mg concentrations were lower for generalists than for specialists of MDF (P < 0.05; Figs 1, 2) but not significantly different between generalists and HF species (P > 0.05; Figs 1, 2). For the sample of 42 more abundant species, foliar K concentrations were also lower for generalists than for specialists of MDF (P < 0.05; Fig. 1) but not significantly different between generalists and HF species (P > 0.05; Fig. 1). Associations of foliar nutrient and Al concentrations with soil chemistry MDF soils had a higher pH and greater values of total and exchangeable Al, and total K, than those of HF (Table 2). Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

Generalist

(f) Foliar Mg

a ab

HF

Habitat classification

(e) Foliar Ca b

a

Generalist

(d) Foliar K

a

Habitat classification Fig. 1 Boxplots of log10 foliar: (a) aluminium (Al), (b) nitrogen (N), (c) phosphorus (P), (d) potassium (K), (e) calcium (Ca) and (f) magnesium (Mg) concentrations for 42 tropical tree species in Brunei Darussalam classified as specialists of either mixed dipterocarp forest (MDF; 21 species) or heath forest (HF; 13 species) or as generalists (eight species). Foliar nutrient concentrations (mg Al g1 dry mass) were log10-transformed before analysis. Red horizontal lines display the fitted mean values based on phylogenetic generalized least-squares models, as described in the text. The different letters within a panel denote significant differences among group mean values (P < 0.05).

HF

Habitat classification

(c) Foliar P

MDF

a

b

Habitat classification

a

(b) Foliar N

a

MDF

HF

Generalist

Habitat classification

However, concentrations of total C, N and Mg, exchangeable Ca and Mg, and available P were higher in the HF soils, and concentrations of total Ca and P, and acid-extractable K did not differ between soils from the two forest types. About 93% of the total variation in soil pH and nutrient concentrations was explained by the first six axes of the soil chemistry PCA (Table 3), of which the first and second axes explained 42% and 21% of variation, respectively. Concentrations of total N, total and available P, and total and exchangeable Mg were positively inter-correlated and strongly associated with PC axis 1 (Table 3). Similarly, concentrations of total and exchangeable Al and total K were negatively associated with PC axis 2. The third PC axis was associated with total and exchangeable Ca concentrations, PC axis 4 with pH New Phytologist (2014) www.newphytologist.com

New Phytologist

MDF

HF

0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5

–0.5

0.0

0.5

b

log10 foliar N concentration

ab

1.0

1.5

(a) Foliar Al

a

–1.0

log10 foliar Al concentration

6 Research

Generalist

MDF

Habitat classification

Generalist

MDF

Habitat classification

Generalist

ab

0.8

1.0

1.2

(f) Foliar Mg

a

0.4

0.6

b

MDF

HF

Generalist

Habitat classification

and total C concentration, and PC axis 5 with pH and total C and Al concentrations. A biplot of the first two PC axes showed that MDF and HF subplots were partitioned in ordination space and differentiated on the basis of soil Al and K concentrations (Fig. 3). The MDF subplots were clustered lower on PC axis 2 than the HF subplots, and were therefore associated with subplots possessing greater top-soil concentrations of total Al and K, and exchangeable Al (Table 2; Fig. 3). HF subplots were differentiated on the basis of soil pH and concentrations of total N and C, total and available P, acid-extractable K, and total and exchangeable Ca and Mg, which all varied with PC axis 1. New Phytologist (2014) www.newphytologist.com

Generalist

0.2

0.5

log10 foliar Mg concentration

1.0

ab

0.0

HF

Habitat classification

HF

a

Habitat classification

–0.5

log10 foliar Ca concentration

–1.0

MDF

Generalist

0.0

(e) Foliar Ca b

a

HF

a

0.7 0.8 0.9 1.0 1.1 1.2 1.3

log10 foliar K concentration

0.0 –0.5 –1.0 –1.5

log10 foliar P concentration

a

HF

c

(d) Foliar K

a a

MDF

b

Habitat classification

(c) Foliar P

a

(b) Foliar N

a

Fig. 2 Boxplots of log10 foliar: (a) aluminium (Al), (b) nitrogen (N), (c) phosphorus (P), (d) potassium (K), (e) calcium (Ca) and (f) magnesium (Mg) concentrations for 58 tropical tree species in Brunei Darussalam classified as specialists of either mixed dipterocarp forest (MDF; 27 species) or heath forest (HF; 14 species) or as generalists (17 species). Foliar nutrient concentrations (mg Al g1 dry mass) were log10-transformed before analysis. Red horizontal lines display the fitted mean values based on phylogenetic generalized least-squares models, as described in the text. The different letters within a panel denote significant differences among group mean values (P < 0.05).

After phylogenetic correction, foliar concentrations of all elements except P correlated significantly with one or two of the soil chemistry PC axes (Table 4). Foliar Al concentrations correlated negatively with soil PC axes 2 and 4, which equates to a positive correlation with subplots containing higher concentrations of total and exchangeable Al and higher pH, and a negative correlation with soil C concentrations (Tables 3, 4). Foliar N concentrations correlated negatively with soil PC axis 1, while foliar Ca concentrations showed the opposite trend. These patterns may be indicative of higher foliar concentrations of N in species associated with MDF and of Ca in HF species. Foliar K concentrations were negatively correlated with soil PC axis 5, which is indicative Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist

Research 7

Table 2 Mean pH and total carbon (C; %) and nutrient concentrations (mg g1 dry mass) (mean  SE of mean (SEM)) of soils from mixed dipterocarp forest (MDF) at Andulau and heath forests (HFs) at Badas and Sawat in Brunei Darussalam Soil chemical trait

Habitat

Mean  SEM

t (df = 70)

Total N

MDF HF MDF HF MDF HF MDF HF MDF HF MDF HF MDF HF MDF HF MDF HF MDF HF MDF HF MDF HF MDF HF

0.95  0.31 2.09  0.36 0.06  0.29 0.06  0.30 1.93  0.34 0.11  0.23 0.69  0.31 0.37  0.22 0.15  0.26 0.19  0.25 0.06  0.31 0.11  0.40 0.002  0.374 0.013  0.300 0.04  0.31 0.04  0.26 0.14  0.35 0.02  0.25 0.01  0.32 0.03  0.41 0.02  0.34 0.08  0.49 2.34  0.41 14.42  1.99 4.42  0.03 4.07  0.03

3.99

< 0.001

0.43

0.672

24.85

< 0.001

6.07

< 0.001

2.00

0.500

3.27

0.002

11.96

< 0.001

1.17

0.247

15.98

< 0.001

4.37

< 0.001

5.85

< 0.001

4.25

< 0.001

7.48

< 0.001

Total P Total K Total Al Total Ca Total Mg Available P Acid-extractable K Exchangeable Al Exchangeable Ca Exchangeable Mg Total C (%) pH

Principal component axes

P

All soil nutrient concentrations, except total soil C and soil pH, were log10transformed before analysis but means and SEM were back-transformed and presented. P-values in bold are significant (P < 0.05). Al, aluminium; Ca, calcium; K, potassium; Mg, magnesium; N, nitrogen; P, phosphorus.

of a negative relationship with representation on subplots possessing greater soil C concentrations and higher pH. Foliar concentrations of Mg were most strongly correlated with soil PC axis 4, which reflects a positive relationship with subplots possessing greater soil pH and a negative relationship to total soil C concentrations. Foliar P concentrations were not related significantly to any measured soil chemical property. The soil PC axes summarizing variation in soil chemistry were also used as explanatory variables in linear regression models to test whether localized differences in soil nutrient and Al concentrations explain within-species variation in foliar nutrient and Al concentrations. For all foliar nutrients, these differences were not strongly correlated with local variation in soil nutrient concentrations within species (Table S3). Associations of foliar Al with nutrient concentrations Foliar Ca and Mg concentrations were highly collinear (r = 0.90) for Al accumulators, but not among the non-Al accumulators or for all species combined. Comparison of AIC values suggested that foliar Mg rather than Ca concentration should be removed Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

Table 3 Eigenvalues from principal components analysis (PCA) of 13 soil chemical traits (soil pH and nutrient concentrations) for 24 subplots in mixed dipterocarp forest (MDF) and 48 subplots in heath forests (HFs) and percentage of total variation explained by each principal component axis

Parameters

1

2

3

4

5

6

Eigenvalue % of total variation explained Cumulative % variation explained Loadings of soil chemical traits Total N Total P Total K Total Al Total Ca Total Mg Available P Acid-extractable K Exchangeable Al Exchangeable Ca Exchangeable Mg Total C pH

2.3 41.8

1.7 21.0

1.3 13.2

1.1 8.5

0.8 4.8

0.7 3.4

41.8

62.8

76.1

84.6

89.4

92.8

0.41 0.35 0.16 0.08 0.21 0.39 0.36 0.27 0.11 0.27 0.40 0.16 0.13

0.05 0.20 0.53 0.48 0.01 0.10 0.09 0.31 0.53 0.02 0.03 0.19 0.13

0.01 0.17 0.07 0.03 0.63 0.14 0.28 0.25 0.11 0.56 0.11 0.19 0.20

0.08 0.14 0.03 0.03 0.24 0.03 0.11 0.11 0.02 0.22 0.002 0.60 0.70

0.04 0.11 0.07 0.43 0.02 0.03 0.07 0.04 0.12 0.03 0.03 0.67 0.57

0.12 0.30 0.07 0.53 0.01 0.21 0.32 0.44 0.31 0.04 0.18 0.17 0.32

Loadings and signs of the correlation coefficient (trait loading) of each soil chemical trait for the first six principal component axes are presented. Al, aluminium; C, carbon; Ca, calcium; K, potassium; Mg, magnesium; N, nitrogen; P, phosphorus.

as an explanatory variable. Foliar N and P concentrations were also collinear but the magnitude of the correlation coefficient (r = 0.60) was not sufficient to justify excluding either variable from the list of explanatory variables before model selection (Zuur et al., 2009b). After accounting for phylogeny, foliar Al concentration was negatively correlated with foliar P concentration and positively correlated with foliar Ca concentration in Al accumulators (P < 0.05; Table 5, Fig. 4a,b). Because of the collinearity between foliar Ca and Mg concentrations, foliar Al was also positively correlated with foliar Mg in Al accumulators (P < 0.05; Fig. 4c). There were no relationships between foliar Al and nutrient concentrations among non-Al accumulators or for all species combined (P > 0.05; Table 5). Differences in foliar nutrient concentrations between Al accumulators and non-Al accumulators A phylogenetic gls model suggested that, as expected, Al accumulators and non-Al accumulators differed significantly (F1,56 = 115.90; P < 0.001) in foliar Al concentration (mean  standard error: 8.18  1.11 versus 0.43  1.12 mg g1, respectively), but not in the foliar concentration of N, P, K, Ca or Mg (P > 0.05 in all cases). New Phytologist (2014) www.newphytologist.com

New Phytologist

8 Research 0.2 ++ +++++ + + + + ++ C ++ ++ + + + pH + + + +++ + + Av. P ++ o + + + ++ + Ca + Ex.Ca + Ex. Mg o o ++ + oo + N o + Mg+ oo o o + P ooooo o o o oo + Ex. K

PC axis 2

0.0

–0.2

Al K Ex. Al

oo

–0.4

o

–0.6 –0.2

–0.1

0.0

0.1

0.2

0.3

0.4

0.5

PC axis 1 Fig. 3 Biplot of eigenvectors (scores) for principal component (PC) axes 1 and 2 from principal components analysis (PCA) of 13 top-soil chemistry variables (soil pH and nutrient concentrations) across 24 mixed dipterocarp forest (MDF) subplots and 48 heath forest (HF) subplots. PC axis 1 and PC axis 2 accounted for 41.8% and 21.0% of the total variation, respectively. The arrows show the loadings of each variable on the first two PC axes. Symbols ○ and + represent the soils from MDF and HF, respectively. Aluminium (Al), nitrogen (N), phosphorus (P), potassium (K), magnesium (Mg), carbon (C) and calcium (Ca) are total soil Al, N, P, K, Mg, C and Ca concentrations, respectively; Ex. Al, Ex. Mg and Ex. Ca are exchangeable soil Al, Mg and Ca concentrations, respectively, Ex. K is acid-extractable soil K concentration and Av. P is available soil P concentration. Table 4 Summary of minimum adequate phylogenetic generalized least squares (gls) models accounting for variation in foliar aluminium (Al) and nutrient concentrations among 58 species of tropical forest trees as a function of soil chemistry variables for subplots where they were growing Response variable

Explanatory variables

Coefficients

SE

t (df = 56)

P

Foliar Al

Soil PC axis 2 Soil PC axis 4 Soil PC axis 1 Soil PC axis 4 Soil PC axis 5 Soil PC axis 1 Soil PC axis 4 Soil PC axis 5

0.17 0.40 0.04 0.07 0.10 0.10 0.30 0.16

0.05 0.13 0.01 0.06 0.04 0.02 0.07 0.08

3.26 3.03 3.86 1.24 2.70 4.38 4.07 2.14

0.002 0.004 < 0.001 0.222 0.009 < 0.001 < 0.001 0.037

Foliar N Foliar P Foliar K Foliar Ca Foliar Mg

SE, standard error of the mean; P, probability values. Thirteen soil chemistry traits were summarized using principal component (PC) analysis and PC axis scores were used as explanatory factors in the gls models. P-values in bold are significant (P < 0.05). Ca, calcium; K, potassium; Mg, magnesium; N, nitrogen; P, phosphorus

Discussion Differences in foliar nutrient and Al concentrations between habitat specialists After phylogenetic correction, species in our sample that had distributions biased towards greater abundance in MDF had greater New Phytologist (2014) www.newphytologist.com

foliar Al and N concentrations, but lower Ca concentrations, than specialists of HF. A separate analysis based on the weighted number of individuals sampled per subplot suggested that species with high foliar Al concentrations were associated with subplots containing high total and exchangeable soil Al concentrations, and reinforced the suggestion that foliar N concentrations were greater for species in MDF and that foliar Ca concentrations were greater for species in HF. Similarly, Masunaga et al. (1998b) found that trees in an Indonesian rain forest possessing high concentrations of Al, K, Ca and Mg in bark tissue were more frequent in subplots possessing higher exchangeable soil Al, K, Ca and Mg concentrations, respectively, although the analytical methods employed in that study did not account for phylogenetic nonindependence of the tree species. Concentrations of K and Mg in the bark of these trees were correlated with soil exchangeable K and Mg concentrations, respectively, while bark concentrations of P were negatively correlated with soil available P concentrations (Masunaga et al., 1998c). Habitat differentiation in foliar elemental concentrations could arise either because soil environments select for species with specific foliar nutrient concentrations or because soil chemistry influences phenotypic expression of differential nutrient concentrations. However, our data provide very little evidence that soil Al concentrations directly influence within-species variation in foliar Al concentrations, as this trend was observed in only two out of 24 species. Studies of Al accumulators across a range of environments have found that they retain their ability to accumulate Al even in soils with low Al concentrations (Watanabe et al., 2008). In Brunei, the HF specialist Baccaurea sumatrana (Phyllanthaceae) accumulated a mean Al concentration of 7.5 mg Al g1 in leaves even when growing on the Al-poor HF soils (Table S1; Metali, 2010). Similarly, Callisthene fasciculate (Spr.) Mart. (Vochysiaceae), Qualea grandiflora Mart. (Vochysiaceae) and an unidentified species of Eugenia (Myrtaceae) accumulated Al in leaves (3.3, 2.6 and 7.3 mg Al g1, respectively) when growing both on moderately fertile (i.e. richer in base cations than in Al) and less acidic soils and on the infertile, highly acidic and Al-rich soils of the Brazilian Cerrad~ao vegetation (Haridasan & Ara ujo, 1988). In the Philippines, an unidentified species of Memecylon (Melastomataceae) accumulated Al, despite growing on the ultramafic soils of Mount Bloomfield, which are low in Al and high in pH (Proctor et al., 2000a,b). In these examples, Al accumulation may be related to the maintenance of a low enough rhizosphere pH to permit the existence of Al ions in soil solution (Haridasan & Ara ujo, 1988). A synthesis of research on the drivers of element accumulation in plants concluded that soil element concentrations play a minor role relative to quantitative genetic variation in determining differences in element concentrations among species (Pollard et al., 2002). This perspective supports the conclusion that foliar Al concentration is an intrinsic property of a species that is relatively insensitive to soil conditions within the species’ natural range, and can be regarded, therefore, as a plant trait (Thompson et al., 1997; Jansen et al., 2002; Watanabe et al., 2007; Metali et al., 2012). Species with high foliar Al concentrations were more common in MDF than HF. This association is related to the greater total Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist

Research 9

Table 5 Summary of minimum adequate phylogenetic generalized least squares models describing the relationships between foliar concentrations of aluminium (Al) and five nutrients for (a) 16 species of Al accumulators, (b) 42 species of non-Al accumulators and (c) all 58 species of tropical trees from Brunei Darussalam Tree species

Response variable

Explanatory variables

Coefficients

SE

t

df

P

(a) Al accumulators

Foliar Al

Foliar Al Foliar Al

1.67 3.69 0.82 0.32 0.35 0.84

0.80 0.78 0.23 0.29 0.38 0.62

2.10 4.72 3.54 1.09 0.91 1.37

12

(b) Non-Al accumulators (c) All 58 tree species

Foliar N Foliar P Foliar Ca Foliar Mg Foliar N Foliar N

0.058 < 0.001 0.005 0.298 0.368 0.178

40 56

SE, standard error; df, degrees of freedom; P, probability values. Foliar nitrogen (N), phosphorus (P), potassium (K), calcium (Ca) and magnesium (Mg) concentrations were used as explanatory variables. Al classification of tree species is the result of cluster model analysis (see Metali et al., 2012). P-values in bold are significant (P < 0.05).

(b) Foliar Al and Ca concentrations

–0.50 –0.45 –0.40 –0.35 –0.30 –0.25

log10 foliar P concentration

1.4 1.2 1.0 0.8 0.6 0.4

log10 foliar Al concentration

1.4 1.2 1.0 0.8 0.6 0.4

log10 foliar Al concentration

(a) Foliar Al and P concentrations

–0.5

0.0

0.5

1.0

log10 foliar Ca concentration

1.4 1.2 1.0 0.8 0.6 0.4

Fig. 4 Relationships between log10transformed foliar concentrations of aluminium (Al) and (a) phosphorus (P), (b) calcium (Ca) and (c) magnesium (Mg) for 16 Al-accumulating tropical trees in Brunei Darussalam. The data points are not corrected for phylogeny and the best fit lines were obtained by least-squares linear regressions. Before transformation, foliar nutrient concentrations were expressed in milligrams per gram dry mass. The coefficients and significance values from phylogenetic generalized linear models on these data are presented in Table 5.

log10 foliar Al concentration

(c) Foliar Al and Mg concentrations

0.0

0.2

and exchangeable Al concentrations in soils of MDF. It has been suggested that intense leaching of the sandy podsolized soils underlying Bornean HFs has resulted in a greater loss of Al ions from the soil–plant system than in the more clay-rich soils underlying MDFs (Whitmore, 1984; Proctor, 1999; Jansen et al., 2002). As a result, the acidity of HF soils has become dominated by the high soil solution concentrations of hydrogen ions rather than Al. The availability of Al in HF soils might be further reduced by complexation with organic chelates, which are likely to be abundant in the thick surface organic horizons (Hargrove & Thomas, 1981). The higher soil Al concentrations of MDF Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

0.4

0.6

0.8

1.0

log10 foliar Mg concentration

than in adjacent more acidic HF soils in Brunei is supported by a similar contrast in soils from lowland dipterocarp and HFs at Barito Ulu in Central Kalimantan (Proctor, 1999). These patterns suggest that underlying variation in soil Al concentrations may represent an environmental filter that influences species composition via the phylogenetic and physiological factors that determine plant tissue Al concentration. One outcome of this environmental filter is that the distribution of Al-accumulating trees is biased towards soils with high exchangeable soil Al concentrations. A similar pattern was found for tree species of lowland dipterocarp forest in West Sumatra, New Phytologist (2014) www.newphytologist.com

10 Research

Indonesia, where Al accumulators were more common in subplots with high exchangeable soil Al concentrations (Masunaga et al., 1998b). However, there was no correlation between tree species’ bark Al concentrations and soil Al concentrations in the dipterocarp forests of Indonesia, despite the presence of high exchangeable Al concentrations (0.6 versus 0.1 mg Al g1 in MDF of Brunei) in the top-soils of this forest (Masunaga et al., 1998c). This discrepancy could arise because Masunaga et al. (1998c) sampled bark and not leaves at their study site in West Sumatra, Indonesia. Although elemental concentrations in bark were correlated with those in leaves from 352 trees of 65 tree species in other studies in Sumatra (Masunaga et al., 1998a,b), it is preferable to analyse leaf tissue, because concentrations of Al are higher in leaves, and contamination by dust and soil is more readily mitigated for leaves than for bark by washing with distilled water or ethanol before analysis. Foliar N concentration was greater for species of MDF and therefore varied inversely with the major gradient in total soil N concentration, which was greater in HF soils. This finding corroborates research showing lower total N and P concentrations in leaves taken from an HF in Bako National Park in Sarawak than in lowland rain forests in Brazil and New Britain (Peace & Macdonald, 1981). MDFs in Sarawak, Brunei and Sabah have greater annual fluxes of N via litterfall than adjacent HFs (Proctor et al., 1983b; Moran et al., 2000; Dent et al., 2006), and research in Sabah has reported greater rates of leaf litter decomposition and litter turnover, and higher N and P concentrations in leaf litter, for lowland dipterocarp forests than for HF (Dent et al., 2006). These patterns reflect the lower nutrient status of HF soils, despite their greater total N concentration in the thick surface organic horizon. The oligotrophic status of Bornean HFs is linked to their extreme acidity and the absence of a buffering of soil acidity by Al ions (Proctor, 1999). These factors limit the rate of humus mineralization and litter decomposition, which leaves nutrients, in particular N and P, less available for plants (Grubb, 1977). Water-logging may also contribute to the accumulation of a thick surface organic horizon in some Bruneian HF soils (Moran et al., 2000). In cross-species comparisons across multiple sites, foliar N and P concentrations correlate with maximum photosynthetic rates and relative growth rate (Reich et al., 1997; Wright et al., 2004; Poorter & Bongers, 2006), while plants of nutrient-poor sites maintain low N foliar concentrations and slow growth rates, and extend leaf lifespan in order to conserve nutrients and increase nutrient-use efficiency (Thompson et al., 1997; Vitousek, 1997; Aerts & Chapin, 2000). Our comparison of species specialized to MDF and HF environments in Brunei support these patterns for foliar N but not for foliar P concentrations. Specialists of HF had greater foliar Ca concentrations than specialists of MDF, and an analysis based on the weighted number of individuals sampled per subplot suggested that species with high foliar Ca concentrations were associated with subplots containing greater exchangeable soil Ca concentrations. Similarly, tree species with high bark Ca concentrations in an Indonesian rain forest were more abundant in subplots with greater exchangeable soil Ca concentrations (Masunaga et al., 1998b). HFs in Brunei and Sarawak had greater annual fluxes of Ca via New Phytologist (2014) www.newphytologist.com

New Phytologist leaf litter than adjacent MDFs (Proctor et al., 1983b; Moran et al., 2000), while fluxes of Ca via fine litterfall at a site in Sabah were greater for an HF than for an adjacent dipterocarp forest on sandstone soils, but highest for a dipterocarp forest on alluvial soils (Dent et al., 2006). The greater foliar Ca concentrations in species of HFs may arise because their thicker and more sclerophyllous leaves have a greater demand for Ca to fulfil a structural role in cell walls and membranes (White & Broadley, 2003; Marschner, 2012) and/or because the low soil Al concentrations reduce Al-induced inhibition of Ca uptake. Foliar P concentrations did not differ between specialists of HF and MDF at our study sites in Brunei, despite differences in available P between the soils of these forests. Similar to Brunei, HF soils in Sarawak and Sabah also possess relatively high available P concentrations (Proctor et al., 1983b; Dent et al., 2006). A high proportion of the P in tropical soils is bound to Al and Fe oxides, which are the main products of chemical weathering in these environments (Grubb, 1977; Coomes, 1997). Therefore, the difference in available P concentration between soils of MDF and HF could arise because the available P in soil solution is lowered by irreversible binding with Al ions to a greater extent in the MDFs (Kochian et al., 2004). The lack of a difference in foliar P concentrations suggests that the plants of MDFs are effective at obtaining P from inaccessible pools. This may be facilitated by mycorrhizal colonization (Alexander & Lee, 2005), or other mechanisms, such as the ability of some Al-tolerant plants to exude organic acid anions from the roots, which could detoxify Al and remobilize P (Foy et al., 1987). Associations of foliar Al with nutrient concentrations Foliar concentrations of Al were positively correlated with either Ca or Mg in Al-accumulating trees in Brunei. Similar relationships have been observed for Al-accumulating tree species sampled in the Brazilian Cerrad~ao (Haridasan, 1982) and Indonesian tropical rain forest (Masunaga et al., 1998a) and in tea (Camellia sinensis) growing in solution culture (Fung et al., 2008), although the opposite trend was obtained for Al and Ca concentrations in the roots of tea. Al accumulators in the Rubiaceae show no relationship between foliar Al and Ca concentrations (Jansen et al., 2003). A positive correlation between foliar Al and Ca/Mg concentrations was unexpected because Al can block Ca uptake and compete with Ca for cation binding sites in the cell and cell wall (Marschner, 2012), and total Ca concentrations in soil are not significantly different between the MDF and HF. By contrast, the lack of correlation between foliar Al and Ca or Mg concentrations in non-Al-accumulating plants does not conform to expectation because studies on Al-sensitive plants have shown that Al impairs base cation uptake by roots (Marschner, 2012). The mechanisms for this impairment include Al-induced displacement of base cations in cell walls and membranes, a reduction in root growth and absorptive area for nutrients by Al and interference in the transport of ions to shoots (Bennet et al., 1985; Rengel, 1990; Kinraide et al., 1994; Osaki et al., 1998, 2003; Cumming et al., 2001; Marschner, 2012). The means by which Al-accumulating plants overcome, or even reverse, these Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist negative impacts of Al represent an important element of their nutritional physiology, but remain unknown. In our study there was a negative association between foliar Al and P concentrations in Al accumulators, which implies an inhibition of P uptake by Al. This finding conformed to expectation because Al may fix P at root surfaces and in the rhizosphere, thus inducing P limitation for plants growing in acidic soils (Foy et al., 1978). Al may compete with P, and thus it may lead to inhibition of P uptake when Al is present in excess concentrations (Naik et al., 2009) and Al limits the uptake and translocation of P from root to shoot (Bennet et al., 1985; Naik et al., 2009). However, low foliar P concentrations may not translate into reductions in growth (Bollard, 1983) because Al accumulators possess efficient mechanisms to maintain growth in the absence of high foliar P concentrations (Masunaga et al., 1998a). A positive, rather than negative, correlation between foliar Al and P correlations was found by Masunaga et al. (1998a) for Alaccumulating tropical forest trees in Indonesia, but that study did not correct for the phylogenetic nonindependence among the species sampled. Foliar nutrient (N, P, K, Ca and Mg) concentrations did not differ significantly between Al accumulators and non-Al accumulators. Although this finding contradicts reports of differential foliar concentrations of Ca, P, K, Mg, S, Si, Fe, Mn and Zn between Al accumulators and non-Al accumulators in Brazilian Cerrad~ao (Haridasan, 1982) and Indonesian tropical rain forest (Masunaga et al., 1998a), again these studies are difficult to compare with ours because they did not correct for phylogenetic nonindependence. Further research is required to resolve these contrasting patterns among sites and species samples. Conclusions Specialists of MDF and HF and habitat generalists manifested contrasting foliar concentrations of N, K, Ca, Mg and Al. For foliar nutrients, these differences were not strongly correlated with local variation in soil nutrient concentrations within species, which supports the view that element concentrations are intrinsic traits of species that may contribute to environmental filtering and species distribution. The Al accumulator trait was much more common among specialists of the MDF habitat than among specialists of HF. Although possession of this trait did not affect mean foliar concentrations of any of the five macronutrients, Al accumulators manifested relationships between foliar Al and nutrient concentrations (positive with Ca and Mg; negative with P) that were not present for non-Al accumulators. The functional significance of the Al accumulator trait and relationships to nutrient concentrations in leaves is currently unknown, but these characteristics may contribute to species habitat partitioning and scale up to ecosystem-level differences in biogeochemical cycles.

Acknowledgements This research work was funded by the Brunei government (Brunei In-Service Training Scholarship to F.M.) and the Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

Research 11

University of Aberdeen. We thank the Department of Forestry, Brunei Darussalam and the Universiti Brunei Darussalam for giving us access to work in the three permanent plots in the forest reserves of Brunei Darussalam. We are grateful to David Robinson, Francis Brearley and three anonymous reviewers for their constructive comments on a previous version of this paper.

References Aerts R, Chapin FS. 2000. The mineral nutrition of wild plants revisited: a re-evaluation of processes and patters. In: Fitter AH, Raffaelli DG, eds. Advances in ecological research, vol. 30. London, UK: Academic Press, 1–67. Alexander IJ, Lee SS. 2005. Mycorrhizas and ecosystem processes in tropical rain forest: implications for diversity. In: Burslem D, Pinard M, Hartley S, eds. Biotic interactions in the tropics. Cambridge, UK: Cambridge University Press, 165–203. Allen SE, Grimshaw HM, Parkinson JA, Quarmby C. 1989. Chemical analysis of ecological materials. Oxford, UK: Blackwell Scientific Publications. Andrews S. 2002. Aquifoliaceae. In: Soepadmo E, Saw LG, Chung RCK, eds. Tree flora of Sabah and Sarawak, vol. 4. Kuala Lumpur, Malaysia: Sabah Forestry Department, Forest Research Institute, Sarawak Forestry Department, 1–28. Ashton PS. 1964. Ecological studies in the mixed dipterocarp forests of Brunei state. Oxford Forestry Memoirs 25. Ashton PS. 2004. Dipterocarpaceae. In: Soepadmo E, Saw LG, Chung RCK, eds. Tree flora of Sabah and Sarawak, vol. 5. Kuala Lumpur, Malaysia: Sabah Forestry Department, Forest Research Institute, Sarawak Forestry Department, 61–382. Baillie IC, Ashton PS, Court MN, Anderson AR, Fitzpatrick EA, Tinsley J. 1987. Site characteristics and the distribution of tree species in mixed dipterocarp forest on tertiary sediments in Central Sarawak, Malaysia. Journal of Tropical Ecology 3: 201–220. Baker AJM, Brooks RR. 1989. Terrestrial higher plants which hyperaccumulate metallic elements – a review of their distribution, ecology and phytochemistry. Biorecovery 1: 81–126. Bennet RJ, Breen CM, Fey MV. 1985. Aluminum uptake sites in the primary root of Zea mays L. South African Journal of Plant Soil 2: 1–7. Bollard EG. 1983. Involvement of unusual elements in plant growth and nutrition. In: L€auchli A, Bieleski RL, eds. Encyclopedia of plant physiology, vol. 15B. New York, NY, USA: Springer, 695–755. Broadley MR, Bowen HC, Cotterill HL, Hammond JP, Meacham MC, Mead A, White PJ. 2003. Variation in the shoot calcium content of angiosperms. Journal of Experimental Botany 54: 1431–1446. Broadley MR, Bowen HC, Cotterill HL, Hammond JP, Meacham MC, Mead A, White PJ. 2004. Phylogenetic variation in the shoot mineral concentration of angiosperms. Journal of Experimental Botany 55: 321–336. Broadley MR, White PJ, Hammond JP, Zelko I, Lux A. 2007. Zinc in plants. New Phytologist 173: 677–702. Broadley MR, Willey NJ, Wilkins JC, Baker AJM, Mead A, White PJ. 2001. Phylogenetic variation in heavy metal accumulation in angiosperms. New Phytologist 152: 9–27. Coomes DA. 1997. Nutrient status of Amazonian caatinga forests in a secondary dry area: nutrient fluxes in litter fall and analyses of soils. Canadian Journal of Forest Research 27: 831–839. Crawley MJ. 2007. The R book. Chichester, UK: John Wiley & Sons Ltd. Cumming JR, Swiger TD, Kurnik BS, Panaccione DG. 2001. Organic acid exudation by Laccariabicolor and Pisolithustinctorius exposed to aluminum in vitro. Canadian Journal of Forest Research 31: 703–710. Davies SJ, Becker P. 1996. Floristic composition and stand structure of mixed dipterocarp and heath forests in Brunei Darussalam. Journal of Tropical Forest Science 8: 542–569. Davies SJ, Tan S, LaFrankie JV, Potts MD. 2005. Soil-related floristic variation in a hyperdiversedipterocarp forest. In: Roubik DW, Sakai S, Hamid A, eds. Pollination ecology and rain forest diversity, Sarawak studies. New York, NY, USA: Springer-Verlag, 22–34.

New Phytologist (2014) www.newphytologist.com

12 Research Dent DH, Bagchi R, Robinson D, Majalap-Lee N, Burslem DFRP. 2006. Nutrient fluxes via litterfall and leaf litter decomposition vary across a gradient of soil nutrient supply in a lowland tropical rain forest. Plant and Soil 288: 197–215. DeWalt SJ, Ickes K, Nilus R, Harms KE, Burslem DFRP. 2006. Liana habitat associations and community structure in a Bornean lowland tropical forest. Plant Ecology 186: 203–216. Fernando DR, Woodrow IE, Bakkaus EJ, Collins RN, Baker AJM, Batianoff GN. 2007. Variability of Mn hyperaccumulation in the Australian rainforest tree Gossia bidwillii (Myrtaceae). Plant and Soil 293: 145–152. Foy CD, Chaney RL, White MC. 1978. The physiology of metal toxicity in plants. Annual Review of Plant Physiology 29: 511–566. Foy CD, Lee EH, Wilding SB. 1987. Differential aluminum tolerances of two barley cultivars related to organic-acids in their roots. Journal of Plant Nutrition 10: 1089–1101. Fung KF, Carr HP, Zhang J, Wong MH. 2008. Growth and nutrient uptake of tea under different aluminium concentrations. Journal of the Science of Food and Agriculture 88: 1582–1591. Godbold DL, Fritz E, H€ uttermann A. 1988. Aluminium toxicity and forest decline. Proceedings of the National Academy of Sciences, USA 85: 3888–3892. Grime JP, Thompson K, Hunt R, Hodgsom JG, Cornelissen JHC, Rorison IJ, Hendry GAF, Ashden TW, Askew AP, Band SR et al. 1997. Integrated screening validates primary axes of specialization in plants. Oikos 79: 259–281. Grubb PJ. 1977. Control of forest growth and distribution on wet tropical mountains: with special reference to mineral nutrition. Annual Review of Ecology and Systematics 8: 83–107. Hargrove WL, Thomas GW. 1981. Extraction of aluminum from aluminium-organic matter complexes. Soil Science Society of American Journal 45: 151–153. Haridasan M. 1982. Aluminium accumulation by some Cerrad~ao native species of Central Brazil. Plant and Soil 65: 265–273. Haridasan M, Ara ujo GMD. 1988. Aluminium-accumulating species in two forest communities in the Cerrad~ao region of Central Brazil. Forest Ecology and Management 24: 15–26. Hodson MJ, White PJ, Mead A, Broadley MR. 2005. Phylogenetic variation in the silicon composition of plants. Annals of Botany 96: 1027–1046. Jansen S, Broadley MR, Robbrecht E, Smets E. 2002. Aluminium hyperaccumulation in Angiosperms: a review of its phylogenetic significance. The Botanical Review 68: 235–269. Jansen S, Dessein S, Piesschaert F, Robbrecht E, Smets E. 2000a. Aluminium accumulation in leaves of Rubiaceae: systematic and phylogenetic implications. Annals of Botany 85: 91–101. Jansen S, Robbrecht E, Beeckman H, Smets E. 2000b. Aluminium accumulation in leaves of Rubiaceae: an additional character for the delimitation of the subfamily Rubioideae? International Association of Wood Anatomists Journal 21: 197–212. Jansen S, Watanabe T, Dessein S, Smets E, Robbrecht E. 2003. A comparative study of metal levels of some Al-accumulating Rubiaceae. Annals of Botany 91: 657–663. John R, Dalling JW, Harms KE, Yavitt JB, Stallard RF, Mirabello M, Hubbell SP, Valencia R, Navarrete H, Vallejo M et al. 2007. Soil nutrients influence spatial distributions of tropical tree species. Proceedings of the National Academy of Sciences, USA 104: 864–869. Kerkhoff AJ, Fagan WF, Elser JJ, Enquist BJ. 2006. Phylogenetic and growth form variation in the scaling of nitrogen and phosphorus in the seed plants. American Naturalist 168: 103–122. Kinraide TB, Ryan PR, Kochian LV. 1994. Al3+–Ca2+ interactions in aluminium rhizotoxicity II. Evaluating the Ca2+ -displacement hypothesis. Planta 192: 104–109. Kochian LV, Hoekenga OA, Pi~ neros MA. 2004. How do crop plants tolerate acid soils? Mechanisms of aluminum tolerance and phosphorous efficiency. Annual Reviews in Plant Biology 55: 459–493. Lawrence D. 2001. Nitrogen and phosphorus enhance growth and luxury consumption of four secondary forest tree species in Borneo. Journal of Tropical Ecology 17: 859–869. Ma LQ, Komar KM, Tu C, Zhang W, Cai Y, Kennelley ED. 2001. A fern that hyperaccumulates arsenic. Nature 409: 579. New Phytologist (2014) www.newphytologist.com

New Phytologist Marschner H. 2012. Marschner’s mineral nutrition of higher plants, 3rd edn. London, UK: Academic Press. Martins EP, Hansen TF. 1997. Phylogenies and the comparative method: a general approach to incorporating phylogenetic information to the analysis of interspecific data. American Naturalist 149: 646–667. Maschwitz U, Fiala B, Davies SJ, Linsenmair KE. 1996. A south-east Asian myrmecophyte with two alternative inhabitants: Camponotus or Crematogaster as partners of Macaranga lamellata. Ecotropica 2: 29–40. Masunaga T, Kubota D, Hotta M, Wakatsuki T. 1998a. Mineral composition of leaves and bark in aluminium accumulators in a tropical rain forest in Indonesia. Soil Science and Plant Nutrition 44: 347–358. Masunaga T, Kubota D, William U, Hotta M, Shinmura Y, Wakatsuki T. 1998b. Spatial distribution pattern of trees in relation to soil edaphic status in tropical rain forest in West Sumatra, Indonesia. I. Distribution of accumulating trees. Tropics 7: 209–222. Masunaga T, Kubota D, William U, Hotta M, Shinmura Y, Wakatsuki T. 1998c. Spatial distribution pattern of trees in relation to soil edaphic status in tropical rain forest in West Sumatra, Indonesia. II. Distribution of non-accumulating trees. Tropics 8: 17–30. Metali F. 2010. Factors controlling Al accumulation in plants: effects of phylogeny, soil conditions and external nutrient supply. PhD thesis, University of Aberdeen, Aberdeen, UK. Metali F, Salim KA, Burslem DFRP. 2012. Evidence of foliar aluminium accumulation in local, regional and global datasets of wild plants. New Phytologist 193: 637–649. Middleton DJ. 2004. Apocynaceae. In: Soepadmo E, Saw LG, Chung RCK, eds. Tree flora of Sabah and Sarawak, vol. 5. Kuala Lumpur, Malaysia: Sabah Forestry Department, Forest Research Institute, Sarawak Forestry Department, 1–60. Moran JA, Barker MG, Moran AJ, Becker P. 2000. A comparison of the soil water, nutrient status and litterfall characteristics of tropical heath and Mixed-dipterocarp Forest sites in Brunei. Biotropica 32: 2–13. Naik D, Smith E, Cumming JR. 2009. Rhizosphere carbon deposition, oxidative stress and nutritional changes in two poplar species exposed to aluminum. Tree Physiology 29: 423–436. Ng FSP. 2002. Ebenaceae. In: Soepadmo E, Saw LG, Chung RCK, eds. Tree flora of Sabah and Sarawak, vol. 4. Kuala Lumpur, Malaysia: Sabah Forestry Department, Forest Research Institute, Sarawak Forestry Department, 29– 100. Olivares E, Pena E, Marcano E, Mostacero J, Aguiar G, Benitez M, Rengifo E. 2009. Aluminum accumulation and its relationship with mineral plant nutrients in 12 pteridophytes from Venezuela. Environmental and Experimental Botany 65: 132–141. Osaki M, Watanabe T, Ishizawa T, Nilnond C, Nuyim T, Shinano T, Urayama M, Tuah SJ. 2003. Nutritional characteristics of the leaves of native plants growing in adverse soils of humid tropical lowlands. Plant Foods for Human Nutrition 58: 93–115. Osaki M, Watanabe T, Ishizawa T, Nilnond C, Nuyim T, Sittibush C, Tadano T. 1998. Nutritional characteristics in leaves of native plants grown in acid sulfate, peat, sandy podzolic, and saline soils distributed in Peninsular Thailand. Plant and Soil 201: 175–182. Paradis E. 2006. Analysis of phylogenetics and evolution with R. New York, NY, USA: Springer. Paradis E, Claude J, Strimmer K. 2004. APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20: 289–290. Peace WJH, Macdonald FD. 1981. An investigation of the leaf anatomy, foliar mineral levels and water relations of trees of a Sarawak forest. Biotropica 13: 100–109. Pollard AJ, Powell KD, Harper FA, Smith JAC. 2002. The genetic basis of metal hyperaccumulation in plants. Critical Reviews in Plant Sciences 21: 539–566. Poorter L. 1999. Growth responses of 15 rain-forest tree species to alight gradient: the relative importance of morphological and physiological traits. Functional Ecology 13: 396–410. Poorter L, Bongers F. 2006. Leaf traits are good predictors of plant performance across 53 rain forest species. Ecology 87: 1733–1743. Proctor J. 1999. Heath forests and acid soils. Botanical Journal of Scotland 51: 1–14. Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

New Phytologist Proctor J, Anderson JM, Chai P, Vallack HW. 1983a. Ecological studies in four contrasting lowland rain forests in Gunung Mulu National Park, Sarawak. I. Forest environment, structure and floristics. Journal of Ecology 71: 237–260. Proctor J, Anderson JM, Fogden SCL, Vallack HW. 1983b. Ecological studies in four contrasting lowland rain forests in Gunung Mulu National Park, Sarawak. II. Litterfall, litter standing crop and preliminary observations on herbivory. Journal of Ecology 71: 261–283. Proctor J, Baker AJM, Bruijnzeel LA, van Balgooy MMJ, Fairweather GM, Madulid DA. 2000a. Foliar chemistry and leaf herbivory on Mount Bloomfield, Palawan, Phillipines. Botanical Journal of Scotland 52: 79–89. Proctor J, Baker AJM, van Balgooy MMJ, Bruijinzeel LA, Jones SH, Madulid DA. 2000b. Mount Bloomfield, Palawan, Philippines: forests on greywacke and serpentinized peridotite. Edinburgh Journal of Botany 57: 121–139. R Development Core Team. 2009. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. ISBN 3-900051-07-0. [WWW document] URL http://www.R-project.org. Reich PB, Walters MB, Ellsworth DS. 1997. From tropics to tundra: global convergence in plant functioning. Proceedings of the National Academy of Sciences, USA 94: 13730–13734. Rengel Z. 1990. Competitive Al3+ inhibition of net Mg2+ uptake by intact Lolium multiflorum roots. II. Plant age effects. Plant Physiology 93: 1261–1267. Soepadmo E, Julia S, Rusea G. 1996. Fagaceae. In: Soepadmo E, Wong KM, Saw LG, eds. Tree flora of Sabah and Sarawak, vol. 2. Kuala Lumpur, Malaysia: Sabah Forestry Department, Forest Research Institute, Sarawak Forestry Department, 1–118. Thompson K, Parkinson JA, Band SR, Spencer RE. 1997. A comparative study of leaf nutrient concentrations in a regional herbaceous flora. New Phytologist 136: 679–689. Viani RAG, Rodrigues RR, Dawson TE, Lambers H, Oliveira RS. 2014. Soil pH accounts for differences in species distribution and leaf nutrient concentrations of Brazilian woodland savannah and Seasonally Dry forest species. Perspectives in Plant Ecology, Evolution and Systematics 16: 64–74. Vitousek PM. 1997. On regression and residuals: response to Knops et al. Oecologia 116: 557–559. Vitousek PM, Sanford RL Jr. 1986. Nutrient cycling in moist tropical forest. Annual Review of Ecology and Systematics 17: 137–167. Watanabe T, Broadley MR, Jansen S, White PJ, Takada J, Satake K, Takamatsu T, Tuah SJ, Osaki M. 2007. Evolutionary control of leaf element composition in plants. New Phytologist 174: 516–523. Watanabe T, Misawa S, Hiradate S, Osaki M. 2008. Characterization of root mucilage from Melastoma malabathricum, with emphasis on its roles in aluminium accumulation. New Phytologist 178: 581–589. Watanabe T, Osaki M. 2001. Influence of aluminum and phosphorus on growth and xylem sap composition in Melastoma malabathricum L. Plant and Soil 237: 63–70. Watanabe T, Osaki M. 2002. Mechanisms of adaptation to high aluminum condition in native plant species growing in acid soils: a review. Communications in Soil Science and Plant Analysis 33: 1247–1260. White PJ. 2001. The pathways of calcium movement to the xylem. Plant and Soil 248: 257–268. White PJ, Broadley MR. 2003. Calcium in plants. Annals of Botany 92: 487–511. White PJ, Broadley MR, Thompson JA, McNicol JW, Crawley MJ, Poulton PR, Johnston AE. 2012. Testing the distinctness of shoot ionomes of

Ó 2014 The Authors New Phytologist Ó 2014 New Phytologist Trust

Research 13 angiosperm families using the Rothamsted Park Grass Continuous Hay Experiment. New Phytologist 196: 101–109. Whitmore TC. 1984. An introduction to tropical rain forests. Oxford, UK: Oxford University Press. Whitmore TC. 2008. The genus Macaranga: a prodromus. Kew, UK: Royal Botanic Gardens Kew. de Wilde WJJO, Dufyjes BEE. 2007. Polygalaceae. In: Soepadmo E, Saw LG, Chung RCK, Kiew R, eds. Tree flora of Sabah and Sarawak, vol. 6. Kuala Lumpur, Malaysia: Sabah Forestry Department, Forest Research Institute, Sarawak Forestry Department, 221–295. Wong KM, Madani L. 1995. Anisophylleaceae. In: Soepadmo E, Wong KM, eds. Tree flora of Sabah and Sarawak, vol. 1. Kuala Lumpur, Malaysia: Sabah Forestry Department, Forest Research Institute, Sarawak Forestry Department, 15–26. Wright IJ, Reich PB, Westoby M, Ackerly DD, Baruch Z, Bongers F, Cavender-Bares J, Chapin T, Cornelissen JHC, Diemer M et al. 2004. The worldwide leaf economic spectrum. Nature 428: 821–827. Zuur AF, Ieno EN, Elphick CS. 2009a. A protocol for data exploration to avoid common statistical problems. Methods in Ecology & Evolution 1: 1– 12. Zuur AF, Ieno EN, Walker NJ, Saveliev AA, Smith GM. 2009b. Mixed effects models and extensions in ecology with R. New York, NY, USA: Springer.

Supporting Information Additional supporting information may be found in the online version of this article. Table S1 Mean foliar potassium (K), aluminium (Al), calcium (Ca), magnesium (Mg), nitrogen (N) and phosphorus (P) concentrations for 58 tropical tree species sampled in the mixed dipterocarp forest (Andulau) and heath forests (Badas and Sawat) in Brunei Darussalam Table S2 Tropical tree species analysed for habitat associations (MDF, HF or generalist) in Brunei Darussalam using randomization tests Table S3 Summary of linear regression models describing the effects of variation in soil chemistry on foliar Al and nutrient concentrations within species Please note: Wiley Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

New Phytologist (2014) www.newphytologist.com

Controls on foliar nutrient and aluminium concentrations in a tropical tree flora: phylogeny, soil chemistry and interactions among elements.

Foliar elemental concentrations are predictors of life-history variation and contribute to spatial patterns in biogeochemical cycling. We examined the...
495KB Sizes 0 Downloads 10 Views