Leaf pH as a plant trait: species-driven rather than soil-driven variation

Authors

  • Johannes H. C. Cornelissen,

    Corresponding author
    1. Systems Ecology, Department of Ecological Science, Faculty of Earth and Life Sciences, VU University Amsterdam, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands
    Search for more papers by this author
  • Florus Sibma,

    1. Systems Ecology, Department of Ecological Science, Faculty of Earth and Life Sciences, VU University Amsterdam, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands
    Search for more papers by this author
  • Richard S. P. Van Logtestijn,

    1. Systems Ecology, Department of Ecological Science, Faculty of Earth and Life Sciences, VU University Amsterdam, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands
    Search for more papers by this author
  • Rob A. Broekman,

    1. Systems Ecology, Department of Ecological Science, Faculty of Earth and Life Sciences, VU University Amsterdam, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands
    Search for more papers by this author
  • Ken Thompson

    1. Department of Animal and Plant Sciences, University of Sheffield, Western Bank, Sheffield, South Yorkshire, UK
    Search for more papers by this author

Correspondence author. E-mail: hans.cornelissen@falw.vu.nl

Summary

1. Interspecific variation in plant functional traits is fast becoming popular as a tool for understanding and predicting ecosystem biogeochemistry as dependent on vegetation composition. Leaf pH has recently been shown to be a promising new candidate trait for this purpose. But how robust is leaf pH as a species trait in the face of environmental variation? We hypothesized that inherent interspecific variation in leaf pH should be greater than phenotypic variation of given species in response to soil environments.

2. We tested this hypothesis in a temperate herbaceous flora by growing 23 species experimentally in three soils of contrasting pH (ranging by almost three pH units) and related chemistry.

3. As predicted, there was large and consistent variation in leaf pH among these species, which was robust to the differences between soil types. Indeed both the species rankings and the absolute species values for leaf pH were remarkably constant in comparisons between soil types.

4. The fact that a given species can maintain a leaf pH very different from that of their soil environment, combined with the great interspecific variation in leaf pH, indicates that leaf pH really is largely a species-specific trait. Linked with recent field evidence we suggest that interspecific variation in leaf pH, while easy and cheap to assess, has important predictive power of biogeochemical properties and processes in ecosystems.

Introduction

There is broad support for the critical roles that vegetation composition plays in ecosystem functions such as carbon, nutrient and water cycling, in processes like regeneration and succession and ecosystem properties such as stress resistance and resilience to disturbance (Vitousek et al. 1987; Hobbie 1992; Chapin et al. 1997; Hättenschwiler & Vitousek 2000; Grime 2001; Farley, Jobbagy & Jackson 2005; Hooper et al. 2005; Wardle et al. 2008). More specifically, these roles depend importantly on the differences in functional traits among plant species in ecosystems (Grime 2001; Craine et al. 2002; Lavorel & Garnier 2002; Eviner & Chapin 2003; Diaz et al. 2007). Measuring plant traits on multiple species is now very popular, and large trait databases are being broadly employed as a tool to formulate and test predictions about changes (differences) in ecosystem functioning as driven by differences or shifts in vegetation makeup (e.g. Reich, Walters & Ellsworth 1991; Grime et al. 1997; Díaz et al. 2004; Wright et al. 2004; Cornwell et al. 2008; Kleyer et al. 2008). In addition to previous shortlists of popular traits for such analyses (Weiher et al. 1999; Westoby et al. 2002; Cornelissen et al. 2003), foliar pH was recently introduced as a potential new candidate trait for its predictive power of (and partly its causal link to) carbon cycling processes, e.g. herbivory, litter decomposition, and mycorrhizal symbiosis (Cornelissen et al. 2006). Indeed, leaf pH may have ‘afterlife effects’ on leaf litter acidity (Cornelissen et al. 2006), which in turn can drastically alter the pH of the soil organic matter (Grubb, Green & Merrifield 1969; Finzi, Canham & van Breemen 1998). This can easily be measured and observed in pine plantations, the acidic needle litter of which may significantly acidify the original soil organic matter within decades. For example, Zinke (1962) found that the soil pH close to Pinus contorta trunks was 5·7 and increased to 7·2 further away from them. Leaf pH depends on a myriad of underlying chemistry, where for instance higher basic cation concentrations tend to increase (see Lang et al. 2009 for lichens) and organic acid concentrations decrease it. Indeed, interspecific variation in leaf pH in a subarctic flora corresponded (negatively) with that in C/N ratio, where carbon content probably represented organic acids (Cornelissen et al. 2006). Variation in organic acids probably also explains the consistent interspecific variation in acidity of wide-ranging subarctic bryophytes (Soudzilovskaia et al. 2010). Recently tissue pH of leaves, fine roots and fine stems alike was found to form part of a ‘plant economics spectrum’ of coordinated whole-plant traits relevant to carbon and nutrient cycling in subarctic ecosystems (Freschet et al. 2010).

However, how far is leaf pH really a species trait? Or, how much of the variance in leaf pH can be attributed to the genetic makeup of species and how much to intraspecific variation, e.g. through phenotypic plasticity? The possible contributing factors to variation in tissue pH of a given species include (i) diurnal or seasonal physiological variation, an extreme example of which may be nocturnal pH reduction through malic acid formation as a step in the photoassimilation of CAM plants (Lambers, Pons & Chapin 2008), and (ii) several environmental factors that control a plant’s physiology and foliar chemistry (e.g. ion uptake by roots). The availability of soil cations, particularly calcium, is well known to be a controlling factor for leaf cation concentration (Rorison & Robinson 1984), with likely consequences for leaf pH. To understand any feedback effects of vegetation pH composition to other biota and soil chemistry, and to predict changes therein with local or global environmental changes, we need to unravel interspecific and intraspecific patterns of variation in leaf pH. This is the first study to do this, testing the hypothesis that variation in leaf pH is predominantly determined by overall inherent differences among species and to a lesser degree by phenotypic variation within species. We do this experimentally by simultaneously growing a broad range of temperate herbaceous species in a greenhouse at three contrasting soil pH and chemistry regimes and subsequently comparing their leaf pH values.

Materials and methods

Seed provenances and germination treatment

We used seeds mostly collected from several ecosystems in the Sheffield area, United Kingdom, including limestone grassland, heathland and grassland on acidic sandstone, woodlands and ruderal sites (see Grime, Hodgson & Hunt 2007). We collected seeds of a few additional species in the river IJssel floodplain and ruderal sites near Zwolle, The Netherlands (Table S1). Together they cover a broad range of herbaceous plant strategies in terms of competitiveness, stress tolerance and ruderality in response to fertility and disturbance regimes (Grime, Hodgson & Hunt 2007). They also show a broad taxonomic range, with both monocots (Poaceae, Cyperaceae) and eudicots (several other families) well represented.

Seeds of some species were stratified (moist, at 4 °C) or scarified with sand paper (hard-coated Lotus corniculatus) based on our prior knowledge about any dormancy. Seeds of all species were then left to germinate on moist filter paper in petri-dishes in the greenhouse (environmental regime see below). Once sufficient numbers of a seed population had germinated, the seedlings were transferred to 10 × 15 cm trays filled with 3–4 cm of continuously moist, chemically inert quarried river sand (Imabo, Aalsmeer, The Netherlands). Although there were interspecific differences in germination timing among species, these differences were very small compared to the duration of the plant growth period.

The experiment

After a few days of acclimatizing in the trays, the seedlings were transferred into 275 mL plastic plant pots with drainage holes. Each pot stood in its own saucer and was watered (with demineralized water) daily if and when needed to ensure there was always some water left in the saucer. For most species we planted five seedlings per pot, but for some species known to grow slowly (Hunt & Cornelissen 1997) these numbers were doubled or tripled in order to grow sufficient green biomass per pot during the course of the experiment, which ran from 25 January through 15 March 2007. We employed three treatments in terms of soil pH regime. The ‘Higher soil pH’ treatment comprised a 50–50% mixture (by volume) of inert river sand (see above) and base-rich potting compost (Jongkind, Aalsmeer, The Netherlands); the ‘Lower soil pH’ a 50–50 mixture of the river sand and an acidic peat compost from northern Sphagnum bogs (Turba Rubia; Agroterra, Madrid, Spain); the ‘Medium soil pH’ treatment comprised a 25–25–50% mixture of potting compost, peat compost and river sand. All combinations were thoroughly mixed before use.

Each pot received five aliquots of 20 mL of a nutrient stock, which contained ammonium nitrate and potassium phosphate dissolved in demineralized water, over the course of the experiment (approx. once a week, starting 2 days after planting up the seedlings). This added up to cumulative amounts of 5 g m−2 N, 1·25 g m−2 K and 0·5 g m−2 P.

The experimental design consisted of five blocks in a semi-climatized glasshouse at VU University Amsterdam, each of which contained one pot for each soil treatment for each species, with all pots randomized and regularly reshuffled within blocks (Fig. 1). Each block had additional PAR lighting provided by one of five large lamps that were suspended about 0·8 m above the pots (Fig. 1), and which were turned on for 12 h a day. These lamps elevated the light regime to above 1000 μEinstein m−2 s−1 at plant height right below the lamp on sunny days, but on two partially cloudy measurement days in February and March the minimum light intensities ranged between 60 and 200 and the maxima between 100 and 420 μEinstein m−2 s−1 as dependent on individual lamps, pot position relative to the lamps and time of measurement. The temperature was monitored in the glasshouse throughout the experiment. The mean daily average was 16·8 °C, the mean daily minimum 11·2 and the mean daily maximum 22·4 °C.

Figure 1.

 The experimental plants growing in a randomized blocked design in the greenhouse, with additional lighting.

Soil chemistry

We took soil samples both before initial fertilization and halfway through the experiment. For each soil pH sampling event, three 10 g samples per soil type were mixed with 25 mL demineralized water, then shaken for 2 h and centrifuged (5 min, 1275 g). The supernatant was measured with a WTW Inolab pH meter (see below). To measure soil base cation content, approx. 500 mg of soil was mixed with 2 mL HCl HNO3 to digest for 4 h at 140 °C. Then 8 mL of demineralized water was added and undissolved soil was left to sediment for half a day. Subsequently, the supernatant was carefully decanted and measured for calcium (Ca) and magnesium (Mg) by atomic absorption spectroscopy under addition of 1% LaNO3, and potassium (K) by atomic emission spectroscopy (both: 1100B Spectrometer; Perkin Elmer Inc., Waltham, MA, USA).

Leaf measurements

Around 15 March all plants that looked healthy, without symptoms of major stress (e.g. stunted growth, leaf colours) were harvested. Such symptoms were common in the Lower soil pH treatment and indicated to us that the plants were unlikely to have survived for much longer, let alone reproduce. In a few borderline cases leaves of several remaining plants from different blocks were pooled to make one bulk sample (see Results; not used in statistical analyses). For several species, including some strictly calcicolous ones such as Centaurea scabiosa and Leontodon hispidus (Grime, Hodgson & Hunt 2007), all plants had died in that treatment, leaving lower species numbers for analysis (see Fig. 2). At harvest all leaves in each pot were cut off and kept fresh until pH measurement soon after (see below). The methodology broadly followed Cornelissen et al. (2006). From each sample we took one subsample for fresh and one for dry measurement. Each fresh subsample was manually chopped into approx. 1 mm pieces and mixed with demineralized water in a 2·5 mL eppendorf tube (volume ratio 1 : 8). After 1 h of shaking at 250 rpm the solution was centrifuged at 12000 g and the supernatant measured using a narrow (5 mm diameter) SenTix Mic electrode connected to an Inolab Level 2 pH meter (both: WTW, Weilheim, Germany). We calibrated the pH meter against buffer solutions (pH 4 and 7) before each measurement series. Because fresh measurements had to be done very soon after harvest, with little time difference between samples within and between blocks, there was not sufficient time to process all species. Alternate subsamples were finely ground in a ball mill (Mixer Mill MM 200; Retsch, Haan, Germany), then dried at 60 °C for 48 h. One part was immersed in water and measured for pH as described for fresh measurements (see also Cornelissen et al. 2006). The other part was used for C and N analysis by dry combustion in a Carlo Erba NA1500 (CE Elantech Inc., Rodana, Italy) elemental analyser. We wanted to test whether the data for pre-dried leaves could represent data for fresh leaves, since the pre-drying treatment allows for preparing and testing larger numbers of samples without the time constraints of measurements on fresh leaves, and because the pre-dried and finely milled material can be used for other chemical analyses as well (see Cornelissen et al. 2006).

Figure 2.

 Leaf pH of wide-ranging temperate herbaceous species grown in three soil types of contrasting pH. (a) Fresh leaf pH, two-way anova results: soil type, F2 = 0·30, P = 0·97; species, F14 = 148, P < 0·001, interaction, F22 = 1·95, P = 0·010. (b) Pre-dried leaves, soil type, F2 = 1·25, P = 0·30; species, F22 = 124, P < 0·001, interaction, F34 = 1·60, P = 0·025. Species abbreviations: Am, Achillea millefolium; Ac, Agrostis capillaris; Ao, Anthoxanthum odoratum; Ae, Arrhenatherum elatius; Bp, Brachypodium pinnatum; Cs, Centaurea scabiosa; Ca, Chamerion angustifolium; Dg, Dactylis glomerata; Dp, Digitalis purpurea; Eh, Epilobium hirsutum; Ev, Eriophorum vaginatum; Fo, Festuca ovina; Ga, Galium aparine; Km, Koeleria macrantha; Lh, Leontodon hispidus; Lp, Lolium perenne; Lc, Lotus corniculatus; Pr, Papaver rhoea; Pl, Plantago lanceolata; Pa, Poa annua; Ra, Rumex acetosella; Tc, Tanacetum vulgare; To, Taraxacum officinale.

Data analysis

We employed one-way Analyses of Variance and subsequent Games-Howell post-hoc tests to compare means of soil pH, soil base cation content and soil C %, respectively, between the three soil types (fixed factor), all on untransformed data (Levene’s tests: no significant heterogeneity of variance). Using two-way anovas, we tested for the effects of soil type (fixed factor) versus species (random factor) on (untransformed) fresh and pre-dried leaf pH respectively. Levene’s tests showed heterogeneity in variance, but scatterplots showed no trend between the predicted values and the standardized residuals, thus satisfying the main requirement for applying anova. We employed linear regression to test for cross-species relationships between (fresh or pre-dried) leaf pH of plants grown at higher soil pH versus plants grown at medium or lower soil pH; between fresh leaf pH and pre-dried leaf pH for each of the soil types; and between leaf C/N ratio and leaf pH for each of the soil types.

Results

Measured soil pH (one-way anova, F2 = 205, P < 0·001) and soil base cation contents (one-way anova, F2 = 66·8, P < 0·001) followed the a priori order of soil types: Higher > Medium > Lower soil pH (Table 1), while soil carbon content was not related to soil type (one-way anova, F2 = 2·38, P = 0·13). Soil carbon was generally rather low, owing perhaps to a combination of organic matter mineralization and the very low initial weight fraction of peat or compost compared to sand in the mixtures, which were standardized by volume.

Table 1.   Soil chemistry and leaf pH in three soil types
 Soil pHBase cation (μmol g−1)C %Leaf pH (fresh)Leaf pH (pre-dried)
  1. Soil pH (water extraction) is the average ± SE over three dates during the plant growth period (i.e. with fertilizer added), where five subsamples per treatment were pooled on each date (Values in parentheses represent initial soil pH before fertilization). Mean Base cation content (K + CA + Mg, μmol g−1) and mean C % over five blocks per soil treatment; for C% each block replicate was based on three samples. For sample size for leaf pH see Fig. 1. Different superscript letters indicate significantly different means in post-hoc tests.

Higher soil pH6·51a ± 0·16 (5·77 ± 0·07)66·9x ± 1·01·72 ± 0·285·60 ± 0·175·80 ± 0·13
Medium soil pH5·22b ± 0·06 (3·78 ± 0·07)48·8y ± 2·61·92 ± 0·315·60 ± 0·155·76 ± 0·14
Lower soil pH3·67c ± 0·03 (2·76 ± 0·01)39·5z ± 1·02·80 ± 0·495·50 ± 0·245·86 ± 0·24

Across wide ranges of herbaceous species there was highly significant heterogeneity in both fresh and pre-dried leaf pH due to species, while soil type had no significant effect (Fig. 2). The interaction effect was significant but very small compared to the species effect. These patterns were confirmed by the close cross-species correspondence of fresh leaf pH between the Higher soil pH treatment and either the Medium (n = 15, R2 = 0·996, Y = 0·89X + 0·63) or Lower soil pH treatments (n = 9, R2 = 0·967, Y = 0·85X + 0·82). Similarly strong correspondence was seen for pre-dried leaf pH (Fig. 3; Higher versus Medium, n = 23, R2 = 0·987, Y = 1·02X + 0·17; Higher versus Lower soil pH, n = 13, R2 = 0·929, Y = 1·03X + 0·25). Indeed, mean leaf pH across species was not significantly different between soil types (Table 1). For the three species where we pooled leaves from the few surviving (and relatively poorly developed) plants from several blocks in the Lower soil pH treatment in order to obtain sufficient material for one bulk sample (Achillea millefolium, Plantago lanceolata, Tanacetum vulgare), leaf pH differed by maximum 0·1 (Achillea millefolium), 0·2 (Plantago lanceolata) and 0·1 units (Tanacetum vulgare) from mean leaf pH at Medium or Higher soil pH.

Figure 3.

 Leaf pH (pre-dried) of herbaceous species grown at Higher soil pH versus at Medium (n = 23, R2 = 0·987) or Lower soil pH (n = 13, R2 = 0·929).

Fresh and pre-dried leaf pH were themselves strongly related across species, in any given soil type, although the strengths of the linear regression coefficients were strongly influenced by the presence or absence of the extremely acidic species, Rumex acetosella (Table 2).

Table 2.   Linear regressions of leaf pH of pre-dried tissues (X) versus leaf pH of fresh tissues (Y), at three different soil pH regimes
 R2nSignificanceRegression equation
  1. Results in parentheses are without the strongly acidic Rumex acetosella (see text).

  2. n, number of species; NS, not significant.

Higher soil pH0·87 (0·29)9 (8)P < 0·001 (NS)Y = 0·92X + 0·32
Medium soil pH0·87 (0·45)15 (14)P < 0·001 (P < 0·01)Y = 0·87X + 0·00
Low soil pH0·88 (0·47)15 (14)P < 0·001 (P < 0·01)Y = 0·99X + 0·09
Overall0·87 (0·40)39 (36)P ≤ 0·001 (P ≤ 0·001)Y = 0·93X ± 0·27

Interspecific ranking of leaf C/N ratio was robust to differences between the three soil types (Higher versus Medium soil pH, n = 23, R2 = 0·890, Y = 1·01X + 2·28; Higher versus Lower soil pH, n = 13, R2 = 0·667, Y = 0·82X + 5·22). However, the negative relationships between (fresh) leaf pH and leaf C/N across species were strongly driven by Rumex acetosella, and were no longer significant with this species excluded (at Higher soil pH, n = 15, R2 = 0·30, P < 0·05, without R. acetosella R2 = 0·014, not significant; Medium soil pH, n = 15, R2 = 0·30, P < 0·05, without R. acetosella R2 = 0·016, NS; Lower soil pH, n = 9, R2 = 0·72, P < 0·01, without R. acetosella R2 = 0·0034, NS). For pre-dried leaf pH versus leaf C/N there were no significant regressions either with (all R2 < 0·024 for all species per soil type) or without R. acetosella (all R2 < 0·15).

Discussion

Among 23 wide-ranging temperate herbaceous species we obtained strong support for our hypothesis: variation in leaf pH is predominantly determined by overall inherent differences among species rather than by phenotypic variation of each species in response to differences in soil chemistry. This implies that variation in leaf pH between different species growing in soils varying in chemical composition may be a robust and useful predictor of species effects on biogeochemical cycling processes. For instance, it provides a potentially easy and powerful tool for predicting species composition effects on soil acidification (Finzi, Canham & van Breemen 1998), plant N and P uptake modes (via mycorrhizal types), susceptibility to herbivory and litter decomposability (Cornelissen et al. 2006). Below we will discuss the possible mechanisms involved in controlling species leaf pH, and the extent to which our findings may or may not be generalized beyond our study.

As all plants in our experiment were grown from seed, we can be confident that leaf pH was not substantially determined by responses to conditions prior to the experiment. Therefore, our results clearly demonstrate that leaf pH really is a species trait to a large extent, at least for the range in soil pH between 3·7 and 6·5 and in soil base cations between 40 and 70 μmol g−1 in our study. These findings corroborate the consistent interspecific variation in leaf pH reported from a subarctic study with similar (but not experimentally controlled) soil pH range (Cornelissen et al. 2006). We have to be careful, however, in extrapolating these findings to for instance calcium- or magnesium-rich soils, with soil pH ≥ 7. Dicot herbs in particular have been shown to take up and transport Ca and Mg partly passively through the apoplast and therefore may accumulate these base cations in rather high concentrations on particularly base-rich soils (Rorison & Robinson 1984; Marschner 1995; Thompson et al. 1997). This could partly explain why there was no obvious correlation between leaf pH of 18 of our species in the experiment and leaf base cation concentrations measured on the same species in the Sheffield flora, partly sampled from base-rich soils (K. Thompson & J. H. C. Cornelissen, unpublished data, see Cornelissen & Thompson 1997; Thompson et al. 1997). Also, woody species were excluded from our study because of methodological constraints (slow recruitment and establishment, ontogenetic patterns, special mycorrhizal requirements). However, woody species include several important taxa that extend particularly into the lower leaf pH range, e.g. ericoid and ecto-mycorrhizal species and gymnosperms (Cornelissen et al. 2006). Rumex acetosella turned out to be the only truly acidic species in our study and Anthoxanthum odoratum the only truly high-pH species. On the other hand, the fact that leaf pH of both species was robust to differences in soil chemistry suggests that the strong species control of leaf pH extends along a broad range of at least four pH units. Indeed, leaf pH did not differ significantly between soil types, neither within given species nor across our broad range of herbaceous species.

Our results, in combination with those of a subarctic flora (Cornelissen et al. 2006; Freschet et al. 2010), suggest that tissue pH itself is tightly controlled for a given species, because of its direct or indirect functions in the plant. For instance, low pH corresponded with poor digestibility and may therefore act as an antiherbivore defence in the same subarctic flora (Cornelissen et al. 2006). On the other hand, multiple interacting aspects of a plant’s physiology and chemistry in and between different cell compartments together determine the overall pH of a tissue (Marschner 1995; Lambers, Pons & Chapin 2008). Indeed, tissue pH is a somewhat artificial measure in this respect, since different cell compartments such as cytoplasm, vacuoles and cell walls may all have a different pH (P. M. Ray, pers. comm.; Lambers, Pons & Chapin 2008). Thus, it is possible that plants need to be physiologically and chemically plastic in response to environmental variation in order to maintain optimal pH for processes (e.g. enzyme activity) in different cell compartments and thereby keep the pH of whole leaves rather constant.

Our findings may have important implications for vegetation feedbacks to soil chemistry and processes that are mediated by plant pH (see also Grubb, Green & Merrifield 1969; Finzi, Canham & van Breemen 1998). By creating a different leaf pH than the soil they grow in, as we have demonstrated here for the first time, and given the cross-species correspondence of green leaf pH and leaf litter pH (Cornelissen et al. 2006), plant species can alter the pH of the soil they grow in. This also suggests that the decline in soil pH that is often seen over long-term chronosequences (e.g. Read 1991; Wardle et al. 1997) may not only be a function of gradual abiotic soil acidification by weathering agents including rainwater, but also of biotic acidification by the accumulation of organic matter derived from the different plant species during succession.

Acknowledgements

We thank Stuart Band for collecting seeds in the Sheffield area, Jurgen van Hal for advice and assistance in the greenhouse and Jacintha Ellers and Will Cornwell for useful discussion points.

Ancillary