Should Phosphorus Availability Be Constraining Moist Tropical Forest Responses to Increasing C02 Concentrations?

J. Lloyd Summary 95

Max Planck Institute lor 1. Introduction 96

Biogeochemistry „t-.ii • . „ -i r i „ t ■ rr, •

lei 1a Germany Phosphorus in the Soils ot the Moist Tropics 96

M I Bird States and Fluxes of Phosphorus in Moist Tropical Forests 98

Australian National University 4. Linking the Phosphorus and Carbon Cycles 103

Canberra, Australia References 109

E.M. Veenendaal

Harry Oppenheimer Okavango

Research Centre Maun, Botswana

B. Kruijt

Alt en a Green World Research

Foundation Wageningen, The Netherlands

Moist tropical forests account for a substantial amount of global plant productivity. And several lines of evidence suggest that they may be sequestering significant amounts of anthropogenically released carbon at the present time. But there are also indications that the productivity of many of these forests is limited by low phosphorus availability. This has led to suggestions that moist tropical forests may be constrained in their ability to increase their growth rates in response to increases in atmospheric carbon dioxide concentrations. This notion is examined in this chapter.

Several factors should prevent low levels of available phosphorus significantly constraining moist tropical forest [C02]/growth responses. One of the main reasons for low soil-solution P concentrations in many tropical soils is the adsorption of most of the phosphate ions onto iron and aluminum oxides and clay minerals. This adsorption is, to a large extent, reversible. This means that, in response to increased rates of removal of P from the soil solution, such as would be required to sustain faster plant growth with increasing [CO,], phosphate ions should be desorbed from their fixation sites and released into the soil solution, thus maintaining the concentration of P in the soil solution at a more-or-less constant level. This contrasts with the situation for nitrogen in temperate and boreal forests, where the rate of entry of nitrogen into the soil solution is closely linked to the rate of carbon mineralization.

The nature of P mineralization in soils is a second factor mediating towards phosphorus availability not constraining tropical forest [C02] responses. This is because, unlike nitrogen, phosphorus is mineralized independent of carbon in most soils. Thus, it has less potential to be "locked up" in the larger soil carbon pool that should occur as a result of increased plant productivity at higher [CO,].

Third, most of the available evidence suggests that, at a given soil P concentration, plants growing at elevated [C02] are capable of maintaining their tissue phosphorus concentrations. This is in contrast to nitrogen and occurs because of the positive effects of larger root systems on the extent of root mycorrhizal colonization, root organic acid efflux per plant, and root acid phosphatase activity. All three processes play important roles in phosphorus acquisition.

An additional phenomenon may also be important in the tropical forest C/P interaction. Humic molecules and organic acids actively compete with phosphorus for soil fixation sites. This means that increases in soil carbon density at higher [CO 2j may serve to displace phosphate ions from sorption sites and into the soil solution, where they can then be utilized by plants. It is not inconceivable that this effect could give rise to a "runaway" positive feedback: CO,-induced increases in tropical forest plant growth


Copyright 0 2001 by Academic Press. Ail rights of reproduction in .my form reserved.

giving rise to increases in soil carbon content, which in turn liberates previously adsorbed phosphorus — this in turn giving rise to even more substantial increases in plant growth.

1. Introduction

By virtue of their large area and year-round favorable growing conditions, moist tropical forests may account for as much as 50% of the global net primary productivity (Grace et al, 2000). This high productivity combined with reasonably long carbon residence times means that tropical forests are likely to be a substantial component of the terrestrial sink for anthropogenic CO, (Lloyd and Farquhar, 1996; Lloyd, 1999a; Malhi et al., 1999).

Although this notion contrasts with some global [CO,]/5l3C inversion studies (e.g., Ciais et al., 1995), it has received experimental support from eddy covariance studies above undisturbed forests in Brazil (Grace et al, 1995, 1996; Malhi et al., 1998) and from biomass inventory data (Phillips et al, 1998). It is also in accordance with some synthesis atmospheric inversions (Enting et al, 1995; Rayner et al, 1999) and current interpretations of the rate of increase and latitudinal gradients in atmospheric 02/N2 ratios (Keeling et al, 1996). In both Keeling et al (1996) and Rayner et al (1999), the carbon balance of tropical regions appears to be more or less netural, despite the fact that many tropical regions are significant sources of anthropogenic CO, because of deforestation associated with land-use change or shifting cultivation. The magnitude of this source is substantial, currently an estimated 0.1-0.2 Pmol C year-1 (Houghton, 1996). By simple mass balance, then, undisturbed tropical regions must be substantial sinks for anthropogenically released C02 of about the same amount.

But the dogma also has it that tropical forests are usually found on heavily weathered soils that are low in nutrients, particularly phosphorus. In a manner similar to possible interactions between the nitrogen and carbon cycles in temperate and boreal forests (McGuire et al, 1995) it has thus been suggested that low P availability may limit the extent to which tropical forests are able to increase their productivity in response to increases in atmospheric C02 concentrations (Friedlingstein et al, 1995; McKane et al, 1995).

We therefore examine here in some detail the relationships between the biogeochemical cycling of carbon and phosphorus in moist tropical forests. Our main purpose is to examine whether low P availability should actually be constraining the ability of moist tropical forests to increase their productivity and carbon stocks as a consequence of increasing atmospheric [C02j.

2. Phosphorus in the Soils of the Moist Tropics

That tropical soils are highly weathered and infertile is a generalization at best. Indeed, only about 50% of the total tropical soil area can be considered to consist of highly weathered leached soils such as oxisols, ultisols, and alfisols (Sanchez, 1976; Richter and ie. ultisol Babbar, 1991). But when one considers the moist tropical regions with an annual precipitation greater than 1500 mm only, these soils account for about 75% of the total area (Sanchez, 1976).

That such soils should be phosphorus deficient is a prediction from pedogenic theory (Walker and Syers, 1976). This is because, in contrast to carbon, nitrogen, and sulfur, P is cycled mainly on geological time scales. That is to say, the only substantial primary source of P for plants is from the weathering of parent material at the base of the soil. As soil development proceeds, there is a loss of this weathered P as a consequence of leaching. The rate of leaching is quite small on an annual basis, even in the tropics (typically 0.1-1 mmol P irT2 year-1: Bruijnzeel, 1991) but it occurs over several thousand years. Moreover, as soils become older, not only does the total amount of phosphorus decline, but there is also a transfer of phosphate from labile pools to nonlabile pools (Walker and Syers, 1976).

The chemistry of soil phosphorus transformations giving rise to this situation is complex (Sanyal and DeDatta, 1991). But even for the simplest understanding of soil phosphorus, it is necessary to consider labile and nonlabile pools of phosphorus in both the organic and inorganic forms as well as the significant fluxes through the microbial pool (Brookes et al, 1984; Singh et al, 1989; Lodge et al, 1994; Gijsman et al, 1996).

From the outset, we need to define the terms used to define the states and fluxes of phosphorus in the soil. Following Barrow (1999), sorption is taken to mean the transfer of a material from a liquid phase (such as the soil solution) to the solid phase, the soil itself. Sorption includes «¿sorption, which means that the sorbed material is on the outside of the soil particle.

When a labile pool is being discussed it is considered that the material in question is liable to displacement or change. Likewise, the nonlabile pool is considered to be in a stable state for the time scales of interest here (years to centuries).

It should also be emphasized from the outset that, in both a physical and a chemical sense, soils are strongly heterogeneous media and elements such as phosphorus do not really partition into such simple compartmented states (Barrow, 1999). Likewise when attempts are made to fractionate P into pools of varying stability, the exact nature of the different P pools within the soils that these chemically isolated fractions represent is also not entirely clear (Gijsman et al, 1996).

2.1 Soil Organic Phosphorus

For soils underneath moist tropical forests, organically bound phosphorus generally accounts for 20-80% of the total P (Westin and de Brito, 1969; Sanchez, 1976; Tiessen et al, 1994a; Newberry et al, 1997). This organic P represents a wide spectrum of compounds, reflecting the diverse biological origins of soil organic matter (Magid et al, 1995). Labile forms include nucleic acids and phospholipids (of primarily bacterial origin). Inositol phosphates often constitute the bulk of the nonlabile organic-P pool, forming sparingly soluble salts with ions such as iron, aluminum, and cal cium. They can also form strong complexes with proteins and can be strongly adsorbed by clay minerals, typically constituting about 50% of organic P (McLaren and Cameron, 1996).

Organic phosphorus is considered to play a key role as a source of P for plants in tropical soils (Sanchez et al., 1976; Sec. 2.3). In this context it is important to note that, in contrast to nitrogen, phosphorus is to a large degree mineralized independent of carbon (McGill and Cole, 1981). This is a result of the production of phosphatases by plant roots, mycorrhizae, and microbes. These specifically hydrolyze phosphate ester linkages on soil organic compounds, releasing phosphorus and making it available for plant uptake (Sec. 3.3.3).

According to Gijsman et al. (1996), data of Ognalaga et al. (1994) also suggest that organic P can be stabilized into nonlabile forms independently of organic carbon. A similar conclusion was also reached by McGill and Cole (1981). This means that there is much more chance for variation in C/P ratios of the labile soil organic pool than is the case for C/N ratios. This has important implications for the response of P-limited systems to increases in atmospheric carbon dioxide concentrations (Sec. 4.3).

2.2 Soil Inorganic Phosphorus

The labile component of the inorganic phosphorus pool is generally taken to comprise calcium-bonded phosphates, aluminum-bonded phosphates, and iron-bonded phosphates. For highly acid and highly weathered tropical soils such as oxisols and ultisols, iron and aluminum phosphates tend to dominate and thus adsorption capacity for P is usually quite high (Sanchez, 1976). Crystalline clay minerals are also able to specifically adsorb P through a ligand-exchange reaction with the (OH)H groups coordinated with the A1 ion on the edge of the crystal (Muljadi et al, 1966).

The high content of aluminum and iron oxides in the oxisols and ultisols typically found underneath moist tropical forests is the reason for the ability of these soils to "fix" significant amounts of phosphorus when applied as a fertilizer after conversion of these systems to agriculture. Most of the added phosphorus is adsorbed within the first few days of application, although subsequent continued long-term sorption also occurs (Sample et al., 1980; Barrow, 1999). It is important to recognize that this adsorption is a more or less reversible reaction, with the amount of sorbed phosphorus being dependent on the soil solution P concentration (Barrow, 1983). This accounts for the long-term beneficial effects of massive initial applications of phosphorus fertilizers applied to some tropical soils (e.g., Younge and Plucknett, 1966). Phosphorus "fixed" by these soils is subsequently released to the soil solution and utilized for plant growth over many years. This is because desorption occurs in response to the diffusion gradient that typically occurs around any plant root or microbe actively acquiring P (Mattingly, 1975).

Barrow (1983) has pointed out the complexities of the sorption process onto and within soil particles. He suggests that the relatively rapid adsorption of P onto the soil surface is followed by a slow diffusive penetration. Support for this idea comes from the observation that the relative rates of penetration of different adsorbed ions into reacting particles are correlated with the affinity of the surface (Barrow and Whelan, 1989). More recently, Strauss et al. (1997) have shown that the extent of the slow reaction between goethite and phosphate depends on the crystallinity of the geothite. Strong evidence was provided that the mechanism for slow phosphate sorption was a slow penetration of the spaces between the crystal domains. Importantly, the longer a sorption reaction takes to occur, the slower the subsequent desorption reaction and the smaller the amount desorbed after a given period of time (Barrow, 1999).

For many tropical soils, the amount of labile inorganic phosphorus in the sorbed form is typically more than a thousand times greater than the amount of P in the soil solution (Sanchez, 1976). As the net movement of phosphate ions between these pools will always be toward a new equilibrium, this much greater amount of sorbed P means that soil solution P concentrations are strongly buffered against any changes in the rates of entry of P into, or removal of P from, this pool, such as changes that might occur due to changes in P mineralization rates or variations in plant P uptake rates. Thus, the inorganic labile phosphorus pool in many tropical soils can almost be looked upon as a slow-release fertilizer pool whose rate of release is determined by the rate of plant phosphorus utilization. As is shown in Sec. 4.3, this has important implications for the ability of tropical forests to maintain increasing growth rates in response to increases in atmospheric [C02].

The nonlabile fraction of inorganic phosphorus not available to plants is sometimes divided into the occluded and reductant soluble forms. Occluded phosphorus consists of aluminum- and/or iron-bonded phosphates surrounded by an inert coat of another material such as oxides or hydrous oxides of iron or aluminum. Reductant soluble forms are covered by a coat that may be partially or totally dissolved under anaerobic conditions (Uehara and Gillman, 1981). The opportunities for occlusions to occur increase dramatically with soil age (Walker and Syers, 1976). This is because substantial amounts of Fe and Al oxides tend to be present only in heavily weathered soils in which the secondary silicate minerals have already dissolved (Fox et al., 1991). Data from tropical forest chronosequence studies in Hawaii are more or less in accordance with this view: the fraction of P present in the "occluded" form increases with soil age (Crews et al, 1995). Nevertheless, that study also showed high amounts of nonoc-cluded (i.e., labile and accessible) inorganic phosphorus to be present, even in forests growing on the oldest soils.

2.3 Soil Carbon/Phosphorus Interactions

Tropical agronomists have long realized the importance of organic phosphorus as the main source of phosphorus in nonfertilizer agriculture, such as that occurs in traditional systems (Nye and Bertheux, 1957; Sanchez, 1976). In addition to being a source of phosphorus for plant uptake after mineralization, the importance of organic matter in tropical crop productivity is associated with the critical relationship between organic matter content and soil fertility in highly weathered tropical soils (Tiessen et al., 1994b). These soils typically have a very low cation-exchange capacity (CEC) or even a dominant anion-exchange capacity (Sanchez, 1976; Sollins et al., 1988). Soil organic matter performs a vital function in these soils by reacting with Fe and Al oxides, coating the surfaces of oxide particles. This gives rise to a net negative charge and hence a dominant cation-exchange capacity (Uehara and Gilman, 1981; Sollins et al, 1988). This strong association between organic matter content and soil fertility has led to the suggestion that the rapid decline in soil carbon stocks after conversion of forest to agriculture is the prime cause for the subsequent leaching of essential elements out of the active rooting zone (Tiessen et al, 1994b).

The coating of Al - and Fe-oxides by soil organic matter in many tropical soils probably increases phosphorus availability as well. Adherence of large humic molecules to the surfaces of clays and metal hydrous oxide particles (Hughes, 1982; Bonde et al, 1992) should mask the phosphorus fixation sites and prevent oxide particles from interacting with phosphorus ions in solution. In addition, the organic acids in the soil that are produced during microbial degradation of organic matter and directly by plants themselves (Sec. 3.3.2) actively compete with phosphorus ions for soil fixation sites (Dalton et al, 1952; Lopez-Hernandez et al, 1986; Sibanda and Young, 1989; Foxef«/., 1990; Bhatti et al, 1998; Jones, 1998). To date, the relationship between tropical soil organic matter content and plant phosphorus availability has concentrated mostly on the significant declines in soil C and P that usually occur after forest clearance (Mueller-Harvey et al, 1985; Tiessen et al, 1992, 1994b). This decline in soil organic matter has also been associated with an increase in the proportion of phosphorus in less labile forms (Tiessen et al, 1992).

If tropical forests are indeed responding to increases in [C02] by increasing their growth (Grace et al, 1995; Phillips et al, 1998), then much of this extra carbon fixed will eventually end up in the soil (Lloyd and Farquhar, 1996). Thus, a crucial question is whether the positive relationship between soil organic matter and soil phosphorus fertility will hold when soil carbon stocks are increasing? If this were the case, then irrespective of the mechanisms tropical trees may employ to acquire the extra phosphorus needed for increased growth in response to increases in [C02] (Sec. 3.3), improved phosphorus fertility would currently be occurring, merely by virtue of increases in soil carbon density.

Indeed, there is some evidence that the relationship between soil organic matter and phosphorus fertility holds for natural rain forest as well as for degrading systems. For example, there are strong correlations between plant available phosphorus and soil organic matter concentration where natural spatial variability is the primary source of variation (Burghouts et al, 1998; Silver et al, 1999). However, this relationship might also arise from a stimulating effect of soil phosphorus availability on above-ground carbon acquisition being reflected in the soil carbon pool. Likewise, correlations between soil organic matter content and maxi mum degree of phosphorus adsorption for tropical soils (Sanyal and De Datta, 1991) may reflect effects of phosphorus availability on plant productivity and hence soil carbon content rather than vice versa. A more specifically targeted experiment is therefore required to test for this phenomenon. The possible magnitude of the effect is modeled in Sec. 4.3.

3. States and Fluxes of Phosphorus in Moist Tropical Forests

From Sec. 2 it can be concluded that, due to the highly weathered state and high phosphorus sorption capacity of many moist tropical forests soils, the level of readily plant available phosphorus is low. Discussion on whether this means that phosphorus availability actually limits productivity of moist tropical forests is reserved until Sec. 4.1. Here we limit our concerns to a discussion of the phosphorus cycle in moist tropical forests and methods by which plant phosphorus acquisition can occur in environments characterized by low levels of available P. The main aim of this section is to quantify the amounts and annual input/output fluxes of P for leaves, branches, boles, and roots of moist tropical vegetation. The inputs of phosphorus into moist tropical forests from rock weathering and wet and dry deposition, as well as from leaching losses, are also considered. This information is then used for model simulations in Sec. 4.3.

3.1 Inputs and Losses of Phosphorus through Rainfall, Dry Deposition, and Weathering: Losses via Leaching

3.1.1 Atmospheric Deposition

Atmospheric inputs of mineral elements into tropical rain forests may constitute an important input of plant nutrients, especially for soils of low inherent fertility (Proctor, 1987; Bruijnzeel, 1991). Such atmospheric inputs are traditionally divided into wet deposition (input of mineral elements dissolved in rainwater) and dry deposition (inputs from deposited aerosol particles or as dust). For large particles such a distinction may be obvious, for example, in examining effects of Saharan dust on overall forest nutrient balances in West African rain forests (Stoorvogel et al, 1997) or in examining long-range advection of particles such as the deposition of Saharan dust into the vegetation of the Amazon Basin (Swap et al, 1992). But for marine, anthropogenic, and biogenic aerosols, entrainment into atmospheric water vapor may occur during the convective mixing of the lower troposphere, with the elements of such particles then being deposited during rainfall events as well as by dry deposition. The separation of dry versus wet deposition is fraught with technical difficulties (Lindberg et al, 1986), but for many tropical forest studies a simple combined measure of the two has been obtained by sampling in a forest clearing or sometimes above the canopy. In this way the bulk nutrient content of the precipitation has been obtained, at least for the collector itself (Bruijnzeel, 1991). As is also discussed by Bruijnzeel (1989; 1991), due to several complications, this method does not necessarily give the amounts as the amounts of nutrients deposited on the proximal forest canopy. In attempts to deduce external nutrient inputs into a forest, a further complication may be that tropical forests themselves produce aerosols (Crozat 1979; Artaxo et al, 1988, 1990).

Given the above uncertainties, and even after unreasonably high values have been excluded, the high variability in reported rates of P deposition onto tropical forests, 0.3-7 rnmol P m-2 year-1 (Bruijnzeel, 1991; Lesalc and Melaclc, 1996; Stoorvogel et al., 1997; Williams et al., 1997), is not all that surprising. What is surprising is the magnitude of this input relative to the annual litter fall flux, which, from the summary of Proctor (1987), typically ranges from 19 to 44 mmol P m-2 year-1 (see also Sec. 3.2). Indeed, comparisons of lowland forest sites where both bulk precipitation inputs and litter fall measurements have been made suggest that the input of P into tropical ecosystems from the atmosphere above is 0.27 ± 0.17 (n = 6) of the annual litter fall P (Nye, 1961; Bernhard-Reverset, 1975; Golley et al, 1975; Brinkmann, 1985). These relatively high rates of P deposition onto tropical forests contrast with the standard view that atmospheric inputs of P into these ecosystems are not significant (Vitousek et al, 1988; Kennedy et al, 1998).

This atmospheric P deposition cannot be supported by long-term transport of P from tropical oceans, as these typically have very low P concentrations in their surface waters (Graham and Duce, 1979). One possibility is the intrusion of dust from arid regions (Swap et al, 1992). The importance of dust as a nutrient source is likely for West African rain forests (Stoorvogel et al, 1997) but the significance of occasional long-term transport of Saharan dust into Amazonia has been questioned (Lesak and Melack, 1996). For Amazonia, it appears that biogenic emissions from the tropical forests themselves are the main source of atmospheric P in the region (Artaxo et al, 1998; Echalar et al, 1998).

3.1.2 Retention of Atmospherically Derived P

Irrespective of the source(s), hydrological studies have shown that a significant proportion of the atmospherically derived phosphorus appears to be retained by moist tropical forests, rather than being leached out of the system (Bruijnzeel, 1991). This rate of retention seems to be between 0.05 and 0.95 of the rate of input. This probably reflects, as much as anything else, the many sources of error in making such measurements. Bruijnzeel (1991) suggests that this general pattern of phosphorus accumulation in the forest/soil system is real and that it may arise as a consequence of P "fixation" onto iron and aluminum oxides (Sec. 2.3). But although the sorption mechanism is undoubtedly complex (Barrow, 1999) and perhaps less rapidly reversible in highly leached tropical soils than elsewhere (Gijsman et al, 1996), the rate at which P is actually transformed into nonlabile forms is likely to be substantially less than this rate of atmospheric input. Indeed, it is not at all clear whether this external phosphorus arriving at the forest floor would even reach the soil sorption sites. This is because of the extensive root mat near and above the soil surface in many rain forests that can effectively trap dissolved and fine-litter nutrient inputs (Sec. 3.3.1).

3.1.3 Throughfall and Stemflow

In addition to substantial inputs of phosphorus occurring as a consequence of wet and dry deposition, substantial enrichment of rainwater phosphorus concentrations occurs during the passage of rainwater through tropical forest canopies (Vitousek and Sanford, 1986; Proctor 1987; Veneklass, 1990; Forti and Moreira-Norde-mann, 1991; McDowell, 1998). Again, exact values for the enrichment in this throughfall are subject to considerable uncertainties as a consequence of methodological problems. For example, it is not always clear whether this enrichment estimate includes accumulation of elements deposited during dry deposition. But both Marschner (1995) and Richards (1996) consider "canopy leaching" to provide the main source of nutrient additions to rainfall as it passes through the canopy. From data summarized by Vitousek and Sanford (1986) and Proctor (1987), canopy leaching can contribute as much as 20 mmol P m-2 year-1, with average values around 8 mmol P m-2 year-1. As for the P input in rainfall itself, this amount is significant compared with an average litterfall value around 25 mmol P m-2 year-1 (Sec. 3.2). Such rates of P leaching are higher than those that typically occur in temperate regions (Parker, 1983; Marschner, 1995), as would be anticipated on the basis of the much higher rainfall amounts and intensities in tropical regions. According to Marschner (1995) canopy leaching for elements such as P can arise as a consequence of the passage of water through the apoplast of intact leaf tissue as well as through damaged leaf areas, with rates of leaching greater at high temperatures. Proctor (1987) has also pointed out the possible importance of insect frass. Generally speaking, nutrient enrichments during stemflow are much less significant source of nutrients to the soil than canopy throughfall (Parker, 1983; Vitousek and Sanford, 1986; Proctor 1987; Richards, 1996).

3.1.4 Weathering as a Source of Biologically Available Phosphorus

From basin-wide studies in South America, phosphorus weathering rates of 0.3-1.0 mmol P m-2 year-1 have been reported (Lewis et al, 1987; Gardner, 1990). The degree to which such weathering of parent material may supply nutrients for plant growth in moist tropical forests has been considered by Burnham (1989) and Bruijnzeel (1989). They point out that for already highly weathered soils, the active zone of rock weathering occurs a considerable distance below the zone where active root uptake of any nutrients released by the weathering process is likely. Nevertheless, there are some cases where moist tropical forest roots can penetrate the underlying weathered rock (Bailie and Mamit, 1983), and this would certainly be expected to be the case for montane forests. Clearly more experimental work is required, but available evidence indicates that because of the great depth at which weathering generally occurs in moist lowland tropical forests, it is unlikely to be a significant source of biologically available phosphorus in most cases.

3.2 Internal Phosphorus Flows in Moist Tropical Forests

The subject of the cycling of mineral nutrients in tropical forests, particularly the degree to which systems are closed with little leakage of nutrients out of them, is a long-standing area of interest and controversy for tropical ecologists (Hardy, 1935; Walter, 1936, 1971; Jordon and Herrera, 1981; Vitousek and Sanford, 1986; Proctor, 1989; Whitmore, 1989; Silver, 1994; Richards, 1996). In general, the earlier paradigm of closed nutrient cycles with little or no leakage out of them (Hardy, 1935; Walter, 1936) has given way lo an appreciation of the diversity of nutrient cycles in different tropical forests, with effects of natural variations in soil fertility now being a central emphasis (Vitousek and Sanford, 1986; Whitmore, 1989).

3.2.1 Above-Ground Phosphorus Stocks and Soil Fertility

Vitousek and Sanford (1986) grouped lowland forests according to the underlying soil fertility and showed that forests growing on moderately fertile soils (about 15% of the total moist tropical forest area) tend to have foliar N, P, K, Ca, and Mg concentrations higher than do those growing on the more common oxisol or ulti-sol soil types of moderate to low fertility (63% of the total moist tropical forest area). Forests on the latter tend to have foliar nutrient concentrations not very different from forests growing on the very low-fertility spodosol or psamment soil types (7% of the total moist tropical forest area: Sanchez, 1976; Vitousek and Sanford, 1986).

The relationship between above-ground carbon density and above-ground phosphorus density (taken from Table 2 of Vitousek and Sanford (1986) with additional data from Hughes et al 1999) is shown in Figure 1. This shows a remarkably strong relationship between the two parameters, but with a different relationship for moderately fertile soils versus the infertile oxisols/ultisols. For both forests the relationship between above-ground carbon density and above-ground phosphorus density is stronger than that for other nutrients such as nitrogen (not shown). Importantly, forests growing on soils with a low level of phosphorus availability are still capable of achieving substantial above-ground carbon densities, despite having much lower phosphorus stocks than forests growing on more fertile soils. As has been pointed out by Vitousek and Sanford (1986), at least part of this difference in phosphorus is due to much higher foliar P concentrations for trees growing on more fertile soils (1.1 ± 0.2 mmol P mol-1 C) than for those growing on the less fertile soils (0.5 ± 0.1 mmol P mol-1 C).

Foliar phosphorus concentrations typically decline with canopy depth in tropical rain forests (Lloyd et al., 1995) and so it is not straightforward to relate bulked canopy values to physiological measurements made on individual leaves. But similar


0 1200


1 800

0 200 400 600 800 1000 Above ground P [mmol m2]

FIGURE 1 The relationship between above ground carbon density and above ground phosphorus density for moist tropical forests growing on moderately fertile (•) and infertile(B) soils.

magnitude differences in foliar nutrient concentrations often occur between primary and secondary successional rain forest species and this is reflected in differences in plant photosynthetic rates (Raaimakers et al., 1995; Reich et al, 1995). Given the laboratory gas-exchange data of the phosphorus dependency of photosynthesis for leaves of warmer-climate trees (Kirschbaum and Tompkins, 1990; Cromer et al, 1993; Sec. 4.1) it seems likely that rain forests growing on more fertile soils have higher gross primary productivities.

Most likely there are also differences in the general growth strategies employed by trees on the different soil types. For example, Veenendaal et al (1996) examined growth responses of tropical tree seedlings from low- and high-fertility soils in Ghana. Although there were some exceptions, they found that seedlings whose natural distribution was limited to low-fertility soils were not capable of faster growth rates when grown on the higher nutrient soil. Likewise, species restricted to high-fertility soils grew poorly on the lower fertility soil.

Along with faster growth rates and higher phosphorus and nitrogen requirements for the species from the higher nutrient soil (Veenendaal et al, 1996), a picture emerges of species adapted to higher nutrient soils being successful by virtue of high potential growth rates and an ability to rapidly acquire nutrients. Likewise, the moist tropical forest species usually found growing on poorer soils are probably successful on these soils as a consequence of low nutritional requirements, particularly with respect to phosphorus. Also associated with these plants should be specific physiological adaptions allowing high phosphorus uptake rates despite low levels of readily available P (Sec. 3.3). This is similar to the relationships between plant growth strategy and soil fertility proposed for temperate, arctic, and boreal ecosystems (Chapin, 1980).

Tropical forest foliage typically accounts for less than 15% of the above-ground P pool (Folster et al, 1976; Klinge, 1976; Hase and Folster, 1982; Uhl and Jordan 1984). Therefore, most of the differences between the C/P relationships in Figure 1 are attributable to differences in the phosphorus concentrations in twigs, branches, and boles. For example, the average concentration of phosphorus in the boles of the forests growing on the more fertile soils is 0.30 ± 0.06 mrnol P-1 mol-]c (n = 4: Greenland and Kowal, 1960; Golley et al, 1975; Hase and Folster, 1982; Hughes et al, 1999), whereas for above-ground woody tissues on the less fertile ox-isols/ultisols this figure is 0.16 ±0.05 mmol P mol-1 C (« = 6: Bernhard-Reversat, 1975; Folster et al, 1976; Klinge, 1976; Uhl and Jordan, 1984).

Explaining this twofold difference between the two forest types is difficult. Despite the fact that woody components constitute the dominant above-ground pool for P in moist tropical forests, the role of P in woody tissue is not well defined. Most likely its functions relate to its being a structural constituent of the growing sapwood, as well as inorganic phosphorus being associated with general energy transfer reactions in sapwood and phloem-associated cells. In both cases, a general positive relationship between high plant growth rates and woody tissue P concentrations would be expected.

As is the case for nitrogen (Lloyd and Farquhar, 1996), one might expect a decrease in bole P content with increasing plant size. This is because most of the P would be expected to be in the physiologically active sapwood tissue. This constitutes a progressively smaller portion of the total stemwood as trees become bigger. Nevertheless, when compared across sites, there seems to be no general pattern of lower bole P concentrations in forests with increasing carbon density (data not shown). However, for individual tropical forest species, such a trend of deceasing P concentrations with increasing bole size has been observed (Grubb and Edwards, 1982).

Along with the likely higher photosynthetic rates discussed above, the greater phosphorus content of woody tissue from forests growing on more fertile soils suggests higher potential gross and net primary productivities than those of less fertile forests. This then begs the question of how the above-ground carbon density of nutrient-poor forests can generally be higher than that of forests growing on more nutrient-rich soils (Figure 1).

In considering the observed lack of correlation between forest biomass and soil nutrient status for moist tropical forests, Vi-tousek and Sanford (1986) proposed that previous natural and anthropogenic stand-level disturbances may have been responsible. Differences in site water balance might also be important. Three of the four sites in Figure 1 growing on moderately fertile soils are moist semideciduous forests and are characterized by the presence of some drought deciduous species (Greenland and Kowal, 1960; Golley et al, 1975; Hase and Folster, 1982). This reflects a greater than average seasonality in water supply. Even for moist evergreen forests, marked effects of soil water deficit on photosynthetic productivity during the dry season can occur (Malhi et al., 1998). This observation, combined with the observation that biomass of dry tropical forests is positively related with annual rainfall up to at least 1500 mm per annum (Martinez-Yrizar, 1995), suggests that the lower biomass of "moist" forests on the more fertile soils could in some cases be a consequence of more prolonged soil water deficits during the dry season than is the case for the forests growing on the more highly leached oxisol and ultisol soil types. Indeed a negative association between soil fertility and soil water balance (Veenendaal et al, 1996) is likely. Soils in areas exposed to lower rainfalls are likely to be less leached and therefore higher in nutrient status (Burnham, 1989). Confounding this rainfall/fertility correlation at the stand productivity level is the observation that even on the same soil, tropical drought-deciduous species typically have higher N and P concentrations than do proximal evergreen species (Medina, 1984).

It is also possible that plant growth traits associated with potentially faster growing trees on higher nutrient soil predispose such forests to lower carbon densities. For example, Phillips et al (1994) showed a positive relationship between soil fertility and tree turnover rates for tropical forests. Likewise, leaves of inherently slower growing species tend to be longer lived (Chabot and Hicks, 1992). Studies with different successional species have confirmed this pattern for moist tropical forests (Reich et al, 1995). Thus, despite their lower productivities, slower tree turnover rates might contribute to the attainment of high above-ground carbon densities for forests growing on nutrient-poor soils.

3.2.2 Phosphorus Content of Coarse and Fine Root Tissue

Not surprisingly, the available information on root P content is less than that on the above-ground biomass. Nevertheless, the available data suggest that the effects of soil fertility on root P concentrations are similar to those discussed above for leaves and above-ground woody tissue. For the two high-fertility sites where data are available (Greenland and Kowal, 1960; Golley, 1975), the average value is 0.37 ± 0.04 mmol P mol-1 C, whereas for the low-fertility oxisol sites for which data are available (Klinge, 1976; Uhl and Jordan, 1984) the average value is 0.15 ± 0.05 mmol P mol-1 C. These values are remarkably similar to the average values for above-ground woody tissue given above: 0.30 ± 0.05 and 0.16 ± 0.05 mmol P mol-1 C, respectively.

The values cited above represent a pooled average for coarse and fine roots. Greenland and Kowal (1960) separated out roots of varying diameter from a forest in Ghana. They showed that P concentration increased with decreasing root size with the finest size category (<6 mm) having a concentration of 0.59 mmol P mol-1 C, much higher than their coarsest size category (>25 mm), which contained 0.10 mmol P mol-1 C. For fine roots in poorer soils, fine root P concentrations seem to be similar. For a forest growing on an oxisol in Venezuela, Medina and Cuevas (1989) give a fine root concentration of 0.85 mmol P mol-1 C. Vitousek and Sanford (1986) cite a fine root concentration of 0.55 mmol P mol-1 C, also for a Venezuelan forest. Thus, unlike foliar tissue or structural woody biomass, it seems that there is little systematic effect of soil fertility on fine root P concentrations. This, along with the tendency of lower fertility sites to have a greater proportion of their total biomass below ground (Vitousek and Sanford, 1986), indicates a need for plants on low-nutrient soils to allocate a greater proportion of their carbon and nutrient resources to the acquisition of limiting elements (Chapin, 1980).

0 0

Post a comment