| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a The Ecosystems Center, Marine Biological Lab., Woods Hole, MA 02543 USA
b Centro de Energia nuclear na Agricultura, Avenida Centenário 303, CEP 13416000, Piracicaba, Sao Paolo, Brazil
dgarcia{at}mbl.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMaterials and methodsResultsDiscussionREFERENCES |
|---|
Abbreviations: ANOVA, analysis of variance • i, inorganic • o, organic
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResultsDiscussionREFERENCES |
|---|
The widespread adoption of slash and burn clearing for pasture establishment in the humid tropics alters not only the highly efficient nutrient-conserving mechanisms that characterize the forest (Jordan, 1982; Medina and Cuevas, 1989), but also the patterns of P cycling within the ecosystem. The commonly observed pulse of soil-available P (Sanchez et al., 1983; Uhl, 1987; Jordan, 1989) associated with forest cutting and burning initiates processes leading to the long-term transformations of soil P. The initial post-burn increase in soil P fertility is probably derived from both rapid ash incorporation into the soil and the slower release of P from the residual unburned or incompletely burned slashed biomass (Sanchez et al., 1991; Hands et al., 1995)
In slash and burn systems, the initial post-burn pulse of released P is rapidly incorporated into the soil available P pool, where it can then be sorbed by the mineral soil or into the soil microbial biomass and later into slower turnover organic P pools (Sanchez et al., 1991). It has been claimed that crop abandonment in slash and burn crop systems is usually preceded by a decline in the initially available P pool (Jordan, 1989; Sanchez et al., 1991). In pasture systems, the introduction of forage grasses after forest clearing and burning may alter the P fertility cycle characteristic of the traditional slash and burn agricultural systems. Because of the low ratio of above- to belowground biomass of grasses and the faster turnover of their roots, P fertility under a pasture system may be maintained by a higher rate of recycling of soil organic P, as has been reported for other unfertilized cultivated systems (Beck and Sanchez, 1994), and this pattern may last for longer period of time.
The conversion of moist tropical forest to cattle (Bos taurus) pasture is now widespread in the humid tropics. In the Brazilian Amazon, land clearing rates average 
20000 km2 yr-1 and the total area cleared reached 500000 km2 (or 10% of the Basin's total area) in the mid 1990s (Instituto Nacional de Pesquisas Espaciais, 1998). The conversion to pasture represents the most common fate of deforested land (Fearnside, 1987). Despite this high rate of conversion and the importance of P in controlling plant production in natural humid tropical ecosystems (Vitousek, 1984; Herbert and Fownes, 1995) and in planted Amazonian pastures (Serrão et al., 1978; Gonçalves and da Oliveira, 1984), very little quantitative information is available about the P transformations that occur after forest is cleared for pasture and how these P transformations affect the P availability in older pasture soil.
In this study, we investigated how stocks of total soil P and different soil P fractions changed after the conversion of moist tropical forest to pasture in the Brazilian State of Rondônia in the southwest of Amazonia. Chronosequences of forests to pastures of different ages but similar history and management were used to document patterns of change in the stocks of P in soil solution, labile inorganic, occluded, and organic pools using sequential P fractionation techniques. The patterns in these pools were then used to develop a conceptual model of P transformation following forest clearing for pasture.
We were also interested in placing the P transformations following land use change in the context of changes to soil P stocks that occur over the course of soil formation. The Walker and Syers (1976) model of P transformations in soil through geological time states that, over the course of soil development, P should be transformed from primary minerals that are dominant in young soils, to organic and occluded forms that are dominant in old, highly weathered soils. The model was recently corroborated by Crews et al. (1995) in a chronosequence of soils from the Hawaiian archipelago. This model also provides a powerful tool for understanding soil P transformations that can occur under the influence of human disturbance. Phosphorus pools as defined by Walker and Syers (1976) can be determined by chemically extracted P fractionation methods (Hedley et al., 1982). The Walker and Syers (1976) model was developed assuming that the major source of P in young soils is P released from primary minerals. When applying this model to older, highly weathered tropical soils, the source of new P is the biomass that is burned after forest clearing. However, when P transformations are restarted in a highly weathered soil system, some of the components that induce P transformations in soils, such as low pH and high Fe and Al oxides are already present. In this study, we compared and contrasted our data on P transformations after forest clearing for pasture with changes to forms of P that occur over the course of soil development. Instead of the thousands of years required for P transformations to occur over geologic time, we hypothesized that vastly less time is required to initiate a similar sequence of events after human disturbance.
|
Materials and methods |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResultsDiscussionREFERENCES |
|---|
4°C (Bastos and Diniz, 1982). The predominant vegetation consists of open moist tropical forest with a large number of palm trees. The primary forest vegetation of Rondônia has been classified as ombrotrofic open submontane forest (RadamBrasil, 1978). The forests used for this study had been altered by selective logging, which removed one to three trees per hectare during the 10 yr prior to the sampling. This level of disturbances is very typical of the accessible forests that remain in large ranches in Rondônia (Pedlowski et al., 1997).
Chronosequence 1 included a forest and six pastures that ranged in age from 3 to 41 yr old (Table 1)
. Chronosequence 2 contained a forest and three pastures that ranged from 3 to 20 yr old. All sites were selected in areas at an elevation of 150 m with minimal relief. All pastures were formed by direct conversion of the original forest with no intermediate cropping phase. Pastures were all created following a similar process: marketable wood was first extracted, the brush was cut during March, large trees were cut between June and July, and the slash from the downed forest biomass was burned in August and September. Pasture grass seed was dispersed by airplane in December or January. In Chronosequence 1, the 3-, 5-, and 20-yr-old pastures were dominated by brizantha [Brachiaria brizantha (Hochst) Stapf.], the 13- and 41-yr-old pastures contained guineagrass (Panicum maximum Jacq.), and the 9-yr-old pasture was dominated by P. maximum but contained some B. humidicola (Rendle) Schweickt. In Chronosequence 2, the 3- and 5-yr-old pastures were dominated by B. brizantha and the 20-yr-old pasture contained predominantly P. maximum. Periodic burning every 4 to 7 yr was used to control woody vegetation. Pasture creation involved no mechanical practices, chemical fertilization, or liming. Younger pastures (3 to 5 yr old) differed from the older pastures in that they usually contained larger number of palms and woody debris left from the clearing of the forest.
|
22% of the Brazilian Amazon Basin (Moraes et al., 1995). Morphological characteristics of these soils have been described in greater detail by Moraes et al. (1995) and are summarized in Table 1. Soil particle distribution showed minimal differences between sites of each chronosequence (Table 1). Clay content ranged between 22 and 29% in the top 10 cm soil of all sites of Chronosequence 1 and between 11 and 23% in all sites of Chronosequence 2. Soil clay content differences with depth were also reported by Moraes et al. (1995) and were small among sites within chronosequences. Similar particle-size distribution, topography, and former vegetation all indicated that preexisting differences among these sites were small. Textural differences are the primary criteria for selection of chronosequences (Sanchez et al., 1985). Soil pH ranged from <5 in the forests to 5.1 to 7.2 in young pastures. Following the burning of the slashed forest biomass, Moraes et al. (1995) reported an increase in soil pH and a decrease in Al3+ activity in the 0- to 10-cm depth of soil (Table 1). Exchangeable K, Ca, and Mg also increased after burning, and Ca increased between two- and fourfold in 5-yr-old pastures in both chronosequences (Table 1). Effective cation-exchange capacity and percentage base saturation increased in young pastures and then declined in older pastures (Table 1). Carbon and N stocks of these soils are described in greater detail in Moraes et al. (1995) and Neill et al. (1997), while N dynamics are described in Piccolo et al. (1994) and Neill et al. (1995). Soil C concentrations in the 0 to 10 cm ranged from 11 to 21 g kg-1, and soil N concentrations ranged from 0.9 to 1.6 g kg-1 (Moraes et al., 1995). Soil C stocks at 0 to 10 cm increased from 1.5 kg m-2 in the forest to 2.5 kg m-2 in the 41-yr-old pasture of Chronosequence 1 and from 1.4 kg m-2 in the forest to 2.3 kg m-2 in the 20-yr-old pasture from Chronosequence 2 (Table 1; Neill et al., 1997). Soil N stocks at 0 to 10 cm ranged from 0.08 to 0.18 kg m-2 but showed no consistent trend after conversion of forest to pasture (Table 1; Neill et al., 1997).
Sample Collection and Preparation
Soil samples were collected in 1992. Soils were obtained at each site along the chronosequences from a pit (80 by 150 by 150 cm) and with an auger at four other points 25 m in each direction from the pit. Within the pit, soil samples were collected at the 0- to 5- and 5- to 10-cm depths, and then every 10 to 150 cm. The four auger samples were collected at 0 to 5, 5 to 10, 10 to 20, and 20 to 30 cm. Phosphorus determinations (total P and fractions) were conducted on all five samples collected from the 0- to 5- and 5- to 10-cm depths from all sites. To examine changes with depth, we analyzed five samples collected from the 20- to 30-cm depth from the forests and the 5- and 20-yr-old pastures and one sample collected from the 40- to 50- and 80- to 90-cm depths in the forests and the 5- and 20-yr-old pastures. Soil samples for P determinations were homogenized, air-dried, passed through a 2-mm mesh sieve, and ground with a mortar and pestle.
Phosphorus Determinations
Total soil P content was determined by the sodium hydroxide fusion method (Smith and Bain, 1982). Air-dried soil (0.15 g) was weighed into a 30-mL nickel crucible. After the addition of 2 g of NaOH, the soil plus NaOH mixture was gently ignited over a bunsen flame for 3 to 4 min until the NaOH completely dissolved. To ensure complete release of P, the mixture was then heated for another 5 min at a dull red color. After cooling, the melted mixture in the crucible was dissolved with deionized water, transferred to a 100-mL volumetric flask, and brought to volume with deionized water. An aliquot was adjusted to pH 4 using p-nitrophenol and P concentrations were determined colorimetrically by the molybdate-ascorbic acid procedure of Murphy and Riley (1962).
We determined P content of different soil fractions using the sequential P fractionation method of Hedley et al. (1982) as modified by Tiessen and Moir (1993). Resin-extractable P was determined by placing 1 g of air-dried soil into a 50-mL polypropylene centrifuge tube and shaking for 16 h with two anion resin strips (9 by 62 mm per strip) in 30 mL of deionized water. The P held on the anion exchange sites of the resin was then removed by shaking for 1 h in 30 mL of 0.5 mol L-1 HCl solution that was then collected for P determination. Bicarbonate-extractable P was determined by adding 1.26 g of NaHCO3 to the deionized water and soil residue left after the extraction with the resin strips. This created a solution 0.5 mol L-1 of NaHCO3 that was adjusted to a pH of 8.2 with 0.1 mL of 4 mol L-1 NaOH solution (modification suggested by H. Tiessen, 1997, personal communication). Soils were shaken for 16 h, centrifuged for 10 min at 10000 rpm, 1000 rpm and 0°C, and P determined on the supernatant. The NaOH-extractable P was obtained by adding 30 mL of 0.2 mol L-1 NaOH to the soil residue remaining in the centrifuge tube after the NaHCO3 extraction. This was shaken for 16 h, centrifuged (10000 rpm at 0°C), and the supernatant collected for P determinations. A subsequent extraction with 1 mol L-1 HCl solution followed the same procedure as the extraction with the solution NaOH. After the extraction with 1 mol L-1 HCl, soil residue was extracted by boiling in concentrated HCl at 80°C for 10 min (Tiessen and Moir, 1993) and the extract was analyzed for P.
Total P was determined in aliquots of the NaHCO3, NaOH, and the hot concentrated HCl extracts by digestion with ammonium persulphate and H2SO4 to convert all the P present as organic (Po) forms to inorganic (Pi) forms. Another subsample of the NaHCO3 and NaOH fractions was used to measure Pi after acidification with H2SO4 to precipitate organic matter. Organic P fractions were estimated as the difference between the total P and Pi in the digested extracts. Phosphorus remaining in the soil residue after the extraction with hot concentrated HCl was estimated as the difference between total P and the sum of all P fractions. The pH of the final extracted solutions was adjusted and P concentration determined with the colorimetric procedure of Murphy and Riley (1962). Phosphorus concentration values were expressed on an areal basis using measured bulk densities.
We recognize that P fractions determined by this sequential extraction represent chemical rather than strictly biologically defined pools; however, they can be associated with different levels of P availability for plant uptake (Tiessen and Moir, 1993). The resin-extractable P is associated with P in solution that is rapidly available for plant uptake, the NaHCO3-extractable Pi is a labile pool in equilibrium with soil solution, the NaOH-extractable Pi pool is a moderately labile P associated with the surfaces of amorphous Al- and Fe-oxides and sequioxides, and the dilute HCl-extractable Pi represents Ca-bound P. The dilute HCl-extractable Pi would be expected to be negligible in the highly weathered soils of the humid tropical forest; however, because of the large amount of Ca that is released from the burning of the forest vegetation we expected to observe an increase of P in this fraction, at least during the first 3 to 5 yr after the burning. The residual P pool has been considered the occluded phosphate physically encapsulated by amorphous minerals. The extraction with hot-concentrated HCl allowed the separation of Po from the rest of the residual Pi (Tiessen and Moir, 1993). We present organic P as the sum of all forms of extracted Po: the NaHCO3-extractable Po, the NaOH-extractable Po, and the hot concentrated-HCl-extractable Po.
We evaluated these extracted P fractions within the context of Walker and Syers' (1976) model for P transformation in soil by grouping fractions in the following way: nonoccluded Pi = resin-extractable P + (NaHCO3- extractable Pi) + (NaOH-extractable Pi); Ca-bound P = dilute HCl-extractable Pi; organic P = (NaHCO3-extractable Po) + (NaOH-extractable Po) + (concentrated HCl-extractable Po); and occluded P = (concentrated HCl-extractable Pi) + residual P. These groupings followed Crews et al. (1995).
The effects of age, site, and the interaction of age with site on soil P stocks and fractions were evaluated with two-way analysis of variance (ANOVA) using the General Linear Model of SYSTAT 7.0 program (SPSS, 1997). To provide true age replication, the two-way ANOVA test was conducted on the P fraction values obtained from the forests, and the 3-, 5-, and 20-yr-old pastures from both chronosequences. One-way ANOVA tests were used to evaluate the effects of forest and pasture age on soil P for each individual chronosequence. These one-way ANOVA tests included all seven sites from Chronosequence 1 or all four sites from Chronosequence 2. All ANOVAs were run as n = 5 for five samples at each depth for each age group within each chronosequence. Data were properly transformed when requirements of normality and homocedasticity were not met. When the resulting ANOVA tests were significant for the age effect, Bonferroni multiple pairwise comparisons were conducted.
|
Results |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResultsDiscussionREFERENCES |
|---|
30% higher than those found in the original forest. In Chronosequence 2, soils from the 3- and 5-yr-old pastures had lower total P stocks than soils from the forest (Table 2). Total P in the 0- to 10-cm depth from the 20-yr-old pasture was 19.4 g m-2, 34% lower than in the forest soils (Table 2).
|
|
|
The dilute HCl-extractable Pi increased 25-fold compared with the levels in the forest (P < 0.001; Fig. 1c) in the 3-yr-old pasture of Chronosequence 1, but in pastures older than 9 yr, dilute HCl-extractable Pi declined to near forest levels (Table 3, Fig. 1c). In Chronosequence 2, dilute HCl-extractable Pi increased 4.3-fold in the 5-yr-old pasture, and as in Chronosequence 1, it declined to near forest levels in the older pastures. The effect of age was highly significant (P < 0.001) for this dilute HCl-extractable Pi fraction; however, differences in the time when P increases occurred in younger pastures also resulted in a significant chronosequence effect (P = 0.04). Dilute HCl-extractable Pi in soils from the 3-yr-old pasture of Chronosequence 1 represented 8% of the total P at 0 to 10 cm, and 4% of the total P in the 5-yr-old pasture of Chronosequence 2 (Table 3). Dilute HCl-extractable Pi represented <1% of the total P in soils from the older pastures of both chronosequences (Table 3). The largest increase in dilute HCl-extractable Pi always occurred in the top 0 to 5 cm, where P content was typically between 8- to 20-times greater than at 5 to 10 cm (Table 3).
The hot concentrated HCl-extractable Pi showed a significant increase during the first 5 yr after forest clearing in Chronosequence 1 (P < 0.001) and during the first 3 yr after forest clearing in Chronosequence 2 (P < 0.01; Fig. 1d). In the 20-yr-old pastures from both chronosequences, hot concentrated HCl-extractable Pi declined, but to levels above the forest in Chronosequence 1 and to levels below the forest in Chronosequence 2 (Fig. 1d). These slight differences in the hot concentrated HCl-extractable Pi trends resulted in a highly significant interaction effect of age with chronosequence for this fraction (P < 0.001). However, because trends in concentrated HCl-extractable P observed in younger pastures are followed by the decline in older pastures in both chronosequences, the effect of chronosequence alone was not significant (P > 0.5). The relatively large increase that was observed over a relatively short period of time (3 to 5 yr after forest clearing) in this supposedly recalcitrant fraction was not the result of a methodological artifact. Separate fractionation analysis of samples from the 3- and 5-yr-old pastures that included different combinations of the five-replicated samples per site and per soil depth always produced the same consistent trends, with <30% coefficient of variation.
Residual P also showed different trends related to age (Fig. 1e) that resulted in a significant effect of chronosequence, age, and the interaction of both (P < 0.02). In Chronosequence 1, residual P increased in the 5-yr-old pasture and then declined to levels similar to the forest, but in Chronosequence 2 residual P continuously declined from the forest to the older pastures (Fig. 1e). Residual P plus the hot concentrated HCl-extractable Pi comprised the larger proportion of the total P in the 0- to 10-cm depth of soils from all sites in both chronosequences (Table 3).
Phosphorus bound in organic forms seems to increase with pasture age (Fig. 1f). Phosphorus bound in organic forms was consistently affected by pasture age (P < 0.001) or chronosequence (P = 0.02), but the interaction of age and chronosequence was not significant (P > 0.2).
Phosphorus Transformations
Phosphorus fractions grouped to fit the model of Walker and Syers (1976) and following Crews et al. (1995) showed that forest clearing for pasture caused the redistribution of different P forms (Fig. 2)
. In forests, the distribution of total P among nonoccluded (resin- + NaHCO3- + NaOH-extractable Pi), organic (NaHCO3- + NaOH-Po), and occluded (concentrated HCl- extractable + residual P) pools in the top 0 to 10 cm of soils was very similar, with 14 to 16% of the total P found in nonoccluded forms, 21% in organic forms, and 63 to 65% as occluded P. Calcium-bound P (dilute HCl-extractable Pi) was negligible in the forest soils, comprising only 0.4 to 0.8% of the total soil P. After burning, nonoccluded P increased in the 3-yr-old pasture of Chronosequence 1 and the 5-yr-old pasture of Chronosequence 2, where they reached 20 to 21% of total P (Fig. 2). Calcium-bound P increased in the 3-yr-old pasture of Chronosequence 1 to 8% of total P and to 4% of total P in the 5-yr-old pasture of Chronosequence 2. Changes in the proportions of organic P were negligible during the first 3 yr after clearing, but the proportion of residual or occluded P declined in both chronosequences. Inorganic P extracted with NaOH contributed between 59 to 82% of the P in nonoccluded forms at both chronosequences (Table 3).
|
In the older, well-established pastures of both chronosequences, the proportion of P in the organic pool increased to 29 to 37% of total P (Fig. 2 and 3) . In these older pastures, the organic NaOH-extractable P comprised around 68% of the P in the organic pool. In contrast to the organic P, the proportion of P in the occluded pool decreased to 44 to 55% of total P in pasture soils compared with the forest.
|
|
|
Discussion |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResultsDiscussionREFERENCES |
|---|
Distribution of Soil Phosphorus after Forest Clearing for Pasture
The burning of tropical humid forest releases P that is bound in forest biomass and initiates a series of changes to ecosystem and soil P pools. These changes involve the redistribution of P between living plant biomass and soils and the redistribution within the soil among nonoccluded, occluded, and organic pools. The results from this study and the results of related studies of P contained in forest biomass in Rondônia (Kauffman et al., 1995) suggested that P redistribution after the clearing and burning of the forest for pasture occurs in a series of phases (Fig. 5)
.
|
Despite the high biomass of these ecosystems, vegetation P accounts for only 10 to 40% of total P stocks in the ecosystem when soils to a depth of <30 cm are considered (Klinge et al., 1975; Uhl and Jordan, 1984; Kauffman et al., 1995). The majority of P in forest soils occurs in the least-available form as occluded P. We found that 64 to 65% of the total P occurred in occluded forms (Fig. 3), while 14 to 16% was in nonoccluded forms and the remaining 20 to 21% was organic P. These proportions are within the range of those reported by Tiessen et al. (1994) for the upper Rio Negro region of the Amazon Basin, where total occluded P constituted the largest proportion (40–70%) of the total P in the first 40 cm of mineral soil.
Phase 2: Post-Burning Period and Pasture Establishment
The immediate post-burning period is characterized by a sharp decline in P in plant biomass and a rapid increase in soil P in the most available, nonoccluded, and Ca-bound pools (Fig. 5) derived from the ash deposited on the soil surface. The combustion efficiency of slashed biomass in Rondônia in typical slash and burn practices has been reported in the range of 42 to 57% (Kauffman et al., 1995; Graça, 1997). This results in the immediate release of 3 to 5 g P m-2 from the burned biomass of two forests in Jamari and Santa Barbara in Rondônia, of which 2 to 3.5 g m-2 P was deposited in ash (Kauffman et al., 1995). In studies from other locations in the Amazon, Sanchez et al. (1991) reported the deposition of 0.6 g m-2 P in ash generated from burning a primary forest near Manaus and 1.7 g m-2 P in ash after burning of a 25-yr-old regrowing forest in Peru. These reported values were slightly lower than the increase in nonoccluded plus Ca-bound P that we observed in Chronosequence 1 of 4.8 g P m-2, 3 yr after pasture creation and suggest an additional slow incorporation of P from the residual biomass. The incorporation of P from ash alone cannot account for the large increase in total P (8.8 g m-2) that we observed 3 yr after burning in Chronosequence 1. Because much of the slashed, residual aboveground biomass remains in place after burning, the mineralization of this material probably provides further input of P to the soil. In Rondônia slash and burn sites of Jamari and Santa Barbara, this aboveground residual material contained 2 to 3 g m-2 P (Kauffman et al., 1995). At Nova Vida, the majority of this large woody debris remaining from slash fires typically decomposes within a period of 5 to 10 yr (personal observation). Phosphorus input from the decomposition of belowground root material could also add another equal or slightly lesser amount of P. The time required for the decomposition of residual belowground material is not known, but would presumably be equal to or more rapid than that for aboveground biomass material. These two sources could potentially account for the measured increases in Chronosequence 1 in our study. Spatial variability between sites of the chronosequences and other locations in Rondônia or in the rest of the Amazon may have also contributed to the observed discrepancies between the results from our study and those produced from other studies. In Chronosequence 2, a decrease in total P but a relative increase in the nonoccluded plus Ca-bound P suggests that a transfer of P from biomass to soil also occurred, but indicates that the forest site may had a higher initial total P than the pasture sites. Aboveground biomass of grasses that become established within 6 mo of the original forest clearing was in the range of 1 kg m-2 (Gonçalves and da Oliveira, 1984) and contained 0.4 g m-2 of P (Kauffman et al., 1998). Even if pasture grass belowground biomass was equal to aboveground biomass during Phase 2, the total transfer of P back to vegetation during this period was relatively small.
The increase of soil P in the resin-extractable P, NaHCO3-extractable Pi, NaOH-extractable Pi, and dilute HCl-extractable Pi observed during the first 3 to 5 yr after burning in both chronosequences was accompanied by increases in base cations and pH (Table 1). Soil pH probably contributes to the maintenance of high P availability during this period. In the highly weathered soils of the humid tropics, pH influences P solubility and increased pH decreases the complexation of phosphate by Fe and Al, resulting in greater P solubilization (Lindsay, 1979; Uehara and Gillman, 1981). Increased soil pH and greater P availability for a few years after slashing and burning of forest are reported in different regions of the humid tropics (Nye and Greenland, 1964; Brinkman and Nascimento, 1973; Ewel et al., 1981; Sanchez et al., 1983; Stromgaard, 1984; Jordan, 1989). However, direct comparisons of soil P are hampered by the different P extraction methods used in different studies. Differences can also arise from the age and the biomass of the slashed forest, the length of time the slashed biomass is allowed to dry before burning, combustion efficiencies, leaching resulting from timing of burning, crop establishment, and rainfall regime. At our sites, this phase appeared to last <5 yr. Burning of a mature terra firme forest in San Carlos de Rio Negro raised the available P extracted with an acid solution from 4.8 mg P kg-1 in the control forest to 13.0 mg P kg-1 20 mo after burning (Jordan, 1989). Burning of a 17-yr-old forest in the lowlands of the Peruvian Amazon increased extractable Olsen P between 9 and 14 mg P kg-1. If we consider our resin plus NaHCO3-extractable Pi pools as equivalent to the P pool extracted by the Olsen procedure, then the increase that we observed 3 yr after burning in Chronosequence 1 resulted in a total pool of 34 mg P kg-1 (2.3 g P m-2) in the upper 10 cm of soil. In Chronosequence 2, the resin- plus NaHCO3-extractable Pi in soil reached a concentration of 30 mg P kg-1 (1.9 g P m-2) in soil in the upper 10 cm of soil. The Olsen-equivalent Pi concentrations in soils of our 3- to 5-yr-old pastures in Rondônia were similar to, or higher than, the concentrations reported by others (Brinkman and Nascimento, 1973; Sanchez et al., 1983; Jordan 1989) and above the values of 8 to 15 mg P kg-1 judged to be critical values for the cropping of maize (Zea mays L.) and cowpea (Vigna unguiculata L.) (Smyth and Cravo, 1990; Sanchez et al., 1983, 1991).
Phase 3: Established Pastures
Rapid declines of available P characterize a phase that occurs several years after establishment of pasture grasses (Fig. 5). This phase occurred 3 to 5 yr after forest clearing in Chronosequence 1 (Fig. 2). Soil pH remained elevated, exchangeable Al remained low, and exchangeable Ca increased during this phase compared with the original forest (Table 1). Declines of available P in Chronosequence 2 occurred 5 yr after forest clearing; however, because pastures between the ages of 5 and 20 yr old could not be included in this chronosequence, we could not test for the repeatability in the pattern of increasing occluded P that was observed for Chronosequence 1 during the decline of the nonoccluded and Ca-bound P pools. However, the increase in occluded P forms observed in Chronosequence 1 supports the interpretation that the decline in available P that is commonly observed several years after forest clearing in traditional slash and burn agriculture (Jordan, 1989) is caused predominantly by the transformation of P from more- to less-available, predominantly occluded, forms. The decline in crop productivity in slash and burn crop systems usually coincides with a decrease in available P and the invasion of weeds that out-compete crop species (Jordan, 1989; Hands et al., 1995). We suggest that the increasing dominance of weeds during this phase in crop systems could be caused by a lower P requirements of weeds or their ability to access less-available forms of P, such as the occluded P that increases during this period. In slash and burn crop pasture systems, grasses might have a similar ability to access less-available forms of P that allows the establishment and maintenance of pastures systems for a longer period of time. This pattern of increasing occluded P early during pasture establishment that is suggested by Chronosequence 1 deserves further investigation.
Phase 4: Aging Pastures
We found that the occluded P that increased during Phase 3 of Chronosequence 1 declined in the oldest pasture. This was accompanied by an increase in organic P, suggesting that the occlusion of P by secondary minerals was acting as a temporary reservoir for some soil P that was ultimately converted to organic P when land was placed in continuous grass cover. We do not know for certain if the increase in occluded P observed for Chronosequence 1 during Phase 3 also occurred in Chronosequence 2; however, a decline in occluded P and increase in organic P was consistently observed in the aging pastures of both chronosequences. In shifting slash and burn agricultural systems, new inputs of available P provided by biomass burning seem to be essential for the cropping of species with high P requirements. It is possible that in pasture systems, some of the demand for P by pasture grasses may be satisfied by a potential larger competitive ability of pasture to access the so-called occluded P pools and by a faster recycling of P throughout the organic P pool. This competitive ability of pasture grasses may allow the maintenance of pastures systems for a longer period of time than the slash and burn agricultural systems. This hypothesis is consistent with the fact that pastures in Nova Vida are usually maintained for a period of 10 to 20 yr without fertilization or liming. It does not contradict the early premise of pasture limitations by P suggested by the increase of grass productivity after P fertilization, but only suggests a potential larger competitive ability of pasture than crop species to access occluded P and to maintain faster recycling of organic P that can maintain pasture productivity to support livestock. Forage production can in fact be further increased if fertilization or liming is implemented.
Many years after pasture establishment, pasture grasses have clearly influenced the distribution of the forms of P in soils (Fig. 3). In pastures 20 yr old and older from both chronosequences, the occluded P pool comprised a lower fraction of total soil P (41–55%) than in the original forests (63–65%). The proportion of P in organic forms was increased (29–35%) compared with forest (20–21%). Other studies in laboratory incubations and crop systems have demonstrated that fertilization with P and the addition of a C source to soils low in available P is necessary to induce significant increases in soil organic P (Hedley et al., 1982; Tiessen et al., 1992). The increase in organic P in older pastures is consistent with increased stocks of organic matter in these soils. Total soil C in the top 30 cm increased from 3.2 kg m-2 in the forest to 4.7 kg m-2 in the 41-yr-old pasture of Chronosequence 1 and from 2.7 kg m-2 in the forest to 3.9 kg m-2 in the 20-yr-old pasture of Chronosequence 2 (Neill et al., 1997). Organic matter derived from grass roots makes up the majority of total soil C in these older pastures (Neill et al., 1997).
The conversion of forest to pasture appeared to change the pattern of P cycling where P is recycled between aboveground biomass through litter to a system where P cycling is more dependent on transfers through grass belowground biomass. Decomposition of root biomass in established pastures releases P that is either recycled or transformed into soil organic P. Currently we know little about the role of occluded and organic forms of P and their roles in P cycling in pastures on the highly weathered soils of the Amazon. At Nova Vida, we suggest that the availability of P in older pastures is probably maintained by faster recycling of organic P and some equilibrium between organic and occluded P forms, possibly mediated by mycorrhizae associated with grass roots.
Parallels with Phosphorus Transformations During Soil Formation
A model of P transformation during geological soil development from parent material in the absence of human influence (Walker and Syers, 1976) provides an excellent framework for understanding P transformations that occur following tropical land use change. This model states that during the initial weathering during primary succession, P is released from Ca-phosphate primary minerals and transferred into nonoccluded or organic forms by soil biota. As soil weathering proceeds, Al and Fe oxides and sesquioxides begin to accumulate and nonoccluded P forms are encapsulated and transformed into occluded forms. At advanced stages of soil formation, when soil has been highly weathered and part of the initial P has been lost, most of the soil P is found in occluded forms or bound into recalcitrant organic forms. At this point, P available to support plant production depends on very efficient cycling of phosphate (Smeck, 1985).
On cleared lands, the one-way process by which P in primary minerals is ultimately converted to occluded and organic forms is reset by the cutting and burning of the forest, but instead of being released from primary minerals, P is released from the burned and decomposing biomass. This restarts a process of P transformation from burned material and residual above- and belowground organic debris. Because this occurs in an already weathered soil, conversion of P from nonoccluded forms to occluded and organic forms occurs in <50 yr instead of the thousands of years required during the primary succession described by Walker and Syers (1976). We found that the pulse of nonoccluded P that was generated by the burning of forest biomass was depleted in <10 yr. However, in contrast to the Walker and Syers model, nonoccluded P comprised a significant fraction of total soil P in both forests and pastures. A similar result was reported for a highly weathered soil in Hawaii by Crews et al. (1995), who found that almost 25% of the nonoccluded P pool persisted in a 4.1 million-yr-old Oxisol.
The primary succession P transformation model predicts that the increase in organic P that occurs during soil formation should occur while the available nonoccluded P is also increasing (Walker and Syers, 1976) and that this should be followed by a decline in soil organic P in response to P demand by plants and soil organisms (McGill and Cole, 1981). This decline in organic P, accompanied by a decrease in nonoccluded P, was observed by Crews et al. (1995). In our human-disturbed systems, the increase in organic P lagged behind the increase in nonoccluded P forms and coincided with the decline of the occluded P pool. The decline in the nonoccluded pool was followed by an increase of the occluded pool, suggesting more of a temporary role for occluded P than predicted by the Walker and Syers model. The exact nature of occluded P in weathered tropical soils is not known (Sanchez et al., 1991). Mycorrhyzal associations could play an important role in the cycling of P in these soils by tapping an occluded pool. It is also possible that what responds to chemical extraction as occluded P in the Phase 3 when pastures are established is not the same material that comprises occluded P in the forest or in very old pastures. Inputs of atmospheric dust over geologic time have been suggested as an agent that prevents P limitation in soils (Crews et al., 1995). In the long term, redeposition of P within the Amazon Basin by present and historical biomass burning may play a role in maintaining soil P availability in the landscape. However, we suggest that a redistribution of biomass belowground in pastures and recycling of organic and perhaps occluded P exert the major short-term controls on P availability in older pastures.
The time over which the redistribution of soil P after burning occurs may differ among soil types, topography, rainfall regimes, and pasture management. Inputs of C derived from pasture drive the process of organic matter accumulation in the soils of pastures at Nova Vida (Neill et al., 1997). But the trend toward increased pasture soil C does not occur in all Amazonian pastures after forest clearing. Lower soil clay content and initial C content seem to be important indicators of where soil organic matter accumulation will occur (Neill and Davidson, 1999). We currently know little about how these factors influence P distribution, but we hypothesize that transformations such as the decline in nonoccluded P in Phase 3 would occur more rapidly in soils with higher clay content and P absorption capacities (typically Oxisols) and that the absence of organic matter accumulation in the high-clay soils would result in more rapid declines in fertility and grass productivity.
Future Land Uses
Pasture now represents the largest single use of cleared forest land in most of the Amazon Basin (Fearnside, 1992; Instituto Nacional de Pesquisas Espaciais, 1998). Increasing areas of pasture in the Amazon are now degraded or experiencing declining grass production. Because P availability is the proximate limiting nutrient for grass growth, a better understanding of P turnover, particularly through organic P that accumulates under grass cover, will be essential for developing pasture management systems that maintain soil fertility and high forage production and maximize efficiency of fertilizer use. Reformation of pastures by disking, liming, and fertilizing and conversion of pastures to row crops are becoming more common (Nepstad et al., 1991; Smith et al., 1995). An understanding of P transformations in these systems and long-term studies of soil P and agricultural yield are likely to be key to sustainable management and recovery of these systems.Neill Davidson 2000; SPSS Inc 1997
|
ACKNOWLEDGMENTS |
|---|
Received for publication October 7, 1999.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResultsDiscussionREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
nia.
