Recuperation of nitrogen cycling in Amazonian
forests following agricultural abandonment
Eric A. Davidson1, Cla ´udio J. Reis de Carvalho2, Adelaine Michela Figueira3, Franc ¸oise Yoko Ishida3,
Jean Pierre H. B. Ometto3, Gabriela B. Nardoto3, Renata Tuma Saba ´2, Sanae N. Hayashi4, Eliane C. Leal4,
Ima Ce ´lia G. Vieira4& Luiz A. Martinelli3
Phosphorus (P) is generally considered the most common limiting
nutrient for productivity of mature tropical lowland forests grow-
land-use change may alter the stoichiometric balance of nutrient
cycling processes. In the Amazon basin, about 16% of the original
forest area has been cleared6, and about 30–50% of cleared land is
estimated now to be in some stage of secondary forest succession
ical soilsexhibit conservative N-cycling properties much like those
of N-limited forests on younger soils in temperate latitudes. As
secondary succession progresses, N-cycling properties recover
and the dominance of a conservative P cycle typicalof mature low-
land tropical forests re-emerges. These successional shifts in N:P
initially lower and then gradually increasing soil emissions of the
during secondary forest succession, demonstrated here over deca-
revealing that N availability in terrestrial ecosystems is ephemeral
and can be disrupted by either natural or anthropogenic distur-
bances at several timescales.
Ecologists have long noted that tropical forests growing on highly
weathered soils exhibit conservative P-cycling processes, whereas
conservative N-cycling properties are more common on younger
soils, including most temperate forests and montane forests1,2. This
pattern was demonstrated along a soil age chronosequence in the
Hawaiian Islands3, where N, which is derived primarily from the
atmosphere, is in short supply in the youngest volcanic soils and
abundant in young soils but becomes bound in unavailable forms to
of years. In global-scale analyses, the N:P ratios of green foliage4and
ing latitude, indicating generally increasing P conservation and
decreasing N conservation with soil age.
Although these stoichiometric generalizations seem robust for
mature forests, accelerating land-use change is altering tropical land-
ary forests are unclear8. Secondary tropical forests are playing an
increasingly important part in maintaining genetic diversity9and
hydrological functioning of altered landscapes10, but biogeochemical
Forest clearing causes an initial loss of nutrients from Amazonian
terrestrial ecosystems through fire, erosion, soil emissions of gases,
harvesting of timber and hydrologic leaching of nutrients8,11.
Additional losses occur as cattle or crops are harvested and as fire
is used as a management tool to prepare fields for planting and to
control pasture weeds12. Both N and P can be lost as particulates
during biomass burning13, but, in contrast to P, N is also volatilized
as a gas. Nitrate also generally leaches from soils more readily than
does phosphate. As a consequence of these N losses, net N miner-
alization, net nitrification, nitrate leaching and soil efflux of N2O
often decline as tropical cattle pastures age14–17. Fertilization can
maintain agricultural productivity,butwherefertilizationisnoteco-
forest begins to grow. Aggrading forests create a strong demand for
essentialplantnutrients. Theobjective ofthisstudywasto usespace-
for-time substitutions in secondary forest age chronosequences to
tropical secondary forest succession.
Three forest-age chronosequences, including stands ranging in age
from 3 to 70yr and remnant mature forests, were established in the
Brazilian state of Para ´, in eastern Amazonia. A complete set of seven
indicators of N-cycling rates was obtained for the chronosequence in
our main study site in the municipality of Sa ˜o Francisco do Para ´. To
provide true replication, a second chronosequence was established in
the municipality of Capita ˜o Poc ¸o, and a third chronosequence was
cipality of Paragominas. The Sa ˜o Francisco do Para ´ and Paragominas
municipalities are about 200km apart, with Capita ˜o Poc ¸o roughly in
the middle. Not all of the indicators could be measured in the second
sequences, but was relatively uniform within each chronosequence
(Supplementary Table 1). The dominant vegetation of the region
was once moist lowland tropical forest, but is now a mosaic of sec-
ondary forests, agricultural fields, cattle pastures and tree crops18(see
Supplementary Information for more site information).
All of the indicators derived from analysis of green foliage, litter-
fall, soil and trace gas emissions are consistent with a conservative N
cycle in the young successional forests, recovery of N-cycling pro-
cesses as succession proceeds, and a leaky N cycle in advanced stages
ofsecondary successionandinmature forests(Fig.1).Thelog-linear
occur early during succession and that the rate of change declines as
the secondary forests mature.
The first indicator, foliar15N, increases with increasing forest age
in both Sa ˜o Francisco do Para ´ and Capita ˜o Poc ¸o chronosequences
1The Woods Hole Research Center, 149 Woods Hole Road, Falmouth, Massachusetts 02540-1644, USA.2EMBRAPA Amazo ˆnia Oriental, C. P. 48, Bele ´m, PA 66.095-100, Brazil.
3CENA,UniversityofSa ˜oPaulo,AvenueCentena ´rio,303,Piracicaba,SP13.416-000,Brazil.4DepartamentodeBota ˆnica,MuseuParaenseEmı ´lioGoeldi,Bele ´m,PA66.040-179,Brazil.
Vol 447|21 June 2007|doi:10.1038/nature05900
(Fig. 1a). The foliar15N values of the mature forests of these sites are
within the range commonly measured in mature tropical forest foli-
age (3.7%63.5; error ofonestandard deviation19), butthe valuesin
the youngest forest at Sa ˜o Francisco do Para ´ (20.5%) are as low as
those usually measured in temperate forest foliage (22.8%6 2.0;
error of one standard deviation19). Enriched foliar15N in the older
forests is indicative of a leaky N cycle, in which isotopically light N
is lost from the ecosystem owing to fractionation during nitrifi-
cation and denitrification, leaving isotopically enriched N behind20.
In contrast, little N is lost in the conservative N cycle of young
secondary forests, resulting in little N fractionation or15N enrich-
increasing forest age (Fig. 1b), provide a second indicator of increas-
ing N availability as secondary forest succession progresses.
Third, the N:P ratios of fine litterfall increased with forest age
(Fig. 1c) for all three chronosequences. The mean litterfall N:P ratios
in the mature forests (50–68) bound the global mean of 62 for trop-
ical forests5. Incontrast, the mean litterfall N:P valuesof 31and 24in
the youngest forests of the Sa ˜o Francisco do Para ´ and Paragominas
chronosequences are closer to the global mean of 29 for broadleaf
temperate forests5. The mean N:P ratio in fresh foliage ranged from
27 to 35 and did not vary systematically across forest ages. Hence,
only in the advanced successional and mature forests did the trees
reabsorb more P relative to N before leaf abscission. Fourth, the
litterfall mass:N ratio also declined with forest age (Fig. 1d). The
values in the young forests (100–140) are similar to N-limited tem-
perate forests, whereas the advanced and mature forests values (68–
88)are common for maturelowland tropical forests1.Although both
foliar N and P may be reabsorbed before leaf fall, the balance shifts
from N economy to P economy in litterfall during secondary forest
Fifth, extractable soil nitrate increased with forest age (Fig. 1e),
indicating increasing availability of soil N as the forests mature.
Ammonium (which tended to decrease with forest age, but not sig-
nificantly so) is often the dominant form of inorganic soil N in
N-limited systems, whereas nitrate accumulates where available N is
able nitrate to extractable ammonium increased from values#1 for
forests#20yr to values.1 for forests$40 yr (Fig. 1f).
Seventh, although the more clay-rich Paragominas site exhibited
higher soil emissions of N2O than the sandy Sa ˜o Francisco do Para ´
site, N2O emissions increased with forest age at both sites (Fig. 1g).
Higher rates of nitrate leaching inmature forests compared to young
forests have also been measured at the Paragominas site22. These
differences in N losses across forest ages reflect current biogeochem-
ical fluxes and are consistent with the time-integrated indicator of
ecosystem N loss and fractionation provided by foliar15N analyses20,
which reflects decades of previous slash-and-burn management in
the young forests and decades of regrowth in the older forests.
These results have important implications for rates of regrowth of
which dependsontheaccumulationofnutrients infallowvegetation
as the source of nutrients for the next cycle of slash-and-burn agri-
culture. Rates of secondary forest growth in Amazonia have been
inversely correlated with the number of fires during the agricultural
phases23. A large fraction of biomass N is often lost during fires13,
depletingthepoolofactively cyclingecosystem NandprovokingaN
limitation and a conservative N cycle after repeated fire8. The time
agricultural phase increases when the pool of available nutrients,
such as available N, declines with each fire cycle. Further evidence
for nutrient limitation comes from nutrient amendment experi-
ments in which biomass accumulation in young Amazonian second-
ary forests responded to N and not P additions in one study24and
showed species-specific responses in another study25.
Although lower emissions of N2O in secondary compared to
mature tropical forests have been previously observed14,17,26,27, here
Foliar 15N (‰)
Foliar N (g kg–1)
1 10 100
Soil NO3– (mg N kg–1)
Forest age (years)
N2O (kg N ha–1 yr–1)
Figure 1 | Indicators of N and P cycling along secondary successional
forest chronosequences. a–g, Foliar delta15N (a); foliar N
nitrate (e); soil nitrate:ammonium ratios (f); and annual soil emissions of
nitrous oxide (g). Error bars indicate s.e.m. within each forest age at each
chronosequence location. The chronosequences are represented as circles
for Sa ˜o Francisco do Para ´, squares for Capita ˜o Poc ¸o and triangles for
Paragominas. The effect of forest age, either as a ranking factor or as the
logarithm of age (assuming an age of 200yr for the mature forest; see
Methods) is significant in analysis of covariance (ANCOVA) for all seven
indices (P,0.05; see Supplementary Table 3 for all P values). The site-
effect is significant for foliar15N, litterfall N:P, litterfall mass:N and N2O.
Results are nearly identical when the mature forests (of unknown age) are
omitted from the analysis, except that the age-effect is not significant for
N2O and is only marginally significant for the nitrate:ammonium ratio
(P50.119 and P50.057, respectively).
NATURE|Vol 447|21 June 2007
we show that increasing emissions of N2O as the successional forests
age can be understood in terms of gradual recuperation of several
N-cycling processes. A legacy of large N losses during an agricultural
phase would probably result in slow rates of recovery of N cycling8
and slow rates of increases of N2O emissions during secondary suc-
cession. Conversely, acceleration of N-cycle recuperation and rapid
increases in N2O emissions could result from the emergence of a
dominant N-fixing species during secondary forest succession27.
Recent isotopic evidence suggests that biological N fixation may
25yr of secondary succession of Amazonian forests28, although
no difference in foliar15N enrichments between legume and non-
legume species was observed (Supplementary Table 2), although
leguminous species had higher foliar N concentrations and lower
often exhibit N-rich tissues without necessarily fixing significant
amounts of nitrogen29. We can neither rule out nor support the
importance of N fixation in this study. Atmospheric deposition
inputs of nitrogen in this region (2–6kgha21yr21; ref. 22) are in
the same range of 3–8kgha21yr21that accumulates in woody bio-
mass of these successional forests, but are not enough to account for
about 11kgha21yr21that accumulates in foliar biomass during
canopy development22,24. The top 10cm of mineral soil contains
ample organic-N stocks ($1000kgha21of N atoms) that could
supply the regrowing forest if a small fraction is gradually minera-
lized to a bioavailable form, and considerably more is present at
lower soil depths22. The rate of recuperation of N-cycling processes
during secondary succession may reflect, in part, the kinetics of
mobilization of recalcitrant forms of soil N to an actively cycling N
pool, as well as the legacy of the degree of degradation during agri-
The patterns of N and P cycling shown here for secondary succes-
sion parallel those previously demonstrated for primary succession.
Actively cycling N interrestrial ecosystems canbelost either byland-
use change, such as forest clearing, burning and agricultural prac-
tices, or by natural processes such as fires, landslides, glaciers and
volcanic activity. Just as accumulation of total ecosystem N alleviates
an N limitation as soils age over thousands and millions of years,
actively cycling N accumulates over decades and centuries during
secondary forest succession, resulting in a similar successional tra-
jectory from a conservative N cycle following agricultural abandon-
ment to the leaky-N and conservative-P cycles expected in mature
lowland tropical forests on old soils.
In the Sa ˜o Francisco do Para ´ municipality, two farms were selected to represent
eachforestageof3,6,10,20,40and70yr.IntheCapita ˜oPoc ¸omunicipality,one
farm was identified for each forest age of 6, 10, 20 and 40yr. Two 20m320m
plots were established on each farm. These forests were regrowing after aban-
donment of crop fields. Two plots were also establishedin eachremnant mature
forest at Sa ˜o Francisco do Para ´ and Capita ˜o Poc ¸o. Measurements were made
from October 2000 through to June 2002 in Sa ˜o Francisco do Para ´, and from
February 2004 to January 2005 at Capita ˜o Poc ¸o. For the Paragominas chrono-
the effects of site (Sa ˜o Francisco do Para ´, Capita ˜o Poc ¸o and Paragominas) and
of uncertainty of the ages of the mature forests, the effects of forest age were
analysed three ways. First, age was converted to a rank score from 1 to 7 (3-, 6-,
thelogarithm of forest age was used as a continuous variable. Third, thisanalysis
wasrepeatedwiththedatafrom mature forests excludedtotest theeffectsoflog-
of these statistical tests are presented in the Fig. 1 legend and in Supplementary
Full Methods and any associated references are available in the online version of
the paper at www.nature.com/nature.
Received 11 April; accepted 2 May 2007.
1. Vitousek, P. M. Litterfall, nutrient cycling, and nutrient limitations in tropical
forests. Ecology 65, 285–298 (1984).
15, 1–19 (1976).
of tropical forest development. Ecology 84, 2231–2255 (2003).
Reich, P. B. & Okeksyn, J. Global patterns of plant leaf N and P in relation to
temperature and latitude. Proc. Natl Acad. Sci. 101, 11001–11006 (2004).
McGroddy, M. E., Daufresne, T. & Hedin, L. Scaling of C:N:P stoichiometry in
forests worldwide: implications of terrestrial redfield-type ratios. Ecology 85,
INPE (Instituto Nacional de Pesquisas Espaciais) Monitoramento Ambiental da
Amazo ˆnia por Sate ´lite Æhttp://www.obt.inpe.br/prodes/
Hirsch, A. I., Little, W. S., Houghton, R. A., Scott, N. A. & White, J. D. The net
carbon flux due to deforestation and forest re-growth in the Brazilian
Amazon: analysis using a process-based model. Glob. Change Biol. 10, 908–924
McGrath, D. A., Smith, C. K., Gholz, H. L. & de Assis Oliveira, F. Effects of
land-use change on soil nutrient dynamics in Amazo ˆnia. Ecosystems 4, 625–645
Vieira, I. C. G., de Paiva Saloma ˘o, R., de Araujo Rosa, N., Nepstad, D. C. & Roma,
the wake of agriculture.) Cie ˛ncia Hoje 20, 38–45 (1996).
10. Nepstad, D. C., Moutinho, P. & Markewitz, D. in Biogeochemistry of the Amazon
Basin and its Role in a Changing World (eds McClain, M. E., Victoria, R. L. & Richey,
J. E.) 139–155 (Oxford Univ. Press, New York, 2001).
11. Davidson, E. A. et al. in Ecosystems and Land Use Change. Geophysical Monograph
Series 153 (eds DeFries, R., Asner, G. & Houghton R.) 147–158 (American
Geophysical Union, Washington DC, 2004a).
Basin (eds McClain, M. E., Victoria, R. L. & Richey, J. E.) 84–105 (Oxford Univ.
Press, New York, 2001).
13. Kauffman, J. B., Cummings, D. L., Ward, D. E. & Babbitt, R. Fire in the Brazilian
104, 397–409 (1995).
14. Davidson, E. A., Keller, M., Erickson, H. E., Verchot, L. V. & Veldkamp, E. Testing a
conceptual model of soil emissions of nitrous and nitric oxides. Bioscience 50,
trace-gas emissions from a deforested area of Costa Rica. Nature 365, 244–246
16. Melillo, J. M. et al. Nitrous oxide emissions from forests and pastures
of various ages in the Brazilian Amazon. J. Geophys. Res. 106, 34,179–34,188
17. Verchot, L. V. et al. Land use change and biogeochemical controls of nitrogen
oxide emissions from soils in eastern Amazonia. Global Biogeochem. Cycles 13,
18. Vieira, I. C. G. et al. Classifying successional forests using Landsat spectral
properties and ecological characteristics in eastern Amazonia. Remote Sens.
Environ. 87, 470–481 (2003).
19. Martinelli, L. A. et al. Nitrogen stable isotopic composition of leaves and soil:
Tropical versus temperate forests. Biogeochemistry 46, 45–65 (1999).
20. Amundson, R. et al. Global patterns of the isotopic composition of soil and plant
nitrogen. Global Biogeochem. Cycles 17, 1031 (2003).
21. Vitousek, P. M., Gosz, J. R., Grier, C. C., Melillo, J. M. & Reiners, W. A. A
comparative analysis of potential nitrification and nitrate mobility in forest
ecosystems. Ecol. Monogr. 52, 155–177 (1982).
22. Markewitz, D., Davidson, E. A., Moutinho, P. & Nepstad, D. C. Nutrient loss and
redistribution after forest clearing on a highly weathered soil in Amazonia. Ecol.
Appl. 14, S177–S199 (2004).
23. Zarin, D. J. et al. Legacy of fire slows carbon accumulation in Amazonian forest
regrowth. Front. Ecol. Environ. 3, 365–369 (2005).
24. Davidson, E. A. et al. Nitrogen and phosphorus limitation of biomass growth in a
tropical secondary forest. Ecol. Appl. 14, S150–S163 (2004b).
25. Gehring, C., Denich, M., Kanashiro, M. & Vlek, P. L. G. Response of secondary
vegetation in Eastern Amazonia to relaxed nutrient availability constraints.
Biogeochemistry 45, 223–241 (1999).
26. Keller, M. & Reiners, W. A. Soil–atmosphere exchange of nitrous oxide,
nitric oxide, and methane under secondary succession of pasture to forest in
the Atlantic lowlands of Costa Rica. Global Biogeochem. Cycles 8, 399–409
27. Erickson, H., Keller, M. & Davidson, E. Nitrogen oxide fluxes and nitrogen cycling
during post-agricultural succession and forest fertilization in the humid tropics.
Ecosystems 4, 67–84 (2001).
28. Gehring, C. Biological nitrogen fixation in secondary regrowth and mature
rainforest of central Amazonia. Agr. Ecosyst. Environ. 111, 237–252 (2005).
NATURE|Vol 447|21 June 2007
29. Vitousek, P. M. et al. Towards an ecological understanding of biological nitrogen
fixation. Biogeochemistry 57, 1–45 (2002).
Supplementary Information is linked to the online version of the paper at
for assistance with establishing the chronosequences, R. Figueiredo for assistance
with project and data management, and the Large-Scale Biosphere–Atmosphere
(LBA) training and education programme for student stipends. This work was
supported by grants from the LBA-Ecology programme of NASA.
supervised the field work of F.Y.I. and R.T.S., and wrote the paper. C.J.R.d.C.
conductedlaboratory analyses of soils and litter. A.M.F. collected foliar samples at
Sa ˜oFrancisco andconducted isotopic analyses,underthesupervision ofJ.P.H.B.O.
S.N.H. conducted litterfall studies and E.C.L. collected soils and fresh foliage at
Capita ˜o Poc ¸o, both under the supervision of I.C.G.V. A.M.F., F.Y.I., J.P.H.B.O.,
G.B.N., and L.A.M. contributed to an early draft of the manuscript in Portuguese,
andL.A.M.supervisedthework ofA.M.F.I.C.G.V. establishedthechronosequence
Author Information Reprints and permissions information is available at
www.nature.com/reprints. The authors declare no competing financial interests.
Correspondence and requests for materials should be addressed to E.A.D.
NATURE|Vol 447|21 June 2007
Site descriptions. The remnant mature forests in the municipalities of Sa ˜o
Francisco do Para ´ and Capita ˜o Poc ¸o are located at 1u159S, 47u479W and
02u129S, 47u229W, respectively, and the secondary forests are located in those
vicinities. The Fazenda Vitoria ranch in the municipality of Paragominas is
located at 2u599S, 47u319W. Mean annual precipitation is 2,200mm for Sa ˜o
Francisco do Para ´ and Capita ˜o Poc ¸o and 1,800mm for Paragominas. The dom-
inantsoilsareTypicHapludultsinSa ˜oFranciscodoPara ´ andCapita ˜oPoc ¸o,and
Typic Hapustox at the Paragominas ranch (see Supplementary Table 1 for sur-
face soil characteristics). Human settlement in this region expanded during the
for distributing rural lands for agricultural development, mostly by small land-
now undergone nine or more cycles of slash-and-burn agriculture18. Cattle
ranching and logging were the main forces of deforestation in the 1960–80s in
Paragominas, including the ranch used in this study, which was established in
1969 (ref. 22).
Method details. Soil emissions of N2O were measured at Sa ˜o Francisco do Para ´
and Paragominas using syringe sampling of static chambers and gas chromato-
perdateineachofthetwenty-sixplotsatSa ˜oFranciscodoPara ´,withfivedatesin
each of the dry and wet seasons at Sa ˜o Francisco do Para ´. Three collections
(0.25m2per collection) of fine litterfall were made monthly for a year in each
plot at both Sa ˜o Francisco do Para ´ and Capita ˜o Poc ¸o. Soil inorganic N was
extracted in 1M KCl from triplicate soil samples collected from the top 10cm
20- and 40-yr-old successional forests and the mature forest of Sa ˜o Francisco do
Para ´ and Capita ˜o Poc ¸o. At Sa ˜o Francisco do Para ´, fully expanded leaves were
collected for the dominant species at each site, according to the species import-
ance values indices18. At Capita ˜o Poc ¸o, all leaves within 1m32m miniplots
were harvested and composited. Finely ground foliar samples were analysed for
C and N concentrations using a Carlo-Erba CHN analyser, for C and N stable
for P by acid digestion followed by colorimetric spectrophotometry.