Methane Fluxes

1
Fluxes of methane from soils in the Central
Amazon The role of agroforestry systems
Marco A. Rondón, Erick C.M. Fernandes, Elisa
Wandelli and Rubenildo Lima Da Silva
Department of Crop and Soil Sciences, Cornell
University, Ithaca, NY 14853 Centro de Pesquisa
Agroflorestal da Amazônia (EMBRAPA-CPAA), AM-10,
km 29, Manaus, AM, 69011-970 Correspondence
mar31_at_cornell.edu
1. INTRODUCTION Forests soils have been found to
be net sinks for atmospheric methane during most
of the annual cycle in several tropical regions
(Davidson et al, 1995) and are considered to play
an important role in global methane budgets (Dorr
et al, 1993). When primary forest is converted to
agriculture or pastures, the methane sink is
usually reduced and eventually soils could become
net sources (Keller et al, 1997). In the last
three decades at least 30 x 106 ha of Amazon
forest has been converted to pasture. Poor
management leading to a drastic decline in
pasture productivity and invasions by persistent,
herbaceous and woody weeds have led to most being
abandoned within a few years of use. This
provokes clearing of additional primary forest
for new pasture establishment. Considering the
vast area involved, net changes in fluxes of
methane and other GHG associated with the
continued conversion of Amazonian forest into
pasture could have effects at regional and global
scales. It is well known that termites emit
methane to the atmosphere. In degraded lands in
the Central Amazon, termite mounds are abundant
and they appear to negatively impact primary
productivity by inhibiting re-colonization by
vegetation. They may also contribute to net
methane fluxes. Our LBA project considers two
strategies for regenerating degraded pasture land
aimed at breaking the vicious circle of
deforestation Restoring productivity through
modest inputs of P and Ca and the establishment
of agroforestry systems (AFS) see McCaffery et
al, this session. Regenerated pastures are
expected to enhance methane sinks, while AFS
growing on formerly degraded lands have been
proven to improve soil quality, increase soil
faunal activity and diversity (Fernandes et al,
1995) and would likely also result in positive
impacts on methane balances. However, data on
methane fluxes between soils and the atmosphere
from AFS in the tropics is notably scarce.

• 2. OBJECTIVES
• To assess net fluxes of methane from
  soils under three agroforestry systems (AS1,AS2
  and ASP), primary rainforest and secondary
  forest (SF).
• To determine the effect of fertilization
  (P, Ca and Gypsum) on fluxes of methane from
  pastures.
• To make preliminary estimates of methane
  fluxes from termite mounds under agroforestry
  systems and secondary vegetation.

5. DISCUSSION Due to its influence on gas
diffusion, soil texture is an important factor
associated with fluxes of methane under native
vegetation (Dorr et al, 1993). Campina forest
located in a soil with 94 sand (high gas
diffusivity) showed higher sink fluxes compared
to primary "terra firme" forest on a clayey soil
(75 clay). Native forest (campina and terra
firme) are clearly the strongest methane sink
(figure 6), with an estimated annual oxidation
rate of 4,5 to 6,9 g CH4.ha -1y-1 (Only the drier
months have been evaluated so far). These
estimates coincide with reported values for
tropical rainforest (Keller et al,
1997). Conversion of forest to pastures
resulted in drastic net reductions in methane
oxidation by soils, almost eliminating net sinks
during the sampled period. This is in agreement
with findings for other rainforests (Keller and
Reiners, 1994, Goureau et al, 1993). Secondary
forest also showed only negligible net methane
sinks, comparable to those of degraded pastures.
This could be attributed to "leaks"coming from
abundant termite mounds that are found in this
vegetation. Figure 6 shows how AS1 and AS2
Agroforestry systems allowed a recovery of
methane sinks to around 50 of forest values, but
the pasture based ASP reached only nearly 25 of
forest sink strength. Initial data suggest that
the application of modest levels of P and Ca to
recover degraded pastures could enhance methane
sinks in the region. Fluxes of methane by
termites. Though the proportion of the area
covered by termite mounds is less than 2 of
total area (Queiroz, 2001), methane fluxes in SF
and in Pasture sites are at least two orders of
magnitude higher than recorded sinks by soils in
the same plots. This suggests that termites
could control methane budgets in such areas. Our
estimates resulted in higher emission rate from
soil termite mounds than reported data by Martius
et al (1993) for above-ground termite mounds. Our
data set is still limited and the does not yet
cover the full annual cycle thus net annual
estimates are provisional. Aknowledgements
Thanks to Antonio Nobre, Eleusa Barros, and
INPAs Agronomy laboratory personnel for
allowing us to use their GC and for providing a
friendly working environment. Thanks to all
students and staff of the LBA ND0-4 project for
continued support and productive discussions.
Photographs by E. Fernandes, K. McCaffery and M.
Renjifo.
4. PRELIMINARY RESULTS Precipitation Figure 1
shows weekly rainfall distribution for the
experimental area. (data from McCaffery, 2001).
Methane fluxes in primary and secondary forests
Figure 2 shows that in agreement with results
from other tropical areas (Scharfe et al, 1990),
forest soils in the Central Amazon are net sinks
of methane during the dry season. Sandy
"Campina"soils constitute the largest sink within
the land uses studied, although we only have late
dry season data evaluated for this ecosystem.
Early wet season data are currently being
processed. SF soils exhibit small net methane
fluxes during the early dry season, but later
become a small sink again until the onset of
rains. It is feasible that observed fluxes in SF
are related to high termite populations and
activity. Intense rainfalls in late December
triggered a change in fluxes from sinks to small
sources. Methane fluxes in Agroforestry
systems As shown in figure 3, during the dry
season all the AFS sites oxidized atmospheric
methane. As the soil becomes more humid with the
arrival of the rainy season, soils reduced their
sink strength and became a net source of
atmospheric methane. Both AS1 and AS2 were
successful in enhancing methane sink compared to
ASP which consistently showed reduced sink
capacity. Temporal variability in flux estimates
was high especially in pastures and was more
homogeneous in AS2. Methane fluxes in fertilized
pastures Figure 4 illustrates that fertilization
appears to slightly enhance methane sinks at the
end of the rainy season. Fertilizer was applied
in September and therefore a time lag seems to
exist before an effect is evident. Pastures,
including ASP, showed consistently lower methane
sinks or higher sources. This is in agreement
with studies done in the region (Goreau et al,
1988). Methane emissions from termite mounds
Figure 5 shows average fluxes from 12 mounds
sampled in each land use. All mounds sampled in
SF were active (emitting methane) while in other
sites only 60 to 70 were active.
3. METHODS Site Three repetitions of four AFS
were planted (3/92) at the EMBRAPA/CPAA Research
station, km 54 on BR-174 north of Manaus, on a
Xanthic Hapludox in an area of abandoned degraded
pastures Palm-based (AS1) Fruit/Timber-based
(AS2) Silvopastoral-High Inputs (ASP) were
established after a slash and burn in 11/91 and
three secondary forest (SF) control plots
demarcated. Primary forest on the same soil type
(Forest) and "Campina", a dwarf forest on a
Quartzipsament were also studied. In the same
area, 12 year old pastures of Brachiaria sp. were
fertilized on 10/00 with P, Ca and gypsum (Welch,
2000). Twelve termite mounds from each land use
were sampled in early January 2001. Gas sampling
Four closed vented chambers (25cm diameter) were
used to sample gas in each of AS1, AS2, ASP, SF,
Forest and Campina, as well as in fertilized and
control pastures. Weekly to bi-weekly
measurements were initiated on August 2000 and to
date, they cover part of the dry season and the
onset of the rainy season. Gas samples were
analyzed within 24 hours after collection using a
Perkin Elmer model GC400 gas chromatograph. (FID
detector, 3m Porapak Q column, 75C, carrier gas,
He).
6. REFERENCES CITED see Handout