Content uploaded by John A. Parrotta
Author content
All content in this area was uploaded by John A. Parrotta on Feb 24, 2019
Content may be subject to copyright.
An evaluation of direct seeding for reforestation of degraded
lands in central Sa
Äo Paulo state, Brazil
Vera Lex Engel
a,1
, John A. Parrotta
b,*
a
Natural Resources Department, Universidade Estadual Paulista, UNESP/FCA, P.O. Box 237, 18603-970 Botucatu, SP, Brazil
b
International Institute of Tropical Forestry, USDA Forest Service, P.O. Box 25000, Rõ
Âo Piedras, PR 00928-5000, USA
Received 7 June 2000; accepted 2 September 2000
Abstract
As part of a larger study evaluating several silvicultural techniques for restoring tropical moist forests on abandoned
agricultural lands in southeastern Brazil, direct seeding with ®ve early-successional Atlantic forest species was tested at three
degraded sites, characterized by different soil types and land-use histories, within the Environmental Protection Area at
Botucatu, SP. The species used in this study were Chorisia speciosa,Croton ¯oribundus, Enterolobium contorstisiliquum,
Mimosa scabrella, and Schizolobium parahyba. Scari®ed seeds of each of these species were sown in prepared seed spots in
replicated, 0.25 ha mixed-species plots at an initial espacement of 1 m 1 m at each site. Of the ®ve species planted, only two,
Enterolobium and Schizolobium, showed good seed germination, seedling survival, and early growth rates, averaging 4.1±
4.6 cm stem diameter and 1.5±1.7 m height growth during the ®rst 2 years after sowing. These two species constituted 88±
100% of the total stand density, which ranged from 1050 to 1790 stems ha
1
at 2 years. Despite the poor performance of the
other species tested, we observed that the natural regeneration of native forest species originating from remnant forests in the
general vicinity of our study sites was signi®cantly greater within the direct-seeded plots than in unplanted control plots that
were protected from ®re and other disturbances. Published by Elsevier Science B.V.
Keywords: Forest restoration; Native species; Natural regeneration; Seedling growth; Tropics
1. Introduction
The Atlantic forest formation in Brazil, which
includes dense evergreen forests, semideciduous sea-
sonal forests and gallery forests, is one of the worlds's
greatest centers of tropical biodiversity as well as one
of the most threatened by deforestation and degrada-
tion (FundacËa
Äo SOS Mata Atla
Ãntica, 1992; Bryant
et al., 1997). These forests formerly occupied a large
area of eastern Brazil from Bahia to Rio Grande do
Norte (Ca
Âmara, 1991). This region, home to an esti-
mated 70% of Brazil's total population of 150 million,
has a long history of deforestation and soil resource
degradation associated with numerous cycles of agri-
cultural development in Brazil since the colonial era.
Today, the total area of native forests is estimated
at less than 10% of its original extent (FundacËa
Äo SOS
Mata Atla
Ãntica/INPE, 1993). Outside of protected
conservation areas, most remaining forest fragments
are disturbed with respect to their structure and
function, and are under constant threat from ®re
and other human disturbances. Current environmental
Forest Ecology and Management 152 (2001) 169±181
*
Corresponding author. Tel.: 1-787-766-5335;
fax: 1-787-766-6263.
E-mail addresses: veralex@fca.unesp.br (V.L. Engel),
jparrotta@fs.fed.us (J.A. Parrotta).
1
Tel.: 55-14-6802-7168; fax: 55-14-6821-3438.
0378-1127/01/$ ± see front matter. Published by Elsevier Science B.V.
PII: S 0378-1127(00)00600-9
legislation in Brazil aimed at reversing deforestation
trends and protecting the region's agricultural soils,
rivers, and their hydroelectric generation capacity,
requires protection of remaining natural forests in this
region and the restoration of forests on 20% of the total
land area on all rural properties, particularly in ripar-
ian sites. However, the owners of small- and medium-
sized land holdings, who generally have severely
limited ®nancial resources, rarely participate in envir-
onmental rehabilitation projects due to their high costs
and lack of direct ®nancial returns.
A proportion of the deforested lands in the Atlantic
forest region can and should be rehabilitated for
agricultural production. However, there are signi®cant
areas of degraded, agriculturally marginal lands that
cannot be economically rehabilitated for either agri-
cultural or intensive commercial forestry production
in the near term, as well as lands of high potential
value for conservation and watershed protection.
These include degraded rangelands and pastures,
riparian areas, steep slopes subject to severe erosion,
or sites that could serve as corridors linking forest
fragments of high biodiversity value. While some of
these lands in the vicinity of remnant native forests
would naturally revert fairly quickly to secondary
forest if the pressures on them such as ®re and live-
stock grazing were reduced, other more isolated or
severely degraded sites will require some form of
management to facilitate their recovery. There is a
need to develop techniques for native forest restora-
tion in this region that are inexpensive to implement
and provide some level of direct, short-term economic
return to farmers and other landholders.
Forest plantings can play a key role in harmonizing
long-term forest ecosystem rehabilitation (or restora-
tion) goals with near-term socio-economic develop-
ment objectives (Brown and Lugo, 1994; Lamb and
Tomlinson, 1994). Recent studies have shown that
plantations established for the production of timber
and other forest products can facilitate, or ``catalyze''
native forest succession in their understories on sites
where persistent ecological barriers to succession
would otherwise prelude recolonization and enrich-
ment by native forest species (see, for example Soni
et al., 1989; Lu
Èbbe and Geldenhuys, 1991; Lugo, 1992;
Lugo et al., 1993; Parrotta, 1992, 1993, 1995, 1999;
Mitra and Sheldon, 1993; Kuusipalo et al., 1995; Lamb
et al., 1997; Parrotta and Turnbull, 1997; Lamb,
1998; Parrotta and Knowles, 1999). Such studies
have shown that the changes in understory microcli-
mate, increased vegetation structural complexity and
habitat diversity, and the development of litter and
humus layers that occur during the early years of
plantation growth often result in increased seed inputs
from neighboring native forests (by seed dispersing
wildlife attracted to the plantations), suppression of
competing grasses, and alterations in microclimatic
conditions that favor seedling survival and growth.
In the absence of silvicultural management aimed
at eliminating woody understory regeneration, the
plantation system is replaced by a mixed forest
comprised of the planted species and an increasing
number of early and late successional tree species
and other ¯oristic elements drawn from surrounding
forest areas.
To date, most of the silvicultural approaches deve-
loped in this region for forest restoration have not
been applied beyond an experimental scale, and those
that have been used over larger areas (by mining
and hydroelectric companies) have a limited applic-
ability for most landholders due to their high establish-
ment and maintenance costs (Maschio et al., 1992;
Kageyama et al., 1994). In 1997±1998 a research
project was established at three contrasting degraded
sites in the Environmental Protection Area at Botucatu
in Sa
Äo Paulo State, Brazil to evaluate ®nancial costs
and bene®ts, productivity and ecological impact of
four different silvicultural techniques for forest
rehabilitation that can be easily adopted by private
small- and medium-sized landowners in the region.
The objectives of this project are to evaluate the
ecological impact and the economic viability of sev-
eral plantation models utilizing a total of 47 native
tree species for restoring seasonal, semideciduous
Atlantic forests under different site conditions (vari-
able soil fertility and landscape ¯oristic patterns)
and management regimes. Speci®c objectives include
the evaluation of: the adaptability and productivity
of native Atlantic forest species on deforested sites
characterized by different levels of soil degradation;
the effects of these management systems on soil
productivity and belowground ecological processes;
patterns of natural regeneration and development
of plant and animal species diversity; and the
®nancial costs and bene®ts associated with each
model, or treatment.
170 V.L. Engel, J.A. Parrotta/ Forest Ecology and Management 152 (2001) 169±181
One of the treatments included in this experiment is
direct seeding using ®ve quick-growing, early succes-
sional, native tree species established in mixed-species
plantation blocks. The potential advantages of direct
seeding over other plantation establishment techni-
ques (i.e. planting of nursery-grown seedlings, wildl-
ings or rooted cuttings) include cost savings associated
with nursery care and planting, as well as the possi-
bility that trees established by this means may develop
more naturally, and quickly, than would transplanted
seedlings or cuttings. There are, however, signi®cant
disadvantages of direct seeding that usually outweigh
these advantages, i.e. typically very low germination
survival percentages resulting in either inadequate
plantation stocking and/or increased seed costs to
compensate for poor germination and survival, poor
early seedling growth relative to nursery-grown seed-
lings that receive daily care, and increased seedling
mortality associated with weed competition (or
increased weeding costs to overcome this) in addition
to increased susceptibility to poor weather conditions
(Evans, 1982). While direct seeding failures (rarely
reported in the literature) greatly outnumber suc-
cesses, it has been successfully used in many parts
of the world for establishment of tree crops such as
Acacia,Anacardium occidentale,Azadirachta indica,
Gmelina arborea,Swietenia, and Pinus in commercial
plantations, and for numerous other tropical/subtropi-
cal species in non-commercial plantations established
for various environmental rehabilitation purposes
(Laurie, 1974; Evans, 1982; Barbosa and Barbosa,
1992; Francis, 1993; Knowles and Parrotta, 1995; Sun
et al., 1995; Sun and Dickinson, 1995, 1996; Fer-
nandes et al., 1998; Parrotta and Knowles, 1999).
Despite generally rather poor record to date, we feel
that the potential cost savings associated with direct
seeding, particularly for species whose seeds are
readily available and amenable to this method of
establishment, could outweigh its disadvantages and
offer a more economical means for reestablishing
forest cover over large areas of degraded lands
(Thomson, 1992; Applegate et al., 1993). What is
required is a more systematic screening of potential
species and their response to direct seeding under ®eld
conditions, and evaluations of costs associated with
plantation establishment and aftercare relative to
those of more commonly used planting stock such
as nursery-grown seedlings.
In this paper, we present the early results of our
experiments with direct seeding at three contrasting
sites. Speci®cally, we examine the ®nancial costs for
establishment and maintenance, seed germination
rates, survival and tree growth during the ®rst 2 years
after establishment, and patterns of natural regenera-
tion by within plantation plots compared to unplanted
control plots.
2. Materials and methods
2.1. Site description
The study sites are located on two adjacent pro-
perties (Fazenda Lageado, Fazenda Edgardia: Fig. 1)
within the campus of the State University of Sa
Äo Paulo
at Botucatu (UNESP-FCA) in the south-central region
of Sa
Äo Paulo State (228500S; 488240W). The natural
vegetation in this area is classi®ed as seasonal semi-
deciduous tropical forest. The experimental sites
receive an average annual rainfall of 1300 mm, mostly
between the months of October and March. The mean
annual temperature is 19.48C, with monthly means
ranging from 16.38C in July to 21.98C in January. The
local topography is hilly, with elevations ranging from
464 to 775 m. Due to a high degree of geologic and
topographic variability, soils at these sites are also
variable, ranging from fertile clays of basaltic origin
to nutrient-poor, acidic sands (Table 1).
Site 1 (Fazenda Lageado) is located at 700 m
elevation in an area with red clay (``terra roxa'') soils
of optimal fertility and good physical properties but
with a tendency towards compaction. Prior to installa-
tion of the experiment the vegetation on this aban-
doned pasture site was dominated by very tall (2±3 m)
grasses (principally Napier grass, Pennisetum purpur-
eum) and pasture herbs. This site is situated near the
Ribeira
Äo Lavape
Âs ravine in a relatively isolated valley
surrounded by agricultural ®elds. Riparian forest frag-
ments are present approximately 200 m downslope
from the experiment site towards the river. Site 2
(Fazenda Edgardia) is located at 574 m elevation in
an area with red-yellow podzolic soils of medium
fertility. In recent decades the site was used as a Citrus
orchard. It is located approximately 50 m from a forest
remnant in a good state of conservation. The ground
cover is dominated by grasses, mainly Brachiaria
V.L. Engel, J.A. Parrotta/ Forest Ecology and Management 152 (2001) 169±181 171
decumbens. Site 3 (Fazenda Edgardia) is located at
540 m elevation in an area with moderately acidic and
leached quartz sand soils of very low fertility and
subject to severe erosion. The site was formerly
used (for several decades) as a pecan and mango
orchard; many of these now senescent trees remain
in the vicinity of experimental plots. Most recently,
the area has been used for livestock grazing and is
Fig. 1. Location of study sites at the Fazenda Lageado (Site 1) and Fazenda Edgardia (Sites 2 and 3), State University of Sa
Äo Paulo at
Botucatu (UNESP-FCA), Sa
Äo Paulo, Brazil.
Table 1
Physical and chemical properties of soils at study sites, 0±20 cm depth
a
Site 1 Site 2 Site 3
Soil texture
b
Clay Loamy sand Sand/loamy sand
Bulk density (g/cm
3
) 1.37 1.55 1.52
pH
c
5.3 4.8 4.5
Organic matter
d
(%) 1.99 0.64 0.67
N (%) 0.155 0.06 0.048
Exchangeable ions
e
(mmol/1000 cm
3
)
P
extractablee
(mg/1000 cm
3
) 9.3 8.1 2.3
K 3.1 1.2 1.3
Ca 46.5 10.7 7.0
Mg 14.9 4.5 2.5
Acidity (H Al) 32.8 20.4 18.5
Cation exchange capacity
KCaMgHAl
97.1 36.8 29.4
Base saturation (%) 66.4 44.6 36.7
a
Average values for all plots at each site (n15/site).
b
USDA classi®cation system.
c
Determined in 0.01 N CaCl
2
(Raij and van Quaggio, 1983).
d
% Organic carbon 1.75 (Camargo et al., 1986).
e
Ion exchange resin extraction (Camargo et al., 1986).
172 V.L. Engel, J.A. Parrotta/ Forest Ecology and Management 152 (2001) 169±181
dominated by ``colonia
Äo'' grass (Panicum maximum)
and B. decumbens. This site is located approximately
50±100 m from a secondary forest fragment which has
been degraded by ®re and wood extraction in the
recent past.
2.2. Experimental design
At each site, a set of ®fteen 0.25 ha (50 m 50 m)
experimental plots were established using a rando-
mized complete block design, with three replications
each of the following ®ve treatments: (1) an unplanted
control; (2) direct seeding of ®ve fast-growing early-
successional species using minimal tillage practices;
(3) a modi®ed taungya (agroforestry) system invol-
ving mixed-species line plantings of 22 fast-growing
fuelwood and slower-growing timber species, inter-
planted with annual crops during the ®rst 2±4 years;
and (4) a mixture of 27 multi-purpose and commercial
timber species; and (5) restoration plantings of 40
native forest species, including early-fruiting unders-
tory species.
In the direct seeding treatment, ®ve early succes-
sional, secondary forest species were chosen for study,
based on their ecological and silvicultural character-
istics (including availability of seed local forests or
commercial sources and ease of propagation) and
inferences on their adaptability to study site condi-
tions. These species are:
Chorisia speciosa St. Hil (local name ``paineira'':
family Bombacaceae), a deciduous forest gap
species up to 30 m tall;
Croton floribundus Spreng (``capixingui'': Euphor-
biaceae), a deciduous or semideciduous pioneer
attaining heights of 6±10 m;
Enterolobium contortisiliquum (Vell.) Morong.
(``timboril'': Leguminosae, Mimosoideae), a pio-
neer species 20±35 m tall at maturity;
Mimosa scabrella Benth. (``bracatinga'': Legumi-
nosae, Mimosoideae), a semidiciduous pioneer up
to 15 m tall; and,
Schizolobium parahyba (``guapuruvu'': Legumino-
sae, Papilionoideae), a deciduous pioneer species
found exclusively in the Atlantic forest region and
typically in riparian forests (Lorenzi, 1992).
With the exception of M. scabrella, whose seeds
were obtained from EMBRAPA's National Center for
Forestry Research in Curitiba, Parana
ÂState, the seed
sources for all other species were purchased from
various locations within Sa
Äo Paulo State through the
Forestry Institute of Sa
Äo Paulo. Prior to planting,
appropriate seed treatments were applied to break
dormancy: for Chorisia,Croton and Enterolobium,
seeds were soaked in water at room temperature for
12 h immediately prior to sowing; for Mimosa and
Schizolobium, seeds were immersed in hot water
(808C) for 2 min followed by soaking in water at room
temperature for 12 h. Germination tests were conducted
in the nursery for comparison with germination data
obtained in the ®eld. Seeds were sown in groups of
1±6 per planting spot (depending on species) at appro-
ximately 5 cm depth in prepared lines located along
slope contours. Each line was planted with a single
species, with a 1 m spacing between planting spots and
between lines, and each plot (at all sites) included the
same sequence of species in successive lines.
2.3. Site preparation, planting and
maintenance
Experimental plots at each site were fenced to
provide protection against grazing, and ®re lines
established around the periphery of the fences enclos-
ing each experimental area. Prior to planting, a post-
emergent herbicide (glyphosate) was applied to all
treatment and control plots to suppress grasses; at Site
1, the exuberant P. purpureum growth required roto-
tilling in advance of herbicide application. During the
®rst 2 years, additional spot applications of this her-
bicide were used, in addition to manual weeding as
required around seedlings to ensure seedling survival
and good early growth. Ant traps containing a for-
micide (Myrex
TM
, active ingredient sul¯uoramide)
were set up at selected spots within the plantations
to reduce herbivory damage to seedlings. Planting
lines were prepared (to a depth of 40 cm) using a
subsoiler pulled by a tractor, a practice commonly
used in minimal tillage agriculture and commercial
forestry in this region. Planting was carried out in early
1997 during the wet season (March).
2.4. Data collection and analysis
Seed germination and seedling survival were eval-
uated 1.5 and 24 months, respectively, after planting.
V.L. Engel, J.A. Parrotta/ Forest Ecology and Management 152 (2001) 169±181 173
Stem basal diameters (at 5 cm above the root collar)
and tree heights were measured were recorded for all
trees in each plot at 24 months. The species composi-
tion of the soil seed bank was characterized indirectly
at the start of the experiment by identifying and
enumerating all plants in ®ve randomly located
1m1 m quadrats in each of the study plots 60 days
after site preparation (and post-emergent herbicide
application). Approximately 2.5 years after planting,
all woody plants (trees and shrubs) that had regener-
ated naturally (i.e. those not planted) since plot estab-
lishment were identi®ed and enumerated in the direct
seeding and control plots.
Operational costs and labor requirements (worker
hours) for activities related to plantation establishment
and maintenance were recorded during throughout the
study period. These included all material, machinery
and labor costs associated with production of planting
material (cost of seeds and nursery treatments), site
preparation (roto-tilling and deep-ripping, herbicide
application), plantation establishment (seed sowing),
and maintenance (manual weeding, herbicide and
formicide application). These data were used to cal-
culate the initial (®rst 2 years) plantation costs at each
site on a per-hectare basis.
Analyses of variance (ANOVA) were used to com-
pare seed germination, seedling survival and tree
growth data for each species among sites. Germination
percentage data were arcsin square-root transformed
prior to analysis (Mead and Curnow, 1983). Bonfer-
roni's least signi®cant difference (LSD) procedure was
used to separate the means of dependent variables that
were signi®cantly affected by site.
3. Results and discussion
3.1. Plantation costs
The establishment and maintenance costs during
the 2-year study period ranged from $747 (at Sites 2
and 3) to $912 (at Site 1) per hectare (Table 2).
Establishment costs constituted 63±68% of the total
costs; seeds ($182 ha
1
) and sowing activities
($117±164 ha
1
) were the most costly operations.
Establishment costs were approximately $150 higher
for Site 1 than for the other two sites, due to the need
to physically suppress the dominant Napier grass
(P. purpureum) at this site by roto-tilling prior to
herbicide application, and the higher costs of manual
seeding. Maintenancecosts (for manual weeding, herbi-
cide and formicide applications) were $137±155 ha
1
during the ®rst year (higher at Site 1 due to higher
manual weeding requirements) and $137 ha
1
for all
sites during the second year.
These costs compare favorably with those for
plantation establishment and maintenance at this site
using nursery-grown seedlings of native tree species in
other plantation treatments included in this project,
which averaged $1200±2500 ha
1
(unpublished data).
Table 2
Establishment and maintenance costs for direct seeding through
®rst 2 years of study
a
US$/ha
Site 1 Site 2 Site 3
Establishment costs
Roto-tilling 86 0 0
Subsoiling 31 31 31
Herbicide 20 20 20
Herbicide application 78 59 59
Formicide 21 26 21
Formicide application 9 9 9
Irrigation 29 29 29
Seeds 182 182 182
Manual seeding 164 117 117
Subtotal $620 $473 $468
Maintenance costs (year 1)
Manual weeding 53 35 35
Herbicide 37 37 37
Herbicide application 35 35 35
Formicide 21 21 21
Formicide application 9 9 9
Subtotal $155 $137 $137
Maintenance costs (year 2)
Manual weeding 35 35 35
Herbicide 37 37 37
Herbicide application 35 35 35
Formicide 21 21 21
Formicide application 9 9 9
Subtotal $137 $137 $137
Total costs $912 $747 $742
a
During the period of study, 1US$ Reis 1.70; manual labor
costs averaged $8.80 per day (8 h); costs for mechanical site
preparation and herbicide application averaged $16.65±$17.65 per
day; herbicide and formicide material costs averaged $5.30/l and
$5.30/kg, respectively.
174 V.L. Engel, J.A. Parrotta/ Forest Ecology and Management 152 (2001) 169±181
Establishment and early maintenance costs for com-
mercial Eucalyptus plantations average $700 ha
1
.
3.2. Germination rates and seedling survival
Seed germination and survival varied considerably
among species and, for one species (C. speciosa),
between sites (Table 3). Forty-®ve days after sowing,
average germination rates across all sites ranged
from 19±24% for Enterolobium contorstisiliquum
and S. parahyba to less than 1.3% for C. ¯oribundus
and M. scabrella; for these four species, there were
no signi®cant site effects (ANOVA, P<0:05).
Intermediate germination rates were observed for C.
speciosa (4.5±13.1%), and were signi®cantly higher at
Site 1 than at Site 2 or 3. These values were much
Table 3
Seed germination, seedling survival and early growth in direct sowing treatments
a
Site 1 Site 2 Site 3 All sites ANOVA
b
Percent seed germination
c
Chorisia speciosa 13.1 1.8a 5.1 1.9b 4.5 1.4b 7.6
Croton floribundus 000 0ns
Enterolobium contorstisiliquum 18.1 4.4 20.8 1.8 18.7 1.8 19.2 ns
Mimosa scabrella 3.3 1.7 0.5 0.5 0 1.3 ns
Schizolobium parahyba 18.8 5.3 21.6 2.3 30.6 13.0 23.7 ns
All species 10.6 1.5 9.6 1.0 10.8 2.9 10.3 ns
Stand density at 45 days (No./ha)
d
Chorisia speciosa 752 171 344 110 308 97 468 ns
Croton floribundus 0000ns
Enterolobium contorstisiliquum 924 153 1209 49 1009 127 1047 ns
Mimosa scabrella 268 135 39 39 0 102 ns
Schizolobium parahyba 564 150 540 42 611 256 572 ns
All species 2508 233 2132 200 1928 330 2189 ns
Stand density at 2 years (No./ha)
d
Chorisia speciosa 136 89 1 1 0 46 ns
Croton floribundus 20 20 0 5 58ns
Enterolobium contorstisiliquum 992 259 1291 289 613 200 965 ns
Mimosa scabrella 25 13 0 0 8 ns
Schizolobium parahyba 365 99 495 145 433 229 431 ns
All species 1538 395 1787 402 1051 174 1459 ns
Tree height (m)
Chorisia speciosa 0.4 0.3 0.1 0.1 0 0.2 ns
Croton floribundus 1.0 1.0 0 0.5 0.5 0.5 ns
Enterolobium contorstisiliquum 1.6 0.2a 2.2 0.2a 0.8 0.1b 1.5
Mimosa scabrella 1.3 1.0 0 0 0.4 ns
Schizolobium parahyba 2.0 0.4 1.4 0.02 1.8 0.4 1.7 ns
Stem basal diameter (cm)
Chorisia speciosa 1.2 0.8 0.4 0.4 0 0.5 ns
Croton floribundus 1.4 1.4 0 0.1 0.1 0.5 ns
Enterolobium contorstisiliquum 4.1 0.4ab 6.5 1.2a 1.8 0.3b 4.1
Mimosa scabrella 1.9 1.0 0 0 0.6 ns
Schizolobium parahyba 4.5 1.2 4.8 0.6 4.6 0.7 4.6 ns
a
Values represent means of replicate plot averages at each site standard error (n3); zeros indicate 100% mortality in all three replicate
plots at each site.
b
ANOVA results: ns Ð site effect not signi®cant;
P< 0.05, similar letters within a row indicate that means were similar (P< 0.05; LSD,
t-test).
c
Based on total number of seeds sown in each plot, 45 days after sowing.
d
Based on the number of planting spots with one or more surviving seedlings/trees at 45 days or 2 years after sowing.
V.L. Engel, J.A. Parrotta/ Forest Ecology and Management 152 (2001) 169±181 175
lower than those obtained in nursery trials, in which
germination percentages for Croton averated 35% up
to 90 days after sowing, and those of the other four
species exceeded 70%.
Despite the very low germination and 45-day sur-
vival rates for three of the ®ve species used in this
study, the resultant stand densities (numbers of plant-
ing spots per hectare with one or more surviving
seedings 45 days after sowing) was reasonably high
at all sites, ranging from a mean of 1928 330 (at Site
3) to 2508 233 (at Site 1). Subsequent mortality up
to 2 years after sowing was severe for Chorisia (82%
at Site 1 to 100% at Sites 2 and 3) and Mimosa (91±
100% across sites) but low for both Enterolobium
(averaging 7.8% for all sites) and Schizolobium
(24.7% for all sites). As a result, after 2 years stand
densities averaged 1538, 1787, and 1051 ha
1
, respec-
tively, for Sites 1, 2 and 3, and were composed almost
exclusively of Enterolobium (58±65% among sites)
and Schizolobium (24±41% among sites).
The low initial survival and subsequent mortality
rates for three of the ®ve species tested in this study
can be attributed to several factors. First, sub-optimal
seed quality may have been a problem, the seeds
having been collected from parent trees in a variety
of locations within the region that may or may not
have been similar to study site conditions. Further-
more, in the case of M. scabrella seeds, despite fairly
high germination percentages (>70%) obtained in
nursery tests, early seedling mortality in the nursery
was very high, indicating that these seeds were of poor
quality and probably more than 1 year old. A second
factor was insuf®cient rainfall during the days imme-
diately after sowing: having broken seed dormancy by
pre-treatment immediately prior to planting, this
resulted in desiccation and mortality of germinating
seeds. This effect was particularly marked in M.
scabrella. Given these problems, it is strongly recom-
mended that future plantation establishment efforts by
direct seeding pay close attention to seed quality, using
fresh seeds collected from several parent trees grow-
ing on sites where soils and climatic conditions are
broadly similar to those being reforested, and that
planting be carried out during periods of expected high
rainfall. A third factor that contributed to seedling
mortality in Chorisia, but probably not the other four
species, was competition with grasses; while the early
growth of this species is usually rapid, it is highly
sensitive to competition at the seedling stage, and
requires frequent weeding in its immediate vicinity
during the early growth phase.
3.3. Tree growth
During the ®rst 2 years after sowing, height and
stem diameter growth rates differed greatly among
species, with differences among species broadly par-
alleling those for seed germination and seedling sur-
vival (Table 3). For Chorisia,Croton, and Mimosa,
average heights (for the few surviving seedlings) at 2
years were 0.16, 0.50 and 0.43 cm, respectively; the
corresponding average stem basal diameters at this age
were 0.53, 0.50 and 0.63 cm. Although site differences
were not signi®cant for these growth parameters
(ANOVA, P<0:05), there was a trend towards higher
growth rates at Site 1 than at other sites for Chorisia
and Croton (survival was zero for Mimosa at Sites 2
and 3). Of the two more successful species, Enter-
olobium height and stem diameter growth averaged
1.5 m and 4.1 cm, respectively, across all sites at 2
years; its growth was signi®cantly greater at Site 2
(2.2 m, 6.5 cm) than at either Site 1 (1.6 m, 4.1 cm) or
Site 3 (0.8 m, 1.8 cm). Schizolobium growth rates
were similar across sites and slightly higher than those
for Enterolobium, averaging 1.4±2.0 m in height and
4.5±4.8 cm in stem diameter at 2 years. It appears that
both of these species, in addition to being well-suited
to establishment in grass-dominated habitats in this
region by direct seeding, are adapted to the broad
range of site conditions, especially soil physical and
chemical properties, included in this study (Table 1).
As discussed earlier, there are clearly problems asso-
ciated with the direct seeding techniques employed for
the remaining three species tested in this study which
remain to be resolved.
3.4. Natural regeneration
At the time of plantation establishment, the soil
seed bank at the three experimental sites was com-
prised exclusively of herbaceous species. These
included six grass species and 53 species representing
18 other families (Table 4). Species richness (total
number of species at each site) was greater at Site
1 (55 species) than at Site 2 (48 species) and Site 3
(31 species). The density of plants regenerating from
176 V.L. Engel, J.A. Parrotta/ Forest Ecology and Management 152 (2001) 169±181
Table 4
Composition of viable soil seed bank in experimental areas at time of plantation establishment
a
Family Scientific name Common name Sites where present
123
Acanthaceae Thunbergia alata Amarelinha x
Amaranthaceae Alternanthera brasiliana Sempre viva x x
Alternanthera tenella colla Apaga fogo x x x
Amaranthus hybridus Caruru crista de galo x x x
Amaranthus viridis Caruru verde x x x
Apiaceae Bowlesia incana Erva salsa x x
Comelinaceae Commelina virginica Trapoeraba x x x
Compositae Artemisia sp.Losna x x x
Bidens pillosa Pica
Äo preto x x x
Conyza bonariensis Rabo de foguete x x x
Eclipta alla Lanceta x
Emilia sonchifolia Pincel x x x
Galinsoga parviflora Bota
Äo de ouro x x
Senecio brasiliensis Berneira x x
Sonchus oleraceus Serralha x x x
Xanthium cavallinesii Carrapicho-bravo x x x
Convolvulaceae Ipomoea acuminata Campainha x
Ipomoea purpurea Campainha x
Ipomoea quamoclit Cipo
Âesqueleto x
Ipomoea grandifolia Campainha x
Cruciferae Raphanus sp.NabicËaxx
Cucurbitaceae Momordica charantia Mela
Äo S.Caetano x
Cyperaceae Cyperus esculentus Tiririca
Äoxx
Cyperus ferax Chufa x
Cyperus iria Tiririca x x x
Euphorbiaceae Chamaesyce hirta Erva S. Luzia x x
Euphorbia heterophylla Leiteria x x x
Euphobia prostrata Quebra pedra x x x
Phyllanthus corcovadensis Quebra pedra x x x
Phyllanthus tenellus Quebra pedra x x x
Ricinus communis Mamona x x x
Gramineae Brachiaria decumbens Braquia
Âria x x x
Brachiaria mutica Capim Angola x
Cenchrus echinatus Capim carrapicho x x x
Panicum maximum Colonia
Äoxxx
Rhynchelytrum repens Capim favorito x x x
Setaria geniculata Capim rabo de gato x x
Labiatae Leonotis nepetaefolia Corda
Äo de frade x
Leonurus sibiricus Rubim x x x
Leguminosae (Caesalpinioideae) Cassia patellaria Peninha x x
Indigofera hirsuta Anil x
Indigofera suffruticosa Anileira x x x
Senna obtusifolia Fedegoso x x
Senna occidentalis Fedegoso x x x
V.L. Engel, J.A. Parrotta/ Forest Ecology and Management 152 (2001) 169±181 177
the soil seed bank 60 days after post-emergent herbi-
cide application was 110 m
2
at Site 1, 75 m
2
at Site
2, and 50 m
2
at Site 3. Grasses comprised an average
of 83% of the regeneration (individuals m
2
) at Site 1,
52% at Site 2, and only 15% of the total at Site 3.
Approximately 2.5 years after planting, natural
regeneration of woody perennial species was observed
at all sites in both the direct-seeding and unplanted
control plots (treatments). This included species with
wind-dispersed seeds, as well as those whose seeds are
typically dispersed by birds and/or bats (Table 5). The
majority are shrubs and tree species very commonly
found in early successional forests in this region. The
number and composition of regenerating woody spe-
cies varied considerably among sites and between
treatments at each site. Inter-site differences can be
attributed to differences in the ¯oristic composition
of forest fragments in close proximity to the study
areas and other seed sources (remnant trees) located
within the study areas (such as guava Ð Psidium
guajava Ð present at Site 2). Within-site treatment
differences in woody species composition did not
follow any clear patterns with respect to modes of
seed dispersal; both direct-seeded and control plots
included similar proportions of woody species dis-
persed by wind and fauna.
Naturally regenerating woody species richness was
signi®cantly higher in the direct seeding treatment
than in the unplanted control areas (treatment means:
6.1 versus 2.8 species/plot; P<0:05, ANOVA).
Although the density of regenerating woody species
in the direct-seeded plots was generally much higher
than in the control plots (treatment means: 316 versus
167 ha
1
), these differences were not statistically
signi®cant. These data strongly suggest that while
natural regeneration of woody perennial species is
occurring in unplanted plots protected from ®re, graz-
ing and other disturbances that can preclude or slow
forest succession, the rates of woody species recolo-
nization (both species richness and density) are
enhanced in the (direct-seeded) plantations. This
may be due to increased seed rain through the provi-
sion of perches for frugivorous birds and bats, and/or
to improved understory microclimatic conditions and
decreased grass competition beneath the planted trees
that facilitate woody species seed germination and
seedling growth. Given the fact that wind-dispersed
species (which presumably arrive in both the direct-
seeded and control plots at similar rates) are more
abundant in terms of species richness and regeneration
density in the direct-seeded plots, it would seem that
at this stage of development at least, the increased
Table 4 (Continued )
Family Scientific name Common name Sites where present
123
Leguminosae (Papilionoideae) Crotalaria lanceolata Guizo de cobra x x x
Crotalaria pallida Guizo de cobra x x x
Malvaceae Malvastrum coromandelianum Guanxuma x x
Sida glaziovii Guanxuma branca x x
Sida urens Guanxuma dourada x x
Sidastrum micranthum Malva x x
Portulacaceae Portulaca oleracea Beldroega x x
Talinum patens Maria gorda x x x
Rubiaceae Diodia teres Mata pasto x x x
Solanaceae Solanum aculeatissimum Arrebenta cavalo x x x
Solanum americanum Maria pretinha x x x
Solanum sisymbriifolium Jua
Âxxx
Sterculiaceae Waltheria americana Malva veludo x x
Verbenaceae Latana camara Cambara
Âde espinho x
a
Based on surveys of regeneration in all experimental plots 60 days after site preparation.
178 V.L. Engel, J.A. Parrotta/ Forest Ecology and Management 152 (2001) 169±181
Tab le 5
Natural regeneration of tree and shrub species in direct seeding treatment and control plots
Species Family Local name Seed dispersal
vector
Number of individuals ha
1
Direct seeding treatment Control
Site 1 Site 2 Site 3 Site 1 Site 2 Site 3
Mangifera indica Anacardiaceae Manga Fauna 0 4.0
Schinus terebinthifolius Anacardiaceae Aroeira pimentiera Fauna 1.3 2.7
Peschiera fuchsiaefolia Apocynaceae Leiteira Fauna 28.0 29.3 6.7 4.0 18.7 1.3
Syagrus oleraceae Arecaceae Coqueiro guariroba Fauna 1.3
Tecoma stans Bignoniaceae Ipe
Ãde jardim Wind 2.7
Carica papaya Caricaceae Mama
Äo Fauna 1.3
Gochnatia polymorpha Compositae Cambara
Â-guacu Wind 9.3
Vernonia polyanthes Compositae Assa peixe Wind 1.3 46.7
Croton urucurana Euphorbiaceae Capixingui Fauna 41.3
Nectandra megapotamica Lauraceae Canelinha Fauna 1.3
Peltophorum dubium Leguminosae (C) Canafistula Wind 1.3 9.3
Enterolobium contorstisiliquum Leguminosae (M) Timboril Fauna 8.0
Piptadenia gonoacantha Leguminosae (M) Pau jacare
ÂWind 2.7
Bauhinia forficata Leguminosae (P) Bauhinia Wind 26.7 13.3
Centrolobium tomentosum Leguminosae (P) Arariba
ÂWind 5.3
Lonchocarpus muellbergianus Leguminosae (P) Embira de sapo Wind 90.7 12.0 1.3
Machaerium brasiliensis Leguminosae (P) Sapuva
Äo Wind 42.7
Platypodium elegans Leguminosae (P) Amendoim campo Wind 1.3 1.3 5.3
Cedrela fissilis Meliaceae Cedro Wind 1.3
Psidium guajava L. Myrtaceae Goiaba Fauna 4.0 332.0 30.7 2.7 349.3 4.0
Citrus sp.Rutaceae Lima
Äo Fauna 25.1 2.7
Solanum ciliatum Solanaceae Jua
ÂFauna 52.0
Solanum erianthum Solanaceae Fumo bravo Fauna 38.7 16.0 2.7 6.7
Solanum paniculatum Solanaceae Jurubeba Fauna 26.7
Aloysia virgata Verbenaceae Lixeira Wind 37.3 1.3 17.3 4.0 1.3
Lantana camara Verbenaceae Cambara
Âcolorido Fauna 2.7 48.0 16.0
Lantana lilacina Verbenaceae Cambara
Âroxo Fauna 2.7
Mean no. per ha S.E. (n3 plots/site) 245 124 485 159 217 136 81 69 405 117 13 7
Mean no. species per plot S.E. (n3) 6.7 2.8 5.0 0.6 6.7 3.3 3.0 1.0 3.7 1.2 1.7 0.7
V.L. Engel, J.A. Parrotta/ Forest Ecology and Management 152 (2001) 169±181 179
regeneration rates observed in the planted plots are
due mainly to the effects of the planted species on
understory microclimatic conditions and partial sup-
pression of competion by herbaceous species.
4. Conclusions
The results of this study show that direct seeding,
despite its risks, can be successfully used as an alter-
native to more expensive (though generally less risky)
methods of plantation establishment on grass-domi-
nated, degraded tropical landscapes. Despite good site
preparation, including initial suppression of compet-
ing herbaceous vegetation, timely planting, use of
prepared sowing lines, and post-planting care (weed-
ing), the poor performance of three of the ®ve species
used in this study highlighted the need for preliminary
testing of species adaptability to direct-sowing under
®eld conditions (C. speciosa and C. ¯oribundus were
clearly not suitable) and the need to use high-quality
seeds for all species (in this study, a de®ciency for M.
scabrella). Adequate germination and seedling survi-
val rates, as well as rapid growth in E. contorstisili-
quum and S. parahyba during the ®rst 2 years after
sowing yielded developing forest stands that were
adequately stocked (1000±1800 trees ha
1
) to facil-
itate natural regeneration by native woody shrub
species in their understories.
Acknowledgements
We wish to thank Elder C. Matos, Maurõ
Âcio Scor-
satto Sartori, Marcelo Zamboni Hildebrand, Ciro G.G.
Croce, Julio C.R. Ferreira, Renato D. Lopes, and
Cristina Silva Reis, Claudia Irene de Oliveira Rezende
and Roberta Cristina Zumba for their invaluable assis-
tance in the ®eld. We also acknowledge Roberto
Lyra Villas Bo
Ãas, Leonardo T. Bu
Èll, Helio Grassi
and Maria Helena Moraes of the Natural Resources
Department of UNESP/FCA for their help in soil
analyses (data presented in Table 1), and Luõ
Âs Alberto
B. Jorge for his kind assistance in the preparation of
the site map (Fig. 1). We thank A.E. Lugo and two
anonymous reviewers for their helpful comments on
an earlier draft of this paper. This work was conducted
in cooperation with the University of Puerto Rico and
supported in part by a grants from the World Bank
to the International Institute of Tropical Forestry
(Research Support Budget Grant RPO 680-05: `The
catalytic effect of tree plantings on the rehabilitation
of native forest biodiversity on degraded tropical
lands') through a cooperative research agreement
between IITF and The FundacËa
Äo de Estudos e Pes-
quisas Agrõ
Âcolas e Florestais (FEPAF, UNESP-FCA)
IITF-96-ICA-003. Additional student scholarship sup-
port to C.G.G. Croce, J.C.R. Ferreira was provided
by the Conselho Nacional de Pesquisas Cienti®cas e
Tecnolo
Âgicas (CNPq).
References
Applegate, G.B., Robson, K.J., Kent, G.A., 1993. Afforestation for
rehabilitation of degraded land in the tropics. In: Thwaites,
R.N., Schaumberg, B.J. (Eds.), Australian Forestry and the
Global Environment. The Institute of Foresters of Australia,
Canberra, pp. 146±153.
Barbosa, J.M., Barbosa, L.M., 1992. RecuperacËa
Äodea
Âreas
degradadas de Mata Ciliar a partir de sementes. Revista do
Instituto Florestal (Sa
Äo Paulo) 4, 702±705.
Brown, S., Lugo, A.E., 1994. Rehabilitation of tropical lands: a key
for sustaining development. Rest. Ecol. 2, 97±111.
Bryant, D., Nielsen, D., Tangley, L., 1997. The Last Frontier
Forests: Ecosystems and Economies on the Edge. World
Resources Institute, Washington DC, 42 pp.
Ca
Âmara de Gusma
Äo, I., 1991. Plan de acËa
Äo para a mata Atla
Ãntica.
FundacËa
Äo SOS Mata Atla
Ãntica, Sa
Äo Paulo, 152 pp.
de Camargo, D.A., Moniz, A.C., Jorge, J.A., Valadares, J.M.A.S.,
1986. Me
Âtodos de ana
Âlise quõ
Âmica, minerolo
Âgica e fõ
Ãsica de solos.
Instituto Agro
Ãnomico de Campinas, Campinas, Brazil, 94 pp.
Evans, J., 1982. Plantation Forestry in the Tropics. Clarendon
Press, Oxford, 472 pp.
Fernandes, T.S.D., Folster, H., Fassbender, W., Vielhauer, K., Voss,
K., Bianchi, H., 1998. Recuperation of degraded pasture using
Acacia mangium to return to traditional shifting cultivation
system in northeast of Para
Â, Brazil. In: Proceedings of the third
SHIFT Workshop, held in Manaus, Brazil, 15±19 March 1998.
Bundesministerium fu
Èr Bildung und Forschung, Germany,
pp. 119±124.
Francis, J.K., 1993. Leucaena leucocephala established by direct
seeding in prepared seed spots under dif®cult conditions.
Nitrogen Fixing Tree Res. Rep. 11, 91±93.
FundacËa
Äo SOS Mata Atla
Ãntica, 1992. Dossie
Ãmata atla
Ãntica.
FundacËa
Äo SOS Mata Atla
Ãntica, Sa
Äo Paulo, 119 pp.
FundacËa
Äo SOS Mata Atla
Ãntica and Instituto Nacional de Pesquisas
Espaciais (INPE), 1993. Atlas da evolucËa
Äo dos remnascentes
¯orestais e ecossistemas associados do domõ
Ânio da Mata
Atla
Ãntica no perõ
Âodo de 1985 a 1990. Sa
Äo Paulo, 22 pp.
Kageyama, P., Santarelli, E., Gandara, F.B., GoncËalves, J.C.,
Simionato, J.L., Antiqueira, L.R., Geres, W.L., 1994. Revege-
tacËa
Äodea
Âreas degradadas: modelos de consorciacËa
Äo com alta
180 V.L. Engel, J.A. Parrotta/ Forest Ecology and Management 152 (2001) 169±181
diversidade. In: Anais do 18Simpo
Âsio Sul-Americano & 28
Simpo
Âsio Nacional de RecuperacËa
ÄodeA
Âreas Degradadas held
at Foz do IguacËu. FundacËa
Äo de Pesquisas Florestais do Parana
Â,
Curitiba, Brazil, pp. 569±576.
Knowles, O.H., Parrotta, J.A., 1995. Amazonian forest restoration:
an innovative system for native species selection based on
phenological data and performance indices. Comm. For. Rev.
74, 230±243.
Kuusipalo, J., A
Êdjers, G., Jafarsidik, Y., Antii, O., Tuomela, K.,
Vuokko, R., 1995. Restoration of natural vegetation in degraded
Imperata cylindrica grassland: understory development in
forest plantations. J. Veg. Sci. 6, 205±210.
Lamb, D., 1998. Large-scale ecological restoration of degraded
tropical forest lands: the potential role of timber plantations.
Rest. Ecol. 6, 271±279.
Lamb, D., Tomlinson, M., 1994. Forest rehabilitation in the Asia-
Paci®c region: past lessons and present uncertainties. J. Trop.
For. Sci., 157±170.
Lamb, D., Parrotta, J.A., Keenan, R., Tucker, N.I.J., 1997.
Rejoining habitat remnants: restoration of degraded tropical
landscapes. In: Laurence, W.F., Bierregaard Jr., R.O. (Eds.),
Tropical Forest Remnants: Ecology, Management and Con-
servation of Fragmented Communities. University of Chicago
Press, Chicago, IL, pp. 366±385.
Laurie, M.V., 1974. Tree planting practices in African savanas.
Forestry Development Paper No. 19, FAO.
Lorenzi, H., 1992. A
Ârvores brasileiras: manual de indenti®cacËa
Äoe
cultivo de plantas arbo
Âreas nativas do Brasil. Nova Odessa, SP:
editora Plantarum, 352 pp.
Lu
Èbbe, W.A., Geldenhuys, C.J., 1991. Regeneration patterns in
planted and natural forest stands near Knysa, Southern Cape. S.
Afr. J. For. 159, 43±50.
Lugo, A.E., 1992. Tree plantations for rehabilitating damaged lands
in the tropics. In: Wali, M.K. (Ed.), Environmental Rehabilita-
tion, Vol. 2. SPB Academic Publishing, The Hague, pp.
247±255.
Lugo, A.E., Parrotta, J.A., Brown, S., 1993. Loss of species caused
by tropical deforestation and their recovery through manage-
ment. Ambio. 22, 106±109.
Maschio, L.M. de A., Balensiefer, M., Rachwal, M.F.G., Curcio,
G., Montoya, L.J., 1992. EvolucËa
Äo, esta
Âgio e caracterizacËa
Äode
pesquisa em recuperacËa
Äodea
Âreas degradadas no Brasil. In:
Anais do 18Simpo
Âsio Nacional de RecuperacËa
ÄodeA
Âreas
Degradadas, held at Curitiba, Universidade Federal do Parana
Â/
FundacËa
Äo de Pesquisas Florestais do Parana
Â, Curitiba, Brazil,
pp. 17±33.
Mead, R., Curnow, R.N., 1983. Statistical Methods in Agricultural
and Experimental Biology. Chapman and Hall, New York.
335 pp.
Mitra, S.S., Sheldon, F.H., 1993. Use of exotic tree plantations by
Bornean lowland forest birds. Auk 110, 529±540.
Parrotta, J.A., 1992. The role of plantation forests in rehabilitating
degraded ecosystems. Agric. Ecosyst. Environ. 41, 115±133.
Parrotta, J.A., 1993. Secondary forest regeneration on degraded
tropical lands: the role of plantations as `foster ecosystems'. In:
Lieth, H., Lohmann, M. (Eds.), Restoration of Tropical Forest
Ecosystems. Kluwer, Dordrecht, pp. 63±73.
Parrotta, J.A., 1995. The in¯uence of overstory composition on
understory colonization by native species in plantations on a
degraded tropical site. J. Veg. Sci. 6, 627±636.
Parrotta, J.A., 1999. Productivity, nutrient cycling, and succession
in single- and mixed-species plantations of Casuarina equise-
tifolia,Eucalyptus robusta, and Leucaena leucocephala in
Puerto Rico. For. Ecol. Manage. 124, 45±77.
Parrotta, J.A., Knowles, O.H., 1999. Restoration of tropical moist
forest on bauxite mined lands in the Brazilian Amazon. Rest.
Ecol. 7, 103±116.
Parrotta, J.A., Turnbull, J.T. (Eds.), 1997. Catalyzing native forest
regeneration on degraded tropical lands. For. Ecol. Manage. 99,
1±290.
Raij, B., van Quaggio, J.A., 1983. Me
Âtodos de ana
Âlise de solo para
®ns de fertilidade. Boletim Te
Âcnico 81. Instituto Agro
Ãnomico,
Secretaria de Agricultura e Abastecimento, Campinas, Brazil,
31 pp.
Soni, P., Vasistha, H.B., Kumar, O., 1989. Biological diversity in
surface mined areas after reclamation. Ind. For. 115, 475±482.
Sun, D., Dickinson, G.R., 1995. Direct seeding for rehabilitation of
degraded lands in north-east Queensland. Aust. J. Soil Water
Cons. 8, 14±17.
Sun, D., Dickinson, G.R., 1996. The competition effect of
Brachiaria decumbens on the early growth of direct-seeded
trees of Alphitonia petriei in tropical north Australia.
Biotropica 28, 272±276.
Sun, D., Dickinson, G.R., Bragg, A.L., 1995. Direct seeding of
Alphitonia petriei (Rhamnaceae) for gully revegetation on
tropical northern Australia. For. Ecol. Manage. 73, 249±257.
Thomson, D., 1992. An ef®cient broad acre tree establishment
technique. Landscape Aust. 4, 313±319.
V.L. Engel, J.A. Parrotta/ Forest Ecology and Management 152 (2001) 169±181 181
