Mechanized Direct-Seeding of Native Forests in Xingu, Central Brazil

Abstract

The Y Ikatu Xingu Campaign brought together indigenous people, farmers, researchers, governmental, and non-governmental organizations seeking riparian forest restoration in the Xingu watershed, in west-central Brazil. Forest restoration is challenging in the region because of scarce nurseries, long distances, and high costs associated with the usual technique of planting nursery-raised seedlings. This article describes mechanized direct seeding and compares it with the planting of seedlings, in terms of cost and tree densities at ages of 0.5 until 5.5 yr after planting. Direct-seeding was mechanized using common agricultural machines designed for sowing cereals or grasses, which were loaded with 200,000 seeds of native trees and 150,000 seeds of annual and sub-perennial legumes, plus 50–150 kg sand ha−1. The Campaign restored more than 900 ha by direct-seeding and 300 ha by planting seedlings. The great demand for native seeds was met by the Xingu Seed Network, formed by Indians, small landholders, and peasants, which commercialized 98 tons of native seeds and earned US$500,000 since 2006. Direct-seeding costs less per hectare than planting seedlings (US$1,845 ha−1 against US$5,106 ha−1), results in higher tree densities (2,500–32,250 trees ha−1 against 1,500–1,650 trees ha−1), is more practical, and creates layers of dense vegetation that better resembles natural forest succession.

Full text

This article was downloaded by: [Eduardo M. Campos-Filho]
On: 07 March 2014, At: 11:25
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of Sustainable Forestry
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/wjsf20
Mechanized Direct-Seeding of Native
Forests in Xingu, Central Brazil
Eduardo M. Campos-Filho a , José N. M. N. Da Costa a , Osvaldo L. De
Sousa b & Rodrigo G. P. Junqueira a
a ISA , São Paulo , SP , Brazil
b Cooperafloresta , São Paulo , SP , Brazil
Accepted author version posted online: 27 Jun 2013.Published
online: 09 Sep 2013.
To cite this article: Eduardo M. Campos-Filho , José N. M. N. Da Costa , Osvaldo L. De Sousa &
Rodrigo G. P. Junqueira (2013) Mechanized Direct-Seeding of Native Forests in Xingu, Central Brazil,
Journal of Sustainable Forestry, 32:7, 702-727, DOI: 10.1080/10549811.2013.817341
To link to this article: http://dx.doi.org/10.1080/10549811.2013.817341
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions

Journal of Sustainable Forestry, 32:702–727, 2013
Copyright ©Taylor & Francis Group, LLC
ISSN: 1054-9811 print/1540-756X online
DOI: 10.1080/10549811.2013.817341
Mechanized Direct-Seeding of Native Forests
in Xingu, Central Brazil
EDUARDO M. CAMPOS-FILHO1,JOSÉN.M.N.DACOSTA
1,
OSVALDO L. DE SOUSA2, and RODRIGO G. P. JUNQUEIRA1
1ISA, São Paulo, SP, Brazil
2Cooperafloresta, São Paulo, SP, Brazil
The Y Ikatu Xingu Campaign brought together indigenous peo-
ple, farmers, researchers, governmental, and non-governmental
organizations seeking riparian forest restoration in the Xingu
watershed, in west-central Brazil. Forest restoration is challeng-
ing in the region because of scarce nurseries, long distances,
and high costs associated with the usual technique of planting
nursery-raised seedlings. This article describes mechanized direct
seeding and compares it with the planting of seedlings, in terms
of cost and tree densities at ages of 0.5 until 5.5 yr after plant-
ing. Direct-seeding was mechanized using common agricultural
machines designed for sowing cereals or grasses, which were
loaded with 200,000 seeds of native trees and 150,000 seeds of
annual and sub-perennial legumes, plus 50–150 kg sand ha−1.
The Campaign restored more than 900 ha by direct-seeding and
300 ha by planting seedlings. The great demand for native seeds
was met by the Xingu Seed Network, formed by Indians, small land-
holders, and peasants, which commercialized 98 tons of native
seeds and earned US$500,000 since 2006. Direct-seeding costs
less per hectare than planting seedlings (US$1,845 ha−1against
US$5,106 ha−1), results in higher tree densities (2,500–32,250 trees
ha−1against 1,500–1,650 trees ha−1), is more practical, and
The authors acknowledge the active contribution of seed-gatherers, farmers, and field
technicians to this research. They also recognize the discussions with Rodrigo Junqueira (Farm
São Luiz), Natalia Guerin, Giselda Durigan, Ernst Gostch, Antônio Melo, and others who
visited the authors since 2006.
Address correspondence to Eduardo M. Campos-Filho, Instituto Socioambiental, Rua
Miragaia, 565, Butantan CEP: 05511-020, São Paulo-SP, Brazil. E-mail: edumalta@hotmail.com
702
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

Mechanized Direct-Seeding of Native Forests 703
creates layers of dense vegetation that better resembles natural
forest succession.
KEYWORDS forest restoration, tree mechanized direct-seeding,
watershed, biodiversity management
INTRODUCTION/BACKGROUND
The Xingu River Watershed and the “Y Ikatu Xingu” Campaign
The Amazon forest is the largest continuous tropical forest ecosystem in the
world. While most of it is still intact, it has been progressively deforested
in the south and east of the Amazonian Basin. The Xingu River—one of
the Amazon’s main tributaries—is located in this southeastern portion of the
Amazon Basin, where the Cerrado and Amazon biomes meet. The Cerrado is
a mainly evergreen vegetation occurring typically on oligotrophic soils with,
at least, four dry months yearly, in which it differs from Amazonia, where dry
seasons usually last less than four months. With contorted tree trunks and
thick leaves, Cerrado is one of the world’s biodiversity hotspots and has the
richest flora among the world’s savannas (over 7,000 species), with high lev-
els of endemism. The Xingu Basin covers 51 million ha with extensive water
resources, biodiversity and socio-diversity and is a site of intense agricultural
expansion, especially in the headwaters region of Mato Grosso, where many
of the remaining forests are situated on lands that are suitable for soy pro-
duction and cattle ranching (Lima et al., 2006). While 24 indigenous groups
and dozens of traditional riverine communities have conserved most of the
native vegetation of Xingu in their territories, settlers that arrived in the last
40 yr have deforested large areas of the surrounding landscape (Sanches &
Villas-Bôas, 2005), including riparian zones, which is specifically forbidden
by the Brazilian Forest Code (Velasquez, Queiroz, & Bernasconi, 2010). The
deforestation of 5.7 million ha of native vegetation, including 315,000 ha in
riparian zones (Figure 1), is jeopardizing water quality and water flow reg-
ulation (Coe, Costa, & Howard, 2007), as well as the health, economy, and
culture of people who, for over a thousand years, have used water from
these rivers for drinking, cleaning, cooking, and fishing (Velasquez et al.,
2010).
Concerned with the situation in the watershed, the Socio-Environmental
Institute (ISA, in Portuguese) integrated socioeconomic and environmental
data into maps of the region. ISA worked with several other organiza-
tions to bring farmers, peasants, Indians, NGOs, and governmental actors
together to discuss these issues in 2004 in the city of Canarana (Sanches
& Villas-Bôas, 2005). These partner organizations included the Association
of the Xingu Indigenous Territory (ATIX), the Forum of Mato Grosso for
Environment and Development (FORMAD), and the State University of
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

704 E. M. Campos-Filho et al.
FIGURE 1 Map of the Xingu headwaters–deforestation up to 2007. The central non-deforested
area is the Xingu Indigenous Park and the surrounding darker patches are deforested areas.
More specifically, the map indicates indigenous villages (dark dots); cities (dotted circles);
municipal limits (thin lines); limits of the river basin (dark, thick outer lines); indigenous
territories (lighter thick lines); conservation areas (bordered shaded areas labeled, e.g., “PES
do Cristalino II”); deforestation up to 2007 (dark gray areas); altered forests up to 2007,
which contain vegetation presenting noticeable alteration in its original composition through
exploration of timber or fire, but which was not cleared (lighter gray patches); and original
vegetation up to 2007, which are areas that haven’t suffered noticeable anthropic alteration
(gray background) (color figure available online).
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

Mechanized Direct-Seeding of Native Forests 705
Mato Grosso (UNEMAT), the Xingu Headwater’s Meeting. The discussions
led to the creation of “Y Ikatu Xingu” (YIX—“Save the Fine Water of
Xingu,” in the Kamaiura language): a “shared socio-environmental responsi-
bility” campaign aiming to contain and reverse degradation processes in the
Xingu Basin. Each stakeholder recognized its own agenda for promoting
sustainable agriculture, sanitation, education, interaction with public pol-
icy, and communication (ISA, 2013). Since water conservation and forest
restoration was a common element across all of the stakeholders’ respective
agendas, it became a main focus of the YIX Campaign. The YIX Campaign
has been a learning network for forest restoration, wherein findings are
disseminated through lectures, workshops, field demonstrations, practical
courses, videos, television, magazines, newspapers, interchange expeditions,
and school activities.
Participative Planning and Innovation
The Y Ikatu Xingu Campaign has helped people to plant native trees on
their lands, whether they be private farms, indigenous territories, or agrar-
ian reform settlements. At the same time, YIX has received help from
these people, since they are also acknowledged as researchers in the
Participatory Action Research approach (Castellanet & Jordan, 2002) as well
as entrepreneurs. Innovative models are welcome as they can benefit from
local knowledge and YIX support in order to develop new strategies. YIX
technicians facilitate diagnosis and planning by addressing important plan-
ning details—such as intended goals, site degradation factors and risks, soil
type, vegetation cover, available tools, and available workforce—resulting
in projects that suit people’s different understandings, resources, and goals.
This approach attempts to rely minimally on external technical assistance
and inputs. YIX seeks to empower local entrepreneurship through restora-
tion aimed at goals such as water quality, fruit production, timber production,
carbon sequestration and forest restoration.
Techniques for Landscape-Scale Approach
Autogenic restoration is the preferred restoration process wherever natural
regeneration potential is still available (Rodrigues et al., 2011). However,
many degraded areas lack this potential, or present physical or biological
barriers that delay it significantly or prevent it from developing into suc-
cessional forests (Lugo, 1988; Nepstad, Uhl, & Serrao, 1991)—such as soil
compaction, distance to forest remnants, long dry seasons, competition with
invasive grasses and herbivory. In this scenario, the most usual technique
for forest restoration in the tropics has been the plantation of nursery-raised
tree seedlings (Lamb, Erskine, & Parrotta, 2005; Chazdon, 2008), a method
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

706 E. M. Campos-Filho et al.
that evolved through decades of research and eventually established a very
popular model in Brazil. Presently, it is recommended and conducted by
government agencies, legal agreements, and NGOs in Brazil as a dominant
formula regarding various aspects of forest restoration. These prescriptions
include seedling spacing (3 m ×2 m), density (1,666 seedlings ha−1), and
proportion (50%) between pioneer or “filling” group (fast-growing and wide
canopy) and non-pioneer or “diversity” group (slow-growing and/or nar-
row canopy) tree species (Rodrigues et al., 2011). Planting seedlings may be
a successful choice for forest restoration (Montagnini, 2001; Holl, Zahawi,
Cole, Ostertag, & Cordell, 2010); however, it was found in other regions of
Brazil that planting can be extremely costly (circa US$5,000 ha−1, according
to Aronson et al., 2011) and usually results in low-diversity stands (aver-
age of 35 species) dominated by pioneer trees (2/3 of planted individuals)
with short life-cycles (10–20 yr), which raises concerns about the long-term
diversity of these planted forests (Barbosa et al., 2003).
The main impediments to planting seedlings in Xingu were the scarce
seedling production, poor dirt roads for transportation, high costs associated
with intensive manual labor and long-standing control of invasive grasses,
high mortality associated with long dry seasons, and unmotivated farmers
(Durigan, Guerin, & Costa, 2013). Additionally, the technology for raising
seedlings of regionally native species in forest nurseries is scarce, especially
for Cerrado species, probably due to its early root depth, which might be an
adaptation for long dry seasons (Oliveira et al., 2005). Thus, for a region such
as Xingu, where distances between locales are long, and farms and degraded
areas are extensive, natural regeneration and seedling transplantation tech-
niques were ill-suited to a landscape-scale forest restoration initiative.
Indians and traditional riverine communities of Xingu have, for cen-
turies, planted trees through the ancient technique of direct-seeding (Posey,
1985). YIX combined the direct-seeding technique with the use of common
farm machines such as soybean planter machines, as seen at the São Luis
Farm in southeastern Brazil. After the soy planter got stuck in the mud dur-
ing the first trial in Xingu, the alternative found by the farmers was to throw
seeds manually and then plow over them. Soon afterwards, broadcasting
machines proved also to be a viable option for direct seeding.
Tree direct seeding is an ancient technique, with numerous references
worldwide (Evelyn, 1670; Harmer & Kerr, 1995; Willoughby, Jinks, Gosling, &
Kerr, 2004; Doust, Erskine, & Lamb, 2006, 2008; Cole, Holl, Keene, & Zahawi,
2010) and in Brazil (Franco, Nardoto, & Souza, 1996; Engel & Parrotta, 2001;
Camargo, Ferraz, & Imakawa, 2002; Araki, 2005; Isernhagen, 2010; Santos,
Ferreira, Aragão, Amaral, & Oliveira, 2012). These studies have found that
some species are incompatible with this type of planting, some seeds must
be buried, germination success and plant survival are directly related to seed
size, and that it is cost-effective relative to planting seedlings and, thus, a
promising alternative for forest restoration in the tropics.
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

Mechanized Direct-Seeding of Native Forests 707
This article aims to present a technique for mechanized direct seeding
and compare it to the forest restoration technique most used in Brazil, which
is planting seedlings that are pre-grown in forest nurseries. The hypothesis
is that direct-seeding represents a cheaper, more practical, effective, and
socio-beneficial technique for large-scale forest restoration.
The following presents the primary ecological aspects of the study area
and describes the techniques used for establishing seedlings and direct-
seeding plantations in Xingu, as well as the management and monitoring
methodologies. Structural data of sapling density, diameter, and height sam-
pled at 26 areas are compared and discussed with reference to the literature
and lead to some concluding remarks.
METHODOLOGY
Study Area
This study analyzed 26 restoration areas randomly selected from 198 farms
across the Xingu River Basin headwaters in the state of Mato Grosso. The
Xingu River begins in the Cerrado of Mato Grosso State and flows north
2,700 km through tall, dense, moist forests before emptying into the main
channel of the Amazon River (Figure 2). The Xingu River drains a landscape
of ancient crystalline Precambrian shield, from 600 to 300 m above sea level,
covered predominantly by oxisols. The climate type, according to Köppen
(1948), is tropical AW with two defined seasons: “wet summer” (October to
March) and “dry winter” (April to September). Annual rainfall in the south
of the basin is around 1,400 mm and highly concentrated, with a 6-month
period of severe drought; whereas to the north of the basin in Mato Grosso,
it rains up to 1,900 mm annually, with only a 4-month drought (Ivanauskas,
Monteiro, & Rodrigues, 2008).
PLANTING SEEDLINGS
Seeds of 47 species were obtained from the Xingu Seed Network, pretreated
for breaking dormancy as necessary, and sown in germination beds made
of washed sand in a local forest nursery. After germination, seedlings were
transplanted into plastic bags of 25 cm ×18 cm filled with 40% sand, 40%
clay, and 20% organic matter, plus 50 g of limestone and 25 g of fertilizer NPK
(10-10-10) and raised 4 to 6 months. Restoration site preparation included
spraying a broad spectrum contact herbicide, such as glyphosate, followed
by plowing two times to break soil compaction. Seedlings were planted
in holes of 30 cm ×30 cm ×40 cm, spaced 3 m ×2 m from each other,
resulting in a density of 1,666 seedlings ha−1, with a 50% proportion between
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

708 E. M. Campos-Filho et al.
FIGURE 2 Location of the Xingu River Basin (thick line) in South America, Brazil and its
headwaters in the Mato Grosso State (MT). Thin lines mark the limits of the Brazilian States
(color figure available online).
pioneer (fast-growing with dense shading canopy) for “filling,” and non-
pioneer (slow-growing, shade tolerant) tree species for creating “diversity.”
At farm 2R, four jack beans and three pigeon peas were established around
each seedling in a crown with a radius of 50 cm.
Species planted as seedlings for “filling” were Anadenanthera colub-
rina (Vell.) Brenan, Cecropia pachystachya Trécul., Anacardium nanum
A. St.-Hil., Enterolobium timbouva Mart., Guazuma ulmifolia Lam., Inga
edulis Mart., Inga vera Willd., Mabea fistulifera Mart., Simarouba amara
Aubl., Simarouba versicolor A. St. Hil., and Terminalia argentea Mart.
& Zucc. Seedling species for “diversity’ were Alibertia edulis Rich.,
Andira vermifuga Mart. ex Benth., Annona coriacea Mart., Apuleia leio-
carpa (Vogel) J.F. Macbr., Astronium fraxinifolium Schott ex Spreng.,
Bauhinia sp. L., Buchenavia tomentosa Eichler, Byrsonima cydoniifolia
A. Juss., Calophyllum brasiliense Cambess, Caryocar brasiliense Cambess.,
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

Mechanized Direct-Seeding of Native Forests 709
Copaifera langsdorffii Desf., Cybistax antisyphilitica (Mart.) Mart., Dipteryx
alata Vogel Dypterix odorata (Aubl.) Willd., Eugenia dysenterica DC.,
Eugenia sp. L., Genipa americana L., Handroanthus heptaphyllus (Vell.)
Mattos, Handroanthus ochraceus (Cham.) Mattos, Handroanthus serrati-
folius (A.H.Gentry) S.Grose, Hymenaea courbaryl L., Hymenaea stigono-
carpa Mart. ex Hayne, Jacaranda cuspidifolia Mart. ex A.DC., Lafoensia
pacari A. St. Hil., Machaerium acutifolium Vogel , Mauritia flexuosa L.f.,
Mouriri pusa Gardner, Myracrodruon urundeuva Allemão, Peltogyne confer-
tiflora (Mart. ex Hayne) Benth, Physocalymma scaberrimum Pohl, Protium
heptaphyllum (Aubl.) Marchand., Spondias mombin L., Sterculia striata A. St.
Hil. & Naud. Syagrus oleracea (Mart.) Becc., Tabebuia aurea (Silva Manso)
Benth. & Hook.f. ex S.Moore, Tabebuia roseoalba (Ridl.) Sand.
“MUVUCA” DIRECT-SEEDING
The mix of sand or soil with seeds of crop, green manure (annual and sub-
perennial legumes) and forest species is known as “muvuca de sementes”
(“seed muvuca”). “Muvuca” is an expression made popular in Brazil by the
agroforestry group Mutirão Agroflorestal.Themuvuca direct-seeding tech-
nique was conducted either using planter machines, which mechanically dig
and sow in rows, or broadcasting machines, that launch seeds randomly over
the area, and require plowing in order to bury the seeds afterwards. All aban-
doned croplands were direct-seeded in rows with planter machines, while
all abandoned pastures were direct-seeded with broadcasting machines and
later plowed. Since the restoration areas were not previously designed as
scientific experiments, it was not possible to establish adequate control plots
for these two ways of direct seeding. Therefore, they were counted together
as one methodology, which is compared with the conventional methodology
of planting seedlings.
Soil Preparation
Adequate soil preparation may be accomplished by plowing, chemical weed-
ing with herbicides, or a combination of both. Use of herbicides is only
justifiable if aggressive grasses are dominant in the target area. Pastures are
the most common scenario for restoration projects in Xingu and are usually
formed by African grasses such as Urochloa P. Beauv. and Panicum L. Thus,
site preparation on pastures usually begins with an application of a broad
spectrum contact herbicide, such as glyphosate. After 2 weeks, the soil was
plowed for breaking soil compaction. At least a month after first plowing,
another plowing was conducted for killing the remaining grasses and lev-
eling the terrain. Where slopes are steep, plowing may be done in leveled
strips, leaving non-plowed strips between them in order to avoid erosion.
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

710 E. M. Campos-Filho et al.
Where slopes are too steep, the muvuca may be sown in small seed beds
made by hand. However, no steep areas were sampled for this research. The
second most common scenario in Xingu are croplands, where direct-seeding
machines are usually available, soil is flat, covered with “mulch” and need
not be plowed, since soil compaction is low and weeds are nearly absent.
In this scenario, no soil preparation is necessary.
Species Selection
Species for direct-seeding were selected for each site from a list of
214 species collected by the Xingu Seed Network. Species of seedlings
were selected from a list of 47 species available in local nurseries. The
first criteria guiding species selection is tolerance to water stress such as
extended dry seasons, annual flooding, or permanently swampy areas.
Species were grouped by occurrence in the following vegetation types:
Amazon Forest, Cerrado Forest, Cerrado, Gallery (riparian) Forest, Temporary
Flooded Forest, Permanently Swampy Forest, and Swampy Field. Second, the
planting machine available was tested for seed size, because the size of a
machine’s apertures limits the maximum seed size, thus it may be necessary
to exclude some species or to plant them separately. However, an average
of 70 species were direct-seeded in each area, varying the group of species
according to the factors described above.
Muvuca—Seed Quantity Calculation
Seed calculation aims at establishing a seed bank capable of out-competing
invasive grasses and creating a canopy in which fast-growing plants get grad-
ually replaced by slow-growing trees over time. Within the adequate group
of species for the target site, species with economic value—such as edi-
ble fruits, medicines, and timber—were favored with higher seed densities.
Nevertheless, each muvuca included species from the following groups: (a)
leguminous herbs and shrubs (10 seeds m−2) capable of recovering soil fer-
tility and shading out invasive grasses, (b) fast-growing trees (10 seeds m−2),
and (c) slow-growing trees (10 seeds m−2). Total seeding density averaged
300,000 seeds ha−1; 200,000 seeds ha−1of trees, plus 100,000 seeds ha−1of
annual and sub-perennial legumes; 3 seeds m−2of jack-beans (Canavalia
ensiformis [L.] DC.), 6 seeds m−2of Crotalaria spectabilis Roth, and 1 seed
m−2of pigeon peas (Cajanus cajan [L.] Millsp.).
Seed quantity of each species should be calculated by dividing the
desired quantity of adults by the rate of viable seeds per kilogram and their
survival rates. Since there were no published studies about the rates of open
field seed germination and tree survival in Xingu, calculation of seed quanti-
ties was based on local knowledge, resulting in a rough estimate of between
2 and 10% germination plus survival rate after 3 yr. Seeding density variations
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

Mechanized Direct-Seeding of Native Forests 711
rely on the assumption that larger seeds present greater germination and sur-
vival rates. Between 10 and 50 seeds were sown for each desired adult tree,
and around 3 seeds for each desired annual or sub-perennial adult legume.
Seed Collection
Seeds were ordered through the Xingu Seed Network (XSN), an organiza-
tion of indigenous people and peasants for seed collection (Figure 3). XSN
organizes seed quality, storage, delivery, and payments to 20 groups across a
range of 750 km, providing seeds from 214 species. Annual meetings gather
leaders from all groups to discuss organization agreements and prices, which
vary from US$0.5 kg−1for big and easy seeds to US$125 kg−1for small and
hard to get seeds, with an average of US$15 kg−1. The XSN also provides
information on species ecology and do research on seed collection, cleaning
and storage (http://www.sementesdoxingu.org.br).
Muvuca—Seed Mixture Preparation
Before mixing the seeds with sand, all dormant seeds—due to seed
coat impermeability—were treated with hot water (60◦C) for 5 min
FIGURE 3 Women collecting Amazon forest seeds at Panará Indigenous Territory, Guarantã
do Norte, MT, Brazil. Seeds are bought by neighbor farmers through the Xingu Seed Network
and direct-seeded in degraded riparian areas of their farms in the Xingu Basin (Credit: Dannyel
de Sá) (color figure available online).
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

712 E. M. Campos-Filho et al.
and then immersed in cold water for 10 min. Pre-treatment was
performed on all seeds from tree species of the leguminous fam-
ily within the genera Apuleia,Bauhinia,Dialium,Dimorphandra,
Enterolobium,Hymenaea,Leptolobium,Ormosia,Parkia,Peltogyne,
Samanea,Tachigali,Schizolobium,Senegalia,Senna,andStryphnodendron;
but also of Passiflora (Passifloraceae), Byrsonima (Malpighiaceae) Cecropia
(Cecropiaceae), and Solanum (Solanaceae). Following pre-treatment, sand
was mixed with the seeds in order to favor a more homogeneous distribu-
tion of seeds with different sizes over the restoration site, resulting in the
muvuca. More than one muvuca may be necessary, depending on the type
of machine available. Machines for broadcasting limestone have the largest
apertures and can launch seeds up to 15 cm in diameter. Smaller machines
for broadcasting fertilizers and grass seeds have apertures that can sow seeds
up to 5 cm. Planters designed for sowing soy and corn have the smallest
apertures: 1 cm in the seed boxes and 2 cm in the fertilizer boxes (Figure 4).
Direct-Seeding
Seeds which were larger than the machine’s apertures, as well as expen-
sive and potentially economically valuable tree species, such as Caryocar
brasiliense Cambess., were separated and sown manually in regularly spaced
seedbeds. Large seeds, over 2 cm, were buried in the ground at a depth
between 1 and 10 cm. Medium-size seeds, from 0.2 to 2 cm, were buried
in the ground between 1 and 2 cm, but never deeper than 5 cm. They
represent the majority of seeds available in Xingu and fit in several types
of planting and broadcasting machines. Broadcasting machines designed for
limestone, fertilizers, or grass seeds are used for direct seeding over plowed
land (Figure 5). After broadcast sowing, the soil was superficially plowed
(leveled) to cover the seeds with soil. Planter machines designed for grains
were regulated to plant at a depth of 2–5 cm (Figure 6) and require no
plowing afterward. Seeds that are very small, lightweight, photoblastic, and
winged—such as Maclura tinctoria (L.) D.Don ex Steud., Astronium frax-
inifolium Schott. ex Spreng., and Aspidosperma spp. Mart. & Zucc.—were
separated in another muvuca and broadcast sown as the last operation in
the area, so as to remain aboveground.
MONITORING
Planted areas were rapidly monitored by YIX technicians and landowners or
employees walking together and estimating the percentage cover of invasive
grasses, tree densities, and listing the species growing in the area. From
those data, only the list of species will be presented here, due to a lack of
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

Mechanized Direct-Seeding of Native Forests 713
FIGURE 4 Preparing muvuca, a mixture of seeds of native trees, fast-growing legumes, and
sand for direct-seeding (Credit: Luciana Akemi Deluci) (color figure available online).
systematic sampling for density and coverage estimations. Selection of sites
for systematic sampling sought to include at least three sample areas of each
age class (Table 1) and of each planting technique (direct-seeding in rows,
broadcast direct-seeding, and planting seedlings) in each biome (Cerrado
and Amazonia). However, the 5.5-yr-old areas were all in the Cerrado and
there were no monitored areas planted with seedlings in Amazonia biome.
Systematic sampling was accomplished in 26 areas, installing four permanent
parcels of 20 m ×1 m in each, totaling 104 parcels and 2,080 m2.All
tree individuals in the parcels were counted and identified to the species
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

714 E. M. Campos-Filho et al.
FIGURE 5 Mechanized direct-seeding using a machine originally designed for broadcasting
fertilizers, in northeast Mato Grosso State, west-central Brazil. After launching seeds over the
soil, the area is plowed to bury the seeds (Credit: Luciano Langmantel Eichholz) (color figure
available online).
FIGURE 6 Planter machine used for direct-seeding in-rows, Mato Grosso State, west-central
Brazil (Credit: Luciano Langmantel Eichholz) (color figure available online).
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

Mechanized Direct-Seeding of Native Forests 715
TABLE 1 Number of Sampled Areas in Each of Four Age Intervals and Respective Planting
Technique in Xingu Region, State of Mato Grosso, West-Central Brazil
Age after
planting (years) Planting technique
Sampled
areas Parcels
Sampled
area (m2)
0–0.5 Broadcast direct-seeding 6 24 480
0.6–2 Broadcast direct-seeding 4 16 320
0.6–2 Direct-seeding in rows 3 12 240
2.1–2.5 Direct-seeding in rows 4 16 320
2.1–2.5 Broadcast direct-seeding 2 8 160
2.1–2.5 Planting seedlings 3 12 240
2.6–5.5 Direct-seeding in rows 2 8 160
2.6–5.5 Broadcast direct-seeding 2 8 160
level. Annual and sub-perennial legumes were not sampled because it would
require other experimental designs and measurement methods. Height and
diameter data were missing in some plots due to lack of adequate equipment
at the time of monitoring and were thus excluded from this article.
MANAGEMENT
Only one area (Farm Simoni) did not receive any management measures
(Figure 7). In all other sampled areas weed control after planting was
necessary, because monitoring detected more than 30% of grass coverage
(Melo, 2004). Weed control was done by spraying selective herbicides.
RESULTS AND DISCUSSION
Since 2006, the Y Ikatu Xingu Campaign has facilitated the restoration
of 2,400 ha of riparian forest, on 198 farms in 23 municipalities of the
Xingu watershed in the state of Mato Grosso. This included the planting
of 98 tons of seeds from 214 native species, collected by over 300 peo-
ple from 20 rural and indigenous communities that earned together over
US$500,000. Approximately 1,100 ha underwent natural regeneration, 300 ha
were planted with seedlings, and 1,000 ha were planted with muvuca by
direct seeding (Figure 8).
Data from areas established by direct seeding presented very high varia-
tion of mean tree densities, regardless of the direct-seeding method or biome
(Table 2). Beyond the fact that these areas were not initially established as
scientific experiments, this large variation was not surprising, since areas
feature different climates, soil types, degradation histories, received different
assemblages of species from different lots of seeds, and seeds were randomly
distributed in the areas. Rodwell and Patterson (1994) support the idea that
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

716 E. M. Campos-Filho et al.
FIGURE 7 Area 5 months after being direct-seeded with muvuca. Note pigeon peas, jack
beans, and maize—Farm Simoni, Canarana, MT, Brazil (Credit: Osvaldo Luis de Sousa) (color
figure available online).
FIGURE 8 Same area shown in Figure 7, 3 yr after direct-seeding, with the trees forming a
canopy as the last pigeon peas dies and jack beans and maize are already out of the system at
Farm Simoni, Canarana, MT, Brazil (Credit: Elin Rømo Grande) (color figure available online).
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

Mechanized Direct-Seeding of Native Forests 717
TABLE 2 List of Monitored Areas, Restoration Techniques Applied, Local Biome, Age After
Planting (Years), and Mean Density (trees/hectare) in the Xingu Region of the State of Mato
Grosso, West-Central Brazil
Local Technique Biome Age
Mean
density
Farm Destino Lado B Broadcasting Amazon 0.5 4,850
Farm Destino Lado A Broadcasting Amazon 0.5 6,650
Farm Destino Lado C Broadcasting Amazon 0.5 7,750
Farm São Carlos direito Broadcasting Cerrado 0.5 8,500
Farm São Carlos–esq. Broadcasting Cerrado 0.5 12,650
Farm Destino Lado D Broadcasting Cerrado 0.5 21,750
Farm Brasil Seeding in rows Cerrado 0.6 3,625
Farm Bang-pasto 10 Broadcasting Amazon 1.0 4,000
Farm Bang-pasto 82 Broadcasting Amazon 1.4 5,750
Farm Cajuru Seeding in rows Cerrado 1.5 4,500
Farm Cajuru 2010 Seeding in rows Cerrado 1.5 21,875
Farm São Roque 2010 Broadcasting Cerrado 1.5 32,250
Farm Bang-pasto 71 Broadcasting Amazon 1.9 2,875
Farm Don José Seedlings Cerrado 2.5 1,500
Farm 22 de maio Seedlings Cerrado 2.5 1,500
Sítio 2 R Seedlings and legumes Cerrado 2.5 1,625
Farm Cajuru Seeding in rows Cerrado 2.5 2,500
Farm Candeia Seeding in rows Amazon 2.5 6,000
Farm Schneider–flores Seeding in rows Amazon 2.5 8,375
Farm Roncador Broadcasting Amazon 2.5 9,125
Farm Schneider–natura Seeding in rows Amazon 2.5 9,750
Farm São Roque 2009 Broadcasting Cerrado 2.5 21,375
Casa da criança Seeding in rows Cerrado 2.6 20,000
Farm São Roque 2008 Broadcasting Cerrado 3.5 14,500
Farm Simoni linhas Seeding in rows Cerrado 5.5 7,250
Farm Simoni lanço Broadcasting Cerrado 5.5 7,375
local variation in tree spacing and size, including small areas of randomly
occurring open space, creates a more natural appearance in the developing
new native forest, as the anthropogenic “fingerprint” of regular spacing and
standardized tree size is reduced.
A mean young tree density of 9,535 trees ha−1may be considered high
in comparison to areas planted with seedlings. The lowest tree density in
direct seeding areas, ranging from 2,500 to 32,250 trees ha−1, were still
higher than the highest tree density found in areas planted with seedlings
(1,500 to 1,625 trees ha−1)inXinguandinmostareasplantedwithseedlings
in Brazil (1,666 trees ha−1and 3 m ×2 m spacing) at the same age
(Rodrigues, Lima, Gandolfi, & Nave, 2009). However, high densities of young
trees (dbh ≤5 cm) seem to reflect better what was observed in mature
forests elsewhere. Silva, Leite, Silveira, Nassif, and Rezende (2004) observed
21,482 young trees ha−1in areas of Cerrado riparian forest in central Brazil;
whereas in Amazonian forests in the state of Pará, northern Brazil, Carvalho
(1982) observed 7,653 young trees ha−1. Rayol, Alvino, and Silva (2008)
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

718 E. M. Campos-Filho et al.
observed 70,800 and 80,600 trees ha−1, in 15- and 20-yr-old secondary forests,
respectively.
Initially, high density areas direct-seeded in Xingu showed signs of nat-
ural thinning, as can be inferred by comparing areas of different ages at
the same farm. For instance, at Farm São Roque 32,250 trees ha−1were
found 1.5 yr after planting, 21,375 trees ha−1were found after 2.5 yr, and
14,500 trees ha−1were found after 3.5 yr (Table 2). According to Willoughby
et al. (2004), high tree densities means that occasional damage to trees is
less serious than when establishing seedlings at substantially lower stock-
ing levels. Indeed, natural thinning by extended droughts leave survivors
more resilient, while natural thinning by herbivores indicates that animals are
using the area and possibly contributing to nutrient cycling (Miyawaki, 2004).
Regarding the initiation of a restoration process, Willoughby et al. (2004)
have found that high sapling densities and absence of transplant shock can
result in much earlier canopy closure (3–5 yr after sowing) for direct-seeded
stands compared with traditional seedlings plantation at a wider spacing.
Thus a forest environment is created far more quickly and early canopy clo-
sure reduces the length of time herbicides and other management measures
may need to be applied (Willoughby et al., 2004).
Sub-perennial (Cajanus cajan) and annual legumes (Cannavalia ensi-
formis and Crotalaria spectabilis), regarded as green manure, were present
in all direct seeding treatments and in one of the seedling treatments. C.
spectabilis and C. ensiformis died with 6 to 8 months, while the sub-perennial
legume (Cajanus cajan) survived until the 4th yr after planting. None of
those legumes were observed resprouting or regenerating in these areas
beyond the 5th yr, which was desirable since those are not native species.
Those short-lived, fast-growing, and N-fixing legumes formed an early and
heterogeneous canopy cover over the germinating trees from the 2nd month
after planting (field observation), while in areas planted only with seedlings
canopy cover were not formed after 30 months. In southeast Brazil, canopy
cover of areas planted with seedlings begins to close between 2 and 3 yr
(Brancalion et al., 2010). During the initial 2.5 yr, the appearance of areas
direct-seeded with muvuca (annual plants, sub-perennial legumes, and trees-
forming layers of vegetation with spatial heterogeneity) is very different from
the appearance of areas planted with seedlings (a layer of grasses with reg-
ularly spaced trees). Muvuca areas more closely resemble areas of natural
regeneration and seem to provide a greater diversity of suitable niches for
re-colonization by non-introduced species.
While fast-growing nitrogen-fixing legumes may compete for water and
light with young trees, an adequate density seems to foster growth of trees
by enhancing aeration, de-compaction, and water absorption in the soil
(Dubois & Viana, 1994). Legumes also yield a nitrogen-rich litterfall, possibly
accelerating nutrient cycling and restoration of soil fertility (Peneireiro, 1999).
An excessively dense canopy of fast-growing legumes, such as pigeon pea
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

Mechanized Direct-Seeding of Native Forests 719
and jack bean, slows tree growth during the wet season when its dense
foliage prevents most of the sunlight from reaching the saplings, but it also
seems to slow the growth of invasive grasses, which is desirable. Competition
seems to turn into facilitation during the dry season, when fast-growing
legume species protect young trees and the soil from sunlight and heat,
creating humid and stable microclimates under its canopy.
The large variation in tree density obtained from similar direct-seeding
operations, with high seeding density and diversity of species and guilds,
derives from uneven germination and survival at different sites and complies
with the idea that ecological restoration is a non-deterministic process open
to stochastic events that may not lead to a single pre-defined climax (Pickett,
Collins, & Armesto, 1987; Parker & Pickett, 1999; Pickett & Cadenasso, 2005).
The density sampled in seedling plantations in Xingu and which is usually
recommended in Brazil (1,666 trees ha−1) is based on the average density
of adult trees (dbh >10 cm) in tropical forests, thus ignoring the large
stock of young trees that leads to the inverse J-shaped curve of diameter
distribution typically found in natural forest communities (Barbour et al.,
1987). By planting seedlings at the approximate desired density of adult
trees, it becomes necessary to fight against ecological processes that cause
natural thinning of seedlings and lead to an old-growth forest structure. This
is a problem of major concern when planting seedings, if re-establishing
ecological process is the main goal of a restoration project.
A list of all the native species that were direct-seeded and germinated
is presented in Table 3. Sampling was not extensive enough to determine
species-specific survival rates resulting from direct seeding.
Although the Xingu Seed Network provided 214 species, we could only
find 89 species growing in the direct-seeded areas (42% of the species sown).
Those 214 species germinated in nurseries, although some at very low rates,
some required complicated pretreatments for breaking dormancy that could
not be conducted in the field, some died during storage and some were too
few and could have been overlooked during sampling. Most of the sampled
species have orthodox1behavior, which makes them easier to store and wait
until the beginning of the rainy season to plant. Storing recalcitrant seeds
is still a major challenge. Those species must be either collected just before
seeding or planted as seedlings. The pre-treatment method for breaking seed
dormancy of orthodox seeds is not effective for all species, but was useful to
accelerate the germination of part of the introduced seed bank, while the rest
germinated throughout the following years. However, the number of species
established by direct seeding (89) in Xingu is higher than the richness estab-
lished by planting seedlings in Xingu (47) and in most projects established
in the Brazilian Atlantic Forest (30 species), where forest restoration has a
much longer research history (Barbosa et al., 2003; Rodrigues et al., 2009).
Comparing costs of different techniques is also essential to scaling up
a forest restoration campaign. The overall cost per sapling established in
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

720 E. M. Campos-Filho et al.
TABLE 3 List of Species and Respective Families That Were Sampled in YIX Restoration
Projects Planted by Direct Seeding in the Xingu Region, State of Mato Grosso, West-Central
Brazil
Botanic family Scientific name
Anacardiaceae Myracrodruon urundeuva Allemão
Anacardiaceae Anacardium humile A. St. Hil.
Anacardiaceae Astronium fraxinifolium Schott ex Spreng.
Anacardiaceae Anacardium nanum A. St. Hil.
Anacardiaceae Spondias mombin L.
Apocynaceae Himatanthus obovatus (Müll. Arg.) Woodson
Apocynaceae Himatanthus sucuuba (Spruce ex Müll. Arg.) Woodson
Apocynaceae Aspidosperma macrocarpon Mart.
Apocynaceae Aspidosperma tomentosum Mart.
Apocynaceae Aspidosperma subincanum A.DC.
Arecaceae Attalea maripa (Aubl.) Mart.
Arecaceae Attalea phalerata Mart.exSpreng.
Arecaceae Mauritia flexuosa L.f.
Arecaceae Syagrus oleracea (Mart.) Becc.
Bignoniaceae Cybistax antisyphilitica (Mart.) Mart.
Bignoniaceae Jacaranda copaia (Aubl.) D.Don.
Bignoniaceae Jacaranda cuspidifolia Mart.exA.DC.
Bignoniaceae Handroanthus heptaphyllus (Vell.) Mattos
Bignoniaceae Handroanthus ochraceus (Cham.) Mattos
Bignoniacee Handroanthus serratifolius (A.H.Gentry) S.Grose
Bignoniacee Tabebuia aurea (Silva Manso) Benth. & Hook.f. ex S.Moore
Bignoniacee Tabebuia roseoalba (Ridl.) Sand.
Bixaceae Bixa orellana L.
Calophyllaceae Calophyllum brasiliense Cambess
Caryocaceae Caryocar brasiliense Cambess
Combretaceae Terminalia argentea Mart. & Zucc.
Combretaceae Buchenavia tomentosa Eichler
Combretaceae Buchenavia capitata (Vahl) Eichler
Dilleniaceae Curatella americana L.
Euphorbiaceae Mabea fistulifera Mart.
Euphorbiaceae Mabea angustifolia Spruce ex Benth.
Fabaceae Anadenanthera colubrina (Vell.) Brenan
Fabaceae Andira cujabensis Benth.
Fabaceae Andira vermifuga Mart. ex Benth.
Fabaceae Apuleia leiocarpa (Vogel) J.F. Macbr.
Fabaceae Bauhinia sp. L.
Fabaceae Copaifera langsdorffii Desf.
Fabaceae Copaifera marginata Benth.
Fabaceae Copaifera martii Hayne
Fabaceae Dialium guianense (Aubl.) Sandwith
Fabaceae Dimorphandra mollis Benth.
Fabaceae Dipteryx alata Vogel
Fabaceae Dipteryx odorata (Aubl.) Willd.
Fabaceae Enterolobium timbouva Mart.
Fabaceae Enterolobium schomburgkii (Benth.) Benth.
Fabaceae Enterolobium maximum Ducke
Fabaceae Hymenaea courbaril L.
Fabaceae Hymenaea stigonocarpa Mart. ex Hayne
Fabaceae Leptolobium nitens (Vog.) Yakov.
Fabaceae Machaerium acutifolium Vogel
(Continued)
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

Mechanized Direct-Seeding of Native Forests 721
TABLE 3 (Continued)
Botanic family Scientific name
Fabaceae Mimosa setosa Benth. var.paludosa(Benth.) Barneby
Fabaceae Ormosia excelsa Spruce ex Benth.
Fabaceae Ormosia paraensis Ducke
Fabaceae Parkia pendula (Willd.) Benth. ex Walp.
Fabaceae Peltogyne confertiflora (Mart. ex Hayne) Benth
Fabaceae Plathymenia reticulata Benth.
Fabaceae Platypodium elegans Vogel
Fabaceae Pterodon pubescens (Benth.) Benth.
Fabaceae Pterogyne nitens Tul.
Fabaceae Samanea tubulosa (Benth.) Barneby & J. W. Grimes
Fabaceae Schizolobium amazonicum Huber ex Ducke
Fabaceae Senegalia polyphylla (DC.) Britton & Rose
Fabaceae Senna silvestris (Vel.) H.S.Irwin & Barneby
Fabaceae Stryphnodendron pulcherrimum (Willd.) Hochr.
Fabaceae Tachigali paniculata Aubl.
Loganiaceae Strychnos pseudoquina A. St. Hil.
Malpighiaceae Byrsonima crispa A. Juss.
Malpighiaceae Byrsonima cydoniifolia A. Juss.
Malpighiaceae Byrsonima pachyphylla A. Juss.
Malpighiaceae Byrsonima verbascifolia (L.) DC.
Malvaceae Apeiba tibourbou Aubl.
Malvaceae Ceiba speciosa (A. St. Hil.) Ravenna
Malvaceae Eriotheca gracilipes (K. Schum.) A. Robyns
Malvaceae Guazuma ulmifolia Lam.
Malvaceae Sterculia striata A. St. Hil. & Naud.
Malvaceae Sterculia apetala (Jacq.) H.Karst.
Meliaceae Swietenia macrophylla King
Menispermaceae Abuta grandifolia (Mart.) Sandwith
Moraceae Maclura tinctoria (L.) D.Don ex Steud.
Passifloraceae Passiflora coccinea Aubl.
Passifloraceae Passiflora nitida Kunth
Rubiaceae Genipa americana L.
Sapindaceae Magonia pubescens A. St. Hil.
Simaroubaceae Simarouba versicolor A. St. Hil.
Simaroubaceae Simarouba amara Aubl.
Solanaceae Solanum lycocarpum A. St. Hil.
Solanaceae Solanum crinitum Lam.
Strelitziaceae Phenakospermum guyannense (A.Rich.) Endl. ex Miq.
Urticaceae Cecropia pachystachya Trécul.
direct-seeding in Xingu was US$0.19 per 2.5-yr-old sapling, US$0.25 per
5.5-yr-old sapling, and US$0.74 per 2.5-yr-old sapling in the area with the
lowest resulting density (2,500 saplings ha−1). While the latter density is
comparable to that usually attained by planting seedlings in Brazil, costs per
established sapling by direct-seeding are still 4.6 times cheaper than that of
seedlings, at the same age. Costs per area of direct-seeding in the Xingu
region (Table 4) were US$1,845 ha−1(53% for seeds and 47% for planting
and taking care of each hectare during 3 yr), slightly higher than the costs
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

722 E. M. Campos-Filho et al.
TABLE 4 Costs in Dollars per Hectare of Forest Restoration Using Direct-Seeding, With 3-Yr
Maintenance, in the Xingu Region, State of Mato Grosso, West-Central Brazil
Item Unit
Cost per
unit (US$) Units/hectare
Total
(US$/hectare)
Jack bean seeds kg 2.00 30 60.00
Pigeon pea seeds kg 2.50 6 15.00
Native seeds Kg 15.00 60 900.00
Heavy plowing machine/hour 60.00 2 120.00
Soft plowing machine/hour 60.00 2 120.00
Sowing machine/hour 60.00 0.5 30.00
Herbicide application 50.00 2 100.00
Planting and Management day-work 25.00 20 500.00
Total 1,845.00
TABLE 5 Costs in Dollars per Hectare of Forest Restoration Using Seedlings, With 3 Yr of
Maintenance in the Xingu Region, State of Mato Grosso, West-Central Brazil
Item Unit
Cost per
unit (US$) Units/hectare
Total
(US$/hectare)
Seedlings seedling 1.00 1,666 1,666.00
Seedlings transport freight 75.00 3 225.00
Herbicide application 40.00 4 160.00
Pit opening machine/hour 60.00 8 480.00
Planting day-work 25.00 8 200.00
Irrigation machine-hour 60.00 12 720.00
Formicide application 20.00 24 480.00
Replanting day-work 25.00 2 50.00
Weeding day-work 25.00 45 1,125.00
Total 5,106.00
calculated by Engel and Parrotta (2001) in southeastern Brazil, ranging from
US$760 to US$1,450 ha−1. Planting seedlings in Xingu (Table 5) cost about
US$5,100 ha−1(32% for raising seedlings in a nursery and 68% for plant-
ing and taking care of each seedling) or US$3.4 per established 3-yr-old
sapling. This is comparable to costs calculated among several projects in
the Brazilian Atlantic Forest of southeastern Brazil, which ranged between
US$3,315–US$5,216 ha−1or US$3.45 per sapling, whereas US$0.45 (13%) for
raising one seedling in a nursery and US$3 (87%) for planting and taking care
of each seedling during 3 yr (Rodrigues et al., 2011). The difference in cost
per hectare is such that, with the same money, one can either restore 1 ha
using seedlings or 2.77 ha by direct-seeding. Still, one can either establish
one 3-yr-old sapling with seedlings or between 4 to 18 3-yr-old saplings by
direct-seeding, with the same cost.
Establishing operations for direct seeding requires less time and fewer
workers than seedling transplantation. As an example, six workers have
planted 20 ha in 1 day by direct-seeding, while planting seedlings in the
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

Mechanized Direct-Seeding of Native Forests 723
same area could take about 20 days for the same six workers. Seeds are
easier and cheaper to transport—a pick-up truck can either carry seedlings
for 1.2 ha or seeds for 15 ha—and do not suffer during long trips on poor
roads, while seedlings suffer damages from excessive shaking and dry wind.
In addition to the economic advantage, Willoughby et al. (2004) explain that
direct seeding uses techniques that are more akin in many ways to the pro-
duction of arable crops than to conventional forestry. All ground preparation
and sowing can be carried out using modified agricultural machinery. This
may be more attractive to a farmer, skilled in establishing agricultural crops
but less skilled and perhaps unwilling to engage in the largely labor-intensive
planting and establishment of trees using seedlings.
CONCLUSIONS
Both techniques of mechanized direct-seeding established more saplings
of up to 5.5 yr old with much lower costs and greater expediency than
by transplanting seedlings. High initial densities of introduced trees, herbs,
and shrubs allowed reduction of maintenance costs for controlling herbi-
vores and invasive grasses. Further research is needed to understand the
interactions between densities of species of trees and fast-growing legumes
concerning facilitation and competition processes during dry and wet sea-
sons. Meanwhile, beans can be harvested for seed or to feed animals, while
others are edible to humans and can render commercialization that may
attenuate the restoration costs.
However, effective weed control during the first years is essential to
ensure sapling survival and reduce the need for weeding in subsequent
years. In most large-scale projects, overall spraying of herbicides is currently
the cheapest and most practical option for controlling weeds. Improvements
in machines for cutting grasses and mulching around the trees can be an
alternative means of controlling and benefiting from grasses, which would be
useful for both direct-seeding and seedling techniques. Mulching would pro-
vide more biomass, faster soil restoration, and probably enhance tree growth.
There is further research to be done on regional plants’ life cycles,
growth rates, shade tolerance, associations for nutrient fixation, phenology,
dispersal syndromes, and pollination systems. Such considerations inform
the process of providing adequate densities of individuals from each suc-
cessional phase and forest stratum, composing a mix of species with all the
vegetation ecological features, and thus avoiding waste of seeds and eco-
logical gaps. Adequate seeding density of trees and fast-growing legumes
may enhance seed germination and survival and help to avoid the use of
herbicides for grass control.
The use of machines originally designed for broadcasting limestone,
fertilizers, and grass seeds proved to be very practical and allowed mixing
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

724 E. M. Campos-Filho et al.
of larger seeds. In spite of the fact that in-row planters limit seed size and
require more attention to the plant, they facilitated weed control and seed-
ing of alternative rows with distinct assemblages of species. Overcrowding
may prove detrimental to biomass accumulation and, if that is the objective,
require future thinning of established trees.
Direct seeding may be particularly suited to large-scale areas where
mechanization is technically and economically viable. However, it is impor-
tant to note that direct-seeding and planting seedlings should be complemen-
tary techniques for high-diversity forest restoration, since recalcitrant seeds
can hardly be introduced on a large scale by direct-seeding and nurseries
remain the only reliable way for doing so. We believe that it is possible
that the YIX model of broad community-based seed network associated with
muvuca mechanized direct-seeding, small nurseries, and natural regener-
ation can be multiplied and adapted to accomplish the restoration of the
other 312,600 degraded hectares of riparian areas in the Xingu Basin.
NOTE
1. Orthodox seeds can be stored dry (seed water content between 3–4%) and cold (less than
−15◦C) for long periods (over a year). Typical recalcitrant seeds die if stored with less than 14% of its
water content and if stored at less than 10◦C. Over these limits, recalcitrant seeds either germinate or rot.
REFERENCES
Araki, D. F. (2005). Avaliação da semeadura a lanço de espécies florestais nati-
vas para recuperação de áreas degradadas (Master’s dissertation). São Paulo
University, Piracicaba, SP, Brazil.
Aronson, J., Brancalion, P. H. S., Durigan, G., Rodrigues, R. R., Engel, V. L., Tabarelli,
M.....Scarano, F. R. (2011). What role should government regulation play in
ecological restoration? Ongoing debate in São Paulo State. Brazilian Restoration
Ecology,19, 690–695.
Barbosa, L. M., Barbosa, J. M., Barbosa, K. C., Potomati, A., Martins, S. E., Asperti, L.
M., . . . Plaza, A. P. (2003). Recuperação florestal com espécies nativas no Estado
de São Paulo: Pesquisas apontam mudanças necessárias. Florestar Estatístico,6,
28–34.
Barbour, M. G., Burk, J. H., & Pitts, W. D. (1987). Terrestrial plant ecology (2nd
Edition). Menlo Park, CA: Benjamin/Cummings Publishing.
Brancalion, P. H. S., Rodrigues, R. R., Gandolfi, S., Kageyama, P. Y., Nave, A. G.,
Gandara, F. B., . . . Tabarelli, M. (2010). Legal instruments can enhance high-
diversity tropical forest restoration. Revista Árvore,34(3), 455–470.
Camargo, J. L. C., Ferraz, I. D. K., & Imakawa, A. M. (2002). Rehabilitation of
degraded areas of central Amazonia using direct sowing of forest tree seeds.
Restoration Ecology,10, 636–644.
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

Mechanized Direct-Seeding of Native Forests 725
Carvalho, J. O. P. (1982). Análise estrutural da regeneração natural em floresta tropi-
cal densa na região de Tapajós no Estado do Pará (Master’s dissertation). Federal
University of Paraná, Curitiba, PR, Brazil.
Castellanet, C., & Jordan, C. F. (2002). Participatory action research in natural
resource management: A critique of the method based on five years’ experience
in the Transamazônica region of Brazil. New York, NY: Taylor & Francis.
Chazdon, R. L. (2008). Beyond deforestation: Restoring forests and ecosystem
services on degraded lands. Science,320, 1458–1460.
Coe, M. T., Costa, M. H., & Howard, E. A. (2007). Simulating the surface waters of
the Amazon River Basin: Impacts of new river geomorphic and dynamic flow
parameterizations. Hydrological Processes,22, 2542–2553.
Cole, R. J., Holl, K. D., Keene, C. L. & Zahawi R. A. (2010). Direct seeding of
late successional trees to restore tropical montane forest. Forest Ecology and
Management,261, 1590–1597.
Doust, S. J., Erskine, P. D., & Lamb, D. (2006). Direct seeding to restore rainforest
species: Microsite effects on the early establishment and growth of rainforest
tree seedlings on degraded land in the wet tropics of Australia. Forest Ecology
and Management,234, 333–343.
Doust, S. J., Erskine, P. D., & Lamb, D. (2008). Restoring rainforest species by direct
seeding: Tree seedling establishment and growth performance on degraded
land in the wet tropics of Australia. Forest Ecology and Management,256,
1178–1188.
Dubois, J. C. L., & Viana, V. M. (1994). Da capacidade produtiva da terra. Informativo
Agroflorestal,6(3) editorial.
Durigan, G., Guerin, N., & Costa, J. N. M. N. (2013). Ecological restoration of
Xingu Basin headwaters: Motivations, engagement, challenges and perspec-
tives. Philosophical Transactions of the Royal Society of London, Series B,368,
20120165.
Engel, V. L., & Parrotta, J. A. (2001). An evaluation of direct seeding for reforesta-
tion of degraded lands in central São Paulo state, Brazil. Forest Ecology and
Management,152, 169–181.
Evelyn, J. (1670). Sylva or a discourse of forest trees (Vol. 1). York, United Kingdom:
Doubleday & Co.
Franco, A., Nardoto, G. B., & Souza, M. P. (1996). Patterns of soil water potential
and seedling survival in the cerrados of Central Brazil. In R. C. Pereira & L. C.
B. Nasser (Eds.), Proceedings of the 1st International Symposium on Tropical
Savannas (pp. 277–280). Planaltina, Brazil: Embrapa-CPAC.
Harmer, R., & Kerr, G. (1995). Creating woodlands: To plant trees or not? In R. Ferris-
Kaan (Ed.), The ecology of woodland creation. Chichester, United Kingdom:
John Wiley.
Holl, K. D., Zahawi, R. A., Cole, R. J., Ostertag, R., & Cordell, S. (2010). Planting
seedlings in tree islands versus plantations as a large-scale tropical forest
restoration strategy. Restoration Ecology,19, 470–479.
ISA. (2013). ‘Y Ikatu Xingu’—Salve a água boa do Xingu, na língua
Kamaiurá. Retrieved from http://www.yikatuxingu.org.br/linhas-de-acao/
mapa-de-iniciativas/
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

726 E. M. Campos-Filho et al.
Isernhagen, I. (2010). Uso de semeadura direta de espécies arbóreas nativas para
restauração florestal de áreas agrícolas, sudeste do Brasil (Doctoral thesis). São
Paulo University, Piracicaba, SP, Brazil.
Ivanauskas, N. M., Monteiro, R., & Rodrigues, R. R. (2008). Classificação fitogeográ-
fica das florestas do Alto Rio Xingu. Acta Amazonica,38(3), 387–402.
Köppen, W. (1948). Climatologia. Mexico, DF, Mexico: Fondo de Cultura Económica.
Lamb, D., Erskine, P. D., & Parrotta, J. A. (2005). Restoration of degraded tropical
forest landscapes. Science 310, 1628–1632.
Lima, E., Merry, F., Nepstad, D. C., Amacher, G., Azevedo-Ramos, C., Lefebvre, P. A.,
& Resque, F. (2006). Searching for sustainability: Forest policies, smallholders,
and the Trans-Amazon Highway. Environment,48(2), 26–38.
Lugo, A. E. (1988). The future of the forest. Ecosystem rehabilitation in the tropics.
Environment,30, 17–20, 41–45.
Melo, A. C. G. (2004). Reflorestamentos de restauração de matas ciliares: Análise
estrutural e método de monitoramento no Médio Vale do Paranapanema (SP)
(Master’s dissertation). São Paulo University, São Carlos, SP, Brazil.
Miyawaki, A. (2004) Restoration of living environment based on vegetation ecology:
Theory and practice. Ecological Research,19, 83–90.
Montagnini, F. (2001). Strategies for the recovery of degraded ecosystems:
Experiences from Latin America. Interciencia,26, 498–503.
Nepstad, D., Uhl, C., & Serrao, E. A. S. (1991). Recuperation of a degraded
Amazonian landscape: Forest recovery and agricultural restoration. Ambio,20,
248–255.
Oliveira, R. S., Bezerra, L., Davidson, E. A., Pinto, F., Klink, C. A., Nepstad, D. C.,
& Moreira, A. (2005). Deep root function in soil water dynamics in cerrado
savannas of Central Brazil. Functional Ecology,19(4), 574–581.
Parker, V. T., & Pickett, S. T. A. (1999). Restoration as an ecosystem process:
Implications of the modern ecological paradigm. In K. M. Urbanska, N. R. Webb,
&P.J.Edwards(Eds.),Restoration ecology and sustainable development (pp.
17–32). Cambridge, United Kingdom: Cambridge University Press.
Peneireiro, F. M. (1999). Sistemas agroflorestais dirigidos pela sucessão natural: Um
Estudo de Caso (Master’s thesis). São Paulo University, Piracicaba, SP, Brazil.
Pickett, S. T. A., & Cadenasso, M. L. (2005). Vegetation dynamics. In E. van der Maarel
(Ed.), Vegetation ecology (pp.172–198). Oxford, United Kingdom: Blackwell.
Pickett, S. T .A., Collins, L. S., & Armesto, J. J. (1987). Models, mechanisms and
pathways of succession. Botanical Review,53, 335–371.
Posey, D. A. (1985) Indigenous management of tropical forest ecosystems: The case
of the Kayapó Indians of the Brazilian Amazon. Agroforestry Systems,3(2),
139–158.
Rayol, B. P., Alvino, F. O., & Silva, M. F. F. (2008). Structure and floristic composition
of the natural regeneration of two secondary forests in Capitão Poço, Pará,
Brazil. Amazônia: Ci. & Desenv.,4(7), 103–116.
Rodrigues, R. R., Gandolfi, S., Nave, A. G., Aronson, J., Barreto, T. E., Vidal,
C. Y., & Brancalion, P. H. (2011). Large-scale ecological restoration of high-
diversity tropical forests in SE Brazil. Forest Ecology and Management,261(10),
1605–1613.
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014

Mechanized Direct-Seeding of Native Forests 727
Rodrigues, R. R., Lima, R. A. F., Gandolfi, S., & Nave A. G. (2009). On the restoration
of high diversity forests: 30 years of experience in the Brazilian Atlantic Forest.
Biological Conservation,142, 1242–1251.
Rodwell, J., & Patterson, G. (1994). Creating new native woodlands (Forestry
Commission Bulletin, 112). London, United Kingdom: HMSO.
Sanches, R. A., & Villas-Bôas, A. (2005). Planejando a gestão em um cenário
socioambiental de mudanças: O caso da bacia do rio Xingu. Revista Brasileira
de Administração Pública,39, 365–379.
Santos, P. L., Ferreira, R. A., Aragão, A. G., Amaral, L. A., & Oliveira, A. S. (2012).
Estabelecimento de espécies florestais nativas por meio de semeadura direta
para recuperação de áreas degradadas. Revista Árvore,36 (2), 237–245.
Silva, J. A., Leite, E. J., Silveira, M., Nassif, A. A., & Rezende, S. J. M. (2004).
Phytosociological and floristic characterization and natural regeneration of the
undergrowth at Tamandua Forest Genetic Reserve, Brazil. Ciência Florestal,
14(1), 121–132.
Velasquez, C., Queiroz, H., & Bernasconi, P. (2010). Fique por dentro: A bacia do Rio
XinguemMatoGrosso(Series Cartô Brasil Socioambiental). São Paulo, Brazil:
Instituto Socioambiental, Instituto Centro de Vida.
Willoughby, I., Jinks, R. Gosling, P., & Kerr, G. (2004). Creating new broadleaved
woodland by direct seeding—Practice guide. Edinburgh, United Kingdom:
Forestry Commission.
Downloaded by [Eduardo M. Campos-Filho] at 11:25 07 March 2014