land use change patterns
TRANSCRIPT
-
8/12/2019 Land Use Change Patterns
1/6
Short communication
Land use change patterns in the Ro de la Plata grasslands: The influence ofphytogeographic and political boundaries
Ernesto Vega a,*, German Baldi b, Esteban G. Jobbagy b, Jose Paruelo a
a Laboratorio de Analisis Regional y Teledeteccion, IFEVA-FAUBA/CONICET; Av. San Mart n 4453 (C1417DSE) Buenos Aires, Argentinab Grupo de Estudios Ambientales-IMASL, Universidad Nacional de San Luis/CONICET, Avenida Ejercito de Los Andes 950 1er piso, D5700HHW San Luis, Argentina
1. Introduction
Land Use and Land Cover Change (LULCC) is a key component of
global change (Dale et al., 2000; Vitousek et al., 1997). Its global
importance derived mainly from theimpact of the accumulation of
small local changes. LULCC may affect climate by modifying water
dynamics (Gordon et al., 2008; Nosetto et al., 2005), increasing of
greenhouse gases (CO2 and nitrous oxide) (Searchinger et al.,
2008), or changing the surface energy balance (Pielke and Avissar,
1990). Biodiversity is directly affected by LULCC, through the effect
of agricultural practices on specific populations (Mattison and
Norris, 2005) and habitat losses (Foley et al., 2008).
The plains locatedin the southernportionof the Ro dela Plata
Basin in eastern South America(Fig.1) hold one of the largest area
of temperate grasslands of theworld,the Ro dela Platagrasslands
(RPG onwards) (Soriano, 1991, pp. 367407;Paruelo et al., 2007,
pp. 232248). Although the RPG is a physiognomically homo-
geneous region, eight phytogeographic districts can be identified
(Leon, 1991, pp. 369387) (Fig. 1). Ro de la Plata grasslands have
an area of 3.4 106 km2 at the center-east of Argentina, south of
Brazil and Uruguay. The presence of three countries adds an
important socioeconomic, political and cultural heterogeneity to
the region. A huge transformation process started in century XVI
with the introduction of domestic cattle brought by European
conquerors and it was accelerated with century XIX farming
expansion (Hall et al., 1991, pp. 413450). Nowadays RPG is
perhaps one the regions with highest rates of land use change in
theworld(Paruelo et al.,2005, 2006;Baldiand Paruelo, 2008).The
Agriculture, Ecosystems and Environment 134 (2009) 287292
A R T I C L E I N F O
Article history:
Received 12 May 2009Received in revised form 30 July 2009
Accepted 31 July 2009
Available online 4 September 2009
Keywords:
Afforestation
Crop expansion
Markovian matrix models
Matrix sensitivity analysis
A B S T R A C T
The Ro de la Plata grasslands (RPG) in South America are one of the largesttemperate grasslands regions
of the world. This region plays a key role in international crop production and land use change rates in
some areas are among the highest detected nowadays. Our objective was to characterize the spatial
heterogeneity of land use change dynamics in the RPG and to relate it with biophysical and political
boundaries. Based on Landsat imagery we characterized land use changes at two time periods (1986
1990, 20022005) and we performed a comparison of markovian models and their properties (stable
proportion vector and sensitivity analysis) for each phytogeographic district and country of the region.
Temporal transitions between natural and implanted forests (Fo), crops (Cr), and grasslands (Gr) were
calculated in order to build matrix probabilistic models. We found that 1.2 106 ha of grassland surface
has been transformed into implanted forests and croplands (6% reduction of grasslands, 60% increase of
afforestations, 3% increase of croplands). Transition probability Cr ! Cr displayed the largest spatial
variation, followed by the other three transitions linking croplands and grasslands (Cr ! Gr, Gr ! Gr,
Gr ! Cr). The less variable transition rate among districts was Cr ! Fo.
Projections for Argentina and parts of Uruguay suggest that grassland loss would continue in most of
analyzed territory, whereas in Brazil and parts of Uruguay the projected trend was the opposite,
grassland cover would increase. Forested area increment would continue in Uruguay and partially in
Argentina,while in Brazil it would decrease. Considering the whole region, crop areawould maintain theincreasing trends. Transition probabilities for the same phytogeographical district (Northern Campos)
differed significantly among the three countries that occupy it, showing the relative importance of
human context driving land use change.
2009 Elsevier B.V. All rights reserved.
* Corresponding author. Present address: Laboratorio de Ecologa Genetica,
CIECO-UNAM, antigua carretera a Patzcuaro 8701, col. Ex-Hacienda San Josede la
Huerta, CP 58190, Morelia, Michoacan, Mexico. Tel.: +52 443 322 2704x42608;
fax: +52 443 322 2719.
E-mail addresses: [email protected],
[email protected](E. Vega).
Contents lists available atScienceDirect
Agriculture, Ecosystems and Environment
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / a g e e
0167-8809/$ see front matter 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.agee.2009.07.011
mailto:[email protected]:[email protected]://www.sciencedirect.com/science/journal/01678809http://dx.doi.org/10.1016/j.agee.2009.07.011http://dx.doi.org/10.1016/j.agee.2009.07.011http://www.sciencedirect.com/science/journal/01678809mailto:[email protected]:[email protected] -
8/12/2019 Land Use Change Patterns
2/6
-
8/12/2019 Land Use Change Patterns
3/6
country. Second, we analyzed the future dynamic of land use
change using projections of the observed trends.
2. Materials and methods
2.1. Study zone
Study zone is in Ro de la Plata grasslands and comprises three
countries: Argentina, Brazil and Uruguay (Fig. 1). This region is
occupied by steppes and grasslands co-dominated by C3 and C4(Soriano, 1991, pp. 367407;Burkart et al., 1998; Paruelo et al.,
2007, pp. 232248). Mean annual temperature ranges from 20 8C
in the north to 13 8C in the south. Annual precipitation varied from
600 mm in the southwest to 1500 mm in the northeast ( Paruelo
et al., 2007, pp. 232248). According to geomorphology, soil,
drainage, physiography and vegetation (Leon, 1991, pp. 369387)
eight phytogeograophic districts are identified: Flooding (F), West
Inland (W), Flat Inland (I), Mesopotamic (M), Austral (A) and
Rolling Pampas (R), and Northern (N) and Southern (S) Campos
(Leon, 1991, pp. 369387). The main crops in the region are
soybean, maize, sunflower,wheat, rice and oat (DIEA-MGAP, 2003;
IBGE, 2004; SAGPyA, 2002). There are important stands of
Eucalyptusspp. and Pinusspp. in the more humid zones (Jobbagy
et al., 2006).
2.2. Transition probabilities estimation
This study is based on a land use change characterization
obtained from Landsat TM and ETM imagery classification at two
time periods, 19851989 and 20022005. Total surface analyzed
comprises eight Landsat scenes (185,664 km2, which represent
27% oftotalarea ofRo de la Plata grasslands) distributed along the
main environmental gradients, phytogeographic districts and
countries in the region (Baldi and Paruelo, 2008).
This work stems out from that ofBaldi and Paruelo (2008), who
analyzed the change in abundance and distribution of three cover
types (or land uses): grasslands (natural and seminatural) (Gr),
annual crops (Cr) and forests stands (natural and commercial) (Fo)of eight Landsat scenes (Fig. 1) in two periods (19851989 and
20022005). On the each of the Landsat classifications (see details
inBaldi and Paruelo, 2008) we overlay an 8 km 8 km cell grid.
Transition probabilities were calculated with data gathered from
cells with at least 90% of their surface covered by the three land
uses. A total of 2899 cells were used, representing 85.17% of total
surface analyzed byBaldi and Paruelo (2008).
Cells were grouped according to phytogeographic districts
(eight) and countries (three). Seven districts were analyzed in
Argentina: Northern Campos (N), Austral (A), Floodable (F), Inland
(I), West (W), Rolling (R), Mesopotamic (M) and Flat (F) pampas. In
Brazil and Uruguay the Northern (N) and Southern (S) Campos
were analyzed. Changes of cover were thus evaluated in 11
subscenes (districts countries).
2.3. Markovian models of land use change
Land Use and Land Cover Change (LULCC) characterization was
assessed through the comparison of cover type proportionsin each
8 km 8 km cell. All theproportional changes from one cover type
to another (aij) were arrangedin the transition matrixA to generate
a cover change modelA n(t)=n(t+1), wheren is the vector of the
relative proportions of cover classes. In the context of this work w
is the dominant eigenvector of A and represents the stable
proportions vector of cover classes.
To find out whether matrix models of phytogeographic districts
(PD) of Argentina are different, a lineal generalized model was
applied, using aij as responsevariable, PD as a factorand a binomial
error (Crawley, 2007). In cases where overdispersion was detected,
a quasibinomial distribution was used instead (Crawley, 2007).
Tukey tests were used to compare 21 pairs of PD. Similar analysis
were applied to Northern Campos (comparing between Argentina,
Brazil and Uruguay) and to Southern Campos (comparing between
Brazil and Uruguay). All analyses were performed with R
statistical package (R Development Core Team, 2007).
2.4. Sensitivity analysis of stable vector proportions
The variation ofaij has an effect on w which can be measured
through vector sensitivity analysis:
@w1@ai j
w1 j
Xsm 6 1
vmi
l1 lmwm (1)
@w1/@aijrepresents the change onw1(the dominant eigenvector)
due toa change ofaij; w1j
is thejth element ofw; s is the size of the
matrix;vmi
is theith element ofmth left conjugated eigenvector;
lmis themth eigenvalue ofA. The result of this expression is a set
of matrices of the same size ofA, one for each element ofw. It is
possible, then, to evaluate how the variation of aij affects each
element of the final proportion vector. Sensitivity value (@w1/@aij)
is a measure of the effect that a change ofaij has on the elements ofthe stable proportions vectorw. If it is positive, then an increment
ofaijcauses an increment of an element ofw. If it is negative then
an increment ofaij causes a decrement of an element ofw. Methods
employed to calculatew and(1)were those proposed byCaswell
(2001).
Matrix models analyzed here are markovian, so a change in aijimplies the variation of the values of the rest of the elements of
columnj (or at least in one of them), so the summation along the
column is unity. Therefore, the increment of a particular cover will
always be related with the reduction of other. Sensitivity analysis
took in to account these aspects (Caswell, 2001).
2.5. Rates of land use change
Rate of change (%) (rt) for each cover was calculated as
rt= ((S1 S0)/S0) 100.S0and S1are the areas of a cover class at
(t= 0) and (t= 1) respectively. Two rates of change were obtained,
the observed and the expected. The first one is calculated from the
observed areas in the two observations periods. The expected rate
considered the area in the second period of observation (2002
2005) (S0), whileS1 was the area predicted by model projections.
Values or rtwere multiplied by (1), so a negative value would
mean a loss of cover, while a positive one is a gain.
Annual rate of change was calculated asra
100 1 1 S1 S0
S0
1=t !;
wheret= 10 years. Observed and expected rates were obtained as
above.
Four recent studies of land use change were used to comparethe
rates obtained in this work with them (White et al., 2000; Achard
et al., 2002; Morton et al., 2006; Hansen et al., 2008). These papers
were selected because they present data to calculatertandra.
2.6. Differences between observed and predicted covers
Distance between observed and predicted cover proportions at
time (t) can be measured as
cAt X
s
i1
nt i w
Ai
2(2)
E. Vega et al. / Agriculture, Ecosystems and Environment 134 (2009) 287292 289
-
8/12/2019 Land Use Change Patterns
4/6
where nti
is theith element of proportions vector at time (t) and
wAi is the ith element of thefinal proportionsvectorw of matrixA.
Summation ofcAt along x iterations ofA n(t)=n(t+1), gives the
total distance (Z(A)), a measure of how far is the observed
proportion vector at time (t) fromw, givenA:
ZA Xxt0
cAt (3)
In this context, a particular distanceZ(A) can also be referred asZobs.
A way to evaluate the oddity ofZ(A)derives from the fact that it is
possible to find matrices different of A but having the same
eigenvectorw. To calculate Pr(Z(A)), the probability of occurrence of
Z(A), 1000 markovian matrices X(different ofA) having the same
eigenvectorw ofA were obtained randomly (Hastings, 1970) for
each combination of phytogeographic district and country. Each
matrixXwas iterated 30 times as according to X n(t)=n(t+1), to
obtain itsZ(X). Pr(Z(A)) was calculated as the fraction ofZ(X)lesser
than Z(A). Thus it is possible to estimate how feasible is to find a
shorter route from n(t)to w than the one defined by A.
3. Results
Though in the three countries the highest transition probabil-ities were those to grassland persistence (Fig. 2), the land use
change dynamic showed important differences. Transitions
probabilities towards grasslands were highest in Uruguay,
followed by those of Brazil andArgentina. Transition rates towards
afforestations were similar in Argentina and Brazil, and lower in
Uruguay. Noticeably, in this country afforestation persistence is
clearly higher than in the others. The area occupied by grassland in
19851989 (64%) decreased almost 6% during the period studied.
Projections for Argentina and Uruguay suggest that this trend
would continue until reaching 58%. In Brazil, the projected trend
was the opposite: grassland cover would increase up to 70% (see
Suplementary material, Table ST1).
The comparison of the average matrix models for each country
showed that the largest transitions are those that point towards
grasslands, while the smallest point towards afforestations (Fig. 2).
However projections showed that grasslands surface would
diminish and afforestation would increase. This apparent contra-
diction may derive from the fact that expected trends of each cover
are a linear combination of all transitions and all covers. If a
transition rate is high it does not follow strictly that the associated
cover will increase.
Northern Campos and West Inland Pampa were the phytogeo-
grahic districts that showed the greatest differences with the rest
(seven out nine (77%) and 6.5 transitions (72%) different from the
rest of districts, respectively). The districts with fewer transitions
different from the rest were Flooding (57%) and Rolling pampas
(50%). The most contrasting pairs of districts were the Northern
Campos-Austral Pampa, Northern Campos-Inland Pampa, and
Mesopotamic Pampa-Austral Pampa, with eight out of ninetransition rates different in each case (see Suplementary material,
Fig. S1a).
Transition rate Cr ! Cr was the most variable: it differed in 19
of 21 district-pair comparisons (p
-
8/12/2019 Land Use Change Patterns
5/6
-
8/12/2019 Land Use Change Patterns
6/6
Acknowledgements
This work has been possible through funding of the Inter-
American Institute for Global Change Research (IAI, CRN II 2031),
which is supported by the US National Science Foundation (Grant
GEO-0452325) forEV. This work has also been funded underthe EU
6th Framework Programme for Research, Technological Develop-
ment and Demonstration, Priority 1.1.6.3. Global Change and
Ecosystems (European Commission, DG Research, contract 003874
(GOCE)). Its content does not represent the official position of the
European Commission and is entirely under the responsibility of
the authors.
Appendix A. Supplementary data
Supplementary data associated with this article canbe found, in
the online version, atdoi:10.1016/j.agee.2009.07.011.
References
Achard, F., Eva, H.D., Stibig, H.J., Mayaux, P., Gallego, J., Richards, T., Malingreau, J.P.,2002. Determination of deforestation rates of the worlds humid tropicalforests. Science 297, 9991002.
Baldi, G., Guerschman, J.P., Paruelo, J., 2006. Characterizing fragmentation intemperate South America grasslands. Agric. Ecosyst. Environ. 116, 197208.
Baldi, G., Paruelo, J., 2008. Land use and land cover dynamics in South Americantemperate grasslands (19852005 period). Ecol. Soc. 13, 6., http://www.ecologyandsociety.org/vol13/iss2/art6/.
Burkart, S.E., Leon, R.J.C., Perelman, S.B., Agnusdei, M., 1998. The grasslands of theFlooding Pampa (Argentina): floristic heterogeneity of natural communities ofthe southern Ro Salado basin. Coenoses 13, 1727.
Caswell, H., 2001. Matrix Population Models. Sinauer, USA.Crawley, M.J., 2007. The R Book. John Wiley & Sons.Dale, V.H., Brown, S., Haeuber, R.A., Hobbs, N.T., Huntly, N., Naiman, J., Riebsame,
W.E., Turner, M.G., Valone, T.J., 2000. Ecological principles and guidelines formanaging the use of land. Ecol. Appl. 3, 639670.
DIEA-MGAP, 2003. La Agricultura de Secano en Uruguay. Contribucion a su Con-ocimiento. Direccion de Estadsticas Agropecuarias Ministerio de Ganadera,Agricultura y Pesca, Montevideo, Uruguay.
Foley, J.A., DeFries, R., Asner, G.P., Badford, C., Bonan, G., Carpenter, S.R., Chapin, F.S.,Coe, M.T., Daily, G.C., Gibbs, H.K., Helkowsky, J.H., Holloway, T., Howard, E.A.,Kucharik,C.J., Monfreda, C., Patz,J.A., Prentice, I.C.,Ramankutty, N., Snyder, P.K.,2008. Global consequences of land use. Science 309, 570574.
Guerschman, J.P., Paruelo, J.M., Burke, I.C., 2003. Land use impacts on the normal-ized difference vegetation index in temperate Argentina. Ecol. Appl. 3, 616628.
Gordon, L.J., Peterson, G.D., Bennett, E.M., 2008. Agricultural modifications ofhydrological flows create ecological surprises. Trends Ecol. Evol. 4, 211219.
Hall, A.J., Rebella, C., Ghersa, C., Culot, J., 1991. Field crop systems of the Pampas. In:Pearson, C.J. (Ed.), Field Crop Ecosystems. Ecosystems of the World, vol. 19.Elsevier, Amsterdam.
Hansen, M.C., Stehman, S.V., Potapov, P.V., Loveland, T.R., Townshend, J.R.G.,DeFries, R.S., Pittman, K.W., Arunarwati, B., Stolle, F., Steininger, M.K., Carroll,M., DiMicelli, C., 2008. Humid tropical forest clearing from 2000 to 2005quantified by using multitemporal and multiresolution remotely sensed data.Proc. Natl. Acad. Sci. U.S.A. 27, 94399444.
Hastings,W.K., 1970.Monte Carlosamplingmethods usingmarkov chains and theirapplications. Biometrika 1, 97109.
Helmut, J.G., Lambin, E.F., 2001. What Drives Tropical Deforestation? LUCC ReportSeries No. 4. LUCC International Project Office, Louvain-la-Neuve, Belgium.
IBGE, 2004. Producao Agrcola Municipal. Instituto Brasileiro de Geografia e Esta-tstica, http://www.ibge.gov.br/home/estatistica/economia/pamclo/2004/pamclo2004.pdf.
Jobbagy, E.G., Vasallo, M., Farley, K.A., Pin eiro, G., Garbulsky, M.F., Nosetto, M.D.,Jackson, R.B., Paruelo, J.M., 2006. Forestacion en pastizales: hacia una visionintegral de sus oportunidades y costos ecologicos. Agrociencias 10, 109124.
Imhoff, M.L., Bounoua, L., Ricketts, T., Loucks, C., Harris, R., Lawrence, W., 2004.
Global patterns in human consumption of net primary production. Nature 429,870873.Lesschen, J.P., Verburg, P.H., Staal, S.J., 2005. Statistical Methods for Analyzing the
Spatial Dimension of Changes in Land Use and Farming Systems. LUCC ReportSeries No. 7. International Livestock Research Institute & Wageningen Univer-sity, The Netherlands.
Leon, R.J.C., 1991. Geographic limits of the region, geomorphology and geology,regional subdivisions, floristic aspects, description of the vegetation. In: Coup-land, R.T. (Ed.), Natural Grasslands: Introduction and Western Hemisphere.Elsevier, Amsterdam.
Mattison, E.H.A., Norris, K., 2005. Bridging gaps between agricultural policy, land-use and biodiversity. Trends Ecol. Evol. 11, 610616.
Morton, D.C., DeFries, R.S., Shimabukuro, Y.E., Anderson, L.E., Arai, E., del bonEspirito-Santo, F., Freitas, R., Morisette, J., 2006. Cropland expansion changesdeforestation dynamics in the southern Brazilian Amazon. Proc. Natl. Acad. Sci.U.S.A. 39, 1463714641.
Nosetto, M.D., Jobbagy, E.G., Paruelo, J.M., 2005. Land-use change and water losses:the case of grassland afforestation across a soil textural gradient in centralArgentina. Global Change Biol. 11, 117.
Paruelo, J.M., Guerschman, J.P., Veron, S., 2005. Expansion agrcola y cambios en eluso del suelo. Ciencia Hoy 87, 1423.
Paruelo, J.M.,Guerschman, J.P.,Pineiro, G.,Jobbagy,E.G.,Veron, S.R., Baldi,G., Baeza,S., 2006. Cambios en el uso de la tierra en Argentina y Uruguay: marcosconceptuales para su analisis. Agrociencia 2, 4761.
Paruelo, J.M., Jobbagy, E.G., Oesterheld, M., Golluscio, R.A., Aguiar, M.R., 2007. Thegrasslands and steppes of Patagonia and the R o de la Plata Plains. In: Veblen,T., Young, K., Orme, A. (Eds.), The Physical Geography of South America.Oxford University Press.
Pielke, R.A., Avissar, R., 1990. Influence of landscape structure on local and regionalclimate. Landscape Ecol. 4, 133155.
R Development Core Team, 2007. R: A Language and Environment for StatisticalComputing. R Foundation for Statistical Computing, Vienna, Austria. , ISBN: 3-900051-07-0,http://www.R-project.org.
SAGPyA, 2002. El Quinquenio de la Soja Transgenica. Secretara de Agricultura,Pesca y Alimentos, Argentina. , http://www.sagpya.mecon.gov.ar/new/0-0/prensa/publicaciones/agricultura/Soja.pdf.
SAGPyA, 2004. Indicadores del Sector Sojero 2003/2004. Secretara de Agricultura,Pesca y Alimentos, Argentina. , http://www.sagpya.mecon.gov.ar/new/0-0/
nuevositio/agricultura/cultivos/granos/soja.php.Satorrre, H., 2005. Cambios tecnologicos en la agricultura Argentina actual. Ciencia
Hoy 87, 2431.Searchinger, T., Heimlich,R., Houghton,R., Dong,F., Elobeid, A., Fabiosa, J., Tokgoz,S.,
Hayes,D., Yu,T.H.,2008.Use of U.S. croplandsfor biofuels increasesgreenhousegases through emissions from land-use change. Science 29, 12381240.
Soriano, A., 1991. Ro de la Plata grasslands. In: Coupland, R. (Ed.), Natural Grass-lands: Introduction and Western Hemisphere. Elsevier, Amsterdam.
Trigo, E., 2005. Consecuencias economicas de la transformacion agrcola. CienciaHoy 87, 4651.
Vitousek,P.M., Money, H.A.,Lubchenco,J., Melillo, J.M.,1997. Humandomination ofEarths ecosystems. Science 277, 494499.
White, R.P., Murray, S., Rohweder, M., 2000. Grassland Ecosystems.Pilot Analysis ofGlobal Ecosystems. World Resources Institute, Washington, DC, USA. ,http://www.wri.org/wr2000.
E. Vega et al. / Agriculture, Ecosystems and Environment 134 (2009) 287292292
http://dx.doi.org/10.1016/j.agee.2009.07.011http://www.ecologyandsociety.org/vol13/iss2/art6/http://www.ecologyandsociety.org/vol13/iss2/art6/http://www.ecologyandsociety.org/vol13/iss2/art6/http://www.ibge.gov.br/home/estatistica/economia/pamclo/2004/pamclo2004.pdfhttp://www.ibge.gov.br/home/estatistica/economia/pamclo/2004/pamclo2004.pdfhttp://www.ibge.gov.br/home/estatistica/economia/pamclo/2004/pamclo2004.pdfhttp://www.r-project.org/http://www.sagpya.mecon.gov.ar/new/0-0/prensa/publicaciones/agricultura/Soja.pdfhttp://www.sagpya.mecon.gov.ar/new/0-0/prensa/publicaciones/agricultura/Soja.pdfhttp://www.sagpya.mecon.gov.ar/new/0-0/nuevositio/agricultura/cultivos/granos/soja.phphttp://www.sagpya.mecon.gov.ar/new/0-0/nuevositio/agricultura/cultivos/granos/soja.phphttp://www.wri.org/wr2000http://www.wri.org/wr2000http://www.wri.org/wr2000http://www.wri.org/wr2000http://www.wri.org/wr2000http://www.sagpya.mecon.gov.ar/new/0-0/nuevositio/agricultura/cultivos/granos/soja.phphttp://www.sagpya.mecon.gov.ar/new/0-0/nuevositio/agricultura/cultivos/granos/soja.phphttp://www.sagpya.mecon.gov.ar/new/0-0/prensa/publicaciones/agricultura/Soja.pdfhttp://www.sagpya.mecon.gov.ar/new/0-0/prensa/publicaciones/agricultura/Soja.pdfhttp://www.r-project.org/http://www.ibge.gov.br/home/estatistica/economia/pamclo/2004/pamclo2004.pdfhttp://www.ibge.gov.br/home/estatistica/economia/pamclo/2004/pamclo2004.pdfhttp://www.ecologyandsociety.org/vol13/iss2/art6/http://www.ecologyandsociety.org/vol13/iss2/art6/http://dx.doi.org/10.1016/j.agee.2009.07.011