LUDIMILLA PORTELA ZAMBALDI LIMA 

 
 
 
 

INFLUÊNCIA DA ESTRUTURA DE PAISAGENS 

EM PARÂMETROS DA BIODIVERSIDADE 

COM FOCO EM PEQUENOS FRAGMENTOS E 

CORREDORES DE VEGETAÇÃO NO BIOMA 

DA MATA ATLÂNTICA, MINAS GERAIS, 

BRASIL 

 
 
 
 
 
 

LAVRAS - MG 

2014 



 

 

 

 

LUDIMILLA PORTELA ZAMBALDI LIMA 

 
 
 
 
 

INFLUÊNCIA DA ESTRUTURA DE PAISAGENS EM PARÂMETROS 

DA BIODIVERSIDADE COM FOCO EM PEQUENOS FRAGMENTOS E 

CORREDORES DE VEGETAÇÃO NO BIOMA DA MATA ATLÂNTICA, 

MINAS GERAIS, BRASIL 
 

 

 

Tese apresentada à Universidade Federal de 

Lavras, como parte das exigências do 

Programa de Pós-Graduação em Ecologia 

Aplicada, área de concentração em Ecologia 

e Conservação de Recursos em Paisagens 

Fragmentadas e Agrossistemas, para 

obtenção do título de Doutor. 

 
 
 

Orientador 

Dr. Eduardo van den Berg 

 
 
 
 
 

LAVRAS - MG 

2014 

  



 

 

 

 

  

  

Lima, Ludimilla Portela Zambaldi. 

      Influência da estrutura de paisagens em parâmetros da 

biodiversidade com foco em pequenos fragmentos e corredores de 

vegetação no bioma da Mata Atlântica, Minas Gerais, Brasil / 

Ludimilla Portela Zambaldi. – Lavras : UFLA, 2014. 

                  75 p. : il. 

 

                  Tese (doutorado) – Universidade Federal de Lavras, 2014. 

                  Orientador: Eduardo van den Berg.      

                  Bibliografia. 

                  1. Fragmentação. 2. Biodiversidade. 3. Ecologia da paisagem. 4. 

Floresta Atlântica. I. Universidade Federal de Lavras. II. Título.   

                                               

                                              CDD – 574.52642 

Ficha Catalográfica Elaborada pela Coordenadoria de Produtos e 

Serviços da Biblioteca Universitária da UFLA 

 

 



 

 

 

 

LUDIMILLA PORTELA ZAMBALDI LIMA 

 
 
 
 

INFLUÊNCIA DA ESTRUTURA DE PAISAGENS EM PARÂMETROS 

DA BIODIVERSIDADE COM FOCO EM PEQUENOS FRAGMENTOS E 

CORREDORES DE VEGETAÇÃO NO BIOMA DA MATA ATLÂNTICA, 

MINAS GERAIS, BRASIL 
 

 

Tese apresentada à Universidade Federal de 

Lavras, como parte das exigências do 

Programa de Pós-Graduação em Ecologia 

Aplicada, área de concentração em Ecologia 

e Conservação de Recursos em Paisagens 

Fragmentadas e Agrossistemas, para 

obtenção do título de Doutor. 

 

APROVADA em 18 de fevereiro de 2014 

 

Dra. Gislene Carvalho de Castro      UFSJ 

Dr. Paulo Santos Pompeu  UFLA 

Dr. Marcelo Tavares de Carvalho UFLA 

Dr. Marcelo Passamani   UFLA 

Dr. Júlio Neil Cassa Louzada  UFLA 

 

Dr. Eduardo ven den Berg 

Orientador 

 

LAVRAS –MG 

2014  



 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

À minha sogra Sofia, exemplo de pessoa e 

mãe, por todas as tuas lutas, ao amor e 

carinho que eram transmitidos em tuas 

palavras e gestos. 

DEDICO 

  



 

 

 

 

AGRADECIMENTOS 

 

Agradeço a Deus, em primeiro lugar, por todas as realizações 

alcançadas. 

Ao meu orientador, Eduardo van den Berg, por sua orientação e pelas 

contribuições imprescindíveis na realização deste trabalho, por acreditar e se 

interessar pelo nosso trabalho e pela amizade e exemplo de pessoa.  

 Ao pesquisador, Warren Cohen por me receber em Corvallis e 

possibilitar amplos conhecimentos adquiridos em tão pouco tempo. Ao 

pesquisado Robert Hughes, agradeço por nos ajudar a fazer o estágio 

internacional, que me trouxe muitas alegrias e aprendizados.  

 Aos meus pais, Paulo e Magali, que sempre deram todo apoio e 

incentivo, necessários e essenciais a minha carreira acadêmica. Vocês estão 

sempre presentes na minha vida e constituem-se no meu porto seguro.  

Agradeço às minhas irmãs Grá e Fá, cunhados Helinho e Léo, família 

tão querida e pelos "piolhos" Bianca e Henrique, pelo apoio e presença em todos 

os momentos importantes. 

Ao meu amigo, companheiro, esposo e co-autor Fábio Mineo por todo 

carinho, amor e compreensão nos meus bons e maus momentos. Obrigada por 

me fazer feliz e por me sentir tão amada, todos os dias. Agradeço também à 

Sofia, Joaquim, Patrícia, Rogério, Marcos e Cíntia, que considero minha 

segunda família, formada por pessoas tão especiais.   

Agradeço ao professor e amigo Paulo Pompeu, a quem tenho uma dívida 

eterna, pelo alto-astral de todos os dias, pelos ensinamentos multidisciplinares, 

científicos ou não, pelo apoio e incentivo. Agradeço também a Thaís, pela 

amizade e carinho. 



 

Aos amigos Ruanny e Ivo, pela amizade verdadeira e pelo carinho que 

demonstram. Desejo que larguem esta vida de cigano e venham logo pra mais 

perto de nós!  

Aos amigos especiais, Grá e Giu, por tudo o que vivemos juntos e pela 

grande saudade que deixaram por aqui.  

Aos padrinhos e amigos Amanda e Henrique pela amizade e todos os 

bons momentos que passamos juntos. 

A todos os amigos da Ecologia: Vanesca, Rodrigo, Lívia, Leopoldo, 

Míriam, Lisi e Victor.  

Também agradeço aos meus amigos de Corvallis (Piero, Cassia, Cleuzir, 

Adriana, Mariana, Clarice, Rodrigo, Danielle, Thaís, Ive, Fábio, Mousa, Ali, 

Russ, Mary, Nancy, Daryl), pelos ótimos momentos de convivência e que 

fizeram dessa experiência internacional verdadeiramente inesquecível; 

A todos os professores da Ecologia, em especial ao Júlio pela 

colaboração neste trabalho. 

Aos membros da banca, Dra. Gislene Carvalho, Dr. Luis Marcelo 

Tavares de Carvalho, Dr. Marcelo Passamani e Dr. Júlio Louzada, pelas valiosas 

críticas e sugestões ao trabalho. 

Ao Programa de Pós Graduação em Ecologia Aplicada à UFLA e à 

CAPES pelo suporte durante o doutorado. 

A todos aqueles que ajudaram, diretamente ou indiretamente, na 

elaboração deste trabalho, mas que por desatenção não tiveram seus nomes aqui 

registrados. 



 

 

 

 

RESUMO GERAL 

 

A área de vegetação remanescente, a matriz, o isolamento, a 

conectividade entre as manchas florestais são fatores estruturais intimamente 

relacionados à biodiversidade. Com o avanço da fragmentação e perda de 

habitat, os pequenos fragmentos e conectores de manchas florestais são 

considerados importantes meios de manutenção de espécies. Apesar do tamanho, 

estes elementos possibilitam o aumento da conectividade da paisagem e o 

incremento da área disponível às espécies. Os remanescentes da Floresta 

Atlântica estão distribuídos em pequenos fragmentos (menores que 50 hectares), 

muitas das vezes inseridos em uma matriz antropogênica. Mesmo sob ameaça, 

este bioma possui uma elevada riqueza de espécies e endemismo. Embora os 

pequenos elementos da paisagem sejam importantes na manutenção da 

biodiversidade ainda existente, estudos geralmente focam na quantificação e 

análise de fragmentos preservados e pouco se sabe sobre a abundância e arranjo 

espacial dos demais. O objetivo deste estudo foi analisar a abundância e os 

padrões espaciais de fragmentos e corredores de valos em 49 paisagens 

fragmentadas distribuídas no bioma Floresta Atlântica, no estado de Minas 

Gerais. Avaliamos a relação entre a estrutura das paisagens e à vegetação 

remanescente, matriz, corredores, isolamento e conectividade da paisagem. Nós 

mapeamos as paisagens através da classificação de imagens multiespectrais de 

alta resolução espacial (5m) aplicando o método semi automático de 

classificação orientada a objeto. Resultados mostraram uma porcentagem 

variável de vegetação remanescente nas paisagens (de 4,1% to 69,7%), a maior 

parte distribuídos em fragmentos menores que 1 ha (de 45 a 97% do total de 

fragmentos). O maior número de corredores foi encontrado no sudeste do estado. 

Análises estatísticas baseadas no critério AICc de seleção de modelos indicaram 

influência dos fatores físicos, estruturais e de divisões políticas na quantidade de 

vegetação, isolamento e no tamanho dos corredores das paisagens. Pequenos 

fragmentos (<100 ha) e corredores de valos (largura ≤ 15 m) são importantes 

elementos na conexão entre os fragmentos. 

  



 

ABSTRACT 

 

Structural factors as like vegetation remnants, isolation and connectivity 

between forest patches are intrinsic related to biodiversity. Small patches and 

connectors are considered important in maintaining species in landscapes with 

high level of fragmentation and habitat loss. Despite the size, these element 

increase the landscape connectivity, and the available area to species. Atlantic 

Forest remnants are distributed in small fragments (less than 50 hectares), often 

inserted into an anthropogenic matrix. Even threatened , this biome presents a 

high species richness and endemism. Despite they importance, studies usually 

have focused on the analysis and quantification of preserved fragments, and 

there is a lack of information about the abundance and spatial arrangement of 

small features. The aim of this study was to quantify and analyze the spatial 

distribution of small features in 49 sample sites in Atlantic Forest fragmented 

landscapes at Minas Gerais State, Brazil. We tested the relationship between the 

remnant vegetation, isolation and hedgerows length with distance to landscape 

features, slope, altitude, number and fragments area. We also tested the 

connectivity for several capacity to cross the matrix, the relation of fragments 

with the surrounding matrix and the importance of small fragments to landscape 

isolation. We mapped the landscape features using a multispectral classification 

of high spatial resolution images (5m) applying a object-based semi automatic 

method. Results showed a variable percentage of remaining vegetation in the 

sample sites (4.1% to 69.7%), most of them with fragments smaller than 1 h 

(from 45 to 97%). The largest number of hedgerows was found in the 

southeastern state. Statistical analyzes based on the AICc model selection 

indicated the influence of physical, structural and political divisions in the 

amount of vegetation , isolation and size of corridors landscapes factors. Small 

fragments (< 100 ha) and hedgerows (width ≤ 15 m) are important elements in 

the connection between the fragments. 



 

 

 

 

LISTA DE FIGURAS 

 

ARTIGO 1   

Figura 1 Location of the study sites. The black squares indicates the 

sample sites used for the classification……………………. 

 

30 

Figura 2 Classification scheme providing an overview of the 

methodological process. The classification result consists of 

two spatial levels………………………………………….. 

 

 

32 

Figura 3  Example of a the land cover classification for a landscape. 36 

Figura 4 Percentage of remaining vegetation (left) and fragments 

smaller than 1ha (right) in landscapes analyzed at Atlantic 

Forest domain, in Minas Gerais State……………………. 

 

 

37 

Figura 5 Vegetation cover percentage of fragments lower 1ha (left) 

and above 1ha (right)………………………………………..  

 

38 

Figura 6 Distribution of landscapes by number of hedgerows and the 

area of fragments connect to hedgerows………………….    

 

39 

ARTIGO 2   

Figura 1   Location of the study sites. Atlantic Forest domain inside 

Minas Gerais (MG) state and sub-regions distributions. 

Gray squares indicates the sample sites used on this study. 

 

 

56 

Figura 2 Landscapes isolation (m) for different sub-regions resulted 

from successive removal of small fragments (ha) for all 

class of size (i) and for fragments up to 100ha (ii). Fragment 

size 0 (ha) indicates no exclusion of any fragments in the 

landscape…………………………………………………… 

 

 

 

 

63 

Figura 3  Functional distance (cluster size) according to the expect 

capacity of species to cross the matrix…………………… 

 

64 

Figura 4 Relation between remain vegetation in the landscape and 

number of fragments inserted in a temporary (left) and 

permanent (right) agricultural matrix…………………….. 

 

 

65 



 

 

LISTA DE TABELAS 

 

ARTIGO 1   

Tabela 1 Accuracy assessment from the main and lowest level…….. 

 

35 

ARTIGO 2   

Tabela 1 Fragments and Hedgerows of Atlantic Forest distribution 

over sub-regions at Minas Gerais state. Mean values are for 

normal distributions, median for non-normal. Superscript 

letters indicates the statistical difference……………………    

 

 

 

60 

Tabela 2 Model selection based on Generalized Linear Model (GLM) 

and first six AiCc-based model selected by (i) percentage of 

remain vegetation, (ii) landscape isolation, (iii) hedgerows 

length. A100 - Fragments larger than 100ha; RD - Density 

of rivers in landscape; M- sub-regions; A99 - Fragments up 

to 99 ha; AL -altitude; N -number of fragments; S - slope; D 

- distance to permanent agriculture. Signal inside 

parentheses indicate the effect of each variable……………. 

 

 

 

 

 

 

 

62 

 

  



 

 

 

 

SUMÁRIO 

 

 PRIMEIRA PARTE  

1 INTRODUÇÃO GERAL……………………………………………… 13 

2 REFERENCIAL TEÓRICO.................................................................. 16 

 REFERENCIAS……………………………………..…………………  19 

 SEGUNDA PARTE – ARTIGOS……………………………………... 25 

 ARTIGO 1  THE IMPORTANCE OF SMALL FRAGMENTS 

AND  HEDGEROWS FOR FRAGMENTED 

LANDSCAPES IN SOUTHEASTERN BRAZIL….... 

 

 

26 

 ARTIGO 2     THE RULE OF SMALL FOREST PATCHES AND 

HEDGEROWS ON BIODIVERSITY 

PARAMETER AT LANDSCAPE SCALE……..…… 

 

 

51 

 

 

 

  



 

 

 

 

 

 

PRIMEIRA PARTE 

 

  



13 

 

 

 

1 INTRODUÇÃO GERAL 

 

 Perda de habitat e fragmentação estão intimamente relacionadas à 

conservação da biodiversidade (FAHRIG, 2003; WILCOX; MURPHY, 1985); 

desta forma, área, distribuição espacial e conectividade dos fragmentos são 

considerados fatores chaves na persistência de espécies em paisagens (BEIER; 

NOSS, 1998; METZGER; DE´CAMPS, 1997). O aumento da população 

humana conjuntamente à expansão de atividades antropogênicas provocam 

amplas alterações nas paisagens como remoção de fragmentos, redução no 

tamanho e incremento do isolamento das manchas de vegetação (FAHRIG, 

2001). Os resultados são mosaicos de pequenos fragmentos (menores que 50 ha) 

inseridos em uma matriz antropizada, (FAHRIG, 2003; NEEL; MCGARIGAL; 

CUSHMAN, 2004; TABARELLI et al., 2010). 

 Apesar da influência que a matriz exerce sobre os remanescentes 

vegetacionais (UEZU; BEYER; METZGER, 2008; UMETSU; METZGER; 

PARDINI, 2008), pequenos fragmentos e corredores ecológicos são importantes 

elementos em paisagens, sendo muitas das vezes relacionados à riqueza de 

espécies (DUELLI; OBRIST, 2003) e à influência que eles exercem no grau de 

fragmentação, conectividade, migração e dispersão de espécies (BENNET et al., 

1994). Corredores de paisagens são estruturas lineares de vegetação que podem 

ser utilizados pelas espécies como habitat ou como conectores entre duas 

manchas florestais (ROSENBERG; NOON; MESLOW, 1997; TISCHENDORF, 

2001) possibilitando o uso múltiplo de fragmentos pelas espécies. Dentro da 

classe de corredores, podemos destacar os corredores de cercas e valos de divisa 

(hedgerows), gerados a partir da colonização natural de plantas em valos de três 

metros de largura, formando uma cobertura de dossel de até 15m (CASTRO; 

VAN DEN BERG, 2013). No sudeste do Brasil, estes corredores são elementos 

proeminentes na paisagem, formando habitat e conectores de fragmentos para 



14 

 

 

pequenos mamíferos (CASTRO; VAN DEN BERG, 2013; MESQUITA; 

PASSAMANI, 2012; ROCHA; PASSAMANI; LOUZADA, 2011). 

 Apesar da importância dos pequenos fragmentos e corredores, pouco se 

é conhecido sobre sua abundância, distribuição e função na paisagem 

(HARVEY et al., 2005). Informações sobre a distribuição espacial e presença de 

conectores de fragmentos em grandes áreas geográficas podem ser obtidas 

através da análise de imagens de sensoriamento remoto de alta resolução 

espacial e espectral. A classificação das imagens possibilita o acesso a 

informações estruturais sobre a vegetação e áreas do entorno, assim como área e 

isolamento de fragmentos e as relações a outros componentes da paisagem 

(NEEL; MCGARIGAL; CUSHMAN, 2004; WITH; KING, 1997). A 

quantificação da estrutura espacial da paisagem é um importante aspecto da 

ecologia da paisagem, justificado pela relação entre a estrutura da paisagem e 

processos ecológicos (NEEL; MCGARIGAL; CUSHMAN, 2004; TURNER, 

1989). 

 A Floresta Atlântica originalmente cobria 150 milhões de hectares, hoje 

distribuídos em paisagens altamente fragmentadas (RIBEIRO et al., 2009). A 

perda de habitat e fragmentação reduziram este bioma a paisagens dominadas 

por pequenos fragmentos (<100 ha) (RANTA et al., 1998) com um alto grau de 

isolamento (METZGER, 2000; RIBEIRO et al., 2009). No entanto, apesar de 

ameaçada, a Floresta Atlântica é considerada um hotspot da biodiversidade com 

elevado grau de endemismo e riqueza de espécies (MYERS et al., 2000; 

RIBEIRO et al., 2009). 

Devido à dependência de processos ecológicos à variabilidade espacial 

dos remanescentes florestais e aos demais componentes da paisagem, o presente 

estudo teve como objetivo analisar a estrutura de paisagens do Bioma Mata 

Atlântica em Minas Gerais, avaliando a relação das variáveis estruturais de 

manchas e corredores florestais com fatores que influenciam a vegetação 



15 

 

 

 

remanescente, conectividade, e isolamento das paisagens, resultando na 

identificação de padrões na relação dos componentes da paisagem a processos 

ecológicos que podem afetar a biodiversidade.    

 Os resultados deste trabalho são apresentados em dois artigos escritos na 

língua inglesa e estruturados nas normas da revista Biological Conservation. No 

primeiro artigo foi quantificado e mapeado os elementos de paisagens no estado 

de Minas Gerais, no domínio da Floresta Atlântica, analisando a distribuição dos 

pequenos fragmentos e corredores de valos. 

 No segundo artigo o remanescente de vegetação, isolamento e 

conectividade foram relacionados a fatores físicos e antropogênicos. Avaliamos 

também a importância dos menores fragmentos e da conectividade das 

paisagens. 

 



16 

 

 

2 REFERENCIAL TEÓRICO 

 

 A perda e fragmentação do habitat são consideradas as principais causas 

de redução da biodiversidade (FAHRIG, 2003; HERRMANN, 2011; 

LAURANCE, 1999). Isto é resultado da insuficiência em área disponível aos 

organismos seguidos pelo incremento no isolamento dos fragmentos e a 

consequente redução na conectividade da paisagem, inviabilizando as relações 

ecológicas entre as espécies e afetando negativamente o tamanho das populações 

(AWADE; METZGER, 2008; FAHRIG, 2003). Manchas isoladas de habitat 

podem não ser suficientes para suportar populações viáveis a longo prazo 

(FAHRIG, 2003; SOULÉ, 1987), tornando-as suscetíveis à extinção decorrente 

de fatores tais como endogamia ou flutuações ambientais (ANDERSON; 

JENKINS, 2005). Modificações e extinções de manchas florestais reduzem o 

número de imigrações de espécies que agem como vetores na realização de 

funções ecológicas como dispersão e polinização de espécies vegetais 

(BROOKER; BROOKER; CALE, 1999). 

Neste sentido, a conectividade, definida como o grau no qual uma 

paisagem facilita ou restringe o movimento de organismos, gametas e 

propágulos entre fragmentos (TAYLOR et al., 1993; URBAN; SHUGART, 

1986), é um elemento vital a ser avaliado na paisagem, pois está diretamente 

ligado à viabilidade de populações, sendo considerada crítica para a 

sobrevivência das espécies (NOSS, 1987; PRIMACK, 1993). A conectividade 

de uma paisagem pode ser mensurada de duas maneiras, pela ligação estrutural 

entre as manchas florestais ou pela conectividade funcional. A conectividade 

estrutural representa a distância entre as manchas, densidade de corredores e 

permeabilidade da matriz (ANTONGIOVANNI; METZGER, 2005; BEIER; 

NOSS, 1998) e a conectividade funcional considera as respostas 



17 

 

 

 

comportamentais das espécies aos elementos da paisagem juntamente com a 

estrutura espacial (GOBEIL; VILLARD, 2002; GOODWIN, 2003). 

Corredores de vegetação, compostos por estruturas lineares que ligam 

manchas de vegetação (FORMAN; GODRON, 1986) são elementos da 

paisagem que desempenham um papel fundamental em termos de conectividade 

(PARDINI et al., 2005). Esta interligação apresenta-se como alternativa 

importante na conservação de paisagens, permitindo o movimento de 

organismos entre manchas (FORMAN; COLLINGE, 1997) e reduzindo os 

efeitos do isolamento estrutural especialmente em paisagens dominadas por 

matrizes pouco permeáveis (BEIER; NOSS, 1998; PARDINI et al., 2005). 

Quando estes conectores estão presentes, o tempo gasto pelas espécies para 

colonizar ou recolonizar habitats de fragmentos onde se tornariam extintas pode 

ser minimizado (ANDERSON; JENKINS, 2005), resultando no incremento da 

abundância, da riqueza e diversidade alfa de espécies em pequenos fragmentos 

florestais (PARDINI et al., 2005) e podendo funcionar também como habitat 

para diferentes táxons (BENNET, 1990; DOWNES; HANDASYDE; ELGAR, 

1997). Entretanto, alguns autores citam funções negativas de corredores como: 

facilitação de propagação de distúrbios, aumento à exposição de predadores, 

chegada de espécies exóticas (HOBBS, 1992; LIDICKER, 1999) e propagação 

de doenças (ANDERSON; JENKINS, 2005).  

No estudo do valor biológico dos corredores e manchas de vegetação, é 

imprescindível a avaliação dos aspectos da paisagem (MACDONALD; 

RUSHTON, 2003; SIMBERLOFF; COX, 1987) sendo as imagens de 

sensoriamento consideradas as primeiras fontes de informação para tal fim 

(HERRMANN, 2011). A dependência da funcionalidade de paisagens e 

corredores de vegetação à sua estrutura, escala e contexto (SAUNDERS; 

HOBBS, 1991) indicam a relevância do uso de ferramentas de sensoriamento 

remoto, podendo resultar em um indicativo do movimento e migração de 



18 

 

 

organismos (HERRMANN, 2011; SAUNDERS; HOBBS, 1991). Apesar de não 

oferecerem dados para testar diretamente a hipótese de corredores de vegetação 

como corredores ecológicos, o sensoriamento remoto é uma poderosa ferramenta 

para a efetiva representação de corredores na paisagem e sua interpretação no 

contexto da conectividade (SAUNDERS; HOBBS, 1991), pois a maior parte dos 

processos ecológicos são inerentemente espaciais, interagindo em unidades 

vizinhas (VENEMA; CALAMAI; FIEGUTH, 2005; WAGNER; FORTIN, 

2005), conectando padrões espaciais à biodiversidade (DIAMOND, 1975).  

Os padrões da estrutura da paisagem podem ser analisados através da 

aplicação de índices (TISCHENDORF, 2001; TURNER, 1989) onde, além do 

contexto da localização dos remanescentes, métricas sobre a inferência de 

persistência de espécies podem ser avaliadas, como o tamanho, forma e o grau 

de isolamento das manchas florestais (HERRMANN, 2011; SAUNDERS; 

HOBBS, 1991).  

A área disponível para a colonização pode ser um preditor à persistência 

de espécies, desde que áreas grandes e conectadas provavelmente manterão 

grandes populações (CARVALHO; JÚNIOR; FERREIRA, 2009).   

A Floresta Atlântica brasileira, considerada um hotspot de 

biodiversidade (MITTERMEIER et al., 1998), é um dos biomas mais 

perturbados do Brasil, apresentando o maior número de espécies ameaçadas por 

unidade de área (FONSECA et al., 1994; MYERS et al., 2000). A degradação 

deste bioma já atingiu níveis extremos, restringindo os remanescente a pequenos 

fragmentos, localizados, em sua maioria próximos a áreas abertas (RIBEIRO et 

al., 2009). O histórico de devastação desde a época do descobrimento 

condicionou a maior concentração dos remanescentes florestais em áreas onde o 

terreno dificulta a ocupação humana (RANTA et al., 1998; RESENDE; LANI; 

REZENDE, 2002). 

   



19 

 

 

 

REFERÊNCIAS 

 

ANDERSON, A. B.; JENKINS, C. Applying nature’s design: corridors as a 

strategy for biodiversity conservation. New York: Columbia University Press, 

2005. 256    

 

 

ANTONGIOVANNI, M.; METZGER, J. P. Influence of matrix habitats on the 

occurrence of insectivorous bird species in Amazonian forest fragments. 

Biological Conservation, Essex, v. 122, p. 441–451, Apr. 2005.    

 

 

AWADE, M.; METZGER, J. P. Using gap-crossing capacity to evaluate 

functional connectivity of two Atlantic rainforest birds and their response to 

fragmentation. Austral Ecology, Carlton, v. 33, n. 7, p. 863-871, Nov. 2008.    

 

 

BEIER, P.; NOSS, R. F. Do habitat corridors provide connectivity? 

Conservation Biology, Malden, v. 12, n. 6, p. 1241–1252, Dec. 1998.    

 

 

BENNET, A. F. Habitat corridors and the conservation of small mammals in a 

fragmented forest environment. Landscape Ecology, Dordrecht, v. 4, p. 109-

122,  1990.    

 

 

BROOKER, L.; BROOKER, M.; CALE, P. Animal dispersal in fragmented 

habitat: measuring habitat connectivity, corridor use, and dispersal mortality. 

Conservation Ecology, Boston, v. 3, n. 1, p. 4, Jun. 1999.    

 

 

CARVALHO, F. M. V.; JÚNIOR, P. D. M.; FERREIRA, L. G. The Cerrado 

into-pieces: Habitat fragmentation as a function of landscape use in the savannas 

of central Brazil. Biological conservation, Essex, v. 142, n. 7, p. 1392–1403, 

Jul. 2009.    

 

 

CASTRO, G. C. D.; VAN DEN BERG, E. Structure and conservation value of 

high-diversity hedgerows in southeastern Brazil. Biodiversity and 

Conservation, London, v. 22, n. 9, p. 2041–2056, Aug. 2013.    

 



20 

 

 

DIAMOND, J. The island dilemma: Lessons of modern biogeographic studies 

for the design of natural reserves. Biological Conservation, Essex, v. 7, n. 2, p. 

128-146, Feb. 1975.    

 

 

DOWNES, S. J.; HANDASYDE, K. A.; ELGAR, M. A. The use of corridors by 

mammals in fragmented Australian forests. Conservation Biology, Boston, v. 

11, n. 3, p. 718-726, Jun. 1997.    

 

 

DUELLI, P.; OBRIST, M. K. Regional biodiversity in an agricultural landscape: 

the contribution of seminatural habitat islands. Basic and Applied Ecology, 

Jena, v. 4, n. 2, p. 129–138, Mar. 2003.    

 

 

FAHRIG, L. How much habitat is enough? Biological conservation, Essex, v. 

100, n. 1, p. 65-74, Jul. 2001.    

 

 

______. Effects of habitat fragmentation on biodiversity. Annual Review of 

Ecology, Evolution and Systematics, Palo Alto, v. 34, n. 1, p. 487–515, Nov. 

2003.    

 

 

FONSECA, G. A. B.  et al. Livro vermelho dos mamíferos brasileiros 

ameaçados de extinção. Belo Horizonte: Fundação Biodiversitas, 1994. 459    

 

 

FORMAN, R. T. T.; COLLINGE, S. K. Nature conserved in changing 

landscapes with and without spatial planning. Landscape and Urban Planning, 

Amsterdam, v. 37, n. 1, p. 129-135, Jun. 1997.    

 

 

FORMAN, R. T. T.; GODRON, M. Landscape Ecology. New York: John 

Wiley ans Sons, 1986. 619    

 

 

GOBEIL, J. F.; VILLARD, M. A. Permeability of three boreal forest landscape 

types to bird movements as determined from experimental translocations. Oikos, 

Buenos Aires, v. 98, n. 3, p. 447–458, Sep. 2002.    

 



21 

 

 

 

GOODWIN, B. J. Is landscape connectivity a dependent or independent 

variable? Landscape Ecology, Dordrecht, v. 18, n. 5, p. 687-699, Jul. 2003.    

 

 

HARVEY, C. A.  et al. Contribution of live fences to the ecological integrity of 

agricultural landscapes. Agriculture, Ecosystems and Environment, 

Amsterdam, v. 111, n. 1, p. 200–230, Dec. 2005.    

 

 

HERRMANN, G. Incorporando a teoria ao planejamento regional da 

conservação: A experiência do corredor ecológico da Mantiqueira. Belo 

Horizonte: Valor Natural, 2011.     

 

 

HOBBS, R. J. The role of corridors in conservation: solution or badwagon? . 

Trends Ecology Evolutions, v. 7, n. 11, p. 389-392, Jun. 1992.    

 

 

LAURANCE, W. F. Reflections on the tropical deforestation crisis. Biological 

Conservation, Essex, v. 91, n. 2, p. 109-117, Sep. 1999.    

 

 

LIDICKER, W. Z. Responses of mammals to habitat edges: an overview. 

Landscape Ecology, Dordrecht, v. 14, n. 4, p. 333-342, Aug. 1999.    

 

 

MACDONALD, D. W.; RUSHTON, S. Modelling space use and dispersal of 

mammals in real landscapes: a tool for conservation. Journal of Biogeography, 

Oxford, v. 30, n. 4, p. 607-620, Apr. 2003.    

 

 

MESQUITA, A. O.; PASSAMANI, M. Composition and abundance of small 

mammal communities in forest fragments and vegetation corridors in Southern 

Minas Gerais, Brazil. Revista de Biologia Tropical, San Jose, v. 60, n. 4, p. 

1335–1343, Jul. 2012.    

 

 

METZGER, J. P. Tree functional group richness and landscape structure in a 

Brazilian tropial fragmented landscape. Ecological Applications, Tempe, v. 10, 

n. 4, p. 1147–1161, Aug. 2000.    

 



22 

 

 

METZGER, J. P.; DE´CAMPS, H. The structural connectivity threshold: an 

hypothesis in conservation biology at the landscape scale. Acta Oecologica, 

Paris, v. 18, n. 1, p. 1-12, Apr. 1997.    

 

 

MITTERMEIER, R. A.  et al. Biodiversity hotspots and major tropical 

wilderness areas: approaches to setting conservation priorities. Conservation 

Biology, Boston, v. 12, n. 3, p. 516-520, Jun. 1998.    

 

 

MYERS, N.  et al. Biodiversity hotspots for conservation priorities. Nature, 

London, v. 403, n. 4, p. 853–858, Apr. 2000.    

 

 

NEEL, M. C.; MCGARIGAL, K.; CUSHMAN, S. A. Behavior of class-level 

landscape metrics across gradients of class aggregation and area. Landscape 

Ecology, Dordrecht, v. 19, n. 6, p. 435–455, Aug. 2004.    

 

 

NOSS, R. F. Corridors in real landscapes: a reply to Simberloff and Cox. 

Conservation Biology, Boston, v. 1, n. 2, p. 159-164, Aug. 1987.    

 

 

PARDINI, R.  et al. The role of forest structure, fragment size and corridors in 

maintaining small mammal abundance and diversity in an Atlantic forest 

landscape. Biological Conservation, Essex, v. 124, n. 2, p. 253–266, Jul. 2005.    

 

 

PRIMACK, R. B. Introduction to conservation biology. Sunderland: Sinauer 

Associates, 1993. 661    

 

 

RANTA, P.  et al. The fragmentation Atlantic rain forest of Brazil: size, shape 

and distribution of forest fragments. Biodiversity and Conservation, London, 

v. 7, n. 6, p. 385-403, Jun. 1998.    

 

 

RESENDE, M.; LANI, J. L.; REZENDE, S. B. Pedossistemas  da  Mata  

Atlântica:  Considerações  pertinentes sobre a sustentabilidade. Revista Árvore, 

Vicosa, v. 26, n. 3, p. 261-269, May. 2002.    

 



23 

 

 

 

RIBEIRO, M. C.  et al. The Brazilian Atlantic Forest: How much is left, and 

how is the remaining forest distributed? Implications for conservation. 

Biological conservation, Essex, v. 142, p. 1141-1153, Dec. 2009.    

 

 

ROCHA, M. F.; PASSAMANI, M.; LOUZADA, J. E. A Small Mammal 

Community in a forest fragment, vegetation corridor and coffee matrix system in 

the Brazilian Atlantic forest. PLoS One, San Francisco, v. 6, n. 8, p. 1-8, Feb. 

2011.    

 

 

ROSENBERG, D. K.; NOON, B. R.; MESLOW, E. C. Biological corridors: 

form, function and, efficacy. BioScience, Washington, v. 47, n. 3, p. 677-687, 

May. 1997.    

 

 

SAUNDERS, D. A.; HOBBS, R. J. Nature Conservation 2: The Role of 

Corridors. Chipping Norton: Surrey Beatty & Sons, 1991. 442    

 

 

SIMBERLOFF, D.; COX, J. Consequences and costs of conservation corridors. 

Conservation Biology, Boston, v. 1, n. 1, p. 63-71, May 1987.    

 

 

SOULÉ, M. E. Viable populations for conservation. Cambridge: Cambridge 

Academic Press, 1987. 189    

 

 

TABARELLI, M.  et al. Prospects for biodiversity conservation in the Atlantic 

Forest: Lessons from aging human modified landscapes. Biological 

conservation, Essex, v. 143, n. 10, p. 2328–2340, Oct. 2010.    

 

 

TAYLOR, P. D.  et al. Connectivity is a vital element of landscape structure. 

Oikos, Buenos Aires, v. 68, n. 3, p. 571-573, Mar. 1993.    

 

 

TISCHENDORF, L. Can landscape indices predict ecological processes 

consistently? Landscape Ecology, Dordrecht, v. 16, n. 6, p. 235-254, Aug. 

2001.    

 



24 

 

 

TURNER, M. G. Landscape ecology: the effect of pattern on process. Annual 

Review of Ecology, Evolution and Systematics, Palo Alto, v. 20, n. 1, p. 171-

197, Nov. 1989.    

 

 

UEZU, A.; BEYER, D. D.; METZGER, J. P. Can agroforest woodlots work as 

stepping stones for birds in the 

Atlantic Forest region? . Biodiversity and Conservation, London, v. 17, n. 12, 

p. 1907-1922, Nov. 2008.    

 

 

UMETSU, F.; METZGER, J. P.; PARDINI, R. The importance of estimating 

matrix quality for modeling species distribution in complex tropical landscape: a 

test with Atlantic forest small mammals. Ecography, Copenhagen, v. 31, n. 3, p. 

359-370, Jun. 2008.    

 

 

URBAN, D. L.; SHUGART, H. H. J. Avian demography in mosaic landscapes: 

modeling paradigm and preliminary. In: VERNER, M. L.;MORRISON, M. L., 

et al (Ed.). Wildlife 2000. Modeling Habitat Relationships of Terrestrial 

Vertebrates. Madison: The University of Wisconsin Press, 1986.  p.273-279.   

 

 

VENEMA, H. D.; CALAMAI, P. H.; FIEGUTH, P. Forest structure 

optimization using evolutionary programming and landscape ecology metrics. 

European Journal of Operational Research, Amsterdam, v. 164, n. 2, p. 423–

439, Jul. 2005.    

 

 

WAGNER, H. H.; FORTIN, M. J. Spatial analysis of landscapes: concepts and 

statistics. Ecology, Tempe, v. 86, n. 8, p. 1975-1987, Sep. 2005.    

 

 

WILCOX, B. A.; MURPHY, D. D. Conservation strategy: the effects of 

fragmentation on extinction. American Naturalist, Chicago, v. 125, n. 3, p. 

879-887, Jun. 1985.    

 

 

WITH, K. A.; KING, A. W. The use and misuse of neutral landscape models in 

ecology. Oikos, Buenos Aires, v. 79, n. 2, p. 219-229, Sep. 1997.    

 

 



25 

 

 

 

 

 

 

 

 

 

SEGUNDA PARTE - ARTIGOS 

  



26 

 

 

ARTIGO 1 

 

 

THE IMPORTANCE OF SMALL FRAGMENTS AND HEDGEROWS 

FOR FRAGMENTED LANDSCAPES IN SOUTHEASTERN BRAZIL 

 

Artigo estruturado nas normas da revista “Biological conservation” 

  



27 

 

 

 

Quantification and qualification of small fragments and hedgerows in 

Southeastern Brazil  

 

Ludimilla Zambaldi*
a
, Eduardo van den Berg

 a  

 

a
Biology Department, Federal University of Lavras, Minas Gerais State, Brazil.  

* Correspondending author. Tel: +55 359160 3375. E-mail adress: 

ludzambaldi@hotmail.com 

 

 

Keywords: fragmentation; biodiversity; landscape ecology; Atlantic Forest 

Abstract 

 

 The high level of fragmented landscapes emphasizes the value of small 

fragments and connectors to biodiversity conservation. Despite the size, these 

elements have proved to play important roles in the conservation of biodiversity 

by enhancing landscape connectivity. Although reduced to less than 12% of its 

original extension, the Atlantic Forest still has a high species richness and 

endemism, currently distributed in landscapes dominated by small fragments 

(<50ha). However, studies have focused on quantifying the preserved fragments 

and little is known about small fragments and corridors abundance and their 

spatial arrangement. Therefore, the objective of this study was to characterize 

the abundance and spatial patterns of small fragments and hedgerows in 

fragmented landscapes distributed in Atlantic Forest domain in Minas Gerais 

State. Semi automated hierarchical classification rules were established using an 

object-based classification of multispectral RapidEye images, implemented in 49 

landscapes, mapping all sizes of fragments, hedgerows and agricultural areas, 

with a high level of accuracy. The results showed a variable percentage of 

remaining vegetation (from 4.1% to 69.7%) distributed mainly  fragments 

smaller than 1ha (from 45 to 97% of total fragments). High density of 

hedgerows was found in the south of  Minas Gerais State, and the hedgerows 

connections to one or more fragments have different distributions in the 

landscapes. This paper classified small and linear vegetation features in 

landscapes, which allows for an understanding of the appropriate spatial 

resolution and methods required to extract these patches when mapping using 

remote sensing imagery. In the present scenario of fragmentation of the Atlantic 

Forest, the quantification and spatial distribution of small and linear fragments 

are essential for the study and management for species conservation.  

 

1. Introduction 

 Habitat loss and fragmentation are main concerns to biodiversity 

conservation (Fahrig 2003; Wilcox and Murphy 1985) causing, among other 



28 

 

 

things, a reduction on habitat amount and an increase on isolation and number of 

patches with small area (Carvalho et al. 2009; Fahrig 2003). Under this scenario, 

remnant size and structural connectivity are considered key factors on species 

persistence (Beier and Noss 1998; Fahrig and Merriam 1985, 1994; Metzger 

2000; Metzger and De´camps 1997). Small remnants and hedgerows have an 

ecological value that is proportionally greater than their real extension (Hou and 

Walz 2013). Several studies imply in a close relation between small-scale 

landscape structures and species richness, e.g. birds and arthropods (Duelli and 

Obrist 2003; Hou and Walz 2013), explained by the presence or absence of 

small remnants and linear vegetation patches and their influence on degree of 

fragmentation, connectivity, species migration and dispersal (Bennet 1990). 

Habitat patches connectivity is thought to be important for movement of genes, 

individuals, populations, and species over multiple scales (Minor and Urban 

2007) 

 Spread across landscapes around the world, small fragments and 

hedgerows are also important features in the Brazilian Atlantic Forest biomes, 

one of the largest rainforest biome of the New World. Originally, this biome 

covered around 150 million hectares, in highly heterogeneous environmental 

conditions (Ribeiro et al. 2009). The Atlantic Forest extension was extremely 

reduced, the estimates vary from 11 to 16% (Ribeiro et al. 2009) , 7 to 8% 

according to SOS Mata Atlântica/INPE (1993, 2000) and Galindo-Leal and 

Câmara, (2003b) and 10.6% according to SOS Mata Atlântica/INPE (2008);  

more than 80% of the fragments with areas below 50 ha (Metzger et al. 2009). 

This biome  is considered a hotspot for biodiversity conservation, due to its 

species richness (both plant and animal), high level of endemism (Myers et al. 

2000) and for being  probably one of the most highly threatened tropical forests 

in the world (Metzger et al. 2009).  

 Ecological corridors are linear features in landscapes working as habitats 

or as connectors between patches (Baudry et al. 2000; Forman and Baudry 1984; 

McCollin et al. 2000; Metzger and De´camps 1997; Pardini et al. 2005). As 

ecological corridors, hedgerows can play an important role for the conservation 

of flora without negatively impacting agriculturally landscapes. In the 

Southeastern Brazil, hedgerows can be originated from natural colonization of 

land plot boundary ditches are prominent features of the landscape (Castro and 

van den Berg 2013). These hedgerows hold a high diversity of plant species 

inside a three meters wide ditches (Castro and van den Berg 2013), creating a 

maximum 15 meters of canopy cover, working as well as a fragments connectors 

or habitat for mammals (Castro and van den Berg 2013; Mesquita and 

Passamani 2012; Rocha et al. 2011). Others hedgerows include strip 

vegetation on fences. 

 Although small and linear patches play an important role in landscapes 

structure and ecological process, remarkably little information is available about 



29 

 

 

 

their abundance, distribution and function (Harvey et al. 2005; Hou and Walz 

2013; León and Harvey 2006).  The assessment of landscape features over large 

geographic regions is possible by mapping these structures using passive sensors 

with high spatial and multispectral resolution (Goossens et al. 1991; Vogt et al. 

2007). Inclusion of small features in mapping remnants it is limited by the 

images spatial resolution and classification methods. Satellite images with less 

than 10 meters of spatial resolution are computationally efficient, reliable, and 

valid for detecting small landscape features over large areas (Vannier and 

Hubert-Moy 2008). Small features classifications in large areas require an 

efficient classification methodology to enable the different size and 

characteristics of landscape elements.  

 Object-based methods for image analysis have the advantage of 

incorporating spatial context and mutual relationships between objects 

(Concheddaa et al. 2008). It is possible to classify objects using information 

about each individual object and also about the relations existing between the 

objects (Lewinski and Zaremski 2004) enabling to define corridors in terms of a 

threshold patch width and local context (Metzger and De´camps 1997; Vogt et 

al. 2007). 

 Quantifying spatial landscape structure remnants an important aspect of 

landscape ecology justified by the fundamental reciprocal relationships between 

landscape structure and ecological processes (Neel et al. 2004; Turner 1989). To 

analyze landscape elements with different sizes and contexts, a multi-scale 

strategy was applied to detect different habitat types and quantify the abundance 

and spatial arrangement of small fragments and hedgerows. Presumably 

abundant in the southeast of Brazil (Castro and van den Berg 2013), hedgerows’ 

distribution and quantification was for the first time analyzed in this study. We 

also mapped features (soil use, e.g. agriculture, pastures and urban occupation, 

and hydrography)  associate with agriculture areas, urban areas and water.  

 

2. Methodology 

 

2.1 Study Area 

 Our survey consisted of 49 12×12 km sample square areas randomly 

distributed over the Atlantic Forest domain in Minas Gerais State, Southeast 

Brazil (Figure 1). The sampled area covered 3% of the Atlantic Forest domain in 

the State. Our random selection of sites obeyed the following restrictions: (1) the 

areas had to be completely included in the domain (not overlapping its edges) 

and (2) no square areas could share boundaries. For every sample site, we used 

high spatial resolution image extracts composed of RapidEye images acquired in 

2011 with spatial resolution of 5 m. This resolution enabled the inclusion a large 



30 

 

 

range of fragment sizes and connectivity conditions, required for detecting 

hedgerows.  

 

 
Figure 1. Location of the study sites. The black squares indicates the sample 

sites used for the classification.  

 

 We did not differentiate fragments composed by vegetation in 

secondary, intermediate or advanced stages of succession. The distinction 

between old growth and secondary forest is particularly difficult for the entire 

Atlantic Forest region because information about forest age is very scarce and 

available only at local scales (Ribeiro et al. 2009). The sampling design covered 

the different kinds of vegetation included in the Atlantic Forest domain and the 

different kinds of human pressure.  

  

2.2 RapidEye Acquisition and processing 

 We used multi-spectral RapidEye images with five spectral bands: Blue 

(440-510nm), Green (520-590nm), Red (630-685nm), Red Edge (690-730nm) 

and Near Infra Red (760-850nm) to map land cover and hedgerows. 

Orthorectified and atmospherically corrected images were obtained through a 



31 

 

 

 

partnership between the Federal University of Lavras (UFLA) and Forest 

Federal Institute of Minas Gerais (IEF). Acquisition errors, clouds and shadows 

were removed in the pre-processing phase (Coppin et al. 2004), which also 

included visual evaluation of image registration.   

 

2.3 Methods  

 We classified the landscape elements obtained from satellite images 

using an object-based approach using multi-scale image segmentation (Figure 

2). Image segmentation is the process of partitioning an image into groups of 

pixels that are spectrally similar and spatially adjacent (Desclée et al. 2006; 

Duveiller et al. 2008). Boundaries among these pixel groups delineate ground 

objects in a similar way a human analyst would do based on their shape, tone 

and texture (Duveiller et al. 2008). Multi-scale segmentation (MSS) has been 

introduced by (Baatz et al. 2000) and allows the extraction of image objects at 

different resolutions to construct a hierarchical network of image objects in 

which each object has information about its context, its neighborhood and its 

sub-objects (Benz et al. 2004). 

 



32 

 

 

Figure 2. Classification scheme providing an overview of the methodological 

process. The classification result consists of two spatial levels. 

   

 We segmented the images into two levels using a multi-scale image 

segmentation algorithm (eCognition software), applied to all RapidEye image 

bands using equal  weights  for  all  bands. The segmentation of the images is 

influenced by three parameters: 1) the global size of desired areas also called 

scale; 2) their homogeneity in terms of color 3) a “shape” parameter that is 

related with smoothness and compactness (Broich et al. 2009). We determined 

these parameters using a systematic trial and error approach validated by the 

visual inspection of the quality of the output image objects (Anders et al. 2011; 

Dragut et al. 2010; Mathieu et al. 2007). 

 Before an appropriate scale factor was identified, the shape and color 

criterion were modified to refine the shape of the image objects (Mathieu et al. 

2007) (Mathieu et al. 2007)(Mathieu et al. 2007). For both levels, a weight of 

0.4 was assigned to the color parameter, 0.3 to the shape parameter, and 0.6 to 

the compactness parameter the color was assigned a weight of 0.4, whereas the 

shape received the remaining weight of 0.3 (compactness 0.6). We chose the 



33 

 

 

 

scale factor, determining the size of the objects, in such a way that the edges of 

the delineated areas would correspond with the feature patterns (classes of land 

cover) visible in the image (Lewinski and Zaremski 2004). A first level of 

segmentation was produced with object sizes ranging from 1.01 ha to the largest 

object in the image. A second level of segmentation was computed to produce 

finer objects ranging in size from 0.025 ha to 1 ha. The first level was used to 

stratify the larger patches, using a scale factor of 70 and a second, more detailed 

level, was created to map smaller patches with a scale factor of 40.  

 Once a successfully segmented image was obtained, we applied an 

object-based using Nearest Neighborhood (NN), trained by image samples, to 

the segmentation image in order to assign a class label to each segment. The NN 

classifier allows quick and straightforward classification and can use a variety of 

variables related to spectral, textural,  shape and/or contextual properties of the 

image objects (Mathieu et al. 2007) . On a higher hierarchical level, defined as 

the main level, land cover classification was based on object samples using 

Nearest Neighborhood (NN) classification identifying major land-use types 

(forest, permanent agriculture, temporary agriculture, water body, urban areas 

and others). We based our NN supervised classification on a training dataset 

comprised of 50 visually independent objects in each land cover class, in each 

scene. These training sites were based on RGB (543) composite and were 

selected using published data and field knowledge. At the lowest hierarchical 

level, vegetation patches was further divided into the classes hedgerows or 

fragments, using threshold conditions (Figure 1). Ancillary data included rivers 

and roads maps  (IBGE – Instituto Brasileiro de Geografia e Estatística 2004), 

forest inventory data (Scolforo and Carvalho 2007), GIS topomaps and the 

Digital Elevation Model (DEM). 

 Hedgerows have similar spectral characteristics of forest patches. 

Nevertheless, the hedgerows are long and narrow with an almost constant width, 

since the dimension for the man-made ditches (3 m wide) which originated them 

are very similar between the areas and also mostly invariable (Castro and van 

den Berg 2013). We applied a merge process using a threshold condition to only 

merge the objects with width bigger than 15 m. The objects not merged were 

then hedgerows and vegetation objects inside fragments. Therefore, we applied 

the process to find the objects that are surrounded by vegetation and classified as 

vegetation. Finally, remain objects were classified as hedgerows. This rule 

enable us to identify not only the isolated hedgerows but also the ones connected 

to the vegetation patches. For each mapped hedgerows, a buffer of 5m was 

created identifying if they had one, two or neither extremities linked to a 

fragment. The beginning and end tips of individual hedgerows were defined as 

where the corridor crossed with another corridor, or where the corridor joined 

another habitat (forest patch, agriculture area, or other land use) or landscape 

feature (road or river). 



34 

 

 

 A few wrongly-classified image objects were reassigned manually to the 

correct classes based on knowledge and the RapidEye image. 

 

2.4 Validation 

 Independent data source, randomly located within each class and 

equitably distributed over the 49 scenes was used as reference for the accuracy 

assessment. We used 14083 objects and points stratified according to the size of 

the area covered by each class (Table 1).  

 Descriptive statistics of user’s, producer’s and overall accuracy (Table 

1) were computed and analyzed. The overall accuracy is computed by dividing 

the total correct by the total number of pixels in the error matrix (Congalton 

1991). The overall kappa coefficient represents a measure of agreement between 

the classes represented in the image and the true reality on the ground for the 

whole map, estimating what level of agreement is due to chance (Concheddaa et 

al. 2008). 

 Another validation method consisted in overlap each object used as 

reference to the accuracy assessment to the corresponding object classified 

(Benz et al. 2004). If the complete reference polygon is covered by 

automatically achieved segments, a highest score of 100% are given. For the 

objects that are not completely covered, the percentage was based on the cover 

percentage. We also used visually inspection comparing the reference objects to 

high resolution images available on the web (2006 Google EarthTM). Google 

EarthTM combines different resolution images and updates them on a rolling 

basis (Concheddaa et al. 2008).  

 

  



 

 

 

 

Table1. Accuracy assessment from the main and lowest level. 

Validation indices Fragments > 1ha Fragments ≤ 1ha Water 
Urban 

areas 

Permanent 

Agriculture 

Temporary  

Agriculture 
Hedgerows Others 

Prod. Accuracy 

(%) 

Min. 69 78 80 79 81 68 85 68 

Med. 78 80 85 84 85 70 87 70 

Max. 83 81 88 84 87 71 89 75 

User's accuracy 

(%) 

Min. 75 69 75 69 78 65 81 69 

Med. 77 70 77 69 79 68 83 74 

Max. 78 73 78 70 81 70 85 77 

Overall 

accuracy (%) 

Min. 87 84 78 81 81 67 82 77 

Med. 89 86 80 82 84 70 85 79 

Max. 92 89 82 84 85 72 88 80 

Kappa 

Min. 0.83 0.81 0.75 0.79 0.75 0.63 0.81 0.75 

Med. 0.86 0.83 0.77 0.8 0.77 0.7 0.84 0.76 

Max. 0.89 0.85 0.8 0.82 0.79 0.75 0.86 0.78 

Object accuracy 

(%) 

Min. 81 83 86 81 75 69 82 77 

Med. 84 85 88 84 78 71 84 78 

Max. 87 87 92 87 80 72 85 81 

Reference totals  2450 4900 245 108 1410 1110 1410 2450 

 

 

 

 

 

 

 

 

 

 



36 

 

 

3. Results  

 

3.1 Validation 

 All kappa measures, overall accuracy and object validation showed a 

high level of agreement and confirmed the good accuracy of our classification 

(Table 1). The overall accuracy higher than 80% and the kappa coefficient larger 

than 0.76 for vegetation and hedgerows are considered robust results (Bock et al. 

2005; Fielding and Bell 1997). Data used to validate the results corresponded to 

3.8% of total area, which is above the one percent generally recommended 

(Congalton 1991; Mathieu et al. 2007). Fragments larger than 100 ha present the 

most accurate results, all above 77%. Temporary agriculture present the worst 

values (65%), which suggests that the methodology used in this study was not 

completely efficient to differentiate the agriculture types. Accurate results have 

been obtained in mapping the class hedgerows thanks to the integration to the 

contextual information. 

 The object-based method achieved satisfactory results for mapping land 

cover classes in the study area, identifying the main elements in each landscape, 

including small and linear features, as well as secondary and disturbed 

vegetation (Figure 3).   

    

 
Figure 3. Example of a the land cover classification for a landscape. 



37 

 

 

Spatial Structure 

  

3.2 Forest patches abundance and distribution  

 The sample sites cover a total of 705,600 ha, with 211,778 ha (30%) of 

forest vegetation in fragments and 992 ha (3%) of forest located in hedgerows.  

The other features comprise 12,984 ha of temporary agriculture, 23,566 ha of 

permanent agriculture, 1,733 ha of urban areas, 2,033 of areas covered by water.  

 The largest fragment mapped reached 9,071 ha, inserted in the best-

preserved landscape analyzed (69.7% of remaining forest cover) located in the 

north of the MG state (Figure 4). Just one landscape presented less than 15%, 

with 4.1% of the original vegetation cover (Figure 4). 

 

Figure 4. Percentage of remaining vegetation (left) and fragments smaller than 

1ha (right) in landscapes analyzed at Atlantic Forest domain, in Minas Gerais 

State 

 

 The Atlantic Forest in the analyzed landscapes is distributed in 93,479 

fragments with their size ranging from 0.005 ha to 9,071 ha. Fragments below 

1ha represent the large majority (80%) (Figure 4), with just one landscape ninth 

less than 45% of fragments in this situation. 



38 

 

 

 The fragments equal or below 1ha correspond to 0.5 to 3% of the 

landscape area analyzed (Figure 5), fragments larger than 1ha covered areas 

between 2.6 and 70% of the landscape, depending on the specific region. We can 

found a huge variation in the area of forest patches larger than 1ha, the largest 

one has 9,071.00 ha, followed by one of 6,606.00 ha.  

 

Figure 5. Vegetation cover percentage of fragments lower 1ha (left) and above 

1ha (right).  

 

 Only two sample sites didn't have any hedgerow. We found 3347 

hedgerows distributed over 47 landscapes. Hedgerows occurred in higher 

density in the south of the State (Figure 6), showing landscapes with more than 

250 hedgerows distributed. Of the 3561 hedgerows mapped, 1547 connected at 

least two fragments, 1420 are linked to only a single fragment and 594 of the 

hedgerows were isolated, completely surrounded by anthropogenic matrix. We 

founded a variable classes of fragments size linked to hedgerows (Figure 6), 

they enhance the extension and/or the connectivity of fragments. The hedgerows 

are linked to a variety of classes of fragments size The hedgerows connected to 

at least one fragment, increase the size and extension of the fragments and when 

they are link at least two fragments, they also enable the connection between two 

fragments (Figure 6), increasing the available vegetation areas to organisms. 

 



39 

 

 

Figure 6. Distribution of landscapes by number of hedgerows and the area of 

fragments connect to hedgerows    

 

4. Discussion  

 

4.1 Classification of small fragments and hedgerows 

 Landscapes are widely recognized as complex systems having a 

hierarchical structure where dominant patterns and processes exist at specific 

scales (Meentemeyer 1989; O´Neill 1988; Wu and Marceau 2002). Forest 

fragmentation, as a huge environmental problem, cannot be handled at a single 

scale of observation (Silván-Cárdenas et al. 2009). The hierarchical approaches 

employed in the current study allowed an accurate classification for different 

type and size of habitats. This technique enabled the detection of different 

spatial scales structures on landscapes (Blaschke 2010; Hou and Walz 2013) 

particularly for the studied landscapes where the objects presented a large range 

of variation. Using this technique we were able to distinguish small objects from 

large ones and evaluate their intrinsic traits and patterns.   

 Because of the images with broad spatial resolution like Landsat (30m) 

are not adapted for mapping hedgerows (Vannier and Hubert-Moy 2008) and 

small fragments, high spatial resolution images (5m) used in this study were 

crucial for the interpretation of different features with variable size. If high-

resolution data is used, feature boundaries are more accurately mapped (Anders 

et al. 2011). Despite the very high spatial resolution, sensors with low spectral 



40 

 

 

resolution are not suitable to extract small and linear features (Vannier and 

Hubert-Moy 2008). For that matter, RapidEye images, thanks to their rich 

spectral information including a Red Edge band, allowed a precise classification 

of small and linear features. The spectral and spatial resolution of RapidEye data 

was appropriate for this study and the accuracy assessment showed that 

satisfactory results can be achieved. 

 The use of object-oriented and contextual rules in mapping linear 

landscape features was fundamental (Goossens et al. 1991) and need to be 

differentiate from remnants, composed by the same elements. Compared to 

traditional habitat mapping techniques (air photo interpretation, field study), 

object oriented methods provide accurate results whilst providing at the same 

time necessary spatial detail (Bock et al. 2005) and the possibility of mapping 

extensive areas. In the object-based approach, the segmentation process ensure 

the quality of the multispectral data to be submitted to the next step and the 

classification process offers the capacity of handling higher level of data 

heterogeneity and more complex spatial patterns (Mathieu et al. 2007). The 

hierarchical segmentation afforded the detection of large-scale fragments at 

higher segmentation levels while small-scale habitats such as small fragments 

and hedgerows could be detected at lower and finer segmentation levels. 

Therefore, segments in an image will never represent meaningful objects at all 

scales, for any application (Blaschke 2010) implying in the use of at least two 

levels of scales to cover all sizes of patches on mapping a fragmented 

landscapes. Moreover, we founded that segmentation reduces local spectral 

variation inducing a better discrimination between land cover types (Lobo 

1997). Hedgerows have similar species composition and appearance to 

fragments which they are associated (Castro and van den Berg 2013), making 

the insertion of the additional feature knowledge during the classification 

process a requisite to differentiate hedgerows from fragments. 

 Given the heterogeneity and the large number of fragments, the image 

analysis by a skilled interpreter is indispensable to reduce wrong classification 

and improve the accuracy of the land cover map. In the present methodology, 

visual analyses were restricted to three crucial steps: the selection of the 

segmentation parameters, the choice of color and shape parameters and the 

selection of land cover sample.   

 The high accuracy reached in this study highlights the efficiency of 

using these techniques to map small and linear features, something that most 

studies fail to achieve. The user's and producer's accuracies of the individual 

classes variation is likely caused by a combination of varying segmentation 

accuracy and the quality of the samples for the nearest neighborhood 

classification. We obtained accurate results for mapping the class hedgerows 

thanks to the integration of contextual information. The combination of multi-

scale segmentation, object-based techniques, supervised and contextual 



41 

 

 

classification demonstrates that high spectral e spatial images can optimize the 

classification of fragmented landscapes, with different kind and size of features. 

We also found that the spatial relationship to other classified objects may help to 

improve our ability to classify specific features on landscape. The classification 

process could be further enhanced by implementing class-specific rules for the 

class temporary agriculture where the classification had lower accuracy. 

 Analysis using different scales leads to more realistic quantification of 

fragmentation (Hou and Walz 2013). Our results indicate that this analysis, 

based on the detailed spatial scale and the true surface geometries of fragments, 

produced a realistic and precise representation of landscape structure. Using the 

methods presented in here we also were able to integrate the particular traits of 

the hedgerows to the classification, developed from knowledge-based rules. 

These rules allowed us to distinguish the hedgerows from forest fragments, 

although both classes of objects had similar spectral traits.  

 

4.2 The importance of small fragments and hedgerows 

 The amount of habitat and the fragmentation status are important 

variables to be considered on planning the management of the landscape for 

biodiversity conservation (Fahrig 2003; Ribeiro et al. 2009; Wilcox and Murphy 

1985). The small number of large fragments on the analyzed landscape is related 

to the extensive and ancient human occupation. The largest fragments are 

restricted to locations where the steep terrain made human occupation 

particularly difficult (Ribeiro et al. 2009; Silva et al. 2007) . The landscape 

sampling scheme was designed for capturing the spatial heterogeneity of the 

fragmentation process oriented by anthropogenic activities. The scheme was also 

directed to evaluate the role of small fragments (≤ 1ha) and hedgerows on the 

whole Atlantic Forest landscape in Minas Gerais State. The high number of 

small fragments and the low percent of forest cover present in all landscapes 

corroborate the extreme degradation of the Atlantic Forest, already indicated in 

some studies carried out in this domain (Galindo-Leal and Câmara 2003a; 

Metzger 2000; Metzger et al. 2009; Ranta et al. 1998; Ribeiro et al. 2009)  

 Although many species require large fragments to survive (Barlow et al. 

2007; Gardner et al. 2007; Harris and Pimm 2004; Laurance 2007), secondary 

forests can sustain a significant amount of biodiversity (Develey and Martensen 

2006). More than 50% of Atlantic Forest and most of the tropical regions are 

secondary or disturbed vegetation distributed on small fragments (Wright 2005) 

highlighting the necessity to include this size class on landscapes mapping. The 

present study filled gaps as presented in Ribeiro et al. (2009), in the 

underestimation of forest cover caused by the difficulty to correctly map the 

small fragments (<30 ha). The large number of small fragments (≤1 ha) founded 

here is an indicator that studies using larger scale mapping (≥ 30 ha) of the 

Atlantic Forest are missing an important feature of the landscape.   



42 

 

 

  On the choice of indicators to access biodiversity, one must recognize 

that biodiversity is a multiple-scale concept (Vogt et al. 2007) influenced by 

spatial processes (Hou and Walz 2013)  and the study must incorporate not only 

the large and well preserved fragments but also forest patches small fragments 

and corridors because of their contribution to the landscape connectivity and the 

value to biodiversity conservation theirself. Hedgerows are recognized as an 

easy option to improve connectivity in landscapes (Harvey et al. 2005) and small 

patches are also suggested as important components to improve landscape 

connectivity (Uezu et al. 2008). Disturbed areas containing small fragments 

(“stepping stones”) (Boscolo et al. 2008; Castellón and Sieving 2005; 

Sekercioglu et al. 2006; Uezu et al. 2008) and hedgerows (Mesquita and 

Passamani 2012; Rocha et al. 2011) can facilitate  animal movement. Small 

fragments and hedgerows, acting as habitat patches, can also be as stable source 

of seeds and individuals (Cerboncini et al. 2011; Mesquita and Passamani 2012; 

Ribeiro et al. 2009; Rocha et al. 2011). Because of the absence of information 

about most of threatened species distribution in tropical areas (Tobler et al. 

2007), the fragmentation pattern and spatial distribution of forest patches can be 

used as an effective surrogate to conservation plans and management of the 

landscapes (Carvalho et al. 2009). 

 Besides vegetation mapping, it is important to indicate the land uses of 

the surrounding landscape, once they affect the fragments in diverse ways. 

Therefore, a clear differentiation among areas with diverse agricultural activities 

is important to define alternative conservation strategies (Fonseca et al. 2009; 

Pardini et al. 2009; Uezu et al. 2008; Umetsu et al. 2008; Umetsu and Pardini 

2007). Beyond the matrix characteristics, fragments in landscapes affected by 

human activities requires structures that can promote  make possible the 

permeability for species  and, therefore, improve biodiversity conservation. The 

presence of hedgerows and small fragments contribute to landscape 

connectivity, but the degree of their roles will depend on the nature of the 

corridors, the nature of the matrix and the response of the organisms to both 

(Beier and Noss 1998; Rosenberg et al. 1997). Integration of hedgerows on 

farming systems contribute as a tool for conservation efforts because they 

occupy a small area and they do not interfer on farming activities. 

   The fragment size is fundamental for its species richness in highly 

isolated fragments of Brazilian Atlantic Forest (Christiansen and Pitter 1997; 

Ribon et al. 2003). However, studies have pointed out that connectivity can 

strongly diminish the negative effects of fragment-size reduction on species 

richness (Marsden et al. 2001). Therefore, small fragments and hedgerows can 

positively impact richness of fragmented landscapes. Although hedgerows might 

potentially favor the biotic flux on fragmented landscapes, their narrowness 

(maximum 15-meters width) and consequent extensive edge effect, enhance 



43 

 

 

their vulnerability to the surrounding human activities and increase their risk of 

disappearance (Vogt et al. 2007). 

 While structural connection does not imply necessarily in functional 

connection, there is a large bulk of evidences that structural corridors are 

important for biodiversity conservation (Vogt et al. 2007). Certainly the large 

number of hedgerows found in the studied landscapes can contribute to increase 

structural connection of the landscapes. Besides , we showed that the hedgerows 

are present in the whole area of Minas Gerais 'Atlantic Forest, although, they are 

denser in the South of the state. It is also possible to find these hedgerows in 

other Brazilian's states (Paraná, São Paulo, Rio de Janeiro, Espírito Santo and 

Bahia) (personal observation). Nevertheless, studies of linear vegetation strips 

associated with land division in tropical regions of Central and South America 

are exclusively for live fence (Castro and van den Berg 2013). The diversity of 

fragments size linked to hedgerows suggests the importance of these elements to 

the majority landscapes. Because of the similarity between hedgerows and the 

fragments that they are associated with (Castro and van den Berg 2013) these 

elements are increasing the landscape connectivity for fragments of a variety of 

size or, at least, increasing the size of fragments in the landscapes.    

Although the approach adopted in this study has clearly captured the distribution 

pattern of hedgerows for the Atlantic Forest in Minas Gerais, it did not entirely 

map all the landscapes on Atlantic Forest. The savannas’ areas (“cerrado”) were 

also not searched, although hedgerows exist there (personal observation). 

Therefore, we recommend a more extensive investigation looking of these 

hedgerows and investigating their holes as connectors.  

  

5. Conclusion  

 

 This study applied an object-based approach to map, for the first time, 

the hedgerows and small fragments over a large scale region, the Atlantic Forest 

in Minas Gerais State. We incorporated all size range fragments and correlated 

them to different land uses. Other studies have been only focused on mapping 

the larger and preserved units but small fragments are also crucial for forest 

monitoring and conservation. The results present here can provides a basis for 

improving landscape management, through conciliation of structural pattern and 

ecological processes. The methods developed here can contribute to improve the 

detection of landscape elements in regular monitoring, to improving the 

accuracy of vegetation maps, making conservation and management decisions 

more precise and efficient.  

 We showed here that, at least for the Brazilian Atlantic Forest in Minas 

Gerais State, hedgerows are conspicuous structures, crossing extensive regions 

and possibly promoting biotic fluxes between forest patches. Besides of that, we 

showed that fragments smaller than 1 ha are also a predominant feature for that 



44 

 

 

landscape and cannot be ignored on mapping procedures and conservation 

strategies. Mapping those structures with repeatable and accurate technique, like 

the ones we used here, can improve our understanding of landscape 

organization, and allow ecologists to better address the concept of corridors in 

biological conservation studies and policies (Vogt et al. 2007). Conservation 

activities needs to be include the hedgerows  to legally protected these 

structures, once their existence is not recognized in the Brazilian environmental 

legislation (Castro and van den Berg 2013). 

 Considering their extensive presence on the analyzed landscape, further 

research focused on conservation must include small fragments and hedgerows 

as well as legal background must be providing to protect them.   

 

Acknowledgements 

 

 We thank for the images provided for this study by the partnership 

between the Federal University of Lavras (UFLA) and State Forestry Institute  

(IEF). We thank for Fundação de Amparo a Pesquisa do estado de Minas Gerais 

and Conselho Nacional de Desenvolvido Cientifico e Tecnológico. We also 

thank for the doctorate and sandwich program scholarship provided by Capes 

(Coordenação de Aperfeiçoamento de Pessoal de Nível Superior).     

 

References 

 

Anders, N.S., Seijmonsbergen, A.C., Bouten, W., 2011. Segmentation 

optimization and stratified object-based analysis for semi-automated 

geomorphological mapping. Remote Sensing of Environment 115, 2976–

2985. 

Baatz, M., Schape, A.I.J.S., et al., (Eds.), A.G.I., Salzburg, X.B.g.z.A.-S., 

Verlag., p.K.H.W., 2000. Multiresolution Segmentation: an optimization 

approach for high quality multi-scale image segmentation. 

Barlow, J., Mestre, L.A.M., Gardner, T.A., Peres, C.A., 2007. The value of 

primary, secondary and plantation forests for Amazonian birds. Biological 

Conservation, 212-231. 

Baudry, J., Bunce, R.G.H., Burel, F., 2000. Hedgerows: An international 

perspective on their origin, function and management. Journal of 

Environmental Management 60, 7–22. 

Beier, P., Noss, R.F., 1998. Do habitat corridors provide connectivity? 

Conservation Biology 12, 1241–1252. 

Bennet, A.F., 1990. Habitat corridors and the conservation of small mammals in 

a fragmented forest environment. Landscape Ecology 4, 109-122. 

Benz, U.C., Hofmann, P., Willhauck, G., Lingenfelder, I., Heynen, M., 2004. 

Multi-resolution, object-oriented fuzzy analysis of remote sensing data for 



45 

 

 

GIS-ready information. Journal of Photogrammetry & Remote Sensing 58, 

239– 258. 

Blaschke, T., 2010. Object based image analysis for remote sensing. Journal of 

Photogrammetry and Remote Sensing 65, 2-16. 

Bock, M., Xofis, P., Mitchley, J., Rossner, G., Wissenc, M., 2005. Object-

oriented methods for habitat mapping at multiple scales – Case studies from 

Northern Germany and Wye Downs, UK. Journal for Nature Conservation 

13, 75—89. 

Boscolo, D., Candia-Gallardo, C., Awade, M., Metzger, J.P., 2008. Importance 

of interhabitat gaps and stepping-stones for a bird species in the Atlantic 

Forest, Brazil. Biotropica 40, 273-276. 

Broich, M., Stehman, S.V., Hansen, M.C., Potapov, P., Shimabukuro, Y.E., 

2009. A comparison of sampling designs for estimating deforestation from 

Landsat imagery: A case study of the Brazilian Legal Amazon. Remote 

Sensing of Environment 113, 2448–2454. 

Carvalho, F.M.V., Júnior, P.D.M., Ferreira, L.G., 2009. The Cerrado into-

pieces: Habitat fragmentation as a function of landscape use in the savannas 

of central Brazil. Biological conservation 142, 1392–1403. 

Castellón, T.D., Sieving, K.E., 2005. An experimental test of matrix 

permeability and corridor use by an endemic undestory bird. Conservation 

biology 20, 135-145. 

Castro, G.C.d., van den Berg, E., 2013. Structure and conservation value of 

high-diversity hedgerows in southeastern Brazil. Biodiversity and 

Conservation 22, 2041–2056. 

Cerboncini, R.A.S., Passamani, M., Braga, T.V., 2011. Use of space by the 

black-eared opossum in a rural area in southeastern Brazil. Mammalia 75, 

287-290. 

Christiansen, M.B., Pitter, E., 1997. Species loss in a forest bird community near 

Lagoa Santa in southeastern Brazil. Biological conservation 80, 23-32. 

Concheddaa, G., Durieuxb, L., Mayauxa, P., 2008. An object-based method for 

mapping and change analysis in mangrove ecosystems. Journal of 

Photogrammetry & Remote Sensing 63, 578–589. 

Congalton, R.G., 1991. A Review of Assessing the Accuracy of Classifications 

of Remotely Sensed Data. Remote Sensing of Environment 37, 35-46. 

Coppin, P., Jonckheere, I., Nackaerts, K., Muys, B., Lambin, E., 2004. Digital 

change detection methods in ecosystem monitoring; a review. International 

Journal of Remote Sensing 25, 1565–1596. 

Desclée, B., Bogaert, P., Defourny, P., 2006. Forest change detection by 

statistical object-based method. Remote Sensing of Environment 102, 1-11. 

Develey, P.F., Martensen, A.C., 2006. As aves da Reserva Florestal do morro 

Grande (Cotia, SP). Biota Neotropica 6, 1-16. 



46 

 

 

Dragut, L., Tiedec, D., Levickd, S.R., 2010. ESP: a tool to estimate scale 

parameter for multiresolution image segmentation of remotely sensed data. 

International Journal of Geographical Information Science 24, 859–871. 

Duelli, P., Obrist, M.K., 2003. Regional biodiversity in an agricultural 

landscape: the contribution of seminatural habitat islands. Basic and Applied 

Ecology 4, 129–138. 

Duveiller, G., Defourny, P., Desclée, B., Mayaux, P., 2008. Deforestation in 

Central Africa: Estimates at regional, national and landscape levels by 

advanced processing of systematically-distributed Landsat extracts. Remote 

Sensing of Environment 112, 1969–1981. 

Fahrig, L., 2003. Effects of habitat fragmentation on biodiversity. Annual 

Review of Ecology, Evolution and Systematics 34, 487–515. 

Fahrig, L., Merriam, G., 1985. Habitat patch connectivity and population 

survival. Ecology 66, 1762-1768. 

Fahrig, L., Merriam, G., 1994. Conservation of fragmented populations. 

Conservation biology 8, 50-59. 

Fielding, A.H., Bell, J.F., 1997. A review of methods for the assessment of 

prediction errors in conservation presence/absence models. . Environmental 

Conservation 24, 38-49. 

Fonseca, C.R., Ganade, G., Baldissera, R., Becker, C.G., Boelter, C.R., 

Brescovit, A.D., Campos, L.M., Fleck, T., Fonseca, V.S., Hartz, S.M., Joner, 

F., Käffer, M.I., Leal-Zanchet, A.M., Marcelli, M.P., Mesquita, A.S., 

Mondin, C.A., Paz, C.P., Petry, M.V., Piovezan, F.N., Putzke, J., Stranz, A., 

Vergara, M., Vieira, E.M., 2009. Towards an ecologically sustainable 

forestry in the Atlantic Forest. Biological Conservation 142, 1144–1154. 

Forman, R.T.T., Baudry, J., 1984. Hedgerows and hedgerow networks in 

landscape ecology. Environmental Management 8, 499–510. 

Galindo-Leal, C., Câmara, I.G., 2003a. Atlantic Forest hotspot status: an 

overview, In The Atlantic Forest of South America: biodiversity status, 

threats, and outlooks. eds C. Galindo-Leal, I.G. Câmara, pp. 3-11. Island 

Press, Washington. 

Galindo-Leal, C., Câmara, I.G., 2003b. The Atlantic Forest of South America: 

Biodiversity Status, Threats and Outlook. CABS and Island Press, 

Washington. 

Gardner, T.A., Barlow, J., Parry, L.W., Peres, C.A., 2007. Predicting the 

uncertain future of tropical forest species in a data vacuum. Biotropica 39, 

25-30. 

Goossens, R., D´Haluin, E., Larnoe, G., 1991. Satellite image interpretation 

(SPOT) for the survey of the ecological infrastructure in a small scaled 

landscape (Kempenland, Belgium). Landscape Ecology 5, 175-182. 

Harris, G.M., Pimm, S.L., 2004. Bird species’ tolerance of secondary forest 

habitats and its effects on extinction. Conservation Biology 18, 1607-1616. 



47 

 

 

Harvey, C.A., Villanueva, C., Villacís, J., Chacón, M., Munõz, D., López, M., 

Ibrahim, M., Gómez, R., Taylor, R., Martinez, J., Navasa, A., Saenz, J., 

Sánchez, D., Medina, A., Vilchez, S., Hernández, B., Perez, A., Ruiz, F., 

López, F., Lang, I., Sinclair, F.L., 2005. Contribution of live fences to the 

ecological integrity of agricultural landscapes. Agriculture, Ecosystems and 

Environment 111, 200–230. 

Hou, W., Walz, U., 2013. Enhanced analysis of landscape structure: Inclusion of 

transition zones and small-scale landscape elements. Ecological indicators 

31, 15-24. 

IBGE – Instituto Brasileiro de Geografia e Estatística, 2004. Mapa de biomas do 

Brasil. Escala 1:5.000.000. IBGE, Rio de Janeiro. 

Laurance, W.F., 2007. Have we overstated the tropical biodiversity crisis? 

Trends in Ecology and Evolution 22, 65-70. 

León, M.C., Harvey, C.A., 2006. Live fences and landscape connectivity in a 

neotropical agricultural landscape. Agroforestry Systems 68, 15-26. 

Lewinski, S., Zaremski, K., 2004. Examples of object-oriented classification 

performed on high resolution satellite images. Miscellanea Geographica 11, 

349-358. 

Lobo, A., 1997. Image segmentation and discriminant analysis for the 

identification of land cover units in ecology. IEEE Transactions on 

Geoscience and Remote Sensing 35, 1136-1145. 

Marsden, S.J., Whiffin, M., Galetti, M., 2001. Bird diversity and abundance in 

forest fragments and Eucalyptus plantations around an Atlantic forest 

reserve, Brazil. Biodiversity and Conservation biology 10, 737-751. 

Mathieu, R., Aryal, J., Chong, A.K., 2007. Object-Based Classification of 

Ikonos Imagery for Mapping Large-Scale Vegetation Communities in Urban 

Areas. Sensors 7, 2860-2880. 

McCollin, D., Jackson, J.I., Bunce, R.G.H., Barr, C.J., Stuart, R., 2000. 

Hedgerows as habitat for woodland plants. Journal of Environmental 

Management 60, 77-90. 

Meentemeyer, V., 1989. Geographical perspectives of space, time, and scale. 

Landscape Ecology 3, 163–173. 

Mesquita, A.O., Passamani, M., 2012. Composition and abundance of small 

mammal communities in forest fragments and vegetation corridors in 

Southern Minas Gerais, Brazil. Revista de Biologia Tropical 60, 1335–1343. 

Metzger, J.P., 2000. Tree functional group richness and landscape structure in a 

Brazilian tropial fragmented landscape. Ecological Applications 10, 1147–

1161. 

Metzger, J.P., De´camps, H., 1997. The structural connectivity threshold: an 

hypothesis in conservation biology at the landscape scale. Acta Oecologica 

18, 1-12. 



48 

 

 

Metzger, J.P., Martensen, A.C., Dixo, M., Bernacci, L.C., Ribeiro, M.C., 

Teixeira, A.M.G., Pardini, R., 2009. Time-lag in biological responses to 

landscape changes in a highly dynamic Atlantic forest region. Biological 

conservation 142, 1166-1177. 

Minor, E.S., Urban, D.L., 2007. A Graph-Theory Framework for Evaluating 

Landscape Connectivity and Conservation Conservation biology 17, 1771–

1782. 

Myers, N., Mittermier, R.A., Mittermeier, C.G., Fonseca, G.A.B., Kent, J., 2000. 

Biodiversity hotspots for conservation priorities. Nature 403, 853–858. 

Neel, M.C., McGarigal, K., Cushman, S.A., 2004. Behavior of class-level 

landscape metrics across gradients of class aggregation and area. Landscape 

Ecology 19, 435–455. 

O´Neill, R.V., 1988. Hierarchy theory and global change, In Scales and Global 

Change. eds R. T., W.R. G., R.P. G., pp. 29-45. John Wiley and Sons, 

Melbourne. 

Pardini, R., Faria, D., Accacio, G.M., Laps, R.R., Mariano, E., Paciencia, 

M.L.B., Dixo, M., Baumgarten, J., 2009. The challenge of maintaining 

Atlantic forest biodiversity: a multi-taxa conservation assessment of an agro-

forestry mosaic in southern Bahia. Biological conservation 142, 1178–1190. 

Pardini, R., Souza, S.M.d., Braga-Neto, R., Metzger, J.P., 2005. The role of 

forest structure, fragment size and corridors in maintaining small mammal 

abundance and diversity in an Atlantic forest landscape. Biological 

Conservation 124, 253–266. 

Ranta, P., Blom, T., Niemela, J., Joensuu, E., Siitonen, M., 1998. The 

fragmentation Atlantic rain forest of Brazil: size, shape and distribution of 

forest fragments. Biodiversity and Conservation 7, 385-403. 

Ribeiro, M.C., Metzger, J.P., Martensen, A.C., Ponzoni, F.J., Hirota, M.M., 

2009. The Brazilian Atlantic Forest: How much is left, and how is the 

remaining forest distributed? Implications for conservation. Biological 

conservation 142, 1141-1153. 

Ribon, R., Simon, J.E., Mattos, G.T.d., 2003. Bird extinctions in Atlantic Forest 

Fragments of the Viçosa Region, Southeastern Brazil. Conservation biology 

17. 

Rocha, M.F., Passamani, M., Louzada, J.e., 2011. A Small Mammal Community 

in a forest fragment, vegetation corridor and coffee matrix system in the 

Brazilian Atlantic forest. PLoS One 6, 1-8. 

Rosenberg, D.K., Noon, B.R., Meslow, E.C., 1997. Biological corridors: form, 

function and, efficacy. BioScience 47, 677-687. 

Scolforo, J.R., Carvalho, L.M.T., 2007. Mapeamento e inventário da flora nativa 

e dos reflorestamentos de Minas Gerais. UFLA, Lavras. 



49 

 

 

Sekercioglu, C.H., Loarie, S.R., Brenes, F.O., Ehrlich, P.R., Daily, G.C., 2006. 

Persistence of forest birds in the Costa Rican agricultural countryside. 

Conservation biology 21, 482–494. 

Silva, W.G., Metzger, J.P., Simões, S., Simonetti, C., 2007. Relief influence on 

the spatial distribution of the Atlantic Forest cover on the Ibiúna Plateau, SP. 

Brazilian Journal Biology 67, 403-411. 

Silván-Cárdenas, J.L., Wang, L., Zhan, F.B., 2009. Representing geographical 

objects with scaleinduced indeterminate boundaries: a neural network-based 

data model. International Journal of Geographical Information Science 23, 

295–318. 

SOS Mata Atlântica, I.N.d.P.E., 1993. Atlas da evolução dos remanescentes 

florestais da Mata Atlântica e ecossistemas associados no período de 1985-

1990., São Paulo. 

SOS Mata Atlântica, I.N.d.P.E., 2000. Atlas dos Remanescentes Florestais e 

Ecossistemas Associados ao Domínio da Mata Atlântica, São Paulo. 

SOS Mata Atlântica, I.N.d.P.E., 2008. Atlas dos remanescentes florestais da 

Mata Atlântica, período de 2000 a 2005. 

Tobler, M., Honorio, E., Janovec, J., Reynel, C., 2007. Implications of collection 

patterns of botanical specimens on their usefulness for conservation planning: 

an example of two neotropical plant families (Moraceae and Myristicaceae) 

in Peru. Biodiversity and Conservation 16, 659–677. 

Turner, M.G., 1989. Landscape ecology: the effect of pattern on process. Annual 

Review of Ecology, Evolution and Systematics 20, 171-197. 

Uezu, A., Beyer, D.D., Metzger, J.P., 2008. Can agroforest woodlots work as 

stepping stones for birds in the 

Atlantic Forest region? . Biodiversity and Conservation 17, 1907-1922. 

Umetsu, F., Metzger, J.P., Pardini, R., 2008. The importance of estimating 

matrix quality for modeling species distribution in complex tropical 

landscape: a test with Atlantic forest small mammals. Ecography 31, 359-

370. 

Umetsu, F., Pardini, R., 2007. Small mammals in a mosaic of forest remnants 

and anthropogenic habitats: evaluating matrix quality in an Atlantic forest 

landscape. Landscape Ecology 22, 517–530. 

Vannier, C., Hubert-Moy, L., 2008. Detection of wooded hedgerows in high 

resolution satellite images using an object-oriented method. IGARSS 4, 731-

734. 

Vogt, P., Riitters, K.H., Iwanowski, M., Estreguil, C., Kozak, J., Soille, P., 2007. 

Mapping landscape corridors. Ecological Indicators 7, 481–488. 

Wilcox, B.A., Murphy, D.D., 1985. Conservation strategy: the effects of 

fragmentation on extinction. American Naturalist 125, 879-887. 

Wright, S.J., 2005. Tropical forests in a changing environment. Trends in 

Ecology and Evolution 20, 553-560. 



50 

 

 

Wu, J., Marceau, D.J., 2002. Modelling complex ecological systems: An 

introduction. Ecological Modelling 153, 1-6. 

 

 



51 

 

 

ARTIGO 2 

 

THE RULE OF SMALL FOREST PATCHES AND HEDGEROWS ON 

BIODIVERSITY PARAMETER AT LANDSCAPE SCALE 

 

 

 

Artigo estruturado nas normas da revista “Biological conservation” 

 

  



52 

 

 

The rule of small forest patches and hedgerows on biodiversity parameter 

at landscape scale 

 
Ludimilla Zambaldi*

a
, Eduardo Van den Berg

a 

a
Biology Department, Federal University of Lavras, Minas Gerais State, Brazil.  

* Correspondending author. Tel: +55 359160 3375. E-mail adress: 

ludzambaldi@hotmail.com 

 

Abstract 

 

Landscape structure and biodiversity are strongly dependent on available area, 

isolation and connectivity among remnants. Small fragments and connectors are 

now common features in most of the landscapes related to human activities. 

However, little importance is given to those elements and their relationship to 

the area where they are inserted. In order to evaluate landscape characteristics 

and their association to vegetation remnants, matrix, hedgerows, landscape 

isolation and connectivity, we carried out analyses in 49 landscapes distributed 

over the entire Atlantic Forest included in the Minas Gerais state, Brazil, with a 

variety of vegetation cover and forest patches size. We considered sub-regions, 

distance to anthropogenic activities, relief and abiotic factors as likely important 

variables. Statistical analyses, based on selection model by AICCc value, 

revealed influence of physical structural, relief and political division on the 

remnant vegetation, isolation and hedgerows length. Small fragments (<100 ha) 

and narrow connectors (width ≤ 15 m) appeared as key elements to promote 

connectivity within the landscapes. Based on our results, we emphasize the 

necessity to take into account small fragments and connectors in landscape 

management and biodiversity conservation, even and mainly in areas where 

most of the natural habitat has already been converted to anthropogenic areas.   
Keywords: landscape ecology; Atlantic Forest; fragment size; isolation; 

connectivity 

 

1. Introduction 

 

 Biodiversity effective conservation is positively related to the amount of 

remain habitat and inversely related to habitat fragmentation (Fischer and 

Lindenmayer 2007; Martensen et al. 2008; Wilcox and Murphy 1985). Those 

findings are justified by the impact in biotic and abiotic relationships and 

ecological processes caused by landscape modifications (Bierregaard Jr. et al. 

1992; Pardini 2004). Removal of fragments, reduction in size and increase of 

remnants isolation are considered main factors of global species extinction in the 

present time (Fahrig 2001), mainly resultant from the  expansion of 

anthropogenic activities into natural areas.  



53 

 

 

 Size and distribution of vegetation remnants and their relation to the 

surrounding landscape are fundamental issues on landscape planning and 

management with focus on species conservation, once spatial arrangement of 

landscape has fundamental relationships with ecological processes (Neel et al. 

2004; Turner 1989). Larger area generally offers more resources and more 

environmental variation to harbor more individuals, allowing opportunities for 

niche specialization (Hodgson et al. 2009). 

 Agriculture and cattle raising result in landscapes dominated by small 

and isolated fragments inserted in agricultural mosaics (Fahrig 2003; Neel et al. 

2004; Tabarelli et al. 2010), negatively affecting population and community 

diversity. Because remaining forest is directly influenced by nearby land use, the 

fragments and the surrounding matrix are of particular interest for one who is 

trying to establish conservation strategies in this context. The type and 

permeability of the surrounding matrix influence on the species flux through 

landscapes elements (Uezu et al. 2008; Umetsu et al. 2008). Landscape mosaics 

include different kind of land uses, such as urban areas, roads, water courses, 

agriculture and pastures associated with patches of natural vegetation with 

heterogeneous structure and variable conditions for species occupancy (Carvalho 

et al. 2009; Vandermeer and Perfecto 2007). 

 Some landscape traits and elements can provide connectivity among 

fragments, allowing biological fluxes in fragmented landscapes. The 

connectivity is a key factor in species persistence (Fahrig and Merriam 1985; 

Fischer and Lindenmayer 2007), afforded by structural or functional 

connectivity between fragments (Tischendorf and Fahrig 2000; With and King 

1997). Structural connectivity enables the biological fluxes between patches 

through physical linkages (Forman and Collinge 1997) and functional 

connectivity are dependent of species behavior demands on a particular 

landscape, considering their capacity to cross the matrix    (Tischendorf and 

Fahrig 2000). At many scales, landscape connectivity is important for species, 

individuals and populations moving among patches (Minor and Urban 2007), 

increasing the species survival chances (Boitani et al. 2007). 

 Considered as a main landscape element that enhances the connectivity 

between patches, vegetation corridors (Beier and Noss 1998; Pardini et al. 2005; 

Uezu et al. 2008) are recognized as narrow, continuous  strips of habitat that 

structurally connect two otherwise non-contiguous habitat patches (Saunders et 

al. 2001; Tischendorf 2001) and gives the  opportunity for individuals to use 

different fragments, reducing the influence of fragment size (Martensen et al. 

2008).  

 In Brazil, the hedgerows generated by natural colonization of land plot 

boundaries ditches are a prominent landscape feature (Castro and van den Berg 

2013). These hedgerows exhibit high plant diversity inside a three meters wide 

ditches (Castro and van den Berg 2013) creating a maximum 15 meters of 



54 

 

 

canopy cover, working as well as a fragments connectors or habitat for 

mammals (Castro and van den Berg 2013; Mesquita and Passamani 2012; Rocha 

et al. 2011). Besides increasing connectivity, the hedgerows in agricultural 

landscapes are recognized globally for providing habitat, shelter and resources 

for some plant and animal species (León and Harvey 2006). Because these 

elements can determine the probability of colonization between patches (Baum 

et al. 2004; Fischer and Lindenmayer 2007), they can also have a larger-scale 

influence, on the total diversity of a landscape (Uezu et al. 2008). 

 Therefore, the understanding of the consequences of fragmentation and 

habitat loss to the structural distribution of patches, their area and connectivity is 

an important tool to infer about species persistence (Antongiovanni and Metzger 

2005; Beier and Noss 1998; Carvalho et al. 2009; Metzger and De´camps 1997). 

 Assessing effects of habitat loss and fragmentation is feasible by the 

application of landscape structure metrics on satellite images classifications 

(Neel et al. 2004; Stehman and Wickham 2011; With and King 1997). Those 

tools are useful surrogates for biodiversity assessments and can be used in 

different steps of conservation planning (Fischer and Lindenmayer 2007). 

Specially in broad-scale landscapes, landscape structural analyses are desirable, 

mainly where species inventories and biodiversity distribution patterns are still 

unavailable (Fairbanks et al. 2001), which is the case for most of tropical area 

(Ribeiro et al. 2009).   

 The Atlantic Forest it was considered one of the largest rainforests of the 

Americas, originally covering around 150 million ha, distributed in highly 

heterogeneous environmental conditions. Habitat loss and fragmentation process 

reduced the Atlantic Forest to landscapes dominated by small fragments (<100 

ha; (Ranta et al. 1998) isolated from each other (Metzger 2000; Metzger et al. 

2009) corresponding to less than 12% of the original vegetation, although it  still 

supports one of the highest degrees of species richness and rates of endemism in 

the planet (Myers et al. 2000).   

 In this study we aimed to evaluate the structural distribution of 

fragments and hedgerows and their relationship with anthropogenic and natural 

characteristics in landscapes for the Atlantic Forest in the Minas Gerais state. 

We calculated the amount of remain vegetation, isolation and connectivity of 

landscape and analyzed the relation between the fragments and ecological 

parameters.  

 

2. Methodology 

 

2.1 Study area  

 Most of the Atlantic Forest was originally located in Brazil (92%) 

(Huang et al. 2007) covering 17.4% of the country territory and distributed over 



55 

 

 

distinct topographic and climate conditions, presenting a high variety of forest 

physiognomies and compositions (Metzger et al. 2009). This highly 

heterogeneous forests harbor a high number of species (1 to 8% of species in the 

world) and is considered a biodiversity hotspot (Metzger et al. 2009; Myers et al. 

2000) and one of the most highly threatened tropical forests, with 70% of the 

Brazilian population occupying its territory. The vegetation has been reduced to 

fragments with less than 50 ha, surrounded by anthrogenic areas (Metzger et al. 

2009).  

 Historically, deforestation of the Atlantic Forest has been related to 

economic exploitation, resulting in highly fragmented landscapes and a large 

number of threatened species (Metzger et al. 2009). Minas Gerais State possess a 

highly diversified landscape. Possibly, it is related to historical occupation, 

vegetation composition, climate differences and relief complexities. The 

Brazilian Institute of Geography and Statistics (IBGE), based on these attributes, 

created 12 mesoregions in Minas Gerais state, as a subsidy to administrative, 

economic, social and tributary activities, contributing to planning activities. 

Therefore, it is also possible that the different patterns of fragmentation and 

distribution of fragments in the state relate indirectly to these mesoregions. 

 

2.2 Methods 

 We evaluated 49 landscapes each one with 12 ×12 km, randomly 

distributed at Atlantic forest biome, Minas Gerais state, southeast Brazil (Figure 

1). Sample sites were randomly selected according to the following restrictions: 

(1) The areas had to be completely included inside the domain (Atlantic Forest). 

All randomly selected samples touching the domain edges were excluded; (2) 

The areas could not share boundaries. 

 For every sample site, one high spatial resolution land cover 

classification is available (first chapter), resultant from imagery extracts 

composed of a RapidEye image acquired in 2011, with five meters of spatial 

resolution. This resolution enables us to include a large range of fragment sizes 

and connectivity conditions.  

 Land cover information was obtained by using a multilevel object-

oriented semi-automatic approach based on image segmentation using scale, 

color and shape as parameters to identify landscape elements. The classification 

resulted on land cover maps of all fragments size, agriculture, urban areas, 

water, pasture (native and non-native) and hedgerows. Agriculture class was 

subdivided in permanent agriculture, for perennial plantations like coffee and 

eucalyptus and temporary agriculture, including frequently modified plantations. 

The agriculture areas can influence in different ways the species persistence 

offering different type of permeability in the landscape. 

 We used the Nearest Neighborhood (NN) algorithm, trained by image 

samples in each class, for definition of the classes. With same spectral 



56 

 

 

characteristics of vegetation, the hedgerows were detected by structural and 

contextual rules. We used ancillary data of rivers, roads and conservation units 

to superpose on the land cover maps and calculate the distance to fragments. 

Elevation Model was used to calculate the altitude and slope of landscapes.   

 

2.3 Map validation  

 We used independent data source, randomly located within each class 

and equitably distributed over the 49 scenes for the accuracy assessment. User’s 

accuracy, producer’s accuracy and overall accuracy were computed and 

analyzed. The accuracy assessment also reported overall kappa statistics for each 

class (Concheddaa et al. 2008). All kappa measures, overall accuracy and object 

validation showed a high level of agreement and confirmed the accuracy of this 

classification. The overall accuracy of more than 80% and the kappa coefficient 

greater than 0.76 for vegetation and hedgerows pointed out to a an reliable result 

(Fielding and Bell 1997). 

 

 
Figure 1. Location of the study sites. Atlantic Forest domain inside Minas Gerais 

(MG) state and sub-regions distributions. Gray squares indicates the sample sites 

used on this study. 

 

  



57 

 

 

2.4 Landscape component indices  

 We selected isolation, matrix data, remained vegetation, connectivity 

and hedgerows length as  landscape descriptors. Our choice was based on their 

relevance to forest and ecological conservation.  

 Landscapes configuration was measured by the number and area of  

remnants, length of hedgerows, mean slope and mean altitude of each landscape. 

We used fragments area to create, for each landscape, two classes of variables: 

fragments larger than 100ha and fragments up to 99ha. We computed the mean 

size of these two fragments for each sub-region. Land cover classes were also 

used to access the mean distance of agricultural areas and pastures to each 

fragment and hedgerow as well as mean distance to conservation units, roads, 

rivers and urban areas, provided by ancillary data. We calculate the density of 

rivers in each landscape dividing the total extension of rivers by the area of 

landscape. Sub-regions were used in the analysis as variables, with 13, 12 and 

24 landscapes for sub-region A, B and C, respectively.     

 Remaining vegetation is a predictor of habitat available to species and a 

measure of conservation degree of landscapes (Fahrig 2003). Based on the size 

of all fragments, we calculate the percentage of remaining vegetation for all 

sample sites. Afterwards, we evaluated the relation between the percentage of 

rem