CECÍLIA GONTIJO LEAL MULTISCALE ANTHROPOGENIC IMPACTS ON STREAM CONDITION AND FISH ASSEMBLAGES IN AMAZONIAN LANDSCAPES LAVRAS-MG 2015 CECÍLIA GONTIJO LEAL MULTISCALE ANTHROPOGENIC IMPACTS ON STREAM CONDITION AND FISH ASSEMBLAGES IN AMAZONIAN LANDSCAPES Thesis submitted for the degree of Doctor of Philosophy as a Dual PhD between the Lancaster Environment Centre, Lancaster University, United Kingdom and the Applied Ecology Postgraduate Program, Universidade Federal de Lavras, Brazil. Supervisors Dr. Paulo dos Santos Pompeu Dr. Jos Barlow Co-supervisor Dr. Toby Gardner LAVRAS-MG 2015 Ficha catalográfica elaborada pelo Sistema de Geração de Ficha Catalográfica da Biblioteca Universitária da UFLA, com dados informados pelo(a) próprio(a) autor(a). Leal, Cecília Gontijo. Multiscale anthropogenic impacts on stream condition and fish assemblages in Amazonian landscapes / Cecília Gontijo Leal. – Lavras: UFLA, 2015. 224 p. : il. Tese(doutorado)–Universidade Federal de Lavras, 2015. Orientador(a): Paulo dos Santos Pompeu. Bibliografia. 1. Forest-agriculture frontier. 2. Water quality. 3. Physical habitat. 4. Human-modified tropical forests. 5. Ichthyofauna. I. Universidade Federal de Lavras. II. Título. CECÍLIA GONTIJO LEAL MULTISCALE ANTHROPOGENIC IMPACTS ON STREAM CONDITION AND FISH ASSEMBLAGES IN AMAZONIAN LANDSCAPES Thesis submitted for the degree of Doctor of Philosophy as a Dual PhD between the Lancaster Environment Centre, Lancaster University, United Kingdom and the Applied Ecology Postgraduate Program, Universidade Federal de Lavras, Brazil. APPROVED on the 27th of March 2015. Dr. Jos Barlow LANCASTER UNIVERSITY Dr. Felipe Pimentel Lopes de Melo UFPE Dr. Júlio Neil Cassa Louzada UFLA Dr. Marcelo Passamani UFLA Dra. Carla Rodrigues Ribas UFLA Dr. Paulo dos Santos Pompeu Supervisor LAVRAS-MG 2015 Declaration I herewith declare that this work has been originally produced by myself for the present thesis, and it has not been previously presented to obtain a degree in any form. Collaborations with other researchers, as well as publications or submissions for publication are properly acknowledged throughout the document. Cecília Gontijo Leal, Lavras, Brazil, February 2015. Para minha mãe e para o meu pai por tanto amor sempre To the Amazon, where the powers of life are predominant DEDICATION Charles Darwin, Beagle Diary, on his visit to the Brazilian rainforest: “The delight one experiences in such times bewilders the mind, — if the eye attempts to follow the flight of a gaudy butterfly, it is arrested by some strange tree or fruit; if watching an insect one forgets it in the stranger flower it is crawling over, — if turning to admire the splendour of the scenery, the individual character of the foreground fixes the attention. The mind is a chaos of delight, out of which a world of future & more quiet pleasure will arise.” ACKNOWLEDGEMENTS / AGRADECIMENTOS If I find enough words maybe I’ll be able to express my upmost gratitude to my supervisors Paulo Pompeu, Jos Barlow and Toby Gardner. You taught me (attention to the correct spelling!) much more than I could ever have written in this thesis. The science behind samples, graphs and results are a small detail besides all feelings, memories and experiences I will take with me irrespective of the next steps in my life. You were just fundamental since the beginning. Ao Paulo que como sempre, desde o mestrado, transborda positividade e incentivos à liberdade, às tentativas, aos pensamentos (às vezes megalomaníacos) e sempre está lá vibrando orgulhoso pelas conquistas ou pronto pra dividir qualquer preocupação. Paulo, com a sua orientação eu nunca me senti sozinha. Esse eu já tinha um orientador que entende das angústias, inseguranças e desafios ao longo dos caminhos e que concorda que ser feliz vale mais do que qualquer linha no currículo, então ganhei outros dois durante o doutorado. Jos e Toby obrigada por terem mergulhado tão profundamente comigo nas águas dessa tese sempre mostrando muito além do que meus olhos enfeitiçados pelos peixes poderiam enxergar. Toby thanks for being so friendly and helpful 24h per day and cheerful with every small achievement. Jos your guidance was fundamental; in every single meeting I came out much more inspired than when I entered. This made me confident that it was worth carrying on. A vocês três: thanks for showing me that there is a big wide world out there… and to support me to go for it. À toda Equipe Aquática do RAS, especialmente os companheiros de campo, sempre na busca incansável por igarapés “amostráveis” que por incrível que pareça não eram tão fáceis de encontrar na Amazônia de tantas águas. Manoel Nascimento (Nego), Renilson de Freitas (Graveto), Martinez de Oliveira, Elbin da Silva, Milton, Francisco Cunha (Pita) e ValderirNascimento por tanta dedicação em campo e por me passarem um pouquinho da sabedoria naturalista nata. Aos amigos novos e antigos que no dia a dia do campo entraram nos igarapés sem medo de se molhar e ser feliz:Rafael Leitão, Bob Hughes, Débora de Carvalho, Lucas Pires, Douglas Bastos, Ceceo Chaves, Leandro Brasil, Janaína de Brito, Ceceo Chaves, Lenise da Silva, Túlio Franco, Leandro Juen, Vivian de Oliveira, José Max de Oliveira-Silva, Karina Silva, Miriam de Almeida, Marcos Vinícius da Silva e Rafael Duarte. Érika Berenguer e Bob Solar pelas trocas de figurinha inter-equipes. Aos coordenadores do RAS Toby Gardner, Joice Ferreira, Luke Parry, Alex Lees e Jos Barlow pelo suporte constante em todas as fases do projeto. Especialmente Toby e Joice pelo apoio logístico fundamental em campo, fazendo o máximo para tudo acontecer da forma mais harmoniosa, eficiente e obviamente divertida. To all farmers and workers unions of Santarém, Belterra, and Paragominas and all collaborating private landowners for their support. I am are also grateful to the following institutions for financial support of the RAS project: Instituto Nacional de Ciência e Tecnologia – Biodiversidade e Uso da Terra na Amazônia (CNPq 574008/2008-0), Empresa Brasileira de Pesquisa Agropecuária – Embrapa (SEG: 02.08.06.005.00), the UK government Darwin Initiative (17-023), The Nature Conservancy, Natural Environment Research Council (NERC) (NE/F01614X/1 and NE/G000816/1). To CAPES for my scholarship in Brazil (2943-13-1) and UK (Science without Borders - PDSE-2943/13-1). To my coautors in the thesis manuscripts Rafa Leitão, Jansen Zuanon, Bob Hughes, Phil Kaufmann, Benito, Silvio Ferraz, Jim Thomson and Ralph Mc Nally for the endless patience, help and reasonable contributions trying to decrypt my thoughts. Ao Rafa Leitão, inseparável parceiro do início ao fim da tese, companheiro Amazônia adentro e Inglaterra afora! Incontáveis momentos divertidos juntos. Dividindo nossos desesperos com os peixes “inidentificáveis”, variáveis sem explicação biológica, planilhas com milhares de linhas e colunas sem fim, tive mais fôlego pra seguir em frente! Aos programas de pós-graduação em Ecologia Aplicada da Universidade Federal de Lavras and Science of Tropical Environments at Lancaster University pelo apoio durante todo o doutorado. Aos professores de ambas instituições queinvestiram tanta energia em fazer a dupla titulação acontecer, principalmente Júlio Louzada e Jos Barlow. To the tropical group, the most “vibrant research group (Barlow, 2014)” ever. Our family bond was a constant relief and inspiration to carry on: Anne, Jos, Luke, Mari, Natalie, Macaco-porco, Lisi, Careca-Cabeludo, Hannah, Sujo- safado, Érika, Rachael Carrie, Sâmia, Bobão and Fabinho. Aos pupilos dos Pompeu do Laboratório de Ecologia de Peixes da UFLA, que desde o mestrado deixam o dia-a-dia no laboratório e as viagens de campo mais interessantes e divertidas: Débora, Dani, Marcela, Tatau, Nina, Fábio, Lud, Nara, Ivo, Rafa e toda a trupe que de alguma parte contribuem para esse cardume. To myvery best special special friends in Lancaster who showed me that love and glitter are never too much and can cross any transoceanic barrier: Mari Piva, my husband Natalie Swan, Hannah Griffiths, Anne Toomey, Lisi Zanella, Dany Tregidgo, Guy Crawford, Olivia Havercroft, Beth Brocket, Luke Parry, Tom Walker, Al Campbell, Ali Birket, Victor Careca-cabeludo, Filipe França, Bob Solar, Fabinho, Gina Frauzin, Victoria Frausin, Érika Berenguer, Chris Rhodes, Andrezão, Grazyna Monvid, Ainara Penalver, Artzai Jauregui and so many others from the Lancaster Environment Centre and the Ecology and Conservation Group!!!Aos queridos Lisi, Toru, Filipe e Andrezão que em Lancaster deram à dupla titulação um aconchego de família: sou duplamente grata à dupla amizade de vocês. I am also deeply thankful to the “Spanish house” team (Mari, Maria and Begoña), Érika and Darren, Grazyna and Al Campbell for hosting me and helping with the six house moving during my “one year” in Lancaster. --- Thank you all for bringing constant joy to my life in Lancaster! Aos amigos brasileiros de sempre e que são muitos e importantes demais em todos os momentos. Os de infância, os de família, os da biologia UFMG, os do mestrado UFLA, as da Manacá, os que foram aparecendo e os que reapareceram ao longo do caminho, os da família Sá Fortes Leite. Especialmente aos biólogos que desde o começo me inspiraram a amar expedições, coletas, história natural, essa coisa de embrenhar no campos das gerais e ser meio bicho do mato mesmo: Felipe Leite, Andréa Mesquita, Fernanda Barros, Flávia Pezzini, Letícia Garcia, Pedro Vianna e Thaís Elias. À minha família linda, especialmente pai e mãe, irmãos, cunhadas e sobrinhos sempre tão presentes, positivos, interessados em saber pra onde vou e o que tenho pra contar.Tão unicamente incentivadores do que quer que eu faça onde quer que eu esteja. To the deepest blue eyes I’ve ever seen… who showed me that an ocean away means nothing and taught me to don’t wash spiders down the plughole. Fitter, happier. Obrigada Al. RESUMO Mudanças no uso da terra e degradação florestal têm resultado em severas alterações aos ambientes tropicais no mundo, entretanto as consequências aos cursos d’água permancem pouco conhecidas. Isto é ainda mais crítico para a bacia Amazônica, particularmente sua complexa rede de igarapés. Estes igarapés fazem a conexão entre os ecossistemas terrestre e aquático através da paisagem. Além disso abrigam grande parte da diversidade de peixes, se não a maioria, da bacia que por si só é a mais diversa do mundo. Apesar da incontestável relevância dos igarapés, as consequências das mudanças no uso da terra para seu habitat aquático e fauna de peixes permanece uma grande lacuna de conhecimento. Esta tese tem como objetivo preencher parte desta lacuna investigando os efeitos dos distúrbios antrópicos em diferentes escalas espaciais na condição biológica dos igarapés em paisagens antropicamente modificadas do estado do Pará, Brasil. O estudo começa investigando como o habitat aquático (representado por características de qualidade da água e habitat físico) responde aos distúrbios antropogênicos da paisagem (Capítulo 2). Em seguida são avaliadas mudanças na riqueza de espécies, abundância e composição das comunidades frente a alterações tanto do habitat aquático quanto da paisagem (Capítulo 3). Por último verificou-se a importância relativa das mesmas variáveis ambientais preditoras usadas no Capítulo 3 nas respostas espécie-específicas, avaliando a potencial efetividade da legislação ambiental brasileira em levá-las em consideração (Capítulo 4). Foram amostrados 99 igarapés distribuídos em cinco bacias hidrográficas e duas regiões (Santarém e Paragominas) na Amazônia oriental, região de desenvolvimento agrícola. Foram registrados 25,526 exemplares de peixes pertencentes a 143 espécies, 27 famílias e sete ordens; sendo os igarapés altamente heterogêneos em suas características bióticas e abióticas. Por exemplo, em todas bacias hidrográficas a diversidade beta entre igarapés foi mais representada pela substituição de espécies (ca. 90%) do que pelo aninhamento. De forma geral os resultados encontrados enfatizam a importância de diversos usos da terra e escalas espaciais em influenciar o habitat aquático dos igarapés, incluindo associações entre por exemplo cobertura florestal na drenagem e temperatura da água, ou dos impactos de cruzamentos de estradas na morfologia do canal. Ambos, paisagem e habitat aquático também influenciaram as comunidades de peixes, porém o habitat aquático mostrou-se particularmente importante em explicar os padrões de abundância das espécies quando comparado a características da paisagem geralmente consideradas mais propícias ao manejo (e.g. proteção da floresta ripária). Entretanto os resultados também ressaltam a complexidade dos igarapés e as dificuldades de desvendar os efeitos de indicadores de distúrbios antrópicos em múltiplas escalas espaciais sustentados por uma inerente heterogeneidade ambiental – tanto o habitat aquático quanto as comunidades de peixes foram influenciados por uma ampla gama de variáveis que diferiram nas diferentes bacias hidrográficas e regiões. Os resultados encontrados são utilizados para discutir os desafios e recomendações ao manejo e conservação desses sistemas amazônicos em paisagens antropicamente modificadas. Enfatizando particularmente a necessidade de estratégias coletivas planejadas em escala de drenagem, ou seja, que incorporem mais que a zona ripária dentro de propriedades rurais individuais como priorizado pela legislação ambiental brasileira vigente. Palavras-chave: Fronteira de desenvolvimento agrícola. Qualidade da água. Habitat físico. Florestas tropicais antropicamente modificadas. Ictiofauna. Desmatamento. Cruzamento de estradas. ABSTRACT Land use change and forest degradation are resulting in pervasive changes to tropical ecosystems around the globe, however consequences for freshwater ecosystems remain poorly understood. This is especially true for the Amazon basin, in particular for its complex network of low-order streams. These streams connect terrestrial and aquatic ecosystems throughout landscapes and host much, of the freshwater fish fauna of the Amazon basin. Despite the biological significance of these stream networks, the consequences of land use change for the condition of instream habitat and fish fauna remain very poorly studied and understood. This thesis aims to address part of this knowledge gap by investigating the effects of anthropogenic disturbances occurring at multiple spatial scales on stream condition and fish assemblages from human-modified Amazonian forests in the state of Pará, Brazil. The thesis starts by asking how instream habitat (composed of both water quality and physical habitat features) responds to landscape-scale anthropogenic disturbances and natural features (Chapter 2). Chapter 3 then investigates changes in fish species richness, abundance and composition following changes in both instream habitat and landscape-scale anthropogenic disturbance. Last, in Chapter 4 I attempt to disentangle the relative importance of those multiscale environmental predictor variables on species-specific disturbance responses, and evaluate the potential effectiveness of the Brazilian legislation in accounting for them. A total of 99 low-order streams were surveyed from five river basins in two large regions (Santarém and Paragominas) in the eastern Brazilian Amazon agricultural-forest frontier. I sampled a total of 25,526 fish specimens belonging to 143 species, 27 families and seven orders. Streams appeared to be exceptionally heterogeneous in their abiotic and biotic features. For instance the contribution of turnover to the beta stream site component was much higher than nestedness in all river basins. Overall these findings underscore the importance of multiple land use changes and disturbances, at multiple spatial scales, in shaping instream habitat, including links between catchment-scale forest cover and water temperature, and the impacts of road crossings on channel morphology. Both landscape and instream habitat variables were isolated as having a marked effect on stream fish, but instream habitat differences were shown to be particularly important in explaining patterns of fish species abundance compared to other landscape factors that are more amenable to management such as the protection of riparian forest strips. However the results of the thesis also highlight the complexity of Amazonian stream systems and the difficulties in disentangling the effects of multiscale environmental predictor variables underpinned by naturally heterogeneous biophysical characteristics – with instream habitat and fish assemblages affected by a broad suite of drivers that often varied across river basins and regions. I use the findings of the thesis to discuss challenges and recommendations for the management and conservation of low-order streams in Amazonian human-modified landscapes. In particular I emphasize the need for catchment-wide collective management approaches that go beyond the protection of riparian forests within individual properties as prioritized by existing Brazilian environmental legislation. Keywords: Forest-agriculture frontier. Water quality. Physical habitat. Human- modified tropical forests. Ichthyofauna. Deforestation. Road crossings. LIST OF FIGURES General Introduction Figure 1.1 Macrophytes in low-order amazonian streams due to disturbances to the riparian vegetation (a, b, c; santarém) in contrast to typically shaded channels (d, e; paragominas). all pictures taken by ras aquatic team....................................................................................31 Figure 1.2 Santarém and Paragominas mosaic of land-uses: Eucalyptus sp. monoculture (A), soya plantation (B), small black peppercorn crop (C), primary forest (E) with preserved streams (J), manioc plantation (E) with associated use of streams (F) for manioc flour preparation (G). Logging (H) and fire associated to pasture (I) are also common. Moreover streams are largely used by rural families (L, M, N) including for small-scale hydropower generation, as well as for animal use (K). Pictures D, E, H taken by Érika Berenguer, the other ones by RAS Aquatic Team.............................................36 Figure 1.3 Stream sites from Santarém (A, B, C) and Paragominas (D, E, F) spread across a gradient of anthropogenic disturbances including preserved (A, D), intermediate (B, E) and degraded (C, F) conditions. All pictures taken by RAS Aquatic Team....................38 Figure 1.4 Aquatic team in the field: reconnaissance of prospective sampling areas and preparatory work for field sampling. All pictures taken by RAS Aquatic Team.........................................................................39 Figure1.5 Schematic of the spatial scales considered to obtain the environmental landscape variables. Riparian buffers are referred as network and local............................................................................40 Figure 1.6 Schematic of the sampling design of the instream habitat of Amazonian sites..............................................................................42 Figure 1.7 Measuring instream habitat characteristics in stream sites: substrate type (A), water properties (B), channel slope (C), depth (D), and canopy density (E). All pictures by RAS Aquatic Team................43 Figure 1.8 Fish sampling in low-order Amazonian streams. All pictures by RAS Aquatic Team.........................................................................44 Figure 1.9 Methodological framework representing overall thesis structure and links between chapters....................................................................46 Manuscript 1 Figure 2.1 Methodological framework to investigate the response of instream habitat of low-order Amazonian stream sites to local and landscape- scale human disturbances (see Table 1). Q1, Q2 and Q3 are the research questions referred to in the Introduction; * see section “Selection of response variables” for detailed steps.......................64 Figure 2.2 Contribution of landscape predictor variables to the first two PCA axes for Santarém (A) and Paragominas (B). Variables in bold were selected for further analysis, with excluded highly correlated metrics listed below each of them...................................................77 Figure 2.3 Representation of random forest (RF) models showing the percentage of variation of the instream habitat response variables explained (pseudo-R2) by anthropogenic predictor variables in Amazonian stream sites. Results are from models that included both anthropogenic and natural predictor variables (‘All’ models shown on Table 2.3)...................................................................................82 Figure 2.4 Raw data distribution (dots) and partial contribution of landscape predictor variables (lines) to instream habitat in Santarém (A, B, C) and Paragominas (D, E, F)..............................................................83 Manuscript 2 Figure 3.1 Location of stream site catchments in Paragominas (ca. 1.9 million ha) and Santarém (ca. 1 million ha) regions, Pará state, eastern Brazilian Amazon.........................................................................118 Figure 3.2 Rank of species abundance (A, B, C) and occurrence (D, E, F) in Curuá-Una (A, D), Capim (B, E) and Gurupi (C, F) river basins. Together the indicated species represent 70% of the total abundance in each basin..................................................................................131 Figure 3.3 Stream site-based rarefaction curves for stream fish from Curuá- Una, Capim, and Gurupi River basins..........................................132 Figure 3.4 Additive diversity partitioning for stream fish by region (Santarém and Paragominas, A) and river basin (Curuá-Una, Capim and Gurupi, B).....................................................................................133 Figure 3.5 Nonmetric multidimensional scaling (MDS) of the fish assemblages from five river basins in the eastern Brazilian Amazon. Ordination analysis was based on quantitative (Bray-Curtis, stress= 0.22; A) and qualitative (Sorensen, stress= 0.21; B) dissimilarity matrices.........................................................................................134 Figure 3.6 Nonmetric multidimensional scaling (MDS) of the fish assemblages from Curuá-Una (A), Capim (B) and Gurupi (C) river basins. The MDS analysis was based on Bray-Curtis dissimilarity index scores. Significant environmental vectors from ‘envfit’ represent instream habitat (blue), anthropogenic (green), and natural (red) predictor variables. See Table 1 for the codes of the predictor variables........................................................................................136 Manuscript 3 Figure 4.1 Schematic of the environmetal predictor variables divided into four groups (‘natural’, ‘instream habitat’, ‘riparian network’ and ‘other landscape’) used to investigate stream fish species-specific responses in the eastern Amazon..................................................167 Figure 4.2 Location of stream site catchments in Paragominas (ca. 1.9 million ha) and Santarém (ca. 1 million ha) regions, Pará state, eastern Brazilian Amazon.........................................................................169 Figure 4.3 Partitioning of the variation in occupancy of stream fish species in Curuá-Una (A), Capim (B) and Gurupi (C) River basins, showing the effects of each group of predictor variables when partitioning out the effects of the other groups through redundancy analysis. Blue, light green, dark green and red represent respectively the fractions explained by instream habitat, riparian network, other landscape and natural alone; black represent all other fractions together..........................................................................................184 Figure 4.4 Isolated and shared effects of instream habitat (I), riparian network (R), other landscape (L), and natural (N) predictor variable groups on stream fish represented by mean and standard error (SE) for each river basin: Curuá-Una (A), Capim (B) and Gurui (C).................................................................................................187 Figure 4.5 Partial effect from random forest models (lines) of physical and chemical habitat, riparian network forest cover and other landscape predictors showing positive associations (dots) with disturbed (A) or better preserved condition (B). Other partial effects were not clearly attributed to sites condition as they can be representing size as well as anthropogenic disturbance (C).....................................190 Figure 4.6 Cluster heat-map of species based on random forest (RF) models results for Capim (A) and Gurupi (B) River basins. Each cell is coloured based on the percentage of explanation values generated by RF. The cluster on the left side of each heat-map groups species with similar response patterns according to their relationship with different predictor variables, based on Euclidean distance..........................................................................................191 LIST OF TABLES Manuscript 1 Table 2.1 Landscape variables, natural and anthropogenic, used to predict Amazonian instream habitat condition. Selected variables are highlighted in bold..........................................................................68 Table 2.2 Acronyms and definitions of instream habitat (water quality and physical habitat features) response variables of Amazonian streams............................................................................................73 Table 2.3 Performance of random forest (RF) models showing the percentage of variation of the instream habitat response variables explained (pseudo-R2) by models that included all predictor variables (All), only the anthropogenic (Ant) and only the natural variables (Nat). Note that strong interactions between anthropogenic and natural predictor variables can result in pseudo-R2 values for the combined (All) model that exceed the sum of values for anthropogenic and natural models (e.g. Dgm in STM; highlighted in light grey). Conversely, the combined model can have lower pseudo-R2 values than anthropogenic (medium grey) or natural (dark grey) models because the random inclusion of weaker predictors in individual trees may lower the overall mean predictive performance (e.g. OD in STM and COND in PGM respectively). Values in parentheses in “All” columns show the % contribution of anthropogenic variables to total variance explained in combined models.............................79 Manuscript 2 Table 3.1 Environmental predictor variables (landscape-scale and instream habitat) used to predict fish diversity and composition from Amazonian stream sites................................................................124 Table 3.2 Performance of random forest models showing the percentage of variation of richness and abundance explained by environmental predictor variables in Curuá-Una (CU), Capim (CA) and Gurupi (GU) River basins. Partial effect of single variables greater than 5% in bold...........................................................................................137 Manuscript 3 Table 4.1 Environmental variables (landscape and instream habitat) used to predict fish species-specific abundances from Amazonian stream sites................................................................................................176 Table4.2 Fish species trophic groups (allo= allochthonous, auto= autochthonous, gen= generalist), number of individuals and occurrence in number of stream sites from Curuá-Una, Capim and Gurupi River basins, eastern Brazilian Amazon...........................181 CONTENTS FIRST PART GENERAL INTRODUCTION......................................................... 23 1 GENERAL INTRODUCTION......................................................... 24 1.1 Freshwater biodiversity..................................................................... 24 1.2 Tropical streams................................................................................ 26 1.2.1 Streams in human-modified tropical forests................................... 27 1.3 The study system: the eastern Brazilian amazon............................ 29 1.3.1 Amazonian streams and their fish assemblages.............................. 30 1.3.2 Conservation of Amazonian streams............................................... 32 1.3.3 The sustainable amazon network..................................................... 34 1.4 Data sampling..................................................................................... 37 1.4.1 Landscape environmental variables................................................. 39 1.4.2 Instream habitat................................................................................. 41 1.4.3 Fish...................................................................................................... 43 1.5 Objectives and structure of the thesis.............................................. 44 REFERENCES................................................................................... 47 SECOND PART - MANUSCRIPTS................................................ 56 MANUSCRIPT 1: MULTI-SCALE ASSESSMENT OF HUMAN-INDUCED CHANGES TO AMAZONIAN INSTREAM HABITATS........ 57 1 INTRODUCTION............................................................................. 59 2 METHODS......................................................................................... 65 2.1 Study system....................................................................................... 65 2.2 Sampling............................................................................................. 66 2.2.1 Landscape predictor variables......................................................... 66 2.2.2 Instream habitat response variables................................................ 70 2.3 Data analysis....................................................................................... 71 2.3.1 Selection of landscape predictor variables...................................... 71 2.3.2 Selection of instream habitat response variables............................ 72 2.3.3 Relationships between land use change and instream habitat...... 74 3 RESULTS........................................................................................... 76 3.1 Variation in landscape characteristics of stream sites................... 76 3.2 LUC influences on stream site condition......................................... 78 3.4 Influence of region and landscape scale on instream habitat condition............................................................................................. 80 4 DISCUSSION..................................................................................... 84 4.1 Do human-induced disturbances influence tropical instream habitats as expected?......................................................................... 84 4.2 Challenges in understanding the influences of anthropogenic disturbances on instream habitat in tropical streams.................. 86 4.3 Disentangling the effects of anthropogenic disturbance from natural variation among Amazonian streams............................... 87 4.4 Cumulative effects of multiple drivers........................................... 88 4.5 Accounting for the full gradient of landscape disturbance.......... 90 4.6 Time-lags in disturbance responses................................................ 91 4.7 Implications for the conservation management of Amazonian streams.............................................................................................. 93 REFERENCES................................................................................. 95 APPENDIX……………………....................................................... 105 MANUSCRIPT 2 A LARGE-SCALE ASSESSMENT OF FISH DIVERSITY IN SMALL STREAMS ACROSS HUMAN-MODIFIED AMAZONIAN LANDSCAPES…………….. 112 1 INTRODUCTION........................................................................... 114 2 METHODS....................................................................................... 117 2.1 Study region...................................................................................... 117 2.2 Data sampling................................................................................... 119 2.2.1 Environmental predictor variables................................................ 119 2.2.2 Landscape-scale............................................................................... 119 2.2.3 Instream habitat............................................................................... 121 2.2.4 Fish.................................................................................................... 122 2.3 Data analysis..................................................................................... 123 2.3.1 Selection of environmental predictor variables............................ 123 2.3.2 Analyzing fish assemblage diversity patterns............................... 125 3 RESULTS......................................................................................... 129 4 DISCUSSION................................................................................... 139 4.1 Insights into the biogeography and diversity of Amazonian streams.............................................................................................. 139 4.2 Natural and anthropogenic drivers of fish assemblage structure and composition............................................................... 141 4.3 Challenges and opportunities for the conservation of stream fish assemblages in human-modified Amazonian landscapes...... 143 REFERENCES................................................................................. 146 APPENDIX....................................................................................... 154 MANUSCRIPT 3 A LARGE-SCALE ASSESSMENT OF LOCAL, RIPARIAN AND CATCHMENT- LEVEL IMPACTS ON AMAZONIAN STREAM FISH.................................................. 162 1 INTRODUCTION........................................................................... 164 2 METHODS....................................................................................... 168 2.1 Study region...................................................................................... 168 2.2 Data sampling................................................................................... 170 2.2.1 Fish.................................................................................................... 170 2.2.2 Environmental predictor variables................................................ 171 2.3 Data analysis..................................................................................... 174 2.3.1 Selection of the environmental predictor variables...................... 174 2.3.2 Statistical analysis............................................................................ 177 3 RESULTS......................................................................................... 180 4 DISCUSSION................................................................................... 192 4.1 Understanding anthropogenic disturbances in megadiverse tropical systems................................................................................ 192 4.1.1 Relative importance of environmental drivers............................. 192 4.1.2 Challenges in understanding species-environment relationships in tropical streams........................................................................... 194 4.1.3 Implications for management......................................................... 195 REFERENCES................................................................................. 198 APPENDIX....................................................................................... 205 1 CONCLUDING REMARKS.......................................................... 211 1.1 Synopsis of key findings.................................................................. 211 1.2 Application of research findings: recommendations for the management and conservation of Amazonian riverscapes.......... 215 1.3 Future research priorities............................................................... 217 1.4 Conclusions……............................................................................... 218 REFERENCES................................................................................. 220 APPENDIX: OTHER OUTCOMES.............................................. 222 FIRST PART 23 1 GENERAL INTRODUCTION 24 1 GENERAL INTRODUCTION 1.1 Freshwater biodiversity Freshwater ecosystems occupy less than 1% of the Earth’s surface, make up 0.01% of all water, and provide many vital services relevant to human well-being and poverty alleviation (MILLENNIUM ECOSYSTEM ASSESSMENT, 2005). They also host a large proportion of global biodiversity including ca. 10% of all known species and ca. 33% of all vertebrates (STRAYER; DUDGEON, 2010). Freshwater fish are both highly diverse, with estimates of up to 13,000 fish species in total, and have high levels of endemism (LÉVÊQUE et al., 2008). The Neotropical region alone is responsible for 5,600 recognised species of fish, equivalent to10% of all vertebrate species, which are distributed across some of the most diverse river basins in the world(ALBERT; BART; REIS, 2011; LÉVÊQUE et al., 2008). Furthermore freshwater ecosystems are considered highly threatened, more so than terrestrial and marine equivalents. The main drivers of these threats are linked to anthropogenic activities leading to habitat degradation, pollution, flow regulation and water extraction, fisheries, overexploitation, and alien species introductions (STRAYER; DUDGEON, 2010). However, the scientific knowledge about freshwater systems is incomplete, and human-induced changes remain poorly understood and may be underestimated. This situation is likely to be more critical in tropical systems than temperate ones. First because fish species composition is poorly known for most tropical river basins (DUDGEON et al., 2006). Second, it is in the megadiverse tropics where landscapes are under rapid and penetrating pressure from intensive and rapid development of urban and agricultural lands with irreversible widespread consequences for natural ecosystems (FAO, 2011; MALHI et al., 2014). 25 Most of our understanding of human-impacts on tropical forests has been developed from the study of terrestrial systems. In contrast, aquatic systems have received far less research attention, with the majority of work to date being concentrated in a small number of well-studied regions, such as Costa Rica, Puerto Rico, and Australia (DUDGEON, 2008). Moreover, work tends to focus on lakes and large rivers that are of interest for navigation and power generation, and which host fish species of commercial value (e.g. RIBEIRO&PETRERE JUNIOR 1990; BATISTA&PETRERE Jr. 2003; ARDURA et al. 2010). As a result, tropical low-order streams have been largely neglected by scientific research, and remain poorly understood. An example of this can be seen in a recent review of 62 studies assessing faunal responses to land use change in Amazonia (PERES et al., 2010) that included only one study investigating freshwater systems, in this case, stream fish (see DIAS et al. 2010). Effective conservation strategies should be built on robust scientific information, and the lack of research on tropical low-order streams means they generally carry little weight in management and conservation planning (BENSTEAD et al., 2003). Global conservation planning initiatives aim to prioritise conservation efforts to areas that have unique biological richness (irreplaceability) and high vulnerability to threats (e.g. biodiversity hotspot regions, protected areas). Almost all are defined based on terrestrial parameters (BROOKS et al., 2006) which are unlikely to match priorities set for freshwater systems (ABRAHAM; KELKAR, 2012; HERBERT et al., 2010). The available initiatives that account for freshwater systems are far more rudimentary than their terrestrial counterparts: for example, an attempt to categorize global freshwater units based on fish species distribution and composition, the Freshwater Ecoregions of the World, FEOW (ABELL et al., 2008), is too coarse to assist regional management strategies for low-order streams. Furthermore in Brazil, the IUCN Red Species List only started including freshwater fish in 26 2004, whereas similar information was created for other biotic groups in 1968 and has been refined since then. Furthermore, data availability means that it is much easier to estimate a threat status for large sized and commercially valuable fish species than typical low-order stream fish fauna (MMA, 2014). Given the lack of information outlined above, it is critical to investigate how changes in tropical forest landscapes translate into changes in stream systems, as this will help guide effective interventions in watershed management and biodiversity conservation (MOULTON; WANTZEN, 2006). 1.2 Tropical streams “There is no such thing as a ‘typical’ tropical stream”(DUDGEON, 2008)Tropical streams share some broad similarities in natural features, such as having a high water temperature for a given elevation, and often having high levels of hydrological periodicity with intense rainfall and runoff (BOULTON et al., 2008). However, beyond these broad generalities, tropical streams, like temperate streams, are extremely heterogeneous in their biotic and abiotic characteristics (DUDGEON, 2008), making it very difficult to draw further generalizations. In addition to any environmental distinctions, the high levels of land use change that characterise much of the tropics in recent decades mean that tropical streams can often differ markedly in socio-economic aspects (MOULTON; WANTZEN, 2006). Both temperate and tropical streams networks do have one important defining feature that has profound implications for their conservation and management: their hierarchical spatial organization that determines how local conditions are highly dependent on their regional context (FRISSELL et al., 1986). This network connectivity is not restricted to the watercourses themselves; stream habitats, water quality and aquatic biota are all influenced by 27 nested landscape scales factors through complex and varying pathways (ALLAN; ERICKSON; FAY, 1997; TOWNSEND; DOLE; ARBUCKLE, 2003; WANG; SEELBACH; HUGHES, 2006). For this reason streams and the surrounding lands are increasingly seen as “riverscapes” or riverine landscapes (ALLAN, 2004; FAUSCH et al., 2002; SCHLOSSER, 1991). 1.2.1 Streams in human-modified tropical forests In tropical rainforests conversion of natural habits for agriculture and major infrastructure development continues to be the major driver of environmental change (FAO, 2011), with additional perturbations from widespread timber and wood extraction, changes in fire regimes, landscape fragmentation, expansion of second-growth forests, faunal extinction, species invasion, increasing CO2 and climate change (MALHI et al., 2014). Alongside tropical forests encompass the most diverse fish streams yet the poorest known group among vertebrates (ALBERT et al. 2011a). While our understanding of anthropogenic impacts on tropical forest ecosystems has increased in the last decade (GARDNER et al., 2009) many challenges remain in disentangling their effects and understanding the combined effect of multiple disturbances. Different activities can operate in synergy, resulting in cascading effects that can be manifest over larger spatial and time scales. Moreover different landscapes are distinguished by distinct regional contexts (e.g. history of colonization) leading to divergent environmental responses (GARDNER et al., 2009). The understanding of streams hydrological and biogeochemical responses to anthropogenic disturbances has improved in recent decades (NEILL et al. 2006; DAVIDSON et al. 2004; FIGUEIREDO et al. 2010; MACEDO et al. 2013).Where the effects of tropical deforestation on stream systems have been investigated it is evident that there are myriad consequences for changes in 28 stream condition. Terrestrial-aquatic links occur through multiple pathways (e.g. groundwater flow, surface runoff; NEILL et al 2006) and impacts on small watercourses can result in cascading effects on larger river networks (NEILL et al., 2013). Vegetation removal, particularly in the riparian zone, can lead to alterations in runoff from upstream areas, resulting in increased erosion and sedimentation, a rise in light incidence and consequently water temperature, and a loss of organic matter inputs that are a fundamental source of nutrient for aquatic biota in heterotrophic tropical streams (DAVIDSON et al., 2004; FIGUEIREDO et al., 2010; MACEDO et al., 2013; NEILL et al., 2001, 2006, 2011, 2013). However the vulnerability of the physical stream environment and fish fauna to land use change elsewhere in the catchment remains as a major knowledge gap for streams in human-modified tropical forests. Profound changes in freshwater ecosystems can often have a negative impact on the provision of key ecosystem services, such as the buffering of flood waters, the maintenance of water flow during dry periods, and maintenance of water quality through natural filtration and treatment (BRAUMAN et al., 2007; GREGORY et al., 1991; MILLENNIUM ECOSYSTEM ASSESSMENT, 2005) often increasing environmental vulnerability and hazard to human populations. For instance in Rio de Janeiro metropolitan region, Brazil, land use change together with natural resource exploitation and a mountainous relief has resulted in frequent landslides events in the last decade (SMYTH; ROYLE, 2000). Although far less studied than abiotic responses, changes in stream biota following anthropogenic disturbance are likely to be pervasive. Alterations in energy input from the adjacent vegetation (e.g. leaves, large wood debris, small branches, fruits, flowers) in combination with an increase in primary production due to a loss in channel shading (e.g. algae and macrophytes) can result in shifts in trophic groups and assemblage composition. In Madagascan rainforests, for instance, endemic stream insects have been shown to be particularly vulnerable 29 to changes in food resources and declined in abundance and biomass in deforested landscapes (BENSTEAD; PRINGLE, 2004; BENSTEAD et al., 2003). The composition of fish assemblages responded to deforestation in Ecuadorian Amazonian streams with shifts from omnivorous and insectivorous Characiformes in forested areas to periphytivorous Loricariidae in deforested sites (BOJSEN; BARRIGA, 2002). Rises in water temperature due to increased light incidence resulted in changes in the taxonomic composition of fish and benthic macroinvertebrates in Costa Rican forest streams (LORION; KENNEDY, 2008; ORION, 2009). In African streams, the fish-based index of biotic integrity (IBI) followed changes in stream physical-chemical condition, which in turn reflected a loss in catchment forest cover (KAMDEM TOHAM; TEUGELS, 1999). Moreover human disturbance such as deforestation can also result in invasion by exotic fish species (PUSEY & ARTHINGTON 2003). 1.3 The study system: the eastern Brazilian Amazon The Amazonian rainforest is the largest and most biodiverse expanse of tropical forest on Earth and covers 4.5 of the total 7 million km2 of the Amazon River basin drainage area. Spreading across nine South American countries, the Amazonian rainforest is of local and global relevance for the provision of myriad ecosystem services (e.g. biodiversity conservation and climate regulation; MALHI et al. 2008; PERES et al. 2010). The Amazon River basin is the largest in area and discharge in the world being responsible for 1/5 of the world’s freshwater that reaches the oceans (JUNK, 1983). Brazil holds 60% of the river basin, representing 50% of its territory, and has a comparatively large responsibility for its management and conservation. Until 2012 20% of the original Brazilian Amazon forest extent had already been deforested (INPE, 2013). Deforestation has been particularly intense across the eastern and 30 southern regions of the Amazon that experienced a colonization boom in the 1970s. For instance the eastern state of Pará, the second largest state in the Amazon, accounted alone for 34% (ca. 138.000 km2) of the total loss of Amazonian primary forest between 1998 and 2014 (INPE, 2015). Since 2004 deforestation in Pará has been following a declining trend similar to the pattern observed across the Brazilian Amazon as a whole in response to an array of different strategies and dynamics, including policy interventions, private sector initiatives and changes in market conditions (NEPSTAD et al., 2014). Moreover the coverage of protected areas (PA) has rapidly expanded in the last few decades, especially during the late 90s and early 2000s; Brazil now has the largest PA system of all countries in the world covering 12.4% of its territory (WDPA, 2012). However the future of the Brazilian Amazon remains uncertain in the context of ongoing pressures and management challenges. For instance forest degradation caused by selective logging, fire and edge effects, long overlooked in both science and conservation planning and policy, are increasingly recognised as being of comparable importance to deforestation (BARLOW et al., 2012; BERENGUER et al., 2014; LAURANCE et al., 2002). In addition, Brazilian Amazon PAs are now facing new threats, including expansion of the mining and hydropower sectors (FERREIRA et al., 2014), reflected in a region and global trend of downgrading, downsizing and degazettement (BERNARD; PENNA; ARAÚJO, 2014; MASCIA et al., 2014; WATSON et al., 2014). 1.3.1 Amazonian streams and their fish assemblages Beyond the Amazonas river and its main tributaries, the Amazon basin encompasses an immense and complex network of low-order streams – with 1st to 3rd order streams representing up to 90% of the total river length (MCCLAIN; 31 ELSENBEER, 2001) – connecting terrestrial and aquatic ecosystems across the region (JUNK, 1983). These streams drain large portions of upland (terra firme in Portuguese) forest areas that are dependent on local rainfall (compared to the annual flood pulse associated with large floodplain rivers; CARVALHO et al. 2007). Unlike some of the main river channels that originate in the Andes, these small streams are typically nutrient poor, and depend on the adjacent forest for the input of nutrients, organic material flow and for regulation of sediment input (LOWE-MCCONNELL, 1987). Moreover their channels are shaded by dense vegetation, resulting in low primary production and low coverage by algae and macrophytes outside of disturbed areas (Figure 1.1). Figure 1.1 Macrophytes in low-order Amazonian streams due to disturbances to the riparian vegetation (A, B, C; Santarém) in contrast to typically shaded channels (D, E; Paragominas). All pictures taken by Sustainable Amazon Network (Rede Amazônia Sustentável, RAS) Aquatic Team. The Amazon basin hosts an exceptionally diverse freshwater fish fauna, with some 2,200 species currently known (REIS; KULLANDER; CARL J. FERRARIS, 2003), and estimates suggesting that the true value is maybe twice 32 this. The basin is the most biodiverse in South and Central America, which in turn has the most diversified freshwater fish fauna in the world, representing some 10% of all vertebrate species (LÉVÊQUE et al., 2008; LUNDBERG et al., 1998; VARI; MALABARBA, 1998). Although there is not an assessment of the extent to which low-order streams contribute to the total fish diversity of the Amazon basin, there is mounting evidence that they are highly diverse, and host a distinct ichthyofauna, including rare and locally specialised species (CARVALHO; ZUANON; SAZIMA, 2007; MENDONÇA; MAGNUSSON; ZUANON, 2005; ZUANON; BOCKMANN; SAZIMA, 2006). For instance, up to 45 fish species can be registered in a single 50m stream segment (Jansen Zuanon and Rafael Leitão personal communication). Moreover recent studies suggest a high level of species turnover between adjacent Amazonian low-order streams and river basins (ALBERT; BART; REIS, 2011; ALBERT et al., 2011; BARROS et al., 2013; MENDONÇA; MAGNUSSON; ZUANON, 2005). 1.3.2 Conservation of Amazonian streams Conservation planning for tropical freshwater systems that are characterized by a daunting knowledge gap is challenging. Information from better studied freshwater systems, such as temperate streams, may only partly assist. While the basic principles of ensuring catchment-scale protection of native vegetation, maintenance of hydrological and natural flow regimes, and biodiversity conservation are general to all freshwater systems, regional planning and management need to rely on studies tailored to regional conditions (MOULTON; WANTZEN, 2006). Many tropical countries have some type of environmental legislation to protect freshwater systems against deforestation (DUDGEON, 2008). Usually they include restrictions in use of the riparian zone along stream and river 33 networks. However it is long recognized that a catchment-based management and conservation planning system is needed that can account for the importance of different disturbances at different spatial scales (MACEDO et al., 2014; MARZIN et al., 2012; SÁLY et al., 2011; WANG; SEELBACH; LYONS, 2006). The two Brazilian legal instruments directly concerned with freshwater systems are the Fisheries Code (Federal Law No 11.959 June 29th 2009; BRASIL 2009) and the Water Resources Regulation (Federal Law No 9.433 January 8th 1997; BRASIL 1997). The first focuses on aquaculture and fishing activities, and the second on water quality parameters relevant to human consumption. However, both only permit a narrow legal perspective of stream condition and mask the importance of other degradation processes resulting in potentially misleading conclusions about the biotic integrity of stream systems (CASATTI; LANGEANI; FERREIRA, 2006; CASATTI et al., 2006; KARR; DUDLEY, 1981; PAULSEN et al., 2008). The paramount piece of legislation regarding the protection of the broader stream environment, including adjacent native vegetation, is the Forest Code (Federal Law No 12.651, May 25th 2012; BRASIL 2012). The Forest Code prescribes the majority of environmental regulations for private properties, which in turn together encompass approximately 50% of the country’s native vegetation (SOARES-FILHO et al., 2010). It stipulates that 80% of the native vegetation in properties in the Amazon (reduced to 50% in areas that have been zoned for agricultural activities) should be protected in Legal Reserves, with an obligation to restore the forest area back to 50% for areas that were illegally cleared prior to 2008. The law further requires that, depending on the property size, a minimum buffer of riparian vegetation must be protected alongside all water courses – although the revised Forest Code reduced the extent of riparian vegetation that is mandated to be restored to 5 m for areas that have been 34 declared for agricultural use. Our lack of a comprehensive understanding on how different spatial scales and distinct activities interact and affect Amazonian stream condition hinders our ability to inform adequate management and conservation strategies, and evaluate the effectiveness of the available regulations. Brazil is now facing an enormous window of opportunity regarding conservation of the Amazon. Successful efforts from the last decade (e.g. expansion of protected areas and multiple actions to curb deforestation) are threatened by current proposals that would undermine the protection of the biome; it would include for instance allowing mining activities to occur in protected areas (FERREIRA et al., 2014). Furthermore the planned construction of additional hydropower plants will make irreversible and widespread changes in the Amazon freshwater networks. These threats are superimposed by a recent revision of the Forest Code, that clearly meant several steps back on the environmental protection on private properties (GARCIA et al., 2013; SOARES- FILHO et al., 2014) together with a long history of weak law enforcement in this vast biome. The trade-offs between economic development and environmental conservation can still run towards a more sustainable management of the Amazon but it will depend first on the current government environmental attitude (FERREIRA et al., 2014). 1.3.3 The Sustainable Amazon Network This thesis is part of the Sustainable Amazon Network (Rede Amazônia Sustentável, RAS, www.redeamazoniasustentavel.org), a multidisciplinary research initiative focused on assessing social and ecological dimensions of land use sustainability in the eastern Brazilian Amazon (see GARDNER et al 2013 for details). Different from much of the existing work in the Brazilian Amazon, 35 RAS adopted a mesoscale spatial experimental design (i.e. covering hundreds of kilometres and corresponding with the scale of individual municipalities in the country). Studies were conducted in two regions, Santarém (including parts of the municipalities of Santarém, Belterra and Mojuí dos Campos; hereafter STM) and the municipality of Paragominas (PGM), which encompass approximately 1 and 1.9 million ha respectively. The two regions have distinct histories of human land use and occupation. STM has been occupied by Europeans since 1661, whereas PGM was formally established in 1959. However, there are also many similarities. Both regions are relatively consolidated with regards to land use change, with decreasing rates of deforestation of primary vegetation, although planned highways mean that Santarém will probably experience both increased human colonization and agricultural expansion in the near future. Large-scale, mechanized agriculture became established in both regions only in the early 2000s and has increased rapidly in recent years (usually at the expense of both pastures and secondary forest), currently occupying approximately 40,000 and 60,000 ha in Santarém and Paragominas, respectively. Today they are both characterized by a diverse patchwork of well-established mechanized agriculture, extensive and intensive cattle pastures, silviculture (mostly Eucalyptus spp. and Schizolobium amazonicum, especially in Paragominas), densely populated small-land holder colonies and agrarian reform settlements, and large areas of undisturbed and disturbed primary forests and regenerating secondary forests (GARDNER et al. 2013; Figure 1.2). 36 Figure 1.2 Santarém and Paragominas mosaic of land-uses: Eucalyptus sp. monoculture (A), soya plantation (B), small black peppercorn crop (C), primary forest (D) with preserved streams (J), manioc plantation (E) with associated use of streams (F) for manioc flour preparation (G). Logging (H) and fire associated to pasture (I) are also common. Moreover streams are largely used by rural families (L, M, N) including for small-scale hydropower generation (O), as well as cattle watering (K). Pictures D, E, H taken by Érika Berenguer, the other ones by RAS Aquatic Team. The RAS initiative includes a diverse group of research and non-research partners from Brazil and other countries, allowing us to perform a comprehensive assessment of the aquatic condition of low-order streams. For 37 this reason, I am the first author of the three papers to be submitted from this thesis but they will also include other co-authors. My supervisors Dr. Paulo Pompeu (UFLA), Dr. Jos Barlow (Lancaster University) and Dr. Toby Gardner (Stockholm Environment Institute) gave equal indispensable contributions to all steps of this thesis; they were involved since the first plannings of the study (definition of research questions, methodological design, implementation of the field work) through data analysis, results structuring and discussion, and prof reading of all chapters. Moreover all following partners participated one way or another in the data chapters planning and results discussion. MSc. Rafael Leitão (from INPA, Brazil) helped coordinating the aquatic field work, and conducted fish and instream habitat sampling with me. Dr. Jansen Zuanon (INPA) helped in the planning and implementation of the field work and was responsible for overseeing the identification of all fish specimens. Dr. Robert Hughes (Amnis Opes Institute and Department of Fisheries & Wildlife, USA) and Dr. Phil Kaufmann (EPA, USA) have a large amount of experience in planning stream condition assessments, and the analysis of instream habitat and fish responses; RH also participated in field work. MSc. Felipe Rossetti de Paula (ESALq/USP at the time of the study) and Dr. Sílvio Ferraz (ESALq/USP) were responsible for processing the satellite imagery and obtaining the landscape predictor variables. Dr. Jim Thomson and Dr. Ralph Mac Nally assisted with the statistical analysis. Dr. Joice Ferreira, one of RAS coordinators, played a critical role in planning the experimental design and providing support for all aspects of the field work. 1.4 Data sampling The field work was carried out during the dry season (June to August) in two consecutive years, STM in 2010 and PGM in 2011. In each region we 38 sampled a total of 99 stream sites (all 1st to 3rd Strahler order on a digital 1:100,000 scale map) spread across five river basins: Curuá-Una, Tapajós and Amazonas in STM, Capim and Gurupi in PGM, and a gradient of anthropogenic disturbances (Figure 1.3). We had two teams of five people each working simultaneously, resulting in two stream sites sampled per day. Usually each team was composed by three postgraduate students each responsible for instream habitat and fish data; benthos; or adult Odonata and aquatic Heteroptera, and two local field assistants. Figure 1.3 Stream sites from Santarém (A, B, C) and Paragominas (D, E, F) spread across a gradient of anthropogenic disturbances including preserved (A, D), intermediate (B, E) and degraded (C, F) conditions. All pictures taken by RAS Aquatic Team. In both regions, sampling started after two or three weeks of planning, reconnaissance of prospective sampling areas, and training to make sure all teams were familiar with the methods and the region (Figure 1.4). This preparatory work was very important to ensure that we had a complete set of stream samples that encompassed a broad disturbance gradient and to ensure that the time in the field was managed efficiently. Our intensive sampling method 39 required at least 6 to 8 hours in the site, meaning that previous knowledge of the local area, estimated time needed to reach the site, and established contact with landowner were essential. Figure1.4 Aquatic team in the field: reconnaissance of prospective sampling areas and preparatory work for field sampling. All pictures taken by RAS Aquatic Team. Stream sites were chosen based on three main criteria: (i) only one site per stream; (ii) wadeable streams (with a maximum of approximately 1.5 m depth)to ensure an effective sampling; (iii) spread across the entire region to encompass the mosaic of land uses and a gradient of disturbance. A preference was also given to select study sites from the same areas as the RAS terrestrial sampling, although this was not always possible because some areas lacked in low-order streams. 1.4.1 Landscape environmental variables Landscape environmental variables were measured at three different spatial scales (Figure 1.5): 1) the whole catchment upstream from the stream site 40 (‘catchment’), 2) the 100 m buffer along the entire drainage network upstream from the stream site (‘riparian network’), and 3) a 100 m riparian buffer adjacent to the stream site itself (‘local riparian’). Catchment boundaries, mean elevation, and slope were obtained through use of digital elevation models for Santarém (SRTM images with 90 m resolution; NASA - National Aeronautics and Space Administration) and for Paragominas (TopoData with 30 m resolution; INPE - National Institute for Space Research). The drainage network was constructed using the hydrological model ArcSWAT (Soil and Water Assessment Tool extension for ArcGis) for both regions. Figure 1.5 Schematic of the spatial scales considered to obtain the landscape environmental variables. Riparian buffers are referred as network and local. Percentage of forest cover in each of the three spatial scales was obtained from a land use map (Landsat TM and ETM+ images, 30 m resolution, year 2010) (GARDNER et al 2013; Table 1 for a summary of landscape predictor variables). The history of mechanized agriculture was calculated from 41 annual MODIS data from 2001 to 2010 (see details in GARDNER et al 2013). Finally, rivers cape fragmentation was measured using the number of upstream and downstream road crossings within a 5 km circular buffer from the stream site. The road crossings in the drainage network were identified by photo interpretation using georeferenced colour Rapideye images (2010 for STM and 2011 for PGM, 5 m resolution). 1.4.2 Instream habitat Instream habitat is composed of both the physical and chemical characteristics of streams and can be grouped into water properties (hereinafter “water quality”) and physical habitat properties (e.g. substrate type, channel morphology, sinuosity, slope, discharge, wood and cover). To assess the instream habitat we used a protocol first proposed by PECK et al (2006) and HUGHES;PECK (2008), which provides a standardized, replicable and complete assessment of the physical and chemical characteristics of wadeable streams. The resulting dataset enables the calculation of several instream variables representing key aspects of instream habitat such as stream size, stream gradient, substrate size and stability, instream cover complexity, and stream- floodplain connectivity. Instream habitat was sampled prior to fish sampling, in a 150m segment. The stream site was subdivided into 10 continuous sections, 15 m long, by 11 cross-sectional transects (Figure 1.6). Quantitative and qualitative measurements were repeated across transects and along sections according to the method described in the thesis’ chapters and GARDNER et al. (2013; Figure 1.7). 42 Figure 1.6 Schematic of the sampling design of the instream habitat of Amazonian sites. 43 Figure 1.7 Measuring instream habitat characteristics in stream sites: substrate type (A), water properties (B), channel slope (C), depth (D), and canopy density (E). All pictures by RAS Aquatic Team. 1.4.3 Fish Following the instream habitat assessment, three people sampled the 150 m stream segment for 120 min (12 min per section). Each section was isolated using block nets to prevent fish escaping during sampling. Fish were sampled using seines (6 x 1.5 m, 5 mm stretched mesh size) and semi-circular hand nets (0.8 m in diameter, 2 mm stretched mesh size; Figure 1.8). The use of different equipment and collection techniques was applied to encompass all kinds of meso and microhabitats (e.g., riffles, pools, undercut banks, open waters, wood debris, leaf packs, sand, marginal vegetation), and consequently fish groups. All catches were made during daylight hours. Specimens were killed in an anesthetic 44 solution of Eugenol and then fixed in 10% formalin. In the laboratory, all sampled fishes were transferred to 70% alcohol and identified to species level. Voucher specimens are deposited at INPA (Instituto Nacional de Pesquisas da Amazônia) and UFLA (Universidade Federal de Lavras) fish collections, Manaus and Lavras respectively, Brazil. Figure 1.8 Fish sampling in low-order Amazonian streams. All pictures by RAS Aquatic Team. 1.5 Objectives and structure of the thesis The main objective of this thesis is to disentangle and understand the role played by multiple scale anthropogenic disturbances and natural landscape features in changing the condition of Amazonian streams. I used fish and instream habitat field data, and landscape data from analyses of satellite imagery integrated into a single framework of analysis to investigate three inter-related 45 sets of objectives (Figure 1.9). First, to investigate how instream habitat condition changes in response to past local and catchment level anthropogenic disturbances (Chapter 2). Second, to understand the effects of changes to instream habitat, as well as riparian and catchment-scale disturbances on the richness, abundance and composition of fish assemblages (Chapter 3). Third, to understand species-specific responses to the disturbances at different scales, and the implications of scale-dependent responses for current Brazilian environmental legislation for the management or private lands (Chapter 4). All three chapters are prepared for submission to Landscape Ecology (Chapter 2), Ecography (Chapter 3) and Conservation Biology (Chapter 4). 46 Figure 1.9 Methodological framework representing overall thesis structure and links between chapters. 47 REFERENCES ABELL, R. et al. Freshwater Ecoregions of the World : A New Map of Biogeographic Units for Freshwater Biodiversity Conservation. BioScience, v. 58, n. 5, p. 403–414, 2008. ABRAHAM, R. K.; KELKAR, N. Do terrestrial protected areas conserve freshwater fish diversity? Results from the Western Ghats of India. Oryx, v. 46, n. 4, p. 544–553, 2012. ALBERT, J. S. et al. Aquatic Biodiversity in the Amazon: Habitat Specialization and Geographic Isolation Promote Species Richness. Animals, v. 1, n. 4, p. 205–241, 2011. ALBERT, J. S.; BART, H. J.; REIS, R. . Species richness and cladal diversity. In: ALBERT, J. S.; REIS, R. E. (Eds.). . Historical Biogeography of Neotropical Freshwater Fishes. Berkeley, USA: University of California Press, 2011. p. 89–104. ALBERT, J. S.; REIS, R. E. Historical biogeography of neotropical freshwater fishes. [s.l: s.n.]. ALLAN, J. D. Landscapes and riverscapes: the influence of land use on stream ecosystems. Annual Review of Ecology, Evolution, and Systematics, v. 35, n. 1, p. 257–284, 2004. ALLAN, J. D.; ERICKSON, D. L.; FAY, J. The influence of catchment land use on stream integrity across multiple spatial scales. Freshwater Biology, v. 37, p. 149–161, 1997. ARDURA, A. et al. DNA barcoding for conservation and management of Amazonian commercial fish. Biological Conservation, v. 143, p. 1438–1443, 2010. BARLOW, J. et al. The critical importance of considering fire in REDD+ programs. Biological Conservation, v. 154, p. 1–8, 2012. BARROS, D. F. et al. Effects of isolation and environmental variables on fish community structure in the Brazilian Amazon Madeira-Purus interfluve. Brazilian Journal of Biology, v. 73, n. 3, p. 491–499, 2013. 48 BATISTA, V. D. S.; PETRERE JR., M. Characterization of the commercial fish production landed at Manaus, Amazonas State, Brazil. Acta Amazonica, v. 33, n. 1, p. 53–66, 2003. BENSTEAD, J. P. et al. Conserving Madagascar’s Freshwater Biodiversity. BioScience, v. 53, p. 1101–1111, 2003. BENSTEAD, J. P.; PRINGLE, C. M. Deforestation alters the resource base and biomass of endemic stream insects in eastern Madagascar. Freshwater Biology, v. 49, p. 490–501, 2004. BERENGUER, E. et al. A large-scale field assessment of carbon stocks in human-modified tropical forests. Global change biology, v. 20, p. 3713–3726, 2014. BERNARD, E.; PENNA, L. A. O.; ARAÚJO, E. Downgrading, downsizing, degazettement, and reclassification of protected areas in Brazil. Conservation Biology, v. 28, p. 939–950, 2014. BOJSEN, B. H.; BARRIGA, R. Effects of deforestation on fish community structure in Ecuadorian Amazon streams. Freshwater Biology, v. 47, p. 2246– 2260, 2002. BOULTON, A. J. et al. Are tropical streams ecologically different from temperate streams? In: DUDGEON, D. (Ed.). . Tropical Stream Ecology. London, UK: Elsevier, 2008. p. 257–284. BRASIL. Lei No 9.433 de 8 de janeiro de 1997, 1997. Disponível em: BRASIL. Lei No 11.959 de 29 de junho de 2009, 2009. Disponível em: BRASIL. Lei No 12.651 de 12 de maio de 2012, 2012. Disponível em: BRAUMAN, K. A. et al. The nature and value of ecosystem services: an Overview highlighting hydrologic services. Annual Review of Environment and Resources, v. 32, p. 67–98, 2007. 49 BROOKS, T. M. et al. Global biodiversity conservation priorities. Science, v. 313, n. 5783, p. 58–61, 2006. CARVALHO, L. N.; ZUANON, J.; SAZIMA, I. Natural history of amazon fishes, in International Comission on Tropical Biology and Natural Resources. In: CLARO, K. DEL et al. (Eds.). . Encyclopedia of Life Support Systems (EOLSS), Developed under the Auspices of the UNESCO. Oxford, UK: Eolss Publishers, 2007. p. 1–32. CASATTI, L. et al. Stream fish, water and habitat quality in a pasture dominated basin, southeastern Brazil. Brazilian Journal of Biology, v. 66, p. 681–696, 2006. CASATTI, L.; LANGEANI, F.; FERREIRA, C. P. Effects of physical habitat degradation on the stream fish assemblage structure in a pasture region. Environmental Management, v. 38, p. 974–982, 2006. DAVIDSON, E. A. et al. Loss of nutrients from terrestrial ecosystems to streams and the atmosphere following land use change in Amazonia. In: DEFRIES, R. S.; ASNER, G. P.; HOUGHTON, R. A. (Eds.). . Ecosystems and Land Use Change. Geophysical Monograph Series. Washington, D. C.: American Geophysical Union, 2004. v. 153p. 147–158. DIAS, M. S.; MAGNUSSON, W. E.; ZUANON, J. Effects of reduced-impact logging on fish assemblages in central Amazonia. Conservation biology : the journal of the Society for Conservation Biology, v. 24, n. 1, p. 278–86, 2010. DUDGEON, D. et al. Freshwater biodiversity: importance, threats, status and conservation challenges. Biological reviews of the Cambridge Philosophical Society, v. 81, n. 2006, p. 163–182, 2006. DUDGEON, D. Tropical stream ecology. London, UK: Elsevier, 2008. FAO. The State of Forests in the Amazon Basin , Congo Basin and Southeast Asia. [s.l: s.n.]. FAUSCH, K. D. et al. Landscapes to Riverscapes: Bridging the Gap between Research and Conservation of Stream Fishes. BioScience, v. 52, n. 6, p. 483, 2002. 50 FERREIRA, J. et al. Brazil’s environmental leadership at risk. Science (New York, N.Y.), v. 346, n. 6210, p. 706–707, 2014. FIGUEIREDO, R. O. et al. Land-use effects on the chemical attributes of low- order streams in the eastern Amazon. Journal of Geophysical Research, v. 115, p. G04004, 2010. FRISSELL, C. A et al. A hierarchical framework for stream habitat classification: viewing streams in a watershead context. Environmental Management, v. 10, p. 199–214, 1986. GARCIA, L. C. et al. Restoration challenges and opportunities for increasing landscape connectivity under the new Brazilian forest act. Natureza & Conservacao, v. 11, n. 2, p. 181–185, 2013. GARDNER, T. A. et al. Prospects for tropical forest biodiversity in a human- modified world. Ecology Letters, v. 12, n. 6, p. 561–582, 2009. GARDNER, T. A. et al. A social and ecological assessment of tropical land uses at multiple scales : the Sustainable Amazon Network. Phil Trans R Soc B, v. 368, p. 20120166., 2013. GOULDING, M.; BARTHEM, R.; FERREIRA, E. The Smithsonian atlas of the Amazon. [s.l.] Smithsonian, 2003. GREGORY, S. V et al. An ecosystem perspective of riparian zones: focus on links between land and water. BioScience, v. 41, n. 8, p. 540–551, 1991. HERBERT, M. E. et al. Terrestrial reserve networks do not adequately represent aquatic ecosystems. Conservation biology : the journal of the Society for Conservation Biology, v. 24, n. 4, p. 1002–11, 2010. HUGHES, R. M.; PECK, D. V. Acquiring data for large aquatic resource surveys: the art of compromise among science, logistics, and reality. Journal of the North American Benthological Society, v. 27, n. October, p. 837–859, 2008. INPE. Projeto PRODES: Monitoramento da floresta Amazônica Brasileira por satélite. Disponível em: . 51 JUNK, W. J. Aquatic habitats in Amazonia. The Environmentalist, v. 3, n. 5 Supplement, p. 24–34, 1983. KAMDEM TOHAM, A.; TEUGELS, G. G. First data on an Index of Biotic Integrity (IBI) based on fish assemblages for the assessment of the impact of deforestation in a tropical West African river system. Hydrobiologia, v. 397, p. 29–38, 1999. KARR, J. R.; DUDLEY, D. R. Ecological perspective on water quality goals. Environmental Management, v. 5, n. 1, p. 55–68, 1981. LAURANCE, W. F. et al. Ecosystem decay of Amazonian forest fragments: A 22-year investigation. Conservation Biology, v. 16, p. 605–618, 2002. LÉVÊQUE, C. et al. Global diversity of fish (Pisces) in freshwater. Hydrobiologia, v. 595, n. 1, p. 545–567, 2008. LORION, C. M.; KENNEDY, B. P. Relationships between deforestation, riparian forest buffers and benthic macroinvertebrates in neotropical headwater streams. Freshwater Biology, v. 54, p. 165–180, 2008. LOWE-MCCONNELL, R. H. Ecological studies in tropical fish communities. Cambridge: Cambridge University Press, 1987. LUNDBERG, J. G. et al. The Stage for Neotropical Fish Diversification: A History of Tropical South American Rivers. In: MALABARBA, L. R. et al. (Eds.). . Phylogeny and Classification of Neotropical Fishes. Porto Alegre, Brazil: Museu de Ciências e Tecnologia, PUCRS, 1998. p. 13–48. MACEDO, D. R. et al. The relative influence of catchment and site variables on fish and macroinvertebrate richness in cerrado biome streams. Landscape Ecology, v. 29, n. 6, p. 1001–1016, 2014. MACEDO, M. N. et al. Land-use-driven stream warming in southeastern Amazonia. Philosophical transactions of the Royal Society of London. Series B, Biological sciences, v. 368, n. April, p. 20120153, 2013. MALHI, Y. et al. Climate change, deforestation, and the fate of the Amazon. Science, v. 319, p. 169–172, 2008. 52 MALHI, Y. et al. Tropical forests in the anthropocene. Annual Review of Environment and Resources, v. 39, n. 125, p. 159, 2014. MARZIN, A. et al. Ecological assessment of running waters: Do macrophytes, macroinvertebrates, diatoms and fish show similar responses to human pressures? Ecological Indicators, v. 23, p. 56–65, 2012. MASCIA, M. B. et al. Protected area downgrading, downsizing, and degazettement (PADDD) in Africa, Asia, and Latin America and the Caribbean, 1900–2010. Biological Conservation, v. 169, p. 355–361, 2014. MCCLAIN, M. E.; ELSENBEER, H. Terrestrial inputs to amazon streams and internal biogeochemical processing. In: MCCLAIN, M. E.; VICTORIA, R. L.; RICHE, J. E. (Eds.). . The Biogeochemistry of the Amazon Basin. New York: Oxford University Press, 2001. p. 185–208. MENDONÇA, F. P.; MAGNUSSON, W. E.; ZUANON, J. Relationships Between Habitat Characteristics and Fish Assemblages in Small Streams of Central Amazonia. Copeia, v. 4, p. 751–764, 2005. MILLENNIUM ECOSYSTEM ASSESSMENT. Ecosystems and human well- being: wetlands and water - Synthesis. Washington, D. C.: World Resources Institute, 2005. MMA. Portaria No 445 de 17 de dezembro de 2014, 2014. Disponível em: MOULTON, T. P.; WANTZEN, K. M. Conservation of tropical streams - Special questions or conventional paradigms? Aquatic Conservation: Marine and Freshwater Ecosystems, v. 16, p. 659–663, 2006. NEILL, C. et al. Deforestation for pasture alters nitrogen and phosphorus in small Amazonian streams. Ecological Applications, v. 11, n. 6, p. 1817–1828, 2001. NEILL, C. et al. Hydrological and biogeochemical processes in a changing Amazon : results from small watershed studies and the large-scale biosphere- atmosphere experiment. Hydrological Processes, v. 2476, n. December 2005, p. 2467–2476, 2006. 53 NEILL, C. et al. Runoff sources and land cover change in the Amazon: an end- member mixing analysis from small watersheds. Biogeochemistry, v. 105, n. 1- 3, p. 7–18, 2011. NEILL, C. et al. Watershed responses to Amazon soya bean cropland expansion and intensification. Philosophical transactions of the Royal Society of London. Series B, Biological sciences, v. 368, n. 1619, p. 20120425, 2013. NEPSTAD, D. et al. Slowing Amazon deforestation through public policy and interventions in beef and soy supply chains. Science (New York, N.Y.), v. 344, p. 1118–23, 2014. ORION, C. H. M. L. Riparian forest buffers mitigate the effects of deforestation on fish assemblages in tropical headwater streams. Ecological Applications, v. 19, n. 2, p. 468–479, 2009. PAULSEN, S. G. et al. Condition of stream ecosystems in the US: an overview of the first national assessment. Journal of the North American Benthological Society, v. 27, p. 812–821, 2008. PECK, D. V et al. Environmental monitoring and assessment program- surface waters western pilot study: field operations manual for wadeable streams. Corvallis: US Environmental Protection Agency, 2006. PERES, C. A. et al. Biodiversity conservation in human-modified Amazonian forest landscapes. Biological Conservation, v. 143, n. 10, p. 2314–2327, 2010. PUSEY, B. J.; ARTHINGTON, A H. Importance of the riparian zone to the conservation and management. Marine and Freshwater Research, v. 54, p. 1– 16, 2003. REIS, R. E.; KULLANDER, S. O.; CARL J. FERRARIS, J. Check List of the Freshwater Fishes of South and Central America. Porto Alegre, Brazil: EDIPUCRS, 2003. RIBEIRO, M. C. L. B.; PETRERE JUNIOR, M. Fisheries ecology and management of the jaraqui (Semaprochilodus taeniurus, S. insignis) in Central Amazonia. Regulated rivers: research and management, v. 5, p. 195–215, 1990. 54 SÁLY, P. et al. The relative influence of spatial context and catchment- and site- scale environmental factors on stream fish assemblages in a human-modified landscape. Ecology of Freshwater Fish, v. 20, p. 251–262, 2011. SCHLOSSER, I. J. Stream Fish Ecology: A Landscape Perspective. BioScience, v. 41, p. 704–712, 1991. SMYTH, C. G.; ROYLE, S. A. Urban landslide hazards: incidence and causative factors in Niterói, Rio de Janeiro State, Brazil. Applied Geography, v. 20, p. 95–118, 2000. SOARES-FILHO, B. et al. Role of Brazilian Amazon protected areas in climate change mitigation. Proceedings of the National Academy of Sciences of the United States of America, v. 107, n. 24, p. 10821–6, 2010. SOARES-FILHO, B. et al. Cracking Brazil’s Forest Code. Science (New York, N.Y.), v. 344, n. April, p. 363–364, 2014. STRAYER, D. L.; DUDGEON, D. Freshwater biodiversity conservation: recent progress and future challenges. Journal of the North American Benthological Society, v. 29, n. 1, p. 344–358, 2010. TOWNSEND, C. R.; DOLE, S.; ARBUCKLE, C. The influence of scale and geography on relationships between stream community composition and landscape variables : description and prediction. Freshwater Biology, v. 48, p. 768–785, 2003. VARI, R. P.; MALABARBA, L. R. Neotropical ichthyology: an overview. In: MALABARBA, L. R. et al. (Eds.). . Phylogeny and Classification of Neotropical Fishes. Porto Alegre, Brazil: Museu de Ciências e Tecnologia, PUCRS, 1998. p. 1–12. WANG, L.; SEELBACH, P.; HUGHES, R. Introduction to landscape influences on stream habitats and biological assemblages. In: HUGHES, R. M.; WANG, L.; SEELBACH, P. W. (Eds.). . Landscape influences on stream habitat and biological assemblages. [s.l.] American Fisheries Society Symposium 48, 2006. v. 48p. 1–23. 55 WANG, L.; SEELBACH, P. W.; LYONS, J. Effects of levels of human disturbance on the influence of catchment, riparian, and reach-scale factors on fish assemblages. In: HUGHES, R. M.; WANG, L.; SEELBACH, P. W. (Eds.). . Landscape Influences on Stream Habitats and Biological Assemblages. [s.l.] American Fisheries Society Symposium 48, 2006. p. 199–219. WATSON, J. E. M. et al. The performance and potential of protected areas. Nature, v. 515, n. 7525, p. 67–73, 2014. ZUANON, J.; BOCKMANN, F. A.; SAZIMA, I. A remarkable sand-dwelling fish assemblage from central Amazonia , with comments on the evolution of psammophily in South American freshwater fishes. Neotropical Ichthyology, v. 4, n. 1, p. 107–118, 2006. 56 SECOND PART- MANUSCRIPTS 57 MANUSCRIPT 1 MULTI-SCALE ASSESSMENT OF HUMAN-INDUCED CHANGES TO AMAZONIAN INSTREAM HABITATS (Prepared for submission to Landscape Ecology) 58 ABSTRACT Context. Land use change and forest degradation have myriad impacts on tropical ecosystems. Yet their consequences for low-order streams remain very poorly understood, including in the world´s largest freshwater basin, the Amazon. Objectives. We investigated how the physical and chemical characteristics of the instream habitat of low-order Amazonian streams change in response to past local and catchment level anthropogenic disturbances. Methods. We used field data on instream habitat and surrounding landscapes of 99 streams from two regions in the eastern Brazilian Amazon. We conducted random forest regression trees to assess the relative importance of different predictor variables in determining changes in instream condition. Results. Multiple drivers, operating at different spatial scales, were important in determining changes in the physical habitat and water quality of small Amazonian streams. While we found few similarities in modelled relationships between the two regions we did find strong support for non-linear responses of specific instream characteristics to landscape change, including a potential threshold effect of catchment deforestation on water temperature, with a loss of more than 20-30% resulting in consistently warmer streams. Conclusions. Our results highlight the importance of local riparian and catchment-scale forest cover in shaping instream habitat, but also underscore the importance of other land use changes and activities, such as road crossings and upstream agriculture intensification. In contrast to the local and property-scale focus of the Brazilian Forest code, governing environmental regulations on private land, our results reinforce the importance of a rigorously enforced catchment-wide management strategy to protect the integrity of stream ecosystems. Keywords: Anthropogenic impacts.Water quality.Physical habitat.Random forest.Watershed management.Deforestation.Land use change.Freshwater. Amazon basin. Tropics. 59 1 INTRODUCTION Land use change (LUC) is one of the most important factors altering Earth’s ecosystems (Vörösmarty and Shagian 2000; Foley et al 2005; Ellis 2011) and affecting global biodiversity (Butchart et al 2010) and the conservation of ecosystem services (Millennium Ecosystem Assessment 2005; Russi et al 2013). The impacts of LUC are of greatest concern in many parts of the tropics, where major agricultural and infrastructure development are still undergoing rapid expansion, usually at the expense of natural habitats (Davidson et al 2012; Ferreira et al 2014). While our understanding of the impacts of LUC on terrestrial tropical systems has improved significantly in recent decades (Malhi et al 2014), tropical aquatic systems have received far less research attention than terrestrial systems, with the majority of existing work concentrated in a small number of well-studied regions, such as Costa Rica, Puerto Rico, and Australia (Dudgeon 2008). Recent studies in the Amazon provide mounting evidence that LUC impacts to stream hydrobiogeochemistry can occur far beyond the adjacent forest. Terrestrial-aquatic links occur through multiple pathways (e.g. groundwater flow, surface runoff; Neill et al 2006) and impacts on small watercourses can result in cascading effects on larger river networks (Neill et al 2013). The conversion of forests into pasture and croplands is leading to manifold consequences for stream environments, such as changes in water quality (temperature and concentration of nutrients), transport of dissolved and particulate materials, and stormflow (Neill et al 2001; Davidson et al 2004; Neill et al 2006; Figueiredo et al 2010; Neill et al 2011; Macedo et al 2013; Neill et al 2013). Such changes can have marked impacts on the biotic communities of streams, such as the negative impacts of temperature increases on many aquatic 60 groups (Lorion and Kennedy 2009; Orion 2009; Isaak et al 2011; Thomson et al 2012). However one major knowledge gap in our understanding of the ecology of tropical aquatic systems remain virtually unstudied for the Amazonian low- order streams; the vulnerability of the physical stream environment to land use change (Casatti et al 2006a; Dudgeon 2008). Together, physical habitat and water properties constitute the lotic environment of streams (hereinafter called instream habitat), and are frequently used to detect and monitor anthropogenic changes to stream condition (Kaufmann et al 1999). Although changes in the instream habitat have profound effects on biological assemblages and stream condition, our current knowledge of LUC effects on stream physical environments is mostly confined to temperate zones (Hughes et al 2006; Kaufmann and Hughes 2006; Beschta et al 2013) where impacts include bank erosion and sedimentation, alterations in discharge, reduced amount of wood and increases in light incidence (Gregory et al 1991; Allan et al 1997; Sutherland et al 2002; Allan 2004; Hughes et al 2006; Beschta et al 2013; Yeakley et al 2014). Increases in the concentration of fine sediments can reduce the availability of food resources and habitat for fish and invertebrates by covering hard substrates and filling interstitial spaces (Nerbonne and Vondracek 2001). In addition, the loss of riparian vegetation that often accompanies stream degradation can have a negative impact on the provision of key ecosystem services, such as the buffering of flood waters, the maintenance of water flow during dry periods, and maintenance of water quality through natural filtration and treatment (Gregory et al 1991; Millennium Ecosystem Assessment 2005; Brauman et al 2007). In general terms, the responses of tropical instream habitat to LUC are likely to mirror those of temperate streams, because key processes are governed by similar hydraulic mechanisms. For example, changes in channel substrate are 61 influenced by a combination of stream slope, geology, discharge, river bedform, and the presence of large wood and other organic materials. However, the specific nature of such relationships may be different in tropical regions characterized by recent deforestation, rapid increases in mechanization, and high levels of river fragmentation from poorly planned infrastructure developments. These anthropogenic differences are overlain upon the distinct natural characteristics of many tropical streams (e.g. high water temperature at a given elevation, high levels of hydrological periodicity with intense rainfall and runoff, distinct structural features of tropical vegetation) and high natural heterogeneity (Junk and Wantzen 2004; Ortiz-Zayas et al 2005; Boulton et al 2008; Boyero et al 2009). A major research challenge therefore, is to untangle how rapidly changing disturbance processes interact with high levels of natural environmental heterogeneity to influence the structure and diversity of tropical stream habitats in different regions and over gradients of land use change (Ramírez et al 2008; Boyero and Ramírez 2009). To address these issues, we used field data from 99 stream sites distributed across two large regions in the eastern Brazilian Amazon to conduct a multi-scale assessment of the effects of deforestation and land use change on instream habitat for a low-order tropical stream system. We recorded differences not only in physical and chemical water properties, but also aimed for a comprehensive set of physical habitat characteristics of streams, including substrate type and channel morphology among others. The Amazon is the world´s largest remaining area of continuous tropical forest, but is severely threatened by myriad human activities including agricultural expansion, increases in the frequency and intensity of fire, large infrastructure developments (particularly dams and mining), the unsustainable extraction of timber and other forest products, and an unknown number of small dams in small streams resulting from road construction or built to provide water for cattle 62 (Asner et al 2005; Morton et al 2006; Peres and Palacios 2007; Fearnside and Pueyo 2012; Castello et al 2013; Macedo et al 2013; Ferreira et al 2014). Over the past several years, there has been a decrease in annual deforestation in the Brazilian Amazon resulting from, among other factors, several initiatives led by the government with support from NGOs and the private sector, including an increase in law enforcement and punitive actions, an increase in the protected areas network, and the establishment of moratoria on soya and beef from recently deforested areas (Boucher et al 2013; Nepstad et al 2014). However, despite these positive changes, management strategies have largely failed to address the environmental damage caused by deforestation and LUC on the hydrological connectivity of streams (Castello et al 2013). Moreover, legal protection of stream environments and associated riparian vegetation has been diminished following the revision of the Brazilian national Forest Code in 2012 (Federal Law No 12.651; May 25th 2012; Brasil 2012; Garcia et al 2013; Soares- filho et al 2014). The conservation status of small streams is of particular concern because they receive much less research attention and conservation action compared to major river channels and the impacts of large infrastructure developments such as dams. Yet small streams are thought to be the most diverse and extensive ecosystem type in the Amazon basin (Junk 1983; Castello et al 2013). For instance in the Cuieiras River basin, in the central Amazon, first to third order streams represent ca. 92% of the total stream length for the entire basin (Mcclain and Elsenbeer 2001). This study is part of the Sustainable Amazon Network (Rede Amazônia Sustentável, RAS), a multidisciplinary research initiative focused on assessing both the social and ecological dimensions of land use sustainability in the eastern Brazilian Amazon (see Gardner et al 2013). We collected a detailed dataset on instream habitat characteristics of 99 stream sites covering a wide disturbance gradient in two independent regions (Figure 2.1) to answer three 63 specific questions. 1) What are the relationships among natural and anthropogenic characteristics that may influence instream habitat? (e.g., natural controls such as catchment size and slope, and anthropogenic disturbances such as road crossings, mechanized agriculture, and deforestation). 2) Which of these predictor variables explain most of the observed variation in instream habitat condition? 3) Are relationships between landscape-level predictor variables and differences in instream habitat condition consistent between regions? 64 Figure 2.1 Methodological framework to investigate the response of instream habitat of low-order Amazonian stream sites to local and landscape- scale human disturbances (see Table 1). Q1, Q2 and Q3 are the research questions referred to in the Introduction; * see section “Selection of response variables” for detailed steps. 65 2 METHODS 2.1 Study system We studied two regions in the eastern Brazilian Amazon sta