Resilience of Native Vegetation and Bird Species to Smallholder Land Use Impacts in the Valdivian Coastal Range (Chile)

Resilience of native vegetation and bird species to smallholder land use impacts in the Valdivian Coastal Range (Chile)

Thesis submitted in partial fulfillment of the requirements of the degree Doctor rer. nat. of the Faculty of Environment and Natural Resources, Albert-Ludwigs- Universität Freiburg im Breisgau, Germany

Katja Seis

Freiburg im Breisgau, Germany

March 2014

Dean: Prof. Dr. Barbara Koch

First Supervisor: Prof. Dr. Dr. h.c. Albert Reif

Second Supervisor: Dr. Pablo J. Donoso

Second Reviewer: Prof. Dr. Michael Scherer-Lorenzen

Thesis' defence: November 2014

Statement of orginality

I hereby declare that this thesis has never been submitted to another examination commission in Germany or in another country for a degree in a same or similar form. The material in this thesis, to be best of my knowledge, contain no material previously published or written by another person except where due acknowledgement is made in the proper manner.

Katja Seis

Freiburg, March 2014

Acknowledgements

First of all, I would like to express my deep appreciation and gratitude to my first supervisor, Prof. Dr. Dr. h.c. Albert Reif, for giving me the opportunity to join his research team in Freiburg and realize my thesis project. His patience, trust, and unconditional support made it possible to develop and to finish the project. The freedom, honesty, and openness allowed me to work creatively and feel at home in Freiburg.

At the same time, I want to thank Dr. Pablo Donoso, who agreed to be my second supervisor and was especially supportive of my fieldwork in Chile. Without his goal-oriented focus, this thesis would not have been possible.

I would also like to thank Prof. Dr. Michael Scherer-Lorenzen for taking interest in the project and being my second examiner.

Very special thanks to Dr. Stefanie Gärtner. Although not being involved with the project from the very beginning, she nonetheless gave me critical insight in all chapters of the thesis and became an irreplaceable colleague. I really appreciate her help, especially with statistical issues, and for focusing me during important steps of the writing process. I also hope she enjoyed the virtual journey to the Valdivian rainforests.

I thank all of the landowners in Lomas del Sol and Camán for so kindly giving me the permission to work on their land, and particularly for making me feel so welcome during my fieldwork. I especially thank Señora Leonila Velasquez, Don Belarmino Roller, Don Hugo Hernández and Don Luís Barrío for such interesting conversations during my fieldwork and Señora Aselia Valenzuela and her family for their hospitality.

Many thanks to the DAAD (Deutscher Akademischer Austauschdiest) and the LGFG (Landesgraduiertenförderung Baden Württemberg) for their financial support.

5 Besides my supervisors, I would like to thank Dr. Bodo Maria Möseler, Prof. Dr. Holger Kreft, and Prof. Dr. Jürgen Bauhus for supporting my research proposals.

I want to thank my field assistants, Antonia Acuña, Danisa Paredes, Magdalena Gerhardt, Melanie Welling, and Paulina Puchi for their company in the field, for their endurance, and for the precision they brought to their fieldwork.

Thanks to Marco Sepúlveda for his cooperation in the bird survey. He helped me develop the field design for the bird survey, trained me in bird identification, and helped me greatly in the bird census.

Special thanks to Melanie Welling, whose help I appreciate professionally and personally. She is an outstanding vegetation scientist and a very special friend who accompanied me during both of my fieldworks, and made this time truly unforgettable.

I want to thank my colleagues in Valdivia, with whom I shared my office: Andrea Ríos, Daniel Soto, Louise Baillet, Karla Locher, Jorge Moya and Bernarda Ponce. You made me feel at home at the University.

Time has passed, but emotionally I still feel part of our “Waldbau-Institut”. Thanks to all members for being such nice and easygoing colleagues.

I would like to thank my two thesis students, Iván Medel and Magdalena Gerhardt. Your work was essential to understanding the research and our discussions enriched the entire project.

Thanks to Karina Martin, who helped me with any and all administrative tasks in Valdivia.

I would really like to express my special thanks to Marco Florez, who greatly supported my fieldwork and helped me with any questions I had about my study area, even once I was back in Germany.

I would like to thank Daniel Soto, Karla Locher, Dr. Maria Eugenia Solari, Dr. Ocar Thiers, Dr. Victor Gerding, and Dr. Victor Sandoval for helping me in important steps of my thesis with their extensive knowledge of the subject.

Very special thanks to Dr. Carlos Ramírez. We met for the first time during his stay in Germany in 2009. He encouraged me to do this project and he contributed important initial ideas. Furthermore, his

6 patience in helping me to determine my herbarium, and outstanding knowledge on the Chilean flora was invaluable to my fieldwork.

A great thanks also to Felipe Osorio for his helpfulness during the early fieldwork and the constant exchange of ideas during the entire thesis.

I thank Louis Faúndez and Patricio Saldivia from the Universidad de Chile, who patiently introduced me to the beauty of the Chilean flora during our field trip to the Chiloé Archipel in 2009. This was the start of my passion for Chilean vegetation.

I would like to thank everyone in Bonn who supported me from the very beginning in developing and convincing me to realize this thesis, Anke Stein, Dr. Bodo Maria Möseler, Prof. Dr. Holger Kreft, Dr. Jan Henning Sommer, Dr. Jens Mutke, Kathrin Esswein, and Melanie Welling.

Thanks to Bernhard Thiel, who helped improve my English throughout the entire thesis.

Thanks also to Corinna Stück, Helen Desmond, Mike Repasch, and Vanessa Seis for improving my English in various stages throughout this thesis.

Thanks to my esteemed colleagues Angela De Avila, Dr. Cristabel Durán, Dr. Chunling Dai, Dr. Dimitris Samaras, Dr. Jan Bannister, Dr. Nestor Gutiérrez, Dr. Osvaldo Vidal, Dr. Patrick Pyttel, Dr. Rodrigo Vargas Dr. Somidh Saha, and Tamalika Chakraborty for the continued scientific exchange and stimulating discussions regarding all fields of vegetation ecology, statistics, and of course, my thesis.

I thank my lovely colleagues in the office, Juliane Schulze and Julia Sohn for having such a pleasant time at work. I also thank my former officemates, Dr. Carl Höcke and Yildiz Günes.

Furthermore, I want to thank Adam Benneter, Ana Karol, Dr. Chunling Dai, Dr. Cristabel Durán, Dr. Dimitris Samaras, Hendrik Stark, Dr. Jan Bannister, Jörg Kunz, Jörg Niederberger, Mario Dobner, Dr. Nestor Gutiérrez, Dr. Osvaldo Vidal, Dr. Rodrigo Vargas and Victor Meza for being such great colleagues.

I thank Christian Rötsch, Daniel Birkenheier, Esther Muschelknautz, Dr. Germar Csapek, Dr. Martin Kohler, Mathias Frowein, Sixto Jara, and Ursula Eggert for their help with all technical and administrative affairs during the thesis.

7 Thanks to Dr. Cristabel Durán and Marco Sepúlveda for helping me with the Spanish translation of the thesis summary.

I want to thank all my friends, my parents Elke and Lothar, my sister Vanessa, brother in-law Mike, and my entire family for supporting and encouraging me during the whole time of my thesis.

Lastly, I want to thank Marco for his unwavering support, and especially for bearing with me all the way through the final days of the thesis.

Contents

Summary ...... 13

Resumen ...... 17

Zusammenfassung ...... 21

General introduction and synthesis ...... 25

1. Background ...... 25

1.1. Origin of the Valdivian evergreen forests’ biodiversity ...... 26

1.2. Genesis of the cultural landscape in the Valdivian Coastal Range ...... 28

1.3. Structure and objectives ...... 30

2. Publications and contributions of the coauthors ...... 32

2.1. Chapter 1 ...... 32

2.2. Chapter 2 ...... 32

2.3. Chapter 3 ...... 32

3. Synthesis ...... 34

3.1. Which vascular plant and bird species can be preserved in the cultural landscape? ...... 36

3.2. Driving forces for the conservation and resilience in the cultural landscape of the VCR ...... 37

3.3. Distribution of native vascular plant and bird diversity in the landscape ...... 38

3.4. Projections for the future and management implications ...... 40

3.5. Suggestions for conservation management ...... 41

3.6. Recommendations for further research ...... 43

4. References ...... 45

9 Chapter 1: The effect of small-scale land use on vegetation in the Valdivian Coastal Range (Chile) ...... 55

1. Abstract ...... 55

2. Introduction ...... 56

3. Material and methods ...... 59

3.1. Study site ...... 59

3.2. Vegetation and land use assessment ...... 60

3.3. Data analysis ...... 62

4. Results ...... 64

4.1. Description of the vegetation types ...... 64

4.2. Land use drivers of recovery and degradation ...... 68

4.3. A conceptual model for the vegetation dynamics in the cultural landscape in the VCR ...... 70

5. Discussion ...... 71

5.1. Vegetation types comprising the vegetation mosaic ...... 71

5.2. Land use drivers that influence recovery and degradation processes ...... 73

5.3. A conceptual model for the vegetation dynamics in the traditional rural landscape in the VCR .... 75

6. General conclusions ...... 76

7. Acknowledgements ...... 77

8. References ...... 78

Chapter 2: Small-scale cultural landscapes as a refugee for native birds and plants in the coastal range of Chile ...... 85

1. Abstract ...... 85

2. Introduction ...... 86

3. Study site and methods ...... 89

3.1. Vegetation and land use data collection ...... 90

3.2. Bird survey ...... 92

10 3.3. Plant trait data ...... 93

3.4. Data Analysis...... 96

4. Results ...... 99

4.1. Partitioning of plant and bird species diversity ...... 99

4.2. Multivariate relationships of plant species traits and land use indicators ...... 101

5. Discussion ...... 106

5.1. Importance of the hierarchical levels for species diversity conservation ...... 106

5.2. Role of functional diversity as a surrogate for plant and bird diversity ...... 108

5.3. Multivariate changes in functional traits to the main physical structure components of the vegetation and land use...... 109

6. Conclusions ...... 110

7. Acknowledgements ...... 111

8. References ...... 112

Chapter 3:The resilience of the of evergreen rainforest tree species to smallholders land use in the Valdivian Coastal Range (Chile) ...... 121

1. Abstract ...... 121

2. Introduction ...... 122

3. Methods ...... 125

3.1. Study area ...... 125

3.2. Field sampling ...... 126

3.3. Data Analysis...... 127

4. Results ...... 130

4.1. Regeneration types ...... 130

4.2. Response of the regeneration types to land use impacts ...... 132

4.3. Seedling density in different habitats ...... 138

5. Discussion ...... 138

11 5.1. Which regeneration types could be identified? ...... 138

5.2. How do the regeneration types respond to land use impacts? ...... 139

5.3. How does the density of species specific regeneration vary within the cultural landscape? ...... 141

6. General Conclusions ...... 142

7. Acknowledgements ...... 143

8. References ...... 144

Appendix A...... 152

Summary

The Valdivian temperate evergreen rainforest in the Coastal Range of south-central Chile is floristically among the richest temperate rainforests in terms of species and endemism worldwide. Because the forest ecosystem and its species are under high human threat and lack protection, it is listed as one of the 25 global biodiversity hotspots. In particular the massive establishment of forestry plantations has mostly restricted native species to smallholder settlements. Those rural areas with small-scale land use represent a ‘cultural landscape’ comprised of a vegetation mosaic. Such mosaics have been recognized in many parts of the world for their conservation potential. My thesis involved the study of two cultural landscapes in the mid-elevation zones of the Valdivian Coastal Range (VCR), Camán and Lomas del Sol. They are embedded in a broader, plantation dominated landscape matrix. My thesis was divided into three main chapters that aim to understand the impact of smallholders land use on (1) vegetation patterns, (2) related diversity of plants and birds and (3) native tree species regeneration.

The data was sampled in plots randomly placed within three predefined strata: grasslands, shrublands, and forests. The impacts of land use were quantified in each plot using field indicators for livestock browsing, timber harvesting and coppicing.

The first chapter aims to describe the vegetation types (VT), identify land use impacts that lead to either degradation or recovery processes and to provide a conceptual model. The analyses were based on 102 vegetation plots. First VTs were classified using flexible beta clustering and Bray-Curtis distance and visualized in a NMDS. Furthermore, extended indicator species analysis recursive partitioning, and PERMANOVA were used. Finally a conceptual model for the vegetation dynamics was developed from the results. Four VTs were identified (1) extensively grazed non-native grasslands (EGN); (2), closed and semi-closed grazed Ugni and Berberis shrublands (UBS) belonging to the formation of ‘open and shrubby Agrostis capillaris pastures’; (3) severely impacted evergreen forests and; (4) sparsely disturbed evergreen forest (SDE) grouped to the ‘impacted evergreen forest’ formation. The indicators for livestock browsing were important for differentiating the VTs. In EGN grasslands more than 0.075

13 dung piles/m² were found compared to UBS shrublands and the SDE forest. When there were fewer dung piles (< 0.001 dung piles/m²) cutting frequency became important. Furthermore, cutting frequency was significant in determining overall floristic composition. Shrublands and evergreen forests had high native forest species richness.

For the second chapter a bird survey was conducted in a subset of 30 plots out of 105 vegetation plots comprising 10 plots in each stratum. Additive partitioning of beta diversity was applied to analyze the distribution of bird and plant species diversity among three hierarchical levels within the landscape. Additionally, plant functional diversity was correlated with bird diversity. Indicators of land use practices causing changes in plant functional traits and related vegetation structure were identified using RLQ ordination and fourth-corner statistic. The formation level explained most of the gamma diversity for both, plant and bird species. Plant functional richness was a suitable surrogate for bird richness. Changes in traits were most broadly explained by impacts of large herbivores and cutting.

For the third chapter the presence of tree species regeneration was assessed in 85 plots and additionally, individuals were tallied in 30 plots, 10 in each stratum. Regeneration types were classified based on seven nominal traits that may influence regeneration by applying flexible beta clustering and Gower distance. Recursive partitioning within a conditional inference framework was used to explain variation in the regeneration types as a function of the land use variables. Five regeneration types were identified. For the long lived, shade-tolerant endozoochorous broad-leaved (long-tol-endo) and long lived, shade-tolerant, anemochorous broad-leaved (long-tol-ane) regeneration types, the impacts of large herbivores were most restrictive. The short lived, semi shade-tolerant, endozoochorous broadleaved type (short-semi-endo) were restricted by a combination of large herbivores and cutting effects. (2) Short lived, shade-intolerant, anemochorous broad-leaved (short-int-ane) regenerated only in shrublands and regeneration of long lived, semi shade-tolerant, endozoochorous conifers (long-semi- endo) was restricted to forests. Species with the highest seedling densities belonged to the long-tol- endo and the long-tol-ane regeneration type. And seedling densities were highest in forests and nearly absent in grasslands.

The cultural landscape in the VCR provides habitat for native biodiversity. However, the present land use regime caused a vegetation dynamic dominated by disturbance rather than recovery processes

14 driven by cutting and grazing. Forest remained the habitat with the highest bird diversity and potential for tree species regeneration. Shrublands had the potential to serve as habitat for many native plant and bird species and provided niches for tree species regeneration. They may therefore comprise important corridors. Conservation efforts should include cultural landscapes and especially those forest and shrubland elements that serve as a connecting network. Forest restoration and maintenance requires protection from livestock and sustainable forestry. Sustainable land use practices have to be developed with landowners who need to be financially compensated. Research should focus on species that depend on undisturbed forest habitat conditions and a large-scale study including the forestry plantation matrix is needed.

Resumen

El bosque lluvioso templado siempreverde de Valdivia localizado en la región central-sur de la costa de Chile es uno de los bosques lluviosos a nivel florístico más ricos en especies y endemismo en el mundo. El VTER y sus especies constituyen uno de los 25 hotspots de biodiversidad del planeta por estar altamente amenazados por el impacto antropogénico y debido a la falta de protección. El establecimiento masivo de plantaciones forestales es la mayor razón por la cual las especies nativas han sido restringidas a los asentamientos campesinos. Estas áreas rurales bajo un uso de tierra a pequeña escala representan “paisajes culturales” que comprenden mosaicos de vegetación. Estos mosaicos son reconocidos en diversas partes del mundo por su potencial para la conservación. El área de estudio de mi tesis fue localizada en dos paisajes culturales en las elevaciones intermedias de la región costera de Chile (VCR*) Camán y Lomas de Sol, enclavadas en una amplia matriz de paisaje dominada por plantaciones forestales. La tesis fue dividida en tres capítulos principales que tienen como objetivo entender el impacto del uso de la tierra por parte de pequeños agricultores en (1) los patrones de la vegetación, (2) la diversidad de plantas y aves relacionada a éstos patrones y (3) la regeneración de especies arbóreas nativas.

La información de campo fue muestreada en parcelas que fueron ubicadas al azar en los estratos predefinidos: pastizales, arbustales y bosques. El impacto de uso de la tierra fue cuantificado en cada parcela usando indicadores para el ramoneo del ganado, la tala y aprovechamiento del retoño de árboles talados.

El primer capítulo tiene como objetivo la descripción de los tipos de vegetación (VT*) y la identificación de los impactos del uso de la tierra que conllevan a procesos de degradación y de recuperación del bosque y que proporcionan un modelo conceptual. Los análisis fueron basados en 102 parcelas de vegetación. Primero, las VTs fueron clasificadas usando análisis de “cluster” con el método de “flexible Beta” y el índice Bray-Curtis de semejanza, y fueron visualizadas en una ordenación NMDS. Luego fueron usados el análisis de especies indicadoras extendido, el análisis de partición recursivo y

17 “PERMANOVA”. A partir de los resultados un modelo conceptual de la dinámica de la vegetación fue desarrollado.

Cuatro VTs fueron identificados: (1) pastizales de especies no nativas con pastoreo extensivo (EGN*), (2) arbustales pastoreados cerrados y semi-cerrados de Ugni y Berberis (UBS*) pertenecientes a la formación de “pastizal abierto de Agrostis capillaris con arbustos”, (3) bosques siempreverde severamente perturbados, y (4) bosques siempreverde levemente perturbados perteneciente a la formación “bosque siempreverde perturbado”. El indicador de ramoneo del ganado fue importante para la clasificación de las parcelas en VTs. Los pastizales ENG fueron impactados con de más de 0.075 montículos de estiércol/m² en comparación con los arbustales UBS y los bosques siempreverde levemente perturbados. Cuando las densidades de los montículos de estiércoles fueron menor (< 0.001 montículos de estiércol/m²), la frecuencia de corte de rebrotes pasó a un factor importante. Sin embargo, la frecuencia de corte fue un indicador importante determinando la composición florística en toda el área de estudio. Los arbustales y los bosques siempreverdes tuvieron una riqueza mayor de especies nativas.

Para el segundo capítulo se realizó un monitoreo de aves en el área de estudio en 30 de las 105 parcelas de vegetación, donde 10 parcelas fueron distribuidas en cada estrato. El análisis de partición aditiva de la diversidad Beta fue realizado con el objetivo de examinar la distribución de aves y especies de plantas en los tres niveles jerárquicos del paisaje. Adicionalmente se investigó la correlación entre la diversidad de especies y la diversidad funcional. Indicadores de uso de la tierra que causan cambios en la estructura de la vegetación y los rasgos funcionales de las plantas fueron identificados usando una ordinación “RLQ” y el estadístico “fourth-corner”. El nivel “formación” en la jerarquía del paisaje analizada es el que mejor explica la diversidad Gamma de especies de plantas y aves. La riqueza funcional de las especies de plantas resultó ser un sustituto adecuado para inferir sobre la riqueza de aves. El impacto de herbívoros grandes en la vegetación y el corte de rebrotes explicaron mayormente los cambios en los rasgos funcionales de las plantas.

En el tercer capítulo de la presente tesis la presencia de la regeneración arbórea fue evaluada en 85 parcelas, y en 10 parcelas en cada estrato los individuos fueron contados. El tipo de regeneración fue clasificado con un análisis de “cluster” con el método de “flexible Beta” y el índice Gower de semejanza

18 basado en siete rasgos nominales que probablemente tienen una influencia en la regeneración arbórea. El análisis de partición recursivo bajo un marco de inferencia condicional fue usado para explicar la variación de los tipos de regeneración arbórea como respuesta de las variables de uso de la tierra. Cinco tipos fueron identificados.

El impacto de grandes herbívoros domesticos en la vegetación resultó ser el más restrictivo para los tipos de regeneración de especies de vida larga, tolerantes a la sombra, de hoja ancha y de dispersión de semillas endozoocora (long-tol-endo*). En el mismo modo fue afectada la regeneración de especies de vida larga, tolerantes a la sombra, de hoja ancha y de dispersión anemocora (long-tol-ane*). El tipo de regeneración de especies de vida corta, semi-tolerantes a la sombra, de hoja ancha y de dispersión endozoocora (short-semi-endo*) resultó estar limitado por la combinación del impacto de los herbívoros grandes y el corte de rebrotes. Las especies arbóreas de vida corta y de hoja ancha, intolerantes a la sombra y de dispersión anemocora (short-int-ane*) regeneraran sólo en los arbustales. La regeneración de especies coníferas de vida larga, semi-tolerantes a la sombra y de dispersión endozoocora (long-semi-endo*) estuvo restringida a los bosques. Las especies con la mayor densidad de regeneración arbórea pertenecen a los tipos de regeneración long-tol-endo y long-tol- ane. La mayor densidad de la regeneración arbórea fue encontrada en los bosques y fue casi ausente en los pastizales.

El paisaje cultural en el VCR provee de hábitats para la biodiversidad de plantas del bosque nativo. Sin embargo, el régimen actual de uso de tierra ha causado una dinámica de la vegetación dominada por los procesos de perturbación en vez de los de regeneración, los primeros determinados por el pastoreo y el corte de rebrotes. Los bosques son los hábitats con la mayor diversidad de aves y el mayor potencial para la regeneración de especies arbóreas. Los arbustales tienen el potencial de servir como hábitat para muchas especies nativas de plantas y de aves y provee nichos para la regeneración de especies arbóreas. Tanto los bosques como los arbustales conforman importantes corredores de biodiversidad. Por lo anteriormente mencionado, los paisajes culturales deberían tomarse en cuenta en las estrategias de conservación, especialmente considerando a los pastizales y bosques como una red. La recuperación de bosques y su mantenimiento requiere de la protección del ganado a través de su uso sostenible. Prácticas sostenibles del uso de la tierra deben ser desarrollados junto con los

19 pequeños agricultores y ser compensadas económicamente. Futuros estudios deben enfocarse en las especies que dependen de condiciones de hábitats de bosque no perturbados, considerando también la matriz de paisaje dominada por plantaciones a una mayor escala.

* todas las abreviaturas por sus siglas en inglés.

Zusammenfassung

Der Valdivianische Regenwald in der süd-zentral-chilenischen Küstenkordillere gehört floristisch zu den artenreichsten temperaten Wäldern der Erde, auch aufgrund seiner hohen Anzahl endemischer Arten. Da dieses Waldökosystem und seine Biota stark durch den Menschen bedroht und nur unzureichend geschützt sind, zählt es zu den 25 globalen Biodiversitäts-Hotspots. In der Valdivianischen Küstenkordillere (VCR*) hat insbesondere die massive Etablierung von Forstplantagen die heimischen Arten in kleinbäuerliche Siedlungsgebiete gedrängt. Diese ländlichen Siedlungsräume stellen eine mosaikartige Kulturlandschaft dar.

Solche Vegetationsmosaike sind in vielen Teilen der Erde für ihren Beitrag zum Naturschutz anerkannt. Die vorliegende Doktorarbeit untersucht zwei Kulturlandschaften in den mittleren Höhenlagen der VCR, Camán und Lomas del Sol. Diese beiden Gebiete sind eingebettet in eine ausgedehnte Forstplantagenmatrix. Inhaltlich gliedert sich diese Doktorarbeit in drei Hauptkapitel, mit den Zielsetzungen, den Einfluss der kleinbäuerlichen Landnutzung auf (1) die Vegetationsmuster, (2) damit verbundene Diversität von Pflanzen und Vögeln und (3) die Verjüngung der heimischen Baumarten zu verstehen.

Daten zu Vegetation, Baumstruktur, Avifauna und Verjüngung wurden in Aufnahmeflächen erfasst, die randomisiert in drei zuvor definierten Straten platziert wurden: Grasland, Gebüsch und Wald. Mit Hilfe von Feldindikatoren für Beweidung, Holzeinschlag und Niederwaldnutzung wurde Einfluss der Landnutzung quantifiziert.

Das erste Kapitel hat zum Ziel Vegetationstypen (VT*) zu beschreiben, sowie Landnutzungseinflüsse, die zu Degradierungs- oder Regenerationsprozessen führen, zu identifizieren und aus den Ergebnissen ein konzeptionelles Modell abzuleiten. Die hierzu durchgeführten Analysen basierten auf 102 Vegetationsaufnahmen. Zunächst wurde eine Klassifizierung der VTs mit der Flexible-Beta- Clustermetode und Bray-Curtis Distanzmaß durchgeführt, und anschließend die Ergebnisse mittels NMDS visualisiert. Außerdem wurden eine erweiterte Indikatorartenanalyse, Classification tree

21 Analyse und PERMANOVA angewandt. Schließlich wurde ein konzeptionelles Modell für die Vegetationsdynamik aus den Ergebnissen abgeleitet. Vier Vegetationstypen wurden dabei identifiziert: (1) extensiv beweidetes nicht heimisches Grasland (EGN*); (2) geschlossenes und halbgeschlossenes beweidetes Ugni und Berberis Gebüsch (UBS*), zu der Formation der “offenen und verbuschten Agrostis capillaris Weiden“ gehörend; (3) stark beeinflusster immergrüner Wald sowie (4) wenig beeinflusster immergrüner Wald (SDE), zur Formation der “beeinflussten immergrünen Wälder“ gehörend. Die Ergebnisse zeigten, dass die Beweidungsindikatoren die wichtigsten Variablen für die Differenzierung der Vegetationstypen waren. Insgesamt 0,075 Dung Pellets/m² unterschieden das EGN Grasland vom UBS Gebüsch und dem SDE Wald. Waren weniger Dung Pellets vorhanden (< 0.001 Faeces/m²), wurde die Frequenz der Holzeinschläge wichtig. Diese bestimmte die gesamte floristische Artenzusammensetzung signifikant. Insgesamt waren die UBS Gebüsche und die beeinflussten immergrünen Wälder am artenreichsten.

Für das zweite Kapitel wurde die Avifauna in einer Auswahl von 30 Untersuchungsflächen erfasst, von denen jeweils zehn Untersuchungsflächen in einem Stratum lagen. Hierbei wurde die Additive Partitionierung von Betadiversität angewandt, um die Verteilung der Vogel- und Pflanzenarten auf drei hierarchischen Landschaftsebenen zu analysieren. Zusätzlich wurde die funktionelle Pflanzendiversität mit der Vogeldiversität korreliert. Indikatoren der Landnutzung, die Veränderungen in den funktionalen Merkmalen der Pflanzen und der hiermit zusammenhängenden Vegetationsstrukturen verursachten, wurden mittels RLQ Ordination und Fourth-Corner Statistik identifiziert. Die Formationsebene war ausschlaggebend für die Gammadiversität der Pflanzen- und Vogelarten. Die funktionelle Pflanzendiversität war ein geeignetes Maß zur Herleitung der Vogeldiversität. Veränderungen in den funktionalen Pflanzenmerkmalen konnten hauptsächlich durch Einflüsse großer Weidetiere und Holzeinschlag erklärt werden.

Im dritten Kapitel wurde das Vorhandensein der Verjüngung der Baumarten in 85 Untersuchungsflächen erfasst. Zusätzlich wurden alle Jungindividuen in 30 Untersuchungsflächen, 10 Untersuchungsflächen pro Stratum, gezählt. Regenerationstypen wurden auf Basis sieben nominaler funktionaler Pflanzenmerkmale, die die Verjüngung beeinflussen könnten, mittels der Flexible-Beta- Clustermetode und Gower Distanzmaß klassifiziert. Die Variation in den Regenerationstypen als

22 Funktion der Landnutzungsvariablen wurde mittels Classification tree Analyse erklärt. Fünf Regenerationstypen wurden dabei identifiziert. Für die Regenerationstypen langlebiger schattentoleranter endozoochorer Laubbäume (long-tol-endo*) und langlebiger schattentoleranter anemochorer Laubbäume (long-tol-ane*) waren die Einflüsse von großen Weidetieren am restriktivsten. Für die die Regenerationstypen kurzlebiger halbschattentoleranter endozoochorer Laubbäume (short-semi-endo*) war die Kombination von großen Weidetieren und Holzeinschlag restriktiv. Die Arten mit der höchsten Verjüngungsdichte gehörten zu den long-tol-endo und den long- tol-ane Regenerationstypen. Die Verjüngungsdichte war im Wald am höchsten, im Grasland hingegen fast nicht vorhanden.

Insgesamt konnte gezeigt werden, dass die Kulturlandschaft in der VCR wichtige Habitate für die heimische Artenvielfalt bietet. Allerdings verursacht die vorherrschende Landnutzung eine Vegetationsdynamik, die stärker von Holzeinschlag und Beweidung beeinflussten Störungsprozessen dominiert wird, als von Regenerationsprozessen. Die Wälder besitzen die Habitate mit der höchsten Vogeldiversität und mit dem höchsten Potential für die natürliche Verjüngung der Baumarten. Gebüsche können als Lebensraum für viele heimische Pflanzen und Vogelarten dienen, und erhalten Nischen für die Verjüngung der Baumarten. Sie können deshalb als wichtige Korridorelemente angesehen werden.

Aus den Untersuchungen dieser Arbeit lässt sich schlussfolgern, dass Naturschutzmaßnahmen die Kulturlandschaft berücksichtigen sollten. Insbesondere sollten deren Waldelemente berücksichtigt werden sowie die Gebüschelemente, die der Habitatkonektivität dienen. Waldrestauration und -Erhalt setzt den Schutz vor Beweidung und eine nachhaltige Waldwirtschaft voraus. Strategien für eine nachhaltige Landnutzung müssen gemeinsam mit den Landbesitzern entwickelt und finanziell kompensiert werden. Weitere Naturschutzforschung sollte zukünftig insbesondere solche Arten berücksichtigen, die auf ungestörte Waldhabitate angewiesen sind. Ebenfalls sollte die umgebende Forstplantagenmatrix im großen Maßstab untersucht werden.

*Alle Abkürzungen beruhen auf der englischsprachigen Schreibweise

General introduction and synthesis

1. Background

The Valdivian temperate evergreen rainforest (VTER) in south-central Chile is floristically among the richest temperate rainforests worldwide in terms of species and endemism (Arroyo et al., 1997; Smith- Ramírez, 2004). Its biogeographic isolation from other forest ecosystems within the South American continent have eliminated the possibility of recolonization from outside and makes the VTER susceptible to invasion and local extinctions after habitat destruction (Armesto, 1998). Because the forest ecosystem and its species are under high human threat and lack protection it is one of the 25 biodiversity hotspots with the highest conservation priorities in the world (Myers et al., 2000; Olson and Dinerstein, 1998). The greatest threat lies in the dramatic reduction of the region’s native forest starting in the 1970s during the military dictatorship (Armesto, 1998; Myers et al., 2000). Although habitat degradation and loss of the VTER in Chile is ongoing, it has largely been neglected by the public because political and scientific discourse focuses on deforestation in tropical rather than in temperate rainforests (Frey and Lösch, 2010). Today native species of the VTER are mostly restricted to National Parks or rural areas with smallholder settlements. These areas found in most of south-central Chile and South America, represent a ‘cultural landscape’ that “reflect the interactions between people and their natural environment in space and time” (Plachter and Rössler, 1995 p. 15). In the VTER region, the cultural landscapes are mainly comprised of a mosaic of pastures, shrublands and forests (Echeverría et al., 2007b). Such mosaics are largely the result of small-scale farming practices and have produced a variety of habitats that support many taxa and therefore have been recognized worldwide for their biodiversity conservation potential (Armesto et al., 2007; Daily et al., 2001; Deutschewitz et al., 2003; Farina, 1995; Menzies, 2007). Among the different taxa that benefit from this type of landscape mosaic characterized by high habitat heterogeneity, plants and birds have been especially recognized (Tscharntke et al., 2005). However, vegetation patterns, bird diversity and related ecological process may strongly depend on the specific land use impacts and on the

25 inherent site conditions. Impacts of small-scale land use in cultural landscapes have been very well documented in many parts of the world and especially in Europe (e.g. Ellenberg and Leuschner, 2010). However, little research has been done so far in the VTER region. Crucial ecological patterns including species composition and its related diversity of plants and the associated diversity of birds and key ecological processes like tree species regeneration are not very well understood. With my thesis I aimed to provide a better understanding of vegetation patterns in the cultural landscapes of the VTER and how they are related to bird species diversity and native tree species regeneration. I wanted to provide a scientific baseline for a sustainable management approach and for conservation actions done on a landscape scale.

1.1. Origin of the Valdivian evergreen forests’ biodiversity

The reason for the VTER high rate of endemism is founded in its natural history and especially by the fact that it has been geographically separated from other forest ecosystems since Gondwana broke up in the Mesozoic and the Andes were uplifted in the Tertiary (Villagrán and Hinojosa, 1997). Hence the biogeographic barriers of the Atacama Desert in the Chilean north, the Andes mountains in the east and the Pacific Ocean in the west and the harsh climatic conditions in the south led to the evolution of many endemic taxa (Smith-Ramírez, 2004). When the warm climate in the Tertiary gave way to glacial periods in the Quaternary, the geographic range of the southern temperate forests was considerably reduced. This led to local species extinctions including many tropical plant and bird taxa which could not be restored because of the geographic barriers aforementioned (Smith-Ramírez, 2004). However, there were unglaciated elevational zones in the Chilean Coastal Range between 38° and 40° latitude that served as a refuge for VTER species. From these refuges recolonization began 12,000 ago (Ashworth et al., 1991; Villagrán and Hinojosa, 1997; Villagrán, 1991). The recent flora and fauna of the VTER is characterized by south Gondwanian and Neotropical floral elements (Arroyo et al., 1997; Murúa, 1996; Rozzi et al., 1996; Vuilleumier, 1985) and occurs between 38° and 43° S

26 latitude in areas approximately below 800 m.a.s.l. in both the Andes and the Coast Range of south-central Chile (Schmithüsen, 1956; Veblen et al., 1983).

Figure 1. The Coastal Range in south-central Chile within the Xth administrative district (Region X). The Valdivian Coastal Range (VCR) ranges from approximately 39° - 41° S. Source: Smith- Ramírez (2004).

27 Compared to other analogous forests in America the VTER constitutes the center of maximum forest biomass and arboreal species richness (15 tree species/ha (Veblen and Alaback, 1996)) and a broad array of plant life forms including vines and epiphytes (Arroyo et al., 1997) that define its structural complexity. Plant species richness is especially high at the family level including many monogeneric families (e.g. Aextoxicaceae and Gomortegaceae). Since species diversity is higher in the Coastal than in the Andean mountain ranges, the former is considered the most important ecosystem for the conservation of biodiversity in temperate forests in southern South America (Smith-Ramírez, 2004).

However, there are few protected areas in the Coastal Range because in Chile protected areas were located in rather remote and unproductive areas of the Andes or further south in areas which are generally biodiversity poor (Armesto, 1998).Within the Coastal Range at low and mid- elevations (< 700 m. a.s.l.; Smith-Ramírez, 2004) species richness is higher on the eastern than on the western slopes. On the eastern slopes in lower elevations forest conversions have resulted in the near total eradication of native vegetation so that now it is mostly restricted to mid-elevation rural areas where land-use is on a small-scale. Although these areas are most critical for conservation, more and more they are becoming threatened by direct or indirect human impacts due to land use (Armesto, 2002, 1998; Smith-Ramírez, 2004).

1.2. Genesis of the cultural landscape in the Valdivian Coastal Range

Exploitation of the evergreen rainforest in the Valdivian Coastal Range (VCR; Figure 1) began in the 19th century and led to widespread degradation and fragmentation of the forest (Donoso and Otero, 2005). The trees were used to make timber for use by local industries and land was cleared and converted to farmland or to be used for homesteads. In many areas settlers arrived to a “desert” caused by the intensity and extent of these disturbances (personal communication with the landowners, January 2012). This massive exploitation was followed by a phase of regulated logging driven by laws that also saw the establishment of forestry institutions (Donoso

28 and Otero, 2005). In the 1940s plantations of exotic trees began to be established and peaked during the time of the military dictatorship (1973–1990). The large-scale establishment of exotic tree plantations was subsidized by the state to strengthen the forest economy (Nahuelhual et al., 2012). Public and private land including native forest was converted to forest plantations (Echeverria et al., 2006; Lara et al., 2009; Medel, 2013; Nahuelhual et al., 2012; Neira et al., 2002). The area covered by plantations has increased by 55% between 1998 and 2008 in the Valdivian region (116–179 thousand ha; CONAF-CONAMA, 2008). These conversions have led to a monotonous landscape dominated by non-native forest plantations. In this matrix of forest plantations smallholder settlements remain as islands.

On lands once covered by the Valdivian temperate evergreen forest smallholder settlements consisting of a mosaic of grasslands, shrublands and forest patches exemplify a cultural landscape which is comparable to vast areas of southern Chile and other regions of South America (Aravena et al., 2002; Echeverría et al., 2007b). The role of this cultural landscape mosaic for nature conservation as refuges for native species richness is especially important in regions like the VCR where native forests are under threat, protected areas are few and where most of the land has already been converted into a matrix that is hostile to native species (Dorresteijn et al., 2013; Harvey et al., 2008). In such areas the cultural vegetation mosaic (1) can serve as a refuge for plant and animal species and act as stepping stones between remaining larger native forest areas or protected areas (Armesto, 2002) if they would be established); (2) may safeguard the potential to restore forest functions and services (Aravena et al., 2002; Chazdon, 2003; Guariguata and Ostertag, 2001); and (3) may represent a part of a heterogeneous cultural landscape with unique plant and animal communities, but is especially valuable for species depending on openings and edges (Dorresteijn et al., 2013; Foster and Motzkin, 1998).

However, cultural landscapes resulting from small-scale land uses are a fairly recent phenomenon in Chile compared to other regions e.g.in Europe where traditional cultural landscapes have existed for thousands of years (Deil and Scherer, 1996; Oberdorfer, 1960; Scherer and Deil, 1997). Vegetation patterns and ecological processes may depend strongly on

29 site-specific conditions and the particular land use regime. These relationships remain poorly understood in the VCR. We selected two study areas (‘Camán’ and ‘Lomas del sol’) that are representative of the cultural landscapes found in the mid-elevations of the eastern slope of the VCR between 250 – 500 m.a.s.l. Within a broader region, both study areas were imbedded in an area surrounded by forest plantations. The plantations were mainly dominated by highly productive by non-native Eucalyptus globulus and Pinus radiata managed on short rotation cycles (15 – 20 years). Hence plantation management is intensive involving the application of fertilizer and pesticides. The plantations are then harvested by clearcutting (Donoso et al., 2013) but the growing conditions on the large clearcuts may be unsuitable for native species. Near to Lomas del Sol in the ‘Llancahue watershed’ some old-growth native forest has survived only because it was state-owned. However, in the 1990s neighboring farmers began to use this forest illegally for grazing livestock and tree cutting (Moorman et al., 2013). Since the beginning of the massive conversions of forest to forest plantations in the 1970s more and more private farm land including native forest in Camán and Lomas del Sol has been sold to large forestry companies and the people moved to cities. Consequently, the vegetation mosaics in the smallholders settlement areas may vanish in the future. The potential that cultural landscapes may have for the conservation and restoration of native species diversity in the future, including ecosystem goods for the smallholders should not be lost. Therefore, there is an urgent need to gain an understanding about the vegetation patterns, related plant and bird diversity as well as tree regeneration processes against the background of the recent land use scenario.

1.3. Structure and objectives

To understand the vegetation patterns, related to the diversity of plants and birds and the underlying drivers for key ecological processes in the cultural landscapes, my thesis was structured into three main chapters. For the specific ecological background and methods, please refer to the Introduction and Methods parts of each chapter respectively.

30 The 1st chapter (Chapter 1) focuses on the floristic composition of the vegetation mosaic aiming to:

• describe the vegetation types (VT) that comprise the vegetation mosaic in both study areas

• identify land use drivers that lead to either degradation or recovery processes

• provide an explanation for the vegetation mosaic with a conceptual model.

In the 2nd chapter (Chapter 2) biodiversity patterns and the functional relationship between birds and plants are studied identifying:

• the distribution of bird and plant species diversity among hierarchical levels within the landscape,

• the importance of functional diversity as a surrogate for bird species diversity,

• indicators of land use activities that introduced changes in plant functional traits and related vegetation structure.

The 3rd chapter (Chapter 3) addresses regeneration patterns throughout the landscape:

• classifying tree regeneration types (RTs) based on multiple traits among the regenerating tree species,

• testing the response of those a priori defined RTs to variables measuring cutting and livestock impacts and physical vegetation structure,

• comparing the density of species specific regeneration in grassland, shrubland and forest habitat.

31 2. Publications and contributions of the coauthors

The three data chapters (Chapter 1 – 3) are supposed to be published in peer-reviewed scientific journals. Bibliographic specifications are as follows:

2.1. Chapter 1

Seis K, Gärtner S, Reif A, Donoso PJ (2014). The effect of small-scale land use on vegetation in the Valdivian Coastal Range (Chile). Community Ecology. 15(2): 194-204.

2.2. Chapter 2

Seis K, Gärtner S, Sepúlveda M, Reif A, Donoso PJ (in preparation). Small-scale cultural landscapes as a refugee for native birds and plants in the coastal range of Chile. Planed to be submitted to: Biodiversity and Conservation.

2.3. Chapter 3

Seis K, Gärtner S, Donoso PJ, Vargas R, Reif A (in preparation). The resilience of the of evergreen rainforest tree species in the cultural landscape of the Valdivian Coastal Range (Chile). Planed to be submitted to: Forest Ecology and Management.

The concept of the 1st chapter was developed by all coauthors. I elaborated field design together with Albert Reif and did all data assessment in the field. The data analysis was developed by me and Stefanie Gärtner. I conducted the data analysis and wrote the manuscript together with Stefanie Gärtner. All coauthors edited the manuscript.

I developed the concept for 2nd chapter with the help of Stefanie Gärtner, Marco Sepúlveda and Albert Reif. The field design for the bird survey was elaborated by Marco Sepúlveda and me. Vegetation data were the same as assessed for the 1st . The bird survey was conducted by Marco Sepúlveda and me as observers. I compiled the database for the plant trait data. Stefanie Gärtner helped to develop the data analysis. All analyses were conducted by me. I wrote the manuscript with the help of Stefanie Gärtner. All coauthors edited the manuscript.

32 I developed the concept for the 3rd chapter together with Albert Reif, Stefanie Gärtner and Pablo Donoso. I developed the field design together with Albert Reif and Stefanie Gärtner. I assessd the data in the field. The database for the tree trait data was compiled by myself and Pablo Donoso. I developed the concept for the data analysis together with Stefanie Gärtner and Rodrigo Vargas. The analyses were made by myself. I wrote the manuscript with the help of Stefanie Gärtner. All coauthors edited the manuscript.

33 3. Synthesis

My thesis provides insights into the vegetation patterns and ecological processes of the cultural landscape in the Valdivian Coastal Range (VCR). It is the first study to address the impacts of smallholders’ land use practices directly, by quantifying land use indicators such as livestock browsing, timber cutting and coppice forestry in the VCR. The results (Figure 2) strengthen the hypothesis that the vegetation patterns, resulting from land uses by smallholders, provide the conditions required for the diversity of plants and birds as well as for tree regeneration in the VCR.

In appearance, the VCR vegetation mosaic with its forest patches, shrublands and pastures resemble similar heterogeneous mosaics, a type of cultural landscape, found in other parts of the world. These cultural landscape mosaics have been recognized for their high potential for preserving native biodiversity (e.g. Dorresteijn et al., 2013; Foster and Motzkin, 1998). The roles that cultural landscapes play in both biodiversity conservation and peoples’ livelihoods present an opportunity to integrate conservation and economic goals within cultural landscapes (Farina, 2000). But do the cultural landscapes in the Valdivian temperate evergreen rainforest (VTER) region have the prerequisite conditions to play such a role?

Figure 2. Summary of the results of the three data chapters. Field sampling was done in a stratified random design in three formations (grassland, shrubland, forest). Chapter 1 focuses on the floristic composition of the vegetation mosaic. Vegetation types (boxes) were classified and the impact of landuse (grazing and cutting) on the vegetation dynamics (arrows) was tested. In the second chapter the distribution of bird and plant species diversity in the landscape was analyzed. Bird and plant species diversity were linked through functional plant diversity. Traits that changed along the floristic gradient driven by land use impacts were identified. For the third chapter tree regeneration types (RTs) were classified based on traits that may influence regeneration. The tree regeneration types (ellipses) responded differently to land use and altered vegetation structure. Seedling densities were highest in forests and nearly absent in grassland.

35 3.1. Which native vascular plant and bird species can be preserved in the cultural landscape?

Considering the overall floristic inventory, the cultural landscapes studied contain a high number of plant species native to the VTER including many growth forms. Compared with literature references (Donoso, 1989; Oberdorfer, 1960; Veblen and Schlegel, 1982) the main broad-leaved tree species described for the VTER such as Aextoxicon punctatum, Eucryphia cordifolia or Laureliopsis philippiana are present. All growth forms including shrubs, herbs, lianas and one vascular epiphyte (Fascicularia bicolor) were recorded. Conspicuous, however, was the absence of Nothofagus dombeyi and the nearly total absence of Weinmannia trichosperma. Both species are characteristic emergent trees in natural VCR forest stands (Oberdorfer, 1960). Although scattered relict N. dombeyi were observed in the landscape, they were too rare to be recorded with the methodology being applied. On excursions to the field with landowners and local experts, large stumps found were most frequently identified as Nothofagus dombeyi or W. trichosperma. It is likely that the lack of both of these species in the evergreen forest in the Coastal Range is because of their dependence on large openings created by large-scale disturbance events that expose the mineral soil necessary for their regeneration (Lusk, 1999; Veblen et al., 1981). Large-scale disturbances like volcanic eruptions and landslides triggered by earthquakes seldom occur in the Coastal Range (Veblen et al., 1981). These species were likely reduced when the forests were exploited for the first time using high grading practices which almost certainly extracted the largest, most valuable trees.

Variation in functional diversity and the presence of plant communities with varying structural attributes provide a range of habitats for native birds within cultural landscapes (McElhinny et al., 2005). The bird inventory in this study was complete regarding expected native forest bird species. With a total of 28 species recorded it exceeded the large-scale inventory reported for a forest on Chiloé Island (Díaz et al., 2005; 24 bird species). The bird inventory included species that depend on large trees like the Campephilus magellanicus (Magellanic Woodpecker), Pygarrhichas albogularis (White-throated Treerunner) and endemic understory species of the

36 Rhynocryptidae family like Scelorchilus rubecula (Chucao Tapaculo). This species is considered an umbrella species for other threatened Valdivian rainforest taxa including small mammals and marsupials (Castellón and Sieving, 2007).

3.2. Driving forces for the conservation and resilience in the cultural landscape of the VCR

In all chapters of my thesis, land use impacts were identified as the main driving forces behind the development of vegetation patterns, the related diversity of plants and birds and tree species regeneration. Livestock grazing, especially by large herbivores (cattle and horses), produced the strongest impacts. The impacts from livestock may be greater in ecosystems that evolved without mammalian herbivores compared to ecosystems with a long evolutionary history of mammalian herbivory (Milchunas et al., 1998). The only ‘large’ native herbivore in the Coastal Range that could have affected the evolution of the VTER’s flora is the small deer Pudu puda (Vázquez, 2002). However little is known about the ecology of Pudu puda and its browsing effects on forests have been rarely studied (Meier and Merino, 2007). Considering the small size of the animal (< 40 cm tall) it can be assumed that the main impact due to browsing is on vegetation near to the ground in primary forest. Most non-native species, recorded in my thesis, originated in mid-Europe where they grew in open habitats and were adapted to a regime of heavy grazing (Vera, 2002). This may explain the advantage non-native plant species have when browsed by livestock.

Tree harvesting occurred at different intensities ranging from the selective extraction of especially large trees to clear cuts. The effects were accordingly variable. However, there is no doubt, especially when high grading is involved, that important biodiversity components such as old habitat trees get eliminated. On the other hand, coppicing has selected for those tree species that have a high capacity for vegetative regeneration making them resilient to this kind of disturbance, e.g. Eucryphia cordifolia is able to resprout massively after cutting (Donoso, 1989).

37 But the most extreme changes were driven by the impacts resulting from a combination of cutting and grazing which led to changes in vegetation structure and species composition. Echeverría et al. (2007b) found very similar effects in an area adjacent to my study area. The study found that in forest fragments the degrading impacts of cutting and grazing increased as the forest fragment size decreased. The overall results of my thesis also suggest that the combination of grazing and cutting are unsustainable activities in a stand because they hindered tree regeneration.

3.3. Distribution of native vascular plant and bird diversity in the landscape

The results of my thesis suggest that the strongest differentiation occurred between open grassland on the one hand and closed shrubland and forest vegetation on the other. The overall results justify the prosaic reaction of Oberdorfer (1960), who expressed his disappointment when he found how species poor the meadows in the Chilean lake region were which was in striking contrast to the species rich shrublands and forests with their south hemispherical flora (“Aber wie wird der Botaniker enttäuscht, wenn er sich diesen Wiesen mit den gleichen Erwartungen wie zu Hause nähert! Wie kraß sind die Gegensätze! Während Busch und Wald von den fremdartigen Bäumen und Blumengestalten belebt werden, ist der Vordergrund der Wiesen aus einer ganz artenarmen, außerordentlich gleichartig monoton zusammengesetzten Gras- und Krautflora europäischer Herkunft gebildet, der nur hie und da einmal eine Art südhemisphärischer Verbreitung eingefügt ist“ (Oberdorfer, 1960; pp. 164-165)). This dichotomy between species richness in forests versus open grasslands that he observed throughout the cultural landscape could be confirmed in my thesis. Furthermore, it can be extended to native plant species diversity to bird species diversity to a decreasing potential for regeneration and the recovery of native tree species. Contrastingly, in other parts of the world where grasslands have been created and are maintained by small-scale farming they are valued for their species richness which may exceed that of forests (e.g. Eriksson and Cousins, 2014). Nevertheless, the opposite pattern was observed in the cultural landscapes of the VCR. An

38 impoverishment in plant species richness and a dominance of non-native species in open areas maintained by livestock grazing are not uncommon in South American tropical evergreen forests regions (Gutiérrez B. et al., 2012). This can be explained by the fact that plant species of evergreen tropical forests are intolerant of open sites (Turner, 1996). The Valdivian evergreen flora is also adapted to a tall and relatively continuous evergreen canopy which shades the forest interior producing a continuously cool and humid microclimate which is relatively dark (Arroyo et al., 1997). Like plants, temperate rainforest birds in Chile depend on habitats with highly complex structures (Díaz et al., 2005). The highest habitat complexity, indicated through functional diversity, was found in forest. Simplification of the vegetation structure threatens bird diversity. The forest of Llancahue, adjacent to the community of Lomas del Sol, is a best case scenario of a continuous forest providing habitat for forest specialists like woodpeckers. However, it is no longer primary forest due to disturbances caused by smallholders land uses. In many parts of the cultural landscape forests appear as very small fragments. In some areas of Camán, people mentioned that they have not heard a woodpecker in a long time (personal communication, December 2010). Although birds are generally very mobile because of their ability to fly, for many forest birds e.g. understory birds, grasslands not only lack key habitat structures but are an obstacle that they will not cross (Díaz et al., 2006; Willson et al., 2001). For those birds it might mean their restriction to forest fragments within the landscape and the survival of metapopulations may be uncertain. Because mutualistic interactions between birds and plants are of crucial importance for ecosystem processes such as dispersal and pollination (Armesto and Rozzi, 1989; Smith-Ramírez and Armesto, 1994), the failure of birds to move between fragments can also restrict seed dispersal and related forest succession.

When summarizing the value of the vegetation mosaics’ contribution to the maintenance of biodiversity within the landscape, it has to be stated that not all parts are equally suitable. Generally forest habitats are most favorable; however, they may lose their value after high intensity land use. Grasslands are valueless and constitute a risk from a conservation perspective. Additionally, important ecosystem provisioning processes can also be altered in grasslands such as when soils get compacted and organic material is lost. The water cycle is

39 significantly different compared to forests and conditions become dry in spite of the humid macroclimate (Echeverría et al., 2007a; Ramírez et al., 2003; San Martín et al., 2009). Nevertheless, the potential for restoration was not finalized because regeneration patterns were clearly driven by the intensity of the land use impacts. Land abandonment may still allow native tree establishment through ecological succession.

3.4. Projections for the future and management implications

The future of the cultural landscapes in the Valdivian Coastal Range is strongly dependant on the socioeconomic and related demographic developments. General projections for Latin America predict an increase in the abandonment of rural lands due to migration to urban areas (Grau and Aide, 2008). This may lead to natural succession processes towards forest (e.g. Gutiérrez B. et al., 2013, 2012) and an opportunity for ecosystem restoration (Armesto et al., 2007; Grau and Aide, 2008). Some remote areas in central Chile have experienced land abandonment as a result of rural–urban migrations. What followed were dynamic reductions, especially in shrubland cover which allowed forest recovery (Rojas et al., 2013; Schulz et al., 2011). However, the study of Moorman et al., (2010) carried out in Lomas del sol and my own interviews with the landowners in both study areas indicate that landowners would like to keep on living in the countryside. In all cases people were concerned about the ecological implications of plantation forestry, but also by the impacts of their own management of the native forest. Therefore, in the future people will still depend on ecosystem goods and services provided by forests in cultural landscape. For at least three landowners in Camán, it was clear that although their descendants would not keep living on the farms, they still wanted to keep their family properties for recreation. In such cases people may keep on using their land but not be completely dependent on it for their livelihood and in this way allow for passive restoration.

40 3.5. Suggestions for conservation management

Beyond any doubt forests are the key component of the vegetation mosaic. From a purely conservation perspective, our results support the recommendations that call for conservation actions to be taken in the VCR, actions that would immediately protect the remaining forests (Armesto, 1998; Smith-Ramírez, 2004) and eliminate land use in these forests. In cases of land abandonment within a cultural landscape, the land should be prevented from being converted to forest plantations and instead be allowed to revert to native forest. The restoration of forests on former shrublands to forest due to succession not only restores the forest and native biodiversity, it is also a cost effective alternative to the recent increase in the planting of non- native tree species, which is subsidized by the state. On the other hand tree planting on abandoned grasslands may accelerate forest recovery by providing perches for birds (Bustamante-Sánchez and Armesto, 2012) that disperse seeds. Additionally, trees provide shaded microhabitats suitable for seedlings and saplings. Because plantation forest matrices were found to restrict animal movement (i.e Fahrig, 2003; Lindenmayer et al., 2000) a network of shrubland corridors (Castellón and Sieving, 2007) connecting forest fragments within the forest plantation matrix as well as between “forest islands” within cultural landscapes would be necessary for conservation goals.

A conservation strategy advocating the prevention of human impacts in forests within a cultural landscape cannot be justified when local people are dependent on the goods provided by the forests. The establishment of protected areas failed when the needs of adjacent communities were not considered (Berkes, 2004; Borrini-Feyerabend, 1996; Elbers, 2008). Therefore, a compromise has to be found, one that eliminates land use practices that cause serious degradation, especially of forests. This can be achieved by the implementation of more sustainable management practices. Even though such compromises are unlikely to be as effective as strict protection from alternative uses (Arroyo et al., 1997) they still may be promising for conservation purposes especially when compared to a scenario involving the further conversion of land to monoculture uses exemplified by forest plantations. Integrative land use planning should aim to optimize the spatial arrangement of land use types by

41 connecting most large forest areas with shrubland corridors (Gao et al., 2010). The application of near-to-nature management practices that simulate the local disturbance regime through selective low impact cutting practices (Atlegrim and Sjöberg, 2004) is recommended because they may allow vegetation components to persist at the same as having a low impact on the highly endemic flora and fauna (Díaz et al., 2005). However, before such strategies can be implemented, substitutes for the livelihoods of the farmers need to be found and the long-term supervision of the forest management by high quality consultants needs to be arranged. The use of non-timber forest products is a promising additional income source as most of the tree species provide multiple ecosystem goods. These additional income sources could be a solution to the peoples’ dependence on unsustainable timber extraction.

However, there still may be some actions that may be applied, promptly and easily, in the cultural landscapes that could improve the conservation situation but also prevent economic degradation. Forest fragments should become the nucleus of these actions. Forests and shrublands should be fenced where tree regeneration is desired. This may be cost effective for landowners wanting to ensure the natural regeneration of native tree species until at least the sapling stage when they are resilient to the impacts of browsing. There is an urgent need for better management of the grasslands, one that considers specific site conditions. At present, it is shrublands in particular that get cleared in favor of more grassland for livestock grazing. However, when the low nutritional value of the dominant grasses in those pastures (e.g. Agrostis capillaris, Aira caryophyllea) is considered, it becomes apparent that the clearing of shrublands not only reduces the ecological value of the landscape but also the economic value for landowners. Instead landowners should leave those shrublands composed of palatable species uncleared for the grazing of domestic animals. At the same time the shrublands provide corridors for birds (Castellón and Sieving, 2007). A last suggestion would be to promote scattered trees on pastures that would upgrade the grasslands not only by providing shelter for domestic animals, something they may otherwise search for in the forest, but also to provide transitional landscape elements for birds (Tryjanowski et al., 2011).

42 3.6. Recommendations for further research

Although this study constitutes an ecological baseline for an understanding of vegetation patterns and ecological processes in cultural landscapes in the Valdivian Coastal Range (VCR), several gaps of important information have been encountered and many questions considering conservation or sustainable management practices remain unanswered.

An understanding of the functional responses of the vegetation is the key for linking plant species composition patterns and ecological processes (Pérez-Harguindeguy et al., 2013). The analysis done in my thesis was based on information presently available about the plant species studied. However, even in the case of the trees, which get studied more than other life forms, we had difficulties in finding reliable Information to set up a database with consistent trait information for all species. Studies that aim to provide basic knowledge on the autecology of native species should be encouraged. Studies should focus on expanding the existing collection of species specific ecological data and to prepare them in way that is useful for functional trait analysis.

Furthermore, research should focus on the identification of key forest structures within forest patches. This research should consider edge effects created by forest fragmentation in the study area and clearly define thresholds in terms of forest patch size and spatial arrangement for plant and bird species that depend on forest habitat (e.g. Fascicularia bicolor, Campephilus magellanicus, Podocarpus salignus) and may therefore be considered as umbrella species (Castellón and Sieving, 2007). The effectiveness of corridors and buffer zones that shrublands and forest edge structures may constitute an amelioration of the negative-impacts of fragmentation for these species also needs to be further studied.

Lastly the landscape matrix in the VCR consisting of forestry plantations may play a key role determining the movement and population exchange of plants and animals on a landscape scale. Existing studies of the vegetation in plantations (e.g. Ramírez et al., 1984) were not representative of the intensively managed plantations in our study area. A broad-scale study of forest plantations and the tolerance of matrix conditions including various taxa will be of crucial

43 importance for conservation planning on the landscape-scale in the future. Strategies and management that improve matrix permeability for native species needs to be developed.

4. References

Aravena, J.C., Carmona, M.R., Pérez, C.A., Armesto, J.J., 2002. Changes in tree species richness, stand structure and soil properties in a successional chronosequence in northern Chiloé Island, Chile. Rev. Chil. Hist. Nat. 75, 339–360.

Armesto, J.J., 1998. Conservation Targets in South American Temperate Forests. Science 282, 1271–1272. Armesto, J.J., 2002. Relevancia de las pequeñas áreas silvestres para la conservación de la biodiversidad en el bosque nativo. Ambiente Desarro. 18, 44–50.

Armesto, J.J., Bautista, S., Del Val, E., Ferguson, B., García, X., Gaxiola, A., Godinez-Álvarez, H., Gann, G., López-Barrera, F., Manson, R., Núñez-Ávila, M., Ortiz-Arrona, C., Tognetti, P., Williams- Linera, G., 2007. Towards an ecological restoration network: reversing land degradation in Latin America. Front. Ecol. Environ. 5, w1–w4.

Armesto, J.J., Rozzi, R., 1989. Seed dispersal syndromes in the rain forest of Chiloé: evidence for the importance of biotic dispersal in a temperate rain forest. J. Biogeogr. 16, 219–226.

Arroyo, M.T.K., Cavieres, L., Penaloza, A., Riveros, M., Faggi, A.M., Armesto, J., Villagrán, C., 1997. Phytogeographic relationships and species richness patterns in the temperate rain forest flora of South America. Ecol. Los Bosques Nativ. Chile Editor. Univ. Santiago Chile 71–100.

Ashworth, A.C., Markgraf, V., Villagran, C., 1991. Late Quaternary climatic history of the Chilean Channels based on fossil pollen and beetle analyses, with an analysis of the modern vegetation and pollen rain. J. Quat. Sci. 6, 279–291.

Atlegrim, O., Sjöberg, K., 2004. Selective felling as a potential tool for maintaining biodiversity in managed forests. Biodivers. Conserv. 13, 1123–1133.

Berkes, F., 2004. Rethinking community‐based conservation. Conserv. Biol. 18, 621–630.

Borrini-Feyerabend, G., 1996. Collaborative management of protected areas: tailoring the approach to the context. IUCN-The World Conservation Union, Gland, Switzerland.

45 Bustamante-Sánchez, M.A., Armesto, J.J., 2012. Seed limitation during early forest succession in a rural landscape on Chiloé Island, Chile: implications for temperate forest restoration. J. Appl. Ecol. 49, 1103–1112.

Castellón, T.D., Sieving, K.E., 2007. Patch network criteria for dispersal-limited endemic birds of South American temperate rain forest. Ecol. Appl. 17, 2152–2163.

Chazdon, R.L., 2003. Tropical forest recovery: legacies of human impact and natural disturbances. Perspect. Plant Ecol. Evol. Syst. 6, 51–71.

CONAF-CONAMA, 2008. Catastro de uso del suelo y vegetación: Monitoreo y actualización Región de Los Ríos. Gobierno de Chile, Ministerio de Agricultura, Santiago de Chile.

Daily, G.C., Ehrlich, P.R., Sanchez-Azofeifa, G.A., 2001. Countryside biogeography: use of human- dominated habitats by the avifauna of southern Costa Rica. Ecol. Appl. 11, 1–13.

Deil, U., Scherer, M., 1996. Wirtschaftsstufe und Pflanzendecke - Geobotanische Diferenzierung von Landschaften im Mittelmeerraum und in Chile. Erdkunde 50, 81–99.

Deutschewitz, K., Lausch, A., Kuhn, I., Klotz, S., 2003. Native and alien plant species richness in relation to spatial heterogeneity on a regional scale in Germany. Glob. Ecol. Biogeogr. 12, 299– 311.

Díaz, I., Armesto, J.J., Reid, S., Sieving, K., Willson, M., 2005. Linking forest structure and composition: avian diversity in successional forests of Chiloé Island, Chile. Biol. Conserv. 123, 91–101.

Díaz, I., Armesto, J.J., Willson, M.F., 2006. Mating success of the endemic Des Murs’ Wiretail (Sylviorthorhynchus desmursii, Furnariidae) in fragmented Chilean rainforests. Austral Ecol. 31, 13–21.

Donoso, C., 1989. Regeneración y crecimiento en el tipo forestal siempreverde costero y andino tras distintos tratamientos silviculturales. Bosque 10, 69–83.

46 Donoso, P.J., Frêne, C., Flores, M., Moorman, M.C., Oyarzún, C.E., Zavaleta, J.C., 2013. Balancing water supply and old-growth forest conservation in the lowlands of south-central Chile through adaptive co-management. Landsc. Ecol. 29 (2), 245–260.

Donoso, P.J., Otero, L.A., 2005. Hacia una definición de país forestal: ¿Dónde se sitúa Chile? Bosque Valdivia 26, 5–18.

Dorresteijn, I., Hartel, T., Hanspach, J., von Wehrden, H., Fischer, J., 2013. The Conservation Value of Traditional Rural Landscapes: The Case of Woodpeckers in Transylvania, Romania. PLoS ONE 8, e65236.

Echeverria, C., Coomes, D., Salas, J., Rey-Benayas, J.M., Lara, A., Newton, A., 2006. Rapid deforestation and fragmentation of Chilean Temperate Forests. Biol. Conserv. 130, 481–494.

Echeverría, C., Huber, A., Taberlet, F., 2007a. Estudio comparativo de los componentes del balance hídrico en un bosque nativo y una pradera en el sur de Chile. Bosque Valdivia 28, 271– 280.

Echeverría, C., Newton, A.C., Lara, A., Benayas, J.M.R., Coomes, D.A., 2007b. Impacts of forest fragmentation on species composition and forest structure in the temperate landscape of southern Chile. Glob. Ecol. Biogeogr. 16, 426–439.

Elbers, J., 2008. Synthesis of the Case Studies about Participation, Conservation and Livelihoods for Nine Protected Areas in Eight Latin American Countries [WWW Document]. URL http://cmsdata.iucn.org/downloads/ jelbers_synthesis_eepa_2008.pdf (accessed 3.20.14).

Ellenberg, H., Leuschner, C., 2010. Vegetation Mitteleuropas mit den Alpen in ökologischer, dynamischer und historischer Sicht.(6., vollst. neu bearb. und stark erw. Aufl. ed.). Ulmer Verlag, Stuttgart, Germany.

Eriksson, O., Cousins, S., 2014. Historical Landscape Perspectives on Grasslands in Sweden and the Baltic Region. Land 3, 300–321.

Fahrig, L., 2003. Effects of Habitat Fragmentation on Biodiversity. Annu. Rev. Ecol. Evol. Syst. 34, 487–515.

47 Farina, A., 1995. Cultural landscapes and fauna, in: Cultural Landscapes of Universal Value. Gustav Fischer Verlag, Jena, Germany, pp. 60–79.

Farina, A., 2000. The cultural landscape as a model for the integration of ecology and economics. BioScience 50, 313–320.

Foster, D., Motzkin, G., 1998. Ecology and Conservation in the Cultural Landscape of New England: Lessons from Nature’s History. Northeast. Nat. 5, 111–126.

Frey, W., Lösch, R., 2010. Lehrbuch der Geobotanik, Spektrum, 3rd ed. Elsevier, Heidelberg.

Gao, Q., Kang, M., Xu, H., Jiang, Y., Yang, J., 2010. Optimization of land use structure and spatial pattern for the semi-arid loess hilly–gully region in China. Catena 81, 196–202.

Grau, H.R., Aide, M., 2008. Globalization and land-use transitions in Latin America. Ecol. Soc. 2, 16. [online].

Guariguata, M.R., Ostertag, R., 2001. Neotropical secondary forest succession: changes in structural and functional characteristics. For. Ecol. Manag. 148, 185–206.

Gutiérrez B., N., Gärtner, S., Gaviria R., J.C., Meier, W., Reif, A., 2012. Successional vegetation patterns in abandoned pastures of the lower montane cloud forest zone in the Venezuelan Andes. Phytocoenologia 42, 101–132.

Gutiérrez B., N., Gärtner, S., López H., J.Y., Pacheco, C.E., Reif, A., 2013. The recovery of the lower montane cloud forest in the Mucujún watershed, Mérida, Venezuela. Reg. Environ. Change 1069–1085.

Harvey, C.A., Komar, O., Chazdon, R., Ferguson, B.G., Finegan, B., Griffith, D.M., Martínez- Ramos, M., Morales, H., Nigh, R., Soto-Pinto, L., Van Breugel, M., Wishnie, M., 2008. Integrating Agricultural Landscapes with Biodiversity Conservation in the Mesoamerican Hotspot. Conserv. Biol. 22, 8–15.

Lara, A., Little, C., Urrutia, R., McPhee, J., Álvarez-Garretón, C., Oyarzún, C., Soto, D., Donoso, P., Nahuelhual, L., Pino, M., Arismendi, I., 2009. Assessment of ecosystem services as an

48 opportunity for the conservation and management of native forests in Chile. For. Ecol. Manag. 258, 415–424.

Lindenmayer, D.B., McCarthy, M.A., Parris, K.M., Pope, M.L., 2000. Habitat fragmentation, landscape context, and mammalian assemblages in southeastern Australia. J. Mammal. 81, 787– 797.

Lusk, C., 1999. Long-lived light-demanding emergents in southern temperate forests: the case of Weinmannia trichosperma (Cunoniaceae) in Chile. Plant Ecol. 140, 111–115.

McElhinny, C., Gibbons, P., Brack, C., Bauhus, J., 2005. Forest and woodland stand structural complexity: Its definition and measurement. For. Ecol. Manag. 218, 1–24.

Medel, I., 2013. Detecting Dynamic Changes in Vegetation Patterns in Areas of Natural and Semi-natural Vegetation under Different Property Regimes in Southern Chile (Masterthesis). Albert-Ludwigs Universität, Freiburg.

Meier, D., Merino, M.L., 2007. Distribution and habitat features of southern pudu (Pudu puda Molina, 1782) in Argentina. Mamm. Biol. - Z. Für Säugetierkd. 72, 204–212.

Menzies, N.K., 2007. Our forest, your ecosystem, their timber: communities, conservation, and the state in community-based forest management. Columbia University Press. New York.

Milchunas, D.G., Lauenroth, W.K., Burke, I.C., 1998. Livestock Grazing: Animal and Plant Biodiversity of Shortgrass Steppe and the Relationship to Ecosystem Function. Oikos 83, 65–74.

Moorman, M., Donoso, P., Moore, S., 2010. Is Community-based Conservation an Effective Tool in the Llancahue Watershed, Chile?, in: Rethinking Protected Areas in a Changing World: Proceedings of the 2009 GWS Biennial Conference on Parks, Protected Areas, and Cultural Sities. The George Wright Society, Hancock, Michigan, pp. 227–232.

Moorman, M., Donoso, P.J., Moore, S.E., Sink, S., Frederick, D., 2013. Sustainable Protected Area Management: The Case of Llancahue, a Highly-Valued Periurban Forest in Chile. J. Sustain. For. 32 (8), pp. 783-805.Murúa, R., 1996. Avifauna of south american temperate forests, in: Ecología de Los Bosques Nativos de Chile. Editorial Universitaria, Sanatiago de Chile, pp. 113–133.

49 Myers, N., Mittermeier, R.A., Mittermeier, C.G., Da Fonseca, G.A., Kent, J., 2000. Biodiversity hotspots for conservation priorities. Nature 403, 853–858.

Nahuelhual, L., Carmona, A., Lara, A., Echeverría, C., González, M.E., 2012. Land-cover change to forest plantations: Proximate causes and implications for the landscape in south-central Chile. Landsc. Urban Plan. 107, 12–20.

Neira, E., Verscheure, H., Revenga, C., 2002. Chile’s frontier forests: conserving a global treasure. Global Forest Watch, World Resources Institute.

Oberdorfer, E., 1960. Pflanzensoziologische Studien in Chile: Ein Vergleich mit Europa. Flora et vegetatio mundi. J. Cramer, Weinheim, Germany.

Olson, D.M., Dinerstein, E., 1998. The Global 200: A Representation Approach to Conserving the Earth’s Most Biologically Valuable Ecoregions. Conserv. Biol. 12, 502–515.

Pérez-Harguindeguy, N., Díaz, S., Garnier, E., Lavorel, S., Poorter, H., Jaureguiberry, P., Bret- Harte, M.S., Cornwell, W.K., Craine, J.M., Gurvich, D.E., Urcelay, C., Veneklaas, E.J., Reich, P.B., Poorter, L., Wright, I.J., Ray, P., Enrico, L., Pausas, J.G., de Vos, A.C., Buchmann, N., Funes, G., Quétier, F., Hodgson, J.G., Thompson, K., Morgan, H.D., ter Steege, H., Sack, L., Blonder, B., Poschlod, P., Vaieretti, M.V., Conti, G., Staver, A.C., Aquino, S., Cornelissen, J.H.C., 2013. New handbook for standardised measurement of plant functional traits worldwide. Aust. J. Bot. 61, 167–234.

Plachter, H., Rössler, M., 1995. Cultural Landscapes: Reconnecting Culture and Nature, in: Droste, B., Plachter, H., Rössler, M. (eds.), Cultural Landscapes of Universal Value. Gustav Fischer Verlag, Jena, Stuttgart, New York, p. 15.

Ramírez, C., Ellies, A., Mac Donald, R., 2003. Cambios en la flora y la materia orgánica desde bosques nativos a praderas antropogenicas en suelos volcánicos de la IX región de Chile. RC Suelo Nutr Veg 3, 1–12.

Ramírez, C., Figueroa, S.H., Carillo, L.R., 1984. Estudio fitosociologico de los estratos inferiores en un bosque de pino (Valdivia, Chile). Bosque Valdivia 5, 65–81.

50 Rojas, C., Pino, J., Basnou, C., Vivanco, M., 2013. Assessing land-use and -cover changes in relation to geographic factors and urban planning in the metropolitan area of Concepción (Chile). Implications for biodiversity conservation. Appl. Geogr. 39, 93–103.

Rozzi, R., Armesto, J.J., Correa, A., Torres-Mura, J.C., Sallaberry, M., 1996. Avifauna de bosques primarios templados en islas deshabitadas del archipiélago de Chiloé, Chile. Rev. Chil. Hist. Nat. 69, 125–139.

San Martín, C., Villagra, J., Novoa, C., 2009. Comparación de manejos pratenses del Centrosur de Chile utilizando valores bioindicadores de Ellenberg. Gayana Bot 66, 158–170.

Scherer, M., Deil, U., 1997. Floristische Diversität und Vegetationsstrukturen in traditionellen und modernen Kulturlandschaften untersucht an Beispielen aus Chile und dem westlichen Mittelmeergebiet. Z. Für Ökol. Naturschutz 6, 19–31.

Schmithüsen, J., 1956. Die räumliche Ordnung der chilenischen Vegetation. Bonn. Geogr. Abh. 17. 1–89.

Schulz, J.J., Cayuela, L., Rey-Benayas, J.M., Schröder, B., 2011. Factors influencing vegetation cover change in Mediterranean Central Chile (1975-2008): Factors influencing vegetation change. Appl. Veg. Sci. 14, 571–582.

Smith-Ramírez, C., 2004. The Chilean coastal range: a vanishing center of biodiversity and endemism in South American temperate rainforests. Biodivers. Conserv. 13, 373–393.

Smith-Ramírez, C., Armesto, J.J., 1994. Flowering and Fruiting Patterns in the Temperate Rainforest of Chiloe, Chile-Ecologies and Climatic Constraints. J. Ecol. 32. 353–365.

Tryjanowski, P., Hartel, T., Báldi, A., Szymański, P., Tobolka, M., Herzon, I., Goławski, A., Konvička, M., Hromada, M., Jerzak, L., Kujawa, K., Lenda, M., Orłowski, G., Panek, M., Skórka, P., Sparks, T.H., Tworek, S., & A.W., Żmihorski, M., 2011. Conservation of Farmland Birds Faces Different Challenges in Western and Central-Eastern Europe. Acta Ornithol. 46, 1–12.

51 Tscharntke, T., Klein, A.M., Kruess, A., Steffan-Dewenter, I., Thies, C., 2005. Landscape perspectives on agricultural intensification and biodiversity-ecosystem service management. Ecol. Lett. 8, 857–874.

Turner, I.M., 1996. Species Loss in Fragments of Tropical Rain Forest: A Review of the Evidence. J. Appl. Ecol. 33, 200.

Vázquez, D., 2002. Multiple Effects of Introduced Mammalian Herbivores in a Temperate Forest. Biol. Invasions 4, 175–191.

Veblen, T., Alaback, P., 1996. A Comparative Review of Forest Dynamics and Disturbance in the Temperate Rainforests of North and South America, in: Lawford, R., Fuentes, E., Alaback, P. (Eds.), High-Latitude Rainforests and Associated Ecosystems of the West Coast of the Americas, Ecological Studies. Springer New York, pp. 173–213.

Veblen, T.T., Schlegel, F.M., 1982. Reseña ecológica de los bosques del sur de Chile. Bosque 4, 73–115.

Veblen, T.T., Schlegel, F.M., Oltremari, J.V., 1983. Temperate broad-leaved evergreen forests of South America, in: Ovington, J.D. (ed.), Temperate Broad-Leaved Forests, Elsevier, Amsterdam, pp. 5-31.

Veblen, T.T., Z., C.D., Schlegel, F.M., R., B.E., 1981. Forest Dynamics in South-Central Chile. J. Biogeogr. 8, 211–247.

Vera, F.W., 2002. The dynamic European forest. Arboric. J. 26, 179–211.

Villagrán, C., 1991. Historia de los bosques templados del sur de Chile durante el Tardiglacial y Postglacial. Rev. Chil. Hist. Nat. 64, 447–460.

Villagrán, C., Hinojosa, L.F., 1997. Historia de los bosques del sur de Sudamérica, II: Análisis fitogeográfico. Rev. Chil. Hist. Nat. 70, 1–267.

Vuilleumier, F., 1985. Forest birds of Patagonia: Ecological geography, speciation, endemism, and faunal history. Ornithol. Monogr. 36, 255–304.

52 Willson, M.F., Morrison, J.L., Sieving, K.E., De Santo, T.L., Santisteban, L., Díaz, I., 2001. Patterns of predation risk and survival of bird nests in a Chilean agricultural landscape. Conserv. Biol. 15, 447–456.

Chapter 1: The effect of small-scale land use on vegetation in the Valdivian Coastal Range (Chile)

1. Abstract

Today, native vegetation in the Valdivian Coastal Range (VCR) is restricted to areas where small- scale land use dominates resulting in a vegetation mosaic. This study (1) provides a description of the vegetation types (VT) within the vegetation mosaic, (2) identifies land use drivers that lead to either degradation or recovery processes and, (3) attempts to provide an explanation for the vegetation mosaic with a conceptual model.

In two regions of the VCR we sampled 102 plots for composition of vegetation and indicators of livestock browsing, timber cutting and recovery from coppice forestry. We classified the vegetation using a flexible beta method and Bray-Curtis distance. Diagnostic species were identified by an extended indicator species analysis. The clustering results were visualized in NMDS and recursive partitioning was used to explain variations in the VTs as a function of the land use variables. Differentiating effects were tested using PERMANOVA and a conceptual model for the vegetation dynamics was developed from the results.

Four VTs such as (1) extensively grazed non-native grasslands (EGN); (2), closed and semi-closed grazed Ugni and Berberis shrublands; (3) severely impacted evergreen forests and; (4) sparsely disturbed evergreen forest were recognized. The browsing indicators were important for differentiating the VTs. The EGN grasslands were differentiated by having more than 0.075 dung piles/m². Few dung piles but direct browsing effects had the greatest impact on vegetation. Forests were preserved when the mean browsing index was equal to or lower than 0.5. The

55 cutting frequency was significant in determining overall floristic composition. We showed that shrublands and evergreen forests within the vegetation mosaic and the result of small-scale farming had a high native forest species richness. This makes the vegetation mosaic especially valuable in a landscape dominated by exotic tree monocultures.

Keywords: small-scale anthropogenic disturbance, vegetation mosaic, floristic composition, degradation, succession, native forest species, Valdivian evergreen rainforest.

2. Introduction

Human activities are a major cause of vegetation change in contemporary landscapes (Foster and Motzkin 1998, Hobbs et al. 2006, Armesto et al. 2010). Unfavorable socioeconomic conditions often lie behind the conversion of forests to pastures and the degradation of forests to shrublands. Conversely, improvements in socioeconomic conditions often lead to less intensive land use and the abandonment of farmland on which forest returns (Gutiérrez et al. 2012, Gutiérrez et al. 2013, Rojas et al. 2013). These changes in vegetation affect biodiversity and environmental services (Foley 2005).

Cultural landscapes characterized by traditional small-scale farming and livestock grazing support a wide range of more or less artificial to semi-natural vegetation types (VTs). The VTs, created and maintained by particular land use practices, often support high native species diversity (Deutschewitz et al. 2003, Menzies 2007). There are various reasons why the vegetation mosaic created by small-scale land use activities is valuable for nature conservation. The vegetation mosaic (1) serves as a refuge for plant and animal species and connects large native forest areas and protected areas (Armesto 2002); (2) it supports the restoration of forest functions and services during succession (Guariguata and Ostertag 2001, Armesto 2002, Chazdon 2003); and (3) it represents a part of a heterogeneous cultural landscape with unique

plant and animal communities which are especially valuable for species that depend on open and edge conditions (Foster and Motzkin 1998). An understanding of the effects of land use on vegetation patterns and dynamics in areas where the vegetation has been altered by humans at a large-scale is valuable. It provides a scientific baseline for management and conservation of such ecosystems on a landscape scale. Land use modifies biotic and abiotic conditions, thereby affecting species establishment, competition and growth resulting in changes in the floristic composition. Changes in the composition that lead to a decrease in structural complexity and floristic diversity often result in a loss of ecosystem services (Thompson 2011). However, changes may bring about an increase in structural complexity, usually called “succession” as defined by Clements (1916) tends to increase the naturalness of the ecosystems and assist in the recovery of degraded ecosystems. Cultural landscapes and the impacts of small-scale land use have been very well documented in many parts of the world and especially in Europe (e.g. Ellenberg and Leuschner 2010). However, little research has been done in south-central Chile even though it is a global biodiversity hotspot (Myers et al. 2000). It is home to the Valdivian evergreen rainforest, a threatened ecosystem with outstanding biodiversity and endemic species richness (Olson and Dinerstein 1998, Myers et al. 2000). In this region the highest rates of biodiversity and endemism occur on the eastern slopes of the Valdivian Coastal Range (VCR) below 600 m in altitude (Armesto 1998, Smith-Ramírez et al. 2006). In the 1970s the conversion of the Valdivian rainforest into exotic tree plantations began (Armesto et al. 2001, 2010, Armesto 2002, Smith-Ramírez and Armesto 2002,). Today’s remaining rainforest ecosystem is insufficiently represented in the Chilean network of protected areas (SNASPE, Sistema Nacional de Áreas Silvestres Protegidas del Estado; Armesto 1998, Smith-Ramírez 2004, Pliscoff et al. 2005). This lack of protection in a landscape dominated by exotic tree plantations has limited most of the native taxa to extreme sites or to rural areas where disturbances by landowners are on a small-scale.

There are only few rural areas left within the VCR where land use is still carried out on a small- scale. The smallholders mainly use their land for extensive livestock grazing and the selective cutting of trees for firewood. The purchase of land by large forest companies that convert it to

57 exotic tree plantations is ongoing (Neira et al. 2002) and threatens the existence of the evergreen forest in the VCR (Armesto 1998, 2002, Smith-Ramírez 2004). Most Chilean ecosystem studies have been carried out from a forestry perspective, subsequently, the VCR forests were classified on the basis of the composition of the dominant tree species (e.g. Donoso 1981, Gajardo 1994) the same way that many temperate forests were classified. However, this approach is not suitable for obtaining an ecological understanding of the ways in which rural farming systems, with an agroforestry component, affect vegetation and biodiversity on a regional scale. In a pioneer work, Oberdorfer (1960) did a small-scale phytosociological study in which he set out a classification of plant communities. His proposal described a successional pathway from Agrostis pasture and Rhaphithamno-Aristotelietum shrubland to the Laurietalia climax forest. A more intensive survey and thorough description of shrubland communities followed (Hildebrand 1983, Hildebrand-Vogel 2002, Amigo et al. 2007). These studies highlighted the close floristic relationship that existed between various shrub and woodland communities, and suggested that they were the result of land use practices. However this assumption was not further investigated or tested, e.g. by comparing the shrubland communities with other related formations on similar sites. Such a comparison was only done in one study that analyzed the floristic similarity of different formations as an indicator of site degradation on a lee slope in the VCR (Ramírez et al. 1984). The study involved Acaena-Agrostis pastures, Ulex shrublands and Aristotelia chilensis shrublands, Dombeyo-Eucryphietum forests and Pinus spp. plantations. After the Dombeyo-Eucryphietum forests were cut, the area developed into a Aristotelia chilensis shrubland. It was assumed that the original forest would regenerate, although in the same study, no evidence was found for forest regeneration on Acaena-Agrostis pastures when sheep grazing was abandoned.

Based on the previous studies it can be assumed that the vegetation mosaic, in rural areas, can be classified into VTs. We hypothesized that these VTs were the result of, and maintained by land use activities and that furthermore, the VTs were indicators of either succession or degradation processes. In order to test our hypothesis and to increase the understanding of the disturbance and recovery processes we selected the ‘Lomas del Sol’ and the ‘Camán’ regions on

the eastern slopes of the VCR. These study areas were representative of the small-scale land use in this region. We collected extensive vegetation, environmental and land use indicator data with the following objectives in mind:

1. to provide a description of the vegetation types within the vegetation mosaic in the study area;

2. to identify the land use drivers responsible for recovery and degradation processes;

3. and to develop and discuss a conceptual model, based on previous results, that attempts to explain the vegetation dynamics in the cultural landscape of the VCR.

Our results are supposed to contribute to the current debate on the potential of traditional rural landscapes to integrate biodiversity conservation with sustainable land use.

3. Material and methods

3.1. Study site

Our study was conducted on the eastern slopes of the Valdivian Coastal Range (VCR), Chile. Two rural regions with similar climatic and soil characteristics were selected ‘Camán (39° 58' S 73° 00' W)’ and ‘Lomas del Sol (39° 50' S 73° 07 'W)’. The potential natural vegetation of these two regions is Valdivian evergreen forest (Ramírez and San Martín 2005). The climate is temperate, perhumid with an average annual rainfall of about 2,100 mm. Rainfall occurs throughout the year but peaks in winter (April-October). The average annual temperature is 12.2 C with a maximal monthly average of 23.3 C in January and minimal average of 4.6 C in August (Fuenzalida 1965). Soils are well drained Ultisols of volcanic origin on metamorphic bedrock; soil texture varies from silty to sandy loam (CIREN 2001). All sites were located at altitudes between 250 and 500 m. The primary vegetation was Valdivian evergreen forest dominated by ‘Coihue’ (Nothofagus dombeyi) and ‘Ulmo’ (Eucryphia cordifolia); (‘Bosque de Coihue-Ulmo’;Ramírez and San Martín 2005), which was described by Oberdorfer (1960) as the ‘Dombeyo-Eucryphietum’.

59 This forest is especially rich in evergreen woody species (Myrtaceae and Proteaceae) but also in vines, lianas and epiphytes that exemplify its structural complexity. The tallest emergent trees in these forests, attaining a height of 45 m, are N. dombeyi and E. cordifolia (Oberdorfer 1960, Ramírez and San Martín 2005).

Around 1900, Chilean and European settlement began in Camán and Lomas del Sol. Settlement resulted in the exploitation and destruction of the native forests. Much of the forested land was converted to subsistence uses. In Camán a second massive conversion of state-owned forest, this time to exotic tree plantations, began in the 1970s. Additionally, more and more private farm land is being sold to large companies for plantations even though native forest in the Camán region is restricted to private properties. In the Lomas del Sol region a similar conversion of land to exotic tree plantations occurred, but the adjacent forest in the Llancahue watershed with some old-growth native forest survived because it was state-owned (Moorman et al. 2013b). However, in the 1990s neighboring farmers began to use this forest illegally for livestock grazing and tree cutting (Moorman et al. 2013a).

3.2. Vegetation and land use assessment

Stratified random sampling was conducted with three predefined strata based on vegetation height; 1) open pastures (vegetation height < 1 m, scarce presence of woody species); 2) shrublands (height 1 m – 5 m, with greater than 20% shrub cover), and 3) forests (> 5 m tall). To handle the heterogeneity caused by the presence of different formations in the landscape, Modified-Whittaker nested plots (Stohlgren et al. 1995) were established. The plot and subplot sizes were adapted to the tree heights in our study area. Within each tree plot (tree layer (> 5 m) 160 m²) a subplot for the shrub layer (1 m – 5 m) 80 m² and for the herb layer (for the layer < 1 m) was 20 m² was established. The smaller subplots were centered in the middle of the larger plot to capture the influence of the layer above. To cope with the evergreen rainforests vertical complexity, the tree layer was subdivided into a lower tree layer (5 – 10 m) and an upper tree layer (> 10 m). Extreme sites like swamps and excessive slopes were excluded from sampling

because they had not been harvested or grazed. Sampling points were randomly distributed according to the total cover proportional to the area of each stratum in all study regions (Table 1).

Table 1. Distribution of the plots for vegetation sampling among strata and study regions.

Study region

Stratum Camán Lomas del Sol Total

Grassland 23 13 36

Shrubland 17 14 31

Forest 18 17 35

Total 58 44 102

All vascular plant species were identified and their cover was estimated using the Londo decimal scale (Londo 1976). Nomenclature follows the International Plant Names Project (2012). For every layer the cover of each plant species was estimated in each of the subplots. In the subplot, for the tree layers, the diameter at breast height (dbh) of all trees with a dbh > 5 cm and their heights were measured. Terrain variables (altitude, slope and aspect) were recorded and soil variables for the upper soil layer (bulk density, depth of Ah horizon, pH, and texture) were assessed using field methods (Arbeitskreis Standortskartierung 1996). Litter cover was estimated in the 20 m² subplots (same plots where the herbs were measured) in three ordinal classes (1) none, (2) little and (3) thick. In the same subplots the impacts of small-scale land use (livestock farming and cutting) were assessed. Cutting included the extraction of larger trees for timber and coppice for fire wood.

3.2.1. Livestock farming

The number of dung piles/m² has been used as a proxy for grazing intensity as herbivores defecate in the same area in which they feed (Relva and Veblen 1998). Therefore, dung piles

61 were counted in the 20 m² subplots and expressed in dung piles/m². Browsing damage caused by livestock was assessed for the woody species in the herb and shrub layers. The mean browsing index (in the following browse index) was then calculated (Veblen et al. 1989). Dung piles and signs of browsing are variables that describe livestock damage on vegetation which can be readily measured and have been used frequently (Echeverría et al. 2007). As an indicator for trampling, the animal tracks, if present, were measured.

3.2.2. Cutting

Stumps were counted to assess the intensity and frequency of cutting. Deadwood age was estimated with a knife test (Rouvinen et al. 2002) done for each stump. Cutting intensity was then calculated as stumps/m² and cutting frequency as the number of different age classes present in a plot.

3.3. Data analysis

3.3.1. Description of the vegetation types

The floristic group structures in the vegetation data were examined using flexible beta agglomerative algorithm (Lance and Williams 1967) with ß= - 0.25 and Bray-Curtis distance measure, using the R package cluster (Maechler et al. 2013, R Core Team 2013). An appropriate cut level for the dendrogram was chosen by conducting an indicator species analysis (Dufrene and Legendre 1997) at each hierarchical group level. The group level that showed a high number of indicator species along with a low average p-value across all species was chosen for cutting the dendrogram (McCune et al. 2002). The significant floristic differences of the resulting vegetation types were then verified using a multi response permutation procedure from the R package vegan (Oksanen et al. 2013). For these and all subsequent analyzes, plant cover values were transformed from the Londo ordinal scale (Londo 1976) to percentages of the class averages (Leyer and Wesche 2007). We combined the subplots representing the different layers within one plot. Species that occurred at frequencies lower than 5% were excluded from

following analyses. To characterize the vegetation types defined via diagnostic species, the extended indicator species analysis, proposed by De Cáceres et al. (2010) was applied (R package indicspecies; De Cáceres and Legendre 2009). This index identifies diagnostic species not only for each single vegetation type but also for combinations of vegetation types. Species that are diagnostic for more than one vegetation type may therefore indicate broader habitat conditions (De Cáceres et al. 2010) than those that indicate combinations of vegetation types because they are more restricted in their niche breadth to a certain habitat.

To visually explore the floristic gradient in our dataset non-parametric multidimensional scaling (NMDS) was performed on the Bray-Curtis dissimilarity matrix (R package vegan, Oksanen 2013). Land use and environmental variables were fitted on the ordination and plotted on the ordination diagram. The NMDS results link the explorative analysis of the vegetation mosaic (first objective) to the second objective aiming at identifying the land use drivers for land recovery and degradation.

3.3.2. Identification of the land use drivers of degradation and recovery

Classification tree analysis was applied to determine which land use variables were influencing the categorical response variable vegetation type. We used recursive partitioning within a conditional inference framework to explain variation in the response variables as a function of the explanatory variables. This was accomplished with the non-parametric conditional inferences tree methods (function ctree in the R package party, Hothorn et al. 2006b). At each step of the analysis, one explanatory variable was selected from all the available variables, based on the best separation of two homogeneous groups using a permutation test; this point is determined by a numerical value (threshold) of the explanatory variable (Hothorn et al. 2006a, 2006b, Hothorn and Zeileis 2008). The relationships between the response variable and explanatory variables are presented in a dichotomous tree diagram with nodes that represent split points, branches that connect nodes, and leaves or terminal nodes that represent the final groups. To test the effect of all land use variables and their interactions on overall floristic patterns, permutational multivariate analysis of variance (PERMANOVA; Anderson 2001) was performed with the R package vegan (Oksanen et al. 2013) on the floristic

63 dissimilarity matrix (Bray-Curtis distance measure) and the land use variables with 999 permutations. Additionally, we performed a PERMANOVA on pairs of vegetation types to investigate separating effects between groups.

4. Results

4.1. Description of the vegetation types

Based on the floristic composition we identified four vegetation types (VTs) (Figure 1): (1) extensively grazed non-native grasslands (EGN), (2) closed and semi closed Ugni and Berberis shrublands (UBS), severely impacted evergreen forest (SIE) and sparsely disturbed evergreen forest (SDE). The dendrogram shows that two VTs are aggregated in two formations respectively: (1) open and shrubby variants of Agrostis capillaris pastures and (2) variants of the impacted evergreen forests.

Figure 1. Dendrogram for the vegetation data (species and cover) in the study area. The dendrogram was generated with a flexible beta agglomerative algorithm (ß = -0.25). Boxes delimit the vegetation types (EGN = grazed non-native grasslands, SDE = sparsely disturbed evergreen forest, SIE = severely impacted evergreen forest, UBS = closed grazed Ugni and Berberis shrublands).

4.1.1. Open and shrubby Agrostis capillaris pastures

The formation of open and shrubby Agrostis capillaris pastures has 11 diagnostic species (Appendix A). All diagnostic species were herbs and non-native species except for Leptostigma arnottianum. The physiognomy comprised a transition from the grazed non-native (EGN) grasslands to the closed grazed Ugni and Berberis (UBS) shrublands ). The EGN grasslands were indicated by two diagnostic species. The floristic composition was dominated by exotic herbaceous species which formed an herb layer covering, on average, 85% and rarely reached more than 1 m in height. These plots in the EGN grasslands were located on flat terrain (average slope 4°).

65 The Ugni and Berberis shrublands (UBS) had 7 diagnostic species representing different life forms: trees (Embothrium coccineum), shrubs (Ugni molinae, Baccharis racemosa, Genista monspessulana, Berberis microphylla), and herbs (Vulpia bromoides, and Leucanthemum vulgare). Trees with a dbh > 5 cm were rare. The average herb layer cover was 60% while the shrub layer averaged 65% with a maximum height of 5 m.

4.1.2. Impacted evergreen forests

The impacted evergreen forests were represented by plots situated in a stand with trees up to 100 years old. The formation was characterized by 21 diagnostic species (Appendix A) which are typical native tree, shrub, liana and herb species of the evergreen forest.

The severely impacted evergreen forest (SIE) had 6 diagnostic species: Hypericum androsaemum, Blechnum hastatum, Aristotelia chilensis, Azara lanceolata, Ribes trilobum and Greigia sphacelata. The basal area (6.8 m2/ha) and tree density (820 stems/ha) were still relatively low. The tree layer was simple or stratified with a maximum tree height of 7.3 m and a crown closure (11%) that was significantly lower than in the SDE forests.

The sparsely disturbed evergreen forest (SDE) had 5 diagnostic species. Study sites were often located on steeper slopes (12° on average). Basal area and tree densities were highest among the four VTs (basal area = 41 m²/ha, 2500 stems/ha). The tree layer could be stratified into a lower tree layer (5-10 m) and an upper tree layer (>10 m) up to 18.5 m in height with a 60% crown closure.

The vegetation types were ordered on a successional gradient along the first NMDS-axis (Fig. 2) from open and shrubby Agrostis pastures to impacted evergreen forests. The little disturbed variant of the evergreen forest (SDE) was floristically most dissimilar to the extensively grazed Agrostis (EGN) pastures. Increasing stem density and litter amount as well as age, basal area and crown closure were mostly correlated with the impacted evergreen forest formations in the NMDS. However, the closeness of the arrows suggests a correlation between stem density, litter and age as well as basal area and crown closure. The variables that had the highest correlation with the floristic gradient in the NMDS (Figure 2) were dung piles, a sign of open and shrubby

Agrostis capillaris pastures, while cutting frequency pointed in the opposite direction on this axis, towards forests.

Figure 2. Ordination diagram from non-metric multidimensional scaling (NMDS). The composition dataset was based on cover values per plot by species. Bray-Curtis distance was used for the ordination (30 maximum random starts, iteration for 3 dimensions, stress: 0.084). We choose axes 1 and 3 for the presentation of the results. Symbols represent individual plots coded by vegetation type that are further delimited with polygons. The arrows represent the explanatory variables fitted onto the ordination (age: age of the trees in the plot). Polygons delimit the vegetation types (EGN = grazed non-native grasslands, SDE = sparsely disturbed evergreen forest, SIE = severly impacted evergreen forest, UBS = closed grazed Ugni and Berberis shrublands).

67 4.2. Land use drivers of recovery and degradation

The importance of the land use variables for differentiating the vegetation types was reflected by the results shown by the classification tree (Figure 3). The number of dung piles per m² and the browse index were the two most important splitting variables. At the first split

Figure 3. Classification tree to predict the vegetation type based on the conditional inference tree (cTree) model. The encircled explanatory variables are those showing the strongest association to the response variable. Values on lines connecting explanatory variables indicate splitting criteria; for example, the first split separated plots of the EGN grasslands (right split) from those in the other three vegetation types (left side of the split). Numbers in boxes above the explanatory variable indicate the node number. The p-values listed at each node represent the test of independence between the listed independent variable and the response variable. “n =” next to terminal nodes indicates the number of plots classified in that node. Bar graphs illustrate the proportion of plots in a vegetation type within that node. 1 = sparsely disturbed evergreen forest (SDE), 2 = closed grazed Ugni and Berberis shrublands (UBS), 3 = severly impacted evergreen forest (SIE) and 4 = grazed non-native grasslands (EGN).

the number of dung piles was selected from the variables. A value higher than 0.075/m² for dung piles explained 100% of the EGN grassland plots. When the number of dung piles was less or equal to 0.075/m², the number of dung piles were again selected for the second split. If the number of dung piles was higher than 0.006/m² at the second split, about 55% of the plots belonged to the UBS shrublands. If the number of dung piles was equal or lower than 0.006/m² the third splitting variable was browse index. Values higher than 0.5 for the browse index explained 60% of the SDE forests and 40% of the UBS shrublands. When the browse index was lower than or equal to 0.5, 80% of the plots belonged to the SDE forests.

When all species of all vegetation types were included in the PERMANOVA, there was a significant differentiation based on the land use variables (number of dung piles F = 10.698, pr = 0.012, cutting frequency F = 5.07, pr = 0.002). There was also an interaction between cutting frequency and browse index (F = 1.873, pr = 0.004). Furthermore, the results proved that cutting frequency (F = 2.289, pr = 0.003) and the number of dung piles (F = 4.816, pr = 0.025) significantly differentiated the UBS shrublands from the extensively grazed non-native grasslands. For the differentiation of UBS shrublands from SDE forests, cutting frequency (F = 2.380, pr = 0.035) and the number of dung piles (F=5.203, pr=0.022) were the most significant single variables. Furthermore, interactions between cutting frequency, mean browsing index, cutting intensity and tracks were also significant (cutting frequency and browse index (F=1.506, pr=0.004), cutting intensity and browse index (F=3.193, pr = 0.017); tracks, cutting frequency and browse index (F=1.254, pr = 0.043); tracks, cutting intensity and browse index (F=1.308, pr = 0.033)). For the separation of SIE forests from the SDE forests the following individual and interactions between land use variables were significant: the number of dung piles (F=1.998, pr=0.014); tracks and number of dung piles (F=2.556, pr=0.005); tracks, cutting frequency and browse index (F=1.633, pr=0.021); cutting frequency and browse index (F=1.424, pr=0.018)).

69 4.3. A conceptual model for the vegetation dynamics in the cultural landscape in the VCR

Based on the results described above we developed a conceptual model (Figure 4) that links the identified VTs along a successional gradient through regeneration and diverting degradation gradients driven by land use activities and the associated disturbances.

Figure 4. Conceptual model showing the dynamic relationships, disturbances and recovery processes between the vegetation types in the traditional rural landscape of the Valdivian Coastal Range (EGN = grazed non-native grasslands, SDE = sparsely disturbed evergreen forest, SIE = severely impacted evergreen forest, UBS = closed grazed Ugni and Berberis shrublands).Arrow size approximates the probability of recovery from disturbance or maintenance under the recent scenario.

5. Discussion

5.1. Vegetation types comprising the vegetation mosaic

The identified floristic gradient (Figure 2) is a structural gradient from grassland to forest. Along the first NMDS axis the diagnostic species change from open land species, mostly non-native annuals, to native forest species. Extensively grazed non-native grasslands (EGN) contain a low number of diagnostic species. This reflects the high variability in floristic composition due to the variable microsite conditions or different disturbances which have a strong influence on the floristic composition of grasslands. The presence of the two diagnostic species indicates that disturbances were either low or extremely high impact in this vegetation type. Trifolium repens (Fabaceae) is known to be common in anthropogenic grasslands in the Valdivian Coastal Range (VCR; Ramírez et al. 1992) indicating low disturbance because it disappears when soils become degraded due to soil compaction (Ramírez et al. 1992). On the other hand, Chevreulia sarmentosa (Asteraceae) indicates sites that are degraded and dry due to soil compaction (Ramírez et al. 1992).

The closed and semi-closed grazed Ugni and Berberis shrublands (UBS) were the VT floristically most similar to the EGN grasslands. UBS shrublands were mainly characterized by shrub species (Ugni molinae, Berberis microphylla, Genista monspessulana, and Baccharis racemosa). U. molinae and B. microphylla are rapidly colonizing thermophilous shrub species. Within one to three years they can form a closed shrub layer (Hildebrand 1983). These two shrub species often begin colonizing from forest fringes (Amigo et al. 2007). The presence of Embothrium coccineum indicates that succession is occurring in the UBS shrublands. This suggests that shrub establishment, in turn, facilitates the regeneration and establishment of trees. This could be because some spiny (e.g. B. microphylla), or unpalatable shrubs (G. monspessulana; Hildebrand 1983) prevent browsing (Cornelissen et al. 2003), or because a dense shrub layer acts as a barrier to livestock. The herbs Vulpia bromoides and Leucanthemum vulgare indicate disturbance. They grew in places where there was no shrub layer, such as on livestock created tracks between shrub patches. The closed sward in these places is typical for open land species

71 like those diagnostic for Agrostis capillaris pastures (Appendix A). Some of them are known to be especially trample-resistant species which lends support to the thesis that Agrostis capillaris pastures are strongly affected by disturbance.

Severely impacted evergreen forests (SIE) comprise the transition from open shrubby Agrostis pastures to the impacted evergreen forests (Figure 2). The shared diagnostic species of the UBS shrublands and the SIE forests are, with one exception (Centaurium littorale), characteristic species of Aristotelia chilensis shrublands (sensu Hildebrandt (1983)). In this study A. chilensis was also a diagnostic species for the SIE forests. However, it showed a higher floristic similarity to the SDE forests than to the UBS shrublands. This indicates that SIE forests represent an advanced state of succession towards closed forest. The species diagnostic for the SIE forests only indicate open conditions in forests and forest fringes, i.e., of openings created by cutting. Therefore, species diagnostic for the SIE forests appear to be less resistant to grazing and trampling. This includes species such as Blechnum hastatum (Godoy et al. 1981, Hildebrand- Vogel 2002, Saldaña et al. 2005, Amigo et al. 2007), A. chilensis and Azara lanceolata (Veblen and Schlegel 1982, Hildebrand, R 1983, Hildebrand-Vogel 2002, Amigo et al. 2007). The diagnostic species Hypericum androsaemum, an exotic species in Chile native to temperate regions of Europe and Western Asia, has escaped from gardens and fields and become naturalized in Chile (Robson 1985). Local landowners recognized that Greigia sphacelata frequently occurs on burned sites (personal communication December 01, 2010) and can be considered a ruderal species indicating disturbed sites.

Diagnostic species of the sparsely disturbed evergreen forest (SDE) reveal less disturbed conditions compared to the SIE forest. Among others, the tree species Aextoxicon punctatum and Drimys winteri are typical for the evergreen temperate rainforest (Oberdorfer 1960, Donoso 1981, Veblen and Schlegel 1982). The fact that the epiphyte Fascicularia bicolor was a diagnostic species emphasizes that the structural complexity of the SDE forests is comparatively high. F. bicolor is the dominant epiphyte of temperate rainforests in Chile (Díaz et al. 2010). It was present in 24% of the SDE forest plots and may indicate less disturbed conditions. Since vascular epiphytes are slow growing (Zotz 1995, Zotz 1998, Hietz et al. 2002) they are susceptible to

anthropogenic disturbances that alter their habitat (tree crowns) and changes in microclimatic conditions on which epiphytes depend (Freiberg 1997). In contrast, the tall fern Lophosoria quadripinnata is associated with the typical hyperhumid conditions of Valdivian woodlands occurring on steep slopes (Oberdorfer 1960). Previously it was also associated with disturbances and clearings in the forest (Amigo et al. 2007). Therefore, it may indicate disturbed sites in the VT.

Remarkable was the total absence of Notofagus dombeyi from the impacted evergreen forests. The site conditions there suggest good growing conditions for this species characteristic of the ’Coihue-Ulmo’ forests. There are many plausible reasons for its absence including past selective cutting of Nothofagus (Ramírez and San Martín 2005), a lack of large-scale disturbances in the coastal range (Smith-Ramírez 2004), an absence of fire, a deficit of seed trees and the grazing of livestock on N. dombeyi (Coihue) seedlings.

Eucryphia cordifolia (Ulmo), although present in our dataset, could not be verified as an indicator species for either the impacted evergreen forests or one of its variants. This is because the species was not restricted to forests in our dataset. E. cordifolia is no longer the typical emergent once found in Coihue-Ulmo forests because of the past and present exploitation of E. cordifolia (Ramírez and San Martín 2005).

5.2. Land use drivers that influence recovery and degradation processes

The number of dung piles and cutting frequency correlated best with overall floristic composition. Dung piles and browse index were the most significant variables for differentiating the discrete vegetation types. However, the browse index interacted strongly with other land use variables. High intensity frequent land use leads to degradation, but when the land use declines or stops, the process of succession ensues. Land use activities therefore drive the

73 degradation and recovery processes that lead to changes in floristic composition and turnover in vegetation types.

Forest recovery through succession requires that grazing and cutting becomes significantly reduced. The forests existing today have developed despite extensive land use activities in the past. On smallholders land this development has taken place on small patches. Fragmented forests in Chile are often simplified in structure because older trees have been harvested. Succession is reset to an earlier forest stage with low basal area and a high abundance of saplings and young trees (Echeverría et al. 2007). This phenomenon has also been reported from Amazonian forests where the vegetation dynamics were accelerated due to fragmentation which is likely to exacerbate changes in forest structure, floristic composition and forest microclimate which could in turn cause the extinction of disturbance-sensitive species locally (Laurance et al. 1998, 2006). An example of such changes in structure and composition can be seen in the severely impacted evergreen forests (SIE). SIE forests are often the only wood- producing vegetation types on small farmsteads and are therefore harvested. In contrast, a large number of plots placed in the sparsely disturbed evergreen forest (SDE) were a part of the Llancahue watershed adjacent to the Lomas del Sol where interventions, after initial heavy cutting took place in the early 1900s, were reduced for a long time but have increased since the 1990s (Moorman et al. 2013a). In Camán, SDE forests were found on steeper slopes or in remote locations and were therefore less attractive to smallholders. Thus, forest recovery through natural succession depends largely on socioeconomic circumstances. A scenario that resulted in the abandonment of farmsteads allowed forests to recover in the Concepción metropolitan area (Rojas et al. 2013). On the other hand if the current economic model of the farmsteads continues, the SDE forests will also face degradation.

The SDE forests that developed after land use activities declined still have a significant portion of the original native flora and have a complex structure although they don’t have the emergent Eucryphia cordifolia and Nothofagus dombeyi trees. This challenges Ramírez et al. (1984) framework for describing forest degradation and the forest succession model proposed by Oberdorfer (1960) which assumes a recovery to previous “climax” conditions in a foreseeable

time frame. But it supports Guariguata and Ostertag (2001) who suggest that tropical forest functions and services may recover long before floristic composition does. This is especially important when considering that these little impacted evergreen forests have many native species of the Valdivian evergreen forest that provide timber and non-timber products for smallholders.

5.3. A conceptual model for the vegetation dynamics in the traditional rural landscape in the VCR

Grazed non-native grasslands (EGN) are maintained by extensive grazing. Dung piles distinguished the two variants of open and shrubby Agrostis capillaris pastures. When grazing is abandoned, floristic composition changes towards the closed and semi-closed grazed Ugni and Berberis shrublands (UBS). However, abandonment in our study area did not occur as a deterministic process, like that presumed in the study of Ramírez et al. (1984). Instead it occurred gradually, depending on the level of avoidance of the sites by browsers (Baudry 1991). However this fast-growing shrub layer will recede if grazing is intensified and floristic composition will revert to EGN. UBS shrublands and SIE forests did not differ significantly in terms of impacts from land use. This may be due to the origin of SIE forests which were a product of degradation of SDE forests rather than a regeneration of the UBS shrublands. Cutting and grazing impede the forest recovery on EGN grasslands and UBS shrublands. Recovery is indicated by their transitional status. Apart from the recovery, we observed landowners clearing shrubs from pastures. This suggests that under the present scenario, succession is interrupted at an early stage. The disturbance regime in SIE forest is not created by a single land use activity, but by the interaction of cutting and grazing. This may be because although livestock generally avoid the forest due to its inaccessibility, wood cutting makes it more accessible which results in browsing, grazing and trampling of tree seedlings. This activity might be more intense in winter when grasslands are unproductive leaving only evergreen trees and shrubs left to provide forage (personal communication with landowners, who mentioned specifically the species

75 Dasyphyllum diacanthoides as being very palatable during winter). Although SDE forests are not impacted as much, they are still browsed and cut. The absence of floristic indicators for these impacts may be due to a high resilience of these forest ecosystems to disturbance.

Our findings generally fit the pathways proposed by Oberdorfer (1960) and Ramírez et al. (1984). However, for the first time, the intensity of anthropogenic impacts caused by grazing and cutting are reported in this study using measurable variables directly related to vegetation composition and species turnover. Our data confirm Oberdorfer’s (1960) theory about a “diversity of human influence” on floristic composition. The complex factor ‘human influence’ can promote either succession or degradation depending on the intensity and interactions resulting from land use activities.

6. General conclusions

The occurrence of the different vegetation types (VTs) (1) grazed non-native grasslands (EGN), (2) closed and semi-closed grazed Ugni and Berberis shrublands (UBS) belonging to the formation of ‘open and shrubby Agrostis capillaris pastures’, (3) severely impacted evergreen forests (SIE) and (4) sparsely disturbed evergreen forest (SDE) grouped to the ‘impacted evergreen forest’ formation in small-scale rural landscapes in the VCR depends on land use type, intensity and interactions between land use impacts. Livestock farming in this case, is especially significant. This was highlighted by the importance of dung piles for group differentiation of EGN grassland, UBS shrubland and the little impacted evergreen forest in the classification tree. When the value for dung piles was low (< 0.001 dung piles/m²) cutting frequency becomes important in differentiating between the forest vegetation types. Furthermore, cutting frequency was significant in determining overall floristic composition. The present land use regime indicates that the vegetation dynamics are being affected by disturbance rather than recovery processes. A lack of fencing has led to impacts due to interactions involving livestock and cutting, particularly in SIE forests. The indicator species found there indicated forest degradation. For the other VTs, succession or degradation determined floristic composition. We

therefore recommend the use of our study as a baseline for research beyond the species level (e.g. functional traits) and to study regeneration processes for tree species on a landscape scale to understand which forest structures and functions will recover from these VTs and at which time scale. We found high native species richness in shrublands and forests. This makes these UBS shrublands, SIE forests and SDE forests valuable landscape elements in rural areas with small-scale activities dominated by monoculture exotic tree plantations. We therefore recommend an intensification of research on other taxa that may be associated with the three VTs, such as birds, and to integrate these or future elements of rural landscapes in conservation planning for the future.

7. Acknowledgements

We thank the landholders in Camán and Lomas del Sol who gave us permission to work on their land. Carlos Ramírez helped to identify the collected plant species. We thank the Landesgraduiertenförderung Baden-Württemberg (Germany), the German Academic Exchange Service (Germany) and the Fondo de Fomento al Desarrollo Científico y Tecnológico (Chile) for their financial support. Marco Flores helped with the field work logistics. Danisa Paredes, Magdalena Gerhardt, Marco Sepúlveda and Melanie Welling assisted with field work. Bernhard Thiel and Helen Desmond corrected the English. Helen Desmond and Rodrigo Vargas helped to improve early versions of the manuscript.

77 8. References

Amigo, J., Ramírez, C., and Quintanilla, L.G. 2007. Mantle communities of the temperate woodlands of South Central Chile: a phytosociological study of the order Aristotelietalia chilensis. Phytocoenologia 37:269–319.

Anderson, M.J. 2001. A new method for non-parametric multivariate analysis of variance. Austral. Ecol. 26:32–46.

Arbeitskreis Standortskartierung. 1996. Forstliche Standortsaufnahme. 6th ed. IHW-Verlag, Eching.

Armesto, J.J. 2002. Relevancia de las pequeñas áreas silvestres para la conservación de la biodiversidad en el bosque nativo. Ambiente y Desarrollo 18:44-50.

Armesto, J.J. 1998. Conservation Targets in South American Temperate Forests. Science 282:1271–1272.

Armesto, J.J., Manuschevich, D., Mora, A., Smith-Ramirez, C., Rozzi, R., Abarzúa, A.M. and P.A. Marquet. 2010. From the Holocene to the Anthropocene: A historical framework for land cover change in southwestern South America in the past 15,000 years. Land Use Policy 27:148–160.

Armesto, J.J., Smith-Ramírez C. and R. Rozzi. 2001. Conservation strategies for biodiversity and indigenous people in Chilean forest systems. J. R. Soc. New Zealand 31:865–877.

Baudry, J. 1991. Ecological consequences of grazing extensification and land abandonment: Role of interactions between environment, society and techniques. Ciheam - Options Méditerranéennes 15:13–19.

Chazdon, R.L. 2003. Tropical forest recovery: legacies of human impact and natural disturbances. Perspect. Plant Ecol. Evol. Syst. 6:51–71.

CIREN. 2001. Descripciones Suelos, materiales y símbolos. Estudio agrologico X Región. CIREN Centro de Información de Recursos Naturales (CIREN), Santiago de Chile.

Clements, F.E. 1916. Plant succession: An analysis of the development of vegetation. Carnegie Institution, Washington.

Cornelissen, J.H.C., Lavorel, S., Garnier, E., Díaz, S., Buchmann, N., Gurvich, D.E., Reich, P.B., Steege, H., Morgan, H.D., Heijden, M.G.A. van der, Pausas, J.G. and Poorter, H. 2003. A handbook of protocols for standardised and easy measurement of plant functional traits worldwide. Aust. J. Bot. 51:335 p.

De Cáceres, M. and P. Legendre. 2009. Associations between species and groups of sites: indices and statistical inference. Ecology 90: 3566–3574.

De Cáceres, M., Legendre, P. and M. Moretti. 2010. Improving indicator species analysis by combining groups of sites. Oikos 119:1674–1684.

Deutschewitz, K., Lausch, A., Kühn, I. and S. Klotz. 2003. Native and alien plant species richness in relation to spatial heterogeneity on a regional scale in Germany. Glob. Ecol. Biogeogr. 12:299–311.

Díaz, I.A., Sieving, K.E., Peña-Foxon, M.E., Larraín, J.and Armesto, J.J. 2010. Epiphyte diversity and biomass loads of canopy emergent trees in Chilean temperate rain forests: A neglected functional component. For. Ecol. Manage. 259:1490–1501.

Donoso, C. 1981. Tipos forestales de los bosques nativos de Chile. Publicacion FAO, Santiago de Chile.

Echeverría, C., Newton, A.C., Lara, A., Benayas, J.M.R. and Coomes, D.A. 2007. Impacts of forest fragmentation on species composition and forest structure in the temperate landscape of southern Chile. Glob. Ecol. Biogeogr. 16:426–439.

Ellenberg, H. and C. Leuschner. 2010. Vegetation Mitteleuropas mit den Alpen in ökologischer, dynamischer und historischer Sicht. 6th ed. Eugen Ulmer Verlag, Stuttgart.

Foley, J.A. 2005. Global consequences of land use. Science 309:570–574.

Foster, D, and Motzkin, G. 1998. Ecology and conservation in the cultural landscape of New England: Lessons from nature’s history. Northeast Nat. 5:111–126.

79 Freiberg, M. 1997 Spatial and temporal pattern of temperature and humidity of a tropical premontane rain forest tree in Costa Rica. Selbyana 18:77–84.

Fuenzalida, H. 1965. Clima. Geografia económica de Chile. Corparación de Fomento de la Produción (CORFO). 99-152.

Gajardo, R. 1994. La vegetación natural de Chile: clasificación y distribución geográfica. Editorial Universitaria Santiago, Santiago de Chile.

Godoy, R., Ramírez, C., Figueroa, H. and E. Hauenstein. 1981. Estudios ecosociologicos en pteridofitos de comunidades boscosas Valdivianas, Chile. Bosque 4:12–24.

Guariguata, M.R. and R. Ostertag. 2001. Neotropical secondary forest succession: changes in structural and functional characteristics. For. Ecol. Manage. 148:185–206.

Gutiérrez B., N., Gärtner, S., Gaviria R., J.C., Meier, W. and A. Reif. 2012. Successional vegetation patterns in abandoned pastures of the lower montane cloud forest zone in the Venezuelan Andes. Phytocoenologia 42:101–132.

Gutiérrez B., N., Gärtner, S., López H., J.Y., Pacheco, C.E.and A. Reif. 2013. The recovery of the lower montane cloud forest in the Mucujún watershed, Mérida, Venezuela. Reg. Environ. Change. Available from http://link.springer.com/article/10.1007%2Fs10113-013-0413-y.

Hietz, P., Wanek, W., Wania, R. and N. Nadkarni. 2002. Nitrogen-15 natural abundance in a montane cloud forest canopy as an indicator of nitrogen cycling and epiphyte nutrition. Oecologia 131:350–355.

Hildebrand, R. 1983. Die Vegetation der Tieflandsgebüsche des südchilenischen Lorbeerwaldes unter besonderer Berücksichtigung der Neophytenproblematik. Phytocoenologia 11:145–223.

Hildebrand-Vogel, R. 2002. Structure and dynamics of southern Chilean natural forests with special reference to the relation of evergreen versus deciduous elements. Fol. Geobot. 37:107– 128.

Hobbs, R.J., Arico, S., Aronson, J., Baron, J.S., Bridgewater, P., Cramer, V.A., Epstein, P.R., Ewel, J.J., Klink, C.A., Lugo, A.E., Norton, D., Ojima, D., Richardson, D.M., Sanderson, E.W., Valladares,

F., Vila, M., Zamora, R. and Zobel, M. 2006. Novel ecosystems: theoretical and management aspects of the new ecological world order. Glob. Ecol. Biogeogr. 15:1–7.

Hothorn, T., Hornik, K., Van De Weil, M.A. and A. Zeileis. 2006a. A lego system for conditional inference. Amer. Stat. 60:257-263.

Hothorn, T., Hornik, K., and A. Zeileis. 2006b. Unbiased recursive partitioning: A conditional inference framework. J. Comput. Graph. Stat. 15:651–674.

Hothorn, T. and A. Zeileis. 2008. Generalized maximally selected statistics. Biometrics 64:1263– 1269.

Laurance, W.F., Ferreira, L.V., Merona, J.M.R. and S.G. Laurance. 1998. Rain forest fragmentation and the dynamics of Amazonian tree communities. Ecology 79:2032–2040.

Laurance, W.F., Nascimento, H.E.M., Laurance, S.G., Andrade, A.C., Fearnside, P.M., Ribeiro, J.E.L., Robson L. Capretz and P.M. Fearnside 2006. Rain forest fragmentation and the proliferation of successional trees. Ecology 87:469–482.

Leyer, I. and K. Wesche. 2007. Multivariate Statistik in der Ökologie. Springer. Berlin.

Londo, G. 1976. The decimal scale for releves of permanent quadrats. Vegetatio 33:61–64.

Maechler, M., Rousseeuw, P., Struyf, A., Hubert, M. and K. Hornik. 2013. Cluster: cluster analysis basics and extensions, R Package Version 1144.

McCune, B., Grace, J.B. and D.L. Urban. 2002. Analysis of ecological communities. MjM Software Design, Gleneden Beach, OR.

Menzies, N.K. 2007. Our forest, your ecosystem, their timber: communities, conservation, and the state in community-based forest management. Columbia University Press. New York.

Moorman, M.C., Peterson, N., Moore, S.E. and P.J. Donoso. 2013a. Stakeholder perspectives on prospects for co-management of an old-growth forest watershed near Valdivia, Chile. Soc. Nat. Resour. 26(9): 1022-1036. Moorman, M., Donoso, P.J., Moore, S.E., Sink, S. and D. Frederick. 2013b. Sustainable Protected Area Management: The Case of Llancahue, a Highly-Valued

81 Periurban Forest in Chile. J. Sustain. For. Available from http://www.tandfonline.com/doi/full/10.1080/10549811.2013.803916#. Ub8FXZxHD2k.

Myers, N., Mittermeier, R.A., Mittermeier, C.G., Da Fonseca, G.A. and J. Kent. 2000. Biodiversity hotspots for conservation priorities. Nature 403:853–858.

Neira, E., Verscheure, H. and C. Revenga. 2002. Chile’s frontier forests: conserving a global treasure. Global Forest Watch, World Resources Institute. Washington DC.

Oberdorfer, E. 1960. Pflanzensoziologische Studien in Chile: Ein Vergleich mit Europa. J. Cramer. Weinheim.

Olson, D.M. and E. Dinerstein. 1998. The Global 200: A Representation Approach to Conserving the Earth’s Most Biologically Valuable Ecoregions. Conserv. Biol. 12:502–515.

Oksanen, J., Guillaume Blanchet, F., Kindt, R., Legendre, P., Minchin, P.R., O'Hara,R.B., Simpson,G.L., Solymos, P., Henry, M., Stevens, H. and H. Wagner. 2013. vegan: Community Ecology Package.

Pliscoff, P., Tecklin, D., Farías, A. and J. Sáez. 2005. Análisis de paisaje de conservación para la Cordillera de la Costa de la Región de los Lagos. Com. Nac. Medio Ambiente (Conama), Valdivia.

R Core Team (2013). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/.

Ramírez, C., San Martín, C., Verdugo, M. and F. Lorena. 1992. Malezas en plantaciones forestales de la cordillera costera del centro-sur de Chile. Ciencias Forestales 8:27–46.

Ramírez, C., Moraga, M. and H. Figueroa. 1984. La similitud florística como medida de degradación antrópica del bosque valdiviano. Agro Sur 12:127–139.

Ramírez, C. and C. San Martín. 2005. Asociaciones vegetales de la Cordillera de la Costa de la Región de Los Lagos. In: Smith-Ramírez, C., Armesto, J.J. and C. Valdovinos (eds.), Historia, biodiversidad y ecología de los bosques costeros de Chile. Editorial Universitaria, Santiago de Chile. pp 206–223.

Relva, M.A. and T.T. Veblen. 1998. Impacts of introduced large herbivores on Austrocedrus chilensis forests in northern Patagonia, Argentina. For. Ecol. Manage. 108:27–40.

Robson, N.K.B. 1985. Studies in the genus Hypericum L. (Guttiferae). 3. Sections 1. Campylosporus to 6a. Umbraculoides. Bull. Br. Mus. Nat. Hist. Bot. 12:163–325.

Rojas, C., Pino, J., Basnou, C. and M. Vivanco. 2013. Assessing land-use and -cover changes in relation to geographic factors and urban planning in the metropolitan area of Concepción (Chile). Implications for biodiversity conservation. Appl. Geogr. 39:93–103.

Rouvinen, S., Kuuluvainen, T. and L. Karjalainen. 2002. Coarse woody debris in old Pinus sylvestris dominated forests along a geographic and human impact gradient in boreal Fennoscandia. Can. J. For. Res. 32:2184–2200.

Saldaña, A., Gianoli, E. and C.H. Lusk. 2005. Ecophysiological responses to light availability in three Blechnum species (Pteridophyta, Blechnaceae) of different ecological breadth. Oecologia 145:251–256.

Smith-Ramírez, C. and J.J. Armesto. 2002. Importancia biológica de los bosques costeros de la décima región: el impacto de la carretera costera sur. Ambiente y Desarrollo. 18:6-14.

Smith-Ramírez, C. 2004. The Chilean coastal range: a vanishing center of biodiversity and endemism in South American temperate rainforests. Biodivers. Conserv. 13:373–393.

Smith-Ramírez C., Díaz I, Pliscoff P, et al. (2006) Distribution patterns of flora and fauna in southern Chilean Coastal rain forests: Integrating Natural History and GIS. Biodivers. Conserv. 16:2627–2648.

Stohlgren, T.J., Falkner, M.B. and L.D. Schell. 1995. A Modified-Whittaker Nested Vegetation Sampling Method. Vegetatio. 117 113–121.

The International Plant Names Index (2012). Available at: http://www.ipni.org. Accessed May 28, 2013.

Thompson, I. 2011. Biodiversity, ecosystem thresholds, resilience and forest degradation. Unasylva 238:25–30.

83 Veblen, T.T., Mermoz, M., Martin, C. and E. Ramilo. 1989. Effects of Exotic Deer on Forest Regeneration and Composition in Northern Patagonia. J. Appl. Ecol. 26:711–724.

Veblen, T.T. and F.M. Schlegel. 1982. Reseña ecológica de los bosques del sur de Chile. Bosque 4:73–115.

Zotz, G. 1998. Demography of the epiphytic orchid, Dimerandra emarginata. J. Trop. Ecol. 14:725–741.

Zotz, G. 1995. How fast does an epiphyte grow? Selbyana 16:150–154.

Chapter 2: Small-scale cultural landscapes as a refugee for native birds and plants in the coastal range of Chile

1. Abstract

In the mid elevations of the Valdivian Coastal Range of Chile, native species are often restricted to areas where there is also small-scale farming. This study (1) analyzed the distribution of bird and plant species diversity among hierarchical levels within the landscape, (2) evaluated the importance of functional diversity as a surrogate for bird species diversity and (3) identified important indicators of land use activities that introduced changes in plant functional traits. Additive partitioning of beta diversity was applied to analyze the contribution of the plot level, formation level and landscape level to gamma diversity. The importance of functional diversity as a surrogate for bird species diversity was analyzed through correlation analysis. Traits that reflected land use impact were identified using RLQ and fourth-corner statistic. The formation level had the highest contribution to gamma diversity while functional richness was shown to be a suitable surrogate for bird richness. Forests and shrublands are important landscape elements and deserve special attention especially in formulating conservation strategies on a landscape scale involving the integration of rural farming areas.

Keywords: Temperate rainforest - additive partitioning - functional traits - functional diversity - rural landscape – bird diversity

85 2. Introduction

Rural landscapes resulting from small-scale farming have been recognized for their biodiversity conservation potential due to their variety of habitats that support many taxa worldwide (Farina 1995; Daily et al. 2001; Deutschewitz et al. 2003; Menzies 2007). In small-scale cultural landscapes vegetation is changed to assemblages of floristically and physically different vegetation types occurring in a spatial mosaic. This vegetation mosaic (1) can serve as a refuge for plant and animal species and function as stepping stones between remaining large native forest areas or protected areas (Armesto 2002); (2) may have the potential to restore forest functions and services (Guariguata and Ostertag 2001; Armesto 2002; Chazdon 2003); and (3) may represent a part of a heterogeneous cultural landscape, with unique plant and animal communities, but is especially valuable for species depending on openings and edges (Foster and Motzkin 1998).

Cultural landscapes “reflect the interactions between people and their natural environment in space and time” (Plachter and Rössler 1995 p. 15). Their role as refuges for native species preservation and richness is especially important in regions lacking protected spaces and areas where most of the land has already been converted to a matrix type having a low proportion of natural forest (Harvey et al. 2008). Besides vascular plant species, birds are one of many other taxa benefiting from this type of landscape mosaic characterized by high habitat heterogeneity (Tscharntke et al. 2005). However, small-scale cultural landscapes developed fairly recently in Chile compared to other regions e.g. the traditional cultural landscapes in Europe. Therefore, the functional role of Chile small-scale farming areas to provide similar conservation services has not been studied yet. This is especially significant when considering the growth of industrial agriculture and forestry matrices.

Our study area is an example of a recent small-scale cultural landscape in the Valdivian Coastal Range (VCR) in southern Chile. The area’s native temperate rainforest along with its vegetation is a threatened ecosystem with outstanding biodiversity (Myers et al. 2000). The avifauna in this temperate rainforest is especially important for the ecosystem because birds perform a number

of ecosystem services such as pollination (Smith-Ramírez et al. 2005), seed dispersal (Armesto et al. 1987; Armesto and Rozzi 1989) and pest control (Soto-Mora and Urrutia 2010). However, bird species richness depends on forest structural properties (Díaz et al. 2005). Therefore, mutualistic interactions between birds and plants are a characteristic feature of this temperate rainforest (Armesto et al. 1987; Smith-Ramírez et al. 2005). In the mid elevations of the VCR nature conservation areas are lacking (Smith-Ramírez 2004). This has restricted native vegetation to rural settlements with small-scale agricultural land use as most of the area has been converted to tree plantations (Armesto 1998; Armesto 2002; Neira et al. 2002; Smith- Ramírez 2004).

The vegetation mosaic in the small-scale cultural landscape is comprised of grassland, shrubland and forest. A first description of the plant communities that comprise the mosaic was done in a small-scale pioneering work by Oberdorfer (1960). Ramírez et al. (1984) hypothesized that the floristic similarities between these communities reflect a degradation gradient. It could be proved that the vegetation types that comprise the mosaic in our study area have a dynamic relationship, involving disturbance and regeneration processes caused by disturbances from livestock farming and wood harvesting (Seis et al. 2014). An indication of this dynamic is that shrubland communities are known to be very variable in their life forms (Hildebrand-Vogel 2002), because shrublands are a transitional stage between grasslands and forests they combine characteristics of both ecosystems. Similar to grasslands is their tendency to be invaded by non-native species (Ramírez et al. 1989). They may also share forest ecosystems processes like the regeneration of native tree species. Furthermore, the presence of fleshy fruited species may be an especially important food for birds (Armesto et al. 1987). On Chile’s Chiloé Island, shrublands and forest remnants have been shown to serve as habitats for certain bird species within an agricultural mosaic landscape. The remnants include several forest understory species (Willson et al. 2001; Díaz et al. 2006) as well as important mutualistic bird species (Willson et al. 2001).

87 However, the diversity patterns of plant and bird species and the causal relationships between the taxa on a landscape scale, generally remains unknown. Our study aimed to make a contribution towards filling this knowledge gap by:

1. Analyzing the distribution of bird and plant species diversity among hierarchical levels within the landscape,

2. Determining the usefulness of functional diversity as a surrogate for bird species diversity,

3. Identifying important variables of land use and vegetation structure that provoke changes in plant functional traits.

Plants are important structural components of terrestrial ecosystems. They are the base of food webs and key structural elements that determine habitat configurations for many animal species, including birds (MacArthur and MacArthur 1961). Plant species composition and diversity are undoubtedly important transmitters of ecological information, yet they are often only very remotely relevant to understanding ecosystem processes (Campetella et al. 2011). The aggregation of plant functional traits to plant functional diversity comprises an important ecosystem component that serves as a proxy for ecosystem complexity and resource diversity crucial for other taxa such as birds (Kissling et al. 2008).

The individual plant functional traits (PFT) that comprise plant functional diversity have been proven to be effective environmental filters sensitive to disturbances caused by land use (Navarro et al. 2006) and link processes that shape patterns and dynamic pathways within ecosystems (Lavorel and Garnier 2002; Garnier et al. 2004). PFTs allow generalizations to be made among ecosystems, for example, they have been used successfully to evaluate complex ecosystem responses to human disturbance in grasslands (e.g. Lavorel et al. 1998; Barbaro et al. 2000; Kahmen and Poschlod 2008; Mládek et al. 2010), forest ecosystems (e.g. Graae and Sunde 2000; Verheyen et al. 2003; Aubin et al. 2007) and other ecosystems (Campetella et al. 2011).

Gaining some knowledge of the scale within a landscape that contains most of the total species diversity (gamma diversity) is of crucial importance for effective conservation management

(Whittaker et al. 2001). For example, although an individual plot may not be very species rich, it may still contribute a part of the richness of a hierarchically higher total (Chandy et al. 2006). Gamma diversity can be partitioned into additive components within and between hierarchical components (Allan 1975; Lande 1996), which makes it possible to identify the relative contribution of alpha, beta and gamma diversity across a range of spatial scales (Wagner et al. 2000; Gering et al. 2003). Such patterns could differ or coincide between taxa.

In this study the concept of additive partitioning of beta diversity proposed by Crist et al. (2003) was applied to analyze the contribution of the plot, formation and landscape level to the maintenance of gamma diversity. In the next step the relationships between bird diversity and plant functional and species diversity were analyzed by correlation analysis. Traits relevant to mirror the land use impact were identified using RLQ and fourth-corner statistic.

3. Study site and methods

The study was situated on the eastern slopes of the VCR in the vicinity of two rural communities: Camán (39° 58'S 73° 00'W) and Lomas del Sol (39° 50'S 73° 07'W) with similar climate and soil conditions (Fig. 1). The climate is classified as temperate perhumid. Annual rainfall in Llancahue is 2,357 mm, with a peak in July (Núñez et al. 2006). The average annual temperature is 12.0 °C at the weather station at 25 m a.s.l. in the Valdivia city.Soils are well drained Ultisols of volcanic origin above metamorphic bedrock.

The soil texture varies from silty to sandy loam (CIREN 2001). All study sites were located between 250 and 500 m.a.s.l. European settlement of the area around Camán and Lomas del Sol started around 1900 and was accompanied by a massive exploitation of native forest. During the 1970s state owned forests in Camán began to be converted to exotic tree plantations on a large scale, and the conversion process continues on private farmland today. Nevertheless, the only native forest remaining in Camán exists on private property. Even though the area covered by exotic tree plantations in Lomas del Sol also expanded, the adjacent forest, part of the Llancahue watershed, remained state owned and was not converted. But in the 1990s

89 neighboring farmers started to graze livestock and harvest timber in the state forest (Moorman et al. 2013). As a general result of the land conversions in the mid elevations of the easternslope of the VCA today approximately 70% of the land is exotic tree plantations and 30% is a mosaic consisting of grasslands, forests and shrublands.

Fig. 1 Study area in Chilean Northern Patagonia. The sites are located in the region of Camán and Lomas del Sol between the cities of Valdivia and Paillaco. Each triangle represents a vegetation sample plot.

3.1. Vegetation and land use data collection

Vegetation data collection plots were placed in a stratified random sampling design in grassland, shrubland and forest formations. Plots were modified Whittacker plots (Stohlgren 2007) with a 20 m² plot used for the 0-1 m high herb layer, 80 m² for the >1 m – 5 m high shrub layer and a

160 m² plot for the lower and higher tree layer respectively. Cover was estimated using a decimal scale (Londo 1976) for each vascular plant species in each layer. Vegetation structure was assessed by estimating the cover of each layer in percent. The crown closure was estimated visually and tree diameters were measured at breast height (dbh) for all trees >= 5 cm dbh in the plots. In each plot site parameters (elevation, aspect and slope) were measured. Tree density and basal area were calculated for each plot. Total cover of the higher and the lower tree layers were added together. The formation, representing gradient in vegetation complexity, was added as a nominal structural variable.

Land use (LU) indicators for livestock farming and tree harvesting (cutting) were assessed in each plot to determine the variability in site use by animals and humans.

The number of dung piles/m² served as a proxy for grazing intensity as herbivores defecate in the same area in which they feed (Relva and Veblen 1998). Therefore, dung piles were counted in the 20 m² subplots and expressed as dung piles/m². Browsing damage caused by livestock was assessed for the woody species in the herb and shrub layers. The mean browsing index was then calculated (Veblen et al. 1989). Dung piles and browsing damage were variables used for describing livestock damage on vegetation. These indicators can be readily measured and have been used frequently (Echeverría et al. 2007). As an indicator for trampling the amount of surface area covered by animal tracks was measured. Trampling intensity was estimated based on the amount of bare soil resulting from tracks. The number of droppings per m² was calculated separately for large herbivores (cattle and horses), small herbivores (sheep and pigs) and European hares. Track surface area in m² was calculated for each plot and weighted by trampling intensity. The mean browsing index (Veblen et al. 1989) was then calculated from the browsing damage recorded on woody species in the herb and shrub layers.

To assess the intensity and frequency of tree cutting, the number of stumps was counted. Deadwood age was estimated with a knife test (Rouvinen et al. 2002) for each stump. Cutting intensity was then calculated as stumps/m² and cutting frequency as the number of different age classes present in a plot. Coppice was recorded by counting tree individuals that apparently originated from coppice.

3.2. Bird survey

Bird data was recorded in a subset of 30 vegetation plots and surveyed with point-counts (Bibby et al. 1992) conducted from centers of vegetation plots. Point counts allow bird abundances to be analyzed with vegetation features (Petit et al. 1995). Sampling points were equally distributed by putting 10 points in each formation. Although a sample size of 10 plots has been described as being adequate for estimating bird density and richness in Chilean rainforests, very rare species may not be included (Jiménez 2000). The vegetation sampling plots chosen for the bird census had to match the following criteria:

1. To be in a homogeneous formation extending for at least 25 m in each direction from the point to avoid edge effects,

2. Plots in the same formation had to be at least 200 m away from each other (plot center to plot center).

3. Plots of different formations had to be separated by at least 500 m (Petit and Petit 2003).

The plots had a 25 m radius because the probability of bird detection is fairly uniform within this distance from an observer, even across extremely different habitat types (e.g., (Scott and Ralph 1981). Census records were taken during the breeding season (November – March) and in the bird’s main activity daytime, after sunset (06:00 – 10:00). Surveys were only carried out on clear days with birds being recorded acoustically and visually by two observers. The effect of birds flying away before they could be recorded was countered by waiting two minutes after arriving. During the next ten minutes all birds that could be detected within and outside the plot were recorded. This ten minute detection time is recommended for Chilean rainforest ecosystems (Jiménez 2000). Additionally, the bird behavior was recorded by categorizing it as: aggressive encounters, flying, feeding, perched, moving between perches, nesting and territorial (Bibby et al. 1992). Each plot was surveyed on two different days at different times. Birds recorded outside of the plot or flying over the plot, without evidence of “active” behavior e.g. feeding

within the plots were deleted from the dataset. Nomenclature follows the classification of the American Ornithologists Union (Remsen et al. 2014).

3.3. Plant trait data

We chose 8 categorical plant functional traits (PFTs) that matched the following criteria.

1. Potential response to grazing and cutting impacts

2. Importance for habitat complexity, or resource diversity for birds

3. Available information in the literature

Information on traits obtained from the literature will help to make our study easier to repeat for further studies and have applications for ecosystem managers. Trait information was available for 85% of the plant species (129 out of 151).

3.3.1. Dispersal

Species were classified according to their main natural dispersal type. The main dispersal type is strongly associated with anthropogenic disturbance (Cornelissen et al. 2003). For example when domestic animals utilize forests, the number of epizoochor dispersed plants is likely to increase. In undisturbed forests seeds dispersed by gravity and over short distances (e.g. ballistochor) should dominate. Timber harvesting and/or animal trails decrease the forest density thereby increasing access to forests and shrublands so that endozoochorous and wind dispersed seeds are more frequent (Halpern 1989). There should be a general trend towards anemochorous species on wind exposed sites. Changes in environmental conditions in coppice forests act to filter dispersal traits (Mason and MacDonald 2002; Decocq et al. 2004; Bartha et al. 2008; Canullo et al. 2010). Each species recorded was assigned one of the following dispersal types: anemochor, ballistochor, endozoochor, epizoochor, hydrochor, and multiple (if there was no tendency towards one single dispersal type).

3.3.2. Fleshy fruits

A fleshy fruit is a nutritious "package", which contains seeds and is "designed" to be eaten by animals (Snow 1971). The richness of frugivorous animals is largely dependent on the availability of fruit (Herrera 1985; Fleming et al. 1987; Bleher et al. 2003; Kissling et al. 2007). In the temperate evergreen rainforest of Chile, fleshy fruits are found in 60% to 70% of the species among different growth forms (Armesto et al. 1987) and are most important as a food resource for birds, but also are important for mammals and the endemic marsupial Dromiciops australis (Armesto et al. 1987). Trees are especially important for the provision of fleshy fruits (Smith- Ramírez and Armesto 1994). We classified all species as either fleshy or non-fleshy fruit bearing.

3.3.3. Growth form

Growth form is a morphological classification of plants. There is no general hypothesis about the influence of grazing on growth form among ecosystems (Díaz et al. 2005). Cattle are known to feed on trees, shrubs, graminoids, and forbs in the Patagonian forests. However, exact feeding strategies depend on the season and habitat types (Vila and Borrelli 2011). Trees and epiphytes are especially affected by cutting. When trees are cut and extracted, epiphytes lose some crown habitat and must adapt to changes in microclimate (Freiberg 1997) due to the canopy gap. Climbing plants are similarly affected by cutting. Opening the tree canopy by cutting may promote shrub and herb growth forms. Growth forms were classified as either: climber, epiphyte, herb, shrub, or tree.

3.3.4. Life cycle

The number of species with perennial life cycles generally increases from grasslands (many annuals) to shrublands and forests. Grazing promotes annuals, especially in humid ecosystems (Díaz et al. 2007). Grazing opportunities may be enhanced by disturbances, like cutting, that cause canopy openings. Similarly, by opening the shrublayer, vegetation becomes more accessible for animals. Life cycles were classified by type: annual, biennual, or perennial.

3.3.5. Phytogeographic origin

The non-native/native relationship is an important indicator of the conservation state of an ecosystem. In the VCR non-native herbs are especially invasive in grasslands (San Martín et al. 2009; Seis et al. 2014). Disturbance by land use can lead to a considerable increase in non-native species. Large ungulate grazers may play a role in spreading and establishing introduced invasive species (Vavra et al. 2007). Forests that are frequently disturbed by cutting and domestic herbivores have a higher number of non-native species. The non-native species occur mainly on tracks in shrublands and in forests (open spaces). Coppicing itself might not have an effect on phytogeographic origin relationship. Species were categorized as either non-native or native.

3.3.6. Raunkier life-form

The life-form concept classifies plants based on the position of their dormant buds. For disturbances that are unpredictable spatially and in their intensity, such as livestock grazing, the classification of species by life-form helps to explain their survival strategies (Cornelissen et al. 2003). Chamaephytes and phanerophytes increase in shrublands and forests if grazing intensity declines (Ellenberg and Leuschner 2010) whereas with increasing amounts of grazing, grasslands become dominated by hemicryptophytes and therophytes. Our data contains chamaephytes, hemicryptophytes, phanerophytes and therophytes.

3.3.7. Spinescence

Grazing and browsing likely increase the abundance of spinescent species because spines make plants less palatable to herbivores (Cornelissen et al. 2003). Spiny plants may also increase if forest vegetation is disturbed, for example, by humans through cutting (Halpern 1989). We classified species as either: spiny or non-spiny.

95 3.3.8. Woodiness

The condition of woodiness is fundamental for ecosystem structure and related functions and is considered a “core trait” in ecology (Weiher et al. 1999). Woody plants are essential for birds in that they provide foraging substrates and nesting sites (Cody 1985). Livestock browsing reduces woody plants (Ellenberg and Leuschner 2010). Because there was no quantitative wood density data available for our dataset, it was classified as either: woody or non-woody.

3.4. Data Analysis

The method of additive partitioning of beta diversity proposed by Crist et al. (2003) was applied to analyze the contribution of the plot, formation and landscape level to the maintenance of gamma diversity. The additive partitioning of species diversity is considered as a promising approach for analyzing patterns of diversity among different ecological scales in a hierarchical order (Lande 1996; Loreau 2000; Godfray and Lawton 2001). We applied this statistical approach for plant and bird gamma diversity on 3 different hierarchical levels (Fig. 2), testing the null hypotheses that observed partitions of species diversity do not differ from those expected by chance (Crist et al. 2003). The diversity at each hierarchical scale was compared against the expected diversity that was generated by the null distribution obtained by a randomization procedure. The adipart function of the R package “vegan” (Oksanen, et al. 2013) was used for the additive partitioning procedure.

For plant and bird species we chose species richness and Simpson’s Index as measures of diversity. Richness by definition considers the absolute number of species whereas Simpson’s Index is the probability of two individuals belonging to the same species. Both measures can be considered independent from each other (Magurran 2004). Indices for species diversity were calculated using the diversity function of the R package “vegan” (Oksanen, et al. 2013).

For the analysis of functional diversity we used the measures functional richness and Rao’s quadratic entropy (RaoQ), proposed by Villéger et al. (2008). RaoQ is influenced by both, species-abundance based diversity and differences among species, and may therefore decrease

if richness increases (Botta-Dukát 2005). This independency of RaoQ from richness made us choose this index for functional diversity complementary to Simpson’s Index for species diversity. The functions to calculate functional diversity indices are implemented in the dbFD function of the R package “FD” (Laliberté and Legendre 2010; Laliberté and Shipley 2010).

Fig. 2 The proposed hierarchical model of plant and bird species diversity. At each hierachical level alpha and beta components are additively linked to form diversity of the next higher level.

In the next step the relationships between bird diversity, plant functionality and plant species diversity were analyzed by correlation analysis. Spearman rank correlation was used to analyze the relationship between plant functional diversity and bird species diversity using the rcorr function of the package “Hmisc” (Harrell and Dupont 2007). Differences in the distribution of functional diversity indices between the formations were tested with the Wilcoxon sign test.

97 We used RLQ ordination to analyze the relationships between species traits and land use indicators based on species cover (Dolédec et al. 1996). To test the significance of the explicit trait – land use indicator relationships we used the fourth-corner statistic (Dray and Legendre 2008).

RLQ analysis (Dolédec et al. 1996; Dray et al. 2002) performs a multivariate analysis between three different matrices: the R matrix represents the land use characteristics of each plot (environmental data); L represents the species cover on each plot (species abundance data) and Q represents the trait attributes of each species (trait data). The purpose of this analysis is to extract the joint structure between the two matrixes R and Q using the third matrix L. The species cover (L) is supposed to measure the intensity of the relationships between land use indicators (R) and species traits (Q). Three independent ordinations were done before the RLQ analysis. Correspondence Analysis (CA) was performed on the L table that provides a joint scaling of plot and species scores and aims to maximize the correlation between them; Principle component analysis (PCA) for the R table and Q table. The final RLQ analysis was performed by a three-table co-inertia analysis. RLQ analysis selects axes that maximize the co-variance between the plot scores constrained by the land use variables (environmental variables; R) and the species scores constrained by the species traits (Q; (Doledec and Chessel 1994)). These axes scores are a compromise between maximizing the correlation and explaining the variation in each matrix. RLQ analysis was conducted using the rlq function of the ade4 (Chessel et al. 2004; Dray et al. 2007; Dray and Dufour 2007).

The additional fourth-corner statistic (Legendre et al. 1997; Dray and Legendre 2008) was performed again on the land use indicators (R), species cover (L) and species trait (Q) tables. With the fourth-corner method, the link is measured by a Pearson correlation coefficient for two quantitative variables (trait and land use variables), by a Pearson Chi2 and G statistic for two qualitative variables and by a Pseudo-F and Pearson r for one quantitative variable and one qualitative variable.

The second step of the analysis offers a multivariate statistic (equal to the sum of eigenvalues of RLQ analysis) and measures the link between two variables by a square correlation coefficient

(quantitative/quantitative), a Chi2/sum (L) (qualitative/qualitative) and a correlation ratio (quantitative/qualitative). Five different permutation models can be used to test significance. We applied permutation model 1 (with 999 permutations) to test the null hypothesis (H0) that species are distributed according to their tolerance (“preferences”) to different land uses, but independent of their traits (Dray and Legendre 2008). Functions fourth-corner, fourthcorner2 and tablematrix in the R package “ade4” (Chessel et al. 2004; Dray et al. 2007; Dray and Dufour 2007)were applied. All analyses were done using the R version 3.0.2(R Core Team 2013). For RLQ and fourth-corner analyses we included all plant species that were present in more than 5% of the plots. Therefore only 72 of 129 plant species were used.

4. Results

4.1. Partitioning of plant and bird species diversity

All observed diversity measures used for partitioning on the plot, formation and landscape level were significantly different from simulated diversities (Fig. 3). For bird richness (Fig. 3a) the landscape level was expected to have the highest richness. However, observed richness exceeded the simulated values for richness on the plot and formation level and was lower than the expected contribution of the landscape level. The formation level contributes most to gamma diversity (50%). Plant richness (Fig. 3b) was significantly higher at the plot and formation level and significantly lower at the landscape level than the simulated data. The formation level contributed most (44%) followed by the plot level (31%) to gamma diversity. Simpson’s Index for bird diversity (Fig. 3c) was significantly higher than expected at the plot level and significantly lower at the formation and landscape level. The plot level contributed 90% to gamma diversity and 9.5% at the formation level. Simpson’s Index for plants (Fig. 3d) was similar to that of birds in that it was significantly higher at the plot level (almost 100% of total gamma diversity) with a marginal contribution of the formation level to the total gamma diversity.

Fig. 3 Partitioning of species richness (a and b) and Simpson’s index (c and d) for bird and plant species Significance values for r^2 are represented with stars for each level of partitioning between the bars (significance codes: p > 0 ‘***’; p > 0.001 ‘**’; p > 0.01 ‘*’ ¸V0.05 ‘.’ ).

4.1.1. Relationships between bird diversity and plant species and functional diversity

Within the bird survey 439 individuals belonging to 29 species were recorded. Correlation analysis showed that between plant functional diversity (functional richness, Rao’s Q) and plants

100

species diversity (species richness, Simpson’s index) versus bird diversity (richness, Simpson’s index) relationships were weak. However, the strongest correlations were found between functional richness and bird richness (r=0.55, p=0.022), Simpson’s Index for plants and bird richness (r=0.44, p=0.014) and plant richness and bird richness (r=0.46, p=0.01). The other correlations were weak (RaoQ/Simpson’s birds: r=-0.01, p=0.966, RaoQ/bird richness: r=0.12, p=0.527, functional richness/Simpson’s birds: r=0.29,p=0.122, Simpson’s plants/Simpson’s birds: r=0.32, p=0.091, plant richness/Simpson’s bird : r=0.2, p=0.279).

Functional diversity (richness and RaoQ) varied significantly between the formations (Fig. 4). The highest significant functional richness was found in shrubland and the lowest in grassland.

Fig. 4 Functional richness (a) and Rao’s Q quadratic entropy (RaoQ) (b) of the three formations (F=forest, G=grassland, S=shrubland). Wilcoxon two-sampled rank sum test confirmed significant differences in functional diversity between all formations and indices (p < 0.05).

4.2. Multivariate relationships of plant species traits and land use indicators

The first ordination RLQ-axis explained 93% of the total variance (Fig. 5). This first axis is dominated by the main structural gradient from open grassland (negatively related) to closed

101

Fig. 5 RLQ diagram for land use indicators and plant functional traits. The importance of each variable is shown by arrow length; the ankle visualizes the correlation between the variables.

forest (positively related). The land use variables shaping this gradient are correlated with this axis. In case of grazing the variables : number of droppings from small and large herbivores and the European hare are correlated with the left part of the axis, and indicators for cutting, the number of coppiced trees and the amount of decayed dead wood (dead.wood.3) with the right part of the axis. Traits strongly associated with the first axis are woodiness (non-woody plants

102

on the left part and woody plants on the right part), Raunkier life form (hemicryptophytes and chamaeophytes on the left part - and phanerophytes on the right part), phytogeographic origin (non-natives are related to the left part of the axis, natives are related to the right part of the axis) and growth form (herbs are related with the left part of the axis). A strong relation to the upper part of the second axis showed the shrub growth form, ballistochorous dispersal and spiny plants whereas non-spiny plants were related with the lower part of the second axis. an annual life cycle were positively related to browsing by European hare and large herbivores. Non-native plants were positively related to droppings from European hare, small and large herbivores. Hemicryptophytes and Therophytes were positively related to impacts of European hare, large- and small herbivores. Shrublands and the cover of the shrublayer were positively associated with spiny plants. Fleshy fruits were positively related to shrublands but also to forests, however showing a negative response to the forest related structural variables basal area and crown closure. Basal area and crown closure coincided to forest in their relation to most traits. This was true for example for the positive relation of the growth forms trees, climbers and epiphytes as well as to native plants, life cycle and anemochorous seed dispersal. Coppice and browsing impacts were negatively associated with the abundance of climbers and epiphytes. However, the number of coppiced trees was positively related to species with an annual and biennial life cycle.

103 Tab. 1 Fourth-corner statistic. Significance values are *** p <0.001, ** p<0.01 and * p >0.05.

eur. hare herb layer large herb. shrub layer small herb. tracks tree density tree layer

F r F r F r F r F r F r F r F r main dispersal type 162.08 *** 381.01 *** 170.41 *** 229.27 *** 239.66 ** 15.67 . 304.73 *** 317.56 *** anemochor -0.11 ** -0.23 ** -0.14 ** 0.13 ** -0.13 ** . . 0.23 ** 0.24 ** ballistochor -0.02 . 0.04 . 0.00 . 0.03 . -0.03 . . . -0.03 . -0.01 * endozoochor -0.01 . 0.06 . 0.03 . -0.01 . -0.04 . . . -0.10 ** -0.07 . epizoochor 0.19 ** 0.23 ** 0.17 ** -0.22 ** 0.22 ** . . -0.18 ** -0.19 ** hydrochor -0.02 . -0.04 . -0.03 . 0.08 . 0.05 * . . 0.05 . -0.01 . multiple 0.05 . 0.07 ** 0.06 * -0.04 . 0.07 . . . -0.07 ** -0.09 ** fruit type 377.67 *** 937.02 *** 477.50 ** 1151.02 *** 728.99 *** 155.64 ** 51.56 . 248.55 *** fleshy -0.15 ** -0.23 ** -0.17 ** 0.25 ** -0.20 ** 0.10 * . . 0.12 ** non-fleshy 0.15 ** 0.23 ** 0.17 ** -0.25 ** 0.20 ** -0.10 * . . -0.12 ** growth form 511.05 *** 3970.68 *** 1299.12 *** 1525.33 *** 622.95 *** 133.38 ** 2257.02 *** 3602.99 *** climber -0.06 ** -0.14 ** -0.09 ** 0.08 ** -0.06 ** 0.04 . 0.13 ** 0.14 ** epiphyte -0.01 . -0.02 * -0.01 * 0.01 . -0.01 . 0.02 . 0.00 . 0.03 ** herb 0.33 ** 0.66 ** 0.48 ** -0.51 ** 0.36 ** -0.17 ** -0.49 ** -0.56 ** shrub -0.10 ** -0.03 . -0.09 * 0.27 ** -0.12 ** 0.07 . -0.15 ** -0.17 ** tree -0.24 ** -0.60 ** -0.39 ** 0.30 ** -0.26 ** 0.11 * 0.55 ** 0.63 ** life cycle 105.93 *** 37.09 *** 28.37 ** 21.42 * 6.25 . 0.47 . 46.56 ** 62.88 *** annual 0.11 ** 0.07 ** 0.06 * 0.05 * . . . . -0.07 ** -0.08 ** biennial -0.01 . 0.01 * 0.00 . -0.01 . . . . . -0.01 . -0.02 ** perennial -0.11 ** -0.07 ** -0.06 * 0.05 * . . . . 0.07 ** 0.09 ** phytogeo. origin 1397.27 *** 10459.80 *** 3158.08 *** 4750.92 *** 2592.67 *** 324.09 *** 4538.91 *** 5821.86 *** non-native 0.28 ** 0.62 ** 0.40 ** -0.47 ** 0.37 ** -0.14 ** -0.46 ** -0.51 ** native -0.28 ** -0.62 ** -0.40 ** 0.47 ** -0.37 ** 0.14 ** 0.46 ** 0.51 ** Raunkier life form 775.96 *** 4889.11 1757.84 *** 2290.86 969.57 *** 142.05 *** 1963.68 *** 2509.90 ***

chamaephyte 0.06 * 0.14 ** 0.16 ** -0.12 ** -0.01 . -0.02 . -0.10 ** -0.10 ** hemicryptophyte 0.27 ** 0.62 ** 0.40 ** -0.48 ** 0.38 ** -0.15 ** -0.46 ** -0.50 ** phanerophyte -0.33 ** -0.68 ** -0.49 ** 0.54 ** -0.37 ** 0.15 ** 0.51 ** 0.56 ** therophyte 0.15 ** 0.09 ** 0.10 ** -0.07 ** 0.02 . -0.01 . -0.09 ** -0.10 ** spinescence 188.93 ** 197.86 ** 232.44 ** 1390.21 *** 250.12 ** 11.17 . 17.18 . 102.18 . non-spiny 0.11 ** 0.11 * 0.12 ** -0.28 ** 0.12 ** . . . . 0.08 . spiny -0.11 ** -0.11 * -0.12 ** 0.28 ** -0.12 ** . . . . -0.08 . woodiness 2000.22 *** 12691.50 *** 4894.13 *** 6607.62 *** 2441.17 *** 366.20 *** 5363.34 *** 6838.29 *** non-woody 0.33 ** 0.66 ** 0.48 ** -0.53 ** 0.36 ** -0.15 ** -0.49 ** -0.54 ** woody -0.33 ** -0.66 ** -0.48 ** 0.53 ** -0.36 ** 0.15 ** 0.49 ** 0.54 **

104

browse coppiced crown formation basal.area dead wood 1 dead wood 2 dead wood 3 index trees closure forest grassland shrubland F r F r F r F r F r F r F r G r G r G r main dispersal 218.60 *** 31.71 . 169.52 ** 265.95 *** 31.36 . 25.18 . 52.40 . 2581.84 *** type anemochor 0.18 ** . . 0.19 ** 0.21 ** . . . . 0.06 . + * - * + * ballistochor . . . . -0.04 ** 0.01 . . . . . 0.00 . . . - . . . endozoochor -0.04 . . . -0.07 . -0.04 . . . . . 0.03 . . . - * . . epizoochor -0.18 ** . . -0.12 ** -0.18 ** . . . . -0.11 . - * + * . . hydrochor -0.02 . . . -0.04 . -0.06 . . . . . -0.02 ...... multiple -0.07 ** . . -0.06 ** -0.08 ** . . . . -0.03 . - . . . . . fruit type 372.53 *** 150.63 * 19.21 . 424.89 *** 70.50 . 147.80 * 153.46 * 1610.87 *** fleshy 0.15 ** 0.09 * . . 0.16 ** . . 0.09 * 0.10 * + ** - ** + ** non-fleshy -0.15 ** -0.09 * . . -0.16 ** . . -0.09 * -0.10 * . . . . - * growth form 2589.93 *** 128.86 *** 960.74 *** 3067.86 *** 8.23 . 135.31 *** 306.01 *** 12019 *** climber 0.15 ** -0.04 * 0.07 * 0.14 ** . . 0.07 * 0.04 . + * - ** - * epiphyte 0.04 ** -0.01 . -0.01 . 0.03 ** . . 0.00 . 0.00 . + * . . . . herb -0.52 ** 0.01 . -0.35 ** 0.54 ** . . -0.17 ** -0.24 ** - * + * . . shrub -0.13 ** 0.16 ** -0.11 ** -0.15 ** . . 0.03 . 0.00 . . . - * + * tree 0.56 ** -0.11 . 0.41 ** 0.60 ** . . 0.12 ** 0.23 ** + * - * - * life cycle 53.74 ** 20.21 * 22.40 * 55.49 *** 8.75 . 0.56 . 2.51 . 165.799 *** annual -0.08 ** 0.05 . -0.05 ** -0.08 ** ...... - ** . . . . biennial -0.01 * 0.02 . -0.01 * -0.02 ** ...... perennial 0.08 ** -0.05 . 0.05 ** 0.08 ** ...... + ** - ** . . phytogeo. origin 4841.42 *** 20.78 . 1762.90 *** 4784.35 *** 0.14 . 321.58 *** 833.15 *** 7471.91 *** non-native -0.47 ** . . -0.31 ** -0.47 ** . . -0.14 ** -0.22 ** - ** + ** . . native 0.47 ** . . 0.31 ** 0.47 ** . . 0.14 ** 0.22 ** + ** - ** . . Raunkier life form 1959.33 *** 11.85 . 770.05 *** 2077.60 *** 18.43 . 150.85 *** 365.81 *** 9163.64 *** chamaephyte -0.08 ** . . -0.08 ** -0.09 ** . . -0.03 . -0.04 . . . .1 . . . hemicryptophyte -0.46 ** . . -0.31 ** -0.47 ** . . -0.15 ** -0.23 ** - * + * . . phanerophyte 0.51 ** . . 0.35 ** 0.52 ** . . 0.16 ** 0.25 ** + * - * . . therophyte -0.09 ** . . -0.06 ** -0.09 ** . . -0.01 . -0.03 . - * + * . . spinescence 112.89 . 131.44 * 9.80 . 102.25 . 43.84 . 23.14 . 27.81 . 1542.42 *** non-spiny 0.08 . -0.09 . . . 0.87 ...... + 0.006 - * - * spiny -0.08 . 0.09 . . . 0.13 ...... 0.39 - ** + ** woodiness 5108.19 *** 4.32 . 2141.89 *** 5592.47 *** 41.66 . 515.43 *** 1063.00 *** 8576.87 *** non-woody -0.48 ** . . -0.34 ** -0.50 ** . . -0.17 ** -0.24 ** - 0.006 + ** . . woody 0.48 ** . . 0.34 ** 0.50 ** . . 0.17 ** 0.24 ** + 0.006 - ** . .

105 5. Discussion

5.1. Importance of the hierarchical levels for species diversity conservation

Gamma diversity of different taxa is not necessarily equally distributed either in the environment or at different scales (Fleishman et al. 2003). However, in this study we found a coincidence in the partitioning patterns of plant and bird richness and Simpson diversity between hierarchical levels.

Considering plant and bird species richness, the formation level (beta1) contributed the most to gamma richness. The importance of this intermediate hierarchical level for plant species richness has also been observed in vegetation studies in other biogeographic regions (i.e. Wagner et al. 2000; Chandy et al. 2006; Chiarucci et al. 2008; Chávez and Macdonald 2012). A classification of the vegetation in the same study area resulted in vegetation types which basically coincided with the formations, but were also transient between grassland and shrubland formations but especially between shrubland and forest formations due to the impacts of cutting and grazing (Seis et al. 2014). These results confirm the high importance of the beta1 level to gamma diversity with that of beta2 level resulting from land use caused transitions.

At the landscape level (beta2) birds made a larger contribution to gamma diversity than plants did. That means that the habitat differentiation for birds was more relevant than for vegetation. Heterogenic land use impacts produce small-scale vegetation dynamics like regeneration and disturbance dynamics that may result in transitions between the formations. The great majority of the birds we observed were forest species. Some of them, like the Campephilus magellanicus may be very specialized and depend on continuous heterogeneous forest habitats with high proportions of large trees (Armesto et al. 2009). Habitat specialists also occurred in grasslands; for example Vanellus chilensis is an open area specialist. Although in Chile it naturally inhabits wetlands and the marine coastline, it has expanded its habitat to include anthropogenic

106

pastures. Habitat specialists that were restricted to specific formations made the highest contribution to the beta1 level to the total gamma diversity. However, many bird species were present in more than one habitat type including some that are typical of the temperate rainforest like those of the Rhinocryptidae. Species belonging to this family live in the forest understory but use forest edges and shrublands as corridors (Díaz et al. 2006). Their occurrence in shrublands could be reduced by barriers such as grasslands situated between forest and shrubland habitats or in shrublands lacking a closed shrub layer making the small birds susceptible to predators (Willson et al. 2001). However, these multi-habitat users explain the contribution of the beta2 level to gamma diversity of birds.

Considering the Simpson index, the individual plot scale contains an important amount of bird and plant diversity. Simpson index is the probability of randomly drawing two individuals belonging to the same species from one sample (Gering et al. 2003). It is very sensitive to changes in the abundance of common species (Hill 1973; Magurran 2004). Various species of different plant growth forms show local dominance including the non-native grass species Agrostis capillaris, and native shrubs Gaultheria mucronata, Berberis microphylla but also several tree species including Dasyphyllum diacanthoides and Drimys winteri. This causes a high heterogeneity in dominance at the plot level which is likely due to the heterogeneity of land use impacts on the individual plot. The impacts create diverse habitats and may promote those species that are resistant or resilient to disturbances and at the same time diminish more susceptible species. This heterogeneity in species dominance at the plot level leads to the high contribution that the plot level has on gamma diversity considering Simpson Index. The same pattern seen in plants was also seen in birds, only a little weaker. This is due to the fact that local, highly dominant bird species also occur with a high frequency. A good example of this in our dataset is that of Elaenia albiceps. This habitat generalist species is the only long distance migratory songbird in temperate rainforests that occurs at very high abundances, although only during the breeding season (Brown et al. 2007).

107 The bird diversity patterns among different scales likely reflect the distribution of resources (Fleishman et al. 2003). The fact that these resources are largely provided by the vegetation may lead to the coincidences in diversity patterns of plants and birds.

5.2. Role of functional diversity as a surrogate for plant and bird diversity

A notable feature of South American temperate rain forests is the importance of mutualistic interactions between plants and animals that benefit plant reproduction (Smith-Ramírez and Armesto 1994). This leads to the assumption that a close relationship exists between plant and bird diversity. However, it was not plant species richness but functional richness that correlated best with bird richness.

Plant functional diversity (richness and RaoQ) was highest in shrublands. This may be due to the transitional role that shrublands play within rural landscapes between open grassland and closed forest mosaic elements. Shrublands in the temperate rainforest region are known to provide forest understory- like habitat structures and food resources for birds (Armesto et al. 1987; Willson et al. 2001; Díaz et al. 2006). Forests also showed high values for both diversity indices. Grasslands had significantly lower diversity, due to the lack of structural richness created by life and growth forms other than herbs, and due to the low variability in fruit types. There was a correlation between bird richness and functional richness with increasing diversity in both shrublands and forests. Shrublands and forests offer many habitat and feeding niches for birds and therefore they may use both habitats. The use of forest and shrubland habitats by species that inhabit the forest understory has been proven. Some forest species in Chile such as Sylviorthorhynchus desmursii or the group in the Rhinocryptidae use shrublands as corridors (Willson et al. 2001; Díaz et al. 2006). The Chucao, a bird species, was the only one that used both forest and shrubland habitats. However, Sylviorthorhynchus desmursii and Eugralla paradoxa were found restricted to shrublands and Pteroptochos tarnii to forests. None of these species used the grasslands; in fact the grasslands in the study area were poor in bird species.

108

Grassland birds were restricted to some specialists of open areas like Vanellus chilensis, Sicalis luteola or Zonotrichia capensis that used grassland and shrubland. Some generalists like Turdus falcklandii, Sephanoides galeritusa and Elaenia albiceps were also found in grasslands but at lower abundances than in forests and shrublands. However, we can confirm the results of Diaz et al. (2006) and Willson et al. (2001) that grasslands within the vegetation mosaic seem to be an obstacle for the movement of most forest bird species within the landscape.

Overall the Simpson Index for birds was weakly correlated with all other diversities. This again may be due to the high local abundances of some species (see above) leading to a high plot variability in Simpson Index. However, the results of this study underpin the usefulness of plant functional diversity as a surrogate for bird diversity in rural landscapes impacted by small-scale farming.

5.3. Multivariate changes in functional traits to the main physical structure components of the vegetation and land use

Land use and vegetation structure are associated with changes in plant functional types. The three formations (grassland, shrubland and forest) that were stratified in the field, represent a vegetation gradient that reflects decreasing grazing impacts and increasing cutting impacts from grassland to shrubland and forest respectively (Seis et al. 2014).

In those grasslands shaped by grazing, therophytes and hemicryptophytes survive because they are resilient to grazing. Most importantly, grazing influences the ratio of non-native to native species. High abundance of non-natives in grasslands is a general issue found in grasslands in temperate Chile (Ramírez et al. 1989). Grazing animals may help to spread introduced species (Vavra et al. 2007). The increase of epizoochor dispersal with an increase in grazing impacts proves the role of grazers in the dispersal of non-native species.

109 Shrublands contain many of the plants and their features known to be characteristic of temperate rainforests such as native trees and shrubs, climbers and fleshy fruits. However, all of these traits were shown to be susceptible to grazing impacts indicated not only by dung piles of grazers, but also by the browsing index that was especially connected with shrubland. Spiny species were also strongly correlated with shrublands. However, a direct causal relationship between the effects of grazing and the promotion of unpalatable spiny species (Cornelissen et al. 2003) could not been proven by this study. It is likely that the occurrence of spiny species was due more to historical grazing than to recent grazing, but this differentiation could not be quantified by this study.

From a functional perspective, the forests remnants showed some signs of having preserved the typical traits of temperate rainforests. This was exemplified by the presence of species with native phytogeographic origin, climbers and (vascular) epiphytes. Impacts from past cutting practices, (dead.wood.3 and coppiced trees) were related to this gradient showing that traits associated with the forest are a sign that the forests are recovering from past impacts. However, the traits that characterize forests were all generally negatively impacted by grazing.

6. Conclusions

The inclusion of landscapes that developed as a result of small-scale farming into biodiversity conservation is promising especially in the Chilean temperate rainforest region which is dominated by plantations. For the first time plant and bird species diversity in rural areas in the Valdivian Coastal Range was partitioned on different hierarchical levels. Hierarchical scale is of crucial importance. The formation level should be the focal scale for the conservation of bird and plant species richness. Vegetation and bird richness within the landscape are closely related and can be measured by functional diversity indices. Grasslands were shown to be of minor importance not only for bird diversity but also for plant diversity. Grasslands were dominated by non-native species, an indicator of strong human influence (Sukopp 1969).

110

In contrast forests and shrublands were suitable habitat elements and farmers should be encouraged to maintain these habitats. Proposals to convert shrublands and forests into grasslands should be prevented. The rural landscapes that developed in conjunction with small- scale farming in the Valdivian Coastal Range serve as refuges for native species as do some other types of cultural landscapes. Future research topics in this area should focus on the identification of key forest structures and thresholds in terms of forest patch size for those species that depend on forest habitat. To make further assumptions about the value of these formations for ecosystem restoration, the potential for tree regeneration on the landscape scale should be addressed.

7. Acknowledgements

The authors thank the MPI-BGC Jena, who host TRY, and the international funding networks supporting TRY (IGBP, DIVERSITAS, GLP, NERC, QUEST, FRB and GIS Climate). Marco Flores, Magdalena Gerhardt, Danisa Paredes and Melanie Welling supported the fieldwork. We thank all landowners in Lomas del Sol and Camán for permitting us the fieldwork and Anke Stein, Felipe Osorio, Peter Borchardt and Osvaldo Vidal for critical comments on the early manuscript and Bernhard Thiel for improving the English.

111 8. References

Allan JD (1975) Components of diversity. Oecologia 18:359–367.

Armesto JJ (2002) Relevancia de las pequeñas áreas silvestres para la conservación de la biodiversidad en el bosque nativo. Ambiente Desarro 18:44–50.

Armesto JJ (1998) Conservation Targets in South American Temperate Forests. Science 282:1271–1272.

Armesto JJ, Rozzi R (1989) Seed dispersal syndromes in the rain forest of Chiloé: evidence for the importance of biotic dispersal in a temperate rain forest. J Biogeogr 16:219–226.

Armesto JJ, Rozzi R, Miranda P, Sabag C (1987) Plant/frugivore interactions in South American temperate forests. Rev Chil Hist Nat 60:321–336.

Armesto JJ, Smith-Ramírez C, Carmona MR, et al. (2009) Old-growth temperate rainforests of South America: conservation, plant–animal interactions, and baseline biogeochemical processes. In: Wirth, C., Gleixner, G., & Heimann, M. (eds.). Old-growth forests, 1st edn. Springer, Berlin Heidelberg,pp 367–390.

Aubin I, Gachet S, Messier C, Bouchard A (2007) How resilient are northern hardwood forests to human disturbance? An evaluation using a plant functional group approach. Ecoscience 14:259– 271.

Barbaro L, Corcket E, Dutoit T, Peltier J-P (2000) Réponses fonctionnelles des communautés de pelouses calcicoles aux facteurs agro-écologiques dans les Préalpes françaises. Can J Bot 78:1010–1020.

Bartha S, Merolli A, Campetella G, Canullo R (2008) Changes of vascular plant diversity along a chronosequence of beech coppice stands, central Apennines, Italy. Plant Biosyst - Int J Deal Asp Plant Biol 142:572–583.

Bibby CJ, Burgess ND, Hill DA (1992) Bird census techniques. Cambridge Univ. Press, Cambridge

112

Bleher B, Potgieter CJ, Potgeiter CJ, et al. (2003) The Importance of Figs for Frugivores in a South African Coastal Forest. J Trop Ecol 19:375–386.

Botta-Dukát Z (2005) Rao’s quadratic entropy as a measure of functional diversity based on multiple traits. J Veg Sci 16:533–540.

Brown CE, Anderson CB, Ippi S, et al. (2007) The Autecology of the Fío-Fío (Elaenia Albiceps Lafresnaye and D’Orbigny) in Subantarctic Forests of the Cape Horn Biosphere Reserve, Chile. An. Inst. Patagon. Universidad de Magallanes, pp 29–40.

Campetella G, Botta-Dukát Z, Wellstein C, et al. (2011) Patterns of plant trait–environment relationships along a forest succession chronosequence. Agric Ecosyst Environ 145:38–48.

Canullo R, Campetella G, Mucina L, et al. (2010) Patterns of Clonal Growth Modes Along a Chronosequence of Post-Coppice Forest Regeneration in Beech Forests of Central Italy. Folia Geobot 46:271–288.

Chandy S, Gibson DJ, Robertson PA (2006) Additive partitioning of diversity across hierarchical spatial scales in a forested landscape: Additive partitioning of diversity. J Appl Ecol 43:792–801.

Chávez V, Macdonald SE (2012) Partitioning vascular understory diversity in mixedwood boreal forests: The importance of mixed canopies for diversity conservation. For Ecol Manag 271:19– 26.

Chazdon RL (2003) Tropical forest recovery: legacies of human impact and natural disturbances. Perspect Plant Ecol Evol Syst 6:51–71.

Chessel D, Dufour AB, Thioulouse J (2004) The ade4 package-I-One-table methods. R News 4:5– 10.

Chiarucci A, Bacaro G, Rocchini D (2008) Quantifying plant species diversity in a Natura 2000 network: Old ideas and new proposals. Biol Conserv 141:2608–2618.

CIREN (2001) Descripciones de suelos. Materiales y simbolos. Estud. Agrol. X Región. Centro de Información de Recursos Naturales (CIREN), Santiago, Chile, p 412

Cody ML (1985) Habitat selection in birds. Academic Press, Orlando, Florida

113 Cornelissen JHC, Lavorel S, Garnier E, et al. (2003) A handbook of protocols for standardised and easy measurement of plant functional traits worldwide. Aust J Bot 51:335.

Crist TO, Veech JA, Gering JC, Summerville KS (2003) Partitioning Species Diversity across Landscapes and Regions: A Hierarchical Analysis of α, β, and γ Diversity. Am Nat 162:734–743.

Decocq G, Aubert M, Dupont F, et al. (2004) Plant diversity in a managed temperate deciduous forest: understorey response to two silvicultural systems. J Appl Ecol 41:1065–1079.

Díaz I, Armesto JJ, Reid S, et al. (2005) Linking forest structure and composition: avian diversity in successional forests of Chiloé Island, Chile. Biol Conserv 123:91–101.

Díaz I, Armesto JJ, Willson MF (2006) Mating success of the endemic Des Murs’ Wiretail (Sylviorthorhynchus desmursii, Furnariidae) in fragmented Chilean rainforests. Austral Ecol 31:13–21.

Díaz S, Lavorel S, McINTYRE S, et al. (2007) Plant trait responses to grazing ? a global synthesis. Glob Change Biol 13:313–341.

Doledec S, Chessel D (1994) Co-inertia analysis: an alternative method for studying species- environment relationships. Freshw Biol 31:277–294.

Dolédec S, Chessel D, Braak CJF, Champely S (1996) Matching species traits to environmental variables: a new three-table ordination method. Environ Ecol Stat 3:143–166.

Dray S, Dufour A-B (2007) The ade4 package: implementing the duality diagram for ecologists. J Stat Softw 22:1–20.

Dray S, Dufour AB, Chessel D (2007) The ade4 package II: Two-table and K-table methods. R News 7:47–52.

Dray S, Legendre P (2008) Testing the species traits-environment relationships: the fourth- corner problem revisited. Ecology 89:3400–3412.

Dray S, Pettorelli N, Chessel D (2002) Matching data sets from two different spatial samples. J Veg Sci 13:867–874.

114

Echeverría C, Newton AC, Lara A, et al. (2007) Impacts of forest fragmentation on species composition and forest structure in the temperate landscape of southern Chile. Glob Ecol Biogeogr 16:426–439.

Ellenberg H, Leuschner C (2010) Vegetation Mitteleuropas mit den Alpen in ökologischer, dynamischer und historischer Sicht.(6., vollst. neu bearb. und stark erw. Aufl. ed.). Ulmer Verlag, Stuttgart, Germany.

Fleishman E, Betrus C, Blair R (2003) Effects of spatial scale and taxonomic group on partitioning of butterfly and bird diversity in the Great Basin, USA. Landsc Ecol 18:675–685.

Fleming TH, Breitwisch R, Whitesides GH (1987) Patterns of Tropical Vertebrate Frugivore Diversity. Annu Rev Ecol Syst 18:91–109.

Foster D, Motzkin G (1998) Ecology and Conservation in the Cultural Landscape of New England: Lessons from Nature’s History. Northeast Nat 5:111–126.

Freiberg M (1997) Spatial and temporal pattern of temperature and humidity of a tropical premontane rain forest tree in Costa Rica. Selbyana 18:77–84.

Garnier E, Cortez J, Billès G, et al. (2004) Pland functional markers capture ecosystem properties during secondary succession. Ecology 85:2630–2637.

Gering JC, Crist TO, Veech JA (2003) Additive Partitioning of Species Diversity across Multiple Spatial Scales: Implications for Regional Conservation of Biodiversity. Conserv Biol 17:488–499.

Godfray HCJ, Lawton JH (2001) Scale and species numbers. Trends Ecol Evol 16:400–404.

Graae BJ, Sunde PB (2000) The impact of forest continuity and management on forest floor vegetation evaluated by species traits. Ecography 23:720–731.

Guariguata MR, Ostertag R (2001) Neotropical secondary forest succession: changes in structural and functional characteristics. For Ecol Manag 148:185–206.

Halpern CB (1989) Early Successional Patterns of Forest Species: Interactions of Life History Traits and Disturbance. Ecology 70:704–720.

115 Harrell FE, Dupont MC (2007) The Hmisc Package.

Harvey CA, Komar O, Chazdon R, et al. (2008) Integrating Agricultural Landscapes with Biodiversity Conservation in the Mesoamerican Hotspot. Conserv Biol 22:8–15.

Herrera CM (1985) Habitat-Consumer Interactions in Frugivorous Birds. In: Cody, ML (ed) Habitat selection in birds. Academic Press, Orlando, Florida, pp 341–365.Hildebrand-Vogel R (2002) Structure and dynamics of southern Chilean natural forests with special reference to the relation of evergreen versus deciduous elements. Folia Geobot 37:107–128.

Hill MO (1973) Diversity and Evenness: A unifying notation and its consequences. Ecology 54:427–432.

Jiménez JE (2000) Effect of sample size, plot size, and counting time on estimates of avian diversity and abundance in a Chilean rainforest. J Field Ornithol 71:66–88.

Kahmen S, Poschlod P (2008) Effects of grassland management on plant functional trait composition. Agric Ecosyst Environ 128:137–145.

Kissling WD, Field R, Böhning-Gaese K (2008) Spatial patterns of woody plant and bird diversity: functional relationships or environmental effects? Glob Ecol Biogeogr 17:327–339.

Kissling WD, Rahbek C, Bohning-Gaese K (2007) Food plant diversity as broad-scale determinant of avian frugivore richness. Proc R Soc B Biol Sci 274:799–808.

Laliberté E, Legendre P (2010) A distance-based framework for measuring functional diversity from multiple traits. Ecology 91:299–305.

Laliberté E, Shipley B (2010) FD: Measuring functional diversity (FD) from multiple traits, and other tools for functional ecology. R Package Version 10 9:

Lande R (1996) Statistics and Partitioning of Species Diversity, and Similarity among Multiple Communities. Oikos 76:5–13.

Lavorel S, Garnier E (2002) Predicting changes in community composition and ecosystem functioning from plant traits: revisiting the Holy Grail. Funct Ecol 16:545–556.

116

Lavorel S, Touzard B, Lebreton J-D, Clément B (1998) Identifying functional groups for response to disturbance in an abandoned pasture. Acta Oecologica 19:227–240.

Legendre P, Galzin R, Harmelin-Vivien ML (1997) Relating Behavior to Habitat: Solutions to the Fourth-corner Problem. Ecology 78:547–562.

Londo G (1976) The decimal scale for releves of permanent quadrats. Vegetatio 33:61–64.

Loreau M (2000) Are communities saturated? On the relationship between alpha, beta and gamma diversity. Ecol Lett 3:73–76.

MacArthur RH, MacArthur JW (1961) On Bird Species Diversity. Ecology 42:594. doi: 10.2307/1932254

Magurran AE (2004) Measuring biological diversity. Blackwell Publishing, Oxford, UK

Mason C, MacDonald S (2002) Responses of ground flora to coppice management in an English woodland – a study using permanent quadrats. Biodivers Conserv 11:1773–1789.

Mládek J, Hejcman M, Hejduk S, et al. (2010) Community Seasonal Development Enables Late Defoliation Without Loss of Forage Quality in Semi-natural Grasslands. Folia Geobot 46:17–34.

Moorman MC, Peterson N, Moore SE, Donoso PJ (2013) Stakeholder Perspectives on Prospects for Co-Management of an Old-Growth Forest Watershed Near Valdivia, Chile. Soc Nat Resour 26(9): 1022-–1036.

Myers N, Mittermeier RA, Mittermeier CG, et al. (2000) Biodiversity hotspots for conservation priorities. Nature 403:853–858.

Navarro T, Alados CL, Cabezudo B (2006) Changes in plant functional types in response to goat and sheep grazing in two semi-arid shrublands of SE Spain. J Arid Environ 64:298–322.

Neira E, Verscheure H, Revenga C (2002) Chile’s frontier forests: conserving a global treasure. Global Forest Watch, World Resources Institute

Núñez D, Nahuelhual L, Oyarzún C (2006) Forests and water: The value of native temperate forests in supplying water for human consumption. Ecol Econ 58:606–616.

117 Oberdorfer E (1960) Pflanzensoziologische Studien in Chile: Ein Vergleich mit Europa. J. Cramer, Weinheim, Germany

Oksanen, J, O’Hara RB, Simpson GL, et al. (2013) vegan: Community Ecology Package.

Petit LJ, Petit DR (2003) Evaluating the Importance of Human-Modified Lands for Neotropical Bird Conservation. Conserv Biol 17:687–694.

Plachter H, Rössler M (1995) Cultural Landscapes: Reconnecting Culture and Nature. In: von Droste B, Plachter H, Rössler M (eds) Cultural Landscapes of Universal Value , 1st edn. Gustav Fischer Verlag, Jena, Stuttgart, New York, p 15.

R Core Team (2013) R: A language and environment for statistical computing. R Found. Stat. Comput. Vienna Austria.

Ramírez C, Hauenstein E, Martín JS, Contreras D (1989) Study of the Flora of RuCamánque, Cautin Province, Chile. Ann Mo Bot Gard 76:444–453.

Ramírez C, Moraga M, Figueroa H (1984) La similitud florística como medida de degradación antrópica del bosque valdiviano. Agro Sur 12:127–139.

Relva MA, Veblen TT (1998) Impacts of introduced large herbivores on Austrocedrus chilensis forests in northern Patagonia, Argentina. For Ecol Manag 108:27–40.

Remsen JV, Cadena CD, Jaramillo A, et al. (2014) A classification of the bird species of South America. In: Am. Ornithol. Union Httpwww Mus. Lsu Edu RemsenSACCBaseline Html. Accessed 1 Feb 2014.

Rouvinen S, Kuuluvainen T, Karjalainen L (2002) Coarse woody debris in old Pinus sylvestris dominated forests along a geographic and human impact gradient in boreal Fennoscandia. Can J For Res 32:2184–2200.

San Martín C, Villagra J, Novoa C (2009) Comparación de manejos pratenses del Centrosur de Chile utilizando valores bioindicadores de Ellenberg. Gayana Bot 66:158–170.

Scott JM, Ralph CJ (1981) Estimating numbers of terrestrial birds. Studies in Avian Biology 6.

118

Seis K, Gärtner S, Reif A, Donoso PJ (2014) The effect of small-scale land use on vegetation in the Valdivian Coastal Range (Chile). Community Ecol 15(2):194-204.

Smith-Ramírez C (2004) The Chilean coastal range: a vanishing center of biodiversity and endemism in South American temperate rainforests. Biodivers Conserv 13:373–393. doi: 10.1023/B:BIOC.0000006505.67560.9f

Smith-Ramírez C, Armesto JJ (1994) Flowering and Fruiting Patterns in the Temperate Rainforest of Chiloe, Chile--Ecologies and Climatic Constraints. J Ecol 85: 353–365.

Smith-Ramírez C, Martinez P, Nuñez M, et al. (2005) Diversity, flower visitation frequency and generalism of pollinators in temperate rain forests of Chiloe Island, Chile. Bot J Linn Soc 147:399–416. doi: 10.1111/j.1095-8339.2005.00388.x

Snow DW (1971) Evolutionary aspects of fruit‐eating by birds. Ibis 113:194–202.

Soto-Mora Y, Urrutia J (2010) Evaluación del estado de conservación de cuatro especies de aves de la familia Rhinocryptidae en un área silvestre protegida en el Sur de Chile. Acta Zool Mex 26:401–414.

Stohlgren TJ (2007) Measuring plant diversity: lessons from the field. Oxford University Press, Oxford, UK

Sukopp H (1969) Der Einfluss des Menschen auf die Vegetation. Vegetatio 17:360–371.

Tscharntke T, Klein AM, Kruess A, et al. (2005) Landscape perspectives on agricultural intensification and biodiversity-ecosystem service management. Ecol Lett 8:857–874.

Vavra M, Parks CG, Wisdom MJ (2007) Biodiversity, exotic plant species, and herbivory: the good, the bad, and the ungulate. For Ecol Manag 246:66–72.

Veblen TT, Mermoz M, Martin C, Ramilo E (1989) Effects of Exotic Deer on Forest Regeneration and Composition in Northern Patagonia. J Appl Ecol 26:711–724.

119 Verheyen K, Honnay O, Motzkin G, et al. (2003) Response of forest plant species to land-use change: a life-history trait-based approach. J Ecol 91:563–577.

Vila AR, Borrelli L (2011) Cattle in the Patagonian forests: Feeding ecology in Los Alerces National Reserve. For Ecol Manag 261:1306–1314.

Villéger S, Mason NWH, Mouillot D (2008) New multidimensional functional diversity indices for a multifaceted framework in functional ecology. Ecology 89:2290–2301.

Wagner H, Wildi O, Ewald K (2000) Additive partitioning of plant species diversity in an agricultural mosaic landscape. Landsc Ecol 15:219–227.

Weiher E, Werf A, Thompson K, et al. (1999) Challenging Theophrastus: A common core list of plant traits for functional ecology. J Veg Sci 10:609–620.

Whittaker RJ, Willis KJ, Field R (2001) Scale and species richness: towards a general, hierarchical theory of species diversity. J Biogeogr 28:453–470.

Willson MF, Morrison JL, Sieving KE, et al. (2001) Patterns of predation risk and survival of bird nests in a Chilean agricultural landscape. Conserv Biol 15:447–456.

120

Chapter 3:The resilience of the of evergreen rainforest tree species to smallholders land use in the Valdivian Coastal Range (Chile)

1. Abstract

The regeneration conditions for native trees are altered by land uses. In the Valdivian Coastal Range native tree species are often restricted to areas where smallholders have altered the landscape by small-scale land uses such as livestock grazing, timber cutting and coppice forestry.

This study aimed to (1) classify tree species that regenerate within a vegetation mosaic of forest, shrubland and grassland into regeneration types (RTs) based on their functional traits, (2) test the response of those a priori defined RTs to measured indicators of land use impacts for tree harvesting, livestock grazing and coppice forestry, and (3) compare the density of regeneration by species in grassland, shrubland and forest.

The frequencies of tree species regenerating were assessed in 85 plots. Seedling and sapling densities were assessed in 30 plots equally distributed among the three strata.

Regeneration types were classified based on seven nominal traits that may influence regeneration. For this we applied flexible beta clustering and Gower distance. Recursive partitioning within a conditional inference framework was used to explain variation in the regeneration types as a function of the land use variables. Seedling densities per ha were compared in the three strata for the most abundant regenerating tree species. Five regeneration types were identified. For the long lived, shade-tolerant endozoochorous broadleaved (long-tol-endo) and long lived, shade-tolerant, anemochorous broad-leaved (long-tol-

121 ane) regeneration types, the impacts of large herbivores had the greatest impact. The short lived, semi shade-tolerant, endozoochorous broad-leaved type (short-semi-endo) was restricted by a combination of large herbivores and cutting effects. (2) Short lived, shade-intolerant, anemochorous broad-leaved (short-int-ane) type regenerated only in shrublands and regeneration of long lived, semi shade-tolerant, endozoochorous conifers (long-semi-endo) was restricted to forests. Species with the highest seedling densities belonged to the long-tol-endo and the long-tol-ane regeneration types. Seedling densities were highest in forests and nearly absent in grasslands.

Forest restoration and maintenance requires both protection from livestock and forestry practices that are sustainable. At the same time sustainable land use practices have to be developed with landowners who need to be financially compensated.

Keywords: Cultural landscape, regeneration types, facilitation, Aextoxicon puntatum, Amomyrtus luma, Eucryphia cordifolia, Lomatia ferrunginea, Myrceugenia planipes, natural regeneration.

2. Introduction

Allowing for the natural regeneration of native tree species is the most powerful and cost effective restoration strategy for forest biodiversity and ecosystem services (Chazdon, 2008). Land use activities alter the natural conditions for native tree species to regenerate and in the long term might lead to a loss of ecosystem functions and services. In the Valdivian temperate evergreen rainforest, a threatened ecosystem of outstanding biodiversity (Myers et al., 2000; Olson and Dinerstein, 1998), native tree species are often restricted to areas where smallholders practice small scale land use activities. For the smallholders in the Valdivian Coastal Range (VCR) and for most areas in south-central Chile the selling of timber for firewood and charcoal for the traditional barbecue is the most important source of livelihood income

122

(Moorman et al., 2013). However, trees harvested are not replaced by planting and livestock grazing is allowed in the area leading to forest degradation. The effect of these land use activities on the natural regeneration of native tree species has not been studied even though these activities are very common.

In primary forests of the Valdivian temperate evergreen rainforest in the VCR most native tree species naturally regenerate in gaps (Veblen et al., 1981). However, in the mid elevations (ca. 200 - 700 m) natural conditions have been dramatically altered thereby affecting forest growth and regeneration. The landscape today is the result of a massive exploitation of the primary forest during colonial times and the establishment of farmsteads and settlements. More recent changes involve the large-scale conversion of forest and farmland to exotic tree plantations and these are the main threat to the conservation of native and endemic tree species of south- central Chile (Armesto, 1998; Hechenleitner et al., 2005; Myers et al., 2000). Rural areas with small-scale farming that remain in south-central Chile and South America represent a cultural landscape comprised of a mosaic of pastures, shrublands and forests (Echeverría et al., 2007). The impacts of tree harvesting and livestock grazing affect the floristic composition and vegetation dynamics of the vegetation mosaic (Seis et al. 2014). Twenty percent of the country’s energy supply comes from fuelwood of which 60% comes from native species (INFOR, 2009), largely sold as firewood and charcoal by smallholders (Burschel et al. 2003). For the smallholders the ecological and economic value of the cultural landscape in the VCR depends strongly on the resilience of the population of native tree species and the potential for natural forest restoration.

The majority of studies which address regeneration of tree species in the Valdivian evergreen rainforests have been conducted in primary and managed forests. The regeneration in old growth forest canopy gaps is dominated by very shade tolerant tree species Aextoxicon punctatum and Laureliopsis philippiana (Donoso and Nyland, 2005). Selective cutting that creates small openings can favor the regeneration of late-successional species (Oyarzún et al., 2012) but it can also dramatically promote species that are able to reproduce vegetatively (e.g. Eucryphia cordifolia; (Donoso, 1989a). Recent studies focusing on the vegetation mosaic in rural

123 landscapes investigated tree regeneration in shrublands, especially those in Chiles’ Chiloé Island (e.g. Bustamante-Sánchez and Armesto, 2012; Bustamante-Sánchez et al., 2011; Díaz and Armesto, 2007). In successional Baccharis-Shrublands little or no regeneration was reported (Díaz and Armesto, 2007). Meanwhile the recolonization of open areas was often slow and highly variable (Aravena et al., 2002; Bustamante-Sánchez et al., 2011). One reason for this was found to be low amounts of seed rain although there were nearby seed sources in forest fragments (Bustamante-Sánchez and Armesto, 2012). The species-specific requirement for regeneration of tree species in the evergreen rainforest cannot be explained by a univariate relationship such as shade tolerance (Díaz and Armesto, 2007; Lusk and Del Pozo, 2002). To improve forest management with regards to the sustainability of wood production, but also for habitat conservation it is necessary to study the individual species. A first step could be to classify plant functional types, these summarize groups of plants with similar multiple biological traits. Plant functional types have become a powerful tool to describe the plants functional response to environmental changes driven by land use (Kleyer et al., 2012; Wellstein et al., 2011). For example, to track degradation and recovery processes and for predicting future changes in vegetation characteristics, ecosystem properties, and ecosystem services (Díaz et al., 2007). This study focuses on plant functional types defined by functional characteristics (traits) that potentially influence tree species regeneration and seedlings resilience to land use disturbances. These functional types will hereafter be called regeneration types (RTs).

We selected two study regions (‘Caman’ and ‘Lomas del sol’) that are similar in abiotic site conditions and representative for cultural landscapes in the mid-elevations of the eastern slope of the VCR. By combining a trait and a species based approach, this study aimed to assess the potential of cultural landscapes in the VCR for the natural regeneration of native tree species by:

• classifying tree species that regenerate within the vegetation mosaic into regeneration types (RTs) based on functional traits,

• testing the response of those a priori defined RTs to measured indicators of land use impacts for tree harvesting, livestock grazing and coppice forestry,

124

• comparing the density of species specific regeneration in grassland, shrubland and forest habitat.

This study is the first attempt to explain the patterns of natural tree species regeneration within the vegetation mosaic of the cultural landscape in the VCR. A ‘real scenario’ approach was used considering the influences of timber harvesting (selective cutting and coppice forestry) and livestock grazing.

3. Methods

3.1. Study area

The study was situated on the eastern slopes of the VCR in the vicinity of two rural communities: Camán (39° 58'S 73° 00'W) and Lomas del Sol (39° 50'S 73° 07'W) adjacent to the the Llancahue watershed. All study sites were located between 250 and 500 m.a.s.l.. Climate and soil conditions in both study sites were similar. The climate is classified as temperate and humid. The Llancahue watershed has an average annual precipitation of 2,357 mm, where, on average, July is the rainiest and February the driest month (Núñez et al., 2006). The weather station, at 25 m.a.s.l. in Valdivia, has a mean annual temperature of 12.0 °C, in the warmest month, January (17.0 °C), and in the coldest month, July (7.6 °C). The soils are well drained Ultisols of volcanic origin above metamorphic bedrock. The soil texture varies from silty to sandy loam (CIREN, 2001). The native vegetation in these elevations is found in the threatened Valdivian temperate evergreen rainforest. Around 1900 European settlement of the area began accompanied by the massive destruction and exploitation of native forests. Settlers used the cleared land mainly for agricultural purposes. During the 1970s state owned forests in Camán began to be converted to exotic tree plantations on a large-scale and the conversion process continues on private farmland today. The amount of land in the region classified as plantation forests increased by 55% between 1998 and 2008 (116–179 thousand ha; CONAF-CONAMA (2008) ) and the only native forest surviving in Camán is on private property. Even though the

125 area covered by exotic tree plantations also expanded in Lomas del Sol, the adjacent forest, part of the Llancahue watershed, which remained state owned, was not converted. This remaining native forest is a dense 70–100 year-old secondary forest (Donoso et al., 2013). In the 1990s neighboring farmers started to graze their livestock and harvest timber in this state-owned native forest (Moorman et al., 2013). Residents in the area produce charcoal and firewood as their primary source of income (Donoso et al., 2013). Reforestation is done with non-native tree species.

3.2. Field sampling

A stratified random sampling design with three predefined strata (formations) was used. The stratification was based on vegetation height and structure; 1) open pastures (vegetation height < 1 m, scarce presence of woody species); 2) shrublands (height 1 m – 5 m, with greater than 20% shrub cover), and 3) forests (> 5 m tall). Plots were modified nested Whittacker plots (Stohlgren et al., 1995) with a 20 m² plot used for the 0-1 m high herb layer nested within a 80 m² plot for the >1 m – 5 m high shrub layer which was again centered in a 160 m² plot for the tree layer (> 5 m). Extreme sites like swamps and excessive slopes were excluded from sampling. Sampling points were randomly distributed within the stands of each stratum; the number of plots was proportional to the area of each stratum in both study regions.

In 85 plots the 20 m² plot was further divided into three regeneration-subplots (6.7 m² each). In these regeneration-subplots the seedlings and saplings were assessed using two different approaches. In the first approach the frequency data (Ellenberg and Mueller-Dombois, 1974) was collected by recording the presence of a seedling and saplings of each tree species in two size classes < 50 cm in height and < 5 cm dbh in each regeneration-subplot. All seedlings past the cotyledon stage were considered. For the second approach - the density assessment was done - in 30 randomly chosen plots (10 pastures, 10 shrublands, 10 forests). One of the three regeneration-subplots (6.7 m² each) per plot was chosen randomly. All seedlings and saplings were counted.

126

The cover of each vegetation layer and crown closure was visually estimated in %. The diameter at breast height (dbh) of each tree (dbh > 5 cm) was measured and basal area (m²/ha) and stem density (number of stems/ha) were calculated. Impacts from livestock were assessed by a) counting the dung piles in the 20 m² subplot, b) measuring the surface of animal tracks (m²) as an indicator for trampling and c) assessment of the browsing damage on woody species and calculating the mean browsing index per plot (Veblen et al., 1989). Impacts that resulted from tree harvesting were assessed by estimating the relative stump age with the knife test and counting all stumps in three decay classes (dead_wood_3 = stump highly decayed, dead_wood_2 = stump beginning to decayed, dead_wood_1 = fresh stump; Rouvinen et al., 2002). Coppice was recorded by counting all trees that were obviously resprouting from stumps. Environmental co-variables were assessed for altitude, slope and aspect. Light was estimated with a hemispherical photo, taken at one representative site close to each plot-center. A Nikon Coolpix 990 digital camera (Nikon Corporation, Tokyo, Japan) fitted with a Nikon FC-E8® fisheye converter (Nikon Corporation, Tokyo, Japan) was used to take photos at one meter which was the maximal height of the herb layer. Images were analyzed with WinScanopy (Regent Instruments Inc, 2008) to calculate total percentage of above canopy light (PACL). In grasslands where there was no overstory cover and full light conditions prevailed, the PACL was set at 100%.

3.3. Data Analysis

To describe the functional response of the regenerating tree species to the land use driven environmental changes, regeneration types (RTs) were classified based on individual tree species regeneration traits. Species that were recorded less than 10 times were not included in the further analysis. A database of 7 nominal traits that are likely to influence the natural regeneration of the species and the resilience of seedlings and saplings to land use activities, was compiled from the literature (Table 1). The RTs were classified using hierarchical clustering. Gower coefficient was chosen as an adequate measure to calculate the distance matrix from the

127 nominal variables. The structure of functional types was examined using a flexible beta agglomerative algorithm (Lance and Williams, 1967) with ß= - 0.25 performed by the package cluster (v. 1.14.4; Maechler et al. (2013)) in the R software environment for statistical computing and graphics (R Core Team, 2013). A cut level of 5 was chosen to classify the groups.

Classification tree analysis was applied to explore which explanatorily variables were influencing the frequency of each of the regeneration types, which were the categorical response variables. Before the classification tree analysis the correlation of potential exploratory variables was analyzed using Spearman rank coefficient for non-parametric data (Dormann et al., 2012). If pairwise correlations exceed a threshold of |r| > 0.5 variables one of the variables was excluded from further analysis.

To detect the direct effect of land use indicators for livestock grazing and cutting on the frequency of the RTs and the indirect effects through altered physical vegetation structure, the direct and indirect impacts were analyzed separately. We used recursive partitioning within a conditional inference framework to explain variation in the response variables as a function of the explanatory variables. This was accomplished with the non-parametric conditional inferences tree methods R package party (Hothorn et al., 2006b). At each step of the analysis, one explanatory variable was selected from all the available variables, based on the best separation of two groups using a permutation test; this point is determined by a numerical value (threshold) of the explanatory variable (Hothorn and Zeileis, 2008; Hothorn et al., 2006a, 2006b). The relationships between the response variable and explanatory variables were presented in a dichotomous tree diagram with nodes that represent split points, branches that connect nodes, and leaves or terminal nodes that represent the final proportion of frequencies.

As a last analysis step the seedling and sapling density per ha were calculated for the five most abundant species: Aextoxicon punctatum, Amomyrtus luma, Eucryphia cordifolia, Lomatia ferruginea, and Myrceugenia planipes. Densities per ha were compared among grasslands, forest and shrublands. Significance was tested with Man-Whitney-U Test.

128

Table 1 Traits used in for the classification of regeneration types as functional groups. Literature: 1) Armesto and Rozzi (1989), 2) Bustamante and Castor (1998), 3) Donoso (1989), 4) Donoso et al. (2006). 5) Escobar et al. (2006), 6) Figueroa and Lusk (2001) 7) Gutiérrez et al. 2007), 8) (Hechenleitner et al., 2005), 9) Hoffmann (1982), 10) Hoffmann (1998), 11) Lusk and Del Pozo (2002), 12) Muñoz and Gonzálz (2009), 13) Rodríguez et al. (1983) 14) Smith- Ramírez (1993), 15) Smith-Ramírez and Armesto (1994), 16) Smith-Ramírez et al. (2005), 17) Veblen et al. (2003), * Information is based on field obervations of various experts. For Maximum lifespan trees were classified in short (< 100 years), mediate (100 – 300 years) and and long living (>300 years). Resprouting Maximum Species Family Symbol Spinescence Main dispersal type Dicliny Pollination Shade tolerance capacity lifespan non- Aextoxicon punctatum Aextoxicaceae Ap spiny * sprouting 7 endozoochory 1;15 dioecy 11;23 biotic 2 tolerant 6 long 11 non- Amomyrtus luma Myrtaceae Al spiny * sprouting 3 endozoochory 1;15 monoecy 32 biotic 13;32 ;33 tolerant 6 mediate 11 non- Amomyrtus meli Myrtaceae Am spiny * sprouting 4 endozoochory 1;15 monoecy 11;32 biotic 32;33 tolerant 3 long * non- Aristotelia chilensis Elaeocarpaceae Ac spiny * sprouting 12 endozoochory 1;15 dioecy 12;13 biotic 15 intolerant * short 11 Dasyphyllum diacanthoides Asteraceae Dd spiny 9 sprouting * anemochory 1;9;15 monoecy * biotic 15 tolerant 3 mediate 11 non- Drimys winteri Winteraceae Dw spiny * non-sprouting * endozoochory 1;15 monoecy 13 biotic 1;15 semi-shade 3 mediate 11 non- Embothrium coccineum Proteaceae Ec spiny * sprouting 17 anemochory 1;15 monoecy 13;16 biotic 15;16 intolerant 3 short 5 non- Eucryphia cordifolia Cunoniaceae Euc spiny * sprouting 3 anemochory 1;15 monoecy 13 biotic 15;17 semi-shade 3 long 11 non- Gevuina avellana Proteaceae Ga spiny * sprouting 3 endozoochory 1;15 monoecy 13 biotic 15;18 semi-shade 3 short * non- Laurelia sempervirens Atherospermataceae Ls spiny * sprouting * anemochory 1 monoecy 13 biotic 2 tolerant 3 long * non- Laureliopsis philippiana Atherospermataceae Lp spiny * sprouting 3 anemochory 1;15 monoecy 13 biotic 15 tolerant 6 long 11 non- Lomatia ferruginea Proteaceae Lf spiny * sprouting * anemochory 1;15 monoecy 13 biotic 15 tolerant 3 short * non- Lomatia hirsuta Proteaceae Lh spiny * sprouting * anemochory 1;9;15 monoecy 13 biotic * intolerant 3 short * non- Luma apiculata Myrtaceae La spiny * sprouting 17 endozoochory 1;15 monoecy 16 biotic 1;15 tolerant 3 mediate * non- Myrceugenia planipes Myrtaceae Mp spiny * sprouting 3 endozoochory 1;15 monoecy 13;16 biotic 15;16 tolerant 6 mediate 11 non- Persea lingue Lauraceae Pl spiny * sprouting * endozoochory * monoecy 13 biotic 2 tolerant 8 long * non- Podocarpus salignus Podocarpaceae Ps spiny * non-sprouting * endozoochor 9 dioecy 13 abiotic 2 semi-shade 3 long *

Rhaphithamnus spinosus Verbenaceae Rs spiny 9;10;13 sprouting * endozoochory 1;15 monoecy 13 biotic 14;15 semi-shade 12 short *

129 4. Results

In total 23 regenerating tree species were found. Four species (Caldcluvia paniculata, Pseudopanax laetevirens, Saxegothaea conspicua, and Weinmannia trichosperma) occurred less than 10 times and were therefore not considered for further analysis.

4.1. Regeneration types

Based on the regeneration traits five regeneration types (RTs) were classified (Fig. 1): (1) long lived, shade-tolerant endozoochorous broadleaved trees (long-tol-endo), (2) short lived, shade- intolerant, anemochorous broadleaved trees (short-int-ane), (3) long lived, shade-tolerant, anemochorous broadleaved trees (long-tol-ane), (4) short lived, semi shade-tolerant, endozoochorous broadleaved trees (short-semi-endo), and (5) long lived, semi shade-tolerant, endozoochorous conifer (long-semi-endo). The long-semi-endo type was represented by only one species Podocarpus salignus (Ps).

130

Fig. 1. Regeneration types classified as functional groups based on their regeneration traits. The dendrogram derived from flexible beta agglomerative algorithm. Dissimilarity was calculated from 7 nominal variables using Gower index (Ap = Aextoxicon punctatum, Am = Amomyrtus meli, Pl = Persea lingue, Al = Amomyrtus luma, La = Luma apiculata, Mp = Myrceugenia planipes, Dd = Dasyphyllum diacanthoides, Euc = Eucryphia cordifolia, Ls = Laurelia sempervirens, Lp = Laureliopsis philippiana, Lf = Lomatia ferruginea, Ac = Aristotelia chilensis, Ec = Embothrium coccineum, Lh = Lomatia hirsuta, Dw = Drimys winteri, Ga = Gevuina avellana, Rs = Rhaphithamnus spinosus, Ps = Podocarpus salignus). The diagram was cut into five groups representing the following regeneration types: (1) long lived, shade- tolerant endozoochorous broadleaved trees (long-tol-endo) (2) short lived, shade-intolerant, anemochorous broadleaved trees (short-int-ane), (3) long lived, shade-tolerant, anemochorous broadleaved trees (long-tol-ane) (4) short lived, semi shade-tolerant, endozoochorous broadleaved trees (short-semi-endo) (5) long lived, semi shade-tolerant, endozoochorous conifer (long-semi-endo).

131 4.2. Response of the regeneration types to land use impacts

The classification tree analyses showed significant responses of the five RTs on the direct and indirect land use impacts (Fig. 2 - 6). Most severe land use impacts explained the frequency of species in all three groups were the damages by large herbivores. Relative frequencies of the long-tol-endo (Fig. 2), long-tol-ane (Fig. 4), and short-semi-endo (Fig. 5) were negatively influenced by large herbivores. The long-tol-ane and the short-semi-endo were partly restricted by recent harvesting impacts (dead wood 1 and dead wood 2).

The physical vegetation structure, as a result of land use impacts, showed a more heterogeneous response among the groups than the direct land use indicators. Low cover of the herb layer was most significant for the long-tol-endo type (Fig. 2). If the herb layer cover was >25% then basal area > 1.745 m² explained most of the regeneration of this RT. For the short- int-ane type less than 38% of the total canopy light in combination with a herb layer cover higher than 45% were the best explaining variables (Fig. 3). The shrub layer cover explained clearly the regeneration of the long-tol-ane type (Fig. 4). For the short-semi-endo type and the long-semi-endo type basal area was the most important variable (Fig. 5 and 6). For two types (short-int-ane and long-semi-endo) no direct influence of the land use activities could be detected. The short-int-ane type regenerated only in shrublands (Fig. 4), long-semi-endo type was restricted to forest habitat (Fig. 6).

132

Fig. 2. Classification trees for the relative frequencies of the long lived, shade-tolerant endozoochorous broadleaved trees (long-tol-endo) based on the conditional inference tree model. Direct impacts of livestock and cutting impacts are tested in the classification trees (a) and the indirect impacts of land use through vegetation structure in the classification trees (b). The encircled explanatory variables are those showing the strongest association to the response variable. Values on lines connecting explanatory variables indicate splitting thresholds. Numbers in boxes above the explanatory variable indicate the node number. The p-values listed at each node represent the test of independence between the listed independent variable and the response variable. Box plots show the relative frequencies of tree individuals belonging to the group within that node.

133

Fig. 3. Classification trees for the relative frequencies of short lived, shade-intolerant, anemochorous broadleaved trees (short-int-ane) based on the conditional inference tree model. Direct impacts of livestock and cutting impacts are tested in the classification trees (a) and the indirect impacts of land use through vegetation structure in the classification trees (b). The encircled explanatory variables are those showing the strongest association to the response variable. Values on lines connecting explanatory variables indicate splitting thresholds. Numbers in boxes above the explanatory variable indicate the node number. The p-values listed at each node represent the test of independence between the listed independent variable and the response variable. Box plots show the relative frequencies of tree individuals belonging to the group within that node.

134

Fig. 4. Classification trees for the relative frequencies of long lived, shade-tolerant, anemochorous broadleaved trees (long-tol-ane) based on the conditional inference tree model. Direct impacts of livestock and cutting impacts are tested in the classification trees (a) and the indirect impacts of land use through vegetation structure in the classification trees (b). The encircled explanatory variables are those showing the strongest association to the response variable. Values on lines connecting explanatory variables indicate splitting thresholds. Numbers in boxes above the explanatory variable indicate the node number. The p-values listed at each node represent the test of independence between the listed independent variable and the response variable. Box plots show the relative frequencies of tree individuals belonging to the group within that node

135

Fig. 5. Classification trees for the relative frequencies of short lived, semi shade-tolerant, endozoochorous broadleaved trees (short-semi-endo) based on the conditional inference tree model. Direct impacts of livestock and cutting impacts are tested in the classification trees (a) and the indirect impacts of land use through vegetation structure in the classification trees (b). The encircled explanatory variables are those showing the strongest association to the response variable. Values on lines connecting explanatory variables indicate splitting thresholds. Numbers in boxes above the explanatory variable indicate the node number. The p-values listed at each node represent the test of independence between the listed independent variable and the response variable. Box plots show the relative frequencies of tree individuals belonging to the group within that node.

136

Fig. 6. Classification trees for the relative frequencies of long lived, semi shade-tolerant, endozoochorous conifer (long-semi-endo) based on the conditional inference tree model. Direct impacts of livestock and cutting impacts are tested in the classification trees (a) and the indirect impacts of land use through vegetation structure in the classification trees (b). The encircled explanatory variables are those showing the strongest association to the response variable. Values on lines connecting explanatory variables indicate splitting thresholds. Numbers in boxes above the explanatory variable indicate the node number. The p-values listed at each node represent the test of independence between the listed independent variable and the response variable. Box plots show the relative frequencies of tree individuals belonging to the group within that node.

137 4.3. Seedling density in different habitats

Most species differed in seedling density among forest, grassland and shrubland habitat. Eucryphia cordifolia showed no significant differences. For Aextoxicon punctatum, seedling densities were significantly higher in forests than in any other habitat. The other species showed strongest differences between forest and grassland.

Table 2. Seedling and sapling density per ha (mean, standard deviation in brackets; Ap= Aextoxicon punctatum, Al = Amomyrtus luma, Euc= Eucryphia cordifolia, Lf = Lomatia ferrunginea, Mp= Myrceugenia planipes). Species belong to the regeneration types of long lived, shade-tolerant endozoochorous broadleaved trees (long-tol-endo) and long lived, shade-tolerant, anemochorous broadleaved trees (long-tol-ane). Significant differences in species density between the different formations resulting from Mann-Whitney-U-Tests were indicated with different letters.

Long-tol-endo regeneration type Long-tol-ane regeneration type Ap Al Mp Euc Lf Mean

Forest 11493 (+/-13)a 6269 (+/-4)a 2687 (+/-2)a 746 (+/-1)a 11045 (+/-17)a 6448

Grassland - - b - - b - - b - - a 149 (+/-0.3)b 30

Shrubland 448 (+/-1)b 2239 (+/-4)ab 746 (+/-1)ab 1493 (+/-3)a 4179 (+/-5)ab 1821 Mean 3980 2836 1144 746 5124 2766

5. Discussion

5.1. Which regeneration types could be identified?

We presented a quantitative and objective classification of regeneration types (RTs) for the regenerating tree species in the vegetation mosaic of the cultural landscape, representative of the Valdivian Coastal Range. The regeneration types that resulted from this classification of regeneration relevant traits showed tendencies regarding successional status. However, we identified transitional groups rather than the classical classification in pioneer and climax species (e.g. Swaine and Whitmore, 1988). If species were ordered along a classical successional gradient in the sense of Clements, (1916), the type of short lived, shade-intolerant,

138

anemochorous broadleaved trees (short-int-ane) would be the potential pioneers after the creation of large forest gaps, or would be the first to regenerate after complete forest clearance or in grasslands. The short lived, semi shade-tolerant, endozoochorous broadleaved trees (short-semi-endo) may follow next in succession. Because the species of this type depend on some preexisting vegetation structures like trees or shrubs that provide perches for seed distributing birds (Bustamante-Sánchez and Armesto, 2012). Long lived, shade-tolerant, anemochorous broadleaved trees (long-tol-ane) have a trait related to pioneer species – anemochorous seeds. Anemochorous seeds enable a tree to disperse over long distances (Cornelissen et al., 2003) and to colonize a wide range of habitats including open areas. But the species in this group also have the shade-tolerance trait which enables them to also regenerate under a closed canopy. The long lived, semi shade-tolerant, endozoochorous conifer (long-semi- endo) type was comprised of only one species: Podocarpus salignus. The separation of this species into an exclusive group makes sense from the ecological perspective. Conifers differ in many traits from angiosperm tree species. The fact that the species is long lived but semi shade- tolerant made it fit into the concept of a long lived pioneer (Lusk, 1999). However endozoochorous dispersal and the fact that it is a dioecious species that depends on the existence of large, old seed trees (Hechenleitner et al., 2005) might place it into a later stage of succession. The long lived, shade-tolerant endozoochorous broadleaved trees (long-tol-endo) fit into the concept of a climax species (Swaine and Whitmore, 1988) characterized by long lifespan, crucial for species persistence in the temperate rainforest landscape (Lusk, 1999), and high shade tolerance.

5.2. How do the regeneration types respond to land use impacts?

Generally the regeneration types (RTs) showed a clear difference in frequency related to the direct and indirect impacts (through altered vegetation structure) of land use activities. This

139 showed the usefulness of the functional types to be able to generalize regenerating capacity of the tree species of the Valdivian temperate evergreen rainforest in the cultural landscape.

Overall, grazing impacts by large herbivores were the variable that most explained the regeneration. Most restrictive were the grazing impacts on long lived, shade-tolerant endozoochorous broadleaved trees (long-tol-endo). Although this RT was assumed to represent climax species, the regeneration of these species was not restricted to a late successional forest habitat. The fact that mature forest species were able to colonize early in succession has already been reported for the temperate rainforest in Chile (Donoso, 1989a) and has been shown as a common phenomenon in other ecosystems (Chazdon et al., 2010; Finegan, 1996; Franklin and Rey, 2007; Van Breugel et al., 2007). The short lived, semi shade-tolerant, endozoochorous broadleaved trees (short-semi-endo) and the long lived, shade-tolerant, anemochorous broadleaved trees (long-tol-ane) were shown to be partly able to regenerate in the presence of large herbivores and disturbance by cutting. For the long-tol-ane type this may be explained by the fact that one species belonging to this RT. Young Dasyphyllum diacantoides trees have hard, galling spines, making them unpalatable for livestock (Veblen et al., 2003). Furthermore, species belonging to the long-tol-ane type were all able to resprout vegetatively. Resprouting ability is another strategy that may contribute to species’ resilience to browsing and coppicing. Additionally resprouting enables short-distance migration when seed dispersal and – recruitment is low (Cornelissen et al., 2003). D. diacantoides was observed to be proliferating vegetatively in some plots. A high capacity for vegetative regeneration is also reported for Eucryphia cordifolia which resprouts aggressively after cutting (Donoso, 1989a). The dependency of regeneration from the short-semi-endo and long-tol-ane type on sites with a high shrub cover also implies that facilitation by shrubs plays an important role in protecting regenerating trees from browsing and trampling by livestock.

In the cultural landscape ecological traits like shade tolerance may become of secondary importance because disturbance impacts by livestock are the main drivers or obstructors of tree regeneration. This assumption is supported by the fact that that even trees that belonged to the short lived, shade-intolerant, anemochorous broadleaved trees (short-int-ane) group, meaning

140

typically pioneers and tree species that are known to be characteristic of open sites, forest edges, and old-fields (Donoso, 1989b, 1981) required a layer of shrub cover for regeneration. For the long lived, semi shade-tolerant, endozoochorous conifer (long-semi-endo) the role of livestock impacts could not be clarified with the classification tree analysis. This RT was shown to be completely dependent on forest to regenerate. This conifer, like many other species of the Valdivian evergreen flora, is adapted to survive under a tall, relatively continuous evergreen canopy whose shade provides a continuously cool and humid microclimate (Arroyo et al., 1997). Additionally, there could be a stronger dependence on the frequency of seed trees because Podocarpus salignus is a dioecious species. Because the number of mature trees has been reduced, P. salignus, has a vulnerable conservation status (Hechenleitner et al. 2005), in the Valdivian Coastal Range.

5.3. How does the density of species specific regeneration vary within the cultural landscape?

The regenerating species that occurred with the highest seedling density belonged to two regeneration types and both consist of late successional species: long lived, shade-tolerant endozoochorous broadleaved trees (long-tol-endo) and long lived, shade-tolerant, anemochorous broadleaved trees (long-tol-ane). The fact that the highest differences existed between grassland and forest habitats, the latter had the highest seedling densities, lends support to the belief that environmental conditions may restrict regeneration in grasslands. However, previous classification tree analysis showed that this micro-environmental hypothesis may not apply to a cultural landscape. Grasslands in the same study area were shown to be maintained by continuous grazing (Seis et al. 2014). This strong grazing effect prevented almost any kind of tree regeneration. The facilitation function of shrubland vegetation is shown by the higher regeneration rates in shrublands. One species, Eucryphia cordifolia showed a tendency to regenerate even better in shrublands than in forests. For many landowners this tree species is the most valuable for timber production (personal communication, landowner 2010). Presently,

141 shrublands in particular are most often cleared for pastureland or planted with exotic tree species. These practices may reduce not only the ecological but also the economic value for landowners.

6. General Conclusions

This study presents the first objective and quantitative characterization of tree regeneration in the cultural landscape in the VCR. The approach of classifying species by multiple regeneration traits into regeneration types can be a powerful tool to explain the response of tree species to land use impacts in the cultural landscape in the VCR. The study showed the high impact large herbivores have on tree regeneration. It could be because the VCR vegatation evolved in the absence of mammalian herbivores and therefore their affect is greater than in ecosystems with a long evolutionary history of mammalian herbivory (Milchunas et al., 1998). The only native browsing animal in the natural Valdivian evergreen rainforest is the small Pudu (Pudu puda). This small forest-dwelling mammal is a concentrate selector (Hanley, 1997) with limited ability to ferment and use cellulose. It browses on buds of woody plants (shrubs, trees) and also consumes fruits (Armesto et al., 1987). Because of its small size and low population density it cannot be expected to be significant for the evolution of temperate rainforest tree species.

The most natural regeneration was found in forests. In grasslands there was no regeneration of native tree species detected which implies that that succession has currently been stalled. Shrublands provided a facilitation function for the regeneration of many tree species and have a high potential to protect the regeneration of different functional types including mature forest species. The restoration of forest and native tree diversity through natural succession from shrubland to forest may be a cost effective means towards restoration presuming that regenerating trees are protected from livestock by fencing. This protection would also ensure that smallholders have a sustainable supply of ecosystem goods from forest trees. However, for forest dependent tree species, such as Podocarpus salignus, the complete protection of its

142

forest habitat and its seed trees wherever they are nearby, may be the only possibility for its conservation.

7. Acknowledgements

The Authors thank to Jan Bannister, Marco Florez, Alvaro Promis, Andreas Vogel, and Thomas Veblen for contributing their expert knowledge. Danisa Paredes and Melanie Welling supported the field work. We thank the Landesgraduiertenförderung Baden-Württemberg (Germany), the German Academic Exchange Service (Germany) and the Fondo de Fomento al Desarrollo Científico y Tecnológico (Chile) for their financial support. And we thank Bernhard Thiel for correcting the English.

143 8. References

Armesto, J.J., 1998. Conservation Targets in South American Temperate Forests. Science 282, 1271–1272.

Armesto, J.J., Rozzi, R., 1989. Seed dispersal syndromes in the rain forest of Chiloé: evidence for the importance of biotic dispersal in a temperate rain forest. J. Biogeogr. 16, 219–226.

Armesto, J.J., Rozzi, R., Miranda, P., Sabag, C., 1987. Plant/frugivore interactions in South American temperate forests. Rev. Chil. Hist. Nat. 60, 321–336.

Arroyo, M.T.K., Cavieres, L., Penaloza, A., Riveros, M., Faggi, A.M., Armesto, J., Villagrán, C., Arroyo, M.T.K., 1997. Phytogeographic relationships and species richness patterns in the temperate rain forest flora of South America. Ecol. Los Bosques Nativ. Chile Editor. Univ. Santiago Chile 71–100.

Burschel, H., Hernández, A., Lobos, M., 2003. Leña. Una fuente energética removable para Chile. Editorial Universitaria, Santiago de Chile.

Bustamante, R., Castor, C., 1998. The decline of an endangered temperate ecosystem: the ruil (Nothofagus alessandrii) forest in central Chile. Biodivers. Conserv. 7, 1607–1626.

Bustamante-Sánchez, M.A., Armesto, J.J., 2012. Seed limitation during early forest succession in a rural landscape on Chiloé Island, Chile: implications for temperate forest restoration. J. Appl. Ecol. 49, 1103–1112.

Bustamante-Sánchez, M.A., Armesto, J.J., Halpern, C.B., 2011. Biotic and abiotic controls on tree colonization in three early successional communities of Chiloé Island, Chile: Controls on tree colonization in shrublands. J. Ecol. 99, 288–299.

Chazdon, R.L., 2008. Beyond Deforestation: Restoring Forests and Ecosystem Services on Degraded Lands. Science 320, 1458–1460.

144

Chazdon, R.L., Finegan, B., Capers, R.S., Salgado-Negret, B., Casanoves, F., Boukili, V., Norden, N., 2010. Composition and Dynamics of Functional Groups of Trees During Tropical Forest Succession in Northeastern Costa Rica. Biotropica 42, 31–40.

CIREN, 2001. Descripciones de suelos. Materiales y simbolos, in: Estudio Agrologico X Región. Centro de Información de Recursos Naturales (CIREN), Santiago, Chile.

Clements, F.E., 1916. Plant succession; an analysis of the development of vegetation,. Carnegie Institution of Washington, Washington, USA.

CONAF-CONAMA, 2008. Catastro de uso del suelo y vegetación: Monitoreo y actualización Región de Los Ríos. Gobierno de Chile, Ministerio de Agricultura, Santiago de Chile.

Cornelissen, J.H.C., Lavorel, S., Garnier, E., Díaz, S., Buchmann, N., Gurvich, D.E., Reich, P.B., Steege, H. ter, Morgan, H.D., Heijden, M.G.A. van der, Pausas, J.G., Poorter, H., 2003. A handbook of protocols for standardised and easy measurement of plant functional traits worldwide. Aust. J. Bot. 51, 335.

Díaz, M.F., Armesto, J.J., 2007. Limitantes físicos y bióticos de la regeneración arbórea en matorrales sucesionales de la Isla Grande de Chiloé, Chile. Rev. Chil. Hist. Nat. 80, 13–26.

Díaz, S., Lavorel, S., McIntyr, S., Falczuk, V., Casanoves, F., Milchunas, D.G., Skarpe, C., Rusch, G., Sternberg, M., Noy-Meir, I., Landsberg, J., Zhang, W., Clark, H., Campbell, B.D., 2007. Plant trait responses to grazing ? a global synthesis. Glob. Change Biol. 13, 313–341.

Donoso, C., 1981. Tipos forestales de los bosques nativos de Chile. Corporación Nacional Forestal y Organización de las Naciones Unidas para la Agricultura y la Alimentación, Documento de trabajo No. 38. Investigación y Desarollo Forestal. Publicacion FAO, Santiago, Chile.

Donoso, C., 1989a. Regeneración y crecimiento en el tipo forestal siempreverde costero y andino tras distinos tratamientos silviculturales. Bosque 10, 69–83.

Donoso, C., 1989b. Antecedentes básicos para la silvicultura del tipo forestal siempreverde. Bosque Valdivia 10.

145 Donoso, C., Zúñiga, A., Utreras, F., 2006. Amomyrtus meli, in: Las Especies Arbóreas de Chile Y Argentina. Autecología. Editorial Marisa Cuneo, Valdivia, Chile, pp. 158–165.

Donoso, P.J., Frêne, C., Flores, M., Moorman, M.C., Oyarzún, C.E., Zavaleta, J.C., 2013. Balancing water supply and old-growth forest conservation in the lowlands of south-central Chile through adaptive co-management. Landsc. Ecol. 245–260.

Donoso, P.J., Nyland, R.D., 2005. Seedling density according to structure, dominance and understory cover in old-growth forest stands of the evergreen forest type in the coastal range of Chile. Rev Chil Hist Nat 78, 51–63.

Dormann, C.F., Elith, J., Bacher, S., Buchmann, C., Carl, G., Carré, G., Marquéz, J.R.G., Gruber, B., Lafourcade, B., Leitão, P.J., Münkemüller, T., McClean, C., Osborne, P.E., Reineking, B., Schröder, B., Skidmore, A.K., Zurell, D., Lautenbach, S., 2012. Collinearity: a review of methods to deal with it and a simulation study evaluating their performance. Ecography 36, 027–046.

Echeverría, C., Newton, A.C., Lara, A., Benayas, J.M.R., Coomes, D.A., 2007. Impacts of forest fragmentation on species composition and forest structure in the temperate landscape of southern Chile. Glob. Ecol. Biogeogr. 16, 426–439.

Ellenberg, H., Mueller-Dombois, D., 1974. Aims and methods of vegetation ecology. Wiley, New York.

Escobar, B., Donoso, C., Souto, C., Zúñiga, A., 2006. Embothrium coccineum, in: Las Especies Arbóreas de Chile Y Argentina. Autecología. Editorial Marisa Cuneo, Valdivia, Chile, pp. 233–243.

Figueroa, J.A., Lusk, C.H., 2001. Germination Requirements and Seedling Shade Tolerance Are Not Correlated in a Chilean Temperate Rain Forest. New Phytol. 152, 483–489.

Finegan, B., 1996. Pattern and process in neotropical secondary rain forests: the first 100 years of succession. Trends Ecol. Evol. 11, 119–124.

Franklin, J., Rey, S.J., 2007. Spatial patterns of tropical forest trees in Western Polynesia suggest recruitment limitations during secondary succession. J. Trop. Ecol. 23, 1–12.

146

Gutiérrez, A.G., Barbosa, O., Christie, D.A., Del-Val, E., Ewing, H.A., Jones, C.G., Marquet, P.A., Weathers, K.C., Armesto, J.J., 2007. Regeneration patterns and persistence of the fog- dependent Fray Jorge forest in semiarid Chile during the past two centuries. Glob. Change Biol. 14, 161–176.

Hanley, T.A., 1997. A Nutritional View of Understanding and Complexity in the Problem of Diet Selection by Deer (Cervidae). Oikos 79, 209–218.

Hechenleitner, P., Gardner, M., Philip, T., Echeverría, C., Escobar, B., Brownless, P., Martínez, C., 2005. Plantas amenazadas del centro-sur de Chile: distribución, conservación y propagación. Universidad Austral de Chile y Real Jardín Botánico de Edimburgo.

Hoffmann, A., 1998. Flora Silvestre de Chile, Zona Central. Edición 4. Fundación Claudio Gay, Santiago, Chile.

Hoffmann J., A.E., 1982. Flora silvestre de Chile, zona araucana, árboles, arbustos y enredaderas leñosas, 2nd ed. Ediciones Fundacion Claudio Gay, Santiago, Chile.

Hothorn, T., Hornik, K., van de Wiel, M.A., Zeileis, A., 2006a. A Lego System for Conditional Inference. Am. Stat. 60, 257–263.

Hothorn, T., Hornik, K., Zeileis, A., 2006b. Unbiased Recursive Partitioning: A Conditional Inference Framework. J. Comput. Graph. Stat. 15, 651–674.

Hothorn, T., Zeileis, A., 2008. Generalized maximally selected statistics. Biometrics 64, 1263– 1269.

Kleyer, M., Dray, S., Bello, F., Lepš, J., Pakeman, R.J., Strauss, B., Thuiller, W., Lavorel, S., 2012. Assessing species and community functional responses to environmental gradients: which multivariate methods? J. Veg. Sci. 23, 805–821.

Lance, G.N., Williams, W.T., 1967. A General Theory of Classificatory Sorting Strategies: 1. Hierarchical Systems.

Lusk, C., 1999. Long-lived light-demanding emergents in southern temperate forests: the case of Weinmannia trichosperma (Cunoniaceae) in Chile. Plant Ecol. 140, 111–115.

147 Lusk, C.H., Del Pozo, A., 2002. Survival and growth of seedlings of 12 Chilean rainforest trees in two light environments: Gas exchange and biomass distribution correlates. Austral Ecol. 27, 173–182.

Maechler, M., Rousseeuw, P., Struyf, A., Hubert, M., Hornik, K., 2013. Cluster: cluster analysis basics and extensions. R Package Version 1144.

Milchunas, D.G., Lauenroth, W.K., Burke, I.C., 1998. Livestock Grazing: Animal and Plant Biodiversity of Shortgrass Steppe and the Relationship to Ecosystem Function. Oikos 83, 65–74.

Moorman, M.C., Peterson, N., Moore, S.E., Donoso, P.J., 2013. Stakeholder Perspectives on Prospects for Co-Management of an Old-Growth Forest Watershed Near Valdivia, Chile. Soc. Nat. Resour. 1–15.

Myers, N., Mittermeier, R.A., Mittermeier, C.G., Da Fonseca, G.A., Kent, J., 2000. Biodiversity hotspots for conservation priorities. Nature 403, 853–858.

Núñez, D., Nahuelhual, L., Oyarzún, C., 2006. Forests and water: The value of native temperate forests in supplying water for human consumption. Ecol. Econ. 58, 606–616.

Olson, D.M., Dinerstein, E., 1998. The Global 200: A Representation Approach to Conserving the Earth’s Most Biologically Valuable Ecoregions. Conserv. Biol. 12, 502–515.

Oyarzún, C., Hervé-Fernandez, P., Frene, C., 2012. Effects of land use changes and management practices on water yield in native forests and exotic plantations in southern Chile, in: Forest Management: Technology, Practices and Impact. Nova Science Publishers, Hauppauge, pp. 79– 101.

R Core Team, 2013. R: A language and environment for statistical computing. R Found. Stat. Comput. Vienna Austria.

148

Regent Instruments Inc, 2008. WinSCANOPY-2008a, Canopy Structure and Solar Radiation Analysis. Regent Instruments Inc., Quebec, Canada.

Rouvinen, S., Kuuluvainen, T., Karjalainen, L., 2002. Coarse woody debris in old Pinus sylvestris dominated forests along a geographic and human impact gradient in boreal Fennoscandia. Can. J. For. Res. 32, 2184–2200.

Seis, K., Gärtner, S., Reif, A., Donoso, P.J., 2014. The effect of small-scale land use on vegetation in the Valdivian Coastal Range (Chile). Community Ecol. 15, 194-204.

Smith-Ramírez, C., 1993. Los picaflores y su recurso floral en el bosque templado de la isla de Chiloé, Chile. Rev. Chil. Hist. Nat. 66, 65–73.

Smith-Ramírez, C., Armesto, J.J., 1994. Flowering and Fruiting Patterns in the Temperate Rainforest of Chiloe, Chile-Ecologies and Climatic Constraints. J. Ecol. 82, 353–365.

Smith-Ramírez, C., Martinez, P., Nuñez, M., Gonzalez, C., Armesto, J.J., 2005. Diversity, flower visitation frequency and generalism of pollinators in temperate rain forests of Chiloe Island, Chile. Bot. J. Linn. Soc. 147, 399-416.

Stohlgren, T.J., Falkner, M.B., Schell, L.D., 1995. A Modified-Whittaker Nested Vegetation Sampling Method. Vegetatio 117, 113–121.

Swaine, M.D., Whitmore, T.C., 1988. On the definition of ecological species groups in tropical rain forests. Vegetatio 75, 81–86.

Van Breugel, M., Bongers, F., Martínez‐Ramos, M., 2007. Species dynamics during early secondary forest succession: recruitment, mortality and species turnover. Biotropica 39, 610– 619.

Veblen, T., Kitzberger, T., Raffaele, E., Lorenz, D., 2003. Fire History and Vegetation Changes in Northern Patagonia, Argentina, in: Veblen, T., Baker, W., Montenegro, G., Swetnam, T. (Eds.), Fire and Climatic Change in Temperate Ecosystems of the Western Americas, Ecological Studies. Springer New York, pp. 265–295.

149 Veblen, T.T., Mermoz, M., Martin, C., Ramilo, E., 1989. Effects of Exotic Deer on Forest Regeneration and Composition in Northern Patagonia. J. Appl. Ecol. 26, 711–724.

Veblen, T.T., Z., C.D., Schlegel, F.M., R., B.E., 1981. Forest Dynamics in South-Central Chile. J. Biogeogr. 8, 211–247.

Wellstein, C., Schröder, B., Reineking, B., Zimmermann, N.E., 2011. Understanding species and community response to environmental change – A functional trait perspective. Agric. Ecosyst. Environ. 145, 1–4.

150

151 Appendix A.

The table shows diagnostic species for each of the four vegetation types as well as for combinations of vegetation types. Indicator species were identified with multi pattern analysis (de Cáceres et al. 2010). The table contains relative frequencies of each diagnostic species in all types as well as the group-equalized Indicator Value as association index and a p-value derived from permutations (999). EGN = Extensively grazed non-native grasslands, SIE = severely impacted evergreen forests, SDE = sparsely disturbed evergreen forest, UBS = closed and semi closed grazed Ugni and Berberis shrublands.

152

Open and shrubby Agrostis pastures Impacted evergreen forests Vegetation type Family Origin EGN UBS SDE SIE stat p value n 41 20 33 8 Open and shrubby Agrostis pastures Agrostis capillaris Poaceae non-native 95 100 3 100 0.922 0.005 Anthoxanthum odoratum Poaceae non-native 90 70 0 0 0.914 0.005 Leptostigma arnottianum Rubiaceae native 88 90 3 38 0.873 0.005 Holcus lanatus Poaceae non-native 93 85 6 75 0.855 0.005 Leontodon saxatilis Compositae non-native 90 75 3 38 0.851 0.005 Viola reichei Violaceae native 83 80 18 75 0.845 0.005 Prunella vulgaris Lamiaceae non-native 100 95 9 88 0.821 0.005 Luzula campestris Juncaceae non-native 85 75 0 75 0.792 0.005 Plantago lanceolata Plantaginaceae non-native 71 65 3 50 0.768 0.01 Leptinella scariosa Compositae native 51 40 3 25 0.631 0.01 Cuscuta suaveolens Convolvulaceae native 39 30 0 0 0.601 0.005 EGN Trifolium repens Leguminosae non-native 71 20 0 25 0.742 0.005 Chevreulia sarmentosa Compositae native 24 5 0 0 0.479 0.025 UBS Ugni molinae Myrtaceae native 12 85 12 25 0.876 0.005 Embothrium coccineum Proteaceae native 5 60 0 0 0.764 0.005 Berberis microphylla Berberidaceae native 17 50 3 0 0.701 0.005 Genista monspessulana Leguminosae non-native 0 50 3 0 0.698 0.005 Vulpia bromoides Poaceae non-native 27 65 0 13 0.668 0.005 Baccharis racemosa Compositae native 7 50 0 0 0.661 0.005 Leucanthemum vulgare Compositae non-native 5 30 0 0 0.498 0.025 Intervened evergreen forests Lomatia ferruginea Proteaceae native 12 55 100 88 0.939 0.005 Chusquea quila Poaceae native 2 35 97 63 0.932 0.005 Luzuriaga radicans Alstroemeriaceae native 0 0 97 25 0.911 0.005 Myrceugenia planipes Myrtaceae native 0 10 85 75 0.902 0.005 Lapageria rosea Philesiaceae native 2 25 94 63 0.898 0.005 Gevuina avellana Proteaceae native 2 25 88 75 0.893 0.005 Laureliopsis philippiana Atherospermataceae native 2 10 88 50 0.89 0.005 Amomyrtus meli Myrtaceae native 5 25 82 63 0.851 0.005 Dasyphyllum diacanthoides Compositae native 5 20 67 100 0.85 0.005 Mitraria coccinea Gesneriaceae native 0 5 79 50 0.848 0.005 Rhaphithamnus spinosus Verbenaceae native 10 50 85 88 0.821 0.005 Amomyrtus luma Myrtaceae native 10 50 76 63 0.771 0.005 Boquila trifoliolata Lardizabalaceae native 5 50 79 88 0.752 0.005 Luzuriaga polyphylla Alstroemeriaceae native 2 0 64 25 0.745 0.005 Nertera granadensis Rubiaceae native 0 20 58 63 0.744 0.005 Podocarpus salignus Podocarpaceae native 2 20 61 38 0.714 0.005 Persea lingue Lauraceae native 0 0 42 38 0.644 0.005 Uncinia phleoides Cyperaceae native 0 0 33 50 0.605 0.005 Blechnum blechnoides Blechnaceae native 0 35 45 38 0.589 0.015 Saxegothaea conspicua Podocarpaceae native 0 0 18 38 0.469 0.025 Hydrangea serratifolia Hydrangeaceae native 2 5 24 13 0.443 0.035 SDE Aextoxicon punctatum Aextoxicaceae native 2 30 100 63 0.952 0.005 Drimys winteri Winteraceae native 0 10 64 0 0.793 0.005 Rhamnus diffusa Rhamnaceae native 0 0 45 0 0.674 0.005 Lophosoria quadripinnata Dicksoniaceae native 0 5 39 25 0.601 0.005 Fascicularia bicolor Bromeliaceae native 0 0 24 0 0.492 0.01 SDE & UBS Eucryphia cordifolia Cunoniaceae native 15 80 94 75 0.874 0.005 Lomatia dentata Proteaceae native 7 40 33 0 0.583 0.02 SIE Hypericum androsaemum Hypericaceae non-native 0 5 9 88 0.911 0.005 Blechnum hastatum Blechnaceae native 10 50 33 88 0.82 0.005 Aristotelia chilensis Elaeocarpaceae native 7 55 48 88 0.744 0.005 Azara lanceolata Salicaceae native 0 5 33 75 0.713 0.005 Ribes trilobum Grossulariaceae native 0 0 3 50 0.691 0.005 Greigia sphacelata Bromeliaceae native 0 0 9 25 0.484 0.025 SIE & EGN Acaena ovalifolia Rosaceae native 88 90 21 75 0.787 0.01 Hypochaeris radicata agg. Compositae non-native 85 85 6 50 0.757 0.005 Gnaphalium americanum Compositae native 49 30 0 50 0.646 0.01 Rumex acetosella Polygonaceae non-native 46 30 0 50 0.618 0.02 Hydrocotyle poeppigii Araliaceae native 51 10 15 25 0.612 0.01 Ranunculus repens Ranunculaceae non-native 41 5 0 25 0.602 0.01 SIE & UBS Gaultheria mucronata Ericaceae native 17 95 9 50 0.892 0.005 Rubus constrictus Rosaceae non-native 37 75 9 88 0.827 0.005 Lotus pedunculatus Leguminosae non-native 83 90 9 88 0.811 0.005 Carex fuscula Cyperaceae native 68 75 12 75 0.741 0.01 Galium hypocarpium Rubiaceae native 5 40 9 50 0.613 0.005 Lomatia hirsuta Proteaceae native 12 50 18 38 0.563 0.03 Berberis darwinii Berberidaceae native 10 25 6 38 0.511 0.025 Centaurium littorale Gentianaceae non-native 17 30 0 25 0.502 0.045 Baccharis concava Compositae native 2 25 0 25 0.496 0.03

153

154