Colonization History, Phylogeography and Conservation Genetics of the Gravely Endangered Tree Species Hagenia Abyssinica (Bruce) J.F

Taye Bekele Ayele OPTIMUS

Colonization History, Phylogeography and Conservation Genetics of the Gravely Endangered Tree Species Hagenia abyssinica (Bruce) J.F. Gmel from Ethiopia

Gottingen 2008 k c & - 223 M 1^.

GrsAlZ - UcVfc^Uo- C2V> £>»A T^t^-A U ' f \Je^C<2^ Colonization History, Pliylogeograpliy and Conservation Genetics of tllie Gravely Endangered Tree Species Httgenia abyssinica (Bruce) J.F. Ginel from Ethiopia

Dissertation

submitted for the degree of Doctor of Philosophy (PhD)

Department of Forest Genetics and Forest Tree Breeding Faculty of Forest Sciences and Forest Ecology Georg-August University of Gottingen

Taye Bekele Ayele Born in Kurkura (Harar), Ethiopia

Gottingen, 2008 Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de

Referee: Prof. Dr. Reiner Finkeldey Co-referee: Prof. Dr. Heiko Becker

Date of disputation: 2 September 2008

Printed with generous support from DAAD/gtz

Taye Bekele Ayele: Colonization History, Phylogeography and Conservation Genetics of the Gravely Endangered Tree Species Hagenia abyssinica (Bruce) J.F. Gmel from Ethiopia ISBN 978-3-941274-07-5

No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, scanning, or otherwise without the prior written permission of the Publisher. Request to the Publisher for permission should be addressed to [email protected]. To A tk ilt, Kal & Hamerenoah

with affection Acknowledgements

“How can I repay the LORD for all His goodness to me?” Psalm 116:12. I praise You Holy Father, for being there for me always and for getting me to the finish line.

It was a challenging but fruitful journey; and there were so many people around me that I should recognize and thank. All started with an inspiring reply to my e-mail of 2002 that I received from Prof. Dr. Reiner Finkeldey, expressing his interest to supervise my PhD project. Prof., 1 admire your patience and support until I step on the open-door of your office only three years latter. I have enjoyed a freedom of self management, excellent guidance and encouragement from you in the course of my study. Vielen Dank! 1 am grateful to Prof. Dr. Heiko Becker for willing to be co-referee for my dissertation and disputation and Prof Dr. Ursula Kiies for willing to be member of the examination team.

I am indebted to Dr. Oliver Gailing, for excellent guidance in molecular laboratory work and constructive suggestions throughout the data analysis and writing-up. Your “super” encouragement and pleasant disposition made my work much easier than I expected. A special gratitude goes to Prof. Dr. Hans H. Hattemer for scientific and administrative support all the way through. Many thanks to Dr. Barbara Vomam for the help in aligning the sequence data and for proof-reading the summaries of the thesis. I am grateful to Oleksandra Dolyniska, Olga Artes, Thomas Seliger and Gerold Dinkel who are the champions in the molecular lab and always ready to help. Also, Thomas and Gerold, thanks for keeping my Laptop running. I appreciate the interactive and friendly environ ment in the entire department with special mention to Prof. Dr. Martin Ziehe, Prof. Dr. Hans-Rolf Gregorius, Dr. Elizabeth Gillet, Dr. Ludger Leinemann and Mr. August Ca- pelle. 1 am grateful to Marita Schwahn for administrative support and for comforting me during some difficult times. Many thanks to former PhD students: Drs HT Luu, C-P Cao, AL Curtu, M Mottura, M Pandey, Abayneh D and VM Stefenon, and the current fellow PhD students: Sylvia, Akindele, Nicolas, Hani, Yanti, Nga, Amaryllis, Marius, Lesya and Dorte for the stimulating and useful discussions and memorable time we had. I ex press my gratitude to the former and present coordinators of the “PhD Programme-Wood Biology and Technology”, Drs E Kuersten & G Buettner, for their commendable work, and the fellow PhD students thereof for useful interactions. I thank Klaus Richter for translation of the summary of the thesis into German and Assefa Guchi for the production of the distribution map of the populations of Hagenia. I commend the encouragement and support I received from Drs Girma Balcha, Kassahun Embaye, Demel Teketay and Sileshi Nemomissa. I thank the former and present Ethiopian students of Georg-August University, Goettingen, for the wonderful moments we shared.

The enduring love and care of my wife S/r. Atkilt Gizaw and my sweetie daughters Kal and Hamerenoah has been a mystery of my strength that kept me moving forward. Ti- naye, you valiantly shouldered the responsibility of caring for our kids and managing the multifaceted social challenges during my long absence from home. Kaliye and Bebitaye, you are brave and I am proud of you. My special thanks to my mother Mintwab Wol- deAregay and my father-in-law Aba WoldeAmanuel for their love and blessing, my brother Ketema Bekele and his family for their encouragement and prayer through out my study, and to my brother Daniel Bekele for his great help and charming accompany during the fieldwork. The moral support from all my relatives and friends is gratefully appreciated. The prayers of my spiritual father Aba Gebretsadik, and that of brothers and sisters from Mahibere Selam MedhaneAlem and Mahibere Kidusan kept me energetic. The congregation of the Ethiopian Orthodox Tewahido Church in Germany particularly the brothers and sisters at the Keraniyo MedhaneAlem Sunday School in Kassel kept me spiritually warm. There are a number of wonderful people whom I want to recognize their thoughtfulness and contribution but the space just isn’t enough. May God bless you all!

Finally, I would like to acknowledge some institutions key to my achievement: the Ethio pian Institute of Biodiversity Conservation (IBC) granted me the study leave. My project was generously funded by the German Federal Ministry of Economic Cooperation and Development (BMZ) through the German Technical Cooperation (gtz). The German Academic Exchange Service (DAAD) executed the grant. The National Meteorological Service Agency of Ethiopia provided climatic data free of charge. Table of contents

1. General introduction...... 1 1.1 Ethiopia in brief...... 1 1.2 Conservation genetics of tropical tree species...... 1 1.3 Taxonomy and reproductive biology of Hagenia abyssinica...... 3 1.4 Ecology and natural distribution of Hagenia abyssinica...... 5 1.5 Economic and ecological significance of Hagenia abyssinica...... 6 1.6 Rationale...... 7 1.7 Aims and predictions...... 8 1.7.1 Objectives...... 8 1.7.2 Hypotheses...... 9 1.7.3 Major research questions...... 10 2. Research approaches...... 11 2.1 Sampling...... 11 2.2 Morphological assessment...... 11 2.3 DNA isolation...... 11 2.4 Chloroplast microsatellites...... 12 2.5 DNA Sequencing...... 12 2.6 AFLP analyses...... 12 3 Summary of results...... 13 3.1 Morphological data...... 13 3.2 Chloroplast microsatellite data...... 13 3.3 Sequence data...... 14 3.4 AFLP data...... 14 4 General discussion...... 15 5 Conclusions and outlook...... 19 6 Summary...... 21 7. Zusammenfassung...... 24 8 References...... 28 9. Papers submitted to journals...... 33 I. Colonization history and phylogeography of Hagenia abyssinica (Bruce) J.F. Gmel in Ethiopia inferred from chloroplast microsatellite m arkers...... 33 II. Spatial distribution of genetic diversity in Hagenia abyssinica (Bruce) J.F. Gmel from Ethiopia, assessed by AFLP molecular markers...... 57 III. Conservation genetics of African redwood (Hagenia abyssinica (Bruce) J.F. Gmel): a remarkable but gravely endangered tropical tree species...... 86 10 Appendices...... I l l

1. General introduction

1.1 Ethiopia in brief

Within an altitudinal range of 126 meters below sea level at Afar Depression to 4,620 meters above sea level (m asl ) at the spectacular mountaintops of Ras Dejen, Ethiopia’s varying physiographic features endowed the country with diverse fauna and flora. The climate of Ethiopia is varying from cool to hot and fundamentally governed by the Inter- tropical Convergence Zone (ITC). The rainfall pattern is influenced by two wind systems: monsoon from south Atlantic and the Indian Ocean, and winds from the Arabian Sea. The country is devided in 21 major Tree Seed Zones and 27 sub-Tree Seed Zones that were delineated based on ecological criteria to facilitate seed transfer within the country (Aalbask 1993). The vegetation of the Ethiopian mountains belongs to the Afromontane phytogeographical region (White 1983). Ethiopia is a severely deforested country with only about 3.5% of its land currently covered by closed forests (WBISPP 2004). The low living standard of the people coupled with lack of options is the underlying factor causing severe decline in forest cover. There has been increasing pressure on the forest land for crop and animal husbandry, and wood for fuel and construction. New settlements in pri mary forests are becoming commonplace and hence resulted in the conversion of forest land into agricultural and other land use systems, subsequently causing forest fragmenta tion. Precious tree species such as Hagenia abyssinica are the prime victims of such mal practices.

1.2 Conservation genetics of tropical tree species

Deforestation, forest fragmentation and extraction of timber in the form of selective log ging could have serious consequences on the long-term maintenance of genetic diversity and fitness in plants (Finkeldey and Hattemer 2007; Laikre and Ryman 1996; Young et al. 1996). The marvelous biodiversity that has captured our planet is being lost at a pace that is nearly unprecedented in the history of life (Ehrlich and Ehrlich 1991). Biodiversity is in a serious decline, with, for example, approximately 50% of the vertebrate animal

l General introduction species and 12% of all plant species now considered vulnerable to near-term extinction, mostly as a result of effects of habitat alteration associated with human population growth (Franklin et al. 2002).

The analyses of the amount and distribution of genetic variation within and among popu lations of a species can increase our understanding of the historical processes underlying the genetic diversity (Dumolin-Lapegue et al. 1997). The maintenance of natural tree populations with sufficient genetic variation to adapt to future changes in the environ ment is essential. Genetic variation is thought to be positively correlated with popula tions' ability to adapt to short-term environmental change, and populations with the high est levels of genetic variation are expected to suffer least from the negative effects of inbreeding depression or genetic drift (reviewed by Barrett & Kohn 1991, Ellstrand & Elam 1993). Examples of natural and dynamic evolutionary processes that shape genetic di versity are mutation, genetic drift, gene flow, natural selection, speciation and hybridiza tion (Avise 2004). Sound knowledge of the biology and genetics of a given organism is therefore instrumental in providing a scientific basis to its conservation and management. The two major goals of conservation biology are (1) the preservation of genetic diversity at any and all possible levels in the phylogenetic hierarchy and (2) the promotion of the continuance of ecological and evolutionary processes that foster and sustain biodiversity (reviewed by Avise 2004). Conservation genetics is a discipline dealing with the charac terization of a given taxon and the development of conservation measures to maintain its variation in order to adapt to changing environmental conditions. The present study in vestigates the pattern of genetic variation in Hagenia abyssinica at morphological and molecular genetic markers in order to identify populations for conservation and domesti cation.

2 General introduction

1.3 Taxonomy and reproductive biology of Hagenia abyssinica

The monotypic Hagenia abyssinica, fonnerly/synonimously known as

Banksia abyssinica Bruce, Brayera abyssinica Moq.-Tand, Brayera an- thelmintica Kunth and Hagenia an- thelmintica Kunth, is a wind- pollinated (anemogamous) and wind- dispersed (anemochorous) broad leaved dioecious tree species belong ing to the Rosaceae family (Hede- berg 1989; Legesse 1995). It is closely related to the monospecific genus Leucosidea from the same family in its taxonomic position (Eriksson et al. 2003). Locally, the tree is known as Kosso, Heto and Fig. I Excellent quality timber tree growing in Che- Habbi in Amharic, Oromiffa and Ti- cheba (Uraga) forest. Photo: Taye B. Ayele grigna, respectively (major local lan guages in Ethiopia). It is also commonly known as African redwood, Brayera, Cusso, Hagenia, Kousso, and Rosewood in English; Mdobore and Mlozilozi in Swahili (http:/www.worldagroforestry.org), and Ko- sobaum in German (http://de.wikipedia.org/wiki/- Hagenia ab-yssinica). The specific name abyssinica Fig. 2 compound leaf of Hagenia refers to the former name of Ethiopia. Photo: Taye B. Ayele

Hagenia grows up to 35 meters in height (Fig.I). Hagenia trees exhibit varying architec tures from croaked to slender, multi-stems or forked to single stem, and thick to thin crowns. The bark is brownish and readily peels in strips, sometimes very thick in old

3 General introduction stems. Branchlets are covered by silky brown hairs and ringed with leaf scars (Azene et al. 1993). The leaves are compound measuring up to 40 cm in length with 7- 19 narrowly oblong leaflets (Fig. 2), having inconsistent leaflet arrangement in opposite, alternate or mixed patterns. Hagenia has distinct male and female trees that are easily recognized by the appearance and color of the flowers (Figs. 3 & 4). The flowers of the female tree are small and inconspicuous, forming attractive bright pinkish-red drooping panicles (inflorescence) of up to 60 cm length and 30 cm width on aggregate (Azene et al. 1993; Legesse 1995). The female flower heads are

Fig 3 Typical mature male inflo bulkier than the more feathery yellowish male heads. rescence. Photo: Taye B. Ayele Flowering takes place between October and March (Legesse 1995). The attractive and appealing appear ance of the flowers of Hagenia is not typical for wind- pollinated species, which are usually dull in colour (Legesse 1995), suggesting that other pollinating vec tors such as insects (particularly bees) or birds might be involved. Fichtl and Admasu (1994) reported that honeybees collect pollen from the male flowers and Fig 4 Typical mature female inflorescence. Photo: Taye B. nectar from the female flowers.

H. abyssinica has small, hairy and single-seeded fruits, which have a brown syncarp with a single ovoid carpel and a fragile pericarp (Fig. 5). It has fairly small and light seeds (Fig. 5), amounting to 400,000 - 500,000 seeds per kg (Azene et al. 1993). The seeds can germi nate within 21 days with a germination capacity of 40 - 60% without any pre-germination treatment (Azene et al. 1993; Girma 1999). The seeds withstand desiccation Fig 5 seeds (top) and fruits (bottom) o f Hagenia, ruler graduation is in and hence can be stored for a long time (Girma 1999) in cm/mm. Photo: Taye B Ayele

4 General introduction cold chambers and 6-12 months without any proper storage facilities (Azene et al. 1993). Protocols have been successfully developed for the micropropagation (Tileye et al.

2005a), in vitro regeneration (Tileye et al. 2005b) and genetic transformation (Tileye et al. 2007b) of H. abyssinica.

1.4 Ecology and natural distribution of Hagenia abyssinica

Hagenia abyssinica is confined to Africa and its eco logical range stretches from Eritrea in the North to Zimbabwe in the South, including Burundi, Central African Republic, Congo, Ethiopia, Kenya, Malawi, Rwanda, Sudan, Tanzania, Uganda, and Zambia (He- deberg 1989; http:/www.worldagroforestry.org). Fos Fig 6 closed Hagenia forest at Dod- sil pollen records suggested that Hagenia immigrated dola-Dachosa, Photo: Taye B Ayele into Ethiopia from the south during the late Pleisto cene (since 16,700 years Before Present (BP)) and became abundant in the southern regions of Ethiopia about 2500 years BP (Paper I). It grows within an altitudinal range of 1,850 to 3,700 m asl (Hedeberg 1989; Friis 1992; Azene et al. 1993; Legesse 1995) inhabiting the montane forests, montane woodlands Fig 7 Hagenia tree retained on Bonsho and montane grasslands (Fig. 6-9). Tileye et al. farmland (close to Hagere Mariam), Photo: Taye B Ayele (2007b) reported

that Hagenia is a late successional species; but field observations during the present work witnessed sap lings emerging in disturbed areas such as road cuts in Bale and Bonga and hence did not support such a

designation. It was also reported that Hagenia has a Fig 8 Typical wooded grassland at regeneration cycle associated with heavy forest fires Deyu (close to Kofele) dominated by Hagenia. This population is suffering (http://database.prta.org/PROTA-html/Hagenia- from strangling by Ficus spp. P. Photo: Taye B Ayele %20abyssinica En.htm), suggesting that it is a pio-

5 General introduction neer species. Furthermore, Finkeldey & Hattemer (2007) argued that pioneer species have a larger seed shadow (typical for wind-dispersed species like Hagenia) than species of late successional stages.

Fig 9 Hagenia retained on grazing- land at Doddola-Serofta homestead. Photo: Taye B Ayele

1.5 Economic and ecological significance of Hagenia abyssi nica

Hagenia abyssinica is one of the best timber species in Ethiopia and its furniture is preferred for its strength, fine texture and attractive appearance (Fig. 2c). It is also used for producing veneers, flooring, cabinets and fuel wood (Azene et al. 1993; Getachew 2006). The tree is used to place traditional beehives (Fig 2b) and it also attracts

Fig. 10 Traditional beehives birds. The concoction made from the powder of dried placed on Hagenia trees. Photo: Taye B. Ayele female inflorescences is used as a purgative and taenicide against tapeworm in Ethiopia (Pankhurst 1969; Jansen 1981; Hedeberg 1989; Dawit & Ahadu 1993; Berhanu et al. 1999). Despite its dreadful and unpleasant taste, the infusion of Kosso has been most extensively used as vermifuge in rural Ethiopia. Overdose of Kosso may be fatal and may also cause abortion. Honey obtained from beehives located near Hagenia abyssinica trees and collected imme diately after their flowering is also effective in ex pelling tapeworms (http://database.prota.org/PRO- Fig. 11 A part of a wooden stage TAhtml/Hagenia%-20abvssinicaEn-.htm). The med made from Hagenia lumber. Photo: Taye B. Ayele ical use of Kosso was recorded as early as the six General introduction teenth century by an Ethiopian monk known as Aba Bahrey who described that the inha bitants of the Northern provinces took the drug to kill and rid their stomachs of certain little worms (reviewed by Pankhurst 1969). Berhanu et al. (1999) reported that Merck in Germany produced the first crystalline substances called kosins from the female flowers of Hagenia in 1870 and it was then in corporated in the European pharmacopoeia. With the advent of modem medicine that have reliable dosage and action, kosso is no more used as tapeworm expel- lant internationally; but it is still locally traded and used in rural parts of Ethiopia. In some areas farmers Fig. 12 Hagenia generously retain scattered Hagenia trees on their farms because it enriching a farm soil in Bon- sho (close to Hagere Mariam). enriches the soil by generously shedding its leaves dur Photo: Taye B Ayele ing the dry season (personal observation, see Fig. 2d). The leaves, seeds and bark are used as fodder, condiment or spice, and for dyeing textiles to yellowish red, respectively (http://database.prota.org/PRQTAhtml/Hagenia%20- abvssinica En.htm). Hagenia is a graceful and beautiful tree of high aesthetic value, es pecially when in blossom.

1.6 Rationale

Because of its quality timber, H. abyssinica has been logged heavily and selectively. It is one of the endangered tree species in Ethiopia (Legesse 1995). The Forestry Proclama tion No. 94/1994 of Ethiopia prohibits the felling of Hagenia abyssinica, Cordia africana, Afrocarpus falcatus and Juniperus procera (Anonymous 1994). Despite the proc lamation, the destruction of the populations of these species is continuing unabated be cause of the lack of mechanisms to enforce the law. Forest decline has many effects on the giant gene reservoir that is represented within forest trees (Hattemer and Melchior

1993). Old Hagenia trees are dying without recruiting new generation and this has aggra vated the level of threat on the species.

7 General introduction

In order to develop appropriate conservation strategies that, inter alia, preserve maximum genetic diversity, it is imperative to know the extent and distribution of genetic variation within a species (Bawa & Krugman 1990; Loveless and Hamrick 1984). Investigation of intraspecific genetic variation may help to assess extinction risks and evolutionary poten tial (fitness) in a changing world (Bawa & Krugman 1990; Hedrick 2001) and is instru mental to identify appropriate units for conservation of rare and threatened species (New ton et al. 1999). The preservation of germplasm in genebanks and the establishment of in situ and ex situ conservation stands requires sound knowledge of the genetic structure of a given species in order to capture the optimum genetic and demographic variations. The genetic diversity of few populations of H. abyssinica was investigated using anonymous RAPD (Kumilign 2005) and ISSR (Tileye 2007b) markers. Both studies covered small spatial scale contrasting to the widespread distribution of the species in Ethiopia and were also limited by the number of samples per population. The chloroplast DNA (cpDNA) of Hagenia has never been investigated before. Therefore, considering the superior econom ic and ecological importance and the alarming depletion of the species, it is crucial to in vestigate the genetic diversity within and among populations of H. abyssinica at the chloroplast markers and at the total genome level, covering the species' natural distribu tion range in Ethiopia.

1.7 A ims and predictions

1.7.1 Objectives

The research is aimed at the following objectives: • to examine the colonization history of H. abyssinica in Africa • to analyze the phylogeographic pattern of the species in Ethiopia using DNA and fossil pollen data • to assess genetic variation and the association with morphological and ecological diversities • to assess and compare genetic variation levels in both sexes • to use the results of the study to establish conservation strategies for the species. General introduction

1.7.2 Hypotheses

The following major hypotheses were tested using two types of molecular markers:

Chloroplast microsatellites (Paper I: Colonization history and phylogeography of Hagenia abyssinica (Bruce) J.F. Gmel in Ethiopia inferred from chloroplast microsatellite markers)

1) Due to limited seed dispersal and possibly rare long-distance seed dispersal, there is a strong differentiation among populations but low variation within populations 2) Populations show geographic structuring primarily induced by mutation and isolation by distance

3) Based on the existing fossil pollen records, Hagenia immigrated into Ethiopia from the south

AFLP (Paper II: Spatial distribution of genetic diversity in Hagenia abyssinica (Bruce) J.F. Gmel in Ethiopia assessed by AFLP molecular markers)

1) There is high variation within-populations due to effective gene flow from different pollen and seed sources and very low differentiation among-populations due to long distance pollen and seed dispersal 2) The species does not lose genetic diversity during colonization due to effective gene flow that counteracts effects of genetic drift. Likewise, the populations representing the two chloroplast lineages show similar levels of genetic diversity, even though the derived one originated by a single mutational event (from a single seed) 3) Given the wind-dispersed and wind-pollinated nature of Hagenia abyssinica, there is no fme-scale spatial genetic structure.

9 General introduction

1.7.3 Major research questions

The following major research questions were addressed:

1) What are the levels and patterns of genetic variation in Hagenia abyssinica (Papers I, II & III)? 2) Which factors shaped genetic variation patterns of Hagenia in Ethiopia (Papers I & II)? 3) Is there congruence between molecular data and palynological evidences to infer the relationships among genealogical lineages and migration routes of the species (Paper I)? 4) Which conservation strategies are appropriate to save Hagenia from extinction (Paper III)?

10 2. Research approaches

2.1 Sampling

Twenty two natural and three planted populations were sampled from forests, woodlands and farmlands known to have stands of H. abyssinica within the various Tree Seed Zones of Ethiopia. Three of the populations were sampled from church/monastery forests. The description of the Tree Seed Zones of Ethiopia in which H. abyssinica is growing is an nexed (Appendix l). The sampled populations represent most of the extant distribution of the species in the country ranging from 05°5l'39"N (Hagere Mariam) in the south to 13°lriO "N (Debark Mariam) in the north, and from 35°4l'59"E (Wonbera) in the west to 40°l4'32"E (Dindin) in the east. The distance between populations ranges from 2 1 to 806 km and they are located within an altitudinal range of 2200 m asl at Bonga to 3200 m asl at Wofwasha. The pairwise geographic distance matrix for the 22 natural populations of Hagenia is presented in Appendix 3. Temperatures range from an absolute minimum of-l°C at Dinsho to a maximum of 33.5 °C at Kosso Ber. Maps showing the spatial dis tribution of individual trees in each population are provided in Appendix 4.

2.2 Morphological assessment

Dimensional, counted and visually observed morphological variables were assessed from 26-50 trees from each population. Details on the traits assessed are given in Paper I.

2.3 DNA isolation

Young leaves were collected and partially desiccated in paper bags before drying with silica gel and stored at room temperature before DNA extraction. Total genomic DNA was isolated from 20 mg leaves after shipment to Germany following the DNeasy 96 kit protocol of Qiagen® (Hilden, Germany).

11 Research approaches

2.4 Chloroplast microsatellites

Three polymorphic consensus chloroplast microsatellite primers (CCMP2, CCMP6 & CCMP10 (nomenclature according to Weising and Gardner 1999)) were used to screen 273 samples (9-12 individuals per population) from 25 populations. Details on the me thods and data analyses are described in Paper I.

2.5 DNA Sequencing

Comparative sequencing of 18 fragments of the three chloroplast loci was performed to confirm the amplified regions and to determine the molecular basis for size variation. Fragments of CCMP2 (224-235 bp) were sequenced directly while fragments of CCMP6 (140-142 bp) and CCMP10 (96-97 bp) were cloned due to their small sizes. In addition, fragments from three out-group species from the same family were sequenced for comparison. Details on the methods and data analyses are described in Paper I.

2.6 AFLP analyses

A total of 596 samples (23-24 individuals/population) were analysed at the nuclear en coded AFLP markers using the selective primer combination E41-M67 (nomenclature according to Keygene N.V. ®). Details on the methods are described in Paper II.

12 3. Summary of results

3.1 Morphological data

The morphological traits observed in Hagenia abyssinica were highly variable among populations. The ranges of absolute morphological values are presented in Appendix 2. The one-way analysis of variance (ANOVA) revealed a strikingly significant differentia tion (p<0.00l) among the 22 natural populations in all morphological traits. The cluster analysis based on the average taxonomic distances matrix of leaf traits grouped the popu lations into two major clusters and separated four outlier populations. In general, no clear association between geographic regions and taxonomic distances could be observed. The average taxonomic distances for all morphological traits did not show any correlation with the average Euclidean distances of climatic variables (r = 0.17062, p = 0.9281), in dicating lack of association between quantitative morphological traits and climatic vari ables.

The total number of the extant individual Hagenia trees throughout the country (includ ing a rough estimation of scattered trees not included in the present study) is estimated as 7,000, the majority of which are old and dying without recruiting new generations. De tails on the results of morphological and ecological variables are presented in Paper III.

3.2 Chloroplast microsatellite data

The combination of 8 variants from the three chloroplast loci resulted in six haplotypes that were phylogenetically grouped into two lineages. The haplotypes demonstrated a very strong geographic pattern as a result of highly restricted gene flow by seeds. The two lineages were separated by an indel (insertion/deletion) of 10 nucleotides in locus CCMP2. The first lineage encompasses haplotypes H4, H5 & H6, which are distributed in the south-western and northern regions, while the second lineage contains haplotypes HI, H2 & H3 in the central and southern regions. A remarkable subdivision of cpDNA diversity was found in the species as indicated by a high coefficient of genetic differentia-

13 Summary of results

tion (G s t = 0. 899, N st = 0. 926). The analysis of molecular variance (AMOVA) showed that 92.3% of the total genetic diversity is represented among populations. Details on the results of the chloroplast microsatellite analyses are described in Paper 1.

3.3 Sequence data

Sequencing confirmed homology of the three chloroplast loci to the expected regions of the chloroplast genome. The observed variation was due to variable numbers of poly (A) or poly (T) repeats in the microsatellites of all loci and a large indel of 10 bp in the flank ing region of locus CCMP2. In total, there were 4 variable sites: 3 short indels in the mi crosatellite motifs and one large indel in the flanking region. More details on the se quence data is provided in Paper I.

3.4 AFLP data

Moderate to high gene diversities were observed at AFLP loci ranging from 13.9% at Dodola Serofita to 36.2% at Dinsho. The mean gene diversity in subdivided populations of Hagenia abyssinica showed high within population variation (19.5%) and moderate but significant population differentiation (FSt = 7.7%). The phylogenetic tree derived from Nei’s (1978) genetic distances congregated the populations into two major clusters, but does not reflect the geographic origin of the populations. Despite marked differences in genetic diversities for some populations, mean genetic diversities for the two sexes are nearly the same (He = 0.207 ± 0.013 for male, He = 0.201 ± 0.019 for female). A test of association between geographic and genetic distances based on AFLP markers showed a very low and non-significant correlation (r = 0.14607, p = 0.9024). The multivariate taxonomic distances of leaf traits are also not correlated with genetic distances (r= - 0.03484, p = 0.3926), showing that the genetic differentiation at neutral AFLPs is not as sociated with the leaf-morphological differences among populations. Ten out of 21 natu ral populations showed significant spatial genetic structure (SGS). Details on the results of AFLPs are provided in Paper II.

14 4. General discussion

The palynological data obtained from the existing fossil pollen records suggested a northward post-glacial colonization of Hagenia in Africa with the oldest available record from Burundi (ca. 34,000 calibrated years before present (cal yrs BP)). The signal of Hagenia in the pollen records from Burundi was quite high around 11,500 cal yrs BP (Bonnefille et al. 1995), whereas its major expansion in the Bale Mountains (southern Ethiopia) was after 2500 cal yr BP (Mohammed et al. 2004; Mohammed & Bonnefille 1998), suggesting a recent colonization of Ethiopia.

The morphological traits showed a significant differentiation among the 22 natural popu lations of H. abyssinica. However, the amount and distribution of morphological trait variation across different habitats, geographic regions and climatic conditions did not show any pattern. The maximum height of Hagenia has been reported to be 20 meter in the existing literature (Hedeberg 1989; Azene et al. 1993; Legesse 1995; Tileye 2007a). However, the present inventory recorded trees growing up to 35 m (mean maximum height = 21.2 meters, n=l 109). Similarly, the number of leaflets was reported to be be tween 5 and 8 on each side (i.e., between 10 and 16 on both sides) whereas the present inventory provided a wider range of 7 to 19 leaflets on both sides, but always in odd num bers because of the presence of an apical leaflet. The lack of correlation between the Euc lidean distances of morphological and climatic data suggests that the observed morpho logical traits are not involved in the adaptation to different climatic conditions.

The chloroplast DNA analysis revealed a strong differentiation among populations, but low variation within populations. The very high level of genetic variation among popula tions of Hagenia at cpDNA suggested a restricted migration of seeds among regions, which is also reflected in the observed geographic structuring of haplotypes (Fig. 3 of paper I). The coefficient of population differentiation (G st) in Hagenia is higher than or comparable to Gst values recorded for other species with heterogeneous mode of seed dispersal, including wind dispersal, investigated by chloroplast markers (Newton et al. 1999, Petit et al. 2003). The geographic distribution of chloroplast haplotypes and their

15 General discussion genealogical relationships observed in Hagenia demonstrated a highly significant asso ciation. The nested clade phylogeographic analysis (NCPA) inferred that restricted gene flow associated with contiguous range expansion and rare long-distance seed dispersal shaped the genetic structure in the chloroplast DNA of Hagenia. The chloroplast data suggests that Hagenia colonized Ethiopia first through the southwest mountains (popula tion Bonga, BG).

The moderately high genetic diversity at AFLPs of Hagenia reflects effective gene flow within populations from different pollen and seed sources, resulting in a very low popula tion differentiation, which in turn reflects effective long-distance pollen dispersal. Inter estingly, the maximum genetic diversity was recorded for a well-protected Park Forest (Dinsho) whereas the lowest gene diversities were recorded for the two farmland popula tions (Doddola Serofta and Hagere Mariam), pointing to negative human impact on ge netic diversity. Nybom (2004) reported a slightly higher mean within-population diver sity (Hpop) of 0.22 at RAPD, 0.23 at AFLP and 0.22 at ISSR markers. The overall mean gene diversity of Hagenia at AFLPs (He = 0.195) is comparable to some other plant spe cies such as the insect-pollinated Hibiscus tiliaceus (Malvaceae, He = 0.198, Tang et al., 2003) and the wind-pollinated Acanthopanax sessiliflorus (Araliaceae, He = 0.187, Huh et al., 2005) but lower than the insect-pollinated Malus sylvestris (Rosaceae, He = 0.225, Coart et al., 2003). H. abyssinica exhibited higher mean gene diversity than some other tropical and subtropical tree species such as the bird-pollinated Lobelia giberroa (Apocy- naceae, He = 0.066, Mulugeta, et al. 2007) the insect-pollinated Shorea leprosula (Dip- terocarpaceae, He = 0.161, Cao et al. 2006), the insect-pollinated Shorea parvifolia (Dip- terocarpaceae, He = 0.138, Cao et al. 2006), the insect and wind-pollinated Acer skutchii (Sapindaceae, He = 0.15, Lara-Gomez et al., 2005) and the bird-pollinated Pelliciera rhizophorae (Pellicieraceae, Ht = 0.117, Castillo-Cardenas et al. 2005) at AFLP loci. Tileye et al. (2007) reported higher mean gene diversity (0.30) in 12 populations of Hagenia from central and southern regions of Ethiopia at ISSR markers. But Qian et al. (2001) and Nybom (2004) argued that ISSR markers generally over-estimate gene diver sity as compared to other markers. Hagenia also showed lower mean gene diversity than some other tree species growing in Ethiopia investigated with AFLP markers, notably,

16 General discussion

the insect-pollinated Cordia africana (Boraginaceae, He = 0.287, Abayneh 2007) and the wind-pollinated Juniperusprocera (Cupressaceae He = 0.269, Demissew 2007).

No trend of decreasing genetic diversity during colonization was detected, reflecting ef fective gene flow. In contrast, Lobelia giberroa, which entered Ethiopia also from the south (Mulugeta, et al. 2007), Carpinus betulus (Betulaceae) in Europe (Coart et al.

2005) and Ptercarpus officinalis (Fabaceae) in the Caribbean (Rivera-Ocasio et al. 2002) demonstrated decreasing diversity during recolonization (all based on AFLP analyses).

Comparable levels of population differentiation were found at AFLPs of Cordia africana

(Abayneh 2007), Acer skutchii (Lara-Gomez et al. 2005), Acanthopanax sessilijlorus

(Huh et al. 2005) and Carpinus spp (Coart et al. 2005). Tileye et al. (2007b) found a higher coefficient of differentiation among 12 populations of Hagenia at ISSR markers.

Higher coefficients of population differentiation than Hagenia were also reported for

Shorea species (Cao et al. 2006), Hibiscus tiliaceus (Tang et al. 2003) and Pelliciera rhizophorae (Castillo-Cardenas et al. 2005). On the other hand, FSt values lower than that of Hagenia were reported for wild Malus sylvestris (Coart et al. 2003).

The population differentiation is much higher in the chloroplast genome than in the nu clear genome of Hagenia abyssinica, as revealed by ISSR (Gst = 0.25; Tileye et al.

2007a) and AFLP markers (F st = 0.077, Paper II). Likewise, Rendell and Ennos (2002) found a population differentiation that was 10-fold higher in the chloroplast genome of

Calluna vulgaris (L.) Hull (Ericaceae), than in the nuclear genome. In general, maternally inherited genomes experienced considerably more subdivision (mean G st value of -0.64) than biparentally inherited genomes (mean G st value o f-0.18) of angiosperm species (reviewed by Petit et al. 2005). Due to its maternal inheritance, cpDNA in angiosperms is transmitted only through seeds and therefore show a higher differentiation among popula tions than nuclear genes that are biparentally inherited. Consequently, genetic variation in the chloroplast genome often shows a strong geographical structure than the nuclear ge nome (e.g., Cavers et al. 2003).

17 General discussion

A weighted-score population prioritization matrix (WS-PPM) that combines genetic, morphological and demographic criteria is developed and used for the first time to pri oritize populations of Hagenia for conservation and domestication. Paper III describes the prioritization process and also provides separate priority lists for in situ conservation, ex situ conservation, and for tree improvement and domestication programs. 5. Conclusions and outlook

The present study at both morphological and molecular markers contributed valuable re sults that increased our understanding on the patterns of genetic diversity in Hagenia abyssinica and provided useful information for planning conservation, tree improvement and domestication programs. It was possible to infer the phylogeography, colonization history and the factors shaping the genetic variation of the species from chloroplast mi crosatellite and AFLP markers.

The chloroplast haplotypes of Hagenia abyssinica demonstrated a pattern of isolation by distance. Due to the recent colonization of the country by the species, it was possible to identify rare long-distance dispersal and mutation events that contributed in shaping the genetic structure of the species at chloroplast (cp) DNA. A remarkable subdivision of cpDNA diversity was found as indicated by a high coefficient of genetic differentiation. The study demonstrated that restricted gene flow, contiguous range expansion and rare long-distance seed dispersal events shaped the genetic structure in the chloroplast ge nome of Hagenia in Ethiopia. Unlike most of the wind-dispersed tree species, the chlo roplast haplotypes found in Hagenia showed a clear pattern of congruence between their geographical distribution and genealogical relationships.

Despite the relatively recent immigration of Hagenia abyssinica into Ethiopia, popula tions showed moderate to high gene diversities (//e = 0.139-0.362), and moderate but sig nificant genetic differentiation (Fst = 0.077), reflecting high levels of post-colonization gene flow among populations. The moderate to high intraspecific variation and a wide vertical distribution of the populations (2200 to 3200 m asl) may suggest that Hagenia might have occupied wider areas in the past than at present. The sizes of the extant popu lations were reduced to very small patches due to human impact, apparently affecting their genetic structure.

The populations of Hagenia abyssinica are severely decreasing without recruiting young trees except for Bonga, the only viable population in southwest Ethiopia. Hagenia

19 Conclusions and outlook should, therefore, be recognised as a critically endangered tree species and urgent action is needed to save it from extinction.

Analysis of cpDNA types, intraspecific genetic variation and palynological inventories, including all countries where the species is known to grow, 1) would fully resolve the genealogical relationships within the natural distribution range of the species, 2) help to identify the glacial refugia, and 3) is indispensable to fully understand the colonization history of Hagenia in Africa. The screening of a large number of AFLP markers in segre gating populations may help to identify markers for sex determination in Hagenia. The pollination mechanism of Hagenia that has been reported elsewhere (wind) should be re examined in view of the tree’s investment of energy to produce alluring flowers.

20 6. Summary

Deforestation and forest fragmentation in general, and extraction of timber in the form of selective logging in particular have serious consequences on the long-term maintenance of genetic diversity and fitness in plants. It is imperative to know the extent and distribu tion of genetic variation within a given species in order to develop appropriate conserva tion strategies that inter alia preserve “optimum” genetic diversity. The genetic variation of Hagenia abyssinica (Bruce) J.F. Gmel has been investigated at morphological and mo lecular markers in order to identify populations for conservation, tree improvement and domestication programs.

The monotypic species Hagenia abyssinica (Rosaceae) is an anemogamous and anemochorous broad-leaved dioecious tree species that is native to Africa. The major aims of this study are to 1) examine the colonization history of Hagenia abyssinica in Africa, 2) analyze the phylogeographic pattern of the species using DNA and pollen data, 3) assess genetic variation and the association with morphological and ecological diversities, 4) assess genetic variation levels in both sexes, and 5) use the results to establish conserva tion strategies for the species.

The colonization history of Hagenia abyssinica is inferred from the existing fossil pollen records. The fossil pollen evidences suggested that postglacial colonization of Hagenia followed a northward route in Africa and that it immigrated into Ethiopia from the south during the late Pleistocene (since 16,700 years Before Present). Morphological and mole cular genetic analyses were performed in 22 natural and 3 planted populations sampled from the natural distribution range of the species within the Ethiopian highlands. Dimen sional, counted and visually observed morphological variables were assessed for a total of 1109 trees (26-50 individuals per population). Two molecular marker techniques, namely, chloroplast microsatellites and nuclear encoded AFLP markers were employed to investigate genetic diversity and to infer the factors shaping the genetic variation, phylogeography, and colonization history of the species. The genetic variation of 273 indi viduals from 25 populations was analysed at three polymorphic chloroplast microsatellite

21 Summary markers (CCMP2, CCMP6 & CCMP10). Homology of the three loci to the respective regions of the chloroplast genome was confirmed by comparative sequencing of 21 frag ments. The intraspecific genetic variation of 596 individuals from 25 populations was analysed at the AFLP markers using the selective primer combination E41-M67 (nomen clature according to Keygene N.V.®).

The analysis of variance (ANOVA) revealed a significant differentiation among the 22 natural populations of Hagenia abyssinica in all quantitative morphological traits (p<0.001). However, the global multivariate analyses of the entire morphological data set did not clearly separate the individuals among the populations. The average taxonomic distances for all morphological traits did not show any correlation with the average Euclidean distances of climatic variables (r = 0.17062, p = 0.9281), indicating a lack of association between quantitative morphological traits and climatic variables. The cluster analysis based on the average taxonomic distances of leaf characters showed a geograph ical pattern with few exceptions and assembled the populations into two major clusters and separated four outlier populations from the rest.

The analysis of cpDNA using microsatellite markers revealed a total of six haplotypes that were phylogenetically grouped into two lineages. The chloroplast haplotypes identi fied in Hagenia demonstrated a strong pattern of congruence between their geographical distribution and genealogical relationships. Eighty percent of the populations were fixed on one type. A very low haplotype diversity within populations (hs = 0.079, vs = 0.058) and a remarkable subdivision of cpDNA diversity (GSt = 0. 899, NSt = 0. 926) was ob served. The study demonstrated that restricted gene flow through seeds, contiguous range expansion and mutation shaped the genetic structure in the chloroplast genome of Hagenia. Due to the recent colonization of the country by the species, it was also possible to identify rare long-distance dispersal events that contributed in shaping the genetic structure of the species in Ethiopia.

Out of 106 unequivocally scored AFLP markers, 91.5% were polymorphic. Despite the relatively recent immigration of Hagenia abyssinica into Ethiopia, populations showed

22 Summary

moderate to high gene diversities (Hs = 0.139-0.362), and moderate but significant ge netic differentiation (Fst = 0.077), reflecting high levels of post-colonization gene flow particularly by pollen among populations. There were no significant differences in gene diversity between sexes, even though single populations exhibited marked differences. AFLP profiles did not show any diagnostic markers for neither of the two sexes. No trend of decreasing genetic diversity was detected during colonization, confirming effective gene flow among populations. Despite the dispersal of seed and pollen of Hagenia by wind, a significant non-random fine-scale spatial genetic structure (SGS) is observed up to 80 m in some populations.

The multivariate taxonomic distances of leaf traits is not correlated with Nei’s genetic distances (r= -0.03484, p = 0.3926), showing that the genetic differentiation at anony mous AFLPs is not associated with the leaf-morphological differences among popula tions. As expected, the coefficient of population differentiation is found to be much lower for the biparentally inherited nuclear genome (represented by AFLPs) of Hagenia abys sinica than in the maternally inherited chloroplast genome. Comparative analyses of the amount and distribution of the genetic diversity of Hagenia abyssinica with other tree species are provided. In conclusion, population history can be reconstructed by chlorop last microsatellite data reflecting seed dispersal while AFLPs identify geographic regions and populations of high genetic diversity. A weighted-score population prioritization ma trix (WS-PPM) that combines genetic, morphological and demographic criteria was de veloped and used for the first time to prioritize populations for in situ conservation, ex situ conservation, and for tree improvement and domestication programs. Extremely ur gent decision is needed to launch conservation and massive plantation programs of the African redwood to ensure the long-term survival of the species and to boost its economic and ecological values.

23 7. Zusammenfassung

Besiedlungsgeschichte, Phylogeografie und Erhaltungsgenetik der vom Aussterben bedrohten Baumart Hagenia abyssinica (Bruce) J. F. Gmel in Athiopien

Abholzung und Fragmentierung von Waldem im Allgemeinen, und im Besonderen die Entnahme von Holz bei der selektiven Abholzung, haben emste Auswirkungen auf die Langzeiterhaltung genetischer Diversitat und auf die biologische Fitness von Pflanzen. Es ist zwingend notwendig das AusmaB und die Verteilung von genetischer Variation in einer Art zu bestimmen, um angemessene Schutzstrategien zu entwickeln, die unter ande- rem eine hohe genetische Diversitat erhalten. Die genetische Variation von Hagenia abyssinica (Bruce) J. F. Gmel wurde mit Hilfe von morphologischen und molekularen Markem untersucht, um Populationen fur Artenschutz und Zuchtung zu identifizieren.

Die monotypische Art Hagenia abyssinica (Rosaceae) ist eine anemogame und anemo- chore diozische Laubbaumart, die in Afrika heimisch ist. Die Hauptziele dieser Studie sind 1) ihre Besiedlungsgeschichte in Afrika zu untersuchen, 2) ihre phylogeografischen Muster mit Hilfe von DNA- und Pollendaten zu analysieren, 3) die genetische Variation und ihre Beziehung zur morphologischen und okologischen Diversitat zu berechnen, 4) das AusmaB an genetischer Variation in beiden Geschlechtem abzuschatzen, und 5) die Ergebnisse zu nutzen, um ErhaltungsmaBnahmen einzuleiten.

Die Besiedlungsgeschichte von Hagenia abyssinica wurde von anderen Autoren mit Hil fe von fossilen Pollenvorkommen rekonstruiert. Die fossilen Pollen deuten auf eine nordwarts gerichtete postglaziale Kolonisierungsroute von Hagenia in Afrika hin. Die Art ist vermutlich im spaten Pleistozan (ab 16,700 Jahren vor heute) aus dem Siiden nach Athiopien eingewandert. Morphologische und molekulare Analysen wurden in 22 natur- lichen und drei angepflanzten Populationen im natiirlichen Verbreitungsgebiet der Art im Hochland von Athiopien durchgefuhrt. Morphologische Eigenschaften wurden bei insge- samt 1109 Baumen (26 bis 50 Baume pro Population) untersucht. Zwei molekulare Mar ker, Chloroplastenmikrosatelliten und kemkodierte AFLP-Marker, wurden angewandt,

24 Zusammenfassung um die genetische Diversitat zu untersuchen und um die Faktoren zu bestimmen, die ge netische Variation, Phylogeographie und Besiedlungsgeschichte der Art gepragt haben. Die genetische Variation von 273 Baumen wurde mit drei Chloroplastenmikrosatelliten (CCMP2, CCMP6 & CCMP10) analysiert. Die Homologie der drei Genorte zu den zuge- horigen Regionen des Chloroplastengenoms wurde durch vergleichende Sequenzierungen von 21 Fragmenten bestatigt. Die intraspezifische genetische Variation von 596 Genoty- pen aus 25 Populationen wurde mit AFLPs untersucht, wobei die Primerkombination E41-M67 verwendet wurde (Nomenklatur entsprechend Keygene N.V.®).

Durch Varianzanalysen (ANOVAs) wurde eine signifikante Differenzierung zwischen den 22 natiirlichen Populationen von Hagenia abyssinica in alien quantitativen morpho logischen Merkmalen festgestellt (p<0.001). Allerdings konnten durch umfassende mul tivariate Analysen des gesamten morphologischen Datensatzes nicht alle Individuen den Populationen zugeordnet werden. Die durchschnittliche taxonomische Distanz aller mor phologischen Merkmale war nicht mit den durchschnittlichen euklidischen Distanzen klimatischer Variablen korreliert (r=0.17062, p=0.9281. Clusteranalysen, basierend auf einer Matrix aus durchschnittlichen taxonomischen Distanzen von Blattmerkmalen, wiesen ein geographisches Muster mit wenigen Ausnahmen auf. Es konnten zwei „Haupt- cluster“ und 4 davon getrennte „Nebencluster“ unterschieden werden.

Die Analyse von cpDNA-Mikrosatelliten ergab sechs Haplotypen, die phylogenetisch in zwei Linien eingruppiert werden konnten. Die in Hagenia identifizierten Haplotypen zeigten eine starke Ubereinstimmung zwischen ihrer geographischen Verteilung und ihrer Abstammung. Achtzig Prozent der Populationen waren auf einen Haplotyp fixiert. Es wurde eine sehr geringe Diversitat an Haplotypen innerhalb der Populationen (hs=0.079, vs=0.058) und eine deutliche Untergliederung der cpDNA Diversitat zwischen den Popu lationen (Gst=0.899, NSt=0.926) festgestellt. Die vorliegende Studie zeigte, dass einge- schrankter GenfluB durch Samen, ein gleichmaBige Ausbreitung und Mutationen die ge netische Struktur im Chloroplastengenom von Hagenia beeinfluBt haben. Aufgrund der rezenten Besiedlung des Landes durch die Art war es zusatzlich moglich, seltene Verbrei-

25 Zusammenfassung

tungsereignisse tiber weite Strecken zu identifizieren, die an der Ausbildung der genetischen Struktur der Art in Athiopien beteiligt waren.

Von 106 AFLP Markem waren 91.5% polymorph. Die Populationen wiesen eine mittlere bis hohe genetische Diversitat (He=0.139-0.362) und eine niedrige aber signifikante gene

tische Differenzierung auf (Fst=0.077). Diese Ergebnisse spiegeln einen hohen GenfluB nach der Besiedlung besonders durch Pollenflug zwischen den Populationen wider. Es zeigten sich keine signifikanten Unterschiede zwischen mannlichen und weiblichen Indi- viduen hinsichtlich ihrer mittleren genetischen Diversitat, obgleich in den einzelnen Po pulationen deutliche Unterschiede auftreten konnten. Die AFLP- Profile zeigten keine diagnostischen Marker fur die beiden Geschlechter. Anders als in viele anderen Baumar- ten erhohte sich die genetische Diversitat wahrend der Besiedlung. Daraus kann ge- schlossen werden, dass GenfluB und Mutationen in Verbindung mit der Besiedlungsge schichte einen starken Einfluss auf die intraspezifische Variation von Hagenia hatten.

Die multivariate taxonomische Distanzmatrix der Blattmerkmale ist nicht mit Nei's gene tischer Distanzmatrix korreliert (r=-0.03484, p=0.3926), da genetische Differenzierung an neutralen AFLPs nicht mit blattmorphologischen Unterschieden zwischen Populatio nen assoziiert ist. Wie erwartet, ist der Koeffizient der Populations-differenzierung im biparental vererbten Kemgenom (reprasentiert durch AFLPs) sehr viel kleiner als im ma ternal vererbten Chloroplastengenom. Es werden vergleichende Analysen der genetischen Diversitat und der Differenzierung von Hagenia abyssinica mit anderen Baumarten dar- gestellt. Zusammenfassend lasst sich feststellen, dass die Populationsgeschichte mit Chloroplastenmikrosatelliten, welche die Verbreitung der Samen widerspiegeln, rekonstruiert werden kann, wahrend die AFLP-Daten geografische Regionen und Populationen mit hoher genetischer Diversitat identifizieren konnen. Eine Weighted-Score Population Prioritization Matrix (WS-PPM), die genetische und demografische Kriterien vereint, wurde entwickelt und genutzt, um Schwerpunktpopulationen fur in ,v//«-Erhaltung, ex si- /w-Erhaltung, und fur Zucht- und Ertragsprogramme zu finden. Sehr schnelle Entschei- dungen sind notig, um Programme zur Arterhaltung und groBangelegte Plantageprojekte

26 Zusammenfassung von afrikanischem Rotholz zu starten, die ein Uberleben der Art iiber lange Zeitraume und ihren okonomischen und okologischen Nutzen sichem konnen.

27 8. References

Abayneh D (2007) Genetic variation in Cordia africana Lam. in Ethiopia. PhD thesis, Georg-August University of Gottingen.

Aalbaek A (1993) Tree seed zones for Ethiopia. National Tree Seed Project, Addis Ababa

Anonymous (1994) Forest Conservation, Development and Utilization Proclamation. Proclamation No. 94/1994. Negarit Gazeta. The Transitional Government of Ethi opia. Addis Abeba, Ethiopia.

Avise JC (2004) Molecular Markers, Natural History and Evolution, 2nd edn. Sinauer Associates Inc, Sunderland, MA.

Azene BT, Bimie A, Tegnas B (1993) Useful Trees and Shrubs for Ethiopia: Identifica tion, Propagation and Management for Agricultural and Pastoral Communities. Technical Handbook No. 5, Regional Conservation Unit, Swedish International De velopment Authority (SIDA).

Barrett SCH, Kohn JR (1991) Genetic and evolutionary consequences of small popula tion size in plants: implications for conservation. In: Falk DA & Holsinger KE (eds) Genetics and conservation o f rare plants. Oxford University Press, New York.

Bawa KS, Krugman SL (1990) Reproductive biology and genetics of tropical trees in re lation to conservation and management. In: Rain Forest Regeneration and Management (eds Gomez-Pampa A, Whitmore TC, Hadley M), pp. 119-136. The Parthenon Publish ing Group.

Berhanu M. Abegaz, Ngadjui BT, Merhatibeb Bezabih, Mdee LK (1999) Novel natural products from marketed plants of eastern and southern Africa. Pure and Applied Chem istry 71:919-926.

Bonnefille R, Riollet G, Buchet G, Icole M, Lafont R, Arnold M, Jolly D (1995) Gla cial/interglacial record from intertropical Africa, high resolution pollen and carbo data at Rusaka, Burundi. Quaternary Science Reviews 14: 917-936.

Cao, C-P., Finkeldey, R., Siregar, I.Z., Siregar, U.J., Gailing, O. (2006). Genetic diversity within and among-populations of Shorea leprosula Miq. and Shorea parvifolia Dyer (Dipterocarpaceae) in Indonesia detected by AFLPs. Tree Genetics and Genomes 2: 225-239.

Castillo-Cardenas MF, Toro-Perea N, and Cardenas-Henao H (2005) Population genetic structure of neotropical Mangrove species on the Colombian Pacific Coast: Pelliciera rhizophorae(Pellicieraceae). Biotropica 37: 266-273.

Cavers S, Navarro C, Lowe AJ (2003) Chloroplast DNA phylogeography reveals coloni zation history of a Neotropical tree, Cedrela odorata L., in Mesoamerica. Molecular Ecology 12: 1451-1460. References

Coart, E., Van Glabeke, S., Petit, R.J., Van Bockstaele, E. and Roldan-Ruiz, 1. 2005). Range wide versus local patterns of genetic diversity in hornbeam (Carpinus betulus L.). Conservation Genetics 6: 259-273.

Coart E, Vekemans X, Smulders MJM, Wagner I, Huylenbroeck JV, Bockstaele EV and Roldan-Ruiz I (2003) Genetic variation in the endangered wild apple (Mains sylvestris (L.) Mill.) in Belgium as revealed by amplified fragment length polymorphism and microsatellite markers. Molecular Ecology 12: 845-857.

Comps, B., Gomory, D., Letouzey, J., Thiebaut, B. and Petit, R.J. (2001) Diverging trends between heterozygosity and allelic richness during post glacial colonization in the European beech. Genetics 157: 389-397.

Dawit Abebe, Ahadu Ayehu (1993) Medicinal Plants and Enigmatic Health Practices of Northern Ethiopia. BSPE, Addis Ababa, Ethiopia.

Demissew SD (2007) Genetic variation of Juniperus procera Hochst. Ex Endl. popula tions in Ethiopia assessed by using microsatellite (SSR) and AFLP markers. MSc. Thesis. Georg-August University of Gottingen.

Dumolin-Lapegue S, Demesure B, Fineschi S, Le Corre V, Petit RJ. (1997) Phylogeo- graphic structure of white oaks throughout the European continent. Genetics 146: 1475-1487.

Ehrlich PR, Ehrlich AH (1991) Healing the planet. Addison-Wesly, Reading, MA.

Ellstrand NC, Elam DR (1993) Population genetic consequences of small population size: implications for plant conservation. Annual Review of Ecology and Systematics 24: 217-241.

Eriksson T, Hibbs MS, Yoder AD, Delwiche CF, Donoghue MJ (2003) The phytogeny of Rosoideae (Rosaceae) based on sequences of the internal transcribed spacers (ITS) of nuclear ribosomal DNA and the trnl/f region of chloroplast DNA. International Jour nal of Plant Sciences 164: 197-211.

Fichtl R, Admasu A (1994). Honeybee flora of Ethiopia. DED - Margraf Verlag, Wei- kersheim.

Finkeldey R, Hattemer HH (2007) Tropical Forest Genetics. Springer-Verlag, Berlin.

Friis I (1992) Forests and Forest Trees of Northeast Tropical Africa: Their Natural Ha bitats and Distribution Patterns in Ethiopia, Djibouti and Somalia. Kew Bulletin, Additional Series XV. Her Majesty's Stationery Office, London.

Getachew Desalegn (2006) Some Basic Physical And Mechanical Properties Of The Valuable Hagenia abyssinica Timber And Their Interactions: Imlpication For Its Ra tional Utilization, Ethiopian Journal o f Biological Sciences 5 (2): 117-135.

29 References

Girma Balcha (1999) Status of forest seed research in Ethiopia: Consequences for devel opment of forest genetic resources conservation strategy. In: Abebe, et al. (eds). For est Genetic Resources Conservation: Principles, Strategies and Actions. Proceedings of the National Forest Genetic Resources Conservation Strategy Development Work shop, June 21-22, Addis Abeba.

Hattemer HH & Melchior GH (1993) Genetics and its application to tropical forestry. In Pancel L (Ed). Tropical Forestry Handbook Vol 1, Springer-Verlag, Berlin.

Hedeberg O (1989) Rosaceae. In: Inga Hedeberg and Sue Edwards, eds. Flora of Ethi opia, Vol. 3, Pittosporaceae to Arralaceae. Addis Abeba and Asmara, Ethiopia; Upp sala, Sweden.

Hedrick PW (2001) Conservation genetics: where are we now? Trends in Ecology and Evolution 16: 629-636.

Heuertz M, Fineschi S, Anzidei MA et al. (2004) Chloroplast DNA variation and post glacial recolonization of common ash (Fraxinus excelsior L.) in Europe. Molecular Ecology 13: 3437-3452.

Huh M.K, Huh H.W. and Back K. (2005) Genetic diversity and population structure of Acanthopanax sessiliflorus (Araliaceae) using AFLP. Korean Journal o f Genetics 27: 71-79.

Jansen P.C.M. (1981) Spices, condiments and medicinal plants in Ethiopia, their taxono my and agricultural significance. Wageningen.

Jump AS, Penuelas P (2007) Extensive spatial genetic structure revealed by AFLP but not SSR molecular markers in the wind-pollinated tree, Fagus svlvatica. Molecular Ecology 16: 925-936.

King RA, Ferris C (1998) Chloroplast DNA phylogeography of Alnus glutinosa (L.) Gaertn. Molecular Ecology 7: 1151-1163.

Kumilign A (2005) Estimation of sex-related genetic diversity of Hagenia abyssinica (Bruce) J.F. Gmel. M.Sc. thesis, Addis Ababa University.

Laikre L, Ryman N (1996) Effects on intraspecific biodiversity from harvesting and en hancing natural populations. Ambio 25: 504-509.

Lara-Gomez G., Gailing O., and Finkeldey R. (2005) Genetic variation in isolated Mex ican populations of the endemic maple Acer skutchii Rehd. Allg Forst- u J-Ztg 176: 97-103.

Legesse N (1995) Indigenous trees of Ethiopia: Biology, uses and propagation tech niques. SLU, Reprocentralen, Umea.

Loveless MD, Hamrick JL (1984) Ecological determinants of genetic structure in plant populations. Annual Review of Ecology andSystematics 15: 65-95.

Luu H.T. (2005) Genetic variation and the reproductive system of Dipterocarpus cf.

30 References

condorensis Pierre in Vietnam. PhD Thesis, Georg-August University of Gottingen. Meister J, Hubaishan MA, Killian N, Oberprieler C (2005) Chloroplast DNA variation in the shrub Justicia areysiana (Acanthaceae) endemic to the monsoon affected coastal mountains of the southern Arabian Peninsula. Botanical Journal of the Linean Society 148: 437-444.

Mohammed MU, Bonnefille R (1998) A Late Glacial to Late Holocene pollen record from a high land peat at Tamasaa, Bale mountains, South Ethiopia. Global and Plane tary change 16-17: 121-129.

Mohammed U, Dagnachew Legesse, Gasse F, Bonnefille R, Lamb HF, Leng MJ (2004) Late quaternary climate changes in the Horn of Africa. In: Past climate variability through Europe and Africa (eds Battarbee RW, Gasse F, Stickly CE. Kluwer Aca demic Publishers, Dordrecht, the Netherlands.

Mulugeta K, Ehrich D, Taberlet P, Sileshi N and Brochmann C (2007) Phylogeography and conservation genetics of a giant lobelia (Lobelia giberroa) in Ethiopian and Trop ical East African mountains. Molecular Ecology 16: 1233-1243.

Nei. M. (1978) Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89: 583-590.

Newton AC, Allnutt TR, Gillies ACM, Lowe AJ, Ennos RA (1999) Molecular phy logeography, intraspecific variation and the conservation of tree species. Trends in Ecology and Evolution 14: 140-145.

Nybom H (2004) Comparison of different nuclear DNA markers for estimating intras pecific genetic diversity in plants. Molecular Ecology 13: 1143-1155.

Pankhurst R (1969) The Traditional Taenicides of Ethiopia. Journal of the History of Medicine and Allied Sciences (XXIV): 323-334.

Petit RJ, Aguinagalde I, de Beaulieu J-L, et al. (2003) Glacial refugia: hotspots but not melting pots of genetic diversity. Science 300: 1563-1565.

Petit RJ, Duminil J, Fineschi S, Hampe A, Salvini D, Vendramin GG (2005) Comparative organization of chloroplast, mitochondrial and nuclear diversity in plant populations. Molecular Ecology' 14: 689-701.

Qian J, Luscombe NM, Gerstein M (2001) Protein family and fold occurrence in ge nomes: power-law behaviour and evolutionary model. Journal of Molecular Biology 313: 673-681.

Rendell S, Ennos RA (2002) Chloroplast DNA diversity in Calluna vulgaris (heather) populations in Europe. Molecular Ecology 11: 69-78.

Rivera-Ocasio E, Aide TM, McMillan WO (2002) Patterns of genetic diversity and biogeographical history of the tropical wetland tree, Pterocarpus officinalis (Jacq.), in the Caribbean basin. Molecular Ecology 11: 675-683.

31 References

Tang T, Zhong Y, Jian SG, Shi SH (2003) Genetic diversity of Hibiscus tiliaceus (Mal vaceae) in China assessed using AFLP markers. Annals of Botany 92: 409-414.

Tileye F, Welander M, Legesse N (2005a) Microporopagation of Hagenia abyssinica: a multipurpose tree. Plant Cell, Tissue and Organ Culture 80: 119-127.

Tileye F, Welander M, Legesse N (2005b) In vitro regeneration of Hagenia abyssinica (Bruce) J.F. Gmel.(Rosaceae) from leaf explants. Plant Cell Reports 24: 392-400.

Tileye F, Nybom H, Bartish IV, Welander M (2007a) Analysis of genetic diversity in the endangered tropical tree species Hagenia abyssinica using ISSR markers. Genetic Resources and Crop Evolution 54: 947-958.

Tileye F, Zhu L-H, Legesse N, Welander M (2007b) Regeneration and genetic transfor mation of Hagenia abyssinica (Bruce) J.F. Gmel.(Rosaceae) with rolB gene. Plant Cell, Tissue and Organ Culture 88: 277-288.

Vos P, Hogers R, Bleeker M, Reijans M, Lee Th van der, Homes M, Frijters A, Pot J, Peleman J, Kuiper M and Zabeau M (1995). AFLP: a new technique for DNA finger printing. Nucleic Acids Research 23(21): 4407-4414.

WBISPP (2004) Forest resources o f Ethiopia. Woody Biomass Inventory and Strategic Planning project. Addis Ababa.

White F (1983) The vegetation of Africa: A descriptive memoire to accompany the UNESCO/AETFAT/UNESCO vegetation maps o f Africa. UNESCO, Paris.

Young A, Boshier D, Boyle T (ed.) (2000) Forest Conservation Genetics: Principles and Practice. CABI Publishing.

Young A, Boyle T, Brown T (1996) The population genetic consequences of habitat fragmentation for plants. Trends in Ecology and Evolution 11: 413—41

32 9. Papers submitted to journals

I. Colonization history and phylogeography of Hagenia abys sinica (Bruce) J.F. Gmel in Ethiopia inferred from chlorop last microsatellite markers

Taye Bekele Ayele,*1 Oliver Gailing,* Mohammed Umer* and Reiner Finkeldey* Forest Genetics and Forest Tree Breeding, Georg-August University, Biisgenweg 2, 37077 Gottingen, Germany ^Department of Earth Sciences, Addis Ababa University, P.O.Box 1176, Addis Ababa, Ethiopia

Abstract We investigated genetic variation of 273 individuals from 25 populations of the monotypic species Hagenia abyssinica (Rosaceae) from the highlands of Ethiopia at three chloroplast microsatellite markers. The objectives were to infer the factors that shaped the genetic structure and to reconstruct the colonization history of the species. Six haplotypes that were phylogenetically grouped into two lineages were identified. Homol ogy of the three loci to the respective regions of the chloroplast genome was confirmed by sequencing. The chloroplast haplotypes found in Hagenia showed a clear pattern of congruence between their geographical distribution and genealogical relationships. A very low haplotype diversity within populations (hs = 0.079, vs = 0.058) and a very high population differentiation (G st = 0.899, N st = 0.926) was observed. Restricted gene flow through seeds, rare long-distance dispersal, contiguous range expansion and mutation shaped the genetic structure of Hagenia. Fossil pollen records suggested that Hagenia immigrated into Ethiopia from the south.

Key words: chloroplast microsatellite, colonization history, genealogical relation ships, geographical structure, Hagenia abyssinica, haplotype diversity, phylogeo graphy.

33 Paper I: Colonization history and phylogeography

Introduction

Genetic variation is structured not only by the contemporary forces of genetic exchange but also by historical patterns of relationship (e.g., Schaal et al. 1998) Phylogeographic analyses can provide insights into the historical processes responsible for restricted dis tributions of populations (Cruzan and Templeton 2000). Phylogeography characterizes population subdivision by recognizing geographical patterns of genealogical structure across the range of a species (Avise 1994), synthesizing the influence of both history and current genetic exchange (Schaal et al. 1998). Cladistic gene genealogies can form the basis of historical approaches to the study of intraspecific processes (Schaal et al. 1998, Templeton et al. 1987, Templeton 2004). Rare long-distance gene flow events potentially have great evolutionary significance (Le Corre et al. 1997, Schaal et al. 1998, Cain et al. 2000) and great biological relevance by shaping genetic variation (Ouborg et al 1999). Rare dispersal events produce fragmented advancing fronts establishing new populations as a result of dispersal from pioneer populations, as well as from populations that are part of the continuous distribution (Cruzan and Templeton 2000). Cain and co-workers (2000) argued that rare events can control the rate of population spread and that only dispersal via seed directly affects colonization of new populations. For plant populations that have passed through recent episodes of range expansion, long-distance dispersal events are probably the most important factors of spatial genetic structuring at maternally inherited genes at small or medium geographic scales (Le Corre et al. 1997). In the simulation- based study on colonization dynamics of maternally inherited loci in oak, Le Corre and co-workers (1997) demonstrated that stratified dispersal was far more rapid than pure dif fusion, even if long-distance dispersals were very rare events. They also argued that long distance dispersal events influenced the genetic differentiation of populations, leaving a genetic signature that is likely to persist for long periods. This paper demonstrates the significance of such rare events, among others, in shaping the genetic structure of the monotypic species Hagenia abyssinica.

34 Paper I: Colonization history andphylogeography

The chloroplast DNA has been widely used in the investigations of genetic structure (e.g. Meister et al. 2005; Parducci et al. 2001), phylogeography (e.g., Butaud et al. 2005; Meister et al. 2005; King and Ferris 1998; Rendel and Ennos 2002) and colonization his tory (e.g., Cavers et al. 2003, Petit et al. 2002, 2003; Heuertz et al. 2004) of tree species. We employed chloroplast microsatellite markers (Weising and Gardner 1999) to investi gate genetic structure, phylogeography and colonization routes of Hagenia abyssinica within its natural range in Ethiopia. The mode of inheritance of the chloroplast genome of Hagenia abyssinica has not been determined, but it is most likely maternally inherited as in the majority of angiosperms (Harris and Ingram 1991; Birky 1995; Finkeldey and Hat- temer 2007). We also examined the available fossil pollen records in order to determine the colonization history of Hagenia in Africa. Analysis of fossil pollen helps to recon struct past vegetation history, demographic history and dynamics of ecosystems (Darby- shire et al. 2003; Lamb 2001; Mohammed et al. 2004; Olago et al. 1999).

Hagenia is a wind-pollinated and wind-dispersed broad-leaved dioecious tree species that belongs to a monotypic genus in the Rosaceae family (Hedeberg 1989; Legesse 1995). It is confined to Africa and its ecological range stretches from Ethiopia in the North to Zimbabwe in the South (Hedeberg, 1989, http:Avww.worldagroforestry.org). Fossil pol len records indicated that Hagenia immigrated into Ethiopia from the south during the late Pleistocene and became abundant in the southern regions of Ethiopia about 2500 years Before Present (BP). At present, the extant Hagenia populations throughout the country are situated at higher altitudes, often in wetter depressions

The tree has remarkably diversified economic and ecological values (Azene et al. 1993; Berhanu et al. 1999; Dawit and Ahadu 1993; Jansen 1981; Hedeberg 1989). Hagenia has been logged heavily and selectively and it is one of the endangered tree species in Ethi opia (Legesse 1995).

Genetic inventories of Hagenia abyssinica are rare and restricted only to some parts of the species’ distribution range. Kumilign (2005) and Tileye (2007) investigated the ge netic diversity of a few populations of H. abyssinica using anonymous RAPD and ISSR

35 r

Paper I: Colonization history and phylogeography

markers, respectively. The present investigation using cpDNA covered the whole range of the species in Ethiopia and is the first of its kind. We predicted that 1) due to limited seed dispersal and possibly rare long-distance seed dispersal, there is a strong differentia tion among populations but low variation within populations, 2) populations show geo graphic structuring primarily induced by mutation and isolation by distance, 3) based on the existing fossil pollen records, Hagenia immigrated into Ethiopia from the south.

Two main questions are addressed: 1) which factors shaped the maternally inherited ge netic variation of Hagenia in Ethiopia? 2) Is there a congruence between molecular data and palynological evidences to infer the relationships among genealogical lineages and migration routes of the species?

Materials and methods

Sampling and DNA Extraction

Twenty two natural and three planted populations were sampled from all regions where Hagenia is known to grow in Ethiopia. These populations represent most of the extant distribution of the species in the country. The distribution of the populations stretches from 05°51'N (Hagere Mariam) in the south to 13°11 N (Debark Mariam) in the north, and from 35°42'E (Wonbera) in the west to 40°14'E (Dindin) in the east (Table 1; Fig. 3). The distances between populations range from 21 to 806 km. The populations are lo cated within an altitudinal range of 2200 m a.s.l. at Bonga to 3200 m a.s.l. at Wofwasha, and temperatures range from an absolute minimum of -1°C at Dinsho to a maximum of 33.5 °C at Kosso Ber populations. The nearest meteorological stations are situated at lower altitudes than Hagenia populations in most of the cases, and therefore, higher rain fall and lower temperatures are expected than those shown in Table 1.

Young leaves were collected and partially desiccated in paper bags before drying with silica gel and stored at room temperature before DNA extraction. Genomic DNA was iso-

36 Paper I: Colonization history andphylogeography lated from leaves following the DNeasy 96 kit protocol of QiagerT (Hilden, Germany). In an initial test, DNA was isolated from dried leaves of different sizes. A size of 1cm2 (about 20 mg) gave the best results with regard to DNA quantity and quality and was used for all samples.

PCR amplification and genotyping

The ten pairs of consensus chloroplast microsatellite primers (CCMPs) (Weising and Gardner 1999) were tested on 3 samples from 3 geographically separated populations that are far from each other. Seven of them gave amplification products, three of which (CCMP2, CCMP6 and CCMP10) were found to be polymorphic, and were used to screen 273 samples (9-12 individuals from each population). Additional 144 samples were ana lysed to study spatial genetic structure in four polymorphic natural populations. DNA was diluted (1:10) prior to PCR amplification. PCR reactions were performed in a Peltier Thermal Cycler PTC-200 (MJ Research' ), with a volume of 16 pi reaction mixture con taining 2 pi HPLC H2O, 8 pi hot star master mix (containing lOmM Tris-HCL (pH 9.0), 1.5 mM MgCh, 50mM KC1, 0.2 mM each of dNTPs, 0.8U Taq DNA polymerase) (Qiagen*, Hilden, Germany), 2 pi of each forward and reverse primer (5pmol/pl) and 2 pi DNA (about 10 ng). The forward primer was labelled with the fluorescent dyes 6-FAM or HEX. The PCR profile for CCMP2 and CCMP10 was 15 min. initial denaturation at 95 °C, followed by 35 cycles of 1 min. denaturation at 94 °C, 1 min. annealing at 50 °C and 1 min. extension at 72 °C, with a final extension of 10 min. at 72 “’C. The PCR profile for CCMP6 differed in the annealing temperature (52.5 °C). Aliquots of the amplification products were diluted prior to clcctrophoretic separation on the ABI 3100 Genetic Ana lyser (Applied BiosystemsR) depending on the intensity of the bands observed after aga rose gel electrophoresis. Two pi diluted (multiplexed in most cases) PCR product were denaturated for 2 minutes at 90°C with 12 pi HiDi formamide (Applied Biosystems®) containing -0.02 pi internal size standard (GS ROX 500, Applied Biosystems®) before loading on the ABI Genetic Analyser 3100 (Applied Biosystems8) for separation.

37 Paper I: Colonization history and phylogeography

Sequencing

Comparative sequencing of 18 fragments from the three chloroplast loci was performed to confirm the amplified regions and to determine the molecular basis for size variation. The amplification products were purified using the QIAquick Gel Extraction kit (Qiagen®, Germany) following the manufacturer’s specifications. We employed direct sequencing for a locus having relatively larger fragment sizes, CCMP2 (224-235 bp). Cloning was performed for the two loci with smaller fragment sizes, CCMP6 (140-142 bp) and CCMP10 (96-97 bp), using the pBSKS vector and X Bluel competitive cells with the TA cloning method (Invitrogen*). Sequencing followed the dideoxy-chain termina tion method (Sanger et al., 1977) Sequencing reaction of 10 pi total volume containing 1 pi Big Dye (BD vers. 3.1), 1.5 pi sequencing buffer (SB 3.1), 4.8 pi HPLC H20, 0.7 pi forward or reverse primer (5pmol/pl), 2 pi purified DNA (about 10 ng) was used. Since no sequences of CCMP2 from other species of the family Rosaceae were available in ex isting databases, three out-group species from the Rosaceae family that were available in the Botanic Garden of the Georg-August University Goettingen, Germany, were also se quenced for comparison. The sequence data have been stored in the EMBL Nucleotide Sequence Database (http://www.ebi.ac.uk/embl/) with the accession numbers FM174367- 75 and FM 174387 for locus CCMP2 (10 sequences), FM174376-80 for locus CCMP6 (5 sequences), FM174381-83 for locus CCMP10 (3 sequences), and FM1743784-86 for the locus ccmp2 of the out-group species (3 sequences).

38 Paper I: Colonization history and phylogeography

Table I Site characteristics of 22 natural and 3 planted populations of Hagenia sampled from Ethiopia. The planted populations are labelled as plantation.

Population Code Latitude Longitude altitude ARF M in T M ax T M ax M ax n (m asl) (m l) Ht DBH (m ) (cm)

Debark-Mariam DK 3013 1270 8.8 19.7 15 144 11 m r 37 °5 7 '

D ebark- DKP 13°12' 38°01' 3005 1270 8.8 19.7 na na 11 Plantation Kimir-Dingay KDP 1 1°48' 3 8°14' 1350 9.2 21.9 na na 11 plantation W oldiya S e’at WD 1 1°55' 39°24' 3112 908 na na 15 125 11 M ichael K osso B er K.B 10°59' 36°54' 2702 2381 12.9 27.4 17 45 11

D enkoro DR 10°52' 38°47' 3061 896 10.9 2 1 .8 2 0 196 11

Wonbera WB 10°34' 35°42' 2428 1622 na na 18 112.5 11

W o f w asha WW 09 °4 5 ' 3 9 0 4 4 - 3159 941 6.1 19.9 15 122 11

C hilim o CM 09°05' 38°10' 2805 1114 11.5 25.8 35 118 12

Dindin DN 08°36’ 40°14' 2410 989 12.7 28,0 26 156 11

Zequala Abo ZQ 08°32' 38°50’ 2856 1215 na na 23 234 10

B oterbecho BB 08°24' 37°15' 2772 1666 5.7 23.6 28 126 9

C hilalo CL 07 °5 6 ' 39°ir 2815 796 9.8 23,0 15 140 11

Sigmo plantation SMP 07 °5 5 ' 36°10' 2300 1837 11.4 2 1 .6 na na 11

Sigm o SM 07 °4 6 ' 36°05' 2651 1837 11.4 2 1 .6 25 152 11

M unesa M S 0 7 °2 5 ’ 38°53' 2459 1028 10.1 24.3 20 84.5 11

B onga BG 07°17' 36°22' 2238 2217 11.9 26.6 18 93 11

Kofele K.L 07°ir 38°52' 2757 1305 7.7 20.1 26 214 11

Dinsho DO 07°05' 3 9 0 4 7 - 3117 1213 3.4 2 0 .8 19 153 11

Doddola-Serofta DS 06°52' 39°02' 2700 1074 6.7 24.3 23 242 11

D oddola- DD 06 °5 2 ' 39°14' 3039 1074 6.7 24.3 25 2 1 0 11 D achosa Rira RR 06 °4 5 ' 39°43’ 2725 736 na na 23 150 11

Bore BR 06 °1 7 ' 3 3 0 3 9 - 2631 1526 8.3 18.8 24 107 12

Uraga UR 06°08' 38°33' 2508 1228 8.3 18.8 19 96 11

HagereMariam HM 0 5 °5 1 ' 38°17' 2443 1228 12.3 23,0 18 114 10 m asl= meters above sea level; ARF = Mean Annual Rainfall; ml = millilitres; Min T = Mean Minimum temperature; Max T = Mean Maximum temperature; Max Ht (m) = Maximum height in meters; Max DBH (cm) = Maximum diame ter at breast height in centimeters; n= no. of samples; na = not available. Source of climatic data; National Meteorologi cal Agency Service (Ethiopia)

39 Paper I: Colonization history and phylogeography

Data analysis

Amplification products were aligned with the internal size standard using GENESCAN 3.7, and fragments were scored with GENOTYPER 3.7 (Applied Biosystems®). Poly morphisms in fragment size were identified as different length variants that were com bined to define haplotypes. Genetic diversity (hs, hT. vs, vT) and differentiation among populations (G s t , N st ) was computed by PermutcpSSR (available at http://www.pierroton.inra.fr/genetics/labo/Software/PermutCpSSR/index.htmn as de scribed by Pons and Petit (1995; 1996). Distribution of genetic diversity within and among populations was estimated by an analysis of molecular variance (AMOVA) using ARLEQUIN Version 3.0 (Excoffier et al. 2005; available at http: //cmpg. unibe. ch/software/arlequin3).

Sequences were analysed with the sequence analysing software 3.7 (Applied Biosys tems®), edited by the program BIOEDIT (Hall 1999) and aligned with Clustal W applica tion (Thompson, et al. 1994; available at http://www.ebi.ac.uk/clustalw/).

A statistical parsimony network of haplotypes was constructed with the help of a program TCS Version 1.21 (Clement et al. 2000) from DNA sequence data. Large gaps in a se quence due to an indel (insertion/deletion) are coded as a single mutation to avoid theo retical intermediate haplotypes that are created by the program, which interprets each gap as independent mutation event. The sequence data also confirmed that the larger indel was the result of a single mutation event. The TCS program was also used to compute the out-group weights of haplotypes. A nested clade phylogeographic analysis (NCPA) of the spatial distribution of the genetic variation was performed by the program GEODIS (Po sada et al. 2000). Nested clades were plotted manually on the haplotype network based on the algorithms defined by Templeton et al. (1987). The interpretation of statistically sig nificant patterns of distribution was made following the inference key described in Tem pleton (2004).

40 Paper I: Colonization history and phylogeography

Results

Genetic diversity and differentiation

We found 3 alleles in locus CCMP2, 3 alleles in locus CCMP6 and 2 alleles in locus CCMP10 (Tables 2 and 3). The analyses o f within population diversity (hs), total diver sity (ht) and differentiation ( G s t ) yielded 0.079, 0.787 and 0.899, respectively, under the assumption of unordered haplotypes (Pons and Petit 1996). The corresponding values for within population diversity (vs), total diversity (vT) and differentiation ( N s t ) with ordered haplotypes (Pons and Petit 1996), taking genetic distances among haplotypes into ac count, were 0.058, 0.787 and 0.926, respectively. An analysis of molecular variance (AMOVA) showed that 92.3% of the total genetic diversity is represented among popula tions. A test of spatial genetic structure in the four polymorphic natural populations did not show any family or spatial genetic structure, indicating effective seed dispersal by wind at the local level. This is not unexpected for species with very light wind-dispersed seeds. The additional 144 individuals analysed for spatial genetic structure showed simi lar haplotype composition and frequencies, as the sample that was analysed to study ge netic diversity (Table 4). No additional haplotypes were found due to increased sample size.

Table 2 Description of chloroplast microsatellites in Hagenia

Repeat motif Fragment size (bp) source of Gene Forward and reverse primer other other location in variation locus sequences (5' - 3')* Hagenia plants* Hagenia plants* genome*

224, Indel, GATCCCGGAGGTAATCCTG 234, 158-234 micro CCMP2 ATCGTACCGAGGGTTCGAAT (A)9 (A),, 235 5' to trnS satellite 140, ORF 77- 141, ORF 82, micro CGATGCAT ATGT AGAAAGCC satellite CCMP6 CATTACGTGCGACTATCTCC (T)v (T)5C(T)17 142 93- 103 intergenic

TTTTTTTTTAGTGAACGTGTCA 91->300 rpl2-rpsl9, micro CCMP10 TTCGTCGDCGTAGTAAATAG (A),2 (T),4 96, 97 intergenic satellite

*Weising and Gardner, 1999, indel = Insertion/deletion (in the flanking region)

41 Paper I: Colonization history and phylogeography

Table 3 Description of Hagenia haplotypes detected by fragment analysis in three chlo roplast DNA loci

Haplotype CCMP2 CCMP6 CCMP10 Relative np nfp frequency

Hi 224 140 97 0.34 10 7 H2 224 141 97 0.25 8 5 H3 224 142 97 0.05 3 0 H4 234 140 97 0.21 6 5 H5 234 140 96 0.04 1 1 H6 235 140 97 0.11 3 2

np= No. of populations possessing the haplotypes: nfp = No. of populations fixed on one type

Table 4 Number of observations per haplotype from different sample sizes of polymor phic populations

Populations Chilimo (CM) Kofele (KL) Bore (BR) Uraga (UR) Sample size 12 47 11 46 12 47 11 50 Haplotypes HI 10 39 6 19 1 3 H2 4 26 9 38 1 2 H3 1 1 2 6 10 48 H4 2 8

Results o f sequencing

Multiple sequence alignments of loci CCMP2, CCMP6 and CCMP10 are shown in Fig. 1. Sequencing confirmed homology of the three loci to the respective regions of the chloroplast genome. The observed variations were due to variable numbers of poly (A) or poly (T) repeats and a large indel of 10 bp at position 100 bp in the flanking region of the locus CCMP2. In total, there were 4 variable sites (3 short indels in the microsatellites and a large indel in the flanking region). A 10-bp segment is preceded by an identical sequence in the flanking region of the four genotypes in locus CCMP2 (underlined in Fig. 1). The duplication can be explained by a strand slippage during cpDNA replication in a Paper I: Colonization history andphylogeography single mutational event (e.g., Wolfson et al., 1991). The sequences of the three out-group species (Rubus fruticosus, Rubus idaeus and Rosa canina) also showed duplication events of different segment in similar region (Fig. 1).

Phylogeography

Six haplotypes (H1-H6) were identified from the combination of the three loci as de tected by fragment analysis (Table 3). A fully resolved statistical parsimony network of the chloroplast haplotypes of Hagenia, which is reconstructed from DNA sequences (Fig. 2), demonstrates the relationship among the different haplotypes and the minimum num ber of evolutionary events separating them. The third frequent haplotype, H4 (represented in 21% of the individuals), has a larger out-group weight (0.35) than the most frequent haplotype HI (represented in 34% of the individuals) with an out-group weight of 0.26. Out-group weight is a relative weight of haplotypes based on mutational steps and is strongly correlated with actual age (of a haplotype) and thus is a much better indicator of haplotype age than is the haplotype frequency (Castelloe and Templeton 1994). All of the haplotypes are separated by a single mutation step. HI is separated from H4 by a deletion of 10 nucleotides in locus CCMP2.

The observed NSt value is significantly higher than the G st value at p<0.01 (none of the pennutated G st values was higher than the observed N st value), indicating geographical clustering of related haplotypes. Three nested clades were evident from the haplotype network (Fig. 2) and the chi-square (x2) statistic revealed a significant association (p<0.0001) between genealogical and geographic distributions in all of the clades (Table 5). Restricted gene flow was inferred for the haplotypes nested in clade 1-1 (but with some long distance dispersal events over intermediate areas) and in clade 1-2 (with isola tion by distance), while contiguous range expansion was deduced from the total cladogram (Table 5).

43 Paper I: Colonization history and phylogeography

Table 5 Interpretation of the results of the nested clade phylogeographic analysis (NCPA)

Clade X2 P Chain of inference* Inferred demographic events§

1-1 276.5762 0.0000 1 -2-3-5-6-7-8-YES restricted gene flow/dispersal (HI, H2, H3) but with some long distance dispersal over intermediate areas 1-2 136.0000 0.0000 1-2-3-4-N0 restricted gene flow with iso (H4, H5, H6) lation by distance Total cladog- 231.7921 0.0000 1-2-11-12-NO ram contiguous range expansion

*Refers to inference key numbers (YES/NO refers to the answers to the respective last keys), §Interpreted by inference key (Templeton 2004)

Haplotype HI is widely distributed in central Ethiopia, while H2 is common in southern regions (Fig. 3). Haplotype H4 has the longest geographic distribution in south-north di rection stretching from the southwest to the northern regions. The northern population DK was established most likely by single long-distance dispersal event. A recent muta tion at locus CCMP10 resulted in the rarest haplotype (H5) that is restricted to only one population (Wonbera) in the west (fixed on one type), distinguishing it from other popu lations. Haplotype H6 is restricted to the central-northern region, while H3 has only a rare occurrence in the southern region, and is always in association with H2 and/or HI. The domination of the population Uraga (UR) by haplotype H3 as contrasting to the neighbor ing populations (H2) in the south was caused by a single mutational event in locus CCMP6.

The three loci exhibited different impacts on haplotypic variation in the different regions of the country. Locus CCMP6 accounted for the variation in central and southern Ethi opia, while CCMP2 was responsible for the variation in southwestern and northern re gions. Locus CCMP10 caused the variation in west Ethiopia due to the prevalence of a private allele. Eighty percent of the populations are fixed on one haplotype, while the re maining populations share two to three haplotypes. Two populations (KL and BR) con tained three similar haplotypes at different frequencies while three populations (UR, CM

44 Paper I: Colonization history and phylogeography and KDP) possessed two different haplotypes each. Two of the planted populations in cluded in this study (DKP and SMP) showed identical haplotypes as their respective par ent populations (DK and SM) based on the record obtained from the District Office of Agriculture (unpublished data). The source population of a third plantation, (KDP) was not confirmed, but it exhibited a combination of haplotypes from two neighboring popu lations (HI and H6), suggesting that seeds were obtained from the adjacent populations or alternatively, they were procured from the national seed center.

45 ae I Clnzto hsoy n phylogeography and history Colonization I:Paper

30 90 100 110 120 130 1 1 50

DK5 2 3 4 AAG TTTTTTTTATTTATTTA TTTAGTTAATTTTAGTTAATTAAAAAAAAA TATTAATAA TTTAAAGAAGTGG SM7 2 3 4 AAG TTTTTTTTATTTATTTA TTTAGTIAATTTTAGTTAATTAAAAAAAAA TATTAATAA tttaaagaagtgg WB5 2 3 4 AAG TTTTTTTTATTTATTTA TTTAGTTAATTTTAGTTAATTAAAAAAAAA TATTAATAA TTTAAAGAAGTGG DR7 2 3 5 AAG TTTTTTTTATTTATTTA TTTAGTTAATTTTAGTTAATTAAAAAAAAAATATTAATAA TTTAAAGAAGTGG KB5 2 2 4 AAG TTTTTTTTATTTATTTA TTAGTTAATTAAAAAAAAA TATTAATAA TTTAAAGAAGTGG CM30 2 2 4 AAG TTTTTTTTATTTATTTA TTAGTTAATTAAAAAAAAA TATTAATAA TTTAAAGAAGTGG DN7 2 2 4 AAG TTTTTTTTATTTATTTA TTAGTTAATTAAAAAAAAA TATTAAAAA TTTAAAGAAGTGG DD7 2 2 4 AAG TTTTTTTTATTTATTTA TTAGTTAATTAAAAAAAAA TATTAAAAA TTTAAAGAAGTGG KB7 2 2 4 AAG TTTTTTTTATTTATTTA TTAGTTAATTAAAAAAAAA TATTAATAA TTTAAAGAAGTGG UR5 2 2 4 AAG TTTTTTTTATTTATTTA TTAGTTAATTAAAAAAAAA TATTAAAAA TTTAAAGAAGTGG R u f GAGCTCTTTTTTTTATTTAATTA TTTAATTAA TAGTT AAAAATGAA TAGTTAAAGAAGTTTAAGGATGNGG R u i GAGCTTTTTTTTT ATTTAATTAA TAGTT AAAAATGA ATAGTTAAAGAAGTTTAAGGATGTGT R oc GAGCTCTTTTTT ATTATATTATTTTTATTATA TAGTT AAAAATTATATAGTTAAAATAGTT AAAGAAGTGG

(b) 8 0

UR5 142 CTACCTTTTAGTTTTATATAATATATATAGTATTTTTTTTTCTATGGATTATGGATATAGTATTTATTAACGTATTTCTT UR16 141 CTACCTTTTAGTTTTATATAATATATATAGTATTTTTTTT CTATGGATTATGGATATAGTATTTATTAACGTATTTCTT DO5 141 CTACCTTTTAGTTTTATATAATATATATAGTATTTTTTTT ctatggattatggatatagtatttattaacgtatttctt KB7 1 4 0 ctaccttttagttttatataatatatatagtattttttt ctatggattatggatatagtatttattaacgtatttctt KB5 14 0 ctaccttttagttttatataatatatatagtattttttt ctatggattatggatatagtatttattaacgtatttctt

20 30 40 50 60 70 80 90

WB4 96 gtagtaaataggcgagaaaatagaatttgtttcttcctcttaaaaaaaaaaa taggagtaattaattgtgacacgttca WB6 9 6 gtagtaaataggcgagaaaatagaatttgtttcttcctcttaaaaaaaaaaa taggagtaattaattgtgacacgttca KB 5 9 7 gtagtaaataggcgagaaaatagaatttgtttcttcctcttaaaaaaaaaaaataggagtaattaattgtgacacgttca

Fig. 1 Sections of sequences of three Hagenia chloroplast microsatellite loci (a= CCMP2 aligned with Rubus fruticosus, Rubus idaeus and Rosa canina, b= CCMP6, c= CCMP10). Duplications in locus CCMP2 are underlined. Microsatellite repeats are shown in bold. Gaps indicate deletions of nucleotides, g = genotype; f = fragment size. Paper I: Colonization history and phylogeography

Fig. 2 Statistical parsimony network showing nested clades and relatedness among haplotypes (H1-H6) of Hagenia abyssinica at three chloroplast loci. Sizes of circles are proportional to their respective out-group weights. Thick bar indicates indel of 10 bp in a single mutation event; thin bar indicates indel of 1 bp.

47 Paper I; Colonization history and phylogeography

Discussion

Genetic diversity and differentiation

A very low genetic and haplotype diversity within populations (hs = 0.079, vs = 0.058, respec

tively) and a very high population differentiation (GSt = 0. 899, NSt = 0. 926) proved a marked genetic separation of the populations. This result supports our first prediction that there is a strong differentiation among populations, but low variation within populations due to limited seed dispersal and possibly rare long-distance seed dispersal events. The population differentia tion is much higher in the chloroplast genome than in the nuclear genome of Hagenia abyssinica,

as revealed by ISSR (G st = 0.25; Tileye et al. 2007) and AFLP markers (G s t = 0.15, Taye et al. submitted).

Likewise, Rendell and Ennos (2002) found a population differentiation that was 10-fold higher in the chloroplast genome of Calluna vulgaris (L.) Hull (Ericaceae), than in the nuclear genome. In general, maternally inherited genomes experienced considerably more subdivision (mean Gst value o f-0.64) than biparentally inherited genomes (mean G s t value of -0.18) of angiosperm

species (reviewed by Petit et al. 2005). The coefficient of population differentiation (G s t ) in Ha genia is higher than or comparable to G s t values recorded for other species with heterogeneous mode of seed dispersal, including wind dispersal, investigated by chloroplast markers (Newton et al. 1999, Petit et al. 2003).

The high level of genetic variation among populations of Hagenia suggested a restricted migra tion of seeds among regions, which is also reflected in the observed geographic structuring of haplotypes. The demographic history of the species and/or existence of natural barriers (moun tains, valleys and long distances) to seed dispersal might account for the strong phylogeographic pattern. Young and Boyle (2000) reported that for wind-pollinated and dispersed species, the pat tern of gene flow and genetic structure is a function of interfragment distance.

48 Paper I: Colonization history and phylogeography

Fig. 3 Geographic distribution of Hagenia chloroplast haplotypes in Ethiopia. Dotted enclosure shows li neage I; dashed enclosure shows lineage II. Grey dashed lines indicate approximate position of the Great Rift Valley. The inset pie chart shows the relative frequency distribution of the haplotypes. Two letters de signate populations as in Table 1; H1-H6 indicates haplotypes as in table 3. The arrows indicate the putative colonization route of the species. Source map: Assefa G (unpublished).

Though Hagenia is a montane species, its migration is not necessarily along the mountains as evidenced by haplotype HI that is distributed at both sides of the Great Rift Valley (see Fig 3), most likely due to long-distance seed dispersal as intermediate populations are missing.

49 Paper I: Colonization history and phylogeography

Phylogeographical andpalynological interpretation

The geographic distribution of haplotypes (Fig. 3) and their genealogical relationships (Fig. 2) observed in Hagenia demonstrated a marked phylogeographical structure as a result of highly restricted gene flow via seeds. Such patterns arise when scattering is reduced because the novel mutations remain localized within the geographical context of their origins (e.g. Butaud et al. 2005). Both the Gst-Nst test and the NCPA detected a very strong association between genea logical and geographic distributions. The NCPA inferred that restricted gene flow associated with contiguous range expansion shaped the genetic structure of Hagenia. This result allows us to accept the second prediction that populations show geographic structuring primarily induced by isolation by distance, coupled with local mutation events. Two distinct lineages that were separated by an indel of 10 nucleotides in locus CCMP2 are evident from the cladogram (see Fig. 2). The first lineage constitutes haplotype H4 and its derived haplotypes H5 and H6 that are distributed in the south-western and northern regions (referred hereafter as lineage I), while the second lineage embodies HI and its derived haplotypes H2 and H3 in central and southern re gions (lineage II). Such a non-random distribution of haplotypes asserts our prediction on the existence of phylogeographic pattern in Hagenia.

In light of the palynological records discussed at the end of this section, there are two possible scenarios for the immigration of Hagenia into Ethiopia. The first scenario suggests that the line age I of Hagenia colonized Ethiopia first through the southwest mountains (population Bonga, BG) that is situated to the west of the Great Rift Valley whereas the second scenario suggests that lineage II of Hagenia colonized Ethiopia first through the south mountains (population Hagere Mariam, HM) situated east of the Great Rift Valley. However, our data supports the first scenario. The cladogram demonstrated that haplotype H4 is the most probable ancient haplotype that served as a root for the rest of the haplotypes because of its higher out-group weight. Castel- loe and Templeton (1994) argued that the most ancient haplotype should be located at the center of the gene tree and be geographically widespread, whereas the most recent haplotypes should be at the tips of the gene tree and be localized geographically. This ascertains the postulation that haplotype H4 is the most ancient haplotype (followed by HI), whereas H2, H3, H5 and H6 are located at the tips of the gene tree and are highly localized geographically. This observation sug-

50 Paper I: Colonization history and phylogeography gests that Hagenia was first introduced most likely to the Mountains of the southwest Ethiopia (population Bonga, BG) from southern African regions and expanded and diversified in the cen tral, southern and northern regions of the country. Our results demonstrated that lineage I most likely gave rise to lineage II due to a deletion of 10 nucleotides in a single mutational event. This mutational event was quite recent assuming the recent colonization of Ethiopia by Hagenia. All the southern haplotypes were most likely derived from a single seed parent with haplotype H4 and expanded to the central regions and diversified into the southern regions. In general, our re sults confirm that colonization took place only from the south. The inferred colonization routes and putative long-distance dispersal event are shown in Fig. 3.

The long gap observed between the northern and southern populations with the haplotype H4 led to three postulations: 1) the populations that are situated between these two regions diversified to other haplotypes due to mutation (e.g., population WB diversified to H5 and population DR di versified to H6, both based on a single mutation event), 2) some populations containing the same haplotype might be lost due to anthropogenic activities, 3) natural or human mediated coloniza tion events in terms of long-distance seed dispersal or purposeful seed transfer account for this disjunct haplotype distribution. The postulations, however, are not exclusive. Haplotype HI is also widely distributed, and such a widespread distribution of individual haplotypes indicates rapid range expansion (Schaal et al. 1998) and the significant role of rare events, particularly long-distance seed dispersal and mutation, in shaping the genetic diversity in Hagenia. Further more, the patchy structure of the haplotypes of Hagenia, in general, is a result of rare long distance dispersal of seeds during colonization, each patch resulting from a founding event beyond the colonizing front.

Though complete coverage is unavailable from Africa, the palynological data obtained from fos sil pollen stratigraphy often sites in Africa (Table 5) suggested that the post-glacial colonization of Hagenia followed a northward route with the oldest available record from Burundi (ca. 34,000 l4C yrs BP). Major expansion of Hagenia took place around 1 1,500 14C yrs BP in Burundi (Bon- nefille et al. 1995), whereas its major expansion in Bale Mountains (southern Ethiopia) was after 2500 cal yr BP (Mohammed et al. 2004; Mohammed and Bonnefille 1998). Bonnefille and Mo hammed (1994) also reported that Hagenia expanded after 590 yr BP in Arsi Mountains (central Ethiopia). In general, the signal of Hagenia in the pollen records in Ethiopia was quite high in

51 Paper I: Colonization history and phylogeography the late Holocene epoch and the palynological evidences in general suggested a northward mi gration route also within Ethiopia. The fossil pollen records support our third prediction that Hagenia immigrated into Ethiopia northward from southern African regions. The examination of the same pollen diagrams that are described above also indicated a northward colonization of some other tree species such as Podocarpus falcatus, Juniperus procera and Olea species in Af rica. The palynological data also showed that the fossil pollen accumulation of Hagenia abys sinica has been alarmingly declining through time in the African countries other than Ethiopia, suggesting a sequential reduction in the size of the populations.

Table 5 Late Pleistocene to late Holocene ice ages fossil pollen records of Hagenia from ten sites in Africa

Altitude Approximate age Maximum Site, Country, References of core at first appearance record o f (m.a.s.l.) pollen (%)

Rusaka, Burundi’ 2070 34,000 l4C yrs BP 20 Mount Kenya4 2350 >33,350 14C yrs BP 40 Lake Albert, Uganda1 619 12,000 14C yrs BP <2.5 Lake Turkana, Kenya2 375 2,200 14C yrs BP <1 Garba Guracha (Bale Mountains, Ethiopia)5 3950 17,000 cal yrs BP 10 Tamsaa (Bale Mountains, Ethiopia)2,6 3000 15,470 cal yrs BP: 45 Lake Tilo (Southern Rift Valley, Ethiopia)7'8 1545 8000 l4C yrs BP <2.5 Dega Sala (Arsi mountain, Ethiopia)9 3600 1850 l4C yrs BP ca.7 Lake Langeno, Ethiopia10 1583 2370 l4C yrs BP <5 Lake Hardibo (Wello, Ethiopia)11 2150 2500 14C yrs BP <2.5

The sites are arranged from south to north, generally showing a decreasing trend in ages of fossil pollen. Low pollen grain percentages in some sites may suggest that pollen was transported from other forests. Sources: ‘Beuning et al. (1997), ^Mohammed et al. (1996), Bonnefille et al. (1995), 401ago et al. (1999). 5Umer et al. (in press), 6Mohammed and Bonnefille (1998), 7Lamb (2001), 8Lamb et al. (2004), ’Bonnefille and Mohammed (1994), 10 Mo hammed and Bonnefille (1991), "Darbyshire et al. (2003).

It remains uncertain from the pollen data whether there was a forest refugium during the gla cial/late glacial period at lower altitudes (2000-2500m) in southern Ethiopia. There is a lack of information from such sites for the early Holocene.

Conclusions

The joint interpretation of genealogical relationships among the chloroplast haplotypes and the fossil pollen evidences allowed us to accept the hypothesis predicting immigration of Hagenia

52 Paper I: Colonization history and phylogeography into Ethiopia from the south. Based on the cpDNA data, the migration of Hagenia into Ethiopia occurred only once, but the exact dating was not possible. There was no indication of past frag mentation of Hagenia populations from our results, pointing to the effect of random long distance seed dispersal. It is most likely that populations were established from few parent seed trees. Given the mountainous topography of the country that is intermittently dissected by wide valleys, Hagenia did not have a continuous distribution. The cpDNA assay detected sufficient variation for a phylogeographic study of Hagenia abyssinica in Ethiopia. A remarkable subdivi sion of cpDNA diversity in the species was found, as indicated by a high level of genetic differ entiation. The chloroplast haplotypes of Hagenia abyssinica demonstrated a pattern of isolation- by- distance. Unlike most of the wind-dispersed tree species, the chloroplast haplotypes found in Hagenia showed a clear pattern of congruence between their geographical distribution and ge nealogical relationships, allowing us to accept the prediction on geographic structuring. Analysis of cpDNA types and palynological inventories, including all countries where the species is known to grow, would fully resolve the genealogical relationships and help to identify the glacial refugia of Hagenia in Africa. The analysis of pollen records from different sites and altitudes in Ethiopia where Hagenia is growing would help to fully understand the colonization route of the species within the country.

Acknowledgements

This work is supported by the German Federal Ministry of Economic Cooperation and Devel opment (BMZ) through the German Technical Cooperation (gtz) as a component of a project “Support to the Forest Genetic Resources Conservation Project” of the Ethiopian Institute of Biodiversity Conservation (IBC). The German Academic Exchange Service (DAAD) executed the grant as a PhD project of the first author. The National Meteorological Service Agency of Ethiopia provided climatic data free of charge. We are indebted to Oleksandra Dolynska and Thomas Seliger for technical assistance in the laboratory.

53 Paper I: Colonization history and phylogeography

References

Avise JC (2004) Molecular Markers, Natural History and Evolution. 2nd edn, Sinauer Associates Inc, Sunderland, USA Azene BT, Bimie A, Tegnas Bo (1993) Useful Trees and Shrubs for Ethiopia: Identification, Propagation and Management for Agricultural and Pastoral Communities. Technical Hand book No. 5, Regional Conservation Unit, Swedish International Development Authority (SI- DA) Bawa KS, Krugman SL (1990) Reproductive biology and genetics of tropical trees in relation to conservation and management. In: Rain Forest Regeneration and Management (eds Gomez- Pampa A, Whitmore TC, Hadley M), pp. 119-136. The Parthenon Publishing Group Berhanu M Abegaz, Ngadjui BT, Merhatibeb Bezabih, Mdee LK (1999) Novel natural products from marketed plants of eastern and southern Africa. Pure Appl Chem 71: 919-926 Beuning KRM, Talbot MR, Kelts K (1997) A revised 30,000-year paleoclimatic and paleohydrologic history of Lake Albert, East Africa. Palaeogeogr palaeocl 136: 259-279 Birky CW Jr (1995) Uniparental inheritance of mitochondria and chloroplast genes: mechanisms and evolution. Proc Natl Acad Sci USA 92: 11331-11338 Bonnefille R, Mohammed MU (1994) Pollen-inferred climatic fluctuations in Ethiopia during the last 3000 years. Palaeogeogr palaeocl 109: 331-343 Bonnefille R, Riollet G, Buchet G, Icole M, Lafont R, Arnold M, Jolly D (1995) Gla cial/interglacial record from intertropical Africa, high resolution pollen and carbon data at Ru- saka, Burundi. Quaternary Sci Rev 14: 917-936 Butaud J-F, Rives F, Verhaegen D, Bouver J-M (2005) Phylogeography of Eastern Polynesian sandalwood (Santalum insulare), an endangered tree species from the pacific: a study based on chloroplast microsatellites. J Biogeogr 32: 1763-1774 Cain ML, Milligan BG, Strand AE (2000) Long-distance seed dispersal in Plant Populations. Am JB ot 87: 1217-1227 Castelloe J, Templeton AR (1994) Root probabilities for intraspecific gene trees under neutral coalescent theory. Mol Phylogenet Evol 3: 102-113 Clement M, Posada D, Grandall KA (2000) TCS: A computer program to estimate gene geneal ogies. Mol Ecol 9: 1657-1659 Cavers S, Navarro C, Lowe AJ (2003) Chloroplast DNA phylogeography reveals colonization history of a Neotropical tree, Cedrela odorata L., in Mesoamerica. Mol Ecol 12: 1451-1460 Cruzan MB, Templeton AR (2000) Paleoecology and coalescence: phylogeographic analysis of hypotheses from the fossil record. Trends Ecol Evol 15: 491-496 Darbyshire I, Lamb H, Mohammed Umer (2003) Forest clearance and regrowth innorthern Ethiopia during the last 3000 years. Holocene 13: 537-546 Dawit A, Ahadu A (1993) Medicinal Plants and Enigmatic Health Practices of Northern Ethi opia. Addis Ababa Excoffier L, Laval G, Schneider S (2005) Arlequin (version 3.0): An integrated software package for population genetics data analysis. Evol Bioinform 1: 47-50 Finkeldey R, Hattemer HH (2007) Tropical Forest Genetics. Springer-Verlag, Berlin Hall TA (1999) BIOEDIT: a user-friendly biological sequence alignment editor and analysis program for windows 95/98/NT. Nucleic Acids Symposium Series 41: 95-98

54 Paper I: Colonization history and phylogeography

Harris SA, Ingram R (1991) Chloroplast DNA and biosystematics: The effects of intraspecific diversity and plastid transmission. Taxon 40: 393-412 Hedeberg, O (1989) Rosaceae. In: Inga Hedeberg and Sue Edwards, eds. Flora of Ethiopia, Vol. 3, Pittosporaceae to Arralaceae. Addis Abeba and Asmara, Ethiopia; Uppsala, Sweden Heuertz M, Fineschi S, Anzidei MA et al. (2004) Chloroplast DNA variation and postglacial recolonization of common ash (Fraxinus excelsior L.) in Europe. Mol Ecol 13: 3437-3452 Jansen PCM (1981) Spices, condiments and medicinal plants in Ethiopia, their taxonomy and agricultural significance. Wageningen King RA, Ferris C (1998) Chloroplast DNA phylogeography of Alnus glutinosa (L.) Gaertn. Mol Ecol7,1151-1163 Kumilign A (2005) Estimation of sex-related genetic diversity of Hagenia abyssinica (Bruce) J.F. Gmel. M.Sc. thesis, Addis Ababa University Lamb HF (2001) Multi-proxy records of Holocene climate and vegetation change from Ethiopian Crater lakes. Proy Irish Acad B 101: 35-46 Lamb AL, Leng MJ, Mohammed UM, Lamb HF (2004) Holocene climate and vegetation change in the main Ethiopian Rift Valley, inferred from the composition (C/N and Si3C) of lacustrine organic matter. Quaternary Sci Rev 23: 881-891 Le Corre V, Machon N, Petit RJ, Kremer A (1997) Colonization with long-distance seed disper sal and genetic structure of maternally inherited genes in forest trees: a simulation study. Genet Res 69: 117-125 Legesse N (1995) Indigenous trees of Ethiopia: Biology, uses and propagation techniques. SLU, Reprocentralen, Umea Meister J, Hubaishan MA, Killian N, Oberprieler C (2005) Chloroplast DNA variation in the shrub Justicia areysiana (Acanthaceae) endemic to the monsoon affected coastal mountains of the southern Arabian Peninsula. Bot J Linn Soc 148: 437-444 Mohammed MU, Bonnefille R (1991). The recent history of vegetation and climate around lake Langano (Ethiopia). Paleoecol Afr22: 275-286 Mohammed MU, Bonnefille R (1998) A Late Glacial to Late Holocene pollen record from a high land peat at Tamasaa, Bale mountains, South Ethiopia. Global Planet change 16-17: 121-129 Mohammed MU, Bonnefille R, Johnson T (1996) Pollen and isotopic records of Late Holocene sediments from Lake Turkana, N. Kenya. Palaeogeogr palaeocl 1 19: 371-383 Mohammed U, Dagnachew Legesse, Gasse F, Bonnefille R, Lamb HF, Leng MJ (2004) Late quaternary climate changes in the Horn of Africa. In: Past climate variability through Europe and Africa (eds Battarbee RW, Gasse F, Stickly CE), pp. 159-177. Kluwer Academic Publish ers, Dordrecht, the Netherlands Newton AC, Allnutt TR, Gillies ACM, Lowe AJ and Ennos RA (1999) Molecular phylogeogra phy, intraspecific variation and the conservation of tree species. Trends Ecol Evol 14: 140-145 Ouborg NJ, Piquot Y, Van Groenendael JM (1999) Population genetics, molecular markers and the study of dispersal in plants. J Ecol 87: 551-568 Olago DO, Street-Perrott FA, Perrott RA, Ivanovich M, Harkness DD (1999) Late quaternary glacial-interglacial cycle of climatic and environmental change on Mount Kenya. J Afr Earth Sci 29:593-618 Parducci L,- Szmidt AE,- Madaghiele A, Anzidei M, Vendramin GG (2001) Genetic variation at chloroplast microsatellites (cpSSRs) in Abies nebrodensis (Lojac.) Mattei and three neighbor ing Abies species. Theor Appl Genet 102: 733-740 Petit RJ, Aguinagalde I, de Beaulieu J-L, et al. (2003) Glacial refugia: hotspots but not melting pots of genetic diversity. Science 300: 1563-1565

55 Paper I: Colonization history and phylogeography

Petit RJ, Brewer S, Bordacs S, et al. (2002) Identification of refugia and post-glacial colonization routes of European white oaks based on chloroplast DNA and fossil pollen evidence. Forest Ecol Manag 156: 49-74 Petit RJ, Duminil J, Fineschi S, Hampe A, Salvini D, Vendramin GG (2005) Comparative or ganization of chloroplast, mitochondrial and nuclear diversity in plant populations. Mol Ecol 14: 689-701 Pons O, Petit RJ (1995) Estimation, variance and optimal sampling of gene diversity. 1. Haploid locus. Theor Appl Genet 90: 462-470 Pons O, Petit RJ (1996) Measuring and testing genetic differentiation with ordered versus unor dered alleles. Genetics 144: 1237-1245 Posada D, Crandall KA, Templeton AR (2000) GeoDis: a program for the cladistic nested analy sis of the geographical distribution of genetic haplotypes. Mol Ecol 9: 487^488 Rendell S, Ennos RA (2002) Chloroplast DNA diversity in Calluna vulgaris (heather) popula tions in Europe. Mol Ecol 11: 69-78 Ryder AO (1986) Species conservation and systematics, the dilemma of subspecies. Trends Ecol Evol 1: 9-10 Schaal BA, Hayworth DA, Olsen KM, Rauscher JT, Smith WA (1998) Phylogeographic studies in plants: problems and prospects. Mol Ecol 7: 465-474 Sanger F, Nicklen S, Coulson AR (1977) DNA sequencing with chain-terminating inhibitors. Proc Natl Acad Sci USA 74: 5463-5467 Templeton AR (2004) Statistical phylogeography: methods of evaluating and minimizing infe rence errors. Mol Ecol 13: 789-809 Templeton AR, Boerwinkle E, Sing CF (1987) A cladistic analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping. I. Basic theory and an analy sis of Alcohol Dehydrogenase activity in Drosophila. Genetics 117: 343-351 Tileye F, Nybom H, Bartish IV, Welander M (2007) Analysis of genetic diversity in the endan gered tropical tree species Hagenia abyssinica using ISSR markers. Genet Resour Crop Evol 54: 947-958 Thompson JD, Higgins DG, Gibson TG (1994) CLUSTAL W: improving the sensitivity of pro gressive multiple sequence alignment through sequence weighting, position specific gap penal ties and weight matrix choice. Nucleic Acids Res 22: 4673-4680 Umer M, Lamb HF, Bonnefille R, Lezine A-M, Tiercelin J-J, Gibert E, Gazet J-P, Watrin J (2007). Late Pleistocene and Holocene vegetation history of the Bale Mountains, Ethiopia. Qu aternary Sci Rev 26: 2229-2246 Weising K, Gardner RC (1999) A set of conserved PCR primers for the analysis of simple se quence repeat polymorphisms in chloroplast genomes of dicotyledonous angiosperms. Genome 42: 9-19 Wolfson R, Higgins KG, Sears BB (1991) Evidence for Replication Slippage in the Evolution of Oenotheru Chloroplast DNA. Mol Biol Evol 8: 709-721 Young, Andrew G and Boyle, Timothy J 2000. Forest Fragmentation. In: Young, A, Boshier, D and Boyle, T (ed.) 2000. Forest conservation genetics: principles and practice. CABI Publish ing

56 II. Spatial distribution of genetic diversity in Hagenia abyssinica (Bruce) J.F. Gmei from Ethiopia, assessed by AFLP molecular markers

Taye Bekele Ayele*, Oliver Gailing, and Reiner Finkeldey Forest Genetics and Forest Tree Breeding, Georg-August University of Goettingen, Buesgenweg 2, 37077 Goettingen, Germany

Abstract The intraspecific genetic variation of 596 individuals from 25 populations of Hagenia abyssinica sampled from the montane forests of Ethiopia was investigated at AFLP markers. Hagenia is a wind-pollinated and wind-dispersed dioecious tree species belonging to a monotypic genus in the Rosaceae family. We obtained 106 unequivocally scorable AFLP markers out of which 91.5% were polymorphic. Despite the relatively recent immigration of Hagenia abyssinica into Ethio pia. populations showed moderate to high gene diversities (He = 0.139-0.362), and moderate but significant genetic differentiation (Fst = 0.077), reflecting high levels of post-colonization gene flow among populations. No trend of decreasing genetic diversity was detected during coloniza tion, confirming effective gene flow among populations. The observed variation at putatively neutral AFLPs does not reflect clinal variation patterns. As expected, the coefficient of popula tion differentiation was found to be much lower in the nuclear genome of Hagenia abyssinica than in the chloroplast genome. Despite the dispersal of seed and pollen of Hagenia by wind, a significant non-random fine-scale spatial genetic structure (SGS) is observed up to 80 m in some populations. Due to significant genetic differentiation observed among populations, as many populations as possible should be considered for conservation and tree improvement programs.

Key words: AFLP, Hagenia, gene diversity, kinship coefficient, population differentia tion, spatial genetic structure

*Correspondence: Taye B. Ayele; Fax: +49 551 398367; e-mail: [email protected] Permanent address: Institute of Biodiversity Conservation, Fax: +251-11-6613722; P.O.Box: 30726, Addis Abeba, Ethiopia; e-mail: [email protected]

57 Paper II: Genetic diversity at AFLPs

Introduction

The level of genetic diversity in a population is affected by various genetic, life history and eco logical characteristics that collectively define the population’s genetic structure (Yeh 2000). Tree species are generally characterized by high levels of genetic diversity within populations and rel atively low levels of differentiation among populations (Loveless & Hamrick 1984; Finkeldey & Hattemer 2007; White et al. 2007). The geographic variation in genetic diversity has important implications for the ecological and evolutionary potential of populations (for example, Hoffmann & Blows 1994).

The spatial distribution of genetic diversity in plant populations is mainly determined by life his tory traits that influence mating patterns and gene dispersal (Hamrick 1989; Hamrick & Loveless 1989; Ouborg et al. 1999) and by the historical patterns of relationship (e.g., Schaal et al. 1998). Genetic diversity is rarely distributed homogeneously within populations and genetic similarity is higher among neighbouring than among distant individuals (Vekemans & Hardy 2004; Jump & Penuelas 2007). Such fine-scale genetic structure is affected by the mating system (higher in selfing species), life form (higher in herbs than trees), population density (higher under low den sity) and population size of the target species (Vekemans & Hardy 2004, Cavers et al. 2005, Jump & Penuelas 2007, Hardy et al. 2006).

Some studies on colonization history detected a decreasing genetic diversity with increasing dis tance from refugial sources based on contemporary patterns of genetic diversity (Rivera-Ocasio et al. 2002, Coart et al. 2005, Mulugeta, et al. 2007). Such patterns reflect the impact of genetic drift associated with sequential founder effects (Wright 1969, Nei 1987, Rivera-Ocasio et al. 2002). Contrarily, others reported increasing pattern of gene diversity away from source popula tions due to gene flow and population admixture effects (e.g., Comps et al. 2001, Petit et al. 2003).

In this study, the AFLP technique was employed to investigate patterns of genetic diversity, population differentiation and fine-scale spatial genetic structure of Hagenia abyssinica from Ethiopia. AFLP is preferred to other techniques because of its short start-up time and cost- effective generation of data from a large number of loci distributed randomly across the whole

58 Paper II: Genetic diversity at AFLPs genome and the ease to generate anonymous multilocus DNA profiles in most species regardless of origin or complexity without prior sequence knowledge of the target species (Vos et al. 1995; Bensch & Akesson 2005, Mueller & Wolfenbarger 1999). Although it has not been possible to separate heterozygotes (1/0) from homozygotes (1/1), the presence and absence data can be con verted to expected heterozygosity by assuming Hardy-Weinberg equilibrium, to generate esti mates directly comparable to codominant markers (Bensch & Akesson 2005). Also, genetic dif ferentiation (F s t ) values generated from dominant markers (AFLPs and RAPDs) were in general similar to estimates obtained from microsatellites and allozymes (reviewed by Nybom 2004).

Hagenia abyssinica is a wind-pollinated and wind-dispersed broad-leaved dioecious tree species belonging to a monotypic genus in the Rosaceae family (Hedeberg 1989; Legesse 1995). The bright colourful and appealing appearance of the flowers of Hagenia is not typical for wind- pollinated species, which are usually dull in colour (Legesse 1995), suggesting that other polli nating vectors such as insects (particularly bees) or birds might be involved. It was also reported that honeybees collect pollen from the male flowers and nectar from the female flowers (Fichtl & Admasu 1994). The species is found in 12 countries in Africa stretching from Ethiopia in the North to Zimbabwe in the South and inland to Congo (Hedeberg 1989; http://www.worldagroforestry.org/Sites/TreeDBS/aft.asp). Hagenia is a multipurpose tree spe cies bestowed with considerable economic and ecological values; but due to over-exploitation, the species is gravely endangered in its natural range and especially in Ethiopia (Legesse 1995) with only about 7000 individuals left.

According to fossil pollen records (Beuning et al. 1997; Bonnefille et al. 1995; Olago et al. 1999; Umer et al. 2007), Hagenia immigrated into Ethiopia in the late Pleistocene (since 16,700 years before present) from southern African countries (Taye et al. submitted (a). A recent phylogeographic investigation using maternally inherited chloroplast markers supported this coloni zation history and suggested a single entry point into Ethiopia. Due to the recent colonization of Ethiopia, it was possible to reconstruct the colonization route of the species and to identify rare mutation and long distance seed disperal events. For example, six specific haplotypes were iden tified that were grouped in two lineages, lineage II originated most likely from a single muta tional event from lineage I (Taye et al. submitted (a)).

59 Paper II: Genetic diversity at AFLPs

The genetic diversity of few populations of H. abyssinica was investigated by using anonymous RAPD (Kumilign 2005) and ISSR (Tileye 2007) markers. Both studies covered a small spatial scale contrasting to the widespread distribution of the species in Ethiopia and were also limited to a comparatively small number of individuals per population. An investigation on the genetic diversity at the nuclear level covering the natural distribution range would enhance further un derstanding on the phylogeography and the forces shaping genetic variation patterns in Hagenia. We expect that large-scale and fine-scale genetic variation patterns are affected by the aforemen tioned historical processes and by the species' life history traits.

Three hypotheses were tested: 1) there is high variation within-populations due to effective gene flow from different pollen and seed sources and very low differentiation among-populations due to long-distance pollen and seed dispersal. 2) The species does not lose genetic diversity during colonization due to effective gene flow that counteracts effects of genetic drift. Likewise, we ex pect that the populations representing the two chloroplast lineages show similar levels of genetic diversity, even though the derived one originated by a single mutational event (from a single seed). 3) Given the wind-dispersed and wind-pollinated nature of Hagenia abyssinica, there is no fine-scale spatial genetic structure.

Materials and methods

Sampling and DNA isolation

Twenty two natural and three planted populations were surveyed from all regions where Hagenia is known to grow in Ethiopia. The natural populations are represented in 12 closed forests, 8 woodlands and 2 farmlands. The geographic distribution of the sampled populations is illustrated in Fig. 1 and the characteristics of the populations are presented in Table 1. The sizes of the sam pled trees range from 3m to 35m in height and from 2.5cm to 245cm in DBH. The distances be tween trees within the same population range from 0.1m to 730m. The natural populations have densities ranging from 0.7 to 75.7 individuals/ha (Table 5). Young leaves were collected and par tially desiccated in paper bags before drying with silica gel and stored at room temperature until DNA isolation. The sampling of individuals was nearly exhaustive in most cases due to the small

60 Paper II: Genetic diversity at AFLPs number of trees available in the forests, keeping a minimum distance when large numbers of trees were available. Trees were sampled from one spot in larger populations. The locations of each tree were mapped. Sexes of trees were identified only for 12 populations that had flowers at the time of the survey. Genomic DNA was isolated following the DNeasy 96 kit protocol of Qiagen® (Hilden, Germany). Different leaf sizes were tested for DNA extraction. Finally, a size of lctrr (ca. 20 mg) gave the best results and was used for DNA isolation.

DNA restriction, PCR amplification and genotyping

The laboratory protocol followed Vos et al. (1995) with some modifications. Genomic DNA was digested with two different restriction enzymes, a rare-cutter CEcoRI; 5’-G|AATTC-3’) and a frequent-cutter (MseI; 5‘-Tj.TAA-3’), and short DNA fragments (adapters) were ligated to cohe sive ends of the restriction fragments. Four pi genomic DNA (about 10 ng) was added to 6 f.il restriction-ligation reaction containing 1 pi T4-Ligase buffer (lOx), 1 pi NaCl (0.5M), 0.5 pi BSA (1 mg/ml), 3 pi M± Adapter (5 pmol/pl), 0.6 E± Adapter (5 pmol/pl) and to 2 pi restriction- ligation mix containing 0.2 pi T4-Ligase buffer (lOx), 0.2 pi NaCl (0.5M), 0.1 pi BSA (1 mg/ml), 0.08 pi M sel (lOU/pl), 0.6 pi EcoRI (lOU/pl) & 0.82 pi T4-Ligase (4U/pl). The resul tant solution was incubated at room temperature over-night. A pre-amplification PCR was run in a Peltier Thermal Cycler PTC-200 (MJ Research®), with a total volume of 15 pi containing 7.8 pi HPLC H20 (high performance liquid chromatography water), 1.7 pi PCR buffer (lOx), 1 pi dNTPs (2.5mM), 0.25 pi of the pre-selective primer M03 (5 pmol/pl), 0,20 pi of EOl (5 pmol/pl), 0.06 pi Taq polymerase (Qiagen®) (5U/pl) and 4 pi of the restriction-ligation reaction (diluted ~1:4). The pre-amplification PCR profile was 15 min. at 72 °C, followed by 20 cycles of 10 sec. denaturation at 94 °C, 30 sec. annealing at 56 °C and 2 min. extension at 72 °C, with a final extension step of 30 min. at 60 °C. A selective amplification was run with a total volume of 15 pi containing 8.11 pi HPLC H20, 1.6 pi PCR buffer (lOx), 0.4 pi dNTPs (2.5mM), 0.6 pi M- Primer (5 pmol/pl ), 0.25 pi E-Primer (5 pmol/pl ), 1.0 pi MgCl2 (25mM), 0.06 pi Taq poly merase (Qiagen®) (5U/pl), and 3 pi pre-amplification product (diluted 1:10). The selective PCR profile was 15 min. initial denaturation at 94 °C, followed by 9 cycles of 30 sec. denaturation at 94 °C, 30 sec. annealing at 65 °C (but reduced by 1 °C per cycle) and 2 min. extension at 72 °C, followed 24 cycles of 30 sec. denaturation at 94 °C, 30 sec. annealing at 56 °C and 2 min. exten sion at 72 °C, with a final extension of 10 min at 72 °C. Aliquots of the selective amplification

61 Paper II: Genetic diversity at AFLPs products were diluted (1:5) before electrophoretic separation. Two pi diluted selective PCR product was added to 12 pi HiDi formamide dye containing -0.02 pi internal size standard (GS ROX 500, Applied Biosystems®), denaturated for 2 minutes at 90°C, quickly cooled on ice, and separated on a capillary sequencer ABI 3100 Genetic Analyser (Applied Biosystems®).

Eighteen primer combinations (4 E & 6 M primers) were tested in different sets (nomenclature according to Keygene N.V.®). The primer combination E41-M67 (5’-FAM-GAC TGC GTA CCA ATT CAG G-3’and 5-GAT GAG TCC TGA GTA AGC A -3’, respectively) showed a well-resolved banding pattern and a high degree of polymorphism. A total of 596 individuals (23-24 per population) were genotyped with this combination. Reproducibility tests were con ducted on 15 samples randomly selected from each run. Only 100% reproducible loci were

Fig. 1. Distribution of Hagenia populations sampled from Ethiopia, represented by solid circles. Codes of popula tions follow Table 1. Broken lines with arrows indicate the putative colonization route of Hagenia starting from the most likely source population BG as deduced from cpDNA analysis (Taye et al. submitted (a)). Source map: Assefa G (unpublished) Paper II: Genetic diversity at AFLPs

considered in the final analysis, resulting in 106 putative loci. Furthermore, three standard lanes, two containing the same individuals and one holding a negative control were run on each plate to compare the data from different runs and to check for the mobility of fragments.

Table 1 Description of Hagenia populations sampled from the mountains of Ethiopia

altitude M in D ensity *Sex

Populations Code Latitude L ongitude (m asl) ARF T M ax T H n N (ind/ha) index

Debark-Mariam DK 1 3 ° ir 37°57' 3013 1270 8.8 19.7 4 24 26 16 1

Debark-Plantation DKP 13°12' 38°01' 3005 1270 8.8 19.7 4 24 - na

Kimir-Dingay KDP 11°48' 3 8 °1 4 ’ - 1350 9.2 21.9 1 ,6 24 - na plantation

W oldiya S e’at WD 11°55" 3 9 °2 4 ’ 3112 908 na na 6 24 120 na

M ichael

K osso-ber KB 10°59' 36°54' 2702 2381 12.9 27.4 1 24 60 52 na

Denkoro DR 10°52’ 38°47' 3061 896 10.9 2 1 .8 6 24 60 12.5 0.7

Wonbera WB 10°34- 35°41' 2428 1622 na na 5 24 60 75.7 na

W of-w asha w w 09 °4 5 ' 3 9 0 4 4 - 3159 941 6.1 19.9 1 24 45 5.4 na

C hilim o CM 0 9 °0 5 ' 38°10' 2805 1114 11.5 25.8 1 ,4 24 65 17.8 na

Dindin DN 0 8 °3 6 ’ 40° 14’ 2410 989 12.7 28,0 1 24 55 13.9 na

Zequala Abo ZQ 08°32' 38°50' 2856 1215 na na 1 23 60 0.7 0.3

Boter-becho BB 08°24' 37°15' 2772 1666 5.7 23.6 1 24 60 27.8 1.2

C hilalo CL 07°56' 39°11 ’ 2815 796 9.8 23,0 1 24 70 7.1 1.8

Sigmo plantation SMP 0 7 °5 5 ' 36°10' 2300 1837 11.4 2 1 .6 4 24 na

Sigm o SM 0 7 °4 6 ' 3 6 °0 5 ’ 2651 1837 11.4 2 1 .6 4 23 60 23.8 na

Munesa MS 07°25' 38°53" 2459 1028 10.1 24.3 1 24 80 10 na

B onga BG 07°17' 36°22' 2238 2217 11.9 26.6 4 24 80 5 0.9

K ofele KL 07oll' 38°52' 2757 1305 7.7 20.1 1,2,3 24 110 12.5 1.8

Dinsho DO 07 °0 5 ' 3 9 0 4 7 - 3117 1213 3.4 2 0 .8 2 24 260 16.7 3

Doddola-Serofta DS 06°52’ 39°02' 2700 1074 6.7 24.3 2 23 75 10 2.9

Doddola-Dachosa DD 06 °5 2 ' 3 9 0 , 4 - 3039 1074 6.7 24.3 2 24 5000 30.9 na

R ira RR 06°45' 39°43’ 2725 736 11a na 2 23 170 10.5 na

Bore BR 0 6 °1 7 ’ 38°39' 2631 1526 8.3 18.8 1,2,3 24 100 9.1 1.1

Uraga UR 06°08' 38°33' 2508 1228 8.3 18.8 2,3 24 70 13.9 0.7

HagereMariam HM 05°5r 38°17' 2443 1228 12.3 23,0 2 24 55 4.8 0.9

Total 110 6741

M asl= meters above sea level; ARF = Mean Annual Rainfall; ml = millilitres; Min T = Mean minimum tempera ture; Max T = Mean maximum temperature; H = chloroplast haplotype (Taye et al. submitted (a)); n= no. of sam ples analysed ; N= population size; *sex index is determined from the relative numbers of male to female individu als for 26-50 individuals from each population; na = not available. Source of climatic data: National Meteorological Agency Service (Ethiopia)

63 Paper II: Genetic diversity at AFLPs

Data analysis

Data were aligned with the internal size standard using GENESCAN 3.7 and fragments were scored with GENOTYPER 3.7 (Applied Biosystems®). Fragments with sizes ranging from 50- 500 nucleotides (bp) were scored as present (1) or absent (0) and transformed to a 1/0 matrix. Each fragment was controlled and edited manually.

Overall and gender-segregated genetic diversity (estimated as total diversity (Ht), within- population diversity (He) and among-population diversity (//b)), percentage of polymorphic loci

(PPL) at the 5% level, and coefficient of differentiation among-populations (.F s t ) were computed using AFLP-SURV (Vekemans et al. 2002, available at http://www.ulb.ac.be/sciences/lagevA following a Bayesian method with non-uniform prior distribution of allele frequencies (Zhivo- tovsky 1999). The null hypothesis for the Fst test (that there is no genetic differentiation among the populations) is rejected at p<0.001 if the observed FSt is higher than the value of FSt lying at the 1% rightmost part of the distribution (Table 3). This observation leads us to conclude that the actual populations are genetically more differentiated than random assemblages of individuals. Gene flow (Nm) was estimated using the formula: Nm = (1-Fst)/4Fst (Slatkin & Barton 1989). Loci that were found to be monomorphic in all populations were excluded from the final compu tation in order to obtain Nei's unbiased gene diversity that is comparable to expected heterozy gosity (e.g., Nybom 2004). As Hagenia is a diocious and hence completely out-crossing species, Hardy Weinberg equilibrium was assumed in all computations. Partition of genetic diversity and the significance of the differences within and among-populations and different groups were esti mated by the analysis of molecular variance (AMOVA) using ARLEQUIN Version 3.0 based on AFLP phenotypes (Excoffier et al. 2005; http://cmpg.unibe.ch/software/arlequin3). The different groups of the sampled populations that are used to examine the partitioning of genetic diversity are provided in Supplementary Table 1. The fine-scale spatial autocorrelation analysis for 21 natural populations was performed with SPAGeDi 1.2 (Hardy & Vekemans, 2002) using pair wise kinship coefficients (Fy) between individuals (Hardy 2003). The inbreeding coefficient is assumed to be 0 (as for diocious species) following Hardy et al. (2006) and Tero et al. (2005). The significance of the spatial genetic structure (SGS) was tested by upper and lower bounds of the 95% confidence interval of Fy defined after 10 000 random permutations of individuals among geographic locations. Eight distance classes were determined for all populations with one

64 Paper II: Genetic diversity at AFLPs exception (DK) after series of tests in order to obtain a minimum of 30 pairs of individuals that lie within a given distance interval. The distance classes were set to 4 for population DK. The program NTSYS-pc 2.0 (Rohlf 1998) was used to draw a phylogenetic tree using the UPGMA (Unweighted Pair Group Method with Arithmetic mean) clustering method and to examine the correlation between geographic and genetic distances (Mantel test).

R esu lts

Within population genetic diversity

The AFLP analysis of 596 samples from 25 populations of H. abyssinica resulted in a total of 106 unambiguously scorable putative markers in the range from 52 to 496 bps of which 97 (91.5%) were polymorphic. The percentage of polymorphic loci (PPL) within-populations ranges from 29.9% at Dodola Serofta (farmland/homestead population with a size of N = 75) and Uraga (located in a very small forest, N = 70) to 90.7% at Dinsho (located in a well-protected Park For est, N = 260). Moderate to high gene diversities were observed at AFLP loci ranging from 0.139 at Dodola-Serofta to 0.362 at Dinsho (Table 2) with a mean genetic diversity of He = 0.195. The largest remaining population DD (N = 5000) showed only a moderate genetic diversity (Hc = 0.173, PPL = 36.1%). On the other hand, population DK with only 26 remaining individuals showed comparatively high levels of genetic diversity (He = 0.217, PPL = 45.4%). Even though there are marked differences in genetic diversity for some populations, mean genetic diversities for the two sexes are nearly the same (//e = 0.207 ± 0.013 for male, //e = 0.201 ± 0.019 for fe male). The two chloroplast lineages (see introduction) show only minor non-significant differ ences in mean and in total genetic diversity (Lineage 1: //e = 0.193 / 0.206, Lineage II: mean He = 0.197/0.214; Table 2).

Based on chloroplast DNA analyses, the putative entry point of Hagenia into Ethiopia was the southwest mountains of Ethiopia (Taye et al. submitted (a)). The colonization routes of the spe cies were reconstructed from southwest to the north, to the east and to the south (Fig. 1). There was no significant association between migration distance and genetic diversity (//e, PPL) of

65 Paper II: Genetic diversity at AFLPs populations (Spearman's nonparametric correlation r = -0.205, p= 0.186) and thus no indication of loss of genetic diversity during colonization.

Partitioning o f genetic diversity among populations

The measures of gene diversity in subdivided populations (Nei 1987) show high mean within- population variation (Hc = 0.195) and moderate population differentiation (Fsj = 0.077, p< 0.001). The differentiation between populations within the two chloroplast lineages of different age was similar (Fst = 0.063 p< 0.001, Fst =0.083, p < 0.001) (Table 3). The estimate of gene flow computed for all populations based on Fsias the number of migrants per generation was 3.

Analyses of molecular variance (AMOVA) based on AFLP phenotypes was performed for all populations and for different groups (ecosystems, geographic regions, types of forest stands, tree seed zones, chloroplast lineages and sexes, Table 4). The detailed description of the grouping is provided in Supplementary Table 1. The AMOVA performed for all populations revealed that 10.4% of the total variation was attributed to the differences among populations. Very low pro portions of the total variation were distributed among groups representing different ecosystems, geographic regions, forest stands, tree seed zones and the two sexes (Table 4). Only differentia tion among chloroplast lineages was significant (PV = 0.78%, p < 0.05). Within each category, the level of genetic differentiation among-populations is similar ranging from 7.5% to 10.6% (Table 4). No private fragment or fragment fixed in one population was detected.

66 Paper II: Genetic diversity at AFLPs

Table 2 Summary of wilhin-populations genetic diversity of Hagenia abyssinica for 25 popula tions. The populations are sorted north to south.

n PPL He Populations Forest type All All All DK woodland 24 45.4 0.217 DKP plantation 24 47.4 0.226 KDP plantation 24 39.2 0.183 WD woodland 24 43.3 0.194 KB Closed forest 24 45.4 0.206 DR Closed forest 24 39.2 0.189 WB Closed forest 24 43.3 0.211 WW woodland 24 41.2 0.189 CM Closed forest 24 37.1 0.192 DN Closed forest 24 48.5 0.212 ZQ Closed forest 23 45.4 0.205 BB Closed forest 24 44.3 0.213 CL woodland 24 37.1 0.177 SMP plantation 24 33.0 0.146 SM Closed forest 23 38.1 0.170 MS Closed forest 24 49.5 0.200 BG Closed forest 24 46.4 0.198 KL Wooded grassland 24 42.3 0.195 DO Closed forest 24 90.7 0.362 DS Farmland 23 29.9 0.139 DD Closed forest 24 36.1 0.173 RR woodland 23 36.1 0.169 BR Wooded grassland 24 38.1 0.187 UR Closed forest 24 29.9 0.160 HM Farmland 24 35.1 0.168 Total/mean 596 100* 0.195 n = sample size; PPL = percent of polymorphic loci; Hc= Nei's gene diversity; for populations code, see Table 1. *The total PPL is 100 because the monomorphic loci were excluded.

67 Paper II: Genetic diversity at AFLPs

Table 3 Summary of the mean gene diversity and population differentiation in subdivided popu lations of Hagenia abyssinica for all populations and for the two chlorotype lineages

H, Hw Hb Fst All LI LII All LI LII All LI LII All LI LII

Mean 0.212 0.206 0.214 0.195 0.193 0.197 0.016 0.013 0.018 0.077 0.063 0.083 Upper 99% limit 0.013 0.013 0.018

P 0.000 0.000 0.000

H, = total diversity; Hw = within-population diversity; Hb = among-population diversity . F St = population differen tiation; lineage I (LI): DK, DKP, KDP, WD, DR, WB, SM. SMP & BG populations; lineage II (LII): BB, BR, CL, CM, DD, DN, DO, DS, HM, KB, KL, MS, RR, UR, WW & ZQ populations; for population codes, see Table 1. Up per 99% limit = value of FSt lying at the 1% rightmost part of the distribution under the null hypothesis, p = the probability of rejecting the null hypothesis

Relationships among populations

The UPGMA dendrogram (Fig. 2) was calculated from Nei’s genetic distances (Nei 1978). The pair-wise Nei’s genetic distance matrix (Supplementary Table 2) among 25 populations exhibits genetic differences of less than 7% for each pairs of population. In general, the UPGMA dendro gram does not reflect the geographic origin of the populations. While populations BR, DD, KL & HM from the southern region are assembled in the same cluster together with one population from the northern region, populations DS, MS, RR, UR and DO from the same region are dis tributed in different parts of the UPGMA tree. Dinsho population is an outlier being the most dissimilar population with the highest gene diversity (0.362). The planted populations DKP and SMP were not clustered with their putative parent populations DK and SM, respectively (Fig. 2). A test of association between geographic and genetic distances (Mantel test) showed a very low and non-significant correlation (r = 0.14607, p = 0.9024). For example, the highest genetic dis tance was observed between population RR and DO (0.0669) that are geographically close (37.5 km air distance) but separated by a big mountain embracing the second highest peak in the coun try. On the other hand, the three pairs of populations with the lowest genetic distances (from 0 to 0.0005) between them (WD & WW, WD & BB and KL & BR) are widely separated (240 km, 452 km and 103 km, respectively), with some small mountains between them.

68 Paper II: Genetic diversity at AFLPs

Table 4 Partitioning of AFLP variation among Hagenia abyssinica individuals in Ethiopia com puted by analysis of molecular variance (AMOVA).

Source of variation df SS vc pv Is Among-populations 24 557.42 0.71554 10.4 *** Within-populations 571 3521.63 6.16748 89.6 *** Among ecosystem groups 4 89.92 -0.00874 -0.13 ns Among-populations 20 467.50 0.72178 10.49 *** Within-populations 571 3521.630 6.16748 89.64 *** Among geographic groups 3 105.427 0.09441 1.4 ns Among-populations 21 451.993 0.64411 9.3 *** Within-populations 571 3521.63 6.16748 89.3 *** Among stand groups 3 64.062 -0.01630 -0.24 ns Among-populations 21 493.358 0.72650 10.56 * * * Within-populations 571 3521.63 6.16748 89.67 *** Among tree seed zones groups 12 311.133 0.18654 2.7 ns Among-populations 9 168.756 0.52310 7.5 *** Within-populations 502 3132.63 6.24030 89.8 *** Among chloroplast lineage groups 1 37.525 0.05421 0.78 * Among-populations 23 519.895 0.68950 9.98 *** Within-populations 571 3521.630 6.16748 89.24 *** Among sex groups 1 3.702 -0.08835 -1.34 ns Among-populations 22 270.164 0.70368 10.64 *** Within-populations 193 1157.245 5.99609 90.69 *** df = degree of freedom, SS = sum of squares, vc = variance components, pv= percent variation, Is = level of significance, *** = highly significant at p<0.001, * = significant at p<0.5, ns = not significant

Fine-scale spatial genetic structure

In general, most populations from farmlands, wooded grasslands and woodlands (6 out of 8) showed significant spatial genetic structure up to longer distances (36-80 m) whereas only 4 out of 13 closed forest populations showed family structure at shorter distance classes (15 - 44 m) (Table 5, Supplementary Fig. 1). No SGS was observed in the largest remaining Hagenia popu lation (DD) in Ethiopia (~ 5000 individuals) while significant SGS was observed in the second largest population DO (N= 260), harboring the highest genetic diversity. In most (7 out of 10) of the small populations (N = 55-80) of the closed forst type, no SGS was observed (Table 5). Den sity, population size and distance from the nearest population were not associated with the kin-

69 Paper II: Genetic diversity at AFLPs ship coefficient averaged over distance classes F(d) of 21 natural populations (for example, Spearman's nonparametric correlation coefficient (r) for density = -0.065, p = 0.391).

cuB)i_r WD(A)r WW(B) —

MS(D)— I SMPfC)—'

D S ( D ) | _

ITODt

SM(C)-----

KDP(A)----- RR(D)----- Q ZQ(B)----- BG(C), CH(B) DN(B) DR(A) BRD)l)T_ KUD)n - DD(D) WB(A) HMD) DK(A) K»A) DKP(A)

DO(D)

0.00 0A3 N « s genetic

Fig. 2 UPGMA tree drawn from Nei’s (1978) genetic distances computed from AFLPs. Popula tion codes follow Table 1. Letters in parenthesis designate geographic regions: A= northern re gion, B= central region, C= southwestern region, D= southern region.

70 Table 5 Spatial genetic structure in Hagenia populations

D istance

Population Type of forest Stand type Sam ple Pop D ensity Ma M ax D istance Sex in- (km) from

C ode size size Ind./ha X. distance classes* dex nearest

F(d) population

KB Closed forest Hagenia-dominated mixed stand 24 60 52 0.06 15 1-2 na 141.2 (WB) DR Closed forest Mixed, sparse Hagenia 24 60 12.5 ns ns 0.7 131.7 (WD) WB Closed forest Hagenia-dominated mixed stand 24 60 75.7 ns ns na 141.2 (KB) CM Closed forest Mixed, sparse Hagenia 24 65 17.8 0.09 31.6 1-3 na 98.8 (ZQ ) DN Closed forest Mixed, sparse Hagenia 24 55 13.9 ns ns na 136.3 (CL) ZQ Closed forest Mixed, sparse Hagenia 23 60 0.7 ns ns 0.3 78.3 (CL) BB Closed forest Mixed, sparse Hagenia 24 60 27.8 0.07 18 1 1.2 122.2 (CM) SM Closed forest Mixed, sparse Hagenia 23 60 23.8 ns ns na 69.3 (BG ) MS Closed forest Mixed, sparse Hagenia 24 80 10 ns ns na 28.5 (KL) BG Closed forest Hagenia-dominated mixed stand 24 80 5 ns ns 0.9 69.3 (SM ) DO Closed forest Hagenia-dominated mixed stand 24 260 16.7 0.21 44 1 3 37.5 (RR) DD Closed forest Hagenia-dominated mixed stand 24 5000 30.9 ns ns na 22.3 (DS) UR Closed forest Mixed, sparse Hagenia 24 70 13.9 ns ns 0.7 20.6 (BR) ae I: eei dvriy t AFLPs at diversity Genetic II: Paper DS Farm land Pure Hagenia stand 23 75 10 0.12 64 1-2 2.9 22.3 (DD) HM Farm land Hagenia-dominated mixed stand 24 55 4.8 0.09 58 1 0.9 41.0 (U R) KL Wooded grassland Hagenia-dominated mixed stand 24 110 12.5 0.2 56 1 1.8 28.5 (M S) BR Wooded grassland Hagenia-dominated mixed stand 24 100 9.1 ns ns 1.1 20.6 (U R) DK woodland Pure Hagenia stand 24 26 16 0.19 36 I 1 215.0 (WD) WW woodland Hagenia-dominated mixed stand 24 45 5.4 ns ns na 141.7 (DN) CL woodland Hagenia-dominated mixed stand 24 70 7.1 0.08 80 1-2 1.8 61.4 (MS) RR woodland Hagenia-dominated mixed stand 23 170 10.5 0.06 52 1-2 na 37.5 (DO) * Distance classes for only populations that showed family structures are indicated. Paper II: Genetic diversity at AFLPs

Discussion

Genetic diversity and population differentiation The moderate to high genetic diversity in Hagenia reflects effective gene flow from dif ferent pollen and seed sources, resulting in low population differentiation, which in turn reflects effective long-distance pollen and/or seed dispersal among-populations. This ob servation confirms the first hypothesis that predicted high variation within populations and low differentiation among populations. The absence of association between genetic and geographic distances might be explained by a random and long-distance dispersal of pollen. Accordingly, planted populations were not clustered with their putative parent populations unlike for chloroplast markers, where plantations showed the same haplo types as their parent populations (Taye et al. submitted (a)).

Phylogeographic analyses of the same 25 populations at cpDNA revealed two chloroplast lineages (Taye et al. submitted (a)). Most likely, lineage I originated from Lineage II by a deletion in a specific chloroplast region. Thus all plants of lineage I originated from a single seed during colonization of Ethiopia. Assuming restricted gene flow by pollen we would expect a much lower genetic diversity in populations with the derived chloroplast haplotypes of lineage I. However, the two chloroplast lineages demonstrated comparable mean genetic diversities with lineage II exhibiting slightly higher values. Also genetic differentiation (Fst) between populations of chloroplast lineage II and the derived lineage I were similar, showing the harmonizing effect of gene flow.

No trend of decreasing genetic diversity during colonization was detected, reflecting ef fective gene flow. This observation allows us to accept the hypothesis “Hagenia does not lose genetic diversity during colonization due to effective gene flow that counteracts ef fects of genetic drift”. A general trend of increasing genetic diversity away from refugia was observed in European beech based on isozymes (Comps et al. 2001), suggesting a gain in gene diversity during recolonization due to gene flow, population admixture ef fects and selection. Petit et al. (2003) also reported that the mixing of colonization routes and increased levels of seed flow resulted in increased intrapopulation diversity away Paper II: Genetic diversity at AFLPs from refugia in some European woody species. In contrast, Lobelia giberroa, which en tered Ethiopia also from the south (Mulugeta, et al. 2007), Carpinus betulus (Betulaceae) in Europe (Coart et al. 2005) and Ptercarpus officinalis (Fabaceae) in the Caribbean (Rivera-Ocasio et al 2002) demonstrated decreasing diversity during recolonization (all based on AFLP analyses). The level of genetic diversity in a population is affected by an array of genetic, life history and ecological characteristics that collectively define the population’s genetic structure (Yeh 2000). Lobelia giberroa has a giant-rosette growth form, reaching 9m when in flower (Mulugeta, et al. 2007) and it grows in altitudes higher than Hagenia. As Hagenia is a canopy tree, pollen and seeds can disperse over long dis tances contributing to the maintenance of comparatively high levels of gene diversity.

In general, closed forest populations harbored more gene diversity (mean He = 0.207) than woodland (mean He = 0.190) and farmland (mean He = 0.172) populations. The maximum genetic diversity was recorded for the population Dinsho (DO) that is situated in a well-protected Park Forest whereas the lowest genetic diversity was recorded for the farmland populations Doddola Serofta (DS) and Hagere Mariam (HM), suggesting a negative effect of human-induced selection.

Comparison o f genetic diversity with other species

Most genetic diversity studies of trees were done with isozymes (e.g., Hamrick & Godt (1996)) and hence the results are not directly comparable with AFLP based diversity es timates. In a review of the estimation of intraspecific genetic diversity in plant species by using nuclear DNA markers, Nybom (2004) reported a slightly higher mean within- population diversity (Hpop) of 0.22 (RAPD), 0.23 (AFLP) and 0.22 (ISSR) based on the outcome of 60, 13 & 4 studies, respectively. The overall mean gene diversity of Hagenia at AFLPs (He = 0.195) is comparable to some other plant species such as the insect- pollinated Hibiscus tiliaceus (Malvaceae, He = 0.198, Tang et al., 2003) and the wind- pollinated Acanthopanax sessiliflorus (Araliaceae, He = 0.187, Huh et al., 2005) but lower than the insect-pollinated Malus sylvestris (Rosaceae, He = 0.225, Coart et al., 2003). Studies based on AFLP markers are limited to a few tropical tree species and in

73 Paper II: Genetic diversity at AFLPs formation on the method of estimating He is missing in most of the cases, making com parisons difficult. Here, we report a comparative analysis from the available literature applying the same method for the estimation of allele frequencies. The insect-pollinated tropical species Dipterocarpus cf. condorensis (Dipterocarpaceae, He = 0.215, Luu 2005) also showed a slightly higher mean gene diversity than Hagenia at AFLP markers. H. abyssinica exhibited higher mean gene diversity than some other tropical and subtropical tree species such as the bird-pollinated Lobelia giberroa (Apocynaceae, He = 0.066, Mu- lugeta, et al. 2007) the insect-pollinated Shorea leprosula (Dipterocarpaceae, He = 0.161, Cao et al. 2006), the insect-pollinated Shorea parvifolia (Dipterocarpaceae, He = 0.138, Cao et al. 2006), the insect and wind-pollinated Acer skutchii (Sapindaceae, He = 0.15, Lara-Gomez et al., 2005) and the bird-pollinated Pelliciera rhizophorae (Pellici- eraceae, He = 0.117, Castillo-Cardenas et al. 2005) at AFLP loci. Tileye et al. (2007) re ported higher mean gene diversity (0.30) in 12 populations of Hagenia from central and southern regions of Ethiopia at 84 polymorphic ISSR markers. But Qian et al. (2001) and Nybom (2004) argued that ISSR markers generally over-estimate gene diversity as com pared to other markers. Hagenia also showed lower mean gene diversity at AFLPs than some other tree species growing in Ethiopia, notably, the insect-pollinated Cordia afri cana (Boraginaceae, He = 0.287, Abayneh 2007) and the wind-pollinated Juniperus procera (Cupressaceae He = 0.269, Demissew 2007). The wider distribution of both species and the effective dispersal of seeds of Cordia by animals explain the higher diversity than Hagenia. The habitat of Juniperus is closer to Hagenia than Cordia that grows in lower altitudes and warmer climate.

In the present study, the maximum gene diversity is recorded for population Dinsho (DO) in the Bale region, conforming to the highest gene diversity found in wild coffee (Coffea arabica, Rubiaceae; Aga et al. 2005) and Lobelia giberroa (Mulugeta et al. 2007) re ported from the same region at ISSR and AFLP markers, respectively. But the neighbour ing population Rira (RR), which is closer to the aforementioned populations of Coffea arabica and Lobelia giberroa, showed much lower He than Dinsho. In contrast Tileye et al. (2007), Abayneh (2007) and Demissew (2007) reported lower gene diversity in Hagenia abyssinica (ISSR), Cordia africana (AFLP) and Juniperus procera (AFLP), re

74 Paper II: Genetic diversity at AFLPs spectively from the Bale region as compared to other regions. The disagreement between the result of Tileye et al. (2007) and that of the present study on the same population (DO) of H. abyssinica is most likely due to small number of trees sampled by the former.

In accordance with the wind-pollinated and out-crossing mating system of Hagenia, a moderate population differentiation (F s t ) was observed, suggesting high levels of gene flow particularly via pollen. Hamrick and Godt (1996) reported an average Fst value of 0.092 for out-crossing perennials at isozyme loci. High levels of gene flow are not unex pected for out-crossing tree species (Hamrick & Godt 1989) and it is reinforced in Hagenia by a recent divergence of populations as confirmed by cpDNA and palynologi cal evidences (Taye et al. submitted (a)). Tileye et al. (2007) found a higher coefficient of differentiation (Gst = 0.25) among 12 populations of Hagenia using ISSR markers. They sampled fewer individuals (10 trees) per population and also included different popula tions that are smaller in size. This might explain the differences between the two studies. Comparable levels of population differentiation were found at AFLPs in Cordia africana

( ® st = 0.072, Abayneh 2007), Acer skutchii (F st = 0.075, Lara-Gomez et al. 2005), Acanthopanax sessilifloms (Gst = 0.069, Huh et al., 2005) and in two species from the Betulaceae family that have a similar breeding system as Hagenia - Carpinus betulus (Fst = 0.074) and C. orientalis (Fst = 0.0863) (Coart et al. 2005). Higher coefficients of population differentiation were also reported for insect-pollinated Shorea species (FSt =

0.25-0.31, Cao, 2006) and Hibiscus tiliaceus (F s t = 0.152, Tang et al., 2003), and bird- pollinated Pelliciera rhizophorae (F st = 0.265, Castillo-Cardenas et al. 2005) based on AFLP markers. On the other hand, Fst values lower than that of Hagenia were reported for wild Malus sylvestris (Rosacea, Fst= 0.0464, Coart et al. 2003).

Fine-scale spatial genetic structure Despite the dispersal of seed and pollen by wind, significant spatial genetic structure was observed within nearly half of the populations of Hagenia abyssinica, reflecting restricted geneflow within populations and mating of related trees. Positive values of F;j were found at short distances, indicating higher genetic relatedness among neighbor individuals than random pairs of individuals, whereas negative values of Fy occurred at larger distances,

75 Paper II: Genetic diversity at AFLPs showing isolation-by-distance within a population (Tero et al. 2005). Significant spatial genetic structure in Hagenia extends up to 80 m from individual trees. This result allows us to reject the hypothesis that predicts absence of fine-scale genetic spatial patterning in Hagenia. While there was no association between tree density, population size or dis tance from the nearest population and the occurrence of wide-ranging SGS, significant SGS was observed more frequently in farmlands and woodlands as compared to closed forests. The extent of SGS in the present study is possibly underestimated due to low sample size and lower number of AFLP loci as compared to other studies. For example, Jump and Penuelas (2007) observed SGS upto about 30m at 6 SSR loci, while significant SGS upto 110m was observed at 250 AFLP markers in wind-pollinated Fagits svlvatica.

Conclusions and recommendations

The intrapopulation genetic diversity and interpopulation genetic differentiation of Hage nia abyssinica is consistent with earlier predictions based on breeding system, life cycle, population size, density and geographic range. Despite the relatively recent colonization of Ethiopia by Hagenia abyssinica that has been suggested by fossil pollen data (Taye et al. submitted (a)) and the small population sizes, the AFLP analysis detected moderate to high gene diversities within populations with considerable differences in He between populations, and moderate but significant genetic differentiation among populations. Since even little effective pollen per generation is sufficient to counteract loss of genetic diversity (Wright 1931, Finkeldey & Hattemer 2007), the effect of recent colonization and the small population sizes is not reflected in the levels of gene diversity. The ob served variation at putatively neutral markers does not reflect clinal variation patems. Consequently, 1) a seed zone approach is questionable to conserve genetic diversity, 2) it is difficult to capture optimal variation for conservation and tree improvement based on approaches to sample ecological and/or geographic zones, 3) Due to significant genetic differentiation observed among populations, it is necessary to collect seeds from as many populations as possible for gene bank storage, and for the establishment of provenance trials and ex situ plantations. The very high gene diversity in some populations calls for the need to conserve the observed variability. The moderate to high intraspecific variation

76 Paper II: Genetic diversity at AFLPs and a wide vertical distribution of the populations (2200 to 3200 m asl) may suggest that Hagenia might have occupied wider areas in the past than at present. The extant popula tions, on the other hand, harbor quite high levele of gene diversity despite of their small sizes. Nonetheless, our data suggests that human impact in the form of selective removal of trees conversely affects gene diversity, as observed in the two farmland populations. A significant fine-scale spatial genetic structure was observed in some populations despite the dispersal of seed and pollen of Hagenia by wind.

Further work on the intraspecific genetic variation and palynological investigations in other African countries where Hagenia is known to grow is suggested to fully understand the colonization history and to identify the refugia of the species. Paternity analyses to estimate effective pollen-flow distances are also recommended.

Acknowledgements

This project is a component of the “Support to the Forest Genetic Resources Conserva tion Project” of the Ethiopian Institute of Biodiversity Conservation (1BC) supported by the German Federal Ministry of Economic Cooperation and Development (BMZ) through the German Technical Cooperation (gtz). The German Academic Exchange Service (DAAD) executed the grant. The National Meteorological Service Agency of Ethiopia provided climatic data. We thank Oleksandra Dolynska, Thomas Seliger and Olga Artes for kindly assisting in the laboratory.

77 Paper II: Genetic diversity at AFLPs

References

Abayneh, D. 2007. Genetic variation in Cordia africana Lam. In Ethiopia. PhD thesis, Georg-August University of Gottingen. Aalbaek A (1993) Tree seed zones for Ethiopia. National Tree Seed Project, Addis Ababa Aga, E., Bekele, E. and Bryngelsson T. 2005. Inter-simple sequence repeat (ISSR) varia tion in forest coffee tree (Coffea arabica L.) populations from Ethiopia. Genetica, 124:213-221. Azene, B.T., Bimi,e A. and Tegnas, B 1993. Useful trees and shrubs for Ethiopia: identi fication, propagation and management for agricultural and pastoral communities. Technical handbook No. 5, Regional Conservation Unit, Swedish International De velopment Authority (SIDA). Berhanu, M.A., Ngadjui, B.T., Merhatibeb, B. and Mdee, L.K. 1999. Novel natural prod ucts from marketed plants of eastern and southern Africa. Pure Appl. Chem. 71: 919- 926. Bensch, S. and Akesson, M. 2005. Ten years of AFLP in ecology and evolution: why so few animals? Mol. Ecol. 14: 2899-2914. Beuning, KRM, Talbot, M.R., Kelts, K. 1997. A revised 30,000-year paleoclimatic and paleohydrologic history of Lake Albert, East Africa. Palaeogeogr. Palaeocl. 136: 259- 279. Bonnefille, R., Riollet, G., Buchet, G., Icole, M., Lafont, R., Arnold, M., Jolly, D. 1995. Glacial/interglacial record from intertropical Africa, high resolution pollen and carbo data at Rusaka, Burundi. Quaternary Sci. Rev. 14: 917-936. Cao, C-P., Finkeldey, R., Siregar, I.Z., Siregar, U.J., Gailing, O. 2006. Genetic diversity within and among-populations of Shorea leprosula Miq. and Shorea parvifolia Dyer (Dipterocarpaceae) in Indonesia detected by AFLPs. Tree Genet Genomes, 2: 225-239. Castillo-Cardenas, M.F., Toro-Perea, N., and Cardenas-Henao, H. 2005. Population ge netic structure of neotropical Mangrove species on the Colombian Pacific Coast: Pel- liciera rhizophorae (Pellicieraceae). Biotropica, 37: 266-273. Cavers S., Degen B., Caron H., Lemes, M.R., Margis, R., Salgueiro, F., and Lowe, A.J. 2005. Optimal sampling strategy for estimation of spatial genetic structure in tree populations. Heredity, 95: 281-289. Cavers, S., Navarro, C., Lowe, A.J. 2003. Chloroplast DNA phylogeography reveals colonization history of a neotropical tree, Cedrela odorata L., in Mesoamerica. Mol. Ecol. 12: 1451-1460. Coart, E., Van Glabeke, S., Petit, R.J., Van Bockstaele, E. and Roldan-Ruiz, I. 2005. Range wide versus local patterns of genetic diversity in hornbeam (Carpinus betulus L.). Conserv Genet. 6: 259-273. Coart, E., Vekemans, X., Smulders, M..JM., Wagner, I., Huylenbroeck. J.V., Bockstaele, E.V. and Roldan-Ruiz, I. 2003. Genetic variation in the endangered wild apple (Ma- lus sylvestris (L.) Mill.) in Belgium as revealed by amplified fragment length poly morphism and microsatellite markers. Mol. Ecol. 12: 845-857.

78 Paper II: Genetic diversity at AFLPs

Comps, B., Gomory. D., Letouzey, J., Thiebaut, B. and Petit. R.J. 2001. Diverging trends between heterozygosity and allelic richness during post glacial colonization in the European beech. Genetics, 157: 389-397. Dawit, A. and Ahadu A. 1993. Medicinal Plants and Enigmatic Health Practices of Northern Ethiopia. Addis Ababa. Degen, B., Petit, R. and Kremer, A. 2001. SGS - Spatial Genetic Software: a computer program for analysis of spatial genetic and phenotypic structures of individuals and populations. J. Hered. 92: 447-448. Demissew, S.D. 2007. Genetic variation of Juniperus procera Hochst. Ex Endl. popula tions in Ethiopia assessed by using microsatellite (SSR) and AFLP markers. MSc. Thesis. Georg-August University of Gottingen. Excoffier, L., Laval, G., Schneider, S. 2005. Arlequin (version 3.0): An integrated soft ware package for population genetics data analysis. Evol Bioinformatics Online, 1: 47-50. Faris, H. 1998. Seedling biomass, seed germination responses and cytology of Hagenia abyssinica (Bruce) J.F.Gmel. MSc. Thesis, Addis Ababa University. Fichtl R, Admasu A 1994. Honeybee flora of Ethiopia. DED - Margraf Verlag, Weikers- heim. Finkeldey, R., and Hattemer, H.H. 2007. Tropical Forest Genetics. Springer-Verlag, Ber lin. Hamrick, J.L. 1989. Isozymes and the analysis of genetic structure in plant populations. In Isozymes in Plant Biology. Edited by D.E. Soltis, and P.S. Soltis. Dioscorides Press, Portland, pp. 87-105. Hamrick, J.L., and Loveless, M.D. 1989. The genetic structure of tropical trees popula tions: associations with reproductive biology. In Evolutionary Ecology of Plants. Edited by J.H. Bockand, and Y.B. Lindhart. Westview Press, Boulder, pp. 129-146. Hamrick, J. L., and Godt, M. J. 1989. Allozyme diversity in plant species. In Plant popu lation genetics, breeding and germplasm resources. Edited by A.H.D Brown, M.T. Clegg, A.L. Kahler, and B.S. Weir. Sinauer, Sunderland, Mass. Hamrick, J.L., and Godt, M.J.W. 1996. Effects of life history traits on genetic diversity in plant species. Philos. Trans. R. Soc. Lond. B. 351: 1291-1298. Hardy, O.J. Maggia, L., Bandou, E., Breyne, P., Caron, H., Chevallier, M-H., Doligez, A., Dutech, C., Kremer, A., Latouche-Halle, C., Troispoux, V., Veron, V. and De gen, B. 2006. Fine-scale genetic structure and gene dispersal inferences in 10 Neo tropical tree species. Mol. Ecol. 15: 559-571. Hardy, O.J. 2003. Estimation of pairwise relatedness between individuals and characteri zation of isolation-by-distance processes using dominant genetic markers. Mol. Ecol. 12: 1577-1588. Hardy, O. J., and Vekemans X. 2002. SPAGeDi: a versatile computer program to analyse spatial genetic structure at the individual or population levels. Mol. Ecol. Notes, 2: 618-620. Hedeberg, O. 1989. Rosaceae. In Flora of Ethiopia, Vol. 3, Pittosporaceae to Arralaceae. Edited by I. Hedeberg, and S. Edwards. Addis Abeba and Asmara, Ethiopia; Uppsala, Sweden. Hoffmann, A.A. and Blows, M.W. 1994. Species borders: ecological and evolutionary perspectives. Trends Ecol. Evol. 9: 223-227.

79 Paper II: Genetic diversity at AFLPs

Huh, M.K, Huh, H.W. and Back, K. 2005. Genetic diversity and population structure of Acanthopanax sessiliflorus (Araliaceae) using AFLP. Korean J Genetics, 27: 71-79. Jansen, P.C.M. 1981. Spices, condiments and medicinal plants in Ethiopia, their taxono my and agricultural significance. Wageningen.

Jump A.S., and Penuelas J. 2007. Extensive spatial genetic structure revealed by AFLP but not SSR molecular markers in the wind-pollinated tree, Fagus sylvatica. Mol. Ecol. 16: 925-936. Lara-Gomez, G., Gailing O., and Finkeldey R. 2005. Genetic variation in isolated Mex ican populations of the endemic maple Acer skutchii Rehd. Allg. Forst- u J-Ztg. 176: 97- 103. Legesse, N. 1995. Indigenous trees of Ethiopia: biology, uses and propagation tech niques. SLU, Reprocentralen, Umea. Loh, J.P., Kiew, R., Hay, A., Kee, A., Gan, L.H. and Gan, Y.-Y. 2000. Intergeneric and interspecific relationships in Araceae tribe Caladieae and developmol/pl ent of mo lecular markers using amplified fragment length polymorphism (AFLP). Ann. Bot. 85: 371-378. Loveless, M.D., and Hamrick, J.L. 1984. Ecological determinants of genetic structure in plant populations. Annu. Rev. Ecol. Syst. 15: 65-95. Luu, H.T. 2005. Genetic variation and the reproductive system of Dipterocarpus cf. condorensis Pierre in Vietnam. PhD Thesis, Georg-August University of Gottingen. Mueller, U.G. and Wolfenbarger, L. L. 1999. AFLP genotyping and fingerprinting. Trends Ecol. Evol. 14(10): 389-394. Mulugeta, K., Ehrich D., Taberlet P., Sileshi N. and Brochmann, C. 2007. Phylogeogra phy and conservation genetics of a giant lobelia (Lobelia giberroa) in Ethiopian and Tropical East African mountains. Mol. Ecol. 16: 1233-1243. Nei, M. 1987. Molecular evolutionary genetics. Columbia University Press. Pp 512. Nei, M. 1978. Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics, 89: 583-590. Nybom, H. 2004. Comparison of different nuclear DNA markers for estimating intraspecific genetic diversity in plants. Mol. Ecol. 13: 1143-1155. Olago, D.O., Street-Perrott F.A., Perrott R.A., Ivanovich M., Harkness D.D. 1999. Late quaternary glacial-interglacial cycle of climatic and environmental change on Mount Kenya. J Afir Earth Sci. 29: 593-618. Ouborg, N.J., Piquot, Y., Van Groenendael, M. 1999. Population genetics, molecular markers and the study of dispersal in plants. J. Ecol. 87: 551-568. Pankhurst, R. 1969. The Traditional Taenicides of Ethiopia. J. Hist. Med. All. Sci. (XXIV): 323-334. Petit, R.J., Aguinagalde, I., de Beaulieu J-L., et al. (2003) Glacial refugia: hotspots but not melting pots of genetic diversity. Science, 300, 1563-1565. Qian, J., Luscombe, N.M., and Gerstein, M. 2001. Protein family and fold occurrence in genomes: power-law behaviour and evolutionary model. J. Mol. Biol. 313: 673-681. Rivera-Ocasio, E., Aide, T.M. and McMillan, W.O. 2002. Patterns of genetic diversity and biogeographical history of the tropical wetland tree, Pterocarpus officinalis (Jacq.), in the Caribbean basin. Mol. Ecol. 11: 675-683.

80 Paper II: Genetic diversity at AFLPs

Rohlf, F.J. 1998. NTSYS-pc: Numerical taxonomy and multivariate analysis system (Version 2.0). State University of New York, USA. Schaal, B.A., Hayworth, D.A., Olsen, K.M., Rauscher, J.T., Smith, W.A. 1998. Phylo- geographic studies in plants: problems and prospects. Mol. Ecol. 7: 465-474. Slatkin, M. and Barton, N.H. 1989. A comparison of three indirect methods for estimat ing average levels of gene flow. Evolution, 43: 1349-1368. Tang, T., Zhong, Y., Jian, S.G. and Shi, S.H. 2003. Genetic diversity of Hibiscus tiliaceus (Malvaceae) in China assessed using AFLP markers. Ann. Bot. 92: 409-414. Tero, N., Aspi, J., Siikamaki, P., and Jakalaniemi. A. 2005. Local genetic population structure in an endangered plant species, Silene tatarica (Caryophyllaceae). Heredity, 94: 478—487. Tileye, F. Nybom, H., Bartish I.V. and Welander M. 2007. Analysis of genetic diversity in the endangered tropical tree species Hagenia abyssinica using ISSR markers. Genet. Resour Crop Ev. 54: 947-958. Umer, M., Lamb, H.F., Bonnefille, R., Lezine, A-M., Tiercelin, J-J., Gibert, E., Gazet, J- P., Watrin, J. 2007. Late Pleistocene and Holocene vegetation history of the Bale Mountains, Ethiopia. Quaternary Sci. Rev. 26: 2229-2246. Vekemans, X., Beauwens, T., Lemaire, M. and Roldan-Ruiz, I. 2002. Data from ampli fied fragment length polymorphism (AFLP) markers show indication of size homoplasy and a relationship between degree of homoplasy and fragment size. Mol. Ecol. 11: 139-151. Vekemans, X., Hardy. O.J. 2004. New insights from fine-scale spatial genetic structure analyses in plant populations. Mol. Ecol. 13: 921-935. Vos, P., Hogers, R., Bleeker, M., Reijans, M., Lee, Th. van der, Homes, M., Frijters, A., Pot, J., Peleman. J., Kuiper, M. and Zabeau, M. 1995. AFLP: a new technique for DNA fingerprinting. Nucleic Acids Res. 23(21): 4407-4414. White, T.L., Adams, W.T., and Neale, D.B. 2007. Forest Genetics. CABI Publishing, Wallingford and Cambridge. Wright, S. 1969. The theory of gene frequencies. In Evolution and the Genetics of Popu lations, Vol. 2. University of Chicago Press, Chicago. Pp 511. Yeh, F.C. 2000. Population Genetics. In Forest conservation genetics: principles and practice. Edited by A.Young, D. Boshier, and T.Boyle. CABI Publishing. Pp 352. Zhivotovsky, L.A. 1999. Estimating population structure in diploids with multilocus do minant DNA markers. Mol. Ecol. 8: 907-913.

81 Paper II: Genetic diversity at AFLPs

Supplementary materials to Paper II

Supplementary Table 1 The grouping of the sampled Hagenia populations that are used to exam ine the partitioning of genetic diversity at AFLP loci. No. of List of populations1 populations

Micro-ecosystem types of sampled populations Closed forest 12 DO, Z0, BB, MS, SM, UR, KB, DN, DR, BG, CM, DD Open forest/woodland 6 WW, CL, WD, RR, DK, WB Farm land/ Homestead 2 DS, HM Wooded grassland 2 BR, KL Plantation 3 DKP, SMP, KDP

Types of Hagenia forest stands Mixed stand, sparse Hagenia 8 ZQ, BB, MS, SM, UR, DN, DR, CM, //agew'a-dominated mixed 12 DO, KB. BG, DD, WW, CL, WD, RR, WB, stand HM, BR, KL Pure Hagenia stand 2 DS, DK Plantation 3 DKP, SMP, KDP

Geographic regions Northern 7 DK, DKP, DR, KB, WB, WD, KDP Central 5 WW, CL, DN, CM, ZQ South-western 4 BB, BG, SM, SMP Southern 9 BR, KL, DS, HM, DO, MS, UR, DD, RR

Chloroplast linerages Lineage I 9 DK, DKP, KDP, WD, DR, WB, SM, SMP, BG Lineage II 16 BB, BR, CL, CM, DD, DN, DO, DS, HM, KB, KL, MS, RR, UR, WW, ZQ

Tree seed zones2 15.3 1 WD 17 4 CM, DO, DD, DS 19 2 DK, KDP 20.1 2 KB, WB 20.2 2 WW, DR 20.3 1 ZQ 20.4 1 CM 21.1 1 MS 21.2 1 DN 23.2 1 SM 23.3 2 BG, BB 24.1 4 KL, BR, UR, HM 24.2 1 RR InPopulation , .•___ codes, follow „ table 1; “Ffr details on tree seed zone descriptions, see Aalbaek (1993) Supplenentary Table 2 Pairwise matrix showing Nei’s genetic distance among 25 populations of H. abyssinica from Ethiopia assessed by AFLP

82 Supplementary Table 2 Pairwise matrix showing Nei’s genetic distance among 25 populations of H. abyssinica from Ethiopia, assessed by AFLP

BB 0.000 BG 0.012 0.000 BR 0.007 0.0010.000 CL 0.002 0.017 0.006 0.000 CM 0.016 0.000 0.003 0.020 0.000 DD 0.014 0.015 0.001 0.009 0.013 0.000 DK 0.027 0.015 0.016 0.034 0.009 0.027 0.000 DKP 0.014 0.023 0.019 0.020 0.018 0.027 0.018 0.000 DN 0.023 0.002 0.006 0.030 0.005 0.020 0.011 0.031 0.000 DO 0.040 0.036 0.035 0.045 0.036 0.037 0.043 0.053 0.036 0.000 DR 0.018 0.007 0.007 0.017 0.002 0.014 0.009 0.017 0.010 0.039 0.000 DS 0.017 0.036 0.022 0.007 0.037 0.021 0.058 0.038 0.056 0.057 0.037 0.000 HM 0.015 0.016 0.007 0.012 0.020 0.005 0.039 0.034 0.025 0.041 0.019 0.018 0.000 KB 0.015 0.014 0.016 0.025 0.009 0.025 0.011 0.021 0.021 0.042 0.018 0.048 0.031 0.000 KDP 0.011 0.027 0.014 0.004 0.022 0.012 0.037 0.021 0.043 0.043 0.020 0.007 0.008 0.027 0.000 KL 0.011 0.0100.001 0.010 0.009 0.001 0.023 0.025 0.015 0.031 0.014 0.020 0.005 0.017 0.011 0.000 MS 0.005 0.009 0.008 0.005 0.013 0.015 0.023 0.021 0.021 0.036 0.011 0.018 0.014 0.026 0.013 0.018 0.000 RR 0.021 0.044 0.029 0.015 0.044 0.032 0.070 0.041 0.062 0.067 0.042 0.007 0.023 0.056 0.010 0.032 0.024 0.000 SM 0.006 0.017 0.012 0.003 0.019 0.021 0.036 0.020 0.033 0.046 0.022 0.004 0.017 0.029 0.008 0.015 0.009 0.0) 1 0.000 SMP 0.010 0.012 0.008 0.009 0.015 0.015 0.028 0.021 0.024 0.044 0.009 0.021 0.019 0.030 0.020 0.017 0.001 0.027 0.010 0.000 UR 0.012 0.032 0.018 0.004 0.034 0.020 0.054 0.031 0.052 0.055 0.033 0.000 0.016 0.045 0.004 0.018 0.013 0.005 0.002 0.018 0.000 WB 0.013 0.013 0.006 0.017 0.008 0.007 0.009 0.016 0.018 0.033 0.013 0.030 0.011 0.013 0.009 0.003 0.014 0.036 0.022 0.022 0.025 0.000 WD 0.000 0.012 0.004 0.000 0.011 0.011 0.016 0.015 0.022 0.038 0.009 0.011 0.016 0.014 0.009 0.007 0.001 0.021 0.003 0.001 0.008 0.010 0.000 WW 0.003 0.016 0.009 0.002 0.014 0.014 0.028 0.024 0.032 0.038 0.018 0.012 0.017 0.017 0.006 0.012 0.006 0.014 0.004 0.009 0.008 0.016 0.000 0.000 ZQ 0.017 0.040 0.027 0.016 0.041 0.028 0.060 0.035 0.055 0.058 0.038 0.015 0.019 0.054 0.013 0.028 0.019 0.009 0.018 0.0260.008 0.029 0.016 0.019 0.000

BB BG BR CL CM DD DK DKP DN DO DR DS HM KB KDP KL MS RR SM SMP UR WB WD WW ZQ

LO00 Paper II: Genetic diversity at AFLPs ae I: eei dvriy t AFLPs at diversity Genetic II: Paper

Supplementary Fig. 1. Correlograms showing kinship coefficient (F(d)) averaged over distance classes and plotted against the maximum distances of 8 distance classes from AFLPs of 21 natural populations of Hagenia abyssinica. Descriptions of plots: solid line with diamond marks = observed values, broken line with triangle marks = upper bound of 95% confidence interval, broken line with square marks = lower bound of 95% confidence interval. For population codes refer to Table 1.

Lr,00 III. Conservation genetics of African redwood (Hagenia abyssini ca (Bruce) J.F. Gmel): a remarkable but gravely endangered tropical tree species

Taye Bekele Ayele, Oliver Gailing, Reiner Finkeldey Forest Genetics and Forest Tree Breeding, Georg-August University of Goettingen, Buesgenweg 2, 37077 Goettingen, Germany

Abstract

A major challenge for the conservation of a given taxon in nature is a well-defined incorpora tion of genetic, demographic, and political criteria into decision-making processes. This paper describes genetic and demographic factors that are instrumental in planning conservation, tree improvement and domestication programs. A study is presented on a tropical tree species, Hagenia abyssinica, which is prone to extinction using morphological, chloroplast microsatel lite and AFLP markers. The analysis of variance (ANOVA) revealed a significant differentia tion among 22 natural populations of Hagenia abyssinica in all quantitative morphological traits at p<0.001. Multivariate and univariate taxonomic distances of leaf traits between popu lations are not correlated with the corresponding genetic distances (r= -0.03484, p = 0.3926), showing that the genetic differentiation at anonymous and presumably neutral AFLPs is not associated with the morphological differences among populations. The chloroplast microsatel lite data allowed us to identify lineages and to reconstruct population history by analyzing seed dispersal, while the AFLP data enabled us to identify populations of high genetic diversi ty. A weighted-score population prioritization matrix (WPPM) that combines genetic, mor phological and demographic criteria was developed and used for the first time to prioritize populations for conservation and domestication. Action is needed to launch conservation and massive plantation programs of the African redwood to ensure the long-term survival of the species and to boost its economic and ecological uses.

Key words: AFLP, chloroplast microsatellite, conservation genetics, genetic diversity, haplotypes, prioritization criteria, quantitative traits

‘Correspondence: Taye B. Ayele; e-mail: tavele@,ibc-et.org Permanent address: Institute of Biodiversity Conservation, Fax: 251-11-6613722 P.O.Box: 30726, Addis Ababa, Ethiopia;

86 Paper III: Converstion genetics

Introduction

Conservation of forest ecosystems in general and of critically endangered tree species in par ticular is a challenging task in the face of high pressure from local communities on forest land. To develop appropriate conservation strategies that {inter alia) preserve maximum genetic diversity, it is imperative to know the extent and distribution of genetic variation within a spe cies (Bawa & Krugman 1990; Loveless and Hamrick 1984). Investigation of intraspecific ge netic variation may help to assess extinction risks and evolutionary potential (fitness) in a changing world (Bawa & Krugman 1990; Hedrick 2001) and is instrumental to identify ap propriate units for conservation of rare and threatened species (Newton et al. 1999). The pre servation of germplasm in genebanks and the establishment of in situ and ex situ conservation stands requires sound knowledge of the genetic structure of a given species in order to capture the optimum genetic and demographic variations. Whereas genetic variation estimates have been used to formulate some general rules of thumb about viable population size (Franklin 1980; Lande 1995; Lynch at al. 1995), demographic analyses of individual species are more often used to assess short-term population health and to suggest management alternatives (Menges 1990; McCarthy et al. 1995). The ecological processes of migration and colonization are crucial to species survival and can have a profound impact on the spatial organization of genetic structure within and among natural populations (Husband & Barrett 1996).

Higher genetic diversity enhances a population's survival probability over ecological or evolu tionary time (Avise 2004). Small population sizes tend to reduce genetic variation, and might therefore lead to a decreased ability of such populations to adapt to ecological challenges (DeSalle & Amato 2004; Amos & Balmford 2001). When populations are few in number and small in size, the possibility of species extinction through stochastic demographic fluctuations can be of paramount immediate concern (Gilpin and Soule 1986; Hanski and Gilpin 1997). Reduced fitness may be a direct consequence of reduction in the number of heterozygous loci (Amos & Balmford 2001). On the other hand, in some endangered species (such as the north ern elephant seal), low genetic variation has not seriously inhibited population recovery from dangerously low levels (Avise 2004). Genetic inventories can provide conservationists with unprecedented precision and can add greatly to their understanding of the genetic parameters, on the basis of which many decisions are made.

87 Paper III: Conversion genetics

Assessment o f morphological traits is particularly useful to enhance tree improvement and domestication programs by identifying appropriate characters and superior traits. It also as sists in conservation decisions through the identification of population structures based on di ameter/age classes of trees. This paper describes morphological and molecular genetic varia tions in the African redwood (Hagenia abyssinica) and proposes various conservation and domestication measures. Hagenia abyssinica is a monotypic tree species of the Rosaceae fam ily that is native to Africa (Hedeberg 1989; Legesse 1995). It is an anemogamous and anemochorous broad-leaved dioecious tree species with distinctly coloured male and female flowers. Fossil pollen records suggested that Hagenia immigrated into Ethiopia from the south during the late Pleistocene (since 16,700 years Before Present (BP)) and became abundant in the southern regions of Ethiopia about 2500 years BP (Beuning et al. 1997; Bonnefille et al. 1995; Olago et al. 1999; Umer et al. 2007). The tree provided enormous timber and non-timber products and various ecological values. Hagenia has been logged heavily and selectively due to its superior timber and it is one of the endangered tree species in Ethiopia (Legesse 1995). According to the present inventory, about 7000 individuals are left in Ethiopia and only two populations out of 22 natural populations recruited young trees. Furthermore, planting efforts were very limited and were not successful in most of the cases. This is a seriously alarming situation for the genetic resources of the species, eventually leading to extinction. We present information on the amount and distribution of diversity at morphological and molecular ge netic markers with the objective to (1) identify conservation units for in situ conservation, (2) identify populations for collection of germplasm for ex situ conservation. (3) enhance domes tication and tree improvement programs.

Materials and Methods

Sampling Twenty two natural populations were sampled from diverse ecologies including closed forests (12 populations), open forests/woodlands (6 populations), wooded grasslands (2 populations), and farmlands/ homesteads (2 populations), representing most of the extant distribution of the species in Ethiopia. In addition, three plantations were also sampled. The distribution of the sampled populations in the country is illustrated in Fig. 1. Table 1 presents the characteristics of the populations investigated in this study. The distance between populations ranges from 21 to 806 km within an altitudinal range of 2200 masl at Bonga to 3200 masl at Wofwasha. Temperatures range from an absolute minimum o f -1°C at Dinsho to a maximum of 33.5 °C at Kosso Ber. Higher rainfall and lower temperatures are expected than those shown in the table Paper III: Converstion genetics as the nearest meteorological stations are situated at altitudes lower than the actual popula tions in most of the cases.

Morphological and ecological assessment The following dimensional, counted and visually observed variables were assessed from 1109 individuals (26-50 trees per population, see Table 1): total height, bole height, diameter at breast height (DBH), length of petiole, width of serrated tooth of leaf, number of leaflets, number of stipules, number of pairs of leaflets having stipules at the back of their bases, num ber of stipules between two pairs of leaflets, arrangement of leaflets, bole form/timber quality, shape of tree and sex. Twigs were harvested at random from each tree and the largest leaf was chosen for leaf measurements. In addition, distance between trees, compass bearing to next tree, total number of individuals, longitude and altitude were assessed at the population level. Distances between populations were computed from the GPS data.

Molecular inventories

Chloroplast microsatellite Three polymorphic consensus chloroplast microsatellite primers (CCMP2, CCMP6 & CCMP10) were used to screen 273 samples (9-12 individuals from each population) from twenty two natural and three planted populations of Hagenia. Details of the methods are de scribed in Taye et al. submitted (a).

AFLP A total of 596 individuals (23-24 trees/population) from twenty two natural and three planted populations of Hagenia were analysed by using the selective primer combination E41-M67 (nomenclature according to Keygene N.V. ®). Details of the methods are described in Taye et al. submitted (b).

Data analysis The program SPSS 16.0 (SPSS Inc®) was used to perform analyses of variance (ANOVA) of morphological traits and to compute taxonomic distances (as described by Sneath & Sokal 1973) from morphological data by using the Euclidean distance option. The program NTSYS- pc 2.0 (Rohlf 1998) was used to draw dendrograms and to perform Mantel tests (Mantel 1967). Mantel tests were performed between univariate or multivariate taxonomic distances of Paper III: Converstion genetics morphological traits and Nei’s genetic distances among populations. Similarly, the association between average taxonomic distances of morphological traits between populations and Euclidean distances of climatic variables between populations was tested. Molecular genetic data analysis softwares PermutcpSSR (available at http://www.pierroton.inra.fr/genetics/labo/Software/PennutCpSSR/index.html. accessed on 3 February 2008) and ARLEQUIN Version 3.0 (Excoffier et al. 2005; available at http://cmpg.unibe.ch/software/arlequin3. accessed on 10 February 2008) were used to analyze the cpSSR data, while ARLEQUIN Version 3.0, AFLP-SURV (Vekemans et al. 2002, avail able at http://www.ulb.ac.be/sciences/lagev/. accessed on 2 March 2008) and NTSYS-pc 2.0 (Rohlf 1998) were used to analyze the AFLP data. A weighted-score population prioritization matrix (WPPM) that combines genetic, morphological and demographic criteria was devel oped and used for the first time to prioritize populations for conservation and domestication.

oi£bK P

Fig. 1 The distribution of populations of Hagenia abyssinica showing the two chloroplast lineages observed in Ethiopia (Taye et al. submitted (a)). Square-dotted enclosure shows lineage I; long-dashed enclosure shows li neage II. Small filled-circles indicate the locations of the populations; population codes are provided in Table 1. Base map: Assefa Guchi (unpublished).

90 Paper III: Converstion genetics

Table 1 Description of Hagenia populations sampled from the mountains of Ethiopia showing some measures of genetic diversity

Populations Code Lat. Long. M asl ARF Min T Max T n N H He Debark-Mariam DK 1 3 °ir 37°57' 3013 1270 8.8 19.7 26 26 4 0.217

Debark- DKP 13°12' 38o01' 3005 1270 8.8 19.7 50 - 4 0.226 Plantation

Kimir-Dingay KDP 11°48' 38°14' - 1350 9.2 21.9 30 1,6 0.183 plantation Woldiya Se’at WD 11°55' 39°24’ 3112 908 na na 50 120 6 0.194 Michael

Kosso Ber KB 10°59' 36°54' 2702 2381 12.9 27.4 50 60 1 0.206 Denkoro DR 10°52' 38°47’ 3061 896 10.9 21.8 50 60 6 0.189

Wonbera WB 10°34' 35°41' 2428 1622 na na 50 60 5 0.211

Wof washa ww 09°45’ 39°44' 3159 941 6.1 19.9 30 45 1 0.189

Chilimo CM 09°05’ 38°10' 2805 1114 11.5 25.8 50 65 1,4 0.192

Dindin DN 08°36' 40°14' 2410 989 12.7 28,0 30 55 1 0.212

Zequala Abo ZQ 08°32' 38°50' 2856 1215 na na 33 60 1 0.205 Boterbecho BB 08°24' 37°15' 2772 1666 5.7 23.6 50 60 1 0.213 Chilalo CL 07°56' 39°11’ 2815 796 9.8 23,0 50 70 1 0.177

Sigmo plantation SMP 07°55' 36°10' 2300 1837 11.4 21.6 30 - 4 0.146

Sigmo SM 07°46' 36°05' 2651 1837 11.4 21.6 30 60 4 0.170

Munesa MS 07°25' 38°53' 2459 1028 10.1 24.3 50 80 1 0.200

Bonga BG 07°17' 36°22' 2238 2217 11.9 26.6 50 80 4 0.198 Kofele KL 07°11' 38°52’ 2757 1305 7.7 20.1 50 110 1,2,3 0.195

Dinsho DO 07°05' 39047- 3117 1213 3.4 20.8 50 260 2 0.362

Doddola-Serofta DS 06°52' 39°02’ 2700 1074 6.7 24.3 50 75 2 0.139

Doddola- DD 06°52' 39°14' 3039 1074 6.7 24.3 50 5000 2 0.173 Dachosa Rira RR 06°45' 39043- 2725 736 11a na 50 170 2 0.169

Bore BR 06°17' 38°39' 2631 1526 8.3 18.8 50 100 1,2,3 0.187 Uraga UR 06°08' 38°33' 2508 1228 8.3 18.8 50 70 2,3 0.160 HagereMariam HM 05°51' 38°17' 2443 1228 12.3 23,0 50 55 2 0.168 Total/mean 1109 6741 0.195

M asl= meters above sea level; ARF = Mean annual rainfall in milliliters; Min T = Mean minimum temperature (C°); Max T = Mean maximum temperature; n= no. of samples assessed for morphological characters; N= Popu lation size (estimation of total no. of individuals), H = chloroplast haplotypes (Taye et al. submitted (a)); He = gene diversity (Taye et al. submitted (b)). Population codes will be used throughout the paper.

91 Paper III: Conversion genetics

Results and discussions

Morphological diversity The morphological traits observed in Hagenia abyssinica were highly variable among popula tions. Supplementary Table 1 summarizes the mean values of morphological traits observed within populations. The DBH-based population structure (Supplementary Fig. 1) indicated that m ost of the populations (59%) fall under a J-shaped distribution pattern, indicating no/bad reproduction and no/poor recruitment. Populations Munesa (MS) and Wonbera (WB) show nearly complete coverage of diameter classes that was close to normal distribution. Population Chilimo (CM) also shows a nearly complete representation of diameter classes but deviated from a normal distribution. Population Zequala (ZQ) demonstrated an example of a U-shaped distribution that is an indication of a selective removal of middle diameter class trees. In gen eral, Hagenia exhibited unsatisfactory population structure as several diameter classes were missing from the majority of the populations. Natural regeneration was observed in only two populations - Bonga (BG) (112 wildings) and Boterbecho (BB) (5 saplings). A one-way analysis of variance (ANOVA) revealed a significant differentiation among the 22 natural populations of Hagenia abyssinica in all morphological traits at p<0.001 (Table 2). Large proportion of variation (>65%) is allocated within populations for all traits. The highest per centage of variation among populations was observed for DBH (34.1%) while the lowest was observed for the number of leaflets (9% ).

The cluster analysis based on the average taxonomic distances matrix of all leaf traits grouped the populations into two main clusters and separated four outlier populations (Fig. 2). In gen eral, no clear association between geographic regions and taxonomic distances could be ob served. Both main clusters are composed of populations from the main distribution areas of the species. The average multivariate taxonomic distances of all morphological traits in our dataset did not show any correlation with the average Euclidean distances of climatic vari ables (r = 0.17062, p = 0.9281), suggesting that the observed morphological traits are not in volved in the adaptation to different climatic conditions. Similarly, separate tests of associa tion of taxonomic distances of individual morphological traits with the Euclidean distances of climatic variables did not show any correlation (not shown). The pronounced divergence of quantitative morphological traits is likely to be due to different age structures, stand histories and edaphic factors, which were, however, not assessed. It may also reflect different physio logical responses to changes in the environment.

92 Paper III: Conversion genetics

Table 2 Analysis of variance (ANOVA) of morphological traits among 22 natural populations of Hagenia abyssinica.

Source of variation SS df MS F % varia Is tion Tree height between populations 6601.2 21 314.3 17.5 28.5 Within populations 16536.5 919 18.0 71.5 Bole height between populations 1617.2 21 77.0 11.2 20.3 Within populations 6342.8 919 6.9 79.7 DBH between populations 414377.9 21 19732.3 22.7 34.1 Within populations 799144.1 919 869.6 65.9 Petiole length between populations 1175.1 21 56.0 10.7 19.7 Within populations 4780.6 918 5.2 80.3 No. leaflets between populations 198.2 21 9.4 4.3 9.0 Within populations 2011.5 918 2.2 91.0 No. of stipules between populations 10351.1 21 492.9 16.2 27.0 Within populations 27977.3 918 30.5 73.0 No. of back stipules between pops 133.9 21 6.4 10.2 24.2 Within populations 420.3 673 0.6 75.8 Tooth width between populations 49.6 21 2.4 7.5 16.7 Within populations 247.6 786 0.3 83.3 df = degrees of freedom, SS = sum of squares, MS= mean sum of squares, F = computed F value, Is = level of significance, **** = highly significant at pO.OOOl.

Molecular genetic diversity

Chloroplast microsatellites Six haplotypes that were phylogenetically grouped into two lineages were identified from the combination of 8 variants from the three loci (Table 1, Fig. 1). The observed haplotypes showed a strong geographic pattern as a result of highly restricted gene flow via seeds and a rare occurrence of long-distance seed dispersal. The two lineages were separated by an indel (insertion/deletion) of 10 nucleotides in locus CCMP2. The first lineage contains haplotypes H4, H5 & H6, which are distributed in the south-western and northern regions, while the sec ond lineage contains haplotypes H 1, H2 & H3 in central and southern regions. A remarkable subdivision of cpDNA diversity in the species was found, as indicated by a high level of ge netic differentiation (G s t = 0. 899, N s t = 0. 926). Also, the non-hierarchical analysis of mole cular variance (AMOVA) showed that 92.3% of the total genetic diversity is represented among populations (Taye et al. submitted (a))

93 Paper III: Conversion genetics

BB(C) DK(A) WB(A) DR(A) CL(B) KB(A) } BR(D) DD(D) MS(D) Ed - BG(C)- K L(D )- DN(B) - WW(B) - DO(D) - WD(A)- UR(D) ■ D S(D )- H M (D)- ZQ(B)- SM (C)- CM(B)- RR(D)------' i------1-1------1------1------1- 1------1------1------1------1- i------1------1------1------1------1------1------1------1 065 249 433 8.18 8M Euclidean distance (leaf characters) Fig. 2 UPGMA cluster diagram drawn based on average taxonomic distances computed from five leaf characters of 22 natural populations of Hagenia abyssinica from Ethiopia. Population codes follow Table 1. Letters in parenthesis indicate geographic regions: A= northern region, B= central region, C= southwestern region, D= southern region.

AFLP Moderate to high gene diversities were observed at AFLP loci ranging from 0.139 at Dodola- Serofta (DS) to 0.362 at Dinsho (DO) (Table 1). Interestingly, the lowest gene diversities were recorded for the two farmland populations (DS & UR) while the maximum gene diver sity was recorded for a well-protected Park Forest (DO), pointing to negative human impact on genetic diversity. The second largest population (DO) demonstrate remarkably high gene diversity (36.2%), reflecting strong divergence from the rest of the populations. The mean gene diversity in subdivided populations of Hagenia abyssinica showed high within- population variation (0.195) and moderate but significant population differentiation (F St = 0.077) (Taye et al. submitted (b)). The largest remaining population (DD) does not show the highest genetic diversity whereas much smaller populations show higher diversity (Table 1). Ten out of 21 natural populations (KB, CM, BB, DO, DS, HM, KL, DK, CL and RR) showed significant spatial genetic structure (SGS) (Taye et al. submitted (b)).

The non-hierarchical analysis of molecular variance (AMOVA) performed for all populations at AFLP markers revealed that 10.4% of the total variation is represented among populations. Paper III: Conversion genetics

The phylogenetic tree drawn from Nei’s (1978) genetic distances using the Unweighted Pair Group Method with Arithmetic mean (UPGMA) clustering method congregated the 22 natural populations into two major clusters (Fig. 3). The dendrogram reflects a weak spatial distribu tion pattern of the populations. Likewise, a test of association (Mantel test) between the mul tivariate taxonomic distances of combined leaf traits and genetic distances did not show any correlation (r = -0.03484, p = 0.3926). Similarly, separate tests for association of the taxo nomic distances of individual morphological traits (including growth traits) with the genetic distances did not show any correlation (not shown). This result suggests that the genetic dif ferentiation at anonymous AFLP markers is not associated with the morphological differences among populations.

Despite the recent immigration into Ethiopia and small population sizes, Hagenia exhibited moderate to high gene diversity within populations. Since even little effective pollen per gen eration is sufficient to counteract loss of genetic diversity (Wright 1931, Finkeldey & Hatte- mer 2007), the effect of recent colonization and of the small population sizes is not reflected in the levels of gene diversity. Likewise, studies on many rare and endangered animal species such as the spring-dwelling fish (Gambusia nobilis), przewalski’s horse (Equus przewalskii), manatee ( Trichechus manatus) and Stephens's kangaroo rat (Dipodomys stephensi) revealed more or less average levels of genetic variation due to effective mating (reviewed by Avise 2004).

Conservation priorities

We have estimated the total number of the extant individual Hagenia trees throughout the country as not exceeding 7,000 (including the estimation of scattered trees that were not in cluded in the present study), the majority of which are old and dying without recruiting young generations. The three plantations included in the present study are small in size amounting to about 450 individuals in total. There are no records on the existence of large plantations of Hagenia in Ethiopia. Given the present open-access to most of the populations and lack of natural regeneration, Hagenia will unquestionably face extinction in the following decades. Bonga is the only viable population that recruited new generation in southwest of Ethiopia, but with only 80 mature individuals left at the time of this survey. The largest remaining pop ulation (DD) does not have natural regeneration. The fossil pollen stratigraphy from African countries other than Ethiopia also showed that the fossil pollen accumulation of Hagenia

95 Paper III: Conversion genetics abyssinica has declined alarmingly through time (Beuning et al. 1997; Bonnefdle et al. 1995; Olago et al. 1999), suggesting a dramatic reduction in the size of the populations and probable local extinction in some locations (Taye et al. submitted (a)). Surprisingly, despite the evident severe threat on its survival, Hagenia is not included in the red list of the International Union for Conservation of Nature (IUCN) whereas Juniperus procera, which is in a comparatively better conservation status than Hagenia in terms of geographic range, population size and re cruitment of young trees (field observation during the preset survey), was red-listed (http://www.iucnredlist.org/, accessed on 26 May 2008).

Fig. 3 Phylogenetic tree drawn based on Nei’s (1978) genetic distances computed by UPGMA clustering from AFLPs of 22 natural populations of Hagenia abyssinica from Ethiopia. Popu lation codes follow Table 1. Letters in parenthesis designate geographic regions: A= northern region, B= central region, C= southwestern region, D= southern region.

A number of organizations foster the conservation of biodiversity in general and that of threatened species in particular at the global level. The major objectives of the Convention on Biological Diversity (CBD) are the conservation of biological diversity, the sustainable use of its components, and the fair and equitable sharing of the benefits arising out of the utilization of genetic resources (http://www.cbd.int/convention/. accessed on 2 June 2008). The IUCN's Red List Criteria are based on available evidence concerning the numbers, trend and distribu tion of a given species based on changes over periods of time (IUCN, 2001). Such compre hensive time-bound information is lacking in Ethiopia. The Convention on International Trade in Endangered Species (CITES) regulate the complex wildlife trade by controlling species-

96 Paper III: Conversion genetics specific trade levels on the basis of biological criteria (http://www.cites.org/. accessed on 3 June 2008). CITES classified species as prohibited (Appendix I), restricted (Appendix II) and optional (Appendix III) for trade, but Hagenia is not included in any of them. There is a lose end, though, in the convention of CITES, which provides exceptional cases that allow utiliza tion of endangered species categorized even under Appendix I. The Endangered Species Act of the United States of America defined a species as endangered if it is at risk of extinction throughout all or a significant portion of its range, and to be threatened if it is likely to be come endangered in the foreseeable future (http://www.nmfs.noaa.gov/pr/pdfs/laws/esa.pdf. accessed on 2 June 2008).

Given the scanty financial resources the country has and the complex nature of conservation, it is indispensable to prioritize populations of Hagenia for conservation. The need to integrate demographic and genetic criteria in plant conservation has been recognised during the last two decades (e.g., Lande 1988; Ostermeijer et al. 2003; DeSalle and Amato 2004; Delgado et al. 2008; Hoebee et al. 2008). Delgado et al. (2008) standardized phylogenetic, demographic and genetic values to obtain conservation indices for populations of Mexican rare pines. We pro pose a “weighted-score population prioritization matrix” (WPPM), a method that integrates genetic, morphological and demographic criteria to prioritize populations of a single species for conservation and domestication purposes. Our method is similar in approach but different in criteria and scoring from that of Delgado and co-workers (2008) in several ways. Our me thod 1) uses actual values instead of standardized values as our target is a single species, 2) uses morphological and additional demographic criteria, 3) prioritizes populations for differ ent conservation measures and domestication, 4) accords different weights to demographic and genetic criteria for each of the measures, 5) uses only one distance measure, i,e., average genetic distance (computed from Nei’s genetic distances) as suggested by O’Meally and Col- gan (2005) instead of branch-node lengths of a phylogenetic tree, 6) uses information on chlo roplast lineages as a complementary criterion. The values from all criteria will be summed up and the population with the highest value will be accorded the top priority and so forth. It is a simple and straight-forward tool that can easily be understood and applied by forestry experts and decision makers to prioritize populations of a given taxon for in situ conservation, ex situ conservation, and tree improvement and domestication programs. The genetic criteria em ployed are the amount of genetic diversity (He) and average genetic distance (AGD) from the AFLP data (Taye et al. submitted (b)) while the demographic criteria included status of natu ral regeneration, DBH-based population structure, present conservation status, total popula

97 Paper III: Converstion genetics tion size, and timber quality measured as mean bole height. More details are provided in Sup plementary Table 2. The weight attributed to genetic, morphological and demographic criteria differs according to the aim of the prioritization (Supplementary Table 2). Presence of natural regeneration and other demographic factors are comparatively more important to conservation in situ than the genetic criteria whereas the reverse applies for conservation ex situ. Similarily, morphological criteria (particularly timber quality) deserve more weight than the genetic crite ria to select superior trees for domestication and tree improvement programs. Accordingly, the weights of genetic criteria for in situ conservation, ex situ conservation, and for tree improve ment and domestication programs are set in the order of 40%, 80% and 40%, respectively. The remaining proportions in each program are accorded to the demographic/morphological criteria. The information from the chloroplast microsatellite data (Taye et al. submitted (a)) is considered after the outcome of the prioritization in order to represent populations in different chloroplast haplotypes/lineages.

Outcome o f prioritization The supplementary Table 3 (a-c) summarizes the results of the prioritization of the extant populations of Hagenia for in situ conservation, ex situ conservation, and for tree improve ment and domestication purposes. The top two priority populations selected for in situ conser vation are Bonga (with the largest natural regeneration and high genetic diversity) and Dinsho followed by Boterbecho at the third position. Populations Kosso Ber and Zequala equally fol lowed in the fourth position. Since only Bonga represented chloroplast lineage I in the top priority list, population Wonbera from the same lineage, which stood sixth in the rank, should be given at least the fourth priority for in situ conservation. The top candidate population for ex situ conservation is Dinsho (with the highest genetic diversity but no natural regeneration) followed by Kosso Ber, Dindin and Zequala equally at the second position, and Debark at the third position. Wonbera and Boterbecho shared the fourth rank. This top priority list is domi nated by populations from chloroplast lineage II. Therefore, population Wonbera from lineage I should also be considered for ex situ conservation. The top three candidate populations se lected for tree improvement and domestication programs are Kosso Ber, Dinsho and Bore fol lowed by Wonbera, Dindin, Zequala and Boter Becho. This priority list should be maintained because timber quality matters most (at least at present) than the other criteria for tree im provement and domestication programs. Regardless of the differences in the types and/or weights of the criteria, the populations Kosso Ber, Zequala, Dinsho and Boterbecho appeared in the top four ranks of all the three objectives while populations Wonbera and Dindin are se

98 Paper III: Converstion genetics lected for both ex situ conservation, and tree improvement and domestication programs. Spe cial consideration should be given to populations that are chosen for multiple objectives. A multiple population breeding strategy that combines breeding goals and conservation (Nam- koong 1984) is particularly useful in this regard. Some purposefully weighted criteria played influential roles to select populations for different objectives. For example, the presence of natural regeneration influenced the selection of populations for in situ conservation (Bonga and Boterbecho) while the amount of gene diversity was crucial to choose populations for ex situ conservation (Dinsho). Similarly, the criterion bole height was vital to choose populations for tree improvement and domestication. The largest remaining population (DD) does not ap pear in the top priority lists for in situ and ex situ conservation because it does not have natu ral regeneration and has lower genetic diversity than others. Also, it is located in a protected area. But it appeared in the 6th priority for tree improvement and domestication because it har bors good quality timber. The production and recruitment of young trees is often overlooked as a key criterion; but it is essential for the success of gene conservation in situ.

Conclusions and recommendations

Distinctive quantitative traits were observed in Hagenia. The chloroplast microsatellite data allowed us to identify lineages and to reconstruct population history by analyzing seed disper sal while the AFLP data allowed us to identify populations of higher genetic diversity. The morphological data enabled us to identify populations of desirable quantitative traits that can be used in conservation and domestication of the species. The sizes of the extant populations were reduced to very small patches due to human impact, probably affecting the genetic struc ture and increasing the risk of extinction. The absence of natural regeneration in most of the populations, the small sizes of all but one (Doddola-Dachosa) populations and the current high demand and pressure from the people for Hagenia lumber are main reasons to regard the species as prone to extinction at least in Ethiopia in the following decades. Action is needed to launch conservation and massive plantation programs of this remarkably valuable but gravely endangered tree species. The work presented here might serve as a starting point to select ge netic resources and superior individuals of Hagenia. The priority rank should be considered taking into account the availability of resources for conservation, tree improvement and do mestication programs. The populations that lack natural regeneration but selected for conser vation in situ should be enriched by planting seedlings raised from the same stand. Conserva Paper III: Conversion genetics tion decisions depend on a number of factors that go beyond scientific information. A major challenge for the conservation of the genetic resources of Hagenia abyssinica will be the well- defined incorporation of social, cultural and political criteria into the decision-making processes. Seed collection for ex situ conservation and tree improvement programs should consider the information on spatial distribution of genetic structure (SGS) described in Taye et al. (submitted (b)) to minimize collection of seeds from related individuals. In conclusion, the present work allowed us to establish priorities for the conservation and domestication of the African redwood based on genetic, morphological and demographic information. This in formation can serve as a benchmark for monitoring its conservation status in the future. In vestigation into the possible impediments to natural regeneration including, inter alia, the eco logical (moisture, soil, animal browsing) and physiological characteristics (seed quality cha racters and viability) of the relict populations of Hagenia abyssinica is crucial to ensure the long-term survival of the species. Common garden experiments and the establishment of comprehensive provenance trials may help to reexamine the association between morphologi cal and molecular genetic traits by separating the genetic differences from non-genetic envi ronmental effects at important adaptive and economic traits. Similar work is recommended in other African countries where Hagenia is known to grow. International organizations such as IUCN and CITES should consider Hagenia in their appropriate databases/programs as it is at high risk of extinction.

Acknowledgements

This project is a component of the “Support to the Forest Genetic Resources Conservation Project” of the Ethiopian Institute of Biodiversity Conservation (IBC) supported by the Ger man Federal Ministry of Economic Cooperation and Development (BMZ) through the Ger man Technical Cooperation (gtz). The German Academic Exchange Service (DAAD) ex ecuted the grant as a PhD project of the first author. The National Meteorological Service Agency of Ethiopia provided climatic data. We thank Oleksandra Dolynska, Thomas Seliger and Olga Artes for kindly assisting in the laboratory and Daniel Bekele for helping during the fieldwork.

100 Paper III: Conversion genetics

References

Aalbsek A (1993) Tree seed zones for Ethiopia. National Tree Seed Project, Addis Ababa Amos M, Balmford A (2001) When does conservation genetics matter? Heredity 87: 257-265 Avise JC (2004) Molecular Markers, Natural History and Evolution, 2nd edn. Sinauer Associ ates Inc, Sunderland, MA Bawa KS, Krugman SL (1990) Reproductive biology and genetics of tropical trees in relation to conservation and management. In: Rain Forest Regeneration and Management (eds Gomez-Pampa A, Whitmore TC, Hadley M), The Parthenon Publishing Group, pp. 119- 136 Beuning KRM, Talbot MR, Kelts K (1997) A revised 30000-year paleoclimatic and paleohydrologic history of Lake Albert, East Africa. Palaeogeogr Palaeocl 136: 259-279 Bonnefille R, Riollet G, Buchet G, Icole M, Lafont R, Arnold M, Jolly D (1995) Gla cial/interglacial record from intertropical Africa, high resolution pollen and carbo data at Rusaka, Burundi. Quaternary Sci Rev 14: 917-936 Delgado P, Eguiarte LE, Molina-Freaner F, Alvarez-Buylla ER, Pinero D (2008) Using phy logenetic, genetic and demographic evidence for setting conservation priorities for Mex ican rare pines. Biodivers Conserv 17: 121-137 DeSalle R, Amato G (2004) The expansion of conservation genetics. Nat Rev Genet 5: 702- 712 Excoffier L, Laval G, Schneider S (2005) Arlequin (version 3.0): An integrated software package for population genetics data analysis. Evolutionary Bioinformatics Online, 1: 47- 50 Finkeldey, R., and Hattemer, H.H. 2007. Tropical Forest Genetics. Springer-Verlag, Berlin. Franklin IR (1980) Evolutionary change in small populations. In Soule MA & Wilcox BA (eds) Conservation biology: an evolutionary-ecological perspective. Sinauer Associates, Sunderland, MA, pp 135-149 Hedeberg O (1989) Rosaceae. In: Inga Hedeberg and Sue Edwards, eds. Flora of Ethiopia, Vol. 3, Pittosporaceae to Arralaceae. Addis Abeba and Asmara, Ethiopia; Uppsala, Swe den Hedrick PW (2001) Conservation genetics: where are we now? Trends Ecol Evol 16: 629-636 Hoebee SE, Thrall PH, Young AG (2008) Integrating population demography, genetics and self-incompatibility in a viability assessment of the Wee Jasper Grevillea (Grevillea iaspi- cula McGill., Proteaceae). Conserv Genet 9: 515-529 Husband BC, Barrett SCH (1996) A metapopulation perspective in plant population biology. J Ecol 84: 461-469 IUCN (2001) IUCN Red List Categories and Criteria: Version 3.1. IUCN Species Survival Commission, Gland, Switzerland and Cambridge, UK Gilpin MA, Soule ME (1986) Minimum viable populations: Processes of species extinction. In Soule ME (ed) Conservation biology: The Science of scarcity and diversity. Sinauer Associates Inc, Sunderland, MA Hanski IA, Gilpin MA (eds) (1997) Metapopulation biology: Ecology, genetics and evolution. Academic press, New York Lande R (1988) Genetics and demography in biological conservation. Science 241: 1455- 1460 Lande R (1995) Mutation and conservation. Conservation biology 9: 782-791 Legesse N (1995) Indigenous trees of Ethiopia: Biology, uses and propagation techniques. SLU, Reprocentralen, Umea Loveless MD and Hamrick JL (1984) Ecological determinants of genetic structure in plant populations. Ann Rev Ecol Syst 15: 65-95

101 Paper III: Converstion genetics

Lynch M (1996) A quantitative genetic perspective on conservation issues. In: Avise JC, Ha mrick JL (eds) Conservation genetics: case histories from nature. Chapman & Hall, USA, pp 471-501 Lynch M, Convey J and Burger R (1995) Mutation accumulation and the extinction of small populations. Am Nat 146: 489-518 Mantel NA (1967) The detection of disease clustering and a generalized regression approach. Cancer Res 27: 209-220 Menges ES (1990) Population viability analysis for an endangered plant. Conser Biol 4: 52-64 McCarthy MA, Burgman MA and Ferson S (1995) Sensitivity analysis of models population viability. Biol Conserv 73: 93-100 Namkoong G (1984) A control concept of gene conservation. Silvae Genet 33: 160-163 Nei M (1978) Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89: 583-590 Newton AC, Allnutt TR, Gillies ACM, Lowe AJ and Ennos RA (1999) Molecular phy logeography, intraspecific variation and the conservation of tree species. Trends Ecol Evol 14: 140-145 Olago DO, Street-Perrott FA, Perrott RA, Ivanovich M, Harkness DD (1999) Late quaternary glacial-interglacial cycle of climatic and environmental change on Mount Kenya. J Afr Earth Sci 29: 593-618 O’Meally D, Colgan DJ (2005) Genetic ranking for biological conservation using information from multiple species. Biol Conser 122: 395-407 Oostermeijer JGB, Luijten SH, den Nijs JCM (2003) Integrating demographic and genetic approaches in plant conservation. . Biol Conser 113: 389-398 Rohlf FJ (1998). NTSYS-pc: Numerical taxonomy and multivariate analysis system (Version 2.0). State University of New York, USA Sneath PHA and Sokal RR (1973) Numerical taxonomy: the principles and practice of numer ical classification. WH Freeman and Company, San Fransisco Umer M, Lamb HF, Bonnefille R, Lezine A-M, Tiercelin J-J, Gibert E, Gazet J-P, Watrin J (2007) Late Pleistocene and Holocene vegetation history of the Bale Mountains, Ethiopia. Quaternary Sci Rev 26: 2229-2246 Vekemans X, Beauwens T, Lemaire M and Roldan-Ruiz I (2002) Data from amplified frag ment length polymorphism (AFLP) markers show indication of size homoplasy and a rela tionship between degree of homoplasy and fragment size. Mol Ecol 11: 139-151 Wright S (1931) Evolution in Mendelian populations. Genetics 16: 97-159.

102 Supplementary Table 1. Mean values (with standard deviations) for morphological traits observed within the populations of Hagenia in Ethiopia.

Populations D (m ) SI TH (m) BH(m) DBH (cm) LP (cm) NL NS NLBS WT (mm) Debark Mariam 31.0 1 9.6±2.6 2.1±1.4 67.9±31.7 12.4±2.2 14.2±1.2 16.6±4.5 1.4±0.7 1.9±0.6 Debark Plantati na on na na Na na 7.5±1.6 13.0±1.6 22.5±5.7 1.2 ±0.8 2 .2±0.6 Kimir Dengay na Plantation na na Na na 10.8±2.4 12 .8± 1.8 15.7±5.7 0 .6 ±0.8 2 .2±0.6 Woldiya Se’at na Michael na 13.5±2.1 3.5±0.7 114.5±14.8 14.0±2.1 13.8±1.4 18.7±4.6 1.7±0.6 1.8±0.7 Kosso Bcr 9.3 na 1 1.0±2.9 6.4±2.3 24.2±9.7 14.7±3.2 14.8±1.7 17.4±5.2 1.1 ±0.9 2.3±0.6 Denkoro 21.1 0.7 12.9±2.5 4.1±2.2 82.3±31.3 12.7±2.4 13.9±1.7 16.8±4.9 0.6±0.7 2.0±0.7 W onbera 13.2 na 12.3±2.8 4.7±2.4 34.6±15.7 12.3±1.8 14.8±0.9 17.1±4.8 1,4±0.6 1.9±0.5 W ofwasha 27.3 na 7.8±2.7 2.7±1.7 54.5±36.2 14.4±2.4 14.5±1.4 21.7±5.5 1.6 ±0.6 2.7±0.7 Chilmo 28.0 na 17.0±9.2 5.2±4.0 43.1±25.5 1 1.8±2.5 13.6±2.1 9.6±2.2 1.3±0.9 2.4±1.1 Didndin 29.3 na 15.4±5.8 5.9±2.6 64.7±47.2 14.2±2.9 14.9±2.3 20.5±7.3 1.3±0.9 2.2±0.5 Zequala Abo 18.7 0.3 13.1±6.4 4.5±3.2 52.7±57.4 9.0±2.3 11.0± 1.8 18.0±6.1 1.0±0.9 2 .0± 0.6 Boter-Becho 15.0 1.2 12.6 ±6.2 5.8±2.8 32.5±25.3 11.9±2.3 14.0±1.4 16.9±6.2 1.2± 0.8 1.9±0.5 Chilalo 31.1 1.8 10.4±2.7 3.5±1.6 54.5±30.0 13.5±2.2 14.9±1.1 17.7±5.8 1,8±0.5 1.8±0.6 Sigmo Plantation na na na na na 12.1± 1.8 14.8±1.6 22.3±5.6 1.3±-0.9 2 .2± 0.6 Sigmo 22.7 1 17.0±5.3 5.5±2.8 80.2±33.7 13.3±2.6 13.7±1.8 24.8±7.3 1.5±0.9 2.5±0.8 M unesa 18.3 na 11.5±2.9 4.7±2.8 40.5±19.5 14.6±2.4 14.4±1.4 15.2±5.9 1,8±0.5 2.2±0.7 Bonga 22.6 0.9 13.4±2.6 5.3±2.2 44.3±13.8 1 1.8± 2.2 14.0±1.8 21.3±6.8 1. 1±0.8 2.4±0.6 ae II Cnesin genetics Converstion III: Paper Kofele 22.5 1.8 14.6±3.7 5.8±3.3 99.2±32.8 10.4±1.7 14.0±1.1 20.4±4.5 1.4±0.8 1.8±0.5 Dinsho 22.7 3 16.1 ±2.4 3.3±2.3 85.2±24.7 14.4±2.0 14.2±1.1 19.0±4.3 1.6±0.6 2.4±0.8 Dodola-Serofta 21.9 2.9 19.5±2.9 6.7±2.4 88.6±38.5 12.4±1.8 14.7±0.8 19.2±4.6 1.7±0.6 3.0±0.6 Dodola-Dachosa 18.3 na 15.7±4.8 5.0±3.3 84.2±40.3 13.7±2.6 14.4±1.4 16.0±4.8 1.8±0.5 2.5±0.6 Rira 19.3 na 15.2±4.7 4.2±2.8 66.5±31.3 13.3±2.4 14.0±1.2 12.9±4.9 1.3±0.7 2.5±0.7 Bore 24.5 1.1 17.3±3.4 7.2±2.5 64.4±16.1 13.2± 1.7 14.6±1.1 15.9±5.2 2.1±0.7 1.6± 0.6 Uraga 171.3 0.7 14.4±2.0 7.1±2.3 49.7±16.1 13.2±2.2 13.5±1.7 19.4±6.7 l.ldb0.8 1.9±0.5 HagereMariam 28.7 0.9 12.7±3.2 7.0±2.5 54.3±24.6 12.8± 2.1 15.0±1.3 19.8±6.8 1.9±0.9 1.7±0.6 Average 13.8±2.80 5.0±1.44 62.8±23.16 12.6±1.74 14.0±0.86 18.2±3.24 1.4±0.37 2.2±0.34 D= distance between trees, SI= sex index (relative number of male to female), TH= total height, BH= Bole height, DBH= diameter at breast height, LP= maxi- — mum length of petiole, NL= no. of leaflets, NS= No. of stipules, NLBS= No. of pairs of leaflets with stipules at the back of their bases, WT= width of serrated uj edges of leaf tooth. Paper III: Conversion genetics

Supplementary Table 2. Weighted-score population prioritization matrix (WPPM) a) for in situ conservation Critera Weight Score3 Product (out of 10) 0 1 2 3 4 5 (weight x score) 1 Genetic criteria 4

1.1 Within-population gene diversi 3 ty (He) 1.2 Average genetic distance 1 (AGD) 2 Demographic criteria 6 2.1 Status of natural regeneration 3 2.2 DBH-based population structure 1

2.3 Present conservation status 1

2.4 Population size 1 sum b) for ex situ conservation (seed bank & ex situ conservation stands) Critera Weight Score3 Product (out of 10) 0 1 2 3 4 (weight x score) 1 Genetic criteria 8

1.1 Within-population gene diversity 5 (He) 1.2 Average genetic distance (AGD) 3 2 Demographic criteria 2 2.1 Population size 2 sum c) for tree improvement and domestication programs Weight Score3 Product Critera (out of 10) 0 1 2 3 4 (weight x score) 1 Genetic criteria 4

1.1 Within-population gene diver 3 sity (He) 1.2 Average genetic distance 1 AGD) 2 Demographic criteria 6

104 Paper III: Conversion genetics

2.1 Boleform/height 5 2.2 Population size 1 sum aA box under the appropriate score is crossed based on the evaluation of the population based on the predetermined weights and scores of each criterion.

Description of scoring

1. Amount of genetic diversity (AFLP)

1.1 Within-population gene diversity

Scores 1 = Mean population gene diversity values (He) less than the overall mean minus 15% of the overall mean 2 = He greater than or equal to the mean minus 15% of the mean, less than mean minus 5% of the mean 3 = He greater than or equal to the mean minus 5%, less than the mean plus 5% of the mean 4 = He greater than or equal to the mean plus 5% of the mean, less than the mean plus 15% of the mean 5 = He greater than the mean plus 15% of the mean

1.2 Average genetic distances (AGD)

1 =0.010-0.019 2 = 0.020-0.029 3 = 0.030-0.039 4 = 0.040-0.049

2. Status of natural regeneration (total count around the sample trees)

0 = No regeneration 1 = low (1-10) wildings

2 = fair (10-25) wildings 3 = good (25-50) wildings 4 = high (>50) wildings

3. DBH-based population structure

1= > 6 DBH classes missing 2 = 5-6 DBH classes missing 3 = 3-4 DBH classes missing

105 Paper III: Converstion genetics

4 = 1-2 DBH class missing 5 = complete distribution

4. Present conservation status (refers to the pressure from the surrounding community and the current level of protection)

0= well-protected - not threatened at the moment 1 = fair protection - but vulnerable 2= open access -endangered 3 = open access - gravely endangered

5. Population size (y)

1= y < 50 2 = 50 < y >150 3 = 150< y > 250 4 = 250< y > 350 5 = y > 350

6. Bole quality measured as mean bole height (z)

1 = z < 4 m 2 = 4< z < 6 m 3 = z > 6 m

106 Supplementary Table 3 Summary of the results of the prioritization of the extant populations of Hagenia for in situ conservation, ex situ conserva tion, and for tree improvement and production purposes. a) Prioritization of Hagenia populations for in situ conservation

Criteria DK WD KB DR WB WW CM DN ZQ BB CL SM MS BG KL DO DS DD RR BR UR HM Amount of He 12 9 12 9 12 9 9 12 12 12 6 6 9 9 9 15 3 6 6 9 3 6 gene diver sity AGD 2 1 2 1 1 1 1 2 2 1 1 1 1 1 1 4 2 1 3 1 2 1

Status of natural re 0 0 0 0 0 0 0 0 0 3 0 0 0 12 0 0 0 0 0 0 0 0 generation DBH-based populati 1 1 5 1 3 3 4 3 3 4 4 2 5 3 1 1 1 2 3 1 3 4 on structure

Present conservation 1 1 1 1 2 1 1 2 3 1 3 2 1 2 2 0 1 0 1 3 2 3 status

Population size 1 2 2 2 2 1 2 2 2 2 2 2 2 2 2 4 2 5 3 2 2 2 sum 17 14 22 14 20 15 17 21 22 23 16 13 18 29 15 24 9 14 16 16 12 16 Rank 8 11 4 I 1 6 10 8 5 4 3 9 12 7 1 10 2 14 11 9 9 13 9 b) Prioritization of Hagenia populations for ex situ conservation

Criteria______DK WD KB DR WB WW CM DN ZQ BB CL SM MS BG KL DO DS DD RR BR UR HM Amount of ge He 20 15 20 15 20 15 15 20 20 20 10 10 15 15 15 25 5 10 10 15 5 10 ne diversity AGD 6 3 6 3 3 3 3 6 6 3 3 3 3 3 3 12 6 3 9 3 6 3

Population size 2 4 4 4 4 2 4 4 4 4 4 4 4 4 4 8 4 10 6 4 4 4 sum 28 22 30 22 27 20 22 30 30 27 17 17 22 22 22 45 15 23 25 22 15 17 Rank 3 7 2 7 4 8 7 2 2 4 9 9 7 7 7 1 10 6 5 7 10 9 ae II Cnesin genetics Converstion III: Paper

c) Prioritization of Hagenia populations for tree improvement and domestication purposes

Criteria DK WD KB DR WB WW CM DN ZQ BB CL SM MS BG KL DO DS DD RR BR UR HM Amount of Hc 12 9 12 9 12 9 9 12 12 12 6 6 9 9 9 15 3 6 6 9 3 6 gene diversity AGD 2 1 2 1 1 1 1 2 2 1 1 1 1 1 1 4 2 1 3 1 2 1 Bole height 5 5 15 10 10 5 10 10 10 10 5 10 10 10 10 5 15 10 10 15 15 15 Population size 1 2 2 2 2 1 2 2 2 2 2 2 2 2 2 4 2 5 3 2 2 2 sum 20 17 31 22 25 16 22 26 26 25 14 19 22 22 22 28 22 22 22 27 22 24 Rank 7 9 1 6 4 10 6 4 4 4 11 8 6 6 6 2 6 6 6 3 6 5 Population codes follow Table 1; Hc= gene div ersity , AGD = average genetic distance between one population and the rest Paper III: Converstion genetics

9 a « s z t s

109 Supplementary Fig. 1. DBH-based population structure of 21 natural populations of Hagenia abyssinica from Ethiopa. Population codes as in Table 1. 10 Appendices

Appendix l. Description of Tree Seed Zones (TSZ) of Ethiopia in which H. abyssinica is growing

TSZ no. Name of Tree Seed Zones______15.3 Welo Dry Juniperus Forest (Welo in Amhara and South extreme of Tigray) 17 Southeastern High Altitude Juniperus Forest (Chilalo, Kaka & Batu mountains - Western Arsi and North western Sidamo) 18 Upper Wabe Juniperus Forest (Southwestern extreme of Arsi and Northwes tern extreme of Sidamo) 19 Western Highlands Moist Juniperus Forest (Northeastern Gonder, Including Wegera Mts.) 20.1 Gojam Undifferentiated Afromontane Forest (Gojam and southeastern Gonder) 20.2 Northeastern Drier Undifferentiated Afromontane Forest (northeastern Shewa & southwestern Wello) 20.3 Southeastern Shewa Undifferentiated Afromontane Forest (Highlands east and south of Addis Abeba and Gurage Mt.) 20.4 Western Humid Undifferentiated Afromontane Forest (western Welega and western Shewa) 21.1 Arsi Western Escarpment Undifferentiated Afromontane Forest (western Arsi) 21.2 Gelemso Central Undifferentiated Afromontane Forest (western Arsi & north central Hararghe) 23.2 Central Wet Broad-leaved Afromontane Rainforest (northwestern Illubabor, eastern Illubabor & central northeastern Kaffa) 23.3 Eastern Higher Broad-leaved Afromontane Rainforest (northeastern half of Gamo Gofa, southwestern extreme of Shewa & southeastern Keffa) 24.1 Southeastern Upper Wet Broad-leaved Afromontane Rainforest (southern slopes of Batu mountains in areas above 2000 masl, north of Hagere Mariam) 24.2 Southeastern Lower Broad-leaved Afromontane Rainforest (southern slopes of ______Batu mountain in areas below 2000 masl, south of Hagere Mariam)______Source: Aalbaek 1993 Appendices

Appendix 2. Ranges of absolute morphological and some ecological values observed among populations of Hagenia in Ethiopia.

Variables Minimum Maximum Variables Min. Max. record record record record altitude 2221 3193 width of serrated tooth (mm) 1 8 sex ratio 1:0.3 1:30 no. of pairs of leaflets having 0 4 (M:F) stipules at the back of their bases tree height 3.0 35 no. of stipules between two 0 10 (m) pairs of leaflets bole height 0.0 20 diameter at breast height (cm) 2.5* 242 (m) length of 5.5 20.5 distance between trees (m) 0.1 730 petiole (cm) no. of leaf 7 19 distance between populations 20.6 806.4 lets (km)

no. of stipu 2 41 les * Plants having diameter at breast height (DBH) values less than 2.5 cm are recorded as either saplings or seedlings

112 Appendix 3. Pairwise matrix showing geographic distances between 22 natural populations of Hagenia abyssinica from Ethiopia (see Table 1 of papers I & II for population codes)

KB 0 BG 415.70 HM 585.8 263.0 0 BR 555.8 280.2 62.6 0 BB 292.8 157.4 301.9 281.0 0 UR 567.4 273.1 41.0 20.60 286.2 0 CL 424.7 317.5 250.4 190.3 208.3 209.8 0 CM 259.9 276.6 356.0 314.3 122.2 330.5 170.5 0 DR 214.1 475.1 555.6 505.2 317.7 520.8 325.8 207.9 0 KL 480.1 277.6 160.1 103.0 222.7 124.7 87.4 224.5 705.5 0 DN 461.0 450.8 367.9 308.6 327.8 327.8 136.3 239.5 298.5 215.6 0 DO 543.0 378.7 208.3 153.1 314.0 173.8 116.2 286.5 430.1 101.2 175.2 0 DD 528.4 321.4 153.9 92.7 276.0 110.9 116.0 273.1 442.7 51.9 220.8 65.9 0 DS 516.5 301.5 137.2 78.1 257.1 98.3 116.7 263.0 439.8 39.1 230.0 84.6 22.3 0 WB 141.2 369.6 595.9 575.7 295.4 584.4 476.6 314.2 340.6 511.9 544.1 591.5 564.9 549.1 0 DK 264.5 671.2 806.4 763.5 530.5 777.8 593.5 451.5 267.7 664.7 562.5 700.2 708.6 704.7 376.5 0 MS 455.1 277.7 191.0 131.5 210.0 151.1 61.4 200.5 378.1 28.50 197.4 107.7 73.8 65.6 492.9 641.8 0 RR 567.1 376.0 185.2 126.6 326.9 143.6 145.0 311.1 466.8 101.8 210.6 37.5 57.2 77.3 611.5 730.0 121.9 0 WD 294.3 609.3 679.6 627.0 452.2 643.6 439.4 388.2 131.7 524.2 375.3 533.2 556.0 555.0 428.7 215.0 500.0 570.0 0 SM 371.1 69.3 331.6 338.9 155.4 337.3 352.8 276.2 456.6 327.2 476.3 427.0 371.8 353.4 311.4 632.0 324.8 427.8 589.1 0 WW 348.4 463.7 464.2 407.2 312.9 424.4 213.7 191.2 160.6 303.9 141.7 296.7 326.1 329.5 451.3 420.3 276.0 336.3 239.7 470.0 0 ZQ 350.4 304.9 301.5 247.8 175.2 266.4 78.3 98.8 257.1 148.4 151.4 189.4 187.8 182.4 409.7 521.1 120.0 212.0 377.6 323.3 167.3 0 KB BG HM BRBB UR CLCMDR KLDN DO DD DS WB DK MS RR WD SM WW ZQ

Appendices

J TAYE BEKELE AYELE [email protected]

Curriculum Vitae

Date o f birth: 23 July 1965 Sex: Male Nationality: Ethiopian Marital status: Married, two daughters

Education

2005 - 2008 PhD study, Department of Forest Genetics and Forest Tree Breed ing, Georg-August University Goettingen, Germany 1994-1996 Master of Science, Farm Forestry, Swedish University of Agricul tural Sciences (SLU), Uppsala & Wondo Genet. Ethiopia 1988-1990 Bachelor of Science, Forestry Management, Swedish University of Agricultural Sciences (SLU), Uppsala & Wondo Genet, Ethiopia 1983-1985 Diploma, Wondo Genet Forestry Resources Institute, Ethiopia,

Professional experience

06, 2000 - 01, 2005 Head, Department of Forest and Aquatic Plants, Ethiopian Institute of Biodiversity Conservation (IBC) and Counterpart to Forest Ge netic Resources Conservation Project • Chief Editor, Biodiversity Newsletter (2000-2004) • Forestry Team Leader, National Biodiversity Strategy and Action Plan (BSAP) 02,1999-06,2000 Forestry Expert, GTZ- Forest Genetic Resources Conservation Project, Ethiopia 07 1996 - 02, 1999 Programming Senior Expert/Assistant Farm Forestry Program Co ordinator, Ethiopian Orthodox Church Development and Inter- Church Aid Commission (EOC-DICAC), Ethiopia 06, 1990 -08, 1994 Junior Research Officer, Bako Agricultural Research Centre, Insti tute of Agricultural Research, Ethiopia 07, 1985-08, 1988 Technical Assistant, Forestry Research Centre (based in Jimma), Ministry of Agriculture, Ethiopia

Publications

Tave BA. Gailing O, Mohammed U, Finkeldey R. Colonization history and phylogeo graphy of Hagenia abyssinica (Bruce) J.F. Gmel in Ethiopia inferred from chloroplast microsatellite markers. Submitted Tave BA. Gailing O, Finkeldey R. Spatial distribution of genetic diversity in Hagenia abyssinica from Ethiopia assessed by AFLP molecular markers. Submitted

116 Taye BA. Gailing O, Finkeldey R. Conservation genetics of African redwood (Hagenia abyssinica) (Bruce) J.F. Gmel: a remarkable but gravely endangered tropical tree spe cies. Submitted Getachew Berhan and Taye Bekele (2006) Population structure and spatial distribution of four woody medicinal plant species in Bonga forest, Ethiopia. Ethiop. J. Nat. Sci. 8: 19-38 Kumlachew Yeshitela and Taye Bekele (2003) The Woody Species Composition and Structure of Masha-Anderacha Forest, Southwestern Ethiopia. Ethiop. J. Biol. Sci. 2(1): 31-48 Taye Bekele. Getachew Berhan, Matheos Ersado and Elias Taye (2003) Regeneration Status of Moist Montane Forests of Ethiopia: Part II: Godere, Sigmo, Setema and Ti- ro-Boterbecho Forests. Walia 23: 19-32 Taye Bekele (2003) The benefits of Forest Certification to Ethiopia. In Proceedings of the National Stakeholders Workshop on Forest Certification. 25 - 26 August 2003, Addis Abeba, Ethiopia Taye Bekele (2003) The Potential of Bonga Forest for Certification. In Proceedings of the National Stakeholders Workshop on Forest Certification. 25 - 26 August 2003, Addis Abeba, Ethiopia Simon Shibru, Taye Bekele and Girma Balcha (2003) Preliminary Survey of the Effect of Drought on the Forest Resources. Biodiversity Newsletter, Vol. 2, No. 1 Taye Bekele. Getachew Berhan, Elias Taye, Matheos Ersado and Kumlachew Yeshitela (2001) Regeneration Status of Moist Montane Forests of Ethiopia: Consideration for Conservation (Part I). Walia 22: 45-62 Franzel S, Ndufa, JK, Obonyo OC, Taye Bekele and Coe R (2002) Farmer-designed agroforestry trials: farmers' experiences in Western Kenya. In Franzel S and Scherr SJ (eds). Trees on the Farm: Assessing the Adoption Potential of Agroforestry Prac tices in Africa. CABI Publishing, New York Taye Bekele. Kumlachew Yeshitela, Getachew Berhan and Sisay Zerfu (2002) Forest Biodiversity Conservation: Perspectives of the Ethiopian Orthodox Church. In Ishii K, Masumori M & Suzuki K Proceedings of BIO-REFOR Tokyo Workshop. 7-11 October 2001, Tokyo Taye Bekele (2002) Indigenous Knowledge of Medicinal Plants: Perspectives of the Ethiopian Orthodox Church. In Mersha Alehegne, Taye Bekele & Netsanet Tesfaye (eds.). Proceedings of the Workshop on the Ethiopian Church: Yesterday, Today and Tomorrow. 18-19 April 2002, Addis Abeba, Ethiopia Edwards S, Abebe Demissie, Taye Bekele & Haase G (eds.) (1999) Forest genetic re sources conservation: principles, strategies and actions: proceedings of the national forest genetic resources conservation strategy development workshop, June 21-22, 1999, Addis Abeba, Ethiopia Taye Bekele. Haase G & Teshome Soromessa (1999). Forest genetic resources of Ethi opia: status and proposed actions. In Edwards, et al., Forest genetic resources conser vation: principles, strategies and actions: proceedings of the national forest genetic re sources conservation strategy development workshop, June 21-22, 1999, Addis Ab eba, Ethiopia Taye Bekele. 1993. Direct sowing Pigeon pea: A successful low cost establishment tech nique. IAR Newsletter. Vol. 8 No. 4

117 ______The m onotypic tropical tree species Hagenio abyssinica (Rosaceaej is an anemogamous and anemochorous broad-leaved dioecious tree species native to Africa. Fossil pollen evidences sug

gest that it immigrated into Ethiopia from the south during the late Pleistocene. The chloroplast

haplotypes identified in Hagenia are grouped into two lineages and demonstrated a strong pat tern of congruence between their geographical distribution and genealogical relationships. Re

stricted gene flow through seeds, contiguous range expansion, mutation and rare long-distance

dispersal shaped the genetic structure in the chloroplast genome of Hagenia.

Populations showed moderate to high gene diversities and moderate but significant genetic dif

ferentiation at AFLP markers, reflecting high levels of post-colonization gene flow. Despite the

dispersal of seed and pollen by wind, a significant fine-scale spatial genetic structure (SGS) was

observed in some populations. A weighted-score population prioritization matrix(WS-PPM) that

combines genetic, morphological and demographic criteria was developed and used for the first

time to prioritize populations for conservation and domestication. Conservation and massive

plantation programs should be launched to ensure the survival of the gravely endangered Kosso

and to boost its economic and ecological values.

ISBN 13:978-3-941274-07-5

9" 783941 11 274075 02850