Research Collection

Doctoral Thesis

Host-plant resistance in apple ( x domestica Borkh.) to common herbivore pests

Author(s): Stöckli, Sibylle Carmen

Publication Date: 2008

Permanent Link: https://doi.org/10.3929/ethz-a-005711196

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.

ETH Library DISS. ETH NO. 17849

Host-plant resistance in apple (Malus x domestica Borkh.) to common herbivore pests

A dissertation submitted to ETH ZURICH

for the degree of Doctor of Sciences

presented by SIBYLLE CARMEN STÖCKLI Dipl. Natw. ETH

born 02 October 1976 citizen of Luthern (LU), Switzerland

accepted on the recommendation of

Prof. Dr. S. Dorn, examiner Dr. M. Kellerhals, co-examiner Dr. K. Mody, co-examiner

2008

Der Apfelgarten

Komm gleich nach dem Sonnenuntergange,

sieh das Abendgrün des Rasengrunds;

ist es nicht, als hätten wir es lange

angesammelt und erspart in uns,

um es jetzt aus Fühlen und Erinnern,

neuer Hoffnung, halbvergessnem Freun, noch vermischt mit Dunkel aus dem Innern,

in Gedanken vor uns hinzustreun

unter Bäume wie von Dürer, die

das Gewicht von hundert Arbeitstagen

in den überfüllten Früchten tragen,

dienend, voll Geduld, versuchend, wie

das, was alle Masse übersteigt,

noch zu heben ist und hinzugeben,

wenn man willig, durch ein langes Leben

nur das Eine will und wächst und schweigt.

Rainer Maria Rilke Table of Contents

1. Summary……………………………………………………………………………….001

2. Zusammenfassung……………………………………………………………………..004

3. General introduction…………………………………………………………………..007 3.1. -plant relationships………………………………………………...... 007 3.2. From the fruit of paradise to the commercial apple………………………………. 007 3.3. Most common apple tree pests……………………………………………………. 009 3.4. Host-plant resistance in apples……………………………………………………. 011 3.5. Molecular approaches in host-plant resistance…………………………………….012 3.6. Thesis outline……………………………………………………………………... 014

4. QTL mapping of resistance in apple to Cydia pomonella and clerkella, and of two selected fruit traits………………………………….. 015 4.1. Introduction……………………………………………………………………….. 016 4.2. Materials and methods…………………………………………………...... 018 Study site and plant material Phenotypic data Data analysis QTL mapping 4.3. Results…………………………………………………………………………….. 021 Phenotypic data QTL analysis for C. pomonella and L. clerkella resistance in apple QTL analysis for fruit number and -diameter 4.4. Discussion………………………………………………………………………… 027 4.5 Supplementary material (SM)……………………………………………………...031

5. Influence of canopy aspect and height on codling (: ) larval infestation in apple, and relationship between infestation and fruit size………………………………... 032 5.1. Introduction……………………………………………………………………….. 033 5.2. Materials and methods……………………………………………………………. 035 Plant material and orchard locations Codling moth survey Measurement of fruit traits Temperature and wind data Statistical analysis 5.3. Results…………………………………………………………………………….. 037 Codling moth larval infestation Canopy aspect Canopy height Fruit diameter in uninfested apples Temperature and wind data 5.4. Discussion………………………………………………………………………… 045

6. QTL analysis for aphid resistance and growth traits in apple…………………….. 048 6.1. Introduction……………………………………………………………………….. 049 6.2. Materials and methods…………………………………………………...... 050 Orchard location and plant material Aphid survey Plant-growth characteristics QTL mapping Pedigree analysis Spatial distribution of aphids Statistical analysis 6.3. Results…………………………………………………………………………….. 055 Aphid infestation Influence of genotype, site and year on aphid abundance QTLs for aphid resistance QTLs for different plant-growth characteristics Spatial distribution of three aphid species Relationship between different aphid species Relationship between different plant-growth characteristics and aphid infestation 6.4. Discussion……………………………………………………………………….... 069 Genetic background of aphid resistance in apple Environmental factors related to aphid resistance Outlook 6.5. Supplementary material (SM)….…………………………………...... 074

7. Aphis pomi population development, shoot characteristics, and antibiosis resistance in different apple genotypes……………………………... 079 7.1. Introduction……………………………………………………………………….. 080 7.2. Materials and methods……………………………………………………………. 082 Study site and plant material Aphid population development in sleeve cages Shoot characteristics and general tree vigor Antibiosis-based resistance Statistical analysis 7.3. Results…………………………………………………………………………….. 084 Aphid population development Shoot characteristics and general tree vigor Climatic conditions Aphid development and shoot characteristics at different sleeve cage positions Relationship between aphid population development and shoot characteristics Relationship between aphid population development and general tree vigor Antibiosis-based aphid resistance 7.4. Discussion………………………………………………………………………… 091

8. Rust mite resistance in apple assessed by quantitative trait loci analysis………….094 8.1. Introduction……………………………………………………………………….. 095 8.2. Materials and methods…………………………………………………...... 097 Orchard characteristics and plant material Assessment of mites and other herbivores QTL analysis Data analysis 8.3. Results…………………………………………………………………………….. 100 Evaluation of rust mite abundance QTLs for rust mite resistance Spatial distribution of rust mites Relationship between rust mite abundance and co-occurring herbivore species 8.4. Discussion………………………………………………………………………… 109 8.5. Supplementary material (SM).……………………………………...... 112

9. General discussion……………………………………………………………………. 115 9.1. Host-plant resistance……………………………………………………………… 116 9.2. Genetically based resistance in apple to herbivore pests………………...... 117 9.3. Environmental factors influencing the expression of resistance………...... 119 9.4. Perspectives in apple breeding……………………………………………………. 120 9.5. Conclusion………………………………………………………………………... 121

Appendix: Herbivore resistant and susceptible apple selections: genetic background and fruit quality…………..…………………………………….... 122

A.1. Objectives…………………………………………………………………...... 123 A.2. Materials and methods…………………………………………………………..123 Plant material and study sites Herbivore survey and categorization Fruit quality Statistical analysis A.3. Results…………………………………………………………………………… 125 Herbivore resistant apple selections Genetic background of herbivore resistant and susceptible apple selections Fruit quality of herbivore resistant and susceptible apple selections A.4. Conclusion…………………………………………………………………...... 129

10. References……………………………………………………………………………. 131

11. Acknowledgements………………………………………………………………….. 143

12. Curriculum vitae…………………………………………………………………….. 145

Chapters based on:

1 Stoeckli, S., K. Mody, C. Gessler, D. Christen, and S. Dorn. 2009. QTL mapping of resistance in apple to Cydia pomonella and , and of two selected fruit traits. Annals of Applied Biology: in print. 2 Stoeckli, S., K. Mody, and S. Dorn. 2008. Influence of canopy aspect and height on codling moth (Lepidoptera: Tortricidae) larval infestation in apple, and relationship between infestation and fruit size. Journal of Economic Entomology 101:81-89. 3 Stoeckli, S., K. Mody, C. Gessler, A. Patocchi, M. Jermini, and S. Dorn. 2008. QTL analysis for aphid resistance and growth traits in apple. Tree Genetics & Genomes 4:833-847. 4 Stoeckli, S., K. Mody, and S. Dorn. 2008. Aphis pomi population development, shoot characteristics, and antibiosis resistance in different apple genotypes. Journal of Economic Entomology 101:1341-1348. 5 Stoeckli, S., K. Mody, A. Patocchi, M. Kellerhals, and S. Dorn. 2008. Rust mite resistance in apple assessed by quantitative trait loci analysis. Tree Genetics & Genomes doi: 10.1007/s11295-008-0186-5. 1. Summary

In agriculture, herbivores cause high economic losses by plant feeding and transmitting plant pathogens. Host-plant resistance is a persistent and environment-friendly control method, compatible with other integrated pest management methods. The main objective of the present thesis was to evaluate the genetic basis of resistance in apple (Malus x domestica Borkh.) to common herbivore pests through quantitative trait loci (QTLs) analyses. A linkage map of the apple cultivars 'Fiesta' and 'Discovery', saturated with 235 random amplified polymorphic DNA (RAPD), 475 amplified fragment length polymorphism (AFLP) and 129 simple sequence repeats (SSR or microsatellite) markers, was the basis for identifying molecular markers associated with herbivore resistance. A total of 160 highly heterozygous F1 progeny plants (genotypes) were investigated for their resistance to two lepidopteran species, the codling moth (Cydia pomonella) and the apple leaf miner (Lyonetia clerkella); three aphid species, the rosy apple aphid (Dysaphis plantaginea), the leaf-curling aphids (Dysaphis cf. devecta species complex), and the green apple aphid (Aphis pomi); and the rust mite (Aculus schlechtendali). Methods for field evaluations comprised quantification of herbivores per leaf or tree, or a survey of leaves and fruits with damage indicating infestation by a distinct herbivore species. The field survey was carried out at three orchards in different regions of Switzerland and during two consecutive years. The proportion of variation in herbivore infestation that can be explained by the genetic variation among the apple progenies was analyzed by broad sense heritability (H2). H2 was intermediate and ranged between 15.7% (A. pomi) and 50.2% (D. cf. devecta). The presence of a genetically based resistance was indicated by genotype being a significant factor explaining infestation of C. pomonella, D. plantaginea, D. cf. devecta, and A. schlechtendali (mixed model ANOVA), and confirmed by aggregated occurrence of these herbivores on individual apple trees (index of dispersion, ID values). The comparison of the phenotypic field data and the genotypic markers resulted in a set of QTLs. Except for L. clerkella and A. pomi, one or more QTLs could be identified for resistance or susceptibility to each herbivore species investigated. The RAPD marker Z19-350 located at 66.2 cM on the 'Discovery' linkage group (LG) 10 was associated with higher C. pomonella infestation. The SSR CH04g09y (allele '177 bp'), nearest to this QTL and in coupling with the RAPD marker should be considered to study resistance to C. pomonella. The QTL explained only 8.2% of the phenotypic variability.

1 A QTL associated to D. plantaginea resistance was identified on the 'Fiesta' LG 17. The AFLP marker E33M35-0269 at 57.7 cM was in close proximity to this QTL, which explained 8.5% of the phenotypic variability. The SSR marker Hi07h02 (allele '255 bp'), which is coupled with the AFLP marker, was traced back in the pedigree of 'Fiesta' to the apple cultivar 'Wagener'. A gene region associated to D. cf. devecta resistance on the 'Fiesta' LG 7 was confirmed. The AFLP marker E32M39-0195 (4.5 cM) explained between 10.3% and 20.4% of the phenotypic variability in infestation. The SSR Hi03a10 (allele '216 bp'), which is coupled with the AFLP marker, was derived from the cultivar 'Blenheim Orange' in the pedigree of 'Fiesta'. A QTL for A. schlechtendali resistance was identified on 'Fiesta' LG 7. The AFLP marker E35M42-0146 (20.2 cM) and the RAPD marker AE10-400 (45.8 cM) explained between 11.0% and 16.6% of the phenotypic variability. An interaction between the two markers was not found. The SSR Hi03a10 (allele '240 bp') linked to the QTL was traced back in the 'Fiesta' pedigree to the cultivar 'Wagener'. Host-plant resistance to A. pomi was studied in more detail by evaluating antibiosis-based resistance. Population development in sleeve cages was quantified and the results indicated a putative QTL on LG 11 of 'Fiesta'. The SSR CH02d12 (allele '199 bp') at 21.5 cM explained 14.1% of the phenotypic variability. As fruit availability and fruit characteristics are of crucial relevance for fruit-feeding , the genetic basis of fruit number and -diameter was evaluated. A putative QTL associated with fruit number was identified on 'Fiesta' LG 12, with the SSR CH01g12 (43.6 cM) in close proximity. The QTL explained 9.6% of the phenotypic variability and apple genotypes amplifying the allele '156 bp' produced significantly less apples compared to other apple genotypes. A significant QTL for fruit diameter was not detected. Besides fruit traits, also plant-growth traits may influence the expression of herbivore resistance. Knowledge about the genetic basis of plant-growth traits may elucidate the relationship to herbivore abundance. Four markers were identified by QTL analysis, and were significantly linked to stem diameter or shoot length. Increased shoot length was linked to the SSR CH04e05 (allele '199 bp') at 26.7 cM on the 'Fiesta' LG 7. Increased stem diameter was associated with markers on the 'Discovery' LG 1, 13, and 14. Differences in climatic conditions, fruit and plant-growth traits, co-occurring herbivore fauna, spatial pattern, and infestation level of neighbor trees (neighborhood effect) effect may have hindered the expression of QTLs for resistance to L. clerkella and A. pomi, and may have influenced the stability of the QTL associated to C. pomonella infestation. This is especially supposed for A. pomi and C. pomonella. There was a significant positive relationship between A. pomi population development and shoot characteristics (measured as

2 shoot length and shoot growth) in spring and early summer, but not in late summer. This result may be related to a decrease in shoot growth and nitrogen levels, which are highest in spring and decrease over the growing season. Microclimate conditions may cause within-tree variability of fruit diameter and of C. pomonella female oviposition preference and blur differences between genotypes. A significantly lower fruit infestation was observed on the cooler north- compared with the warmer south- or east-facing tree side. These findings are important for future monitoring systems focusing on female . The complex host-finding process of arthropod herbivores possibly impeded the evolution of major resistance genes. Such major resistance genes have been identified for many pathogens, where a host-finding process is inexistent. The presented high-resistance-selections have potential to be introduced into apple breeding combining pest resistance with promising fruit quality. Five selections (13, 22, 189, 265, and 303) have a high potential for breeding apple cultivars with multiple resistance to C. pomonella, D. plantaginea, D. cf. devecta, A. pomi, and A. schlechtendali. To conclude, the present thesis provides molecular markers associated with resistance in apple against herbivores, which can be used in marker-assisted selection or transferred into new varieties by genetic engineering. The intermediate heritability indicates a partial resistance, which is of high value in host-plant resistance. The high selection pressure on a pest of highly resistant plants may lead to adapted biotypes. This risk is minimized in quantitative traits. Molecular markers are a prerequisite to understand the segregation of resistance genes among varieties. Furthermore, the resistance in phenotypically evaluated apple cultivars could be confirmed by testing them for the identified molecular markers.

3 2. Zusammenfassung

Arthropoden verursachen in der Landwirtschaft einen hohen Ernteverlust durch Herbivorie an Pflanzen und durch die Übertragung von Krankheitserregern. Die Pflanzung von resistenten Sorten ist eine dauerhafte und umweltfreundliche Methode, um den Befall durch Schädlinge zu mindern. Pflanzenresistenz ist kompatibel mit anderen Massnahmen der integrierten Produktion ist. Das Hauptziel der vorliegenden Studie war die genetische Grundlage der Resistenz beim Apfel (Malus x domestica Borkh.) gegenüber häufigen Schädlingen zu untersuchen. Die Assoziation zwischen molekularen Markern und Schädlingsbefall wurde mit Hilfe von quantitativen Analysen (QTLs) getestet. Eine genetische Karte der Apfelsorten 'Fiesta' und 'Discovery' gesättigt mit 235 'random amplified polymorphic DNA' (RAPD), 475 'amplified fragment length polymorphism' (AFLP) und 129 'simple sequence repeats' (SSR oder 'microsatellite') Markern war bereits vorhanden. Total

160 hoch heterozygote F1 Nachkommen (Genotypen) wurden auf Befall durch zwei lepidopteren Arten, der Apfelwickler (Cydia pomonella) und die Schlangenminiermotte (Lyonetia clerkella); drei Blattlausarten, die mehlige Apfelblattlaus (Dysaphis plantaginea), der Komplex der Apfelfaltenläuse (Dysaphis cf. devecta) und die grüne Apfelblattlaus (Aphis pomi); sowie eine Milbe, die Rostmilbe (Aculus schlechtendali) untersucht. Für die Bestimmung der Resistenz wurden die Anzahl Individuen per Blatt oder Baum, oder die Aufnahme von Blättern oder Früchte mit einem Arthropod-spezifischen Schaden evaluiert. Die Feldstudie wurde an drei Orten in der Schweiz und während zwei aufeinander folgenden Jahren durchgeführt. Die Variation im Herbivor Befall die durch die genetische Variabilität erklärt werden kann (broad sense heritability H2) war mittelhoch. Sie variierte von 15.7% (A. pomi) bis 50.2% (D. cf. devecta). Der Genotyp war ein signifikanter Faktor für die Erklärung des Befalls von C. pomonella, D. plantaginea, D. cf. devecta, und A. schlechtendali (ANOVA Model), was einen ersten Hinweis auf eine genetisch basierte Resistenz beim Apfel gegenüber Schädlingen gab. Dies wurde durch die Aggregation von Schädlingen an individuellen

Apfelbäumen (index of dispersion, ID values) bestätigt. Der Vergleich des Befalls durch Schädlinge im Feld (Phänotyp) und den molekularen Markern (Genotyp) führte zu mehreren QTLs für Resistenz/Anfälligkeit. Mit Ausnahme von L. clerkella und A. pomi wurden QTLs für Schädlinge identifiziert. Der RAPD Marker Z19-350 auf der 'Discovery' linkage group

4 (LG) 10 (66.2 cM) war mit stärkerem C. pomonella Befall assoziiert. Der SSR CH04g09y (Allel '177 bp') das mit dem QTL verbunden ist, sollte bei weiteren Untersuchungen zur Resistenz bei C. pomonella beachtet werden. Der QTL erklärte 8.2% der phänotypischen Variabilität. Ein QTL assoziiert mit D. plantaginea Resistenz lag auf der 'Fiesta' LG 17. Der AFLP Marker E33M35-0269 bei 57.7 cM war am nächsten positioniert, und erklärte 8.5% der phänotypischen Variabilität. Der SSR Marker Hi07h02 (Allel '255 bp'), der mit dem AFLP Marker verbunden ist, wurde im Stammbaum von 'Fiesta' bis auf die Sorte 'Wagener' zurückverfolgt. Eine Genregion für D. cf. devecta resistance konnte auf der 'Fiesta' LG 7 bestätigt werden. Der AFLP Marker E32M39-0195 (4.5 cM) erklärte zwischen 10.3% und 20.4% der phänotypischen Variabilität. Der SSR Marker Hi03a10 (allele '216 bp') der mit dem AFLP Marker gekoppelt ist, wurde von der Sorte 'Blenheim Orange' im 'Fiesta' Stammbaum weitervererbt. Ein QTL für A. schlechtendali Resistenz wurde auf der 'Fiesta' LG 7 identifiziert. Der AFLP Marker E35M42-0146 (20.2 cM) und der RAPD Marker AE10-400 (45.8 cM) erklärten zwischen 11.0% ud 16.6% der phänotypischen Variabilität. Es wurde keine Interaktion zwischen den beiden Markern gefunden. Der SSR Marker Hi03a10 (Allel '240 bp') wurde im 'Fiesta' Stammbaum auf 'Wagener' zurückgeführt. Pflanzenresistenz durch Antibiosis wurde bei A. pomi im Detail evaluiert. Die Populationsentwicklung in 'sleeve cages' wurde quantifiziert und es gab einen Hinweis auf einen vermeintlichen QTL auf der LG 11 von 'Fiesta'. Der SSR Marker CH02d12 (Allel '199 bp' bei 21.5 cM erklärte 14.1% der phänotypischen Variabilität. Frucht-Vorkommen und Frucht-Merkmale sind wichtig für direkte Fruchtschädlinge, und es wurde deshalb mögliche genetische Basis von Frucht-Zahl und Frucht-Durchmesser untersucht. Ein QTL wurde mit Frucht-Zahl auf der 'Fiesta' LG 12 assoziiert, mit dem SSR Marker CH01g12 (43.6 cM) am nächsten positioniert. Der QTL erklärte 9.6% der phänotypischen Variabilität, und Apfelgenotypen mit dem Allel '156 bp' produzierten weniger Äpfel als andere Genotypen. Ein signifikanter QTL für Frucht-Durchmesser konnte nicht identifiziert werden. Neben Frucht-Merkmalen können auch Pflanzen-Wuchs Faktoren die Ausbildung einer Resistenz verhindern. Das Wissen über die genetische Basis von Pflanzen- Wuchs Faktoren Hilft zum Verständnis der Schädlings-Resistenz. Vier molekulare Marker wurden durch QTL Analyse identifiziert. Längere Äste war mit dem SSR Marker CH04e05 (Allel '199 bp', 26.7 cM) auf dem 'Fiesta' LG 7 assoziiert. Breiterer Stamm-Durchmesser war assoziiert mit Markern auf dem 'Discovery' LG 1, 13 und 14. Unterschiedlichen Klimaverhältnisse, Fruchtmerkmale und Pflanzenwuchs, herbivore Fauna, räumliche Verteilung, oder Befall von Nachbarbäumen (Nachbarschafts-Effekt),

5 könnten die Ausbildung einer genetisch basierten Resistenz gegenüber L. clerkella and A. pomi beeinflussen, auch die Stabilität des QTLs für C. pomonella Befall beeinflussst haben. Es wurde ein signifikanter Zusammenhang zwischen A. pomi Populations-Entwicklung und Astlänge im Frühling und Frühsommer gefunden (Astlänge, Astwachstum), aber nicht mehr im Spätsommer. Dieses Resultat kann durch die Abnahme des Astwachstums und des Stickstoffs erklärt werden, das im Frühling am höchsten ist und bis im Spätsommer abnimmt. Das Mikroklima könnte eine Variabilität von Fruchtdurchmesser und C. pomonella Eiablage verursachen. Es wurde ein signifikant niedrigerer Befall auf der kälteren Nordseite im Vergleich zur wärmeren Süd- oder Ostseite des Baumes gefunden. Dieses Resultat ist auch wichtig für eine repräsentative Bestandesaufnahme, die speziell weibliche Falter beachtet. Der komplexe Wirtsfindungsprozess hat möglicherweise die Evolution monogenetischer Resistenz verhindert. Eine solche Resistenz wurde für viele Pathogene gefunden, wo ein Wirtsfindungsprozess nicht vorhanden ist. Die präsentierten hoch-resistenten Apfelgenotypen haben ein gutes Potential um in die Apfelzucht mit hoher Fruchtqualität integriert zu werden. Mehrere Genotypen (13, 22, 189, 265, and 303) sind geeignet um Apfelsorten zu züchten mit einer multiplen Resistenz gegen C. pomonella, D. plantaginea, D. cf. devecta, A. pomi, and A. schlechtendali. Abschliessend kann bemerkt werden, dass die vorliegende Arbeit molekulare Marker liefert die mit Resistenz gegen Herbivore assoziiert sind und die geeignet sind für Marker unterstützte Selektion, oder eine Grundlage für die Herstellung genetisch transformierter Apfelsorten bilden. Die intermediäre Vererbung ist ein Hinweis auf eine partielle Resistenz, die von hohem Wert in der Apfelzucht ist. Der hohe Selektionsdruck von hoch resistenten Pflanzen könnte schnell zu angepassten Biotypen führen. Dieses Risiko ist bei quantitativen Merkmalen minimiert. Molekulare Marker dienen dazu um die Segregation von Resistenz Genen unter Apfelsorten zu verfolgen. Weiter können phänotypisch evaluierte Apfelsorten auf die identifizierten Marker getestet und die Resistenz durch einen molekularen Ansatz bestätigt werden.

6 3. General introduction

3.1. Insect-plant relationships

Studies that contribute towards elucidating insect-plant relationships are of crucial relevance for various reasons. The taxa of plants and insects are the most diverse groups, representing 50% of all known multicellular species (Strong et al. 1984). Plants and plant- feeding herbivores are considered to largely account for the present natural diversity of plants and , and they are therefore central to biodiversity conservation (Schoonhoven et al. 2005). Another aspect of general concern is the reduction of pest-associated yield losses in agriculture, estimated at 14% of the total agricultural production (Oerke et al. 1994). Besides direct loss due to herbivores, insects are vectors of plant pathogens. In the context of the predicted increase of the human population to 10 billion by the year 2050, insects may have increased significance (Fedoroff and Cohen 1999).

3.2. From the fruit of paradise to the commercial apple

There is no other fruit in temperate climates that is so universally appreciated and extensively cultivated as the apple. Many ancient myths and stories describe the apple as a symbol of life, immortality, love and fertility (Laudert 1998). The story about the tree of paradise whose fruits were said to grant eternal life forms appears in many legends. The Greek goddess Hera, who was married to Zeus, received an apple tree with golden apples as a wedding present. Herakles, the son of Zeus, had to gather the golden apples of the Hesperides as one of his twelve tasks as an adult man. In Christianity, the originally positive symbol of the fruit was changed to mean the opposite, signifying seduction and sin (Morgan and Richards 1993). In the Middle Ages, the apple was used as a sign of terrestrial power for emperors (Laudert 1998). Last but not least, what would William Tell (drama by Friedrich Schiller, 1804) be without the apple, and what would Switzerland be without William Tell? Nowadays, the symbol of apple has changed from this rather mythological meaning to representing commercial products (Laudert 1998). In advertisements for cosmetics the apple stands for health. The city of New York is well known as 'Big Apple'. The computer company

7 Apple Macintosh uses the apple as a symbol for global networks. Apples as fruits are admired by all humans because of the many ways that they can be consumed (e.g. fruit, juice, vinegar, apple crumble and cake) and because of their convenience and durability (Morgan and Richards 1993). The common domesticated apple is putatively an interspecific hybrid complex, usually designated Malus x domestica Borkh. (Luby 2003). Apples are members of the genus Malus Miller, which is placed in the subfamily Maloideae of the family . Pears, quinces and hawthorns are further members of the Maloideae. The origin and ancestry of the M. x domestica complex remain unknown. However, Malus sieversii (Ledeb.) Roem. is hypothesized as the key species at its origin (Juniper et al. 1999). M. sieversii is widespread in the mountains of central Asia. Malus species are mostly diploid with 17 chromosomes (Brown 1992). World apple production has been increasing since the Second World War (O'Rourke 2003), mainly due to the expansion of production in China and the successful spread into warmer climates (Luby 2003). The latter was enabled due to breeding of heat-tolerant cultivars and the popularity of varieties that require a long growing season (e.g. 'Granny Smith'). The estimated world production of apples (for 2006) was 64 million t (http://faostat.fao.org), ranking in fourth position behind bananas (71 million t), grapes (69 million t) and oranges (65 million t). With 26 million t, China produces 40% of the world production. China is followed by the USA (4.6 million t). China’s apple production rapidly increased with the introduction of the cultivar 'Fuji'. It is believed that apple production in China could exceed 35 million t in 2010 as many of the trees are not yet at full bearing maturity (O'Rourke 2003). Managing the expected intensified competition on apple markets will be a challenging task. The cultivation of apples started with the civilization of humans, and there is evidence that in the Neolithic age fruits from wild apple trees were collected and dried (Juniper et al. 1999). The domestication of the apple seems to have been practiced by the Greeks and Romans and, as a result of their travels and invasions, domesticated apples were spread by them throughout and Asia (Luby 2003). Traditionally, the best seedling grown from open-pollinated seeds was selected (Janick et al. 1996). Many of today's popular apple cultivars were selected from chance seedlings that were developed in the 18th and 19th century ('Golden Delicous', 'Cox's Orange Pippin', 'Granny Smith' and 'McIntosh' (Gardiner et al. 2007). Thomas Andrew Knight (1759-1838) produced the first cultivars of known parentage (Janick et al. 1996) and all present-day apple breeding programs are based on this technique.

8 Besides resistance, the marketability of the fruit is the principal selection criteria in apple breeding (Janick et al. 1996). Consumer preference has been related to, for example, flavor, juiciness, sweetness, size, shape, color, or firmness (Eigenmann and Kellerhals 2007). Additionally, farmers and producers are interested in acceptable yield and promising storage qualities (Janick et al. 1996). Many breeding programs exist around the world, but only certain new cultivars are commercially used (Hampson and Kemp 2003). The major constraints in apple breeding are the long juvenile period and the self-incompatibility (Maliepaard et al. 1998, Brown and Maloney 2003, Gessler and Patocchi 2007). This means that only after five years can plants be selected based on the desired traits and that a 'pseudo- inbred' line has to be produced. Therefore the development of new apple cultivars takes 12-15 years. The parents used are usually heterozygous and the F1 progeny plants segregate for a large number of traits with genetically unique seedlings (Gessler and Patocchi 2007).

3.3. Most common apple orchard pests

An arthropod is determined as a pest when it interferes with humans for the same resources (Pedigo 2006). Particularly this is most obvious in agricultural production systems where cause serious economic losses (Pimentel 1997). The term pest is sometimes not only constricted to arthropods, but also to plant-parasitic nematodes, microbial and viral plant pathogens, weeds and vertebrates (Prokopy and Croft 1994). The various structures of an apple tree provide food and/or shelter for a large number of arthropod pests (Schoonhoven et al. 2005). Direct pests of fruits have the most apparent impact on yield because only slight infestation makes the product unmarketable (Beers et al. 2003). Many lepidopterans, especially tortricids, attack fruits. The codling moth (Cydia pomonella L.) is considered as the key species in apple orchards worldwide and infestation levels have even increased within the last years (Blommers 1994, Prokopy and Croft 1994, Dorn et al. 1999). Besides apples, it attacks the fruits of pears, quinces, apricots, peaches, and walnuts (Alford 2007). The damage is caused by the larvae, which burrow into the fruit to feed on the flesh and seeds. A small red-ringed cavity hole filled with dry frass is an indication for larval penetration (Baggiolini et al. 1992). After few weeks, and passing through five instars, the mature larvae leave the fruit to spin a cocoon for pupation under loose bark on the tree (Geier 1963). In northern Europe, usually the last larval instar overwinters and pupation occurs in spring. Five to six generations have been recorded in

9 warmer regions (Audemard 1991). The first moths appear in spring (Beers et al. 1993). After mating, eggs are laid singly on leaves and fruits during warm evenings (above 15°C). The new larvae hatch after 10-14 days. Other lepidopteran fruit pests are the winter moth (Operophtera brumata L.), and the oriental fruit moth ( molesta Busck) (Prokopy and Croft 1994). Furthermore, chewing herbivores include coleopterans such as the apple blossom weevil (Anthonomus pomorum L.), and hymenopterans such as the apple sawfly (Hoplocampa testudinea Klug), of which both infest apple buds and fruits (Blommers 1994). Several phytophagous gracillariid (e.g. the spotted tentiform leafminer, Phyllonorycter blancardella Fabricus) and lyonetid moths (e.g. the apple leaf miner, Lyonetia clerkella L.) mine apple leaves in their larval stage (Beers et al. 2003). L. clerkella frequently occurs in apple orchards of the Palearctic zone (Beers et al. 1993). Fruit growth is affected by reduced photosynthesis when feeding larvae damage the epidermis (Reissig et al. 1982). Outbreaks are not uncommon, as larval parasitoids are highly susceptible to pesticides used in commercial orchards (Maier 1992). Besides tortricid moths, aphids are key pests in apple orchards worldwide (Blommers 1994, Prokopy and Croft 1994). The phloem-sap-feeding behavior causes deformation of leaves, shoots and also of fruits. In addition to these direct injuries, aphids may act as vectors for virus transmission, while the release of honeydew fosters sooty mold infestation (Arbab et al. 2006). Aphids have a special lifecycle with a holocyclic (sexual) and anholocyclic (viviparity and parthenogenesis) part, often including a host-plant change (Beers et al. 1993). This condition makes aphids very adaptable to new environments and efficient in rapid colony build-up. The rosy apple aphid (Dysaphis plantaginea Pass.), the species complex of the leaf-curling aphids (Dysaphis cf. devecta Wlk.), the green apple aphid (Aphis pomi De Geer) and the woolly apple aphid (Eriosoma lanigerum Ham.) are widespread within apple orchards (Blommers 1994, Beers et al. 2003). Almost no infestation by D. plantaginea is tolerated as small and deformed fruits are produced on infested shoots (Blommers 1994). It colonizes apples from late spring until early summer (Beers et al. 1993). The feeding behavior of D. cf. devecta produces red-curled leaves (Beers et al. 1993). It is found on apple trees until late spring (Beers et al. 1993). Comparable to D. plantaginea, D. cf. devecta migrates to an herbaceous weed as a summer host (e.g. Plantago lanceolata in the case of D. plantaginea) (Beers et al. 1993, Blommers et al. 2004). A. pomi uses apple and pear trees as a host-plant the whole year, but population outbursts may occur (Baker and Turner 1916). This aphid causes low leaf deformation, but produces a large amount of honeydew (Beers et al. 2003). E.

10 lanigerum may reduce tree vigor by feeding on roots and shoots, producing galls and waxy coverage of shoots (Prokopy and Croft 1994). The tetranychid mites Panonychus ulmi (Koch) (European red mite) and Tetranychus urticae (Koch) (two-spotted spider mite) and the eriophyid apple rust mite (Aculus schlechtendali Nalepa) are the most destructive mites in apple orchards (Prokopy and Croft 1994). Mite development is rapid under warm climates. Damage is caused by the mites sucking sap and damaging leaf cells (Beers et al. 1993). The most obvious symptoms are speckling and brownish or bronze leaves (Beers et al. 1993). The reduced CO2 assimilation leads to leaf desiccation and early defoliation (Spieser et al. 1998). The mentioned injuries diminish fruit and tree quality in subsequent years and initiate russet formation (Easterbrook and Fuller 1986). Some mites overwinter as eggs (P. ulmi) and some as adults (T. urticae, A. schlechtendali) (Beers et al. 1993). In late summer, several hundred rust mite individuals can be found on the underside of an apple leaf (Spieser et al. 1998). Rust mite infestation has increased along with changes from broad-spectrum insecticides that reduced rust mite populations to selective fungicides and acaricides with no effect on rust mites (Spieser et al. 1998). Besides arthropods, apples are threatened by diseases (e.g. apple scab, powdery mildew and fire blight), vertebrate pests (e.g. voles, birds and porcupines) and weed pests (Prokopy and Croft 1994).

3.4. Host-plant resistance in apples

Integrated pest management (IPM) is the most popular approach to suppress arthropod populations (Kogan 1998). IPM is a pest regulation strategy that uses multiple tactics such as chemical, behavioral, microbial, biological and cultural control, and host-plant resistance (HPR) to prevent a pest from exceeding the economic injury level (which means that the yield loss caused by pest damage exceeds the costs for control) (Beers et al. 1993). The main goal of IPM is to significantly reduce pesticide application and to protect the environment (Avilla and Riedl 2003). Plant resistance to pests is defined as any inherited characteristic of a host-plant that results in a lower damage compared to non-resistant plants (Smith 2005b). One of the earliest important examples of an application of plant resistance is the introduction of American grape varieties resistant to the grape phylloxera (Daktulosphaira vitifoliae) to Europe in 1880 (Pedigo 2006). Without HPR, the French wine industry would have nearly collapsed as this

11 pest reached a catastrophic spread. Plant feeding arthropods (herbivores) are mostly affected by HPR and natural enemies (Schoonhoven et al. 2005). HPR is considered a very useful control method because (once resistant host-plants have been developed) it is (1) specific to a pest with no strong detrimental effect on natural enemies, (2) applicable for a long time, (3) environmental friendly, (4) compatible with other IPM tactics, (5) easy to apply and (6) not expensive (Maxwell 1985). A 120-fold economic return is reported through using wheat plants resistant to the Hessian fly (Mayetiola destructor Say) (Smith and Quisenberry 1994). Serious environmental pollution, the elimination of natural enemies, increased costs of pesticides and the resistance of pests to pesticides caused by heavy reliance on insecticides have all promoted the integration of more sustainable tactics such as HPR (Prokopy and Croft 1994, Pimentel 1997). HPR is based on certain morphological, and/or chemical plant defensive factors (Pedigo 2006). Different varieties of the same plant species express different defensive factors, resulting in variable levels of resistance. There are three different resistance categories to describe the way plants protect themselves from arthropod damage. Antixenosis includes plant properties that reduce colonization by pests seeking food or oviposition sites. Antibiosis comprises biological processes of insects, like survival, growth, generation time, fecundity, and longevity. As in antixenosis, antibiosis involves both insect and plant factors. Tolerance is defined as plants that show a reduced response to damage compared with non- tolerant plants. Tolerance is not always considered as resistance (Leimu and Koricheva 2006), because only plant responses are involved. High-value crops such as apple (Jackson 2003, O'Rourke 2003), have low economic injury levels for pest and disease damage to be marketable. Therefore, pesticide applications to diminish yield loss by pests are still the primary pest control option, emphasizing the value of HPR (Beers et al. 2003).

3.5. Molecular approaches in host-plant resistance

Apple cultivars have been selected for high yields, promising fruit quality and pest resistance since the beginning of agriculture 10'000 years ago (Janick et al. 1996). Resistant host-plants have gained importance as the genetic characterization of the host-plant genome has been accelerated by time-efficient and cost-efficient molecular techniques (Tanksley et al. 1989).

12 Many of the important agronomic traits (e.g. pest resistance) are quantitative traits, which are controlled by multiple genes. Quantitative traits exhibit continuous variation in phenotypic expression and are much more difficult to work with in breeding programs than qualitative traits (controlled by a single gene) (Yencho et al. 2000). The gene regions controlling these traits are called quantitative trait loci (QTLs). Molecular markers represent such gene regions (Mohan et al. 1997a), and reveal genetic differences between plant individuals through variation in DNA nucleotide sequences. This DNA polymorphism is based on insertions, deletions, duplication and substitutions of nucleotides within sexual reproduction (meiosis) (Liu 1998). Random amplified polymorphic DNA (RAPD), amplified fragment length polymorphism (AFLP), restriction fragment length polymorphism (RFLP) and simple sequence repeats or microsatellites (SSR) are used to make the DNA polymorphism visible (Collard et al. 2005). A comprehensive molecular map with developed molecular markers is now available for many crops, such as apple, maize, tomato, potato, rice and wheat. The association between a molecular marker and the phenotypic trait is analyzed by QTL analyses. The detection of QTLs, their localization and the estimation of their effect is the aim of these analyses. In single-marker analysis, the individual plants are divided based on presence and absence of a specific molecular marker. A Kruskal-Wallis test is used to analyze if the phenotypic trait differs between the two subpopulations. Interval mapping (IM) is applied for multiple marker analysis. IM uses a maximum likelihood approach to compare the possibility of a present or an absent QTL and the LOD (logarithm of odds) ratio is presented for statistical significance. Multiple QTL models additionally include interactions between QTLs. QTLs for disease resistance (Durel et al. 2004, Calenge and Durel 2006, Khan et al. 2007) and tree and fruit traits (Conner et al. 1998, King et al. 2001) in apples have been identified. Additionally, molecular markers associated with insect resistance in apple (Roche et al. 1997, Bus et al. 2008), wheat (Myburg et al. 1998), rice (Jairin et al. 2007), bean (Frei et al. 2005), maize (Brooks et al. 2007) and tomato (Rossi et al. 1998) have been described. Nevertheless, information about the genetic basis of resistance in apples to insects is still scarce and underlines the importance of a comprehensive study. The availability of molecular markers that are tightly linked to a phenotypic trait allows (1) the determination of seedlings with desirable traits in an early stage by marker assisted selection, (2) the segregation of important genes among varieties to be monitored, (3) the transfer of novel genes from related wild species and (4) the establishment of genetic relationships between sexually incompatible crop plants.

13 3.6. Thesis outline

The main objective of the present thesis was to investigate the genetic basis of resistance in apples (Malus x domestica Borkh.) to several pest herbivores by quantitative trait loci (QTLs) analysis. A field survey of 160 F1 'Fiesta' x 'Discovery' progeny plants was carried out at three different sites in Switzerland and during two consecutive years. Herbivore infestation was taken as a measure for resistance. The experiment focused on six common apple pests: the codling moth (Cydia pomonella L.), the apple leaf miner (Lyonetia clerkella L.), the rosy apple aphid (Dysaphis plantaginea Pass.), the leaf-curling aphids (Dysaphis cf. devecta Wlk.), the green apple aphid (Aphis pomi De Geer) and the rust mite (Aculus schlechtendali Nalepa). Molecular markers associated with herbivore resistance are identified within the present thesis, and are available for marker assisted selection of new apple cultivars. The QTL analysis was based on the available linkage maps for progenies of the apple cultivars 'Fiesta' x 'Discovery' (Liebhard et al. 2002, Liebhard et al. 2003b, Silfverberg- Dilworth et al. 2006), filled with both dominant (RAPD, RFLP and AFLP) and co-dominant (SSR) markers. Chapter 4 refers to the QTL analysis for resistance to the moth species C. pomonella and L. clerkella, and of two selected fruit traits, the number fruits and fruit diameter. 40,000 fruits have been used for damage assessment on the 160 apple genotypes surveyed over two successive years. This comprehensive data set has been used in Chapter 5 to statistically determine the influence of canopy aspect and height on C. pomonella infestation, and to relate infestation to fruit size. Chapter 6 describes the results from the QTL analysis for resistance to the aphid species D. plantaginea, D. cf. devecta and A. pomi. Furthermore, QTLs for selected plant-growth traits are described and a possible relationship with aphid resistance is evaluated in this chapter. In Chapter 7, resistance to A. pomi based on antibiosis is studied by evaluating population development in sleeve cages in the field. Population development in relation to shoot growth and a possible genetic basis and influence of general tree vigor is analyzed. Chapter 8 presents molecular markers associated with rust mite (A. schlechtendali) resistance identified by QTL analysis. In the Appendix, the 20 most resistant and susceptible progeny plants are selected based on herbivore abundance. The apple selections are evaluated considering available QTLs and fruit quality.

14 4. QTL mapping of resistance in apple to Cydia pomonella and Lyonetia clerkella, and of two selected fruit traits 1

Apple (Malus x domestica) is the most important fruit grown within the temperate zonobiome. It is attacked by both fruit- and leaf damaging lepidopteran pest insects, which require regular control such as the carpophagous codling moth (Cydia pomonella) or frequent control such as the phyllophagous apple leaf miner (Lyonetia clerkella). As many environmentally friendly pest control tactics are only effective at low levels of infestation, host plant resistance is a promising future component of integrated pest management systems, but knowledge is still lacking on such genetically based approaches against lepidopteran pests. The aim of the study was to identify molecular markers linked to C. pomonella and L. clerkella resistance or susceptibility in commercial apple, as well as markers linked to selected fruit traits. The number of C. pomonella infested fruits, and the number of L. clerkella mines were quantified as measures of apple resistance or susceptibility to the studied moth species. Herbivore surveys on 160 apple genotypes, representing a segregating F1-cross of the apple cultivars 'Fiesta' and 'Discovery', were carried out during two consecutive years and at two sites in Switzerland. Broad-sense heritability was 29.9% (C. pomonella), 18.2% (L. clerkella), 21.9% (fruit number), and 16.6% (fruit diameter). A subsequent analysis identified a QTL associated to C. pomonella susceptibility on the 'Discovery' linkage group 10. The closest marker to this QTL was the RAPD marker Z19-350. No significant QTL was identified for resistance to L. clerkella. A putative QTL associated to fruit number was identified on 'Fiesta' linkage group 12. The presented QTL associated with C. pomonella susceptibility and the putative QTL linked to fruit number may facilitate marker assisted breeding of resistant apple cultivars with cropping traits desirable for optimal fruit production.

______1 Stoeckli, S., K. Mody, C. Gessler, D. Christen, and S. Dorn. 2009. QTL mapping of resistance in apple to Cydia pomonella and Lyonetia clerkella, and of two selected fruit traits. Annals of Applied Biology: in print.

15 4.1. Introduction

Moths (Lepidoptera) and aphids (Sternorrhyncha) represent key pests in apple orchards worldwide (Blommers 1994). The most serious yield loss is caused by the carpophagous codling moth (Cydia pomonella L.) (Lepidoptera: Tortricidae), and infestation levels have even increased in recent years in many regions (Dorn et al. 1999). The larvae of this moth feed primarily on apple and several other pome and stone fruits, causing direct fruit injury (Beers et al. 2003). The phyllophagous apple leaf miner (Lyonetia clerkella L.) (Lepidoptera: ), is a frequently occurring moth in fruit orchards of the Palaearctic zone, mainly on apple and cherry trees, and serious infestation can lead to leaf desiccation and affect shoot growth (Beers et al. 2003). In high-value crops such as apple (Malus x domestica Borkh.) there is substantial demand to limit and diminish the amount of yield loss caused by herbivore damage for economic reasons. To keep the populations below the economic injury levels, which are for codling moth as low as 0.3-1.0% infested fruits (Mani et al. 1997, Knight and Light 2005a), integrated pest management (IPM) systems that combine different tactics are considered promising. The reliance on insecticides as primary control option has decreased, as codling moth resistance even against biopesticides increases (Reyes et al. 2007), and control success of leafminer is often limited by the larvae's cryptic habit (Beers et al. 2003). Particularly for the codling moth a number of environmentally friendly means of control have been developed, including mating disruption (Beers et al. 2003), sterile insect technique (Bloem et al. 2006), and the use of natural enemies (Häckermann et al. 2008), or microbial products such as granuloviruses or Bacillus thuringiensis (Bt) (Cross et al. 1999, Lacey and Shapiro- Ilan 2008). However, these methods are only successful at low infestation levels. Thus, planting resistant cultivars that reduce insect pest populations is considered as an important component of IPM systems comprising also biological and cultural control (Avilla and Riedl 2003). Resistant host plants have gained importance since the genetic characterization of the apple genome with molecular markers was accelerated by time and cost efficient molecular techniques (Silfverberg-Dilworth et al. 2006). Molecular markers are used to identify genetically based resistance by QTL (quantitative trait locus) analysis, and are a prerequisite for marker assisted selection of resistant cultivars (Francia et al. 2005).

16 In apple, detailed knowledge has been gained recently on the genetic basis of resistance to a fungal and a bacterial pathogen (Calenge and Durel 2006, Gessler et al. 2006, Khan et al. 2006), while little information is available on the genetic basis of resistance to insect pests, principally arthropods. Major genes and QTLs have been detected for apple resistance to aphid species, e.g. the leaf-curling aphid (Dysaphis cf. devecta Wlk.) (Roche et al. 1997, Stoeckli et al. 2008c), or the woolly apple aphid (Eriosoma lanigerum Hausm.) (Bus et al. 2008). Detailed knowledge about the genetic basis of resistance in apple to lepidopteran herbivores is still missing, despite the high significance of moth species as pest insects in this perennial crop. In contrast, genomic regions related to resistance to lepidopteran pests were found for herbaceous plants, mainly for maize (Zea mays L.) (Brooks et al. 2007), rice (Oryza sativa L.) (Selvaraju et al. 2007), soybean (Glycine max L.) (Komatsu et al. 2005), tomato (Lycopersicon hirsutum f. glabratum Mull.) (Moreira et al. 1999), and cotton (Gossypium hirsutum L.) (Wan et al. 2007). To our knowledge, there exists only one QTL report for moth resistance in a fruit tree, namely for resistance to the citrus leaf miner (Phyllocnistis citrella) in citrus trees (Citrus aurantium L. x Poncirus trifoliata) hybrids (Bernet et al. 2005). Nevertheless, different authors proposed a genetically based resistance in commercial apple to C. pomonella (Cossentine and Madsen 1980, Goonewardene and Kwolek 1985, Maindonald et al. 2001, Lombarkia and Derridj 2002). It is therefore highly promising to evaluate genetic markers associated with apple resistance to lepidopterans as a basis of marker assisted selection. Incorporating QTLs into an apple cultivar has proved to be more efficient than considering only one major gene alone (Gardiner et al. 2007). However, several QTLs are necessary for durable resistance, and low heritability makes it difficult to find a significant association between a marker and a phenotypic trait (Gardiner et al. 2007). Besides resistance to diseases and pests, marketing-relevant tree and fruit qualities (e.g. tree architecture, productivity, storability of fruits, fruit size, or -flavor) are primary factors for cultivar selection (Kellerhals et al. 1997, Beers et al. 2003). Such traits are largely controlled by environmental conditions such as temperature, light or soil fertility, and cultural manipulation such as pruning (Liebhard et al. 2003a). It is known that insect resistance is partly based on morphological plant traits such as e.g. pubescence, fruit color or -shape, and surface waxes (Smith 2005b). Apple varieties resistant to C. pomonella seem to have a small fruit size and a strong fruit firmness (Cossentine and Madsen 1980). The investigation of fruit characteristics is therefore a prerequisite to understand variable resistance to diseases and pests in apple in different environments.

17 The objective of the study was to identify molecular markers in apple linked to resistance or susceptibility to representatives of each a leaf-feeding and a fruit-feeding lepidopteran pest. Fruits infested by C. pomonella, and number of L. clerkella mines were assessed in the field as measures of host-plant resistance using 160 'Fiesta' x 'Discovery' progeny plants. The genetic background of apple resistance to lepidopteran pests, and of two fruit traits (fruit number and fruit diameter), was evaluated by QTL analysis.

4.2. Materials and methods

Study site and plant material Moth surveys and fruit trait measurements were carried out at two different study sites in Switzerland, the Ticino (Cadenazzo; at 46°09'35''N, 8°56'00''E, 203 m altitude) and the Valais (Conthey; at 46°12'30''N, 7°18'15''E, 478 m altitude) sites, and in two consecutive years (2005 and 2006 = Year 1 and Year 2). Orchards were treated with fertilizers and herbicides, but no insecticides, fungicides, or codling moth mating disruption technique were applied. At both study sites, a fruit thinning was carried out without any bias for infested fruits. At the Ticino site fruitlets were removed by hand in June, and at the Valais site blossoms were removed chemically in May (Frufix, Rhodofix). The studied apple genotypes represent a segregating F1 population of the cultivars 'Fiesta' x 'Discovery' described by Liebhard et al. (2003b). Each genotype was bud grafted on M27 rootstocks, and planted once at each site in winter 1998/1999. Mean tree height was 248 ± 3 cm (Ticino; mean ± SE) and 254.8 ± 4 cm (Valais). Mean stem diameter was 12.5 ± 0.2 cm (Ticino) and 10.4 ± 0.2 cm (Valais). All 160 tree genotypes that were present at both sites were surveyed. As some trees died after plantation establishment, and as for specific analyses (e.g. neighborhood effect) not all genotypes were considered, sample size differed from the maximum genotypes for some characters.

Phenotypic data The number of C. pomonella larval penetrations into fruits was inspected only on fruits attached to the tree, not on fallen fruits. The survey was repeated two times in each of the two years to distinguish between first-generation larvae (June/July) and larvae belonging mainly to the second generation (August, when the main cohort reached the second generation) (Mani et al. 1997). All fruits on each apple tree (in total 40,000 fruits) were

18 screened for C. pomonella larval infestation, and infested fruits were marked with a correction pen (white color; edding 7000). Fruits were not removed nor were they damaged in detection of larvae, as further C. pomonella infestation and data collection should not be biased. A small hole covered with dry frass on the surface of the fruit, and marked by a reddish discoloration around the infestation area was used as indication for larval penetration (Baggiolini et al. 1992). When more than one penetration hole per fruit was detected (a rare event), damage was still regarded as a single larval infestation, as usually no more than a single larva is survived per apple (Barnes 1991). Windfall apples were not included in the survey as it was often not possible to allocate fruits to an individual tree, and secondary infestation by other insects or fungi impeded accurate evaluation. The number of fruits per tree was assessed and the maximum fruit diameter measured simultaneously with the C. pomonella survey in July in both years. All leaf mines of L. clerkella were counted for each individual tree by examining every leaf at the end of July in Year 1. At this time of the year, mines from all generations (2- 3 per year) should be visible and the quantified infestation level should therefore be representative for the whole-year (Baggiolini et al. 1992).

Data analysis All statistical analyses were performed with SPSS 16.0 for Mac OS X (SPSS, Inc., Chicago, IL) and R 2.6.0 (R Development Core Team, Vienna). Abundance data of C. pomonella and L. clerkella infestation, and fruit trait data were tested for normality and homogeneity of variances (one-sample Kolmogorov-Smirnov test, Levene test, and analyses of residuals) (Zar 1999). Number of C. pomonella infested fruits and number of L. clerkella th 2 th mines per tree was transformed with log10 (x+1+c), with c = (25 percentile) / 75 percentile. The value 1 has no impact on the mixed model analyses, and it was added to avoid negative values making a QTL analysis impossible. Number of apples and apple diameter was square root transformed. Spearman’s rank tests were used to assess correlations between different study sites and years. When multiple correlation tests were carried out the Benjamini- Hochberg procedure was applied to correct for false discovery rates (type I errors) (Benjamini and Hochberg 1995, Verhoeven et al. 2005). To evaluate the effect of genotype, site, and year on C. pomonella and L. clerkella abundance, as well as on fruit traits, a three factor mixed model analysis was conducted, with genotype as random factor, and site and year (repeated variable) as fixed factors. The restricted maximum likelihood (REML) method and type III sum of squares were used.

19 The potential tendency of C. pomonella and L. clerkella to occur aggregated on individual apple trees was described by the index of dispersion (ID) using BiodiversityPro 2.0 2 (McAleece, Lambshead and Paterson; The Natural History Museum, London), with ID = s (ν- 1)/m, where m and s2 are the sample mean and variance, respectively, and ν is the sample size 2 (Southwood and Henderson 2000). ID values significantly greater than the χ statistic (0.025 probability level, d.f. = ν-1) indicate an aggregated distribution on individual trees of the studied moth species. To evaluate the possible effect of moth infestation on neighbor-trees on the infestation level of an individual tree, Spearman's rank tests were computed for the number of C. pomonella larval penetrations and the number of L. clerkella leaf mines on an individual tree and on the two neighbor-trees (sum) (neighborhood effect). Potential spatial patterns of C. pomonella and L. clerkella infestation within the orchard were visualized by fitting trend surfaces on contour plots by kriging (best unbiased generalized least squares estimation) with an exponential covariance function (Venables and Ripley 2002).

QTL mapping QTL analyses were carried out separately for each site and year using MapQTL® 4.0 (van Ooijen et al. 2002). The mean value of all surveys and separate mean values for each site and year were calculated for analysis. The genetic linkage maps for both 'Fiesta' and 'Discovery' (single parent maps), used in QTL analysis, had already been published (Liebhard et al. 2003b). The maps consist of a total of 345 ('Fiesta') and 389 ('Discovery') markers and include 137 amplified fragment length polymorphism (AFLP), 108 simple sequence repeats (SSR or microsatellite), and 100 random amplified polymorphic (RAPD) markers in 'Fiesta', and 160 AFLP, 103 SSR, 1 sequence characterized amplified region (SCAR) and 125 RAPD markers in 'Discovery'. Simple interval mapping (IM) was used for QTL analysis. Logarithm of odds (LOD) threshold values were determined by 1,000-fold permutation tests (MapQTL® 4.0), and a QTL was declared as significant at the 95% genome-wide confidence level (Churchill and Doerge 1994). As tests on markers from different linkage groups are independent, a Bonferroni correction [α' = 1 - (1-α)(1/n), n = 17, α = 0.05] was applied to obtain an overall significance level per linkage group (α' = 0.003), and to fix the type I error rate over the entire genome to α = 0.05. A QTL was determined as suggestive using a 5% (α' = 0.05) chromosome-wide significance level (van Ooijen 1999). The one-LOD support interval was calculated, which approximates the position of significant QTLs with 95% confidence (Lander and Botstein 1989). The 'Fiesta' x 'Discovery' population was divided into

20 subpopulations based upon the presence/absence of the marker closest to the detected QTL and phenotypic differences were tested considering the two subpopulations (Mann-Whitney test). Possible QTL interactions were tested by multiple QTL mapping (MQM) for QTLs with LOD scores exceeding the LOD threshold values in IM. The proportion of variation in moth infestation and fruit traits that can be explained by the genetic variation among the apple 2 progenies was analysed by broad-sense heritability, which was estimated by the formula H = 2 2 2 2 2 2 σ g/σ p and σ p = ([σ g + σ e]/n), where σ g is the genetic variance (genotype variance 2 2 estimate; Table 4.2), σ p is the phenotypic variance, σ e is the environmental variance (residual variance estimate; Table 4.2), and n is the number of replicates per genotype (Lauter and Doebley 2002, Calenge et al. 2004).

4.3. Results

Phenotypic data The mean infestation rate of the 40,000 fruits screened for C. pomonella was lower in Year 1 compared to Year 2, and it was higher at the Valais site (between 1.1 and 6.5%) than at the Ticino site (between 0.4 and 1.3%) (Table 4.1). Incidence of C. pomonella occurrence varied between 9.0% and 28.0% infested trees at the Ticino site, and between 18.2% and 71.8% at the Valais site (Table 4.1). The number of L. clerkella mines per tree ranged from 0.6 (Ticino) to 28.4 (Valais) (Table 4.1). L. clerkella incidence ranged from 19.5% (Ticino) to 96.6% infested trees (Valais) (Table 4.1). The number of C. pomonella infested fruits per tree did not correlate between years, neither in July nor in August (Spearman's rank test; P > 0.05). C. pomonella infestation per tree within the same year (July and August surveys) was highly correlated at both study sites in Year 1 (Spearman's rank test; Ticino: n = 144, rs = 0.216, P = 0.009; Valais: n = 143, rs =

0.207, P = 0.013) and in Year 2 (Ticino: n = 143, rs = 0.277, P = 0.001; Valais: n = 142, rs = 0.346, P < 0.001). C. pomonella infestation of trees with the same genotype was correlated between the two study sites in August of Year 1 (Spearman's rank test; n =131, rs = 0.305, P < 0.001), but not in July of Year 1 and in July and August of Year 2 (P > 0.05). No relation was found between study sites for L. clerkella infestation (Spearman's rank test; P > 0.05).

21 Table 4.1. Codling moth (Cydia pomonella) and apple leaf miner (Lyonetia clerkella) infestation of apple trees of differing genotype. Number of studied apple genotypes (n), mean ± standard error (SE), maximal value (Max) and incidence (I = % infested trees) are presented. Mean and maximal values refer in C. pomonella to % of infested fruits per tree assessed in July and August of two consecutive years, in L. clerkella to number of mines per tree in July of the first study year.

Ticino Valais n Mean ± SE I n Mean ± SE I (Max) (Max)

Cydia pomonella Year 1 0.4 ± 0.2 1.1 ± 0.3 July 133 9.0 136 18.2 (14) (18) 0.3 ± 0.1 2.4 ± 0.5 August 133 6.3 136 27.3 (9) (38) Year 2 1.3 ± 0.3 4.5 ± 0.5 July 142 28.0 137 56.3 (25) (43) 1.0 ± 0.3 6.5 ± 0.7 August 135 16.1 135 71.8 (25) (50) Lyonetia clerkella Year 1 0.6 ± 0.1 28.4 ± 1.9 July 149 (8) 19.5 149 (119) 96.6

A mixed model analysis revealed that genotype contributed significantly to variance in C. pomonella infestation and fruit number, but not to variance in L. clerkella infestation and fruit diameter (Table 4.2). C. pomonella and L. clerkella infestation and fruit number of apple genotypes varied significantly between sites and years, whereas fruit diameter was stable among sites (Table 4.2). A possible variation in resistance among apple genotypes was indicated for both moth species by the index of dispersion (ID) (Table 4.3). While some trees were highly infested, others were not infested at all. ID values ranged between 152 and 293 for C. pomonella. The strongest aggregation was detected at the Valais site in Year 2 (P < 0.0001), and C. pomonella infestation was randomly distributed at the Ticino site in Year 1 (P > 0.05) (Table 4.3). ID values for L. clerkella were 533 (Ticino) and 2,870 (Valais) and infestation was highly aggregated (P < 0.0001).

22 Table 4.2. Effect of genotype, site and year on C. pomonella and L. clerkella infestation, and on fruit number and -diameter, assessed by mixed model analysis. The model was conducted with genotype as random factor, and site and year (repeated variable) as fixed factors.

Tests of fixed effects Variance components

Factor d.f.1 F value P value Factor Estimates P value

C. pomonella Site 451 175.3 <0.0001 Genotype 0.0053 0.044 Year 412 145.8 <0.0001 Residual 2 0.0251 <0.0001 Year x site 412 101.5 <0.0001

L. clerkella Site 151 1,791.7 <0.0001 Genotype 0.0074 0.241 Residual 2 0.0666 <0.0001

Fruit number Site 480 20.8 <0.0001 Genotype 0.9090 0.013 Year 480 4.4 0.036 Residual 2 4.1400 <0.0001 Year x site 480 8.5 0.004

Fruit diameter Site 440 0.6 0.457 Genotype 0.2568 0.561 Year 440 25.5 <0.0001 Residual 2 1.5407 <0.0001 Year x site 440 1.3 0.263

1 Denominator d.f. (Numerator d.f. = 1) 2 Error term (variance) associated with genotype

Table 4.3. Index of Dispersion ID values assessed to characterize the distribution of C. pomonella (% infested fruits) and L. clerkella (number of leaf mines) infestation on 2 individual apple trees. ID values significantly greater than the χ statistic (0.025 probability level) indicate an aggregated distribution of the studied species. *P < 0.05; ***P < 0.0001, n.s.P > 0.05 (random distribution).

Ticino Valais

d.f. ID values d.f. ID values

C. pomonella Year 1 132 152n.s. 135 178* Year 2 134 179* 134 293*** L. clerkella Year 1 148 533*** 148 2,870***

23 No significant neighborhood effects (relationship between infestation of an individual tree and the two neighbor-trees) were found for C. pomonella in 3 out of 4 analyses and for L. clerkella in all analyses. The only exception occurred at the Valais site in Year 2, where a weak correlation was found (Spearman's rank test; n = 54, rs=0.370, P < 0.05). A visualization of C. pomonella and L. clerkella abundance within the apple orchard did not indicate a strong spatial pattern of these insect herbivores (supplementary material; Figure SM 4.5.1).

QTL analysis for C. pomonella and L. clerkella resistance in apple Broad-sense heritability for moth infestation was 29.9% (C. pomonella) and 18.2% (L. clerkella). A significant QTL for C. pomonella susceptibility in apple was detected by simple interval mapping (IM) on the 'Discovery' linkage group 10 at the Valais site in Year 2 and for the mean value of all surveys (Figure 4.1). Multiple QTL mapping (MQM; data not shown) did not reveal any additional QTLs and results were therefore based on IM. The closest marker to the QTL was the RAPD marker Z19-350 located at 66.2 cM. The LOD score for the mean value of all surveys at the marker position was 2.72 (LOD threshold for a genome wide error rate of 5%: 2.6) and the QTL explained 8.2% of the phenotypic variability. A separate QTL analysis for each site and year was carried out. The LOD scores at the marker position for the two consecutive years were 1.08 and 2.01, respectively (Valais), and 0.32 and 1.25, respectively (Ticino) (LOD thresholds for suggestive/significant linkage: 1.6-1.8 /2.5- 2.8). A higher C. pomonella infestation was found for the apple subpopulation amplifying the marker Z19-350 compared to apple genotypes not amplifying the marker at all sites in both years (Figure 4.2). The difference was significant for the Valais site in two consecutive years. These results are confirmed by the joint segregation analysis of the identified QTL marker and C. pomonella infestation, which revealed that more uninfested genotypes were found for the subpopulation not amplifying the RAPD marker in comparison to the other subpopulation in three of four assessments (Figure 4.2). The nearest SSR marker flanking this QTL was the SSR CH04g09y (67.5 cM). No significant associations between this marker and apple resistance to L. clerkella was detected for any year and site.

QTL analysis for fruit number and –diameter Broad-sense heritability was 21.9% (fruit number) and 16.7% (fruit diameter). A putative QTL for fruit number was found on the 'Fiesta' linkage group 12 using IM. The closest marker to the QTL peak was the SSR CH01g12-112/156 located at 43.6 cM. There

24 was a significant association between the number of fruits and the SSR marker at the Valais site in Year 1. The significant LOD score was 3.53 (LOD threshold for suggestive/significant linkage: 1.8/3.0) and explained 9.6% of the phenotypic variability. The apple genotypes amplifying the allele '156 bp' of the SSR CH01g12-112/156 produced significantly less apples compared to apple genotypes not amplifying the allele at the Valais site in Year 1. A significant QTL for fruit diameter was not found.

Figure 4.1. QTL for susceptibility in apple to Cydia pomonella identified on linkage group 10 of 'Discovery' based on IM results. The x-axis indicates the linkage map of 'Discovery' in cM and the marker names; the y-axis shows the LOD scores. LOD threshold levels for a significant QTL varied between 2.5 (Ticino Year 1) and 2.8 (Valais Year 1&2). LOD threshold levels for a suggestive QTL ranged between 1.6 (Ticino Year 1) and 1.8 (Ticino Year 2). The LOD threshold level for the mean value (of all surveys) was 1.7 (suggestive) and 2.6 (significant). The solid black bar indicates the one- LOD support interval for the position of the QTL (mean value). The marker closest positioned to the QTL is underlined.

25

Figure 4.2. Mean Cydia pomonella infestation (number of infested fruits per tree) of the two subpopulations of the 'Fiesta' and 'Discovery' cross, based on presence and absence of the RAPD marker Z19-350 ('Discovery' linkage group 10). Significant differences between subpopulations are indicated by *P < 0.05 (Mann-Whitney U-test). There was no significant segregation distortion in the number of genotypes (shown as figures within the bars) of the two subpopulations (χ2 statistic; P = 0.11-0.27). The percentage of uninfested genotypes was calculated for the two subpopulations.

26 4.4. Discussion

We identified a quantitative trait locus (QTL) for apple susceptibility to the codling moth, C. pomonella, on the 'Discovery' linkage group 10, which, together with the detection of significant genetic variation (H2=29.9%), provides statistical evidence for the genetic basis of C. pomonella resistance as before suggested (Cossentine and Madsen 1980, Goonewardene and Kwolek 1985, Maindonald et al. 2001, Lombarkia and Derridj 2002). When the mean value of the data from all surveys was analysed, the LOD score indicated a significant linkage between marker and C. pomonella susceptibility. There was only a suggestive linkage using phenotypic data for the Valais site in Year 2. This may reflect variable C. pomonella infestation between years, which can be explained by differing environmental conditions. In contrast to C. pomonella, no significant QTL for resistance to the apple leaf miner, L. clerkella, was identified, either because of lack of sufficiently strong QTLs in the segregating

F1-cross of the apple cultivars studied, or because of strong effects of the biotic and abiotic environment on L. clerkella, which will be discussed for C. pomonella below. The presented results provide the basis for comparisons with expected follow-up studies considering different apple varieties. The identified QTL for C. pomonella explains a relatively small fraction of the variable resistance level only. Major QTLs appear to be more common for resistance in apple to foliar pathogens. For example, in an apple cross between 'Golden Delicious' and 'Murray', all apple scab [Venturia inaequalis (Cke.) Wint] resistant plants amplified the '228 bp' allele of the SSR Hi07h02, whereas the '274 bp' allele was present in all the scab susceptible plants (Patocchi et al. 2005). This perfect co-segregation of the Hi07h02 marker and the Vm scab resistance gene highlights the strong interaction between a plant pathogen and its determinant, similar to other major genes for apple scab (Gessler et al. 2006). In contrast to lepidopteran herbivore-plant relationships, the complex process of host-finding is absent in pathogen-plant interactions, which may facilitate the evolution of major pathogen resistance genes under variable environmental conditions. Most lepidopteran insects are winged and can actively choose the habitat for their progeny before landing, and subsequently adjust the number of eggs deposited per plant after landing. Finally, the host plant can interfere with feeding of the larvae that hatched from the eggs, and only this third step in the interaction with the host parallels the plant-pathogen interaction. The documented QTL for resistance to the codling moth might be expressed in plant traits relevant for one or two steps of its host location process. Plant features mediating apple and codling moth interactions have been discovered

27 mostly during the past decade. Such features may underlie the resistance reported in this study, and their knowledge might facilitate efficient future marker-assisted selection of resistant apple genotypes. Prior to landing, a relatively specialized moth species such as C. pomonella is expected to rely largely on olfaction (Ramaswamy 1988). In fact, while there is no known association of fruit color and codling moth susceptibility (Cossentine and Madsen 1980), females readily select host trees based on olfaction stimuli (Yan et al. 1999, Vallat and Dorn 2005). Different apple cultivars emit qualitatively and quantitatively different blends of volatiles (Hern and Dorn 2003, Mehinagic et al. 2006), and resulting attraction or repulsion of female codling moths depends on the composition of the blend (Vallat and Dorn 2005). Among the single apple compounds with documented attraction or repellent effect on codling moth females in bioassays, the attractant ester butyl hexanoate (Hern and Dorn 2004) was released in significantly lower amounts by the cultivar 'Maigold' than by 'Golden Delicious' (Hern and Dorn 2003), indicating a genetic basis for these behaviorally relevant phenotypic differences between cultivars. After landing, the primary sensory modality of female lepidopteran insects is again considered chemical (reviewed by Honda 1995), and apple cultivars strongly vary in their composition of non-volatile compounds (Wu et al. 2007). Oviposition in C. pomonella is stimulated by the primary metabolites fructose or sorbitol, two sugars present in green leaves and in high amounts in fruits (Lombarkia and Derridj 2002). It is further stimulated by the secondary metabolite (E,E)-α-farnesene (Sutherland et al. 1977), a sesquiterpene exhibiting a female-specific behavioral effect also prior to landing (Hern and Dorn 1999). While no differences in this compound have been found in headspace volatile analyses between different apple cultivars (Hern and Dorn 2003), such differences have been reported for the secondary plant metabolite phloridzin (Hunter and Hull 1993) that mediates apple resistance to the green apple aphid, Aphis pomi (De Geer) (Montgomery and Arn 1974) and the Japanese beetle, Popillia japonica (Newman) (Fulcher et al. 1998). Besides chemical cues, morphological leaf and fruit traits can influence C. pomonella oviposition (Hagley 1976, Plourde et al. 1985). For example epicuticular waxes (Belding et al. 1998) or fruit firmness (Johnston et al. 2001) have been described to vary between cultivars. Different physical plant traits were found to affect the feeding-initiation of neonate C. pomonella larvae. In a growth chamber no-choice experiment, a lower number of C. pomonella larval fruit entries were noted on cultivars with a pubescent lower leaf surface after

28 48 hours of exposure (Plourde et al. 1985). Surface waxes and fruit firmness are additionally described to prevent larval entrance into the fruit (Cossentine and Madsen 1980). Knowledge about the genetic background of plant traits improves the understanding of herbivore-plant relationships. As fruit characteristics are of high relevance for fruit feeding pests including C. pomonella, we focused on the number of apple fruits per tree and fruit diameter. QTLs for fruit traits, such as fruit flesh firmness, fruit weight, fruit color or sugar content have been previously identified (Liebhard et al. 2003a, Chagné et al. 2007). In the current study a putative QTL for the number of apple fruits was identified on the 'Fiesta' linkage group 12. Other QTLs for fruit number have been detected on linkage group 5 and 15 ('Fiesta'), and 5 and 16 ('Discovery') (Liebhard et al. 2003a) but were not confirmed by the current study. In the cited study, the QTLs for fruit number were assessed in 2001-2002, investigating the same apple trees as our study. We therefore postulate that the genotype- phenotype relationship of fruit traits may change when trees reach their full productivity. The effects of plant genotype on such traits may be more pronounced in younger apple trees than in mature apple trees that have been exposed to varying environmental conditions for many years. We consider the QTL for C. pomonella resistance identified in this study as independent of the addressed plant traits because the putative QTLs for fruit number (this study), as well as for shoot length or stem diameter (Stoeckli et al. 2008c) were positioned on different linkage groups within the apple genome compared to the QTL for C. pomonella susceptibility. Seasonal changes and dependency on environmental conditions have been reported for physical (Johnston et al. 2001, Volz et al. 2003) and chemical plant features (Vallat and Dorn 2005, Mody et al. 2008). Environmental conditions may therefore affect the expression of a QTL (Tingey and Singh 1980), and the results of the present study indicate that such conditions are of high relevance in trees, especially considering lepidopteran pests. We did not find a correlation of C. pomonella infestation between years and between study sites. The amount of rainfall, for example, was much higher at the Ticino site compared to the Valais site (www.meteoschweiz.ch). Rainfall had a significant influence on changes in volatile release from apple trees (Vallat et al. 2005), influencing emission rate of compounds with effects on odor-discrimination of adult codling moth females (Vallat and Dorn 2005). Rainfall also affects oviposition, resulting in reduced C. pomonella populations (Hagley 1976). Furthermore, environment-based within-tree variability of herbivore infestation may impede the identification of a possible genetic background of resistance (Unsicker and Mody 2005). In fact, C. pomonella infestation was higher on the southern, warmer part of the tree, a finding

29 which was more distinct at the Ticino site (Stoeckli et al. 2008b), which may explain why the QTL for C. pomonella resistance was not found at this site. Despite the constraints given by the complex host-finding process of the moths, and the variability of relevant cues due to changing seasonal and climatic conditions, we were able to identify a QTL for C. pomonella susceptibility in apple. This finding along with the putative QTL for fruit number is expected to facilitate the development of resistant cultivars with high fruit yield by marker assisted selection. Combining major resistance genes and QTLs in a single apple cultivar helps to establish durable resistance (Gardiner et al. 2007). The putative QTL for fruit number was not associated with fruit diameter, indicating a genetic independence of the two characters. This finding is underlined by the quantitative manner of both fruit traits. However, a negative relationship between the fruit load and fruit weight in the field is reported (Fukuda and Takishita 1998). Specific management techniques, such as fruit thinning, will help to ensure that selection for fruit number does not result in small fruit.

30 4.5. Supplementary material (SM)

Figure SM 4.5.1. Spatial distribution of (a-d) Cydia pomonella and (e, f) Lyonetia clerkella on trees along the (horizontal) rows is described for the study sites by fitting trend surfaces on contour plots with an exponential covariance function. Mean moth infestation from different observations within a year was considered for contour plots (cf. Table 4.1). Infested trees are indicated by filled (orange) circles with infestation level reflected by circle size; uninfested trees are indicated by open (black) circles. No contour lines were plotted for the Valais site due to the orchard design with two rows.

31 5. Influence of canopy aspect and height on codling moth (Lepidoptera: Tortricidae) larval infestation in apple, and relationship between infestation and fruit size 2

Monitoring systems based on traps with female attractants are expected to enhance forecasting of insect population size and damage. The optimal placement of such traps should match the small-scale distribution of ovipositing females. In the codling moth (Cydia pomonella) (Lepidoptera: Tortricidae), fruit infestation takes place in close proximity to the oviposition site. We characterized the within-tree distribution of codling moth infestations as well as the size of uninfested fruit based on a survey of 40,000 apples (Malus x domestica), from trees belonging to 160 different apple genotypes and growing in two different environments. Each tree was subdivided into twelve sectors, considering canopy aspect (north, east, south, west) and canopy height (bottom, middle, top). This study revealed that fruit infestation by the first but not by the second generation of larvae correlated significantly with canopy aspect. Similarly, fruit size differed significantly between the north- and the south-facing tree side for the period of infestation by the first but not by the second larval generation. Significantly lower fruit infestation was observed on the north- compared to the south- or east-facing tree side for the first generation. A significant influence of canopy height on larval infestation was observed in three out of eight assessments, in which the middle height level showed the highest infestations. Significant differences in within-tree distribution of codling moth infestation suggest that oviposition preference is guided by nonrandom factors including microclimate, fruit phenology and wind direction. These cultivar- independent findings should be considered in future monitoring systems that focus on female codling moth.

______2 Stoeckli, S., K. Mody, and S. Dorn. 2008. Influence of canopy aspect and height on codling moth (Lepidoptera: Tortricidae) larval infestation in apple, and relationship between infestation and fruit size. Journal of Economic Entomology 101:81-89.

32 5.1. Introduction

The development of an effective monitoring system for an herbivore pest ultimately relies on accurate identification and exploitation of the pest's needs for resources, including oviposition sites for females and food sources for larvae (Prokopy et al. 1999). Such knowledge on spatial habitat use by adult female herbivores ready to oviposit is of particular significance for key pests such as the codling moth (Cydia pomonella L.) (Lepidoptera: Tortricidae). The codling moth is a major pest in apple (Malus x domestica Borkh.) orchards worldwide that requires frequent control measures (Dorn et al. 1999). In this species, infestation of growing and ripening fruit by neonate larvae takes place in close proximity to the female's oviposition site in the tree canopy (Geier 1963). Trap captures with male pheromone as well as visual inspection to locate infested fruit on trees are common tactics of surveillance (Kellerhals et al. 1997). However, recent advances in ecology and behavior of the codling moth focus on plant-derived volatiles to attract females, either in conjunction with male attraction (Knight and Light 2001) or as a female-specific attractant (Hern and Dorn 1999, 2004). Traps containing a female-attracting lure should be positioned near the major oviposition sites (Knight and Light 2005b). Such lures should be especially attractive for mated females ready to oviposit, as they are known to be more responsive than non-mated females to attractants including the pear ester ethyl (2E, 4Z)-2,4-decadienoate (Knight 2006), butyl hexanoate (Hern and Dorn 2004), and (E, E)-α- farnesene (Hern and Dorn 1999). Meaningful visual inspection for fruit damage should be based on a systematic monitoring that avoids bias. In contrast to these needs, little information on field distribution patterns for codling moth females and larvae, especially with respect to within-tree distribution, is available. The few studies related to within-tree distribution of codling moth females and larvae carried out during the first half of the 20th century did not analyze data statistically, and were limited to spray coverage at different canopy levels (List and Newton 1921, Bobb et al. 1939, Woodside 1944). Other factors related to within-tree fruit infestation patterns, especially the effects of canopy aspect, were not studied in detail, and this gap in knowledge hampers field surveillance of codling moth females. In fact, (Trimble and El-Sayed 2005) considered current positioning of traps containing the pear ester ethyl (2E, 4Z)-2,4-decadienoate, a bisexual codling moth attractant, suboptimal for female captures, and suggest changing trap positioning within the tree canopy. This conclusion highlights the significance of studies into microhabitat distribution of codling moth females and larvae.

33 Several abiotic factors influencing codling moth behavior, such as temperature, light interception, and exposure to rainfall and wind, may vary with canopy aspect and height (Kappel and Quamme 1993, Graf et al. 2001, Kührt et al. 2006a, b). Both, the minimum (15°C) and optimum (25°C) temperatures for oviposition may be reached at a given time in some parts of a tree canopy but not in others (Saethre and Hofsvang 2002). Differences in temperature are well perceived by adult codling moth (Bloem et al. 2006), and temperatures at oviposition sites that accelerate development of the progeny are preferred by mated females in the laboratory (Kührt et al. 2006a, b). In addition, rainfall has a negative effect on codling moth oviposition (Hagley 1976), and this avoidance behavior is most likely mediated by marked changes in the tree's volatile emissions (Vallat and Dorn 2005). Wind may support or disrupt oviposition activity. Wind direction may be used for orientation by the adult females that are moving upwind for kairomone-aided host plant finding (Hern and Dorn 2002, Reed and Landolt 2002, Hern and Dorn 2004), but wind above 5 m/s inhibits codling moth flight (Russ 1961). In addition to directly affecting moth activity, the mentioned environmental parameters hold the potential to affect fruit phenology, and to generate within-tree variation of fruit suitability for oviposition at a specific time of the season. This variation in plant phenology may influence codling moth oviposition behavior. Studies show that blossom and fruit formation early in the season is most advanced at the tree tops and that moth activity is also highest in this part of the tree (Blomefield et al. 1997). This study aimed at characterizing within-tree distribution of codling moth infestation as a basis for improved monitoring of mated, females ready to oviposit, and of infested fruit in the canopy. Previous anecdotal reports were limited to one or few cultivars (List and Newton 1921, Bobb et al. 1939, Woodside 1944). We made detailed assessments on a large number of different apple genotypes under different environmental conditions in order to provide data transferable to different apple cultivars and environments. To interpret the data, we tested the hypothesis that mated female codling moth avoid cool microhabitats that delay development of their progeny (Kührt et al. 2006b). Additionally, we analyzed measurements of wind direction and assessed fruit phenology in order to gain an insight into further potential factors influencing small-scale distribution of infestation sites.

34 5.2. Materials and Methods

Plant material and orchard locations Codling moth infestation was studied on apple trees of 160 different genotypes replicated in two different environments. The trees were grown from seeds of a cross between the apple cultivars 'Fiesta' and 'Discovery', and replicated by bud grafting on M27 rootstocks (Liebhard et al. 2003a). Each genotype was planted in 1998 in two different regions of Switzerland, at the Valais study site (Sion; 46°12'30''N, 7°18'15''E; 478-m altitude) and the Ticino study site (Cadenazzo; 46°09'35''N, 8°56'00''E; 203-m altitude). Each of these sites are located in flat valleys. The trees were planted in rows 3.5 m apart with a tree-to-tree distance of 0.5 m at the Valais study site and of 1.25 m at the Ticino study site. Trees had an average height of 2.5 m (dwarf trees) and were aligned in north-south direction (Valais) and in east- west direction (Ticino). Climatic conditions of the two study sites were similar regarding temperature, but the amount of rainfall was higher at the Ticino study site compared to the Valais study site. In July and August 2006, for example, the average monthly temperature was 23.3 and 16.3°C at the Valais study site, and 23.2 and 18.6°C at the Ticino study site. The total monthly rainfall amounted to 37.6 and 84.2 mm at the Valais study site compared to 152.9 and 199.0 mm at the Ticino study site. Agrochemical treatment of the trees included fertilizers and herbicides but no insecticides for codling moth control.

Codling moth survey Distribution of codling moth larval infestation was used as a measure of effective within-tree distribution of mated, ovipositing codling moth females, as the pattern of larval infestation does not differ from that of egg distribution (Geier 1963). Larval infestation of fruit was assessed for apples attached to the tree. Windfall apples were not considered. The survey was conducted in two consecutive years (2005 = Year 1; 2006 = Year 2). The apples were examined with a focus on first generation larvae (June/July) and on larvae belonging mainly to the second generation (August, i.e. when the main cohort reached the second generation (Mani et al. 1997). For each infested apple, canopy aspect (north, east, south, west) and canopy height (bottom, middle, top) were determined, resulting in a total of 12 investigated tree aspect sectors. All fruit hanging on the 160 apple genotypes were assessed for the presence of codling moth infestation. Apples were neither removed nor damaged during detection of larvae in order to avoid any bias regarding infestation. Indicator for codling moth larval penetration was a small hole covered with dry frass on the surface of the

35 fruit, surrounded by a reddish discoloration. This method was regarded as sufficiently accurate as other insects causing comparable damage were not present at the study sites. Absence of other pests was confirmed using pheromone traps for smaller fruit tortrix (Grapholita lobarzewskii Ragonot) (Deltatrap PET, Andermatt Biocontrol, Grossdietwil, Switzerland), and by means of visual control for the fruitlet mining tortrix (Pammene rhediella Clerck), and the European apple sawfly (Hoplocampa testudinea Klug).

Measurement of fruit traits In addition to the total number of apples per tree (Year 1 and Year 2), the number of apples within the 12 tree sectors was counted (Year 2). The diameter of not infested fruit was assessed 1-2 d after the codling moth survey as a measure of fruit phenology.

Temperature and wind data Leaf temperature at the south- and north-facing tree side was modeled employing the ecosystem model SiB 3 (Simple Biosphere model version 3) for Year 2 (Sellers et al. 1997, Vidale and Stöckli 2005). Surface radiation, energy- and water-balances were calculated at every time step. The model was run with hourly atmospheric forcing data at reference level (10 m). Down-welling radiation (short-wave and long-wave), wind speed, temperature, relative humidity, rainfall and surface pressure, relevant for the Valais and Ticino study site (measured in close vicinity at Sion and Magadino, respectively) were provided by MeteoSwiss (www.meteoschweiz.ch). Long-wave radiation was derived from temperature and humidity (Idso 1981). Surface pressure was derived from temperature and site elevation. A site-specific tuning was only performed for the following parameters: canopy level (2.5 m), bottom of tree crown (1.0 m), vegetation cover fraction (50%), leaf width (4.5 cm), leaf length (10 cm), and leaf area index (2.5). North-south differences in canopy temperature could be modeled accurately only at local solar noon ± 2 h (10:00 - 14:00), when the north-facing tree side corresponded to shaded leaves and the south-facing tree side to sunlit leaves, and during night (22:00 - 02:00), as the remaining thermal radiative forcing is non-directional. At other times the directional attribution of sunlit and shaded leaves would require an explicit and spatially resolved three-dimensional canopy radiative transfer model.

Statistical analysis The number of larval infestations, the number of apples and the fruit diameter were averaged for each tree, and for parameters of canopy aspect and height. As assumptions of

36 parametric tests were not met after data transformations, non-parametric tests were used to analyze the influence of canopy aspect and height on fruit infestation (Sokal and Rohlf 1995, Zar 1999). All statistical analyses were performed with SPSS 11.04 for Mac OS X (SPSS, Inc., Chicago, IL), with the exception of the Friedman's test including Nemenyi post hoc tests, which were calculated with SsS 1.1a for PC (Rubisoft Software GmbH, Eichenau, Germany). Significance levels for the Nemenyi post hoc test were set at P < 0.05. Meteorological forcing datasets were processed with IDL (Interactive Data Language, Creaso GmbH, D-Gilching, Germany). SiB3 was compiled with gfortran 90 (Silicon Graphics, Inc., Mountain View, CA).

5.3. Results

Codling moth larval infestation In total, approximately 40,000 apples were surveyed for codling moth infestation in situ. Larval infestation was lower in Year 1 compared to Year 2, and it was higher at the Valais study site (between 1.4 and 5.2%) than at the Ticino study site (between 0.4 and 1.0%) (Table 5.1).

Table 5.1. Number of examined apple fruit on experimental trees, and percentage of fruit infested by Cydia pomonella larvae in July (first generation larvae) and in August (mainly second generation larvae) at the study sites Valais and Ticino in two consecutive years.

Surveyed apple fruit (No.) Infested apple fruit (%)

Year 1 Year 2 Year 1 Year 2

Valais

July 3,985 6,346 1.4 4.3

August 1,916 6,422 4.7 5.2

Ticino

July 8,169 6,737 0.4 1.0

August 2,029 3,401 0.8 0.9

37 Canopy aspect The lowest mean values for codling moth larval infestation were found for the north- facing tree side in six of eight assessments (Figure 5.1). Infestation by first generation larvae (July) was significantly influenced by canopy aspect in Year 1 at the Ticino study site (Friedman test; n = 21, χ2 = 20.9, P < 0.001), and in Year 2 at both study sites (Valais: n = 74, χ2 = 16.8, P = 0.001; Ticino: n = 35, χ2 = 11.8, P = 0.008). Infestation by first generation larvae was significantly lower at the north-facing than at the south-facing tree side at the

Ticino study site in Year 1 (Nemenyi post hoc; QSouth,North=4.14). In Year 2, infestation by first generation larvae was significantly lower at the north-facing than at the east-facing tree side at the Valais study site (Nemenyi post hoc; QEast,North=4.50). In contrast, infestation by second generation larvae (August) was not significantly influenced by tree aspect in Year 1 (Friedman test; Valais: n = 42, χ2 = 4.1, P = 0.2; Ticino: n =9, χ2 = 2.8, P = 0.4) and in Year 2 (Valais: n = 96, χ2 = 0.6, P = 0.9; Ticino: n = 20, χ2 = 3.0, P = 0.4) (Fig. 5.1).

Canopy height Highest mean values of infestation were found in the top canopy height (Valais, three of four assessments in two years) and in the middle canopy height (Ticino, three of four assessments) (Figure 5.2). In Year 2, canopy height significantly influenced infestation by first generation larvae at the Ticino study site (Friedman test; n =38, χ2 = 16.3, P<0.001) and by second generation larvae at the Valais and Ticino study sites (Valais: n = 92, χ2 = 11.8, P < 0.001; Ticino: n =22, χ2 = 14.5, P = 0.001). At the Valais study site, infestation by second generation larvae was significantly higher in the middle canopy height than in the bottom height (Nemenyi post hoc; QMiddle,Bottom=4.12). Infestation by first- and second generation larvae was significantly higher in the middle canopy height than in the top height at the

Ticino study site (Nemenyi post hoc; first generation larvae: QMiddle,Top=4.06; second generation larvae: QMiddle,Top=5.00). Regarding comparisons of bottom and top canopy height no consistent pattern was detected. Infestation in the top canopy height was generally higher at the Valais and lower at the Ticino study site (Figure 5.2).

38

Figure 5.1. Effect of canopy aspect (N: north, E: east, S: south, W: west) on infestation by Cydia pomonella at the study sites Valais and Ticino in July (first generation larvae) and August (second generation larvae) in Year 1 (A) and Year 2 (B). Number of larval infestations in Year 1 and percentage of larval infestation in Year 2 is given. The percentage was calculated including the number of apples counted at a specific tree side. (n) is the number of infested trees used for the analysis. Different letters a, b indicate significant differences in larval infestation (P < 0.05, Friedman test, Nemenyi post hoc test).

39

Figure 5.2. Effect of canopy height (bottom, middle, top) on infestation by Cydia pomonella at the study sites Valais and Ticino in July (first generation larvae) and August (second generation larvae) in Year 1 (A) and in Year 2 (B). Number of larval infestations in Year 1 and percentage of larval infestation in Year 2 is given. The percentage was calculated including the number of apples counted at a specific canopy part. (n) is the number of infested trees used for the analysis. Different letters a, b indicate significant differences in larval infestation (P < 0.05, Friedman test, Nemenyi post hoc test).

40 Fruit diameter in uninfested apples Canopy aspect had a significant influence on fruit diameter in July at both study sites in Year 1 (Friedman test; Valais: n = 109, χ2 = =18.5, P = 0.001; Ticino: n = 115, χ2 = 44.7, P = 0.001), and at the Ticino study site in Year 2 (n = 35, χ2 = 27.5, P = 0.001) (mean fruit diameter ranging from 3.7 - 4.8 cm; Table 5.2). Fruit at the south-facing tree side were significantly larger than fruit growing at other tree sides at the Ticino study site (Nemenyi post hoc; Year 1: QSouth,North=7.84, QSouth,East=7.73, QSouth,West=7.26; Year 2: QSouth,North=6.48,

QSouth,East=5.89, QSouth,West=4.39). The smallest fruit (mean fruit diameter) were found at the north-facing tree side at the Valais study site in Year 1 (Nemenyi post hoc; QWest,North=5.45,

QSouth,North=3.75, QEast,North=1.34), while no significant differences were found in Year 2. Differences between the east- and south-facing tree side (Valais, Year 1), and the east- and west-facing tree side (Ticino) were not significant. In contrast to these differences found for the period relevant for the main cohort of the first generation codling moth (July), no significant differences in fruit diameter were found for the period relevant for second generation codling moth (August; mean fruit diameter ranging from 4.9 to 5.8 cm), as assessed in Year 2 for the Valais study site (Friedman test; n = 96 trees, χ2 = 1.3, P=0.739) as well as for the Ticino study site (n = 20, χ2 = 2.5, P = 0.479, data not shown). Canopy height did not significantly influence fruit size in July with one exception (Valais Year 1: n = 115, χ2 = 10.3, P = 0.006; Table 5.2), where the bottom canopy height bore significantly smaller fruit than the top height (Nemenyi post hoc; QTop,Bottom=4.43). Similarly, data evaluated in August of Year 2 showed no influence of canopy height on fruit size at the Valais study site (Friedman test; n = 92, χ2 = 3.5, P = 0.170) as well as at the Ticino study site (n = 22, χ2 = 0.6, P = 0.729, data not shown).

Temperature and wind data The maximal difference in modeled temperature between north- and south-facing leaves amounted to 1.6°C (average per hour, 12:00 - 13:00) at the Valais and 1.3°C (11:00 - 12:00) at the Ticino (Figure 5.3). The average temperature difference at local solar noon (± 2 h; 10:00 - 14:00) was 1.2°C and 1.0°C at the Valais and Ticino study sites, respectively. No difference in temperature between tree directions was indicated for the night period (22:00 - 02:00). At other times of the day, the difference in temperature between south- and north- facing leaves was higher than the night value (0°C) and lower than the maximal value (Valais: 1.6°C, Ticino: 1.3°C). The prevalent wind direction at the Valais study site was from east,

41 whereas at the Ticino study site wind was blowing predominantly either from the eastern or western direction, with a higher wind speed for the western direction (Figure 5.4).

Table 5.2. Mean fruit diameter (cm) considering canopy aspect and height at the Valais and Ticino study sites in July a.

Canopy aspect Canopy height

Year 1 Year 2 Year 1 Year 2

Valais North 3.7 ± 0.1b 4.0 ± 0.1a Bottom 3.8 ± 0.0b 3.9 ± 0.1a

East 3.8 ± 0.1b 4.0 ± 0.0a Middle 3.8 ± 0.1ab 4.0 ± 0.1a

South 3.8 ± 0.0ab 4.0 ± 0.1a Top 3.9 ± 0.0a 4.0 ± 0.1a

West 3.9 ± 0.0a 4.0 ± 0.1a

n 109 74 n 115 75

Ticino North 3.9 ± 0.1b 4.4 ± 0.1b Bottom 3.9 ± 0.1a 4.5 ± 0.1a

East 3.9 ± 0.1b 4.5 ± 0.1b Middle 3.9 ± 0.1a 4.6 ± 0.1a

South 4.1 ± 0.1a 4.8 ± 0.1a Top 4.0 ± 0.1a 4.5 ± 0.1a

West 3.9 ± 0.1b 4.6 ± 0.1b

n 115 35 n 104 38 a Means (±SEM) within the same column followed by different letters a, b indicate significant differences in fruit diameter for a study site (P < 0.05, Friedman test, Nemenyi post hoc test). (n) is the number of trees included in the analysis.

42

Figure 5.3. Modeled average leaf temperature at the Valais and Ticino study site for sunlit and shaded leaves in Year 2 from June to August.

43

Figure 5.4. Wind roses for the Valais (A) and the Ticino (B) study site based on six months of hourly wind data (1 May until 1 September in Year 2; provided by MeteoSwiss). The wind roses are divided into 36 sectors, one for each 10 degrees (n: north, e: east, s: south, w: west), and four wind speed categories (<1, 1-3, 3-6 and >6 m/s). The radius of one wedge gives the relative frequency of each of the 36 wind directions (how many times in percent the wind is blowing from that direction).

44 5.4. Discussion

Our objectives when designing this survey were to examine within-tree distribution of codling moth larval infestation, to draw conclusions on small-scale distribution of ovipositing female codling moths in the field, and to understand factors underlying these findings. This study shows that the distribution of codling moth larval infestation varied significantly among canopy aspects. The north-facing tree side was generally less infested than the south- or east- facing tree sides, and the middle canopy height was in several instances the most strongly infested vertical sector. Differences in infestation between canopy aspects were marked in the period of the first codling moth generation, while they disappeared later in the season when the main cohort of the second generation was infesting fruit. Our findings of an non-homogeneous distribution of codling moth fruit infestation within apple tree canopies suggests that female codling moths are not randomly approaching their host trees, provided that egg mortality is comparable among canopy sectors. In fact, oviposition site and infestation site appeared to be highly congruent in a previous study (Geier 1963). These findings are in line with the conclusion that ovipositing codling moths avoid the north-facing canopy and that this choice is significant in the first part of the season. Our results, based on a large dataset of approximately 40,000 apples from 160 apple genotypes in two environments, thereby statistically support the findings reported in other studies that ovipositing females prefer a particular canopy aspect at the beginning of the season but not later (McLellan 1962, Blomefield et al. 1997), and partly contradict the observation that codling moth fruit infestation is randomly distributed in the two or three generations observed (Geier 1963, Blago and Dickler 1990). Regarding canopy height, several studies suggest oviposition preference for the top height, although it is not clear whether highest egg densities at the top height might reflect differences in insecticidal treatment, fruit number or fruit sampling (Wood 1965, Jackson 1979, Blago and Dickler 1990). Furthermore, most of the studies regarding canopy height distinguished between bottom and top canopy heights only. Our study was based on three canopy heights and revealed in several instances that the middle part was most infested. The current field results that codling moth infestation was lower at the north-facing tree side compared to the south- or east-facing side in the first generation, validate recent laboratory data on avoidance of microhabitats with lower temperature for oviposition (Kührt et al. 2006b). In the laboratory, females showed a thermophilous behavior, avoiding low- temperature and choosing high-temperature zones in a circular temperature gradient arena for

45 egg deposition (Kührt et al. 2006b). In the field, we found the lowest level of infestation on the north-facing canopy aspect, which is the coolest canopy area according to modeled leaf temperature. If the temperature gradient was sufficiently large to be perceived by females searching for a suitable oviposition site, the detected variation in infestation levels in the field might be explained as a direct response to these temperature differences. Although being not retrievable from our leaf-temperature model, small temperature gradients that may trigger behavioral responses in insects (0.25°C being sufficient for the apple blossom weevil (Anthonomus pomorum L.) (Hausmann et al. 2004)) may still persist in the evening when codling moth oviposition activity starts. Besides directly affecting codling moth behavior, a temperature gradient within the canopy may also result in a gradient in fruit development, which might indirectly influence oviposition behavior. In line with our hypothesis that codling moth females avoid cool microhabitats for oviposition, our study revealed an avoidance of the north-facing tree side with lower temperature and smaller fruit size in the first codling moth generation in July. As higher temperatures will increase fruit development (Carlé et al. 1987), ovipositing codling moths appeared to select larger (riper) fruit in July when preferring the south-facing tree side. In August, when the main cohort of the second generation larvae appeared (Kührt et al. 2006a), neither larval infestation nor fruit size was different among tree aspects. Physical or chemical changes of the fruit may underlie the observed changes with increasing ripeness and with progressing season. Fruit firmness strongly decreases particularly in the first phase of fruit development (Volz et al. 2003), and volatile emissions undergo qualitative and quantitative changes with fruit ripening (Carlé et al. 1987, Vallat and Dorn 2005). Our data suggest that early in the season tree sectors differ in attractiveness or suitability for codling moth as a result of variable fruit development status and that this difference is relevant for oviposition site selection. Later in the season, when fruit phenology is more similar at different tree sides, within-canopy site selection becomes less relevant. Increased activity of female codling moth in the tree part most advanced in blossom and fruit formation was already suggested (Blomefield et al. 1997) and is now statistically confirmed by our study for a cultivar- and environment-independent data set. In addition to temperature gradients in the tree canopy, prevailing wind direction is expected to influence oviposition site selection by adult females, and thus subsequent larval infestation sites. Wind direction changed regularly between east and the opposite west direction at the Ticino study site, rendering interpretation on its influence difficult. At the Valais study site, prevailing wind direction was consistently from west. At this study site, highest larval infestation levels were found for the canopy side facing east, thus opposite to

46 the prevailing wind direction. Considering that insects respond to kairomones by upwind orientation (Sabelis et al. 1999), and female codling moths are attracted to volatile blends from the apple tree over many weeks in summer (Vallat and Dorn 2005), they are expected to land predominantly at the upwind side of the canopy, in this case at the east-facing side. Our data on larval infestation sites coincide with this expectation. We suppose that the temperature gradient between the east and the south-facing aspect was not sufficient to trigger a stronger preference of the females for the south-facing tree side, thus wind direction may modify the well-documented effect of the temperature gradient (Kührt et al. 2006b) In conclusion, knowledge on the small-scale distribution of oviposition or subsequent infestation sites is highly relevant for meaningful surveillance of the codling moth in orchards. As the current study relies on a large number of different apple genotypes, it can be considered cultivar-independent. Our results suggest that positive thermotaxis of mated codling moth females supports their avoidance of the north-facing direction throughout the season, but observed differences in mean values between infestation of different canopy aspects were significant only in the first part of the season. Our data show, for the first time, that monitoring of the first codling moth generation in July poses more challenges than monitoring of the second generation in August, as levels of larval infestation depended only in July on the canopy aspect. These results provide baseline data for trap placement for surveillance of female codling moth. To efficiently and reliably monitor the first generation, we recommend avoiding the north-facing direction when positioning female-attracting traps as well as for visual inspection. Even slight differences in host plant (fruit) parameters occurring in different canopy aspects, as well as abiotic factors like prevailing wind direction, can affect the distribution of ovipositing codling moth females. Finally, although the effect of the canopy height on the larval infestation was relatively limited, our data indicate that the middle height requires particular attention during surveillance.

47 6. QTL analysis for aphid resistance and growth traits in apple 3

The rosy apple aphid (Dysaphis plantaginea), the leaf-curling aphid (Dysaphis cf. devecta) and the green apple aphid (Aphis pomi) are widespread pest insects, reducing growth of leaves, fruits and shoots in apple (Malus x domestica). Aphid control in apple orchards is generally achieved by insecticides, but alternative management options like growing resistant cultivars, are needed for more sustainable integrated pest management (IPM). A linkage map available for a segregating F1-cross of the apple cultivars 'Fiesta' and 'Discovery' was used to investigate the genetic basis of resistance to aphids. Aphid infestation and plant-growth characteristics were repeatedly assessed for the same 160 apple genotypes in three different environments and two consecutive years. We identified 'amplified fragment polymorphism markers' (AFLP) linked to quantitative trait loci (QTLs) for resistance to D. plantaginea ('Fiesta' linkage group 17, locus 57.7, marker E33M35-0269; heritability: 28.3%), and to D. cf. devecta ('Fiesta' linkage group 7, locus 4.5, marker E32M39-0195; heritability: 50.2%). Interactions between aphid species, differences in climatic conditions and the spatial distribution of aphid infestation were identified as possible factors impeding the detection of QTLs. A pedigree analysis of 'simple sequence repeat marker' (SSR) alleles closely associated with the QTL markers revealed the presence of the alleles in other apple cultivars with reported aphid resistance ('Wagener', 'Cox's Orange Pippin'), highlighting the genetic basis and also the potential for gene pyramiding of aphid resistance in apple. Finally, significant QTLs for shoot length and stem diameter were identified, while there was no relationship between aphid resistance and plant trait QTLs.

______3 Stoeckli, S., K. Mody, C. Gessler, A. Patocchi, M. Jermini, and S. Dorn. 2008. QTL analysis for aphid resistance and growth traits in apple. Tree Genetics & Genomes 4:833-847.

48 6.1. Introduction

Apple (Malus x domestica Borkh.) is the most relevant fruit crop in the temperate region, and production quantity has even increased by 27% between 1995 and 2006 (http://faostat.fao.org). Insects from different orders have a negative impact on the quantity and quality of fruit yield and require control (Beers et al. 2003). Development of sustainable insect pest management is particularly promising in perennial crops such as apple, as measures taken to reduce insect pest populations in one growing season are likely to also affect the next growing season (Dorn et al. 1999). Although quantitative trait loci (QTLs) based approaches to host-plant resistance against insects are common in annual crops (Frei et al. 2005, Fujita et al. 2006), and intensively studied in apple to suppress diseases (Cheng et al. 1998, James et al. 2004, Calenge and Durel 2006, Khan et al. 2006), they have received little attention so far in apple in connection with aphid suppression (Roche et al. 1997, Bus et al. 2008). The rosy apple aphid (Dysaphis plantaginea Pass.) is one of the major insect pests on apple, causing stunting and malformation of leaves and fruits (Graf et al. 2006). The economic threshold level is extremely low with 1% of infested flower buds in spring, because even light infestations produce unmarketable fruits (Blommers 1994). The leaf-curling aphids (Dysaphis cf. devecta Wlk.) belong to a species complex (Stekolshchikov and Lobanov 2004). They are pests of cultivated apple species and cause red-curled leaves leading to high economic damage through malformation of leaves and fruits and inhibited shoot growth (Cevik and King 2002a). The green apple aphid (Aphis pomi De Geer), found in most areas where apple is cultivated, negatively affects the shoot and leaf of apple trees by phloem-sap sucking (Arbab et al. 2006). Population outbursts can lead to serious economic losses as aphids feeding on immature and mature fruits produce deformed apples of reduced quality (e.g. russet fruits or reduced sugar content) (Hamilton et al. 1986, Arbab et al. 2006). Besides direct injuries to leaves, shoots and fruits, aphids may act as vectors for virus transmission, while the release of honeydew fosters sooty mold infestation (Arbab et al. 2006). As a consequence, large amounts of insecticides are needed to manage aphid pests, for example, up to four applications of pirimicarb (250-750 g/ha) per season in the Swiss midlands (Maag 1996, Cevik and King 2002a). For economic and environmental reasons, integrated pest management (IPM) programs have been developed (Zehnder et al. 2007). Potential tools within IPM programs in fruit orchards include natural enemies as biological control measures and orchard management. These tools have, however, in recent examinations been found to be

49 of minor importance for the suppression of the mentioned aphid species (Miñarro et al. 2005, Omkar and Pervez 2005, Simon et al. 2006). Breeding for resistant cultivars is a further possibility. Several authors have found variable susceptibility of different apple cultivars to aphids (Alston and Briggs 1977, Lespinasse et al. 1985, Graf et al. 1998, Habekuss et al. 2000, Qubbaj et al. 2005, Angeli and Simoni 2006). Based on these observations, a genetic basis of aphid resistance in apple was assumed. So far only the Sd1 gene for D. cf. devecta resistance and the ER1, ER2, and ER3 genes for woolly apple aphid (Eriosoma lanigerum Hausm.) resistance were studied for application in apple breeding (Roche et al. 1997, Bus et al. 2008). The detection of resistance-breaking biotypes of D. plantaginea (Rat-Morris et al. 1999) emphasizes the need for a better understanding of the genetic background of aphid resistance, and application of new molecular tools such as QTL analysis could help. The purpose of this study was to investigate resistance in apple to the aphid species, D. plantaginea, D. cf. devecta and A. pomi, using linkage map data available for a segregating

F1-cross of the apple cultivars 'Fiesta' and 'Discovery' (Liebhard et al. 2003b, Silfverberg- Dilworth et al. 2006). Besides detection of potential QTLs, knowledge on the D. cf. devecta resistance gene (Sd1) (Alston and Briggs 1977, Roche et al. 1997, Cevik and King 2002a) should be improved by describing more markers; especially simple sequence repeat markers (SSR) linked to this resistance gene. These could then be used for gene pyramiding in plant breeding. Since plant-growth characteristics could influence aphid infestations and erroneously be interpreted as QTLs for physiological resistance, the relationship of plant traits and aphid infestation was assessed by QTL analyses of plant-growth characteristics. To account for environmental factors causing instable QTL effects at different study sites, interactions between the aphid species studied, the climatic conditions at the study sites and the spatial variability of aphid infestations were examined.

6.2. Materials and methods

Orchard location and plant material The resistance of apple trees to the three aphid species was studied in the field at three different study sites in Switzerland, situated in Zurich (Wadenswil; at 47°13'20''N, 8°40'05''E, 455 m altitude), Valais (Conthey; at 46°12'30''N, 7°18'15''E, 478 m altitude), and Ticino (Cadenazzo; at 46°09'35''N, 8°56'00''E, 203 m altitude) in 2005 and 2006 (= Year 1 and Year 2). The climate from March to August was characterized for each site by monthly average

50 temperature and sum of rainfall obtained from MeteoSwiss (http://www.meteoschweiz.ch) (supplementary material; Table SM 6.5.1). Compared to the standard value (30-year average: 1960-1990), temperature in the study years was 1-2 °C higher, and at the Ticino site the measured sum of rainfall was 60-70% of the standard values (Table SM 6.5.1). Orchards were treated with fertilizers and herbicides, but no insecticides and fungicides were applied. The surveyed apple genotypes were previously described (Liebhard et al. 2003b).

They are a segregating F1-cross of the apple varieties 'Fiesta' (syn. 'Red Pippin') and 'Discovery'. 'Fiesta' was named as a sort 1985 in England (HRI East Malling), as a cross of 'Cox's Orange Pippin' and 'Idared', and only resistance properties to one aphid species (Dysaphis devecta) were yet reported (Roche et al. 1997). 'Discovery' is a cross of 'Worcester Pearmain' as mother sort and probably 'Beauty of Bath' as father sort. It was discovered around 1949 in England and is attributed a high, polygenic resistance to apple scab (Liebhard et al. 2003c). The apple trees were tripled in summer 1998 by bud grafting on M27 rootstocks and planted in winter 1998/1999 at the three sites. The apple trees were planted in rows 3.50 m apart and with a tree-to-tree distance of 1.25 m at the Ticino site and of 0.50 m at the Valais and Zurich site. All tree genotypes that were present at all three sites were surveyed. Sample size occasionally differed from the maximum of 160 apple genotypes, as some trees died since plantation establishment, and as for specific analyses not all genotypes were considered (e.g. neighborhood effect).

Aphid survey Aphid infestation was assessed on all shoots of each apple genotype. To quantify Dysaphis plantaginea infestation, the number of aphid colonies per tree was counted two times per year in both study years (sampling date: May and June). The number of red-curled leaves per tree was used as a measure of Dysaphis cf. devecta infestation (May, June and July). The number of green apple aphids was counted three times (May, June and July) for each shoot to quantify infestation with Aphis pomi. In the second study year, a fourth A. pomi survey was carried out in August at the Valais site. In case of large A. pomi colonies, aphid number was estimated by reference counts. Reference counts were obtained by assessing the number of A. pomi aphids on a part of a shoot to obtain a reference for a shoot section hosting 50 aphids. Extrapolation of this information was then used to quantify the size of large A. pomi colonies.

51 Plant-growth characteristics Stem diameter was measured 1.2 m above soil. For quantification of tree crown volume, the crown diameter at the widest part, the diameter perpendicular to the latter and the crown height were measured (crown volume = 4π/3 abc; a,b,c= the three sides of an ellipsoid). Tree-height was determined as another plant trait, from the soil surface to the top (highest point) of the tree. The term 'new-shoots' was used for the current year's actively growing shoots. The total length of shoots (including new-shoots), and the length of new- shoots separately, were measured and the sum and number were calculated. Shoot length was assessed in May, new-shoot length end of June. Shoots and new-shoots that were very short (shoot length < 20 cm; new-shoot length < 5 cm) were not considered.

QTL mapping Quantitative trait loci (QTLs) analyses of mean aphid infestation per year and location (Krakowsky et al. 2004) were carried out with MapQTL® 4.0 (van Ooijen et al. 2002). The same software was used for QTL analysis of plant characteristics data (per year and location, one replication). The genetic linkage maps for both 'Fiesta' and 'Discovery' (single parent maps), used in QTL analysis, were calculated with 251 apple genotypes and were already published (Liebhard et al. 2003b). The maps consist of a total of 345 ('Fiesta') and 389 ('Discovery') markers and include 137 AFLP, 108 microsatellites (SSR), and 100 RAPD markers in 'Fiesta', and 160 AFLP, 103 SSR, one SCAR and 125 RAPD markers in 'Discovery'. The average linkage group length was 66.96 cM for 'Fiesta' and 84.36 cM for 'Discovery'. Kruskal-Wallis tests and interval mapping (IM) were used for QTL analysis. Logarithm of odds (LOD) threshold values were determined by 1,000-fold permutation tests (MapQTL® 4.0) at a significance level of 95% (genome-wide) (King et al. 2000). The proportion of variation in aphid infestation that can be explained by the genetic variation among the apple progenies was analyzed by broad-sense heritability, which was estimated by 2 2 2 2 2 2 2 2 the formula H = σ g/σ p and σ p = (σ g + σ e/n), where σ g is the genetic variance, σ p is the 2 phenotypic variance, σ e is the environmental variance and n is the number of replicates per genotype (Lauter and Doebley 2002). Variance components were based on mean square ANOVA results (Frei et al. 2005). To test for interactions between different QTLs, multiple QTL mapping (MQM) was carried out for QTLs with LOD scores exceeding the significant LOD threshold in IM. For MQM, markers with the highest likelihood ratios (i.e. LOD test statistic) positioned on the linkage group containing the QTL were selected as the initial set of possible MQM cofactors. A backward elimination procedure was applied to this initial set of

52 cofactors using a significance level of 0.02 (Larson et al. 2006). The 2-LOD support interval was calculated to estimate the position of significant QTLs with 95% confidence (King et al. 2000). The 'Fiesta' x 'Discovery' population was divided into subpopulations based upon the presence/absence of the marker closest to the detected QTL and phenotypic difference was tested considering the two subpopulations (Mann-Whitney U-test).

Pedigree analysis PCR amplifications with the primers of the SSR markers Hi03a10 and Hi07h02 (Silfverberg-Dilworth et al. 2006) were performed in a 10 µl volume containing 5 µl of a DNA solution (1 ng/µl), 1x reaction buffer (Amersham Pharmacia, Dübendorf, Switzerland), 0.1 mM of each deoxyribonucleotide triphosphate (dNTP), 0.2 µM of dye-labelled forward primer and 0.2 µM of reverse primer, and 0.7 U of Taq Polymerase (Amersham Pharmacia, Dübendorf, Switzerland) per reaction. Forward primers were labelled with the following dyes:

Hi03a10f with NED (Applied Biosystems) and Hi07h02f with HEX (Biomers, Germany). PCRs were performed in a Gene Amp PCR system 9600 (Perkin Elmer, Foster City, CA) under the following conditions: 5 min at 96oC, 35 cycles of 30 s at 96oC, 30 s at 60oC and 30 s at 72oC with a final extension of 10 min at 72oC. For each sample 1 µl of PCR product were mixed with 10 µl deionized formamide and 0.2µl 500-LIZ ladder (Applied Biosystems). Microsatellite fragments were separated by electrophoresis on an ABI PRISM 3700 capillary sequencer (Applied Biosystems). Chromatographs were generated using Genescan 3.7 software and microsatellite fragment lengths were scored with Genotyper 3.6 (Applied Biosystems).

Spatial distribution of aphids The distribution patterns of aphids on individual apple genotypes were characterized 2 computing the index of dispersion (ID) (Southwood and Henderson 2000), with ID = s (ν- 1)/m, where m and s2 are the sample mean and variance, respectively, and ν is the sample 2 size. ID values significantly greater than the χ statistic (0.025 probability level) with (ν-1) degrees of freedom indicate an aggregated distribution of the studied species (significant variation considering the level of infestation of individual trees) (Ludwig and Reynolds

1988). The calculation of the ID values was performed with BiodiversityPro 2.0 (McAleece, Lambshead and Paterson; The Natural History Museum, London). Potential effects of the spatial position of trees in the study sites on aphid infestation were inferred from analyses of spatial autocorrelation, computing Moran's I (Legendre and Legendre 1998) and

53 corresponding z values (significance levels) using the software CrimeStat III (Levine 2007) . Moran's I varies between -1.0 for negative spatial autocorrelation (nearby trees have dissimilar aphid infestation level) and +1 for positive spatial autocorrelation (nearby trees have similar aphid infestation level). If no spatial autocorrelation exists, the expected value for Moran's I is eI = -1/(n-1). Values of I greater than the expected I indicate clustering while values of I less than the expected I indicate dispersion. The significance test of the z-values indicates whether these differences are greater than what would be expected by chance. The z values were compared to a standard normal table, and absolute values greater than 1.96 indicate a spatial autocorrelation at a 5% significance level. Additionally, we described the spatial distribution of aphids by fitting trend surfaces on contour plots by kriging (best unbiased generalized least squares estimation) with an exponential covariance function (Venables and Ripley 2002). The distribution of aphids on a small scale level was analyzed by comparing aphid infestation on neighbor trees. The number of aphids on an individual tree was compared to the number of aphids on the neighbor trees (sum) using Spearman's rank tests to estimate neighborhood effects. Only those genotypes were included in this analysis that had direct neighbor genotypes (a dead or missing genotype was not regarded as neighbor genotype).

Statistical analysis

A log10(x+1) transformation was applied to normalize error distribution of aphid infestation. To evaluate the effect of genotype, site, and year, a three factor mixed model ANOVA, with year as within-subject effect (repeated measure), and genotype and site as between-subjects fixed effects, was applied. Spearman's rank tests were used to study relationships between different sites, years, sampling dates, aphid species and plant characteristics. When multiple correlation tests were carried out, the Benjamini-Hochberg procedure was used to correct for false discovery rates (type I errors) (Benjamini and Hochberg 1995, Verhoeven et al. 2005). All statistical analyses were performed with SPSS 16.0 for Mac OS X (SPSS, Inc., Chicago, IL) and R 2.6.0 (R Development Core Team, Vienna).

54 6.3. Results

Aphid infestation Aphid infestation of individual apple trees varied strongly among sites and study years (Table 6.1). The highest number of Dysaphis plantaginea colonies per tree was found at the Ticino site in both study years (mean: 13.5 and 17.8), and the lowest number at the Zurich site (mean: 0.9 and 0.0). Incidence ranged between 0% (Zurich Year 2) and 92% (Valais Year 2) infested trees. The mean number of red-curled leaves per tree caused by Dysaphis cf. devecta varied between 1.6 and 2.8 at the Ticino site and between 0.8 and 4.0 at the Valais site. Incidence ranged from 0% (Zurich Years 1 and 2) to 18% (Valais Year 1) and 59% (Valais Year 2) infested trees. Infestation by Aphis pomi was highest for the Valais site in both study years (average: 443.5 and 63.4 aphids per tree) followed by the Ticino (mean: 62.9 and 45.6) and the Zurich site (average: 5.8 and 10.2). Incidence ranged from 26% (Zurich Year 1) to 87% (Valais Year 1) infested trees.

Table 6.1. Infestation of progeny plants of the cross 'Fiesta' x 'Discovery' by rosy apple aphid (Dysaphis plantaginea), leaf curling aphid (Dysaphis cf. devecta), and green apple aphid (Aphis pomi) at three study sites in two consecutive years. Number of sampled trees (n), mean infestation per season ± standard error (SE), maximum value (Max) and incidence (I) (% of infested trees) are presented.

Ticino Valais Zurich n Mean ± SE I n Mean ± SE I n Mean ± SE I (Max) (Max) (Max)

D. plantaginea 13.5 ± 1.4 0.9 ±0.3 0.9 ± 0.4 Year 1 144 (124) 87 143 (53) 13 153 (91) 11 17.8 ± 1.6 9.8 ± 0.9 0.0 ± 0.0 Year 2 144 75 143 92 153 0 (196) (87) (0) D. cf. devecta 2.8 ± 0.7 0.8 ± 0.1 0.0 ± 0.0 143 22 142 18 153 0 Year 1 (130) (18) (0) 1.6 ± 0.4 4.0 ± 0.5 0.0 ± 0.0 Year 2 143 19 142 59 153 0 (75) (68) (0) A. pomi 62.9 ± 9.8 443.5 ± 61.1 5.8 ± 1.9 143 69 142 87 153 26 Year 1 (1,570) (12,020) (405) 45.6 ± 6.8 63.4 ± 7.5 10.2 ± 1.3 Year 2 143 66 142 85 153 55 (1,150) (2,100) (134)

55 Influence of genotype, site and year on aphid abundance A significant influence of genotype was found for infestation by D. plantaginea and D. cf. devecta but not by A. pomi. Site and year contributed significantly to infestation patterns of all three aphid species (Table 6.2). Broad-sense heritability (H2) was highest for D. cf. devecta resistance (50.2%) and lowest for A. pomi infestation (15.7%) (Table 6.3). For resistance to D. plantaginea, H2 was intermediate (28.3%).

Table 6.2. Effects of genotype, site and sampling date on aphid infestation (Dysaphis plantaginea, D. cf. devecta, Aphis pomi) assessed by a mixed model ANOVA, with year as within-subject effect, and genotype and site as between-subjects fixed effects.

d.f. Mean square F value P value

D. plantaginea Within-subjects effects Year 1 21.338 113.946 <0.0001 Year x genotype 155 0.178 0.949 0.623 Year x site 1 20.917 111.702 <0.0001 Between-subjects effects Genotype 155 0.322 1.394 0.026 Site 1 24.747 107.185 <0.0001 D. cf. devecta Within-subjects effects Year 1 2.263 20.194 <0.0001 Year x genotype 155 0.104 0.929 0.671 Year x site 1 5.017 44.765 <0.0001 Between-subjects effects Genotype 155 0.215 1.504 0.009 Site 1 1.504 10.786 0.001 A. pomi Within-subjects effects Year 1 3.859 9.509 0.002 Year x genotype 157 0.454 1.118 0.210 Year x site 2 24.047 59.264 <0.0001 Between-subjects effects Genotype 157 0.872 1.187 0.109 Site 2 108.249 147.353 <0.0001

56 Table 6.3. Broad-sense heritability (H2) for aphid infestation (Dysaphis plantaginea, D. cf. devecta, Aphis pomi). Calculation 2 2 of genetic variance (σ g) and phenotypic variance (σ p) based on mean square ANOVA results (cf. Table 6.2).

2 2 2 Variance components σ g σ p H (%)

D. plantaginea 0.030 0.107 28.3 D. cf. devecta 0.072 0.144 50.2 A. pomi 0.046 0.291 15.7

No correlation between the three sites was found for D. plantaginea infestation per tree in May and June (P > 0.05, Spearman's rank test, Benjamini-Hochberg procedure). D. cf. devecta infestation per tree was correlated between the Ticino and Valais site in June of Year

1 (n = 130, rs = 0.185, P = 0.035), but it was not correlated between the Ticino and Valais site in May of both years and June of Year 2 (P > 0.05, Spearman's rank test, Benjamini- Hochberg procedure). The number of A. pomi per tree correlated significantly between the Ticino and Valais site in July (Year 1) (Spearman's rank test, Benjamini-Hochberg procedure; n = 129, rs = 0.198, P = 0.025), but no correlation for A. pomi infestation was found in May and June of both study years, and between the Ticino and Valais site, between the Ticino and Zurich site, and between the Valais and Zurich site in July of Year 2. Aphid infestation of the same tree individuals was significantly correlated between the two study years for D. plantaginea at the Ticino site (Spearman's rank test; May: n = 144, rs =

0.19, P = 0.026; June: n = 144, rs = 0.17, P = 0.046), but not at the Valais site (P > 0.05). For D. cf. devecta, infestation of the same tree individuals was correlated between the two consecutive years at the Valais site (May: n = 142, rs = 0.339, P < 0.001, June: n = 142, rs = 0.368, P < 0.0001), but not at the Ticino site. For A. pomi, infestation of the same tree individuals was correlated between the two study years at the Valais site (Spearman's rank test, May: n = 142, rs = 0.184, P = 0.028; June: n = 142, rs = 0.286, P = 0.001, July: n = 142, rs = 0.237, P = 0.005) and the Ticino site in June (n = 143, rs = 0.30, P < 0.0001), but not at the Zurich site (P > 0.05). The within-year infestation of the same tree individuals was significantly correlated in both years for all the three aphid species (P < 0.05, Spearman rank correlation and Benjamini- Hochberg procedure; analysis not shown). Comparing May and June, D. plantaginea

57 abundance was generally higher in June than in May. The same pattern was found for A. pomi. Additionally, the potential to rapidly increase population size was detected for A. pomi at the Valais site. The number of red-curled leaves caused by D. cf. devecta was higher in May compared to June.

QTLs for aphid resistance A significant QTL for D. plantaginea resistance in apple was identified at the Ticino site in Year 1. Multiple QTL mapping (MQM; data not shown) did not reveal any multiple linked QTLs and results were therefore based on interval mapping (IM). The QTL was associated with the AFLP marker E33M35-0269 at 57.7 cM on linkage group 17 of 'Fiesta'. The LOD score at the marker position was 2.85 and explained 8.5% of phenotypic variability (PVE, Table 6.4, Figure 6.1a). The 2-LOD support interval ranged from map positions 45-67 cM (Ticino Year 1). A significant lower D. plantaginea infestation was found for the apple subpopulation amplifying the marker E33M35-0269 compared to apple genotypes not amplifying the marker (Figure 6.2). The origin of the QTL associated to aphid resistance was followed in the pedigree of 'Fiesta'. The allele '255 bp' of the SSR marker Hi07h02 (Silfverberg et al. 2006), which is closely located and in coupling with the AFLP marker E33M35-0269 (10.2 cM distance between Hi07h02 and E33M35-0269), has been inherited from 'Wagener' to 'Idared' (Figure 6.3a). We detected a significant QTL for D. cf. devecta resistance on 'Fiesta' linkage group 7 at the Ticino site in both years, and at the Valais site in Year 2 (Table 6.4, Figure 6.1b). QTL results based on IM as well as MQM mapping (data not shown) did not identify any multiple linked QTLs. The significant LOD scores for the AFLP marker E32M39-195 at different sites and in different years ranged from 3.36 (Ticino, Year 2) to 6.82 (Valais, Year 2), and PVE by this QTL varied between 10.3% (Ticino, Year 1) and 20.4% (Valais, Year 2). The 2-LOD support interval ranged from map position 0-15 cM (Valais Year 2). Apple genotypes containing the marker E32M39-0195 showed significantly lower D. cf. devecta infestation compared to the subpopulation missing the marker (Figure 6.2). The origin of the QTL associated to D. cf. devecta resistance was followed in the pedigree of 'Fiesta'. The allele '216 bp' of the SSR marker Hi03a10, which is in close proximity and in coupling with the AFLP marker E32M39-0195 (9.9 cM distance between Hi03a10 and E32M39-0195) (Silfverberg- Dilworth et al. 2006), was found to derive from 'Cox's Orange Pippin', which is a progeny of 'Blenheim Orange' (Figure 6.3b). No QTL was identified for A. pomi resistance in apple.

58 QTLs for different plant-growth characteristics Significant QTLs were identified at the Ticino site in Year 1 (shoot length and stem diameter) and at the Valais site in Year 2 (stem diameter) (Table 6.5). MQM mapping (data not shown) did not identify any multiple linked QTLs and results were therefore based on IM. The QTL for shoot length was significant at the Ticino site in Year 1 and was positioned on linkage group 7 of 'Fiesta' (SSR marker CH04e05 on locus 26.7). The allele of CH04e05 conferring increased length is '199 bp' long. The highest LOD score for the shoot length QTL obtained at the Ticino site in Year 1 was 2.8, while the PVE by this QTL was 8.8%. The proportion of variation in shoot length that can be explained by the genetic variation (broad- sense heritability) was 44.9% (data not shown). Two markers, linked to QTLs for stem diameter, were identified on linkage groups 1 and 13 of 'Discovery' (AFLP marker E31M38-0080 on locus 15.2, and AFLP marker E35M41-0575 on locus 29.7) at the Ticino site in Year 1. LOD scores were 2.58 and 1.62, respectively. The markers explained 7.1% and 4.5% of the phenotypic variability. Another marker linked to a QTL for larger stem diameter was identified on linkage group 14 of the 'Discovery' chromosome at the Valais site in Year 2 (RAPD marker D01-1200 on locus 10.3). The LOD score and PVE were 2.1 and 6.2%, respectively. Broad-sense heritability for stem diameter was 15.8% (data not shown).

59 Table 6.4. QTLs detected for resistance to Dysaphis plantaginea and D. cf. devecta at the Ticino, Valais and Zurich sites in two consecutive years in the segregating population of 'Fiesta' x 'Discovery' based on interval mapping (IM) results. Linkage group ('Fiesta', parental map), genetic locus (locus in cM), closest marker, linkage phase, LOD score and LOD threshold level (genome-wide) at the locus of the closest marker, and phenotypic variance explained by the QTL (PVE %) are presented. Significant LOD scores are highlighted.

a b c Linkage group Locus (cM) Closest marker LOD score LOD threshold PVE (%)

Dysaphis plantaginea 17 57.7 E33M35-0269 (+) Ticino Year 1 2.85 1.7 8.5 Year 2 0.29 1.8 0.9 Valais Year 1 0.09 1.5 0.3 Year 2 0.05 1.5 0.2 Zurich Year 1 0.20 1.4 0.6

Dysaphis cf. devecta 7 4.5 E32M39-0195 (-) Ticino Year 1 4.65 3.3 14.0 Year 2 3.36 2.7 10.3 Valais Year 1 2.31 3.3 8.5 Year 2 6.82 3.1 20.4 a Molecular marker closest to the likelihood peak of each QTL. Linkage phase information is provided as (+) or (-), indicating on which of the homologous chromosomes the marker is located. b LOD (logarithm of odds ratio) score and LOD threshold level at the position of the closest marker calculated with MapQTL®4.0. LOD threshold levels were calculated with 1,000-fold permutation tests (genome-wide). c Phenotypic variance explained by the QTL.

60 Table 6.5. QTLs identified for different plant-growth characteristics in the segregating population of 'Fiesta' x 'Discovery', based on IM results. Linkage group, genetic locus (locus in cM), closest marker, linkage phase, LOD score and LOD threshold level (genome-wide) at the locus of the closest marker, phenotypic variance explained by the QTL (PVE %), and parental map are presented.

Linkage group Locus Closest marker a LOD score b LOD threshold PVE (%) c Map d

Shoot length 7 26.7 CH04e05-199/227 (199) Ticino Year 1 2.80 2.3 8.8 F Stem diameter 1 15.2 E31M38-0080 (+) Ticino Year 1 2.58 1.5 7.1 D 13 29.7 E35M41-0575 (+) Ticino Year 1 1.62 1.5 4.5 D 14 10.3 D01-1200 (+) Valais Year 2 2.12 2.1 6.2 D a Molecular marker closest to the likelihood peak of each QTL. Linkage phase information is provided as (+) or (-), indicating on which of the homologous chromosomes the marker is located. For the SSR marker CH04e05-199/227 the favorable allele is given in parenthesis. b LOD score (logarithm of odds ratio) and LOD threshold level at the position of the closest marker calculated with MapQTL®4.0. LOD threshold levels were calculated with 1,000-fold permutation tests (genome-wide). c Phenotypic variance explained by QTL. d Parental map on which each QTL was detected: D ('Discovery') and F ('Fiesta').

61

Figure 6.1. QTLs for resistance to Dysaphis plantaginea (a), identified on linkage group 17 of 'Fiesta', and Dysaphis cf. devecta (b), identified on linkage group 7 of 'Fiesta', based on IM results. The x-axis indicates the linkage map of 'Fiesta' in cM and marker names; the y-axis shows the LOD scores. The solid black bar indicates the 2-LOD support interval for the position of the QTL (Ticino Year 1).

Average values of log10(x+1) transformed data were used for QTL analysis. LOD threshold values at different study sites and in different years ranged between 2.7 and 3.3. The closest marker to the QTL is underlined.

62

Figure 6.2. Mean Dysaphis plantaginea and D. cf. devecta infestation of the two subpopulations of the 'Fiesta' and 'Discovery' cross, based on presence and absence of the nearest marker associated to a QTL on the 'Fiesta' chromosome. The subpopulations for AFLP markers E33M35-0269 (D. plantaginea) and E32M39-0195 (D. cf. devecta) are presented. Significant differences between subpopulations are indicated by *P < 0.001 and **P < 0.0001 (Mann-Whitney U-test).

Figure 6.3. Analysis of the pedigree of the apple variety 'Fiesta' with the two SSRs Hi07h02 and Hi03a10 associated with the QTLs for resistance to Dysaphis plantaginea (a) and to D. cf. devecta (b). The SSR marker alleles associated with resistance are highlighted (allele '255 bp' for Hi07h02 and allele '216 bp' for Hi03a10).

63 Spatial distribution of three aphid species The distribution of the aphid species on individual apple trees was highly aggregated, as was indicated by the index of dispersion (ID) (Table SM 6.5.2). This means that some trees were strongly infested, whereas other trees were not infested at all. This finding was consistent for all sites and years. ID values ranged from 946 (D. cf. devecta) to 137,360 (A. pomi). ID values for D. plantaginea varied between 1,093 and 5,600. A higher aggregation of D. plantaginea and D. cf. devecta was found at the Ticino site compared to the Valais site, whereas for A. pomi a higher aggregation was found at the Valais site compared to the Ticino site. Spatial distribution patterns on the orchard-scale varied among aphid species (Table 6.6, Figure SM 6.5.1). For D. plantaginea distribution, a significant positive spatial autocorrelation was found at the Ticino site in both study years and at the Zurich site in Year 1. Infestation levels appeared to be higher at the southeastern part at the Ticino site and lower at the middle row at the Zurich site (contour plots, Figure SM 6.5.1). No spatial autocorrelation was found for D. cf. devecta distribution (Table 6.6). This is in line with the findings from the contour plots (Figure SM 6.5.1). The distribution of A. pomi within an orchard seemed to be most strongly autocorrelated. There was a significant positive autocorrelation at the Ticino site in Year 1, at the Valais site in Year 2, and at the Zurich site in both years (Table 6.6). At the Ticino site, rows at the southern part showed a higher infestation compared to northern rows, and the border rows were more infested compared to the middle row at the Zurich site (Figure SM 6.5.1). Possible effects of neighborhood on aphid infestation of specific apple trees differed among sites and aphid species (Table 6.7). D. plantaginea infestation of a specific tree and of the two neighbor trees was significantly correlated at the Zurich site in Year 1 and the Valais site in Year 2 (no D. plantaginea infestation was detected at the Zurich site in Year 2). For the same aphid species no neighborhood effect was found at the Ticino site (P > 0.05; Spearman's rank test). For D. cf. devecta infestation, no significant correlation between aphid infestation of a specific tree and of the two neighbor trees was found (Table 6.7). For A. pomi infestation, possible neighborhood effects were found at the Ticino site in Year 1, at the Valais site in both study years and the Zurich site in Year 2 (Table 6.7).

64 Table 6.6. Spatial autocorrelation analysis of Dysaphis plantaginea, Dysaphis cf. devecta, and Aphis pomi distribution at the Ticino, Valais and Zurich sites in two consecutive years (mean values of different surveys within a year). Values of I greater than a randomly expected I indicate clustering while smaller values of I indicate dispersion. The z values were compared to a standard normal table, and absolute values greater than 1.96 indicate a spatial autocorrelation at a 5% significance level (highlighted). Sample size was n = 143 for the Ticino, n = 142 for the Valais, and n = 153 for the Zurich site. At the Zurich site there was no infestation of D. plantaginea in Year 2 and of Dysaphis cf. devecta in both years.

Moran's I I randomly Normality Randomization expected (SD) significance (z) significance (z)

D. plantaginea Ticino Year 1 0.064 -0.007 (0.013) 5.607 5.621 Year 2 0.019 -0.007 (0.013) 2.055 2.071 Valais Year 1 0.001 -0.007 (0.023) 0.331 0.441 Year 2 0.023 -0.007 (0.024) 1.275 1.284 Zurich Year 1 0.042 -0.007 (0.019) 2.576 4.215 Year 2 - - - - D. cf. devecta Ticino Year 1 0.002 -0.007 (0.013) 0.703 0.750 Year 2 0.003 -0.007 (0.013) 0.754 0.842 Valais Year 1 -0.005 -0.007 (0.023) 0.086 0.093 Year 2 -0.006 -0.007 (0.024) 0.042 0.043 A. pomi Ticino Year 1 0.013 -0.007 (0.013) 1.522 1.564 Year 2 0.034 -0.007 (0.013) 3.592 5.048 Valais Year 1 0.096 -0.007 (0.024) 4.373 4.516 Year 2 0.034 -0.007 (0.024) 1.759 1.812 Zurich Year 1 0.032 -0.007 (0.019) 2.031 2.390 Year 2 0.143 -0.007 (0.019) 7.921 8.486

65 Table 6.7. Effects of neighborhood on aphid infestation (Dysaphis plantaginea, D. cf. devecta, Aphis pomi) of individual apple trees assessed by Spearman's rank tests comparing aphid infestation of a specific tree and the infestation of the two neighbor trees in Year 1 and Year 2. For each aphid species, the mean of the different infestation surveys (sampling date) at a site within a year was used for correlation analyses. Significant correlations are highlighted.

Ticino Valais Zurich

n rs P n rs P n rs P

D. plantaginea Year 1 53 0.249 0.072 54 0.189 0.171 64 0.414 0.001 Year 2 53 0.064 0.651 54 0.346 0.010 - - -

D. cf. devecta Year 1 52 0.270 0.053 54 0.004 0.978 - - - Year 2 52 -0.126 0.374 54 0.246 0.073 - - -

A. pomi Year 1 52 0.274 0.049 54 0.514 <0.0001 64 0.242 0.054 Year 2 52 0.188 0.182 54 0.293 0.031 64 0.423 0.001

Relationship between different aphid species D. plantaginea infestation per tree was positively correlated with D. cf. devecta and A. pomi infestation per tree at the Ticino site in Year 1 (Spearman's rank test, Benjamini-

Hochberg procedure; n = 143, rs = 0.222, P = 0.008; n = 143, rs = 0.445, P < 0.0001) but not in Year 2 (P > 0.05, Table S3). There was no significant relationship between D. cf. devecta and A. pomi (P > 0.05, Table SM 6.5.3). At the Valais and Zurich site no significant relationship between the three aphid species was found (P > 0.05, Table SM 6.5.3). In general, the relationship between the aphid species was positive, however the correlation was weak and not significant (Table SM 6.5.3).

Relationship between different plant-growth characteristics and aphid infestation Aphid infestation was positively correlated to different plant-growth characteristics (Table 6.8). The influence of plant-growth traits on aphid infestation differed among aphid species. Plant-growth traits were most important for A. pomi and least important for D. cf. devecta. A significant relationship between D. plantaginea infestation per tree and plant- growth characteristics was found for the Ticino site (Spearman's rank test, Benjamini-

66 Hochberg procedure; Table 6.8). Stem diameter, shoot length and new-shoot length were significantly correlated to the number of D. plantaginea colonies. At the Valais site in Year 2, shoot length and new-shoot length were significantly related to D. plantaginea infestation. At the Zurich site no significant relationship was found. The number of red-curled leaves per tree induced by D. cf. devecta was significantly related to tree-height at the Ticino site in Year 2 and at the Valais site in Year 2, to shoot length at the Ticino site in Year 1 and at the Valais site in Year 2 (Table 6.8). There was no significant relationship between D. cf. devecta infestation and stem diameter and new-shoot length, respectively. The length of new-shoots had a significant influence on A. pomi infestation per tree (Table 6.8). A stronger correlation was detected between A. pomi infestation and new-shoot length compared to shoot length (Ticino, Valais Year 1, and Zurich). At the Ticino site the relation between the number of A. pomi per tree and shoot length was not significant, whereas a significant correlation between A. pomi infestation per tree and the length of new-shoots was found. The number of A. pomi aphids was also significantly correlated to stem diameter at the Ticino site in Year 2, at the Valais site and at the Zurich site, to tree-height at the Valais site in Year 1 and at the Zurich site in Year 1 (Table 6.8).

67 Table 6.8. Relationship between infestation of individual apple trees by Dysaphis plantaginea, D. cf. devecta, and Aphis pomi and the tree-growth characteristics stem-diameter, tree-height, crown- volume, shoot-length and new-shoot-length assessed by Spearman's rank tests. The mean of different surveys at a site within a year was used for correlation analysis. Sampled tree number was 144 for the Ticino site, 143 for the Valais site and 153 for the Zurich site, respectively. Significant correlations after FDR correction are highlighted.

Stem Tree height Shoot length a New-shoot diameter length a rs P rs P rs P rs P

Dysaphis plantaginea Ticino Year 1 0.197 0.018 0.060 0.476 0.313 <0.0001 0.328 <0.0001 Year 2 0.236 0.004 0.083 0.322 0.355 <0.0001 0.237 0.004 Valais Year 1 0.078 0.355 0.040 0.635 0.086 0.308 -0.027 0.748 Year 2 0.150 0.076 0.152 0.070 0.287 0.006 0.220 0.008 Zurich Year 1 0.008 0.922 0.069 0.397 -0.014 0.864 -0.029 0.719 Dysaphis cf. devecta Ticino Year 1 0.167 0.046 -0.028 0.738 0.211 0.011 0.068 0.419 Year 2 0.164 0.051 0.228 0.006 0.137 0.136 0.050 0.557 Valais Year 1 0.084 0.318 0.138 0.100 0.105 0.214 0.114 0.175 Year 2 0.167 0.047 0.185 0.028 0.414 <0.0001 0.111 0.188 Aphis pomi Ticino Year 1 0.147 0.080 -0.006 0.948 0.167 0.046 0.418 <0.0001 Year 2 0.231 0.005 -0.090 0.286 0.180 0.033 0.580 <0.0001 Valais Year 1 0.249 0.003 0.377 <0.0001 0.199 0.017 0.586 <0.0001 Year 2 0.287 0.001 0.091 0.280 0.273 0.009 0.252 0.003 Zurich Year 1 0.260 0.001 0.230 0.004 0.301 <0.0001 0.386 <0.0001 Year 2 0.215 0.008 -0.084 0.304 0.265 0.001 0.324 <0.0001 a Shoot length and new-shoot length was highly correlated to the number of shoots and new-shoots

(Spearman's rank test, Benjamini-Hochberg procedure; n = 158-160, rs = 0.907-0.943, P < 0.0001). Shoot

length was also highly correlated to crown volume (n = 158-160, rs = 0.687-0.797, P < 0.0001). Data for the number of shoots and new-shoots and crown volume are not shown as similar results were received for shoot length and new-shoot length, respectively.

68 6.4. Discussion

Genetic background of aphid resistance in apple The present study aimed at investigating the resistance of apple trees to the aphid species Dysaphis plantaginea, D. cf. devecta and Aphis pomi by QTL analysis. The relationships between aphid infestation and plant-growth characteristics, and interactions between the studied aphid species, the climatic conditions at the study sites and the spatial variability of aphid infestation were also assessed. We identified a putative QTL for resistance to D. plantaginea and a QTL for resistance to D. cf. devecta in apple. The markers closely linked to the QTLs were related to significantly lower D. plantaginea infestation at the Ticino site in Year 1 (AFLP marker E33M35-0269), and to significantly lower injuries caused by D. cf. devecta at the Ticino site in Year 1 and 2 and at the Valais site in Year 2 (AFLP marker E32M39-0195). Broad-sense heritability (H2) was higher for D. cf. devecta (50.2%) compared to D. plantaginea (28.3%). We found a strong variation in aphid numbers on individual apple trees. While some trees were highly infested, others were not infested at all. Although in some instances aphid infestation patterns appeared to be spatially autocorrelated, no distinct clusters of highly infested trees were detectable. ANOVA analyses indicated that the genotype was a significant factor for this variation in infestation by D. plantaginea and D. cf. devecta but not by A. pomi. The identified QTLs and the findings on variable infestation of individual apple genotypes highlight the genetic background of aphid resistance in apple. The existence of some genetically based resistance to aphids in apple is also indicated by studies reporting that susceptibility to D. plantaginea, D. cf. devecta, and A. pomi may differ among apple varieties (Alston and Briggs 1970, 1977, Roche et al. 1997, Graf et al. 1998, Cevik and King 2002a, Qubbaj et al. 2005, Angeli and Simoni 2006) (see Table SM 6.5.4 for detailed list). Despite this considerable coverage of aphid resistance in apple, information on the genetic background of this resistance is scarce. MAL 59/9, a Malus robusta derivative, was shown to carry a single dominant gene for hypersensitivity to D. plantaginea (Alston and Briggs 1970), but until today the gene has not been located. Qubbaj et al. (2005) investigated gene expression in a resistant ('Florina') and a susceptible ('Topaz') apple cultivar after infestation with D. plantaginea. The authors describe gene fragments expressed only in the resistant or susceptible cultivar (e.g. RNase-L-inhibitor-like protein), and mention that the identified genes show homologies to genes related to plant stress. For D. cf. devecta, the existence of a resistance

69 locus at the top of linkage group 7 in 'Fiesta' (Sd1) has previously been reported (Roche et al. 1997, Cevik and King 2002a). The mentioned studies used aphid inoculations and the experiments were either carried out in glasshouses (Alston and Briggs 1977, Roche et al. 1997), or in glasshouses and in the field (Cevik and King 2002a). In our study, 140-160 apple genotypes were exposed, in comparison, to aphid field infestation under different environmental conditions. This study is the first QTL analysis of D. plantaginea and A. pomi resistance, and it presents the probable position (QTL identified only once and in a single place) of a D. plantaginea resistance gene within the apple genome. In addition, the present study confirms the results of the mapping of the resistance gene Sd1 (Cevik and King 2002a) in a different genetic background ('Fiesta' x 'Discovery') under natural infestation by different populations of the D. cf. devecta species complex. Cevik and King (2002a) mapped the Sd1 resistance gene to D. cf. devecta within an interval of 1.3 cM delimited by the molecular markers SdSSRa and 2B12a at the top of the 'Fiesta' chromosome 7. Khan et al. (2007) developed a SSR marker, called CH-Sd1, from the sequence of the BAC clone 49N23 belonging to the contig spanning the Sd1 locus (Cevik and King 2002a). CH-Sd1 has been mapped in the 'Fiesta' x 'Discovery' cross used in this study on F7 between the markers P13-2000 and E35M42-0480 at only 2cM from E32M39-0195 (Khan et al. 2007). The CH-Sd1 locus is therefore exactly in the confidence interval of the QTL identified in this study on 'Fiesta' linkage group 7. This resistance is therefore due to Sd1. To evaluate the applicability in apple breeding of the SSR markers Hi07h02 and Hi03a10, which were found to be closely positioned to the QTLs for D. plantaginea and D. cf. devecta resistance, apple cultivars described as resistant to aphids could be used. If the alleles of the SSR markers in coupling with these resistances (allele '255 bp' for Hi07h02 and allele '216 bp' for Hi03a10) are present in the resistant apple varieties, these cultivars and the SSR markers could be used for marker assisted breeding. Based on our pedigree analyses, the alleles in coupling with the resistance occur in 'Idared' and 'Wagener' (Hi07h02-255: D. plantaginea resistance), and in 'Cox's Orange Pippin' and 'Blenheim Orange' (Hi03a10-216: D. cf. devecta resistance). For 'Wagener' and 'Cox's Orange Pippin', resistance to D. plantaginea and D. cf. devecta, respectively, was reported (Alston and Briggs 1977), whereas no information on resistance was found for 'Idared' and 'Blenheim Orange'. In combination, these results suggest that apple varieties with known resistance to aphids should be tested for the presence of the markers, and that the markers could also be used to detect so far unknown resistance in other apple cultivars.

70 Environmental factors related to aphid resistance Our study shows that for field populations of insect pests and especially for aphids, the genetic background of the host plant could partly explain infestation levels but that environmental and neighborhood effects render their identification very difficult. The identified QTLs were not stable for different sites and years, and aphid abundance on specific apple genotypes was not strongly correlated between different sites and years. Different climatic conditions at the sites may impede the identification of genetically based resistance. Highest aphid infestation was found for the Ticino site, which was characterized by dry and warm climate conditions that are known to favor aphid population growth (Blommers et al. 2004). Drought stress in apple increases emission of volatile secondary plant metabolites (Vallat et al. 2005), and plant volatiles were shown to affect aphid distribution (Bernasconi et al. 1998). Field observations showed that plant phenology was more progressed at the Ticino site, allowing an earlier infestation with faster population build-up. This may partly explain the lack of correlations of A. pomi infestation between the sites in May and June, and its presence in July. D. cf. devecta was not present at the Zurich site. Insecticides were applied in neighboring orchards, and this may have reduced population build-up due to low aphid immigration. Besides population fluctuation and climate, various other environmental parameters may affect aphid abundance and blur genetically based resistance in apple. Highly susceptible genotypes may be severely infested by aphids and serve as a source for aphid colonization of neighboring trees (Blommers et al. 2004). A significant spatial pattern and an influence of neighbor tree infestation level were found for D. plantaginea and A. pomi. The positive spatial autocorrelation as well as significant neighborhood effects may partly explain why the QTL for D. plantaginea resistance was not identified in Year 2 at the Ticino site and at the Zurich site, and why no QTL was identified for A. pomi. There was a significant effect of neighbor tree infestation mainly at the Valais and Zurich sites. At these sites, trees were planted at a distance of 0.5 m compared to the 1.25 m at the Ticino site, which can increase the tree-to-tree colonization by aphids. Interestingly, the QTL on the 'Fiesta' linkage group 17 for D. plantaginea resistance was identified at the Ticino site, but not at the Valais and Zurich site, where a neighborhood effect was present. Orchard-wide distribution patterns were also identified for the woolly apple aphid (Eriosoma lanigerum) (Alspach and Bus 1999). The authors mention that computation of genetically based resistance may be influenced by significant spatial autocorrelation.

71 Direct interspecific interactions between aphids or similar resource requirements may also affect the distribution of individual species. At the Ticino site in Year 1, distribution of all the aphid species was positively correlated. In general no negative relationships between the aphid species were found, indicating that competition among aphids was not an important factor affecting their distribution. Similar resource requirements mediated by plant characteristics may, however, help to explain the concordant distribution patterns. If resources commonly used by different species are reflected by specific plant-growth characteristics, distribution of different herbivore species should independently correlate with those characteristics. Indications for such congruent resource use were mainly found for stem diameter and for shoot length and new-shoot-length, showing a significant positive correlation especially with D. plantaginea and A. pomi infestation. The QTL analysis of different plant-growth characteristics revealed that QTLs for shoot length and stem diameter were generally positioned on other linkage groups than the QTLs for aphid resistance. This finding indicates that the identified QTLs for aphid resistance are not likely to be correlative artifacts of plant trait effects. This reasoning may also apply to the putative QTL for shoot length and the QTL for D. cf. devecta resistance. Although both QTLs were positioned on 'Fiesta' linkage group 7, they do not appear to be strongly related. Whereas the QTL for D. cf. devecta resistance was found at different sites in both study years, the putative QTL for shoot length was only found at the Ticino site in Year 1. In addition, the correlations between D. cf. devecta infestation and shoot length for the Ticino site were weak or not significant. Effects of predators on aphid populations are variable (Dixon 2000). In our study, field observations did not reveal relevant predator densities, suggesting that quantification of aphid infestation level was not strongly influenced by predation.

Outlook The identified SSR markers for aphid resistance could be used in breeding programs where there is a demand for increased fruit quality and plant resistance (Avilla and Riedl 2003). The value of the detected markers linked to a resistance QTL applicable for MAS depends on the ease with which they can be introgressed into commercial cultivars. A combined use of the identified markers associated to resistance to D. plantaginea and D. cf. devecta may facilitate the breeding of new apple cultivars resistant to several aphid species. As markers on linkage group 7 of 'Fiesta' are linked to fire blight resistance (Khan et al. 2007), the study also emphasizes the potential for gene pyramiding in broad-spectrum resistance breeding of apple. The same applies to aphid resistance, as the putative QTL for D.

72 plantaginea resistance, identified in the present study, was found on the 'Fiesta' linkage group 17, as was a gene most recently found for E. lanigerum resistance (Bus et al. 2008). Furthermore, apple cultivars described as resistant to aphids and amplifying the SSR alleles associated with increased resistance to aphids, provide a useful basis to facilitate traditional apple breeding. The utilization of apple cultivars with the resistance-related marker alleles can minimize failures in apple breeding as resistance should be genetically based and are not strictly environment-dependent.

73 6.5. Supplementary material (SM)

Table SM 6.5.1. Climatic characterization of the three study sites during the observation period (March - August) in the two years of survey, and comparison to standard values a.

Mean ± SE Deviation from Total rainfall Rainfall %

temperature (°C) standard (°C) (mm) standard

Ticino Year 1 16.8 ± 0.1 +1.7 679 66 Year 2 16.7 ± 0.1 +1.6 720 70 Valais Year 1 15.2 ± 0.1 +1.5 305 109 Year 2 14.9 ± 0.1 +1.2 406 145 Zurich Year 1 13.8 ± 0.1 +1.2 855 109 Year 2 13.6 ± 0.1 +1.0 854 109 a Standard values are based on 30-year average climatic data (1960 to 1990), obtained from MeteoSwiss.

Table SM 6.5.2. Distribution patterns of aphids (Dysaphis plantaginea, D. cf. devecta, Aphis pomi) on individual apple trees characterized by the index of dispersion (ID). All ID values significantly greater than the χ2 statistic (0.025 probability level) with (ν-1) degrees of freedom indicate an aggregated distribution of the studied species (P < 0.001 for all observed ID values).

Ticino Valais Zurich

D. plantaginea Year 1 2,322 2,988 5,600 Year 2 4,276 1,093 - D. cf. devecta Year 1 4,083 946 - Year 2 4,278 1,535 - A. pomi Year 1 34,364 137,360 15,426 Year 2 20,887 24,475 3,877

74 Table SM 6.5.3. Relationship between the aphid species Dysaphis plantaginea, D. cf. devecta, and Aphis pomi assessed by Spearman's rank test at the Ticino, Valais, and Zurich site. The average aphid infestation at a study site and within a year was used for correlation analysis. Significant correlations after FDR correction are highlighted in bold. The number of colonies (D. plantaginea), the number of red-curled leaves (D. cf. devecta) and the number of aphids (A. pomi) was taken as measure of aphid infestation.

D. plantaginea D. cf. devecta A. pomi rs P rs P rs P

D. plantaginea Ticino (n = 143) Year 1 - - 0.222 0.008 0.445 <0.0001 Year 2 - - -0.105 0.211 0.020 0.809 Valais (n = 142) Year 1 - - 0.021 0.801 -0.029 0.733 Year 2 - - 0.147 0.081 0.137 0.105 Zurich (n = 153) Year 1 - - - - 0.083 0.308 D. cf. devecta Ticino (n = 143) Year 1 0.222 0.008 - - 0.005 0.957 Year 2 -0.105 0.211 - - 0.001 0.998 Valais (n = 142) Year 1 0.021 0.801 - - 0.107 0.204 Year 2 0.147 0.081 - - 0.063 0.460 Zurich (n = 153) Year 1 ------A. pomi Ticino (n = 143) Year 1 0.445 <0.0001 0.005 0.957 - - Year 2 0.020 0.809 0.001 0.998 - - Valais (n = 142) Year 1 -0.029 0.733 0.107 0.204 - - Year 2 0.137 0.105 0.063 0.460 - - Zurich (n = 153) Year 1 0.083 0.308 - - - -

75 Table SM 6.5.4. Apple varieties with reported resistance properties to the three aphid species Dysaphis plantaginea, D. cf. devecta and Aphis pomi

Dysaphis plantaginea Cultivar Reference

Baujade Audemard et al. 1992 Cleeve Briggs and Alston 1969 Colwall Quoining Briggs and Alston 1969 Cox's Orange Pippin Briggs and Alston 1969 Delorina Graf et al. 1998 Elise Rathke Briggs and Alston 1969 Florina Audemard et al. 1992, Graf et al. 1992, Rat- Morris 1993, Graf et al. 1998, Dapena and Miñarro 2001, Qubbaj et al. 2005, Angeli and Simoni 2006, Arnaoudov and Kutinkova 2006 Fraise de Buhler Briggs and Alston 1969 Franklyn's Golden Pippin Briggs and Alston 1969 Gelber Münsterländer Borsdorfer Briggs and Alston 1969 Golden Orange Briggs and Alston 1969 Graine Briggs and Alston 1969 Liberty Arnaoudov and Kutinkova 2006 MAL 59/9 Alston and Briggs 1970, Massonie et al. 1981, Lyth 1985 McIntosh Garman 1938 Ontario Briggs and Alston 1969 Pepino Jaune Briggs and Alston 1969 Perle d'Angleterre Briggs and Alston 1969 Reanda Habekuss et al. 2000 Remo Habekuss et al. 2000 Resi Graf et al. 1998 Rewena Habekuss et al. 2000 Saturn Graf et al. 1998 Schöner von Nordhausen Briggs and Alston 1969 Severn Bank Briggs and Alston 1969 Spartan Graf et al. 1992 Thurgauer Weinapfel Briggs and Alston 1969 Vista Bella Graf et al. 1992 Wagener Briggs and Alston 1969

76 Table SM 6.5.4. Continued.

Dysaphis cf. devecta Cultivar Reference

Angold Graf et al. 1998 Cox's Orange Pippin Alston and Briggs 1968, 1977 Delorina Graf et al. 1998 Fiesta Roche et al. 1997, Cevik and King 2002a Lane's Prince Albert Alston and Briggs 1977 MAL 59/9 Alston and Briggs 1977 MAL 68/5 Alston and Briggs 1977 McIntosh Alston and Briggs 1977 Northern Spy Alston and Briggs 1977 Resista Graf et al. 1998 Rewena Graf et al. 1998 Worcester Pearmain Alston and Briggs 1977

Aphis pomi Cultivar Reference

Baldwin Caesar et al. 1934 Bogatyr' Sokolov 1965 Brat Pobedy Sokolov 1965 Florina Graf et al. 1998 La Nationale Briggs and Alston 1969 Langridge I Briggs and Alston 1969 McIntosh Caesar et al. 1934 Northern Spy Caesar et al. 1934 Pobeda Sokolov 1965 Ralls Janet Briggs and Alston 1969 Red Delicious Oatman and Legner 1961 Resi Graf et al. 1998, Habekuss et al. 2000 Suvorovets Sokolov 1965 Vered Oppenheimer 1962

77

Figure SM 6.5.1. Spatial distribution of aphids (Dysaphis plantaginea, D. cf. devecta and Aphis pomi ) described by fitting trend surfaces on contour plots by kriging (best unbiased generalized least squares estimation) with an exponential covariance function. Mean aphid infestation of different observations within a year was considered for contour plots (cf. Table 6.1). Infestation level of individual trees is reflected by circle size. Yellow-colored circles indicate trees with an infestation greater than zero, whereas uninfested trees are represented by open circles. No contour lines were plotted for the Valais site due to the orchard design with two rows.

78 7. Aphis pomi population development, shoot characteristics, and antibiosis resistance in different apple genotypes 4

In high value crops such as apple (Malus x domestica), insecticidal pest control is of high relevance. The use of resistant apple cultivars can increase the sustainability of pest management in apple orchards. Besides variation in plant chemistry that may influence plant resistance by antibiosis or antixenosis, plant growth characteristics can also affect plant susceptibility to pests such as aphids. Variable susceptibility to the green apple aphid (Aphis pomi), has been described for different apple cultivars. These observations were based on phenotypic surveys and no information on genetically based apple resistance to A. pomi is yet available. The objective of this study was to relate shoot growth characteristics with aphid population development, and to assess the genetic background of apple antibiosis-based resistance to A. pomi by quantitative trait loci (QTLs) analysis. Aphid population development was repeatedly studied in the field in sleeve cages attached to 200 apple trees of different genotypes. Aphid population development was positively correlated to shoot length and growth, and was also affected by climatic conditions. Indications for antibiosis-based resistance to A. pomi remained weak in the studied apple genotypes, and the only detected putative QTL on linkage group 11 of the apple cultivar 'Fiesta' was not stable for the different replications of the experiment. This lack of quantifiable resistance may be partly explained by environmental conditions related to aphid development in sleeve cages.

______4 Stoeckli, S., K. Mody, and S. Dorn. 2008. Aphis pomi population development, shoot characteristics, and antibiosis resistance in different apple genotypes. Journal of Economic Entomology 101:1341- 1348.

79 7.1. Introduction

The green apple aphid (Aphis pomi De Geer) (Hemiptera: Aphididae), is one of the most abundant and widespread herbivore pests in apple (Malus x domestica) (Borkh.), orchards of Europe, North America and the (Blackman and Eastop 2000). Infestation with A. pomi can stunt plant growth and stimulate lateral shoot growth, especially on young, nonbearing plants with high infestation levels on shoot tips (Arbab et al. 2006). Mature trees are mostly affected by the honeydew production of the aphids dripping onto foliage and fruit. Honeydew stimulates sooty mold growth, which hinders key leaf functions and fruit ripening (Coleson and Miller 2005). Honeydew, in addition, attracts ants that protect aphid colonies on apple trees from predation (Stewart-Jones et al. 2008). Natural enemies would otherwise maintain aphid populations at low levels, although potential biological control agents may also have undesirable augmentative effects on A. pomi populations (Chouinard et al. 2006). To reduce yield losses, the application of insecticides such as systemic neonicotinoids against A. pomi is of high relevance in apple production (Lowery et al. 2005). A reduction of insecticide use for economic and environmental reasons may be achieved by Integrated Pest Management (IPM) programs (Kogan 1998), which combine insecticide use with biological control measures, orchard management and development of resistant cultivars. Morphological and chemical plant characteristics that result in higher levels of resistance to herbivore damage may be inherited. For apple, information on the genetic basis of aphid resistance is scarce. So far, gene regions related to resistance to the rosy leaf curling aphids (Dysaphis cf. devecta Wlk. species complex) (Roche et al. 1997, Stoeckli et al. 2008c), the rosy apple aphid (Dysaphis plantaginea Pass.) (Stoeckli et al. 2008c), and the woolly apple aphid (Eriosoma lanigerum Hausm.) (Bus et al. 2008), have been reported. The genetic basis of antibiosis- and antixenosis-resistance to aphids has been investigated in more detail for different herbaceous plants such as soybean (Glycine max) (Diaz-Montano et al. 2006, Hill et al. 2006), wheat (Triticum aestivum) (Hawley et al. 2003, Castro et al. 2005), tomato (Lycopersicum esculentum) (Kohler and St. Clair 2005), taro (Colocasia esculenta) (Coleson and Miller 2005), and barrel medic (Medicago truncatula) (Klingler et al. 2005). Plant resistance to aphids may be influenced by different plant characteristics, including plant volatiles (Bernasconi et al. 1998), leaf morphology (Cutright 1930, Hill et al. 2004, Coleson and Miller 2005, Whitaker et al. 2006), secondary plant compounds (Gayler et al. 2004), and nutritional quality (Dorschner et al. 1987, Miyasaka et al. 2007). Climate, crop

80 management practices, soil quality, and herbivore activity can all affect these resistance- related plant traits, evoking variation in resistance to aphids in different environments. Besides effects on plant resistance, environmental parameters such as temperature (Wang and

Tsai 2000, McCornack et al. 2004, Arbab et al. 2006), rainfall and CO2 emission (Newman 2005) are also known to directly affect aphid population growth. Susceptibility to A. pomi has been found to differ between apple cultivars(Oatman and Legner 1961, Oppenheimer 1962, Sokolov 1965, Vanin 1965, Briggs and Alston 1969, Habekuss et al. 2000). The mentioned studies were, however, based on a phenotypic evaluation, so the genetic basis of resistance to A. pomi has not been described yet. To better understand resistance to A. pomi in apple, a combination of methods assessing genotype effects such as quantitative trait loci (QTLs) analysis, and a detailed consideration of environment-dependent plant characteristics appears necessary. QTL analysis uses genetic linkage maps, which show the position of known genes or markers relative to each other in terms of recombination frequency, to identify gene regions associated to quantitative (polygenic) phenotypic traits (Liebhard et al. 2003b). It is known that A. pomi is most abundant on vigorously growing shoots (Cutright 1930, Hull and Grimm 1983, Beers et al. 1993, Whitaker et al. 2006), and that food quality, especially the availability of the macronutrient nitrogen, affects aphid development (Jansson and Smilowitz 1986, Dorschner et al. 1987, Miyasaka et al. 2007). Aphids are phloem feeders and prefer high concentrations of nitrogen (White 1993). In apple plants, nitrogen concentration is higher in younger, actively growing shoots than in older shoots (O'Kennedy et al. 1975, Khemira et al. 1998, Neilsen and Neilsen 2003). We therefore assumed a relationship between morphological apple shoot characteristics and aphid population development. Support for this hypothesis is available from observational studies on A. pomi (Cutright 1930). The objective of this study was therefore to relate shoot growth characteristics and aphid population development, and to assess aspects of the genetic background of apple resistance to A. pomi. We measured shoot length, growth and diameter and related these parameters to aphid population development in 200 apple trees of different genotype. The effects of stem diameter and median shoot length per tree (as measures of general tree vigor), and of climatic conditions on aphid population development were also considered. To elucidate aspects of the genetic background of apple resistance to A. pomi, a QTL analysis testing resistance by antibiosis was conducted.

81 7.2. Materials and methods

Study site and plant material Field experiments were performed in an apple orchard located in northern Switzerland (Wadenswil, 47°13'20''N, 8°40'05''E and 455 m altitude; Agroscope Changins-Wadenswil Research Station ACW) in 2006 and 2007 (Year 1 and Year 2). Climate data for the orchard were retrieved from MeteoSwiss (http://www.meteoschweiz.ch). The 200 apple trees (each representing a different genotype) used in this study were a F1-cross of the apple varieties 'Fiesta' x 'Discovery'. They were randomly selected from 251 progeny trees, which had been bud-grafted on M27 rootstocks and planted in the apple orchard in 1998 (Liebhard et al. 2003b).

Aphid population development in sleeve cages Aphis pomi individuals were collected from the orchard in Wadenswil, transferred to the laboratory and reared on apple plants until needed in field experiments. Aphid population development in the field was quantified in sleeve cages (mesh size: 230 µm3, diameter: 20 cm, length: 60 cm) that were fixed to the different apple genotypes. The sleeve cage experiments were carried out once in Year 1 (Experiment 1; June) and three times in Year 2 (Experiments 2, 3 and 4; June, July and July/August). For Experiments 1, 2 and 4, the same 100 apple trees were used, whereas for Experiment 3 supplementary 100 apple trees were considered. The timing of the experiments was based on weather forecast and aphid abundance in the field. Three (Experiment 1) or five (Experiments 2-4) wingless adult female aphids were randomly selected from the laboratory population and transferred into the sleeve cages. The number of these aphids was increased in Year 2 to reduce variability in population development, which arose in Year 1 possibly because of random, pre-reproduction die-off of the transferred aphids. The resulting reduction in noise in the data was considered to improve the precision of the experiments (comparing Experiment 1 to Experiments 2-4). As microclimate within a tree canopy or in an apple orchard may differ from general site conditions (Landsberg 1973, Stoeckli et al. 2008b), aphid population development was tested on different tree positions. On each tree, three sleeve cages were fixed on the current year's actively growing shoots (shoots). Sleeve cages were restricted to 15 cm shoot length to equalize space for aphid development. Two sleeve cages were attached at a medium height of the canopy (middle tree position) at the northwestern and southeastern canopy aspect, and the third cage was fixed at the top of the tree, equally exposed to all compass directions. Shoots

82 were randomly selected given two restrictions: they had to have a minimal length of 15 cm, and shoots with a pronounced vertical top growth (watersprouts) were not considered. Incidental insects, including other aphids, and mites were removed prior to sleeve cage setup. After 12-14 d of exposure of the aphids to field conditions, the shoots were cut, placed in individual freezable plastic bags and stored at 4 °C. The number of aphids that had developed within the sleeve cages was counted within 2-5 d.

Shoot characteristics and general tree vigor We measured the length, growth and diameter of each selected shoot. Shoot length was assessed at the beginning of the experiment in the field. Shoot growth was quantified as the difference between shoot length measured at the beginning and at the end of the experiment. Shoot diameter at the approximate mid-point of the shoot was assessed at the end of the experiment in the laboratory using a standard caliper. Length of all shoots per tree and stem diameter were quantified in Experiment 1 as measures of general tree vigor. The median length of all shoots per tree, and the stem diameter measured at 1.20 m, were used in the analysis. Aphid population development in sleeve cages was related to shoot growth characteristics and general tree vigor.

Antibiosis-based resistance The number of aphids that had developed within the sleeve cages was used for quantification of antibiosis-based resistance. Quantitative trait loci (QTLs) analyses of aphid population development per genotype (sum, median, maximal value) were carried out with MapQTL® 4.0 (van Ooijen et al. 2002) for each separate experiment (Experiment 1, Experiment 2, Experiment 3, and Experiment 4). Additionally, the summed number of aphids per genotype in all experiments (Experiment 1-4) was used for QTL analysis. Kruskal-Wallis tests and interval mapping (IM) were used for QTL analysis. The genetic linkage maps for both 'Fiesta' and 'Discovery' (single parent maps), used in the QTL analysis, were calculated with 251 apple genotypes and have already been published (Liebhard et al. 2003b). Logarithm of odds (LOD) threshold values were determined by 1,000-fold-permutation tests (MapQTL® 4.0). Significant threshold values were set to declare a QTL significant a the 95% confidence level (genome-wide) (King et al. 2000).

83 Statistical analysis Non-parametric tests were used as data did not meet the requirements (normality, homoscedasticity) for parametric test before and after data transformation. Statistical analyses were performed with SPSS 16.0 for Mac OS X (SPSS, Inc., Chicago, IL), with the exception of the Kruskal-Wallis post hoc test (Dunn's test), which was calculated with SsS 1.1a for PC (Rubisoft Software GmbH, Eichenau, Germany). Spearman's rank tests were used to study relationships between aphid development and shoot- or tree characteristics. When multiple correlation tests were carried out, the Benjamini-Hochberg procedure was applied to correct for false discovery rates (type I errors) (Verhoeven et al. 2005).

7.3. Results

Aphid population development Aphis pomi population development on the different apple genotypes was highly variable, and aphid numbers per sleeve cage ranged from zero (all experiments) to 319 (Experiment 1) (Table 7.1). Overall median aphid numbers were higher in June (Experiments 1 and 2) compared to July and August (Experiments 3 and 4) (Kruskal-Wallis test; d.f. = 3, χ2 = 47.8, P < 0.0001; Dunn's post hoc test, Table 7.1), and the percentage of sleeve cages containing no aphids increased from 34% (Experiments 1 and 2) to 43% (Experiment 3) and 49% (Experiment 4) (Table 7.1). Aphid population development on individual trees was not correlated across Experiments 1, 2 and 4 (Spearman correlation, Benjamini-Hochberg procedure; n = 100, rs = -0.051-0.266, P > 0.05 for all tree positions, data not shown), indicating that aphid population development on a particular tree genotype varied between experiments.

Shoot characteristics and general tree vigor Median length of shoots in sleeve cages differed significantly between experiments (Kruskal-Wallis test; d.f. = 3, χ2 = 143.4, P < 0.0001). Within the same year on the same apple trees, shoot length increased from 22 cm (Experiment 2 in early summer) to 26 cm (Experiment 4 in late summer) (Table 7.1), and it was not significantly different between Experiment 3 and Experiment 4, which were carried out in mid and late summer of the same year (Dunn's post hoc test, Table 7.1). Median growth of shoots in sleeve cages varied significantly across experiments (Kruskal-Wallis test; d.f. = 3, χ2 = 200.2, P < 0.0001), being

84 highest during Experiment 1 in early summer of the first year, and amounting to zero in the other experiments (Dunn's post hoc test, Table 7.1). Median diameter of caged shoots differed significantly across experiments (Kruskal-Wallis test; d.f. = 3, χ2 = 49.7, P < 0.0001). It was smaller in Experiment 1 and Experiment 2, carried out in early summer in two subsequent years, compared to Experiment 3 and Experiment 4 later in summer (Dunn's post hoc test, Table 7.1).

Table 7.1. Aphis pomi population development and shoot characteristics on 200 apple genotypes, and climatic conditions for Experiment 1, 2, 3, and 4. Median, 25th and 75th percentile, and maximal value for aphid population development and shoot characteristics are presented, as well as mean daily temperature (mean ± SE) and mean value of precipitation (mean ± SE) during aphid exposure. Different letters a,b,c indicate significant differences between experiments (P < 0.05).

Experiment 1 Experiment 2 Experiment 3 Experiment 4

Aphid population development (no.) A Median 10 a 4 a 1 b 1 b 25th percentile 0 0 0 0 75th percentile 33 17 6 8 Maximal value 319 152 51 77 Shoot length (cm) A Median 18 c 22 b 26 a 26 a 25th percentile 16 19 21 20 75th percentile 21 27 32 32 Maximal value 30 30 48 54 Shoot growth (cm) A Median 7 a 0 b 0 b 0 b 25th percentile 3 0 0 0 75th percentile 11 2 2 1 Maximal value 35 29 21 13 Shoot diameter (cm) A Median 0.4 a 0.4 a 0.4 b 0.4 b 25th percentile 0.3 0.3 0.4 0.3 75th percentile 0.5 0.4 0.5 0.4 Maximal value 0.6 0.6 0.8 0.7 Temperature (°C) B 20.2 ± 0.7 a 19.7 ± 0.5 a 15.3 ± 0.6 b 18.8 ± 0.5 a

Precipitation (mm) B 2.4 ± 1.7 a 4.2 ± 1.7 a 10.2 ± 3.4 a 6.7 ± 2.7 a

A Kruskal-Wallis test and Dunns' post hoc test B One-way ANOVA and Scheffé post hoc test

85 Shoot length was positively correlated to shoot growth at all three tree positions in

Experiment 1 (Spearman rank correlation; n = 100, rs = 0.395-0.506, P < 0.0001), to the southeast (n = 100, rs = 0.261, P = 0.009) and northwest (n = 100, rs = 0.485, P < 0.0001) tree position in Experiment 2, and to the northwest tree position in Experiment 3 (n = 100, rs = 0.261 and 0.304, P = 0.002). No significant correlation between shoot length and shoot growth was found in Experiment 4 (Table 7.2). Shoot length and shoot diameter were never significantly correlated (rs = 0.061-0.238, P > 0.05; data not shown). Stem diameter as measure of general tree vigor ranged between 7.6 and 17.0 cm, and varied between 6 and 29 cm when quantified as median shoot length (all shoots per tree). There was a high variation of shoot length s within a tree (coefficient of variation = 26-92%, data not shown).

Table 7.2. Relationship between the two apple plant characteristics 'shoot length' and 'shoot growth', assessed by Spearman rank correlation. Analysis was carried out separately for each tree position (n = 100). Significant correlations are highlighted in bold.

Tree position Experiment 1 Experiment 2 Experiment 3 Experiment 4

Middle Southeast rs = 0.395 rs = 0.261 rs = 0.116 rs = -0.021 P < 0.0001 P = 0.009 P = 0.250 P = 0.832 Northwest rs = 0.467 rs = 0.485 rs = 0.304 rs = -0.134 P < 0.0001 P < 0.0001 P = 0.002 P = 0.185 Top rs = 0.506 rs = 0.144 rs = 0.073 rs = 0.051 P < 0.0001 P = 0.154 P = 0.469 P = 0.614

Climatic conditions Mean daily temperature during the 12-14 days of aphid exposure varied significantly among the four experiments (one-way ANOVA, d.f. = 3, MS = 71.491, F3,59 = 12.861, P < 0.0001). During Experiment 3 mean daily temperature was significantly lower compared to the other experiments (Scheffé post hoc test, Table 7.1). Mean daily precipitation ranged from 2.4 mm (Experiment 1) to 10.2 mm (Experiment 3) and was not significantly different across the experiments (one-way ANOVA, d.f. = 3, MS = 165.840, F3,59 = 1.552, P = 0.221).

86 Aphid development and shoot characteristics at different sleeve cage positions As environmental conditions may differ across tree positions, we studied aphid development at the middle (southeast and northwest) and at the top tree position. At the southeast position, aphid numbers varied between 0 and 319, whereas aphid numbers ranged from 0 to 152 at the northwest and from 0 to 202 at the top tree position. Aphid numbers were not congruently highest at the southeast compared to other tree positions. This is in accordance with the finding that there was no significant difference in aphid numbers in sleeve cages positioned at middle tree positions (southeast and northwest) or the top tree position (Friedman test; n = 100, d.f. = 2, χ2 = 2.0-3.5, P > 0.05 for all experiments, data not shown). There was no significant difference in shoot length and growth at the different tree positions for all experiments (Friedman test; n = 100, d.f. = 2, χ2 = 0.5-5.1, P > 0.05, data not shown).

Relationship between aphid population development and shoot characteristics A significant relationship between shoot length and aphid population development was detected for different Experiments (Table 7.3). In Experiments 2 and 3 (June and July), shoot length was significantly positively correlated to the number of aphids at the southeast and northwest middle tree position and also at the top tree position (Spearman rank correlation; n = 100, rs = 0.334-0.421, P < 0.001). There was no relationship between shoot length and aphid population development in Experiments 1 and 4 (n = 100, rs = -0.164-0.198, P = 0.051-0.970), with one exception (Table 7.3). For the northwest tree position in

Experiment 4, the number of aphids was significantly correlated to shoot length (n = 100, rs = 0.417, P < 0.001). Aphid population development was significantly correlated with shoot growth in Experiment 2 (n = 100, rs = 0.265-0.328, P = 0.001-0.008) and Experiment 3 (n =

100, rs = 0.259-0.385, P = 0.0001-0.009), but not in Experiment 1 (n = 100, rs = -0.108 - -

0.005, P = 0.283-0.959) and Experiment 4 (n = 100, rs = -0.047-0.001, P = 0.642-0.994) (Table 7.3). No relationship between diameter of shoots and aphid numbers was detected for Experiments 1, 2, and 4, neither for the two middle tree positions nor for the top tree position

(Spearman rank correlation; n = 100; rs = -0.107-0.166, P = 0.099-0.869, data not shown). In Experiment 3, a correlation was found between shoot diameter and number of aphids that developed in the southeast tree position (n = 100, rs = 0.261, P = 0.009) and at the top tree position (n = 100, rs = 0.258, P = 0.010), but not in the northwest tree position.

87 Table 7.3. Relationship between aphid population development (aphid number) and the shoot characteristics 'shoot length', 'shoot growth' and 'shoot diameter' assessed by Spearman rank correlation. Analysis was carried out separately for each tree position (n = 100). Significant correlations are highlighted in bold. For further details on Experiments 1 to 4 see Table 7.1.

Tree position Experiment 1 Experiment 2 Experiment 3 Experiment 4

Shoot length Middle Southeast rs = -0.004 rs = 0.421 rs = 0.497 rs = 0.198 P = 0.970 P < 0.0001 P < 0.0001 P = 0.048 Northwest rs = 0.067 rs = 0.334 rs = 0.270 rs = 0.417 P = 0.508 P = 0.001 P = 0.007 P < 0.0001 Top rs = -0.164 rs = 0.394 rs = 0.221 rs = 0.188 P = 0.103 P < 0.0001 P = 0.027 P = 0.060 Shoot growth Middle Southeast rs = -0.061 rs = 0.328 rs = 0.259 rs = -0.034 P = 0.546 P = 0.001 P = 0.009 P = 0.749 Northwest rs = -0.005 rs = 0.320 rs = 0.385 rs = 0.001 P = 0.959 P = 0.001 P < 0.0001 P = 0.994 Top rs = -0.108 rs = 0.265 rs = 0.286 rs = -0.047 P = 0.283 P = 0.008 P = 0.004 P = 0.642 Shoot diameter Middle Southeast rs = 0.127 rs = 0.166 rs = 0.261 rs = 0.059 P = 0.198 P = 0.099 P = 0.009 P = 0.559 Northwest rs = 0.091 rs = -0.017 rs = 0.129 rs = 0.099 P = 0.617 P = 0.869 P = 0.202 P = 0.327 Top rs = 0.050 rs = 0.026 rs = 0.258 rs = -0.107 P = 0.451 P = 0.795 P = 0.010 P = 0.287

Relationship between aphid population development and general tree vigor There was no significant relationship between general tree vigor, measured as stem diameter or median shoot length per tree, and aphid population development in the sleeve cages. The correlation between stem diameter and aphid number (sum per tree) was not significant for all experiments (Spearman rank correlation; n = 94-99, rs = 0.059-0.224, P > 0.05 after Benjamini-Hochberg procedure, data not shown). There was also no significant correlation between median shoot length per tree and aphid number (sum per tree) in all experiments (n = 94-99; rs = 0.042-0.249, P > 0.05 after Benjamini-Hochberg procedure, data not shown).

88 Antibiosis-based aphid resistance A putative QTL for antibiosis resistance of apple to A. pomi was found in Experiment 3 (sum of aphid numbers in the three sleeve cages per apple genotype) on the 'Fiesta' linkage group 11. The marker closest positioned to the QTL was the microsatellite (SSR) CH02d12, allele '199 bp', at 21.5 cM. The LOD score for this SSR was 2.15 (LOD threshold level was 1.9), and the marker explained 14.1% of the phenotypic variability. This QTL was not stable among experiments and disappeared in the other experiments, and when aphid population development of Experiment 1-4 were summed (Figure 7.1).

89

Figure 7.1. QTL analysis results for the putative QTL associated to antibiosis-based resistance in apple to Aphis pomi on the 'Fiesta' linkage group 11. The x-axis indicates the linkage map of 'Fiesta' in cM and marker name; the y-axis shows the LOD scores. LOD threshold value was 1.9 (Experiment 3) and ranged between 1.6 and 2.3 for the other QTL analyses. The closest marker to the putative QTL is highlighted.

90 7.4. Discussion

This study aimed at quantifying the relationship between growth characteristics of apple shoots and population development of the green apple aphid (Aphis pomi) and at assessing antibiosis-based resistance in apple to A. pomi by QTL analysis. Indications that A. pomi abundance is highest on vigorously growing shoots were so far based on observations from single apple genotypes (Cutright 1930, Whitaker et al. 2006). Our study supports these observations and presents, for the first time, a statistically significant correlation between aphid population development and growth characteristics of individual shoots over an extended period of time and for a high number (i.e. 200) of different apple genotypes. Aphid numbers in sleeve cages were influenced by shoot length and shoot growth. The detected relationship was stable for the three tree positions studied, and the fluctuation in aphid population development on individual shoots appeared to be independent of stem diameter or median shoot length per tree as measures of general tree vigor. Our findings thus underline the importance of within-tree variation of nutritional quality for the distribution of herbivorous insects in tree canopies (Neilsen and Neilsen 2003, Unsicker and Mody 2005, Stoeckli et al. 2008b) The significant positive relationship between aphid population development and shoot growth was found in June and July (Experiments 2 and 3), but not in late summer (Experiment 4). This finding may be related to a decrease in both, shoot growth (from June to August) and in nitrogen levels, which are highest in spring and decrease over the growing season (Khemira et al. 1998). Later in the season, when shoot growth is terminated, nutritional levels of the different shoots are probably less variable than during spring and summer (Neilsen and Neilsen 2003). Our study is in line with the expectation that a relationship between aphid population development and shoot growth is then no longer expected. Population growth of A. pomi may not only be influenced by changing nutrient levels with progressing season, but also by changing secondary metabolite concentrations in leaves and the surrounding headspace. Numerous secondary metabolites such as polyphenolic compounds including phloridzin (Hunter and Hull 1993), terpenoids and esters (Vallat and Dorn 2005) that can affect fruit tree herbivores (Montgomery and Arn 1974, Hern and Dorn 2004) vary in concentration during the growing season. Ambient climatic conditions, in particular drought or rainfall, further modify release of apple plant chemicals (Sircelj et al. 2005, Vallat et al. 2005). These variations in environmental factors may mask a potential genetic background of apple resistance to A. pomi.

91 Within-season differences in aphid population development may also be related to variations in daily temperature (Graf et al. 1985). The results of our experiments were probably not strongly affected by this parameter as all experiments were carried out during the summer months when temperatures never reached the lower (5.9°C; Graf et al. 1985) or the upper (35.6°C; Arbab et al. 2006) temperature thresholds for A. pomi development. This assumption is also supported by our finding that low aphid population development occurred in both, Experiment 3 (with lowest temperature of all four experiments), and Experiment 4 (with temperature similar to those in Experiment 1 and 2). Strong precipitation, which tends to wash aphids off their plants, partly explains reduced aphid population growth (Newman 2005), but there was no significant difference in precipitation in our four experiments and thus this variable could hardly have exerted a strong influence on the experiments' outcome. As there was no difference in aphid population development between different tree positions, possible effects of within-tree variability of microclimate on aphid population development appear negligible. We found a high variation of A. pomi population development on different apple genotypes and identified a significant QTL for antibiosis-based resistance in apple to A. pomi in June (Experiment 3) on the 'Fiesta' linkage group 11. Resistance to A. pomi has been reported for different apple cultivars, suggesting that resistance of apple to A. pomi is genetically fixed, and supporting the possible genetically based resistance identified in this study. The mentioned cultivars exhibiting resistance to A. pomi were: 'Baldwin', 'McIntosh' and 'Northern Spy' (Caesar et al. 1934), 'Bogatyr', 'Brat Pobedy', 'Pobeda' and 'Suvorovets' (Sokolov 1965), 'Florina' and 'Resi' (Habekuss et al. 2000), 'La Nationale', 'Landgridge 1' and 'Ralls Janet' (Briggs and Alston 1969), 'Red Delicious' (Oatman and Legner 1961), and 'Vered' (Oppenheimer 1962). The QTL determined in the present study was significant in only one of four experiments, it was highly variable across experiments, and it was not expressed when the combined data set of all experiments was considered. It thus seems that a weak genetic background for antibiosis to A. pomi may exist, but that environmental factors such as climate (Graf et al. 1985), orchard management (Simon et al. 2006), or morphological plant traits (Cutright 1930), play a more important role for A. pomi performance on apple trees in the field. The present study identified the strong influence of individual shoot growth characteristics for population development of A. pomi, rather than a major genetic background of apple resistance to A. pomi. Further experiments to assess the effect of different plant characteristics including phytochemicals on A. pomi populations should be conducted over

92 specific windows of time that overlap with major A. pomi development in the field. For this seasonal period our results documented the strongest interactions between plant traits and rapidly growing aphid populations.

93 8. Rust mite resistance in apple assessed by quantitative trait loci analysis 5

The aim of this study was to assess the genetic basis of resistance to the rust mite (Aculus schlechtendali) in apple (Malus x domestica). A. schlechtendali infestation of apple trees has increased as a consequence of reduced side effects of modern fungicides and narrow-spectrum acaricides on rust mites. An analysis of quantitative trait loci (QTLs) was carried out using linkage map data available for a segregating F1-cross of the apple cultivars 'Fiesta' x 'Discovery'. Apple trees representing 160 different genotypes were surveyed for rust mite infestation, each at three different sites in two consecutive years. The distribution of rust mites on the individual apple genotypes was aggregated and significantly affected by apple genotype and site. We identified two QTLs for A. schlechtendali resistance on linkage group 7 of 'Fiesta'. The AFLP marker E35M42-0146 and the RAPD marker AE10-400 were closest positioned to the QTLs and explained between 11.0 and 16.6% of the phenotypic variability. A pedigree analysis revealed that the resistance to A. schlechtendali derived from the apple cultivar 'Wagener'. Additionally, putative QTLs on the 'Discovery' chromosomes 4, 5 and 8 were detected. The SSR marker Hi03a10 identified to be associated to the QTLs will facilitate the breeding of resistant apple cultivars by marker assisted selection (MAS). Furthermore, the genetic background of rust mite resistance in existing cultivars can be evaluated by testing them for the identified SSR marker.

______5 Stoeckli, S., K. Mody, A. Patocchi, M. Kellerhals, and S. Dorn. 2008. Rust mite resistance in apple assessed by quantitative trait loci analysis. Tree Genetics & Genomes doi: 10.1007/s11295-008- 0186-5.

94 8.1. Introduction

The apple rust mite (Aculus schlechtendali Nalepa) is a serious pest in many apple growing regions of the world (Easterbrook and Palmer 1996). A. schlechtendali infestation of apple trees (Malus x domestica Borkh.) has increased along with changes from broad- spectrum to selective fungicides and acaricides. Many formerly applied sulphur-containing products for disease control and products for spider mite control exhibited side effects on rust mites and other non-target organisms, and have now been replaced by selective compounds devoid of such effects (Spieser et al. 1998). High numbers of A. schlechtendali cause browning of leaf undersides and early defoliation (Easterbrook and Fuller 1986), which result in a reduced CO2 exchange and transpiration rate, and negatively affect yield, fruit quality, tree growth and flower formation (Spieser et al. 1998). In addition, feeding A. schlechtendali can initiate russet formation, rendering fruits unmarketable (Easterbrook and Fuller 1986). The use of resistant cultivars is a part of an integrated pest management (IPM) system (Frei et al. 2005), combining chemical, biological and cultural control, and minimizing pesticide applications (Kellerhals et al. 2004). The potential influence of the apple cultivar on A. schlechtendali resistance has been reported by several studies (Downing and Moilliet 1967, Herbert 1974, Höhn and Höpli 1990, Easterbrook and Palmer 1996, Graf et al. 1998, Spieser et al. 1998, Duso et al. 2003). However, the resistance of these apple cultivars was scored by phenotypic evaluation, and no information on the genetic basis is yet available. Consequently, the use of such cultivars for resistance breeding in apple is limited. Phenotypic evaluation is either unreliable, or substantial resources are needed for additional, complex field trials (Brown and Maloney 2003, Francia et al. 2005). Genetic linkage maps allow the identification of quantitative trait loci (QTLs), which can indicate chromosomal regions controlling phenotypic traits (Collard et al. 2005). Such a linkage map should be densely covered with molecular markers, in order to obtain the maximum probability to identify a QTL (Silfverberg-Dilworth et al. 2006). Considering the apple genome, the saturation of linkage maps with molecular markers (AFLP, RAPD, SSR, SCAR markers) was strongly improved within the last years (Liebhard et al. 2003b, Silfverberg-Dilworth et al. 2006). Time and cost efficient molecular techniques, as for example multiplex-polymerase chain reaction (PCR)-based methods (Frey et al. 2004), accelerated the development of molecular markers. Knowledge about the underlying genetics of resistance will facilitate the breeding of resistant cultivars. Molecular markers can be used to select apple cultivars based on their genome (MAS) (Brown and Maloney 2003, Francia et

95 al. 2005), even at the seedling stage (Mohan et al. 1997a). Furthermore, MAS facilitates the combination of resistances to serious diseases and pests and of high fruit quality (Fischer 1994, Mohan et al. 1997a, Varshney et al. 2004), and it offers the possibility to pyramid two or more desirable resistance genes promoting durable resistance (Mohan et al. 1997a, Kellerhals et al. 2004). In the apple system, there was a focus on plant diseases, and QTL studies were carried out for resistance to apple scab [Venturia inaequalis (Cke.) Wint] (Liebhard et al. 2003c), mildew (Podosphaera leucotricha Elllis and Everh.) (Calenge and Durel 2006), and fire blight [Erwinia amylovora (Burrill) Winslow et al.] (Khan et al. 2006). Information on QTLs for tree growth and fruit quality traits is also available (Conner et al. 1998, King et al. 2000, Segura et al. 2006). Recently, some basic information about markers associated to pest resistance was provided for three aphid species, namely the leaf-curling aphid (Dysaphis devecta Wlk.), the rosy apple aphid (Dysaphis plantaginea Pass.) and the woolly apple aphid (Eriosoma lanigerum Hausm.) (Roche et al. 1997, Bus et al. 2008, Stoeckli et al. 2008c). Evidence for a genetic basis of pest resistance in apple was given for several other herbivore species, for example for the brownheaded leafroller (Ctenopseustis obliquana Wlk.) (Wearing et al. 2003), the European red mite (Panonychus ulmi Koch) (Goonewardene et al. 1982), but information on the genetic basis of arthropod resistance in apple is still scarce. No gene region for A. schlechtendali resistance is known so far, and the present study is the first report of a QTL analysis for A. schlechtendali resistance in apple. The aim of this study was to assess the genetic basis of resistance in apple to A. schlechtendali. Apple trees representing 160 different progeny genotypes were surveyed for rust mite infestation at each of three different study sites. Based on rust mite infestation, a

QTL analysis was carried out using linkage map data available for a segregating F1-cross of the apple cultivars 'Fiesta' x 'Discovery' (Liebhard et al. 2003b). Host-plant resistance to A. schlechtendali has been reported based on phenotypic evidence for 'Cox's Orange Pippin', the mother cultivar of 'Fiesta' (Easterbrook and Palmer 1996), which highlights the potential to identify QTLs for A. schlechtendali resistance in the 'Cox's Orange Pippin' pedigree. Effects of environmental variability on A. schlechtendali infestation, which may impede the detection of the genetic basis of resistance, were assessed by considering (i) climatic conditions, (ii) the relationship of A. schlechtendali to other herbivores, and (iii) the spatial variability of A. schlechtendali infestation at the different sites.

96 8.2. Materials and methods

Orchard characteristics and plant material Apple rust mite (Aculus schlechtendali) abundance on apple trees was surveyed in Switzerland at the sites Zurich (Wadenswil; at 47°13'20''N, 8°40'05''E, 455 m altitude), Valais (Conthey; at 46°12'30''N, 7°18'15''E, 478 m altitude), and Ticino (Cadenazzo; at 46°09'35''N, 8°56'00''E, 203 m altitude), during two consecutive years, 2005 and 2006 (= Year 1 and Year 2). Climate data from March to August during the two consecutive years, and standard climate values (1960-1990) were retrieved from MeteoSwiss (http://www.meteoschweiz.ch). Highest mean temperatures (March to August) were measured at the Ticino site (16.8°C), followed by the Valais (15.2°C) and the Zurich site (13.8°C) (supplementary material; Table SM 8.5.1). At the Valais site the measured sum of rainfall (March to August) (300-400 mm) was half of the amount at the other sites (700-800 mm). Temperature was 1-2 °C higher conferred to standard temperature values, and at the Ticino site the measured sum of rainfall was 60-70% of the standard value (Table SM 8.5.1).

The studied plants represent F1 progeny plants of the cultivars 'Fiesta' x 'Discovery'. They were bud grafted on M27 rootstocks in summer 1998 and planted in winter 1998/1999 at the three sites (Liebhard et al. 2003c). Tree-to-tree distance was 0.5 m (Zurich and Valais) and 1.25 m (Ticino), respectively. Rows were planted 3.5 m apart. The maximum number of genotypes present at all three sites was 160, and was lower for some analysis, as some trees died since plantation establishment. Orchards were treated with fertilizers and herbicides, but no insecticides, fungicides and acaricides were applied.

Assessment of mites and other herbivores In Year 1, the abundance of A. schlechtendali on each studied apple tree was quantified by (a) counting the number of all rusty leaves per tree and by (b) evaluating the number of mites per leaf extracted by filtration from a sample of 24 leaves per tree. In Year 2, only the filtration method was applied as the results of both methods were comparable. For the filtration method, the tree was divided into eight sectors (north, south, west, east; each bottom and top) and three leaves were sampled randomly from each sector to obtain a measure of mite abundance representing the whole tree and not only a tree part (Stoeckli et al. 2008b). Young leaves (the top 3-5 leaves of a shoot) were sampled as they generally show the highest A. schlechtendali infestation (Easterbrook and Palmer 1996). The leaves were suspended in 200 ml of a 0.1% Etalfix solution (surface-active agent; gvz-rossat, Otelfingen,

97 Switzerland). After five to seven hours, the Etalfix solution was filtrated, using a multibranch filter system consisting of three filter holder support bases (Sartorius AG, Biotechnology Division, Dietikon, Switzerland), equipped with 'Biosart 250' funnels (250 ml, polypropylene material) and cellulose nitrate filters (diameter: 46 mm, pore size: 8 µm, white with black lines; Sartorius AG, Switzerland). A. schlechtendali on the filters were counted with a binocular. To assess a possible relationship between the number of mites per leaf and the leaf area (sum of the 24 collected leaves per tree), leaves from the Ticino site in Year 1 were exemplarily photographed with a digital camera (Nikon, Coolpix 990) together with a reference area of 1 cm2, and total leaf area was determined using the software Adobe Photoshop CS2 for Mac OS X (following the method described in (Mody and Linsenmair 2004). To study the relationship between A. schlechtendali and different herbivore species, the abundance of three aphid and of two moth species was quantified for the same apple trees that were considered for rust mite assessments. The number of rosy apple aphid (Dysaphis plantaginea Pass.) colonies, the number of red-curled leaves caused by leaf-curling aphids (D. cf. devecta Wlk., species complex), and the number of green apple aphids (Aphis pomi De Geer), were counted 3-4 times from May to July at the three sites in the two consecutive years. The number of codling moth (Cydia pomonella L.) larval penetrations, and the number of mines caused by the apple leaf miner (Lyonetia clerkella L.), were assessed in July and August (C. pomonella) and July (L. clerkella), at the Ticino and Valais site (C. pomonella) and at all three sites (L. clerkella), in Year 1 (L. clerkella) or the two consecutive years (C. pomonella).

QTL analysis Abundance data of A. schlechtendali infestation were not normally distributed and a log10(x+1) transformation was applied to normalize error distribution. QTL analyses of the number of rusty leaves per tree and the number of mites per leaf were carried out separately for each site and year using the software MapQTL® 4.0 (van Ooijen et al. 2002). The genetic linkage maps for both 'Fiesta' and 'Discovery' (single parent maps), used in QTL analysis, were already published (Liebhard et al. 2003b). Kruskal-Wallis tests and interval mapping (IM) were used for QTL analysis. Logarithm of odds (LOD) threshold values were determined by 1,000-fold-permutation tests (MapQTL® 4.0) at a significance level of 95% (genome-wide) (King et al. 2000). The 2-LOD support interval was calculated to estimate the position of significant QTLs with 95% confidence (King et al. 2000). The proportion of

98 variation in mite infestation that can be explained by the genetic variation among the apple 2 progenies was analyzed by broad-sense heritability, which was estimated by the formula H = 2 2 2 2 2 2 2 σ g/σ p and σ p = (σ g + σ e/n), where σ g is the genetic variance, σ p is the phenotypic 2 variance, σ e is the environmental variance and n is the number of replicates per genotype (Lauter and Doebley 2002). The 'Fiesta' x 'Discovery' population was divided into subpopulations based upon the presence/absence of the marker closest to the detected QTL and the phenotypic difference was tested considering the two subpopulations (Mann-Whitney U-test). Possible QTL interactions were tested by multiple QTL mapping (MQM) for QTLs with LOD scores exceeding the LOD threshold values in IM. The SSR marker Hi03a10 (Silfverberg-Dilworth et al. 2006) was used to carry out a pedigree analysis of the identified QTL for rust mite resistance on the 'Fiesta' linkage group 7. PCR amplifications were performed in a 10 µl volume containing 5 µl of a DNA solution (1 ng/µl), 1x reaction buffer (Amersham Pharmacia, Dübendorf, Switzerland), 0.1 mM of each dNTP, 0.2 µM of dye- labelled forward primer and 0.2 µM of reverse primer, and 0.7 U of Taq Polymerase (Amersham Pharmacia, Dübendorf, Switzerland) per reaction. PCRs were performed in a Gene Amp PCR system 9600 (Perkin Elmer, Foster City, CA), and microsatellite fragment lengths were scored with Genotyper 3.6 (Applied Biosystems).

Data analysis Effects of genotype, site and year on Aculus schlechtendali infestation was assessed by a three factor mixed model ANOVA (number of mites per leaf), with year as within- subject effect, and genotype and site as between-subjects fixed effects. A one-way ANOVA was carried out to analyze the number of rusty leaves per tree, with genotype and site as fixed factors. Spearman's rank tests were applied to assess the relationship of (i) A. schlechtendali abundance between different sites and years, and of (ii) A. schlechtendali number on the two neighbor trees (sum) on mite number on the specific individual trees. Only those trees were included in this analysis that had direct neighbor trees (a dead or missing tree was not regarded as neighbor tree). When multiple correlation tests were carried out the Benjamini- Hochberg procedure was applied to correct for false discovery rates (type I errors) (Verhoeven et al. 2005). The distribution of A. schlechtendali on individual apple trees was analyzed by the index of dispersion (ID) (Southwood and Henderson 2000) using BiodiversityPro 1997 (Neil McAleece, P.J.D. Lambshead and G.L.J. Paterson; The Natural 2 History Museum, London). ID values significantly greater than the χ statistic (0.025

99 probability level) indicate an aggregated distribution of the studied species (Ludwig and Reynolds 1988), in our study of A. schlechtendali on individual trees (some trees were strongly infested, whereas other trees were not infested at all). Potential effects of the spatial position of trees in the study sites on A. schlechtendali infestation were inferred from analyses of spatial autocorrelation, computing Moran's I (Legendre and Legendre 1998) and corresponding z values (significance levels) using the software CrimeStat III (Levine 2007). Values of I greater than the expected I indicate clustering while values of I less than the expected I indicate dispersion. Potential spatial patterns of A. schlechtendali infestation within the orchard were visualized by fitting trend surfaces on contour plots by kriging (best unbiased generalized least squares estimation) with an exponential covariance function (Venables and Ripley 2002). All statistical analyses were performed with SPSS 16.0 for Mac OS X (SPSS, Inc., Chicago, IL) and R 2.6.0 (R Development Core Team, Vienna).

8.3. Results

Evaluation of rust mite abundance Highest numbers of apple rust mite (Aculus schlechtendali) per leaf were found at the Zurich site in Year 2 (mean: 6.2; Table 8.1), whereas lower infestation occurred at the Zurich site in Year 1 (mean: 0.6), at the Valais site (mean; both years: 0.6), and at the Ticino site (mean; Year 1: 1.4, Year 2: 0.3). The number of rusty leaves, caused by A. schlechtendali infestation in Year 1, was highest at the Zurich site (mean: 8.5; Table 8.1), compared to 5.6 and 5.4 at the Valais and Ticino site, respectively (Table 8.1). Maximal values showed that some highly infested trees occurred at the Ticino site (maximum number of mites per leaf: 84; Table 8.1). The number of mites per leaf and the number of rusty leaves per tree was significantly positively correlated (Spearman's rank test; Ticino: n = 143, rs = 0.817, P <

0.0001; Valais: n = 142, rs = 0.902, P < 0.0001; Zurich: n = 153, rs = 0.585, P < 0.0001). At the Ticino and the Valais site, the number of A. schlechtendali per leaf of the same trees was significantly correlated in the two consecutive years, but not at the Zurich site (Ticino: n =

143, rs = 0.232, P = 0.005; Valais: n = 142, rs = 0.180, P = 0.003; Zurich: n = 153, rs = 0.123, P = 0.131). For the same tree genotype, the number of A. schlechtendali per leaf and the number of rusty leaves were highly correlated in Year 1 among the Ticino and Valais sites, as well as among the Valais and Zurich sites, but not amongst the Ticino and Zurich sites (Table 8.2). No relation in the number of A. schlechtendali per leaf among the sites was detected in

100 Year 2 (Table 8.2). The number of mites per leaf and the leaf area of the 24 assessed leaves (sum) were not significantly correlated at the Ticino site in Year 1 (Spearman's rank test; n =

141, rs = -0.156, P = 0.065). A. schlechtendali infestation was significantly influenced by apple genotype and site (ANOVA; Table 8.3). Year, as a within-subject effect, was not significant for the number of mites per leaf, but there was a significant year x genotype and year x site interaction (Table 8.3). Broad-sense heritability (H2) for A. schlechtendali infestation was 19.9% (number of mites per leaf), and 46.2% (number of rusty leaves per tree, Table 8.4).

Table 8.1. Aculus schlechtendali infestation of progeny plants of the cross 'Fiesta' x 'Discovery' at three sites in two consecutive years. Number of studied genotypes (n), mean infestation ± standard error (SE), maximum value (Max) and incidence (I = % infested trees) are presented. Mean and maximal values refer to the number of mites per leaf (extraction of 24 leaves per tree) in August in Year 1 and Year 2, and to the number of rusty leaves per tree in August in Year 1.

Ticino Valais Zurich n Mean ± SE I n Mean ± SE I n Mean ± SE I (Max) (Max) (Max)

No. mites per leaf 1.4 ± 0.6 0.6 ± 0.1 0.6 ± 0.1 Year 1 143 55 142 38 153 61 (84) (6) (13) 0.3 ± 0.1 0.6 ± 0.2 6.2 ± 0.8 Year 2 143 19 142 26 153 73 (16) (15) (47)

No. rusty leaves 5.4 ± 1.1 5.6 ± 0.8 8.5 ± 0.9 Year 1 149 (83) 41 148 (51) 45 153 (54) 67

101 Table 8.2. Comparison of Aculus schlechtendali infestation on individual apple genotypes at three sites assessed by Spearman's rank tests. Significant correlations after FDR correction are highlighted.

Correlation Correlation Correlation Ticino-Valais Ticino-Zurich Valais-Zurich rs P rs P rs P

No. mites per leaf Year 1 0.315 <0.0001 0.026 0.758 0.259 0.002 Year 2 0.079 0.374 -0.003 0.975 -0.05 0.558 n 129 138 137

No. rusty leaves Year 1 0.395 <0.0001 -0.082 0.326 0.359 <0.0001 n 140 144 143

Table 8.3. Effects of genotype, site and year on Aculus schlechtendali infestation. Evaluation of the number of mites per leaf assessed by a mixed model ANOVA and the number of rusty leaves per tree by one-way ANOVA.

d.f. Mean square F value P value

No. mites per leaf Within-subjects effects Year 1 0.210 0.463 0.497 Year x genotype 157 0.548 1.205 <0.0001 Year x site 2 24.665 54.252 <0.0001 Error 278 0.455 Between-subjects effects Genotype 157 0.682 1.249 0.050 Site 2 34.661 63.436 <0.0001 Error 278 0.546

No. rusty leaves Genotype 157 0.417 1.865 <0.0001 Site 2 3.896 17.418 <0.0001 Error 290 0.224

102 Table 8.4. Broad-sense heritability (H2) for Aculus schlechtendali infestation. Calculation of genotypic 2 2 variance (σ g) and phenotypic variance (σ p) based on mean square ANOVA results (cf. Table 8.3).

2 2 2 Variance components σ g σ p H

No. mites per leaf 0.045 0.227 0.199

No. rusty leaves 0.064 0.139 0.462

QTLs for rust mite resistance Two significant QTLs were identified for A. schlechtendali resistance in apple on the 'Fiesta' linkage group 7 at the Zurich site in Year 1 (Figure 8.1, Table 8.5). MQM mapping (multiple QTL mapping; data not shown) did not reveal any multiple linked QTLs and results were therefore based on IM. The closest markers to these QTLs were the AFLP marker E35M42-0146 at 20.2 cM and the RAPD marker AE10-400 at 45.8 cM (Table 8.5). The significant LOD scores at the marker positions were 4.3 (E35M42-0146) and 3.1 (AE10-400) for the number of mites per leaf (Zurich Year 1). For the number of rusty leaves per tree, the significant LOD scores at the marker positions were 6.0 (E35M42-0146) and 4.7 (AE10-400) (Zurich). The phenotypic variation explained by these markers ranged between 11-12% (number of mites per leaf) and 16-17% (number of rusty leaves per tree) for significant LOD scores (Table 8.5). The 95% confidence interval (2-LOD support interval) ranged from map position 15-25 cM (E35M42-0146) and 28-47 cM (AE10-400) (Figure 8.1, Zurich Year 1). A lower A. schlechtendali infestation was found for the apple subpopulation amplifying the AFLP marker E35M42-0146 or the RAPD marker AE10-400 compared to apple genotypes not amplifying one of the markers in 14 of 18 surveys (Table 8.5). In the case of a significant QTL (Zurich), the difference was significant (Mann-Whitney, P < 0.05; Table 8.5). A combined analysis of the AFLP marker E35M42-0146 or the RAPD marker AE10-400 did not reveal an interaction between the two markers. A lower A. schlechtendali infestation of the 'Fiesta' x 'Discovery' progeny amplifying the AFLP marker E35M42-0146 was found independently of the presence/absence of the RAPD marker AE10-400 (Table 8.6; Zurich Year 1). Similarly, A. schlechtendali infestation was lower or equal on genotypes amplifying AE10-400 compared to trees not amplifying AE10-400, independent of the presence/absence of E35M42-0146. The effect of the AFLP marker E35M42-0146 was stronger than the effect

103 of the RAPD marker AE10-400 as differences between presence/absence of this marker are higher compared to the latter (Table 8.6). The origin of the resistance QTL (marker E35M42- 0146 at 20.2 cM) was followed in the pedigree of 'Fiesta'. The allele '240 bp' of the SSR marker Hi03a10 (Silfverberg-Dilworth et al. 2006), which is closely located and in coupling to the AFLP marker E35M42-0146 (5.8 cM distance between Hi03a10 and E35M42-0146) has been inherited from 'Wagener' to 'Idared' (Figure 8.2). We additionally identified QTLs that were significant for one survey method, at one site, and in one of the two consecutive years (data not shown). The ALFP marker E35M41- 0148 on linkage group 5 at 63.8 cM of the 'Discovery' chromosome was closest positioned to a QTL that was significant for the number of mites per leaf at the Valais site in Year 2. The QTL had a LOD score of 3.1 and explained 9.5% of the phenotypic variability. Also on the 'Discovery' chromosome, the RAPD marker C05-1000 on linkage group 4 at 35.8 cM, and the allele '112 bp' of the SSR marker CH02g09 on linkage group 8 at 33.5 cM, were closest positioned to QTLs that were significant for the number of rusty leaves at the Zurich site in Year 1. The LOD scores of the QTLs were 3.2 and 2.0, and the phenotypic variability explained was 9.0% and 5.9%.

Spatial distribution of rust mites

The calculation of the index of dispersion (ID) showed that rust mite infestation on individual trees was significantly aggregated (P < 0.0001; Table SM 8.5.2). Some trees were strongly infested, whereas other trees were not infested at all, but there was no significant spatial pattern of A. schlechtendali infestation assessed by Moran's I (Table 8.7). This finding was in line with a visualization of the position of trees infested with A. schlechtendali (Figure SM 8.5.1). No correlation between mite infestation on an individual tree and mite infestation on the two neighboring trees was found, neither for the number of rusty leaves nor for the number of mites per leaf (Table SM 8.5.3).

104 Table 8.5. QTLs identified for Aculus schlechtendali infestation in a segregating 'Fiesta' x 'Discovery' population on linkage group 7 of 'Fiesta'. QTL analysis was carried out for each year and site separately. Site and Year, genetic locus (locus in cM), closest marker, linkage phase, LOD score and threshold level at the locus of the closest marker, and phenotypic variance explained (PVE in %) based on IM are presented. Mean A. schlechtendali infestation for the two subpopulations of the 'Fiesta' x 'Discovery' progeny divided based upon the presence (pres) and absence (abs) of the nearest markers linked to the identified QTL on the 'Fiesta' chromosome. Significant LOD scores are highlighted.

Closest marker LOD score Mean infestation Site and Year Locus PVE c a (threshold) b (marker pres/abs) d Number of mites per leaf 20.2 E35M42-0146 (+) Ticino Year 1 0.8 (2.5) 2.9 0.7 / 1.0 n.s. Year 2 0.7 (2.4) 2.5 0.2 / 0.3 n.s. Valais Year 1 1.3 (3.0) 4.1 0.5 / 0.8 * Year 2 0.3 (3.4) 0.7 0.8 / 0.4 n.s. Zurich Year 1 4.3 (2.6) 12.3 0.5 / 0.9 ***** Year 2 0.2 (4.5) 0.4 6.2 / 6.3 n.s. 45.8 AE10-400 (-) Ticino Year 1 0.8 (2.5) 7.9 1.2 / 0.8 n.s. Year 2 0.7 (2.4) 6.7 0.7 / 0.5 n.s. Valais Year 1 1.1 (3.0) 6.4 0.6 / 0.6 n.s. Year 2 1.2 (3.4) 9.1 1.3 / 0.4 n.s Zurich Year 1 3.1 (2.6) 11.0 0.4 / 0.8 *** Year 2 0.8 (4.5) 2.9 4.0 / 6.6 n.s. Number of rusty leaves 20.2 E35M42-0146 (+) Ticino 0.7 (2.5) 2.2 4.6 / 6.0 n.s. Valais 1.6 (3.4). 4.8 4.0 / 7.2 **** Zurich 6.0 (3.2) 16.6 4.7 / 12.1 ***** 45.8 AE10-400 (-) Ticino 0.8 (2.5) 6.0 5.0 / 6.8 n.s. Valais 1.5 (3.4) 5.7 5.0 / 6.0 * Zurich 4.7 (3.2) 15.9 4.5 / 10.7 **** a Molecular marker closest to the likelihood peak of each QTL. Linkage phase information is provided as (+) or (-), indicating on which of the homologous chromosomes the marker is located. b LOD (logarithm of odds ratio) score and LOD threshold at the position of the closest marker. LOD threshold levels were derived by 1,000-fold permutation tests (genome-wide) c Phenotypic variance explained by the QTL. d Mean A. schlechtendali infestation for the two subpopulations of the 'Fiesta' x 'Discovery' progeny divided based upon the presence/absence of the nearest markers linked to a QTL on the 'Fiesta' chromosome . Different letters indicate significant differences between subpopulations (Mann-Whitney U-test). *: P < 0.05; **: P < 0.01; ***: P < 0.005; ****: P < 0.001; *****: P < 0.0001. Sample size for the marker E35M42-0146 varied between 70-76 (pres) and 69-74 (abs). For the marker AE10-400 sample size varied between 46-52 (pres) and 53-56 (abs).

105

Figure 8.1. QTLs for resistance of apple to Aculus schlechtendali identified on linkage group 7 of 'Fiesta' based on IM results. The x-axis indicates the linkage map of 'Fiesta' in cM and the marker names; the y-axis shows the LOD scores. The solid black bar indicates the 2-LOD support interval for the position of the QTL (Zurich Year 1). Log10(x+1) transformed data were used for QTL analysis. The markers closest positioned to the QTLs are underlined. LOD threshold levels at the Zurich site for the number of mites per leaf were 2.6 (Year 1) and 4.5 (Year 2), and for the number of rusty leaves the threshold level was 3.2 (Year 1).

106 Table 8.6. Combined analysis of the two markers at the peak of the QTLs E35M42-0146 and AE10-400 ('Fiesta') that where significantly linked to Aculus schlechtendali resistance at the Zurich site in Year 1. The phenotypic trait (number of mites per leaf and number of rusty leaves) was divided in four subpopulations based on the presence and absence of the markers, and average values (log10(x+1)-transformed) were analyzed.

No. mites No. rusty AE10-400 AE10-400 per leaf leaves

presence absence presence absence

0.39 ± 0.10 0.38 ± 0.13 0.39 ± 0.07 0.51 ± 0.15

0146 0146 - - presence presence

0.63 ± 0.18 0.92 ± 0.11 0.67 ± 0.17 0.89 ± 0.08

E35M42 E35M42

absence absence

Figure 8.2. Analysis of the pedigree of the apple variety 'Fiesta' with the SSR Hi03a10 associated with the QTLs for resistance to Aculus schlechtendali. The SSR marker allele associated with resistance is underlined (allele '240 bp' for Hi03a10).

107 Table 8.7. Spatial distribution pattern of Aculus schlechtendali infestation at the Ticino, Valais and Zurich site in two consecutive years (mean values) assessed by Moran's I. Values of I greater than a randomly expected I indicate clustering while smaller values of I indicate dispersion. The z values were compared to a standard normal table, and absolute values greater than 1.96 indicate a spatial autocorrelation at a 5% significance level. Significant z-values are highlighted. Sample size was n = 143 for the Ticino, n = 142 for the Valais, and n = 153 for the Zurich site.

I randomly Normality Randomization Moran's I expected (SD) significance (z) significance (z)

No. rusty leaves Ticino -0.012 -0.007 (0.012) -0.398 -0.412 Valais 0.009 -0.007 (0.022) 0.704 0.713 Zurich 0.032 -0.007 (0.019) 1.955 1.981 No. mites per leaf Ticino Year 1 -0.004 -0.007 (0.013) 0.262 0.375 Year 2 -0.011 -0.007 (0.013) -0.338 -0.409 Valais Year 1 0.005 -0.007 (0.024) 0.523 0.532 Year 2 -0.022 -0.007 (0.024) -0.619 -0.676 Zurich Year 1 0.014 -0.007 (0.019) 1.100 1.211 Year 2 0.001 -0.007 (0.019) 0.400 0.405

Relationship between rust mite abundance and co-occurring herbivore species A significant positive relationship was found between the number of A. schlechtendali per leaf and the number of A. pomi per tree at the Ticino site in Year 1 (Spearman's rank test,

Benjamini-Hochberg procedure; n = 143, rs = 0.284, P = 0.001), and at the Zurich site in Year

1 and Year 2 (Year 1: n = 153, rs =0.222, P = 0.006; Year 2: n = 153, rs = 0.200, P = 0.005). A significant positive relationship between A. schlechtendali and A. pomi was additionally found for the number of rusty leaves per tree as measure for mite infestation in Year 1

(Ticino: n = 143, rs = 0.270, P = 0.001; Zurich: n = 153, rs = 0.204, P = 0.011). No significant relationship between infestation of the apple genotypes by A. schlechtendali and by the other surveyed herbivores D. plantaginea, D. cf. devecta, C. pomonella, and L. clerkella was identified (P > 0.05 for all tests).

108 8.4. Discussion

The purpose of this study was to investigate the resistance of apple (Malus x domestica) to the apple rust mite (Aculus schlechtendali). Interactions between A. schlechtendali and other herbivores, climatic conditions at the study sites and the spatial variability of A. schlechtendali infestation were assessed to elucidate QTL effects. We identified two significant QTLs associated with A. schlechtendali resistance on the 'Fiesta' linkage group 7. The AFLP marker E35M42-0146 at 20.2 cM and the RAPD marker AE10- 400 at 45.8 cM were closest positioned to the QTLs. A significantly lower number of mites per leaf and a lower number of rusty leaves per tree were found for apple genotypes amplifying the AFLP marker E35M42-0146 compared to apple genotypes not amplifying the markers at the Zurich site in Year 1 and at the Valais site in Year 2. Referring to the second marker on the 'Fiesta' linkage group 7, the RAPD marker AE10-400, the linkage map at this region is not well saturated with molecular markers. QTL significance may disappear when including more markers in the QTL analysis. We did not find an interaction between the two markers and they seem to be linked, as 70% of the apple genotypes amplifying for E35M42- 0146 additionally amplify the marker AE10-400. These findings may partly explain the QTL at the region of the RAPD marker AE10-400. Broad-sense heritability (H2) amounted to 19.9% (number of mites per leaf), and 46.2% (number of rusty leaves), respectively. While some putative QTLs for resistance to A. schlechtendali were also found on the 'Discovery' linkage groups 4, 5 and 8, the linkage group 7 of 'Fiesta' appears to be strongly related to resistance against different pests and diseases. Besides the two newly identified QTLs associated to resistance to A. schlechtendali, a QTL for fire blight resistance (Calenge et al. 2005, Khan et al. 2006) and one for D. cf. devecta resistance (Roche et al. 1997, Stoeckli et al. 2008c) were previously identified on this linkage group. The specific gene regions of these QTLs are probably not overlapping. The fire blight QTL is positioned on the bottom of the linkage group (46.5-51.5 cM), whereas the QTL for D. cf. devecta resistance was identified at its top (0-5 cM). The genetic basis of apple resistance to A. schlechtendali is underlined by the finding that rust mite infestation of individual trees representing unique genotypes was highly aggregated (ID, Table SM 8.5.2), and that genotype was a significant factor explaining the number of rusty leaves per tree (ANOVA; Table 8.3). Furthermore, we detected a significant correlation of A. schlechtendali infestation of individual apple genotypes between the sites Ticino and Valais, as well as between the sites Valais and Zurich in Year 1. In combination

109 with the identified QTLs, these phenotypic findings provide further evidence for the genetic basis of rust mite resistance in apple, which is supported by results from other studies showing that rust mite infestation may vary on different apple cultivars (e.g. Graf et al. 1998; Spieser et al. 1998; Duso et al. 2003). A number of apple cultivars described as resistant to A. schlechtendali, such as 'Cox's Orange Pippin' (Easterbrook and Palmer 1996), 'Florina' (Graf et al. 1998), 'Glockenapfel' (Höhn and Höpli 1990), 'Golden Delicious' (Herbert 1974, Spieser et al. 1998), 'Jonagold' (Höhn and Höpli 1990), 'McInosh' (Downing and Moilliet 1967), 'N.Y. 18491' (Duso et al. 2003) or 'Red Delicious' (Downing and Moilliet 1967) (Table SM 8.5.4), are highly promising to be tested for the SSR marker allele associated with the described QTL (allele '240 bp' of Hi03a10 on 'Fiesta' linkage group 7). However, the pedigree analysis of 'Fiesta' revealed that the SSR marker allele was inherited from the cultivar 'Wagener' to 'Idared', but was not present in 'Cox's Orange Pippin'. Host-plant resistance to A. schlechtendali has not yet been reported for the cultivars 'Wagener' and 'Idared', and phenotypic field surveys are therefore highly encouraged. The phenotypically observed resistance of 'Cox's Orange Pippin' (Easterbrook and Palmer 1996) may be based on genetic factors, or its expression may be strongly affected by environmental conditions. The identified QTL for the number of mites per leaf was significant for the Zurich site in Year 1 and the Valais site in Year 2, but it was not significant for the second year and the Ticino site. This observation may reflect environmental variability, which may partly explain different infestation patterns and the instability of QTLs among sites (Walde et al. 1997). The potential influence of environmental variability on rust mite infestation is supported by the finding of a lacking correlation of A. schlechtendali infestation of individual apple genotypes amongst the Ticino and Zurich sites in Year 1. In general, all three sites differed markedly in climate conditions. Temperature at the Ticino and the Valais site was higher compared to the Zurich site, and field observations revealed that tree phenology in the Zurich orchard was approximately two weeks delayed compared to the Ticino site. Besides climate, other environmental factors such as the composition of the orchard fauna may affect mite distribution and mask the effects of host tree genotype. Host-tree infestation by aphids, for example, affects leaf growth und thus probably changes habitat and resources for rust mites. There was a significant positive correlation between infestation by A. schlechtendali and the aphid A. pomi at the Ticino site in Year 1, and at the Zurich site in Year 1 and Year 2. A. pomi infestation was higher at the Ticino and Valais sites than at the Zurich site and may have contributed to instable QTL effects considering sites. Although high herbivore infestation of

110 an individual tree may serve as a source of infestation by the same herbivore of the neighbor trees (neighborhood effect), there was no correlation between mite abundance on an individual apple tree and mite abundance on the two neighbor trees. Similarly there was no significant effect of the spatial position of trees in the study plots on distribution of A. schlechtendali. Therefore neighborhood effects or spatial autocorrelation can be ruled out as explanation for the QTLs in different environments and years. The apparent influence of environmental conditions on the variable and partly weak expression of the described QTLs has to be taken into account when considering these QTLs for breeding programs or orchard management decisions. The stability of the QTLs in different genetic backgrounds and under different environmental conditions should be evaluated, and further molecular markers should be developed to saturate the described QTL regions. Consideration of SSR markers that are closely positioned to the QTLs will enhance the reliability and efficiency of marker assisted selection (MAS). Besides a focus on the genetic basis of A. schlechtendali resistance in apple, knowledge about biotic and abiotic factors related to mite resistance will facilitate rust mite control in apple orchards. It is already known that pest management strategies should not solely rely on resistant cultivars developed by MAS, and field evaluations are necessary to receive environment-specific results (Mohan et al. 1997a, Francia et al. 2005). The complexity of breeding pest resistant cultivars is not only due to the mentioned environmental factors but also to the usually quantitative genetic background of arthropod resistance. Nonetheless, SSR marker alleles such as allele '240 bp' of the SSR marker Hi03a10 that are closely positioned and in coupling to the identified QTLs may be used as a starting point to screen existing apple cultivars for resistance to A. schlechtendali and to identify resistant parents that can be used in MAS to develop new resistant apple cultivars.

111 8.5. Supplementary material (SM)

Table SM 8.5.1. Climatic characterization of the three study sites during the observation period (March - August) in the two years of survey, and comparison to standard values a.

Mean ± SE temperature Deviation from standard Total rainfall Rainfall %

(°C) (°C) (mm) standard

Ticino Year 1 16.8 ± 0.1 +1.7 679 66 Year 2 16.7 ± 0.1 +1.6 720 70 Valais Year 1 15.2 ± 0.1 +1.5 305 109 Year 2 14.9 ± 0.1 +1.2 406 145 Zurich Year 1 13.8 ± 0.1 +1.2 855 109 Year 2 13.6 ± 0.1 +1.0 854 109 a Standard values are based on 30-year average climatic data (1960 to 1990).

Table SM 8.5.2. Distribution pattern of Aculus schlechtendali on apple trees 2 characterized by index of dispersion (ID). All ID values significantly greater than the χ statistic (0.025 probability level) with (ν-1) degrees of freedom indicate an aggregated

distribution of A. schlechtendali. All ID values were significant (P < 0.0001).

ID values No. mites per leaf No. rusty leaves

Ticino Year 1 129,469 4,727 Year 2 31,864 - Valais Year 1 7,906 2,660 Year 2 28,714 - Zurich Year 1 12,305 2,127 Year 2 57,665 -

112 Table SM 8.5.3. Effects of neighborhood on Aculus schlechtendali infestation of individual apple trees at three study sites assessed by Spearman's rank tests comparing mite infestation on a specific tree and the infestation of the two neighbor trees.

Ticino Valais Zurich

n rs P n rs P n rs P

No. rusty leaves Year 1 57 -0.154 0.251 58 -0.080 0.551 64 0.181 0.152

No. mites per leaf Year 1 52 -0.229 0.103 54 -0.037 0.789 64 0.195 0.121 Year 2 52 0.005 0.071 54 -0.226 0.101 64 0.205 0.104

Table SM 8.5.4. Apple varieties with reported resistance properties to Aculus schlechtendali.

Cultivar Reference

Cox's Orange Pippin Easterbrook and Palmer 1996

Florina Graf et al. 1998 Glockenapfel Höhn and Höpli 1990 Golden Delicious Herbert 1974, Spieser et al. 1998 Jonagold Höhn and Höpli 1990 McIntosh Downing and Moilliet 1967 N.Y. 18491 Duso et al. 2003 Red Delicious Downing and Moilliet 1967

113

Figure SM 8.5.1. Spatial distribution of Aculus schlechtendali described by fitting trend surfaces on contour plots by kriging (best unbiased generalized least squares estimation) with an exponential covariance function. Mean rust mite infestation of different observations within a year was considered for contour plots (cf. Table 8.1). Infestation level of individual trees is reflected by circle size. Yellow- colored circles indicate trees with an infestation greater than zero, whereas uninfested trees are represented by open circles. No contour lines were plotted for the Valais site due to the orchard design with two rows.

114 9. General discussion

When eating an apple, do you consider the enormous effort invested by the grower and that a tasty apple does not simply grow by itself? The production of fruit with the desired quality starts by selecting an appropriate cultivar and planting system. The various orchard and tree management systems have to be scheduled, and finally post-harvest handling techniques have to be evaluated. Orchard-specific environmental conditions, such as temperature, as well as the availability of water, light, and nutrients all influence fruit set and quality. Arthropod pests cause serious yield losses by feeding on the various structures of an apple tree and by transmitting plant pathogens. Besides producing marketable fruit of high quality, pest control is the main concern when growing apple trees. The presented thesis focused on molecular approaches in host-plant resistance (HPR). The genetic basis of resistance in apple (Malus x domestica Borkh.) to common herbivore pests, such as the codling moth (Cydia pomonella L.), the apple leaf miner (Lyonetia clerkella L.), the rosy apple aphid (Dysaphis plantaginea Pass.), the leaf-curling aphid species complex (Dysaphis cf. devecta Wlk.), the green apple aphid (Aphis pomi De Geer), and the rust mite (Aculus schlechtendali Nalepa), was investigated by quantitative trait loci (QTLs) analysis. The evaluation was based on available linkage maps for the apple varieties 'Fiesta' and 'Discovery' and on phenotypic field evaluation of the mentioned herbivores. Molecular markers are a prerequisite to select pest-resistant apple cultivars in an early stage. Herbivore resistance varied greatly among progeny plants, and the QTL analysis resulted in a set of molecular markers linked to herbivore resistance/susceptibility (Chapters 4, 6, 8). In a field experiment, a specific type of plant resistance, namely resistance by antibiosis, was elucidated for the aphid A. pomi (Chapter 7). Population development in sleeve cages was assessed and the results demonstrated a strong influence of shoot growth, rather than a genetically based resistance. The study indicated that environment and plant related factors, such as climate, co- occurring herbivores, spatial pattern of herbivores, plant growth and fruit traits may impede QTL identification. Environmental factors such as microclimate, and tree characteristics such as phenology, affect within-tree oviposition preference of C.pomonella (Chapter 5). These findings are important for future monitoring systems focusing on female moths and for elucidating HPR in apple.

115 9.1. Host-plant resistance

There is a discrepancy between efficient agriculture and the risk of pest infestation (Risch 1987). Simplified planting systems and commercial practices often have a negative impact on natural enemies, and seem to raise populations of harmful arthropods. Monoculture plantings are primary drivers for pest outbreaks, mainly due to the unlimited availability of food (Hastings 1999). Furthermore, the elimination of small-scale structures such as hedges or wild.f.lower strips deplete the occurrence of natural enemies (Kim and McPheron 1993). This loss of natural biodiversity in plants and animals has caused high genetic uniformity in cultivated crop plants. As a result of this genetic uniformity the sources of host-plant resistance (HPR) in modern cultivars is limited (Schoonhoven et al. 1998). Moreover modern cultivars have lost their natural resistance due to selection for quality (e.g. sugar content) and high yield (Gardiner et al. 2007). The chance of introducing a plant species is higher nowadays. Introduced crop plants are not adapted to the native fauna, and are immediately invaded by arthropods (de Jong and Nielsen 2002). Furthermore, the single reliance on chemical control increased pesticide resistance in pests (Pimentel 1997, Reyes et al. 2007). In IPM, HPR is used as primary tactic to control severe pests observable in economically significant levels, such as resistant wheat varieties to the Hessian fly (Mayetiola destructor Say) (Pedigo 2006). HPR is an environmental friendly and long-lasting tactic, and easy to apply. HPR is compatible with chemical control and usually does not negatively affect natural enemies. Most concerns deal with resistance breaking pest biotypes, the difficulty to identify resistance genes, and plant traits causing resistance to one pest rendering the plant susceptible to another pest (Kogan 1998). For the green leafhopper (Nephotettix virescens Distant) damaging rice (Oryza sativa L.) plants, there was a synergistic effect of HPR and insecticides, resulting in an increased grain yield (Heinrichs et al. 1986). Insecticide treatment in O. sativa may even become unnecessary, as was observed for the brown planthopper (Nilaparvata lugens Stal) (Aquino and Heinrichs 1979). This positive effect may be explained by the minimization of pesticide applications otherwise toxic to natural enemies. Resistant barley (Hordeum vulgare L.) cultivars made the parasitoid wasp (Lysiphlebus testaceipes Cresson) more effective against the greenbug (Schizaphis graminum Rondani) compared to susceptible plants (Starks et al. 1972). There is an indication that even very low levels of resistance increase the efficiency of natural enemies (van Emden 1997). The increase of natural enemy efficiency may be explained by a prolonged susceptible stage of the pests on resistant host-plants (Velten et al. 2007). However, in the case pest density is seriously

116 reduced by HPR, natural enemies cannot find enough food (negative effect) (Sadasivam and Thayumanavan 2003). Furthermore, resistance in some crop plants may induce herbivore enzymes involved in the detoxification of insecticides. The 2-tridecanone from wild tomato (Lycopersicon esculentum Mill.) induces tolerance in corn earworm (Helicoverpa zea Boddie) to the insecticide carbaryl (Kennedy 1984). Furthermore, secondary metabolites from herbivore resistant plants may impede natural enemies of the herbivore. The alkaloid α- tomatine, an antibiotic factor in tomato (L. esculentum) to lepidopterans, is toxic to the endoparasitoid Hyposoter exiguae (Viereck) (Duffey and Bloem 1986). In this case, natural enemies can be favored by cultural control, such as building hedges and wildflower strips (Sadasivam and Thayumanavan 2003). An integration of HPR, cultural practice and chemical control is made by planting trap crops adjacent to resistant varieties. Planting early maturing soybean (Glycine max L.) varieties attracted bean leaf beetle (Cerotoma trifurcate Forster) populations. The trapped population is chemically killed, and often the field planted with a resistant variety requires no additional treatment afterwards (Kogan 1994).

9.2. Genetically based resistance in apple to herbivore pests

The present thesis is the first report evaluating genetically based resistance in apple to the common apple pests C. pomonella, L. clerkella, D. plantaginea, D. cf. devecta, A. pomi, and A. schlechtendali by QTL analysis. The existence of genes on the apple genome associated with arthropod resistance was assumed for moths, such as C. pomonella (Maindonald et al. 2001) or the brownheaded leafroller (Ctenopseustis obliquana) (Wearing et al. 2003), to various aphids (Alston and Briggs 1977, Habekuss et al. 2000, Angeli and Simoni 2006), and to the European red mite (Panonychus ulmi) (Goonewardene et al. 1982). These observations are now supported by results of the present thesis, showing that that the genotype significantly influenced the abundance of C. pomonella, D. plantaginea, D. cf. devecta, and A. schlechtendali, that the infestation level was highly variable among the progeny plants, and that for these species a QTL associated with herbivore infestation was identified. Conferring to D. cf. devecta resistance, a linkage analysis revealed molecular markers at the top of the 'Fiesta' linkage group 7 (Roche et al. 1997, Cevik and King 2002a). The present thesis confirmed this locus in another genetic background by QTL analysis and provided molecular markers for MAS. Molecular markers linked to a QTL for resistance were positioned on 'Fiesta' linkage group 7 (D. cf. devecta, A. schlechtendali) and 17 (D.

117 plantaginea). The molecular marker on 'Discovery' linkage group 10 was associated with increased susceptibility to C. pomonella. The identified molecular markers provide a basis for selecting apple seedlings in an early stage by marker assisted selection (MAS) and evaluating traits for which a phenotypic scoring is extremely difficult and expensive (e.g. agronomic traits) (Mohan et al. 1997a). The segregation of important genes among varieties can be monitored, and genes from related wild varieties with described resistance may be transferred by genetic engineering (Tanksley et al. 1989). Heterozygous individuals can easily be distinguished from homozygous individuals, which is not always possible by phenotypic evaluation (Dirlewanger et al. 1998). Furthermore, a selection may be practiced for several traits simultaneously (Varshney et al. 2004). MAS can be used to pyramid genes (Servin et al. 2004), such as breeding apple cultivars with multiple (Fischer 1994) or durable (Kellerhals and Furrer 1994) resistance. Many breeding projects were initiated in the last decade and the many identified molecular markers associated with fruit and resistance traits resulted in a variety of germplasm databanks. Such databanks can be screened for new sources of resistance with the identified molecular markers (Gardiner et al. 2007). In general the QTLs explained only a small fraction of the phenotypic variability, indicating that herbivore resistance is under polygenic control, as supposed for many insect pests (Gessler and Patocchi 2007). The complex process of host-finding may have hampered the evolution of major resistance genes. Major resistance genes appear to be more common for resistance in apple to diseases such as apple scab (Patocchi et al. 2005). A partial resistance has the advantage of a reduced pressure on biotypes overcoming resistance. Minor genes are of high value regarding gene pyramiding with the aim of breeding apple cultivars with multiple and durable pest resistance (Mohan et al. 1997b). HPR is based on morphological and chemical plant traits. A relationship between high leaf trichome density (Webster et al. 1994, Elden 1997, Shockley et al. 2002), thickness of surface waxes (Stoner 1990), or thickness of plant tissues (Quiros et al. 1977, Mensah and Madden 1991, Beeghly et al. 1997) and herbivore resistance is described. Phenolics and flavonoids are secondary metabolites affecting plant acceptability by herbivores (Takemura et al. 2002, van Loon et al. 2002). Conferring to the studied herbivores, several morphological and chemical plant traits were related to C. pomonella infestation. Lignification of cells, preventing larval entrance into the fruits, is supposed as basis for resistance (Westigard et al. 1975). Apple selections having a pubescent lower leaf surface had significantly reduced number of larval entries (Plourde et al. 1985). High pubescence and wax structure is

118 mentioned to influence oviposition of adult females (Hagley et al. 1980). Seedless apple fruits lead to a lower C. pomonella damage (Goonewardene et al. 1984). However, no association to fruit color and C. pomonella infestation was found (Cossentine and Madsen 1980). The role of host-plant chemicals has been studied intensively for C. pomonella. The attractant ester butyl hexanoate (Hern and Dorn 2004), the primary metabolites fructose and sorbitols (Lombarkia and Derridj 2002), and the secondary metabolite (E, E)-α-farnesene (Sutherland et al. 1977, Hern and Dorn 1999) all modify host searching behavior of C. pomonella. Phloridzin mediates apple resistance to A. pomi (Montgomery and Arn 1974). For a better understanding of HPR, it would be essential discover the function of the identified gene regions associated with herbivore resistance.

9.3. Environmental factors influencing the expression of resistance

Abiotic factors, such as temperature, light intensity, relative humidity, CO2 level, soil fertility, soil moisture, and agrochemicals can alter physiology and influence the expression of HPR (Tingey and Singh 1980). Further biotic factors related to the plant (age, density, height, tissue age, phenology, disease infection) or to the arthropod (age, gender, density, duration, activity period, biotypes) influence the expression of HPR (Smith 2005a). Such biotic factors are likewise influenced by environmental conditions. The presented results revealed that herbivore infestation significantly varied between sites and years. The influence of temperature, rainfall and wind, as well as co-occurring herbivore fauna and spatial distribution of herbivores, was evaluated and discussed. Abiotic and biotic factors mediating the expression of HPR are crucial for the understanding of insect-plant relationships. Furthermore, an accurate monitoring in IPM, fertilizer application, planting system, pest control tactics, and fruit-thinning method may be adjusted based on environmental conditions. Different morphological and chemical plant traits may affect the host-finding process as described before. Such factors have a high dependency on seasonal and environmental conditions (Volz et al. 2003, Vallat and Dorn 2005), and should therefore be evaluated before drawing conclusions.

119 9.4. Perspectives in apple breeding

The latest trend in apple breeding is to focus on the target gene, rather than to rely on molecular markers. Sequencing the target gene will prevent inaccurate selection due to recombination between marker and gene, and therefore enhance the effectiveness of MAS (Gardiner et al. 2007). The different steps for this evaluation are summarized as map-based cloning. The molecular markers identified in the present thesis associated with herbivore resistance are a prerequisite for map-based cloning. Firstly, the specific gene regions need to be saturated with molecular markers (high-density maps). This aims at identifying molecular markers in closest position to the target gene. These markers are then used to screen genomic libraries to isolate clones that hybridize to the marker. Bacterial artificial chromosome (BAC) and expressed sequence tag (EST) are involved in genomic libraries. Such markers reveal specific DNA sequences of the target gene region. Usually the ends of the clones are used to sequence new markers (Morgante and Salamini 2003). Perfect markers are developed, which are located within the actual gene sequence, thus eliminating the option of recombination between marker and gene (Ellis et al. 2002). The gene on this clone is flanked by the identified markers and highly co-segregates with this gene. The candidate gene should be sequenced and cloned. The cloned genes are a prerequisite for genetic engineering to transform commercial apple cultivars with target genes (Sharma et al. 2004). Examples are the Mi nematode pest resistance gene cloned from tomato and introduced into tobacco, Arabidopsis and flax (Rossi et al. 1998). Preliminary map-based cloning was carried out for the apple scab resistance gene Vf (Vinatzer et al. 2001, Belfanti et al. 2004) and the D. cf. devecta Sd1 gene (Cevik and King 2002b). The last step in map-based cloning is an in-depth molecular and biochemical analysis to determine the function of the target genes, and in this way to reveal new sources of resistance. Multiplex PCR for efficient DNA extraction, microarrays for monitoring the expression of thousands of genes simultaneously, have facilitated the development of molecular markers. The availability of more detailed maps will promote whole genome sequencing of apple, as exists for Arabidopsis. First results are available for the genus , and this will be interesting for comparative analysis of the two Rosaceae genera. A further approach is highly valuable for identifying QTLs in a diverse population. Pedigree analysis studies the segregation of target gene with the aim to select an appropriate parent (van de Weg et al. 2004). Further improvements may derive from advanced QTL models and algorithms to enhance the power and precision of QTL analysis (Manly and Olson 1999).

120 9.5. Conclusion

In conclusion, the present thesis demonstrated the occurrence of a genetically based resistance in apple to common herbivore pests. Based on a field survey herbivore abundance, a significant variability among the 'Fiesta' x 'Discovery' apple progenies was found, and QTLs related to C. pomonella, D. plantaginea, D. cf. devecta, and A. schlechtendali infestation in apple were identified. The presented SSRs are a prerequisite for marker assisted selection. A fine mapping, and phenotypic screening under different environmental conditions and genetic background should be carried out in advance. The intermediate heritability indicates a partial resistance, which is of high value in host-plant resistance. The high selection pressure on a pest of highly resistant plants may lead to adapted biotypes. This risk is minimized in quantitative traits. Relating to A. pomi, the present thesis revealed a particularly strong influence of environmental conditions, such as shoot length, as the relationship to co- occurring herbivores, and a spatial pattern. Microclimate conditions and fruit traits have affected withtin-tree infestation of C. pomonella larvae. These findings are relevant to elucidate host-plant resistance in apple and for an efficient and accurate monitoring system in integrated pest management.

121 Appendix: Herbivore resistant and susceptible apple selections: genetic background and fruit quality

Growing herbivore resistant plants is a primary management tool in integrated cropping systems. The present study identified herbivore resistant and susceptible selections in 160 apple progeny plants (Malus x domestica). Five selections turned out to show particularly high potential for multiple resistance breeding targeting the codling moth (Cydia pomonella) the aphid pests Dysaphis plantaginea, D. cf. devecta, and Aphis pomi, as well as the rust mite (Aculus schlechtendali). The presence of molecular markers that were previously identified to be associated with QTLs for susceptibility to C. pomonella or for resistance to D. plantaginea, D. cf. devecta, A. pomi, and A. schlechtendali varied between the 20 selections with highest resistance and the 20 selections with highest susceptibility, emphasising the presence of genetically based resistance. Remarkably, no relationship between resistance and fruit-quality characteristics appeared to exist as no significant differences in fruit weight, fruit firmness, and sugar content between the apple selections were detected. In conclusion, the studied apple selections show high potential to be introduced into breeding for pest resistant cultivars with high fruit quality.

122 A.1. Objectives

A field survey of different herbivores on 160 'Fiesta' x 'Discovery' F1 progeny plants resulted in a set of QTLs associated with herbivore resistance in apple. The RAPD marker Z19-350 on the 'Discovery' linkage group (LG) 10 at 66.2 cM was associated with higher codling moth (Cydia pomonella L.) infestation (Stoeckli et al. 2009). Resistance to the rosy apple aphid (Dysaphis plantaginea Pass.) was related to the AFLP marker E33M35-0269 on the 'Fiesta' LG 17 at 57.7 cM. Progeny plants amplifying the AFLP marker E32M39-0195 ('Fiesta' LG 7 at 4.5 cM) were resistant to the leaf-curling aphids (Dysaphis cf. devecta Wlk., species complex) (Stoeckli et al. 2008c). For a third aphid species, namely the green apple aphid (Aphis pomi De Geer), a putative QTL for antibiosis-based resistance was identified on the 'Fiesta' LG 11 at 21.5 cM (SSR CH02g12) (Stoeckli et al. 2008a). Rust mite (Aculus schlechtendali Nalepa) resistance in apple was associated to the AFLP marker E35M42-0146 on the 'Fiesta' LG 7 at 20.2 cM (Stoeckli et al. 2008d). The objectives of the present study were (a) to quantify the expression of the described molecular markers associated to QTLs for herbivore infestation in the apple selections with strongest and lowest phenotypic resistance to distinct arthropod pest species, and (b) to identify apple genotypes that show a high potential for apple breeding according to their combined resistance and fruit quality traits.

A.2. Materials and Methods

Plant material and study sites

Herbivore resistance was assessed for 160 F1 progeny plants of a 'Fiesta' and 'Discovery' (Malus x domestica Borkh.) cross. The studied dwarf apple trees were bud grafted on M27 rootstocks and replicated at the three study sites Ticino (Cadenazzo), Valais (Conthey) and Zurich (Wadenswil) in Switzerland in 1998 (Liebhard et al. 2003b).

Herbivore survey and categorization Herbivore infestation per tree was taken as a measure of resistance. The number of codling moth (Cydia pomonella L.) larval penetrations in fruits was inspected only on fruits attached to the tree, not on fallen fruits (Stoeckli et al. 2009). The rosy apple aphids (Dysaphis plantaginea Pass.) were assessed by evaluation of the number of colonies (Stoeckli et al. 2008c). The number of red-curled leaves was taken as a measure of leaf-curling aphid

123 (Dysaphis cf. devecta Wlk.) infestation (Stoeckli et al. 2008c). The number of individual aphids was counted to quantify overall-resistance to the green apple aphid (Aphis pomi De Geer) (Stoeckli et al. 2008c). Besides this overall-resistance documented in the field survey, a separate experimental investigation was carried out to characterize antibiosis-based resistance to A. pomi by evaluating population development on artificially infested shoots in sleeve cages (in four field trials) (Stoeckli et al. 2008a). The number of rusty leaves was considered to evaluate rust mite (Aculus schlechtendali Nalepa) resistance in apple (Stoeckli et al. 2008d). Several surveys per year were carried out for evaluation at the three study sites in 2005 and 2006. Herbivore surveys were based on one (L. clerkella and A. schlechtendali) to four (A. pomi) assessments per season. As C. pomonella was controlled by mating disruption at the Zurich site, this site was not considered for evaluation of resistance to this species. For each herbivore and survey, the total infestation (e.g. number of C. pomonella larval penetrations on the studied progeny plants) was calculated, and the relative infestation (RI) in percentage of this total was determined for each progeny plant. RI from several surveys was averaged for statistical analysis. Subsequently, progeny plants were sorted according to their level of resistance to a specific herbivore, and two categories were established for statistical evaluation. The 20 selections with the strongest resistance (maximum RI ranging between 0.0% and 0.1%) and the 20 selections with the strongest susceptibility (maximum RI ranging between 4.1% and 10.6%) to a specific herbivore were selected (high-resistance-selections and high-susceptibility-selections).

Fruit quality Weight, firmness, and sugar content of fruits produced by the study trees had been recorded previously for the same 160 progeny plants used for the herbivore survey (Liebhard et al. 2003a). Fruit quality was measured at the three sites in 2000 and 2001, and fruits of a tree were harvested at once in two-week intervals. This data set was used to analyse fruit quality of high-resistance-selections and high-susceptibility-selections. Fruit flesh firmness was assessed with an automated Magnes-Taylor penetrometer (C. Stevens & Son., St. Albans, UK). The penetration force within a skinned fruit cortex was recorded in kg/cm2. The juice of penetrated fruits was collected using a hand-squeezer. Sugar content was determined with a digital refractometer (Atago, Tokyo, Japan) requiring a few drops of juice per sample. Results were recorded in °Brix, which equals to grams sugar per 100 ml juice. Although an effect of age on fruit set and quality is described (Volz et al. 1994), differences between cultivars or selections remain stable among years (Sturm et al. 2003, Veberic et al. 2007). Possible

124 differences in fruit quality between high-resistance-selections and high-susceptibility- selections should therefore be detectable with the analyzed data set.

Statistical analysis To analyse the presence of previously identified molecular markers associated with QTLs for herbivore resistance or susceptibility in high-resistance-selections and high- susceptibility-selections, a Pearson χ2 test for a 2 x 2 cross-table was applied (Zar 1999). The fruit quality of high-resistance-selections and high-susceptibility-selections was analyzed using a Mann-Whitney test and a Benjamini-Hochberg correction for multiple tests (Verhoeven et al. 2005). All statistics were carried out with SPSS 16.0 for Mac OS X (SPSS, Inc., Chicago, IL).

A.3. Results

Herbivore resistant apple selections The 20 high-resistance-selections and the 20 high-susceptibility-selections for a specific herbivore species are presented in Table A.1. Mean relative infestation (RI) of high- resistance-selections was 0.0 ± 0.0% (mean ± SE; data not shown). The 20 high-resistance- selections were not infested at all by D. cf. devecta and by A. schlechtendali. Some of the high-resistance-selections were infested with C. pomonella (maximum RI: 0.04%), D. plantaginea (0.07%), and A. pomi (0.06%). Mean RI (± SE; data not shown) of high- susceptibility-selections amounted to 2.6 ± 0.4% (C. pomonella), 2.7 ± 0.6% (D. plantaginea), 3.3 ± 0.5% (D. cf. devecta), 2.7 ± 0.3% (A. pomi), and 2.4 ± 0.4% (A. schlechtendali). Maximum RI was 8.7% (C. pomonella), 10.6% (D. plantaginea), 8.6% (D. cf. devecta), 7.4% (A. pomi), and 4.1% (A. schlechtendali). The 20 high-resistance-selections were compared to the 20 high-susceptibility-selections and differences in RI were significant for all studied pest species (Mann-Whitney test; n = 20/20, Z = -5.4 to -5.8, P < 0.0001; data not shown). Remarkably, five apple progeny plants were ranked as high-resistance-selections to four of five studied herbivores (Table A.1): selections no. 13, 265, and 303 to D. plantaginea, D. cf. devecta, A. pomi (overall or antibiosis-based resistance), and A. schlechtendali; selection no. 22 to C. pomonella, D. plantaginea, D. cf. devecta, and A. pomi (antibiosis- based resistance); and selection no. 189 to C. pomonella, D. plantaginea, A. pomi (antibiosis- based resistance), and A. schlechtendali.

125 Table A.1. The 20 high-resistance-selections (R) and the 20 high-susceptibility-selections (S) identified for a specific herbivore (A-E) a. In bold are the selections with multiple resistance to four different herbivore species. Apple selections with multiple resistance to three different herbivore species are underlined.

(A) Cydia (B) Dysaphis (C) Dysaphis (E) Aculus (D) Aphis pomi pomonella plantaginea cf. devecta b schlechtendali

Overall Antibiosis c

R S R S R S R S R S R S

9 50 13 4 13 18 12 342 14 6 13 18

10 53 21 5 21 29 13 330 15 29 25 29

22 65 22 53 22 68 18 324 22 43 26 32

30 69 44 108 65 74 26 318 71 52 73 33 68 92 68 111 71 79 30 299 85 67 79 35 82 109 76 114 82 80 50 292 116 74 83 80 155 126 77 132 86 117 71 278 121 97 95 107 157 133 82 151 162 151 76 270 127 111 120 114 173 138 112 153 172 248 109 262 142 180 140 149 189 139 127 231 210 262 140 237 159 217 162 173 215 149 189 239 223 263 155 230 172 223 172 179 225 231 215 246 226 271 172 201 225 231 175 180 226 238 265 251 243 293 175 130 248 269 235 217 227 241 269 252 251 300 189 112 249 276 248 238 240 243 284 255 265 311 198 74 265 277 255 262 288 257 287 261 280 356 227 69 290 292 265 270 300 261 288 271 292 357 269 65 293 299 280 288 318 282 303 277 303 358 303 33 294 305 292 305 358 307 317 280 326 360 356 32 307 354 303 307 362 357 375 282 375 376 378 4 322 362 375 360 a Mean relative infestation (RI) of the different surveys at the Ticino, Valais and Zurich sites conducted in two consecutive years were used for evaluation. Number of surveys: 10 (D. plantaginea, D. cf. devecta), 16 (A. pomi), 3 (A. schlechtendali), 8 (C. pomonella). b The following selections showed additionally no infestation by D. cf. devecta: 6, 9, 20, 26, 30, 33, 43, 50, 92, 119, 120, 127, 132, 133, 138, 139, 147, 149, 175, 178, 189, 208, 212, 225, 227, 237, 238, 239, 255, 257, 269, 277, 278, 284, 299, 305, 318. c Experimental evidence for antibiosis-based resistance to A. pomi was obtained from four trials that quantified population development in sleeve cages.

126 Genetic background of herbivore resistant and susceptible apple selections The presence of the identified molecular marker associated with a QTL for susceptibility to C. pomonella varied significantly between high-resistance-selections and high-susceptibility-selections (χ2 test, d.f.=1; χ2 = 5.0, P = 0.025; Table A.2). The identified molecular markers associated with a QTL for herbivore resistance were overrepresented in high-resistance-selections compared to high-susceptibility-selections (D. cf. devecta: χ2 = 28.9, P < 0.0001; A. schlechtendali: χ2 = 4.9, P = 0.027; and D. plantaginea: χ2 = 1.6, P = 0.204). The putative QTL marker related to A. pomi antibiosis-based resistance was significantly more present in high-resistance-selections compared to high-susceptibility- selections (χ2 = 7.6, P =0.006).

Fruit quality of herbivore resistant and susceptible apple selections Mean fruit weight was lower for high-resistance-selections (107.2-116.2 g) compared to high-susceptibility-selections (108.5-133.6 g), but the difference was significant for C. pomonella only (Mann-Whitney test; n = 20/20, Z = -2.5, P = 0.012) (Table A.3). For fruit firmness, no consistent pattern existed for high-resistance-selections and high-susceptibility- selections, and the difference was not significant for any herbivore species (Mann-Whitney test; n = 20/20, P > 0.05) (Table A.3). Mean sugar content was not significantly different in high-resistance-selections compared to high-susceptibility-selections with one exception (Mann-Whitney test; n = 20/20, P > 0.05) (Table A.3). For C. pomonella, sugar content in high-resistance-selections was significantly higher compared to high-susceptibility-selections (Mann-Whitney test; n = 20/20, Z = -2.5, P = 0.011) (Table A.3). Mean sugar content was slightly higher in high-susceptibility-selections (14.0-14.7 °Brix) compared to high- resistance-selections (13.8-14.0 °Brix) considering the other herbivore species.

127 Table A.2. Presence and absence of molecular markers associated with QTLs for susceptibility to Cydia pomonella, and for resistance to Dysaphis plantaginea, Dysaphis cf. devecta, Aphis pomi (antibiosis-based resistance), and Aculus schlechtendali in the 'Fiesta' x 'Discovery' apple progeny a. Presence and absence of molecular markers in the high-resistance-selections and high-susceptibility- selections was tested with a χ2 test for a 2 x 2 cross-table (d.f. = 1).

Cydia Dysaphis Dysaphis cf. Aphis pomi Aculus pomonella plantaginea devecta (antibiosis) schlechtendali

Molecular marker Z19-350 b E33M35-0269 c E32M39-0195 d CH02g12-199 e E35M42-0146 f

High-resistance-selections Presence 8 13 18 11 14 Absence 12 7 2 6 6 High-susceptibility selections Presence 15 9 1 4 7 Absence 5 11 19 16 13 χ2 5.0 1.6 28.9 7.6 4.9 P 0.025 0.204 <0.0001 0.006 0.027 a QTL analyses were based on previously developed molecular markers for the 'Fiesta' (345 markers) and 'Discovery' ( 389 markers) apple linkage map (Liebhard et al. 2003a). Interval mapping was used for QTL analysis. The described QTLs associated with herbivore resistance or susceptibility were significant at the 5% genome-wide confidence level, and molecular markers linked to a QTL were positioned within the 95% confidence interval. b The RAPD marker Z19-350 is positioned on the 'Discovery' linkage group 10 (66.2 cM), and was associated with a QTL for C. pomonella susceptibility (phenotypic variation explained: 8.2%) (Stoeckli et al. 2009). c The AFLP marker E33M35-0269 is positioned on the 'Fiesta' linkage group 17 (57.7 cM), and was associated with a QTL for D. plantaginea resistance (8.5%) (Stoeckli et al. 2008c). d The AFLP E32M39-0195 is positioned on the 'Fiesta' linkage group 7 (4.5 cM), and was associated with a QTL for D. cf. devecta resistance (20.4%) (Stoeckli et al. 2008c). e The SSR CH02g12-199 is positioned on the 'Fiesta' linkage group 11 (21.5 cM), and was associated with a QTL for A. pomi antibiosis-based resistance) (14.1%) (Stoeckli et al. 2008a). f The AFLP E35M42-0146 is positioned on the 'Fiesta' linkage group 7 (20.2 cM), and was associated with a QTL for A. schlechtendali resistance (16.6%) (Stoeckli et al. 2008d).

128 Table A.3. Comparison of the fruit characteristics a fruit weight (g), fruit firmness (kg/cm2), and sugar content (°Brix) of high-resistance-selections and high-susceptibility-selections (mean ± SE). Significant P values (Mann-Whitney test) after Benjamini-Hochberg correction for multiple tests are highlighted.

Cydia Dysaphis Dysaphis Aphis pomi Aphis pomi Aculus pomonella plantaginea cf. devecta (overall) (antibiosis) schlechtendali

Fruit weight P = 0.012 P = 0.684 P = 0.620 P = 0.851 P = 0.134 P = 0.024 Resistant 107.2 ± 6.1 108.8 ± 8.0 109.6 ± 7.9 116.2 ± 7.6 218.1 ± 2.4 110.5 ± 7.9 Susceptible 131.5 ± 6.6 108.5 ± 6.3 111.3 ± 5.5 118.0 ± 7.1 219.3 ± 1.8 133.6 ± 6.6 Fruit firmness P = 0.171 P = 0.443 P = 0.707 P = 0.778 P = 0.893 P = 0.041 Resistant 9.3 ± 0.4 9.7 ± 0.5 9.8 ± 0.4 10.0 ± 0.4 9.4 ± 0.5 10.4 ± 0.4 Susceptible 10.3 ± 0.5 10.1 ± 6.3 9.3 ± 0.5 10.1 ± 0.6 9.4 ± 0.4 9.3 ± 0.5 Sugar content P = 0.011 P = 0.708 P = 0.134 P = 0.471 P = 0.578 P = 0.059 Resistant 14.9 ± 0.3 13.8 ± 0.3 13.9 ± 0.3 14.0 ± 0.2 14.2 ± 0.3 13.9 ± 0.2 Susceptible 13.7 ± 0.3 14.0 ± 0.4 14.2 ± 0.2 14.4 ± 0.3 14.2 ± 0.3 14.7 ± 0.3 a Based on data published in Liebhard et al. (2003b).

A.4. Conclusion

The presented results revealed a significant variation in the level of resistance among the 160 apple progeny plants of a 'Fiesta' x 'Discovery' F1-cross. Furthermore, five apple progeny plants were described showing multiple resistance to four of five studied herbivore species (selection number 13, 22, 189, 265, and 303). This finding underlines the potential for breeding apple cultivars with multiple pest resistance. QTLs associated to herbivore resistance and susceptibility had previously been identified (Stoeckli et al. 2008a, Stoeckli et al. 2008c, Stoeckli et al. 2008d, Stoeckli et al. 2009). The expression of molecular markers associated to these QTLs varied between high-resistance-selections and high-susceptibility-selections. The difference was significant for all studied herbivore species, with the exception of D. plantaginea. These results indicate a genetic background associated with herbivore resistance. Molecular markers are a prerequisite to select pest-resistant apple cultivars in an early stage. Marker-assisted selection eliminates unreliable scoring based on phenotypic evaluation. There was no negative relationship in fruit quality between high-resistance-selections compared to the high-susceptibility-selections, as supposed for pest resistant plants. This result indicates that the presented high-resistance-selections are of promising value for breeding pest resistant

129 apple cultivars combined with desired fruit quality. Relating to A. pomi, there was only a weak relationship between field resistance and antibiosis-based resistance, and a more detailed evaluation should be considered in follow-up studies.

130 10. References

Alford, D. V. 2007. Pests of Fruit Crops: A Colour Handbook. Manson Publishing Ltd., London, UK. Alspach, P. A. and V. G. M. Bus. 1999. Spatial variation of woolly apple aphid (Eriosoma lanigerum Hausmann) in a genetically diverse apple planting. New Zealand Journal of Ecology 23:39-44. Alston, F. H. and J. B. Briggs. 1968. Resistance to Sappaphis devecta in apple. Euphytica 17:468-472. Alston, F. H. and J. B. Briggs. 1970. Inheritance of hypersensitivity to rosy apple aphid (Dysaphis plantaginea) in apple. Canadian Journal of Genetics and Cytology 12:257-258. Alston, F. H. and J. B. Briggs. 1977. Resistance genes in apple and biotypes of Dysaphis devecta. Annals of Applied Biology 87:75-81. Angeli, G. and S. Simoni. 2006. Apple cultivars acceptance by Dysaphis plantaginea Passerini (Homoptera: Aphididae). Journal of Pest Science 79:175-179. Aquino, G. B. and E. A. Heinrichs. 1979. Brown planthopper populations on resistant cultivar treated with resurgence causing insecticide. International Rice Research Newsletter 4:12. Arbab, A., D. C. Kontodimas, and A. Sahragard. 2006. Estimating development of Aphis pomi (DeGeer) (Homoptera: Aphididae) using linear and nonlinear models. Environmental Entomology 35:1208-1215. Arnaoudov, V. and H. Kutinkova. 2006. Susceptibility of some apple cultivars to infestation by the rosy apple aphid (Dysaphis plantaginea Pass., Homoptera: Aphididae). Journal of Fruit and Ornamental Plant Research 14:137-142. Audemard, H. 1991. Population dynamics of the codling moth. Pages 329-338 in L. P. S. van der Geest and H. H. Evenhuis, editors. Tortricid Pests: Their Biology, Natural Enemies and Control. Elsevier, Amsterdam, NL. Audemard, H., O. Baudry, D. Bergère, D. Breniaus, and J. P. Gendrier. 1992. Protection intégrée du verger de pommiers de l'an 2000. Acta Phytopathologica et Entomologica Hungarica 27:39-44. Avilla, J. and H. Riedl. 2003. Integrated fruit production for apples: principles and guidelines. Pages 539-549 in D. C. Ferree and I. J. Warrington, editors. Apples: Botany, Production and Uses. CABI Publishing, Wallingford, UK. Baggiolini, M., E. Keller, H. G. Milaire, and H. Steiner. 1992. Visuelle Kontrollen im Apfelanbau. Schweizerische Zentrale für Obstbau, Oeschberg, Koppigen, CH. Baker, A. C. and W. F. Turner. 1916. Morphology and biology of the green apple aphis. Journal of Agricultural Research 5:955-994. Barnes, M. M. 1991. Codling moth occurrence, host race formation, and damage. Pages 313-328 in L. P. S. van der Geest and H. H. Evenhuis, editors. Tortricid Pests: Their Biology, Natural Enemies and Control. Elsevier, Amsterdam, NL. Beeghly, H. H., J. G. Coors, and M. Lee. 1997. Plant fibre composition and resistance to European corn borer in four maize populations. Maydica 42:297-303. Beers, E. H., J. F. Brunner, M. J. Willett, and G. M. Warner. 1993. Orchard Pest Management: A Resource Book for the Pacific Northwest. Good Fruit Grower, Yakima, WA. Beers, E. H., D. M. Suckling, R. J. Prokopy, and J. Avilla. 2003. Ecology and management of apple arthropod pests. Pages 489-514 in D. C. Ferree and I. J. Warrington, editors. Apples: Botany, Production and Uses. CABI Publishing, Wallingford, UK. Belding, R. D., S. M. Blankenship, E. Young, and R. B. Leidy. 1998. Composition and variability of epicuticular waxes in apple cultivars. Journal of the American Society of Horticultural Science 123:348-356. Belfanti, E., E. Silfverberg-Dilworth, S. Tartarini, A. Patocchi, M. Barbieri, J. Zhu, B. A. Vinatzer, L. Gianfranceschi, C. Gessler, and S. Sansavini. 2004. The HcrVf2 gene from a wild apple confers scab resistance to a transgenic cultivated variety. Proceedings of the National Academy of Sciences of the United States of America 101:886-890. Benjamini, Y. and Y. Hochberg. 1995. Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society: Series B (Statistical Methodology) 57:289- 300. Bernasconi, M. L., T. C. J. Turlings, L. Ambrosetti, P. Bassetti, and S. Dorn. 1998. Herbivore-induced emissions of maize volatiles repel the corn leaf aphid (Rhopalosiphum maidis). Entomologia Experimentalis et Applicata 87:133-142. Bernet, G. P., C. Margaix, J. Jacas, E. A. Carbonell, and M. J. Asins. 2005. Genetic analysis of citrus leafminer susceptibility. Theoretical and Applied Genetics 110:1393-1400. Blackman, R. L. and V. F. Eastop. 2000. Aphids on the World's Crops: An Identification and Information Guide. John Wiley & Sons Ltd., Chichester, UK.

131 Blago, N. and E. Dickler. 1990. Neue Methode zur Untersuchung der Ei-Phänologie des Apfelwicklers (Cydia pomonella L.) (Lepidoptera, Tortricidae). Journal of Applied Entomology 109:98-104. Bloem, S., J. E. Carpenter, and S. Dorn. 2006. Mobility of mass-reared diapaused and nondiapaused Cydia pomonella (Lepidoptera: Tortricidae): effect of different constant temperatures and lengths of cold storage. Journal of Economic Entomology 99:707-713. Blomefield, T. L., K. L. Pringle, and A. Sadie. 1997. Field observations on oviposition of codling moth (Cydia pomonella L.) (Lepidoptera: Olethreutidae) in an unsprayed apple orchard in South . African Entomology 5:319-336. Blommers, L. H. M. 1994. Integrated pest-management in European apple orchards. Annual Review of Entomology 39:213-241. Blommers, L. H. M., H. H. M. Helsen, and F. W. N. M. Vaal. 2004. Life history data of the rosy apple aphid (Dysaphis plantaginea Pass.) (Homoptera, Aphididae) on plantain and as migrant to apple. Journal of Pest Science 77:155-163. Bobb, M. L., A. M. Woodside, and R. N. Jefferson. 1939. Baits and bait traps in codling moth control. Virginia Agricultural Experiment Station Bulletin 1:320. Briggs, J. B. and F. H. Alston. 1969. Sources of pest resistance in apple cultivars. Report of the East Malling Research Station for 1968 1:159-162. Brooks, T. D., B. S. Bushman, W. P. Williams, M. D. McMullen, and P. M. Buckley. 2007. Genetic basis of resistance to fall armyworm (Lepidoptera: Noctuidae) and southwestern corn borer (Lepidoptera: Crambidae) leaf-feeding damage in maize. Journal of Economic Entomology 100:1470-1475. Brown, S. K. 1992. Genetics of apple. Plant Breeding Review 9:333-366. Brown, S. K. and K. E. Maloney. 2003. Genetic improvement of apple: breeding, markers, mapping and biotechnology. Pages 31-61 in D. C. Ferree and I. J. Warrington, editors. Apples: Botany, Production and Uses. CABI Publishing, Wallingford, UK. Bus, V. G. M., D. Chagné, H. C. M. Bassett, D. Bowatte, F. Calenge, J.-M. Celton, C.-E. Durel, M. T. Malone, A. Patocchi, A. C. Ranatunga, E. H. A. Rikkerink, D. S. Tustin, J. Zhou, and S. E. Gardiner. 2008. Genome mapping of three major resistance genes to woolly apple aphid (Eriosoma lanigerum Hausm.). Tree Genetics & Genomes 4:223-236. Caesar, L., J. A. Hall, and A. Kelsall. 1934. The influence of cultural practices on orchard insects. Report of the Entomological Society of Ontario 65:40-42. Calenge, F., D. Drouet, C. Denance, W. E. van de Weg, M. N. Brisset, J. P. Paulin, and C.-E. Durel. 2005. Identification of a major QTL together with several minor additive or epistatic QTLs for resistance to fire blight in apple in two related progenies. Theoretical and Applied Genetics 111:128-135. Calenge, F. and C.-E. Durel. 2006. Both stable and unstable QTLs for resistance to powdery mildew are detected in apple after four years of field assessment. Molecular Breeding 17:329-339. Calenge, F., A. Faure, M. Goerre, C. Gebhardt, W. E. van de Weg, L. Parisi, and C.-E. Durel. 2004. Quantitative Trait Loci (QTL) analysis reveals both broad-spectrum and isolate-specific QTL for scab resistance in an apple progeny challenged with eight isolates of Venturia inaequalis. Phytopathology 94:370-379. Carlé, S. A., A. L. Averill, G. S. Rule, W. H. Reissig, and W. L. Roelofs. 1987. Variation in host fruit volatiles attractive to apple maggot fly (Rhagoletis pomonella). Journal of Chemical Ecology 13:795-805. Castro, A. M., A. Vasicek, M. Manifiesto, D. O. Giménez, M. S. Tacaliti, O. Dobrovolskaya, M. S. Röder, J. W. Snape, and A. Börner. 2005. Mapping antixenosis genes on chromosome 6A of wheat to greenbug and to a new biotype of Russian wheat aphid. Plant Breeding 124:229-233. Cevik, V. and G. J. King. 2002a. High-resolution genetic analysis of the Sd1 aphid resistance locus in Malus spp. Theoretical and Applied Genetics 105:346-354. Cevik, V. and G. J. King. 2002b. Resolving the aphid resistance locus Sd1 on a BAC contig within a sub- telomeric region of Malus linkage group 7. Genome 45:939-945. Chagné, D., C. M. Carlisle, C. Blond, R. K. Volz, C. J. Whitworth, N. C. Oraguzie, R. N. Crowhurst, A. C. Allan, R. V. Espley, R. P. Hellens, and S. E. Gardiner. 2007. Mapping a candidate gene (mdMyB10) for red flesh and foliage color in apple. BMC Genomics 8:1-11. Cheng, F. S., N. F. Weeden, S. Brown, K., H. S. Aldwinckle, S. E. Gardiner, and V. G. M. Bus. 1998. Development of a DNA marker for Vm, a gene conferring resistance to apple scab. Genome 41:208- 214. Chouinard, G., S. Bellerose, C. Brodeur, and Y. Moring. 2006. Effectiveness of Hyaliodes vitripennis (Say) (Heteroptera: Miridae) predation in apple orchards. Crop Protection 25:705-711. Churchill, G. A. and R. W. Doerge. 1994. Empirical threshold values for quantitative trait mapping. Genetics 138:963-971. Coleson, J. L. and R. H. Miller. 2005. Antibiosis and antixenosis to Aphis gossypii (Homoptera: Aphididae) in Colocasia esculenta. Journal of Economic Entomology 98:996-1006.

132 Collard, B. C. Y., M. Z. Z. Jahufer, J. B. Brouwer, and E. C. K. Pang. 2005. An introduction to markers, quantitative trait loci (QTL) mapping and marker-assisted selection for crop improvement: the basic concepts. Euphytica 142:169-196. Conner, P. J., S. K. Brown, and N. F. Weeden. 1998. Molecular-marker analysis of quantitative traits for growth and development in juvenile apple trees. Theoretical and Applied Genetics 96:1027-1035. Cossentine, J. and H. Madsen. 1980. Susceptibility of crabapple cultivars to attack by the codling moth (Laspeyresia pomonella). Journal of the Entomological Society of British Columbia 77:27-29. Cross, J. V., M. G. Solomon, D. Chandler, P. Jarrett, P. N. Richardson, D. Winstanley, H. Bathon, J. Huber, B. Keller, G. A. Langenbruch, and G. Zimmerman. 1999. Biocontrol of pests of apples and pears in northern and central Europe: microbial agents and nematodes. Biocontrol Science and Technology 9:125-149. Cutright, C. R. 1930. Additional notes on Aphis pomi. Journal of Economic Entomology 23:738-741. Dapena, E. and M. Miñarro. 2001. Evaluation of the tolerance to the rosy apple aphid (Dysaphis plantaginea Pass.) in descendants of the crossing 'Raxao' x 'Florina'. IOBC/wprs Bulletin 24:247-255. de Jong, P. W. and J. K. Nielsen. 2002. Host plant use of Phyllotreta nemorum: do coadapted gene complexes play a role? Entomologia Experimentalis et Applicata 104:207-215. Diaz-Montano, J., J. C. Reese, W. T. Schapaugh, and L. R. Campbell. 2006. Characterization of antibiosis and antixenosis to the soybean aphid (Hemiptera: Aphididae) in several soybean genotypes. Journal of Economic Entomology 99:1884-1889. Dirlewanger, E., V. Pronier, C. Parvery, C. Rothan, A. Guye, and R. Monet. 1998. Genetic linkage map of peach [Prunus persica (L.) Batsch] using morphological and molecular markers. Theoretical and Applied Genetics 97:888-895. Dorn, S., P. Schumacher, C. Abivardi, and R. Meyhöfer. 1999. Global and regional pest insects and their antagonists in orchards: spatial dynamics. Agriculture, Ecosystems and Environment 73:111-118. Dorschner, K. W., J. D. Ryan, R. C. Johnson, and R. D. Eikenbary. 1987. Modification of host nitrogen levels by the greenbug (Homoptera: Aphididae): its role in resistance of winter wheat to aphids. Environmental Entomology 16:1007-1011. Downing, R. S. and T. K. Moilliet. 1967. Relative densities of predacious and phytophagous mites on three cultivars of apple trees. Canadian Entomologist 99:733-741. Duffey, S. S. and K. A. Bloem. 1986. Plant defense-herbivore-parasite interactions and biological control. Pages 135-183 in M. Kogan, editor. Ecological Theory and Integrated Pest Management Practice. John Wiley & Sons Ltd., New York. Durel, C.-E., F. Calenge, L. Parisi, W. E. van de Weg, L. P. Kodde, R. Liebhard, C. Gessler, M. Thierman, F. Dunemann, F. Gennari, S. Tartarini, and Y. Lespinasse. 2004. Overview on position and robustness of scab resistance QTL an major genes by alignment of genetic maps in five apple progenies. Acta Horticulturae 663:135-140. Duso, C., M. Pasini, and M. Pellegrini. 2003. Distribution of the predatory mite Typhlodromus pyri (Acari: Phytoseiidae) on different apple cultivars. Biocontrol Science and Technology 13:671-681. Easterbrook, M. A. and M. M. Fuller. 1986. Russeting of apples caused by apple rust mite (Aculus schlechtendali) (Acarina: Eriophyidae). Annals of Applied Biology 109:1-9. Easterbrook, M. A. and J. W. Palmer. 1996. The relationship between early-season leaf feeding by apple rust mite (Aculus schlechtendali Nal.), and fruit set and photosynthesis of apple. Journal of Horticultural Science 71:939-944. Eigenmann, C. and M. Kellerhals. 2007. Welche Äpfel wollen die Konsumentinnen und Konsumenten? Agrarforschung 14:6-9. Elden, T. C. 1997. Influence of soybean lines isogenic for pubescence type on twospotted spider mite (Acarina: Tetranychidae) development and feeding damage. Journal of Entomological Science 32:296-302. Ellis, M. H., W. Spielmeyer, K. R. Gale, G. J. Rebetzke, and R. A. Richards. 2002. 'Perfect' markers for the Rht- B1b and Rht-D1b dwarfing genes in wheat. Theoretical and Applied Genetics 105:1038-1042. Fedoroff, N. V. and J. E. Cohen. 1999. Plants and population: is there time? Proceedings of the National Academy of Sciences of the United States of America 96:5903-5907. Fischer, C. 1994. Breeding apple cultivars with multiple resistance. Pages 43-48 in H. Schmidt and M. Kellerhals, editors. Progress in Temperate Fruit Breeding. Kluwer Academic Publishers, Dordrecht, NL. Francia, E., G. Tacconi, C. Crosatti, D. Barabaschi, D. Bulgarelli, E. Dall'Aglio, and G. Valè. 2005. Marker assisted selection in crop plants. Plant Cell, Tissue and Organ Culture 82:317-342. Frei, A., M. W. Blair, C. Cardona, S. E. Beebe, H. Gu, and S. Dorn. 2005. QTL mapping of resistance to Thrips palmi (Karny) in common bean. Crop Science 45:379-387. Frey, J. E., B. Frey, C. Sauer, and M. Kellerhals. 2004. Efficient low-cost DNA extraction and multiplex fluorescent PCR method for marker-assisted selection in breeding. Plant Breeding 123:554-557.

133 Fujita, D., K. Doi, A. Yoshimura, and H. Yasui. 2006. Molecular mapping of a novel gene, Grh5, conferring resistance to green rice leafhopper (Nephotettix cincticeps Uhler) in rice (Oryza sativa L.). Theoretical and Applied Genetics 113:567-573. Fukuda, H. and F. Takishita. 1998. Effect of fruit load on the average fruit size and quality of 'Fuji' apple trees on M9 dwarfing rootstock. Bulletin of the National Institute of Fruit Tree Science 30:63-73. Fulcher, A. F., T. G. Ranney, J. D. Burton, J. F. Walgenbach, and D. A. Danehower. 1998. Role of foliar phenolics in host plant resistance of Malus taxa to adult Japanese beetles. Hortscience 33:862-865. Gardiner, S., V. G. M. Bus, R. Rusholme, D. Chagné, and E. Rikkerink. 2007. Apple. Pages 1-62 in C. Kole, editor. Genome Mapping and Molecular Breeding in Plants: Fruits and Nuts. Springer, Berlin. Garman, P. 1938. Control of the rosy apple aphid in Connecticut apple orchards. Circular of the Connecticut Agricultural Experiment Station 126:1-16. Gayler, S., C. Leser, E. Priesack, and D. Treutter. 2004. Modelling the effect of environmental factors on the 'trade off' between growth and defensive compounds in young apple trees. Trees 18:363-371. Geier, P. W. 1963. The life history of codling moth (Cydia pomonella) in the Australian capital territory. Australian Journal of Zoology 11:323-367. Gessler, C. and A. Patocchi. 2007. Recombinant DNA technology in apple. Green Gene Technology: Research in an Area of Social Conflict 107:113-132. Gessler, C., A. Patocchi, S. Sansavini, S. Tartarini, and L. Gianfranceschi. 2006. Venturia inaequalis resistance in apple. Critical Reviews in Plant Sciences 25:473-503. Goonewardene, H. F., J. Janick, W. F. Kwolek, C. H. Bailey, and L. F. Hough. 1982. Pedigree analysis of apple clones with differences in preference to European red mite (Panonychus ulmi Koch). Hortscience 17:63-65. Goonewardene, H. F. and W. F. Kwolek. 1985. Greenhouse study of the effects of three early season pests on 19 apple selections. Journal of Economic Entomology 78:366-370. Goonewardene, H. F., W. F. Kwolek, and R. A. Hayden. 1984. Survival of immature stages of the codling moth (Lepidoptera: Tortricidae) on seeded and seedless apple fruit. Journal of Economic Entomology 77:1427-1431. Graf, B., J. Baumgärtner, and V. Delucchi. 1985. Life table statistics of three apple aphids Dysaphis plantaginea, Rhopalosiphum insertum, and Aphis pomi (Homoptera: Aphididae), at constant temperatures. Journal of Applied Entomology 99:285-294. Graf, B., H. U. Höpfli, and H. Höhn. 1992. Einfluss der Sortenwahl auf Schädlingsbefall und Raubmilbenbesatz im Apfelanbau. Schweizerische Zeitschrift für Obst- und Weinbau 128:618-622. Graf, B., H. U. Höpfli, H. Höhn, M. Kellerhals, and C. Krebs. 1998. Schorfresistente Apfelsorten: Wie steht's mit der Schädlingsanfälligkeit? Schweizerische Zeitschrift für Obst- und Weinbau 3:71-73. Graf, B., H. Höpli, and H. Höhn. 2001. Die unterschiedlichen Standortansprüche von Apfelwickler und Kleinem Fruchtwickler. Schweizerische Zeitschrift für Obst- und Weinbau 24:670-673. Graf, B., H. Höpli, H. Höhn, and J. Samietz. 2006. Temperature effects on egg development of the rosy apple aphid and forecasting of egg hatch. Entomologia Experimentalis et Applicata 119:207-211. Habekuss, A., G. Proeseler, E. Schliephake, and C. Fischer. 2000. Resistance of apple to spider mites and aphids. Acta Horticulturae 538:271-278. Häckermann, J., A. S. Rott, K. Tschudi-Rein, and S. Dorn. 2008. Cold stored ectoparasitoid of Cydia fruit moths released under different temperature regimes. BioControl 53:857-867. Hagley, E. A. C. 1976. Effect of rainfall and temperature on codling moth oviposition. Environmental Entomology 5:967-625. Hagley, E. A. C., J. F. Bronskill, and E. J. Ford. 1980. Effect of the physical nature of leaf and fruit surfaces on ovipositon by the codling moth (Cydia pomonella) (Lepidoptera: Tortricidae). Canadian Entomologist 112:503-510. Hamilton, G. C., F. C. Swift, and R. P. Marini. 1986. Effect of Aphis pomi (Homoptera: Aphididae) density on apples. Journal of Economic Entomology 79:471-478. Hampson, C. and H. Kemp. 2003. Characteristics of important commercial apple cultivars. Pages 61-90 in D. C. Ferree and I. J. Warrington, editors. Apples: Botany, Production and Uses. CABI Publishing, Wallingford, UK. Hastings, A. 1999. Outbreaks of insects: a dynamic approach. Pages 206-215 in B. A. Hawkins and H. V. Cornell, editors. Theoretical Approaches to Biological Control. Cambridge University Press, Cambridge, UK. Hausmann, C., J. Samietz, and S. Dorn. 2004. Monitoring the dynamics of orchard colonisation by Anthonomus pomorum in spring. Entomologia Experimentalis et Applicata 110:207-216. Hawley, C. J., F. B. Peairs, and T. I. Randolph. 2003. Categories of resistance at different growth stages in halt, a winter wheat resistant to the Russian wheat aphid (Homoptera: Aphididae). Journal of Economic Entomology 96:214-219.

134 Heinrichs, E. A., H. R. Rapusas, G. B. Aquino, and F. Palis. 1986. Integration of host-plant resistance and insecticides in the control of Nephotettix virescens (Homoptera: Cicadellidae), a vector of rice tungro virus. Journal of Economic Entomology 79:437-443. Herbert, H. J. 1974. Notes on the biology of the apple rust mite (Aculus schlechtendali) (Prostigmata: Eriophyoidae), and its density on several cultivars of apple in Nova Scotia. Canadian Entomologist 106:1035-1038. Hern, A. and S. Dorn. 1999. Sexual dimorphism in the olfactory orientation of adult Cydia pomonella in response to α-farnesene. Entomologia Experimentalis et Applicata 92:63-72. Hern, A. and S. Dorn. 2002. Induction of volatile emissions from ripening apple fruits infested with Cydia pomonella and the attraction of adult females. Entomologia Experimentalis et Applicata 102:145.151. Hern, A. and S. Dorn. 2003. Monitoring seasonal variation in apple fruit volatile emission in situ using solid- phase microextraction. Phytochemical Analysis 14:232-240. Hern, A. and S. Dorn. 2004. A female-specific attractant for the codling moth (Cydia pomonella) from apple fruit volatiles. Naturwissenschaften 91:77-80. Hill, C. B., Y. Li, and G. L. Hartman. 2004. Resistance to the soybean aphid in soybean germplasm. Crop Science 44:98-106. Hill, C. B., Y. Li, and G. L. Hartman. 2006. A single dominant gene for resistance to the soybean aphid in the soybean cultivar 'Dowling'. Crop Science 46:1601-1605. Höhn, H. and H. U. Höpli. 1990. Die Apfelrostmilbe: Oft überschätzt, aber kaum prognostizierbar! Schweizerische Zeitschrift für Obst- und Weinbau 126:259-266. Honda, K. 1995. Chemical basis of differential oviposition by lepidopterous insects. Archives of Insect Biochemistry and Physiology 30:1-23. Hull, L. and J. W. Grimm. 1983. Sampling schemes for estimating populations of the apple aphid (Aphis pomi) (Homoptera: Aphididae) on apple. Environmental Entomology 12:1581-1586. Hunter, M. D. and L. Hull. 1993. Variation in concentrations of phloridzin and phloretin in apple foliage. Phytochemistry 34:1251-1254. Idso, S. B. 1981. A set of equations for full spectrum and 8-14 µm and 10.5-12.5 µm thermal radiation from cloudless skies. Water Resource Research 17:295-304. Jackson, D. M. 1979. Codling moth egg distribution on unmanaged apple trees. Annals of the Entomological Society of America 72:361-368. Jackson, J. E. 2003. Biology of Apples and Pears. Cambridge University Press, Cambridge, UK. Jairin, J., K. Phengrat, S. Teangdeerith, A. Vanavichit, and T. Toojinda. 2007. Mapping of a broad-spectrum brown planthopper resistance gene, Bph3, on rice chromosome 6. Molecular Breeding 19:35-44. James, C. M., J. B. Clarke, and K. M. Evans. 2004. Identification of molecular markers linked to the mildew resistance gene PL-d in apple. Theoretical and Applied Genetics 110:175-181. Janick, J., J. N. Cummins, S. K. Brown, and M. Hemmat. 1996. Apples. Pages 1-77 in J. Janick and J. N. Moore, editors. Fruit Breeding: Tree and Tropical Fruits. John Wiley & Sons Ltd., New York. Jansson, R. K. and Z. Smilowitz. 1986. Influence of nitrogen on population parameters of potato insects: abundance, population growth, and within-plant distribution of the green peach aphid (Myzus persicae) (Homoptera: Aphididae). Environmental Entomology 15:49-55. Johnston, J. W., E. W. Hewett, N. H. Banks, F. R. Harker, and M. Hertog. 2001. Physical change in apple textue with fruit temperature: effects of cultivar and time in storage. Postharvest Biology and Technology 23:13-21. Juniper, B. E., R. Watkins, and S. A. Harris. 1999. The origin of apple. Acta Horticulturae 484:27-33. Kappel, F. and H. A. Quamme. 1993. Orchard training systems influence early canopy development and light microclimate within apple tree canopies. Canadian Journal of Plant Science 73:237-248. Kellerhals, M., L. Bertschinger, and C. Gessler. 2004. Use of genetic resources in apple breeding and for sustainable fruit production. Journal of Fruit and Ornamental Plant Research 12:53-62. Kellerhals, M. and B. Furrer. 1994. Approaches for breeding apples with durable disease resistance. Pages 93-97 in H. Schmidt and M. Kellerhals, editors. Progress in Temperate Fruit Breeding. Kluwer Academic Publishers, Dordrecht, NL. Kellerhals, M., W. Müller, L. Bertschinger, C. Darbellay, and W. Pfammatter. 1997. Obstbau. Landwirtschaftliche Lehrmittelzentrale, Zollikofen, CH. Kennedy, G. G. 1984. 2-tridecanone, tomatoes and Heliothis zea: potential incompatibility of plant antibiosis with insecticidal control. Entomologia Experimentalis et Applicata 35:305-311. Khan, M. A., B. Duffy, C. Gessler, and A. Patocchi. 2006. QTL mapping of fire blight resistance in apple. Molecular Breeding 17:299-306. Khan, M. A., C.-E. Durel, B. Duffy, D. Drouet, M. Kellerhals, C. Gessler, and A. Patocchi. 2007. Development of molecular markers linked to the 'Fiesta' linkage group 7 major QTL for fire blight resistance and their application for marker-assisted selection. Genome 50:568-577.

135 Khemira, H., T. L. Righetti, and A. N. Azarenko. 1998. Nitrogen partitioning in apple as affected by timing and tree growth habit. Journal of Horticultural Science and Biotechnology 73:217-223. Kim, K. C. and B. A. McPheron. 1993. Evolution of Insect Pests: Patterns of Variation. John Wiley & Sons Ltd., New York. King, G. J., J. R. Lynn, C. J. Dover, K. M. Evans, and G. B. Seymour. 2001. Resolution of quantitative trait loci for mechanical measures accounting for genetic variation in fruit texture of apple (Malus pumila Mill.). Theoretical and Applied Genetics 102:1227-1235. King, G. J., C. Maliepaard, J. R. Lynn, F. H. Alston, C.-E. Durel, K. M. Evans, B. Griffon, F. Laurens, A. G. Manganaris, E. Schrevens, S. Tartarini, and J. Verhaegh. 2000. Quantitative genetic analysis and comparison of physical and sensory descriptors relating to fruit flesh firmness in apple (Malus pumila Mill.). Theoretical and Applied Genetics 100:1074-1084. Klingler, J., R. Creasy, L. Gao, R. M. Nair, A. S. Calix, H. S. Jacob, O. R. Edwards, and K. B. Singh. 2005. Aphid resistance in Medicago truncatula involves antixenosis and phloem-specific, inducible antibiosis, and maps to a single locus flanked by NBS-LRR resistance gene analogs. Plant Physiology 137:1445- 1455. Knight, A. L. 2006. Assessing the mating status of female codling moth (Lepidoptera: Tortricidae) in orchards treated with sex pheromone using traps baited with ethyl (E, Z)-2,4-decadienoate. Environmental Entomology 35:894-900. Knight, A. L. and D. M. Light. 2001. Attractants from 'Bartlett' pear for codling moth (Cydia pomonella L.) larvae. Naturwissenschaften 88:339-342. Knight, A. L. and D. M. Light. 2005a. Developing action thresholds for codling moth (Lepidoptera: Tortricidae) with pear-ester-and codlemone baited traps in apple orchards treated with sex pheromone mating disruption. Canadian Entomologist 137:739-747. Knight, A. L. and D. M. Light. 2005b. Factors affecting the differential capture of male and female codling moth (Lepidoptera: Tortricidae) in traps baited with ethyl (E, Z)-2,4-decadienoate. Environmental Entomology 34:1161-1169. Kogan, M. 1994. Plant resistance in pest management. Pages 73-128 in R. L. Metcalf and W. H. Luckman, editors. Introduction to Insect Pest Management. John Wiley & Sons Ltd., New York. Kogan, M. 1998. Integrated pest management: historical perspectives and contemporary developments. Annual Review of Entomology 43:243-270. Kohler, G. R. and D. A. St. Clair. 2005. Variation for resistance to aphids (Homoptera: Aphididae) among tomato inbred backcross lines derived from wild Lycopersicon species. Journal of Economic Entomology 98:988-995. Komatsu, K., S. Okuda, M. Takahashi, R. Matsunaga, and Y. Nakazawa. 2005. QTL mapping of antibiosis to common cutworm (Spodoptera litura Fabricius) in soybean. Crop Science 45:2044-2048. Krakowsky, M. D., M. Lee, W. L. Woodman-Clikeman, M. J. Long, and N. Sharpova. 2004. QTL mapping of resistance to stalk tunneling by the European corn borer in RILs of maize population B73 x De811. Crop Science 44:274-282. Kührt, U., J. Samietz, and S. Dorn. 2006a. Effect of plant architecture and hail nets on temperature of codling moth habitats in apple orchards. Entomologia Experimentalis et Applicata 118:245-259. Kührt, U., J. Samietz, and S. Dorn. 2006b. Thermal response in adult codling moth. Physiological Entomology 31:80-88. Lacey, L. A. and D. I. Shapiro-Ilan. 2008. Microbial control of insect pests in temperate orchard systems: potential for incorporation into IPM. Annual Review of Entomology 53:121-144. Lander, E. S. and D. Botstein. 1989. Mapping mendelian factors unerlying quantitative traits using RFLP linkage maps. Genetics 121:185-199. Landsberg, J. J. 1973. Microclimate in an apple orchard. Journal of Applied Ecology 10:881-896. Larson, S. R., X. Wu, T. A. Jones, K. B. Jensen, N. J. Chatterton, B. L. Waldron, J. G. Robins, B. S. Baushman, and A. J. Palazzo. 2006. Comparative mapping of growth habit, plant height, and flowering QTLs in two interspecific families of Leymus. Crop Science 46:2526-2539. Laudert, D. 1998. Mythos Baum: Was Bäume uns Menschen bedeuten. Bayerischer Landwirtschaftverlag, BLV, München. Lauter, N. and J. Doebley. 2002. Genetic variation for phenotypically invariant traits detected in teosinte: implications for the evolution of novel forms. Genetics 160:333-342. Legendre, P. and L. Legendre. 1998. Numerical Ecology: Developments in Environmental Modelling 20. Elsevier, Amsterdam, NL. Leimu, R. and J. Koricheva. 2006. A meta-analysis of tradeoffs between plant tolerance and resistance to herbivores: combining the evidence from ecological and agricultural studies. Oikos 112:1-9. Lespinasse, Y., J. M. Olivier, J. M. Lespinasse, and M. LeLezec. 1985. 'Florina-Quérina': la résistance du pommier à la tavelure. Arboriculture Fruitière 378:43-47.

136 Levine, N. 2007. CrimeStat: A Spatial Program for the Analysis of Crime Incident Locations (v3.1). Ned Levine & Associates and National Institute of Justice, Houston and Washington DC. Liebhard, R., L. Gianfranceschi, B. Koller, R. C. D., R. Tarchini, E. van de Weg, and C. Gessler. 2002. Development and characterisation of 140 new microsatellites in apple (Malus x domestica Borkh.). Molecular Breeding 10:217-241. Liebhard, R., M. Kellerhals, W. Pfammatter, M. Jermini, and C. Gessler. 2003a. Mapping quantitative physiological traits in apple (Malus x domestica Borkh.). Plant Molecular Biology 52:511-526. Liebhard, R., B. Koller, L. Gianfranceschi, and C. Gessler. 2003b. Creating a saturated reference map for the apple (Malus x domestica Borkh.) genome. Theoretical and Applied Genetics 106:1497-1508. Liebhard, R., B. Koller, A. Patocchi, M. Kellerhals, W. Pfammatter, M. Jermini, and C. Gessler. 2003c. Mapping quantitative field resistance against apple scab in a 'Fiesta' x 'Discovery' progeny. Phytopathology 93:493-501. List, G. M. and J. H. Newton. 1921. Codling moth control for certain sections of Colorado. Colorado Agricultural Experiment Station Bulletin 1:268. Liu, B. H. 1998. Statistical Genomics: Linkage, Mapping, and QTL Analysis. CRC Press, Boca Raton, NY. Lombarkia, N. and S. Derridj. 2002. Incidence of apple and leaf surface metabolits on Cydia pomonella L. (Lepidoptera, Tortricidae) oviposition. Entomologia Experimentalis et Applicata 104:79-87. Lowery, D. T., M. J. Smirle, R. G. Foottit, C. L. Zurowski, and E. H. B. Peryea. 2005. Baseline susceptibilities to imidacloprid for green apple aphid and spirea aphid (Homoptera : Aphididae) collected from apple in the pacific Northwest. Journal of Economic Entomology 98:188-194. Luby, J. J. 2003. Taxonomic classification and brief history. Pages 1-15 in D. C. Ferree and I. J. Warrington, editors. Apples: Botany, Production and Uses. CABI Publishing, Wallingford, UK. Ludwig, J. A. and J. F. Reynolds. 1988. Statistical Ecology: A Primer on the Methods and Computing. John Wiley & Sons Ltd., New York. Lyth, M. 1985. Hypersensitivity in apple to feeding by Dysaphis plantaginea: effects on aphid biology. Annals of Applied Biology 107:155-161. Maag, R. 1996. Integrierte Produktion im Obstbau. Dr. R. Maag AG, Dielsdorf, CH. Maier, C. T. 1992. Seasonal development, flight activity, and density of Sympiesis marylandensis (Hymenoptera: Eulophidae), a parasitoid of leaf-mining Phyllonorycter spp. (Lepidoptera: Gracillariidae), in Connecticut apple orchards and forests. Environmental Entomology 21:164-172. Maindonald, J. H., B. C. Waddell, and R. J. Petry. 2001. Apple cultivar effects on codling moth (Cydia pomonella) (Lepidoptera: Tortricidae) egg mortality following fumigation with methyl bromide. Postharvest Biology and Technology 22:99-110. Maliepaard, C., F. H. Alston, G. van Arkel, L. M. Brown, E. Chevreau, F. Dunemann, K. M. Evans, S. Gardiner, P. Guilford, A. W. van Heusden, J. Janse, F. Laurens, J. R. Lynn, A. G. Manganaris, A. P. M. den Nijs, N. Periam, E. Rikkerink, P. Roche, C. Ryder, S. Sansavini, H. Schmidt, S. Tartarini, J. J. Verhaegh, M. Vrielink-van Ginkel, and G. J. King. 1998. Aligning male and female linkage maps of apple (Malus pumila Mill.) using multi-allelic markers. Theoretical and Applied Genetics 97:60-73. Mani, E., T. Wildbolz, W. Riggenbach, and H. Staub. 1997. Population dynamics of the codling moth (Cydia pomonella L.) in undisturbed apple orchards of eastern Switzerland. Mitteilungen der Schweizerischen Entomologischen Gesellschaft 70:117-132. Manly, K. F. and J. M. Olson. 1999. Overview of QTL mapping software and introduction to map manager QTL. Mammalian Genome 10:327-334. Massonie, G., P. Maison, J. C. Meymerit, and Y. Lespinase. 1981. Considérations sur la résistence par hypersensibilié au puceron cendré du pommier (Dysaphis plantaginea Pass.). IOBC/wprs Bulletin 4:67- 71. Maxwell, F. G. 1985. Utilization of host plant resistance in pest management. Insect Science and its Application 6:437-442. McCornack, B. P., D. W. Ragsdale, and R. C. Venette. 2004. Demography of soybean aphid (Homoptera: Aphididae) at summer temperatures. Journal of Economic Entomology 97:854-861. McLellan, C. R. 1962. Mortality of codling moth eggs and young larvae in an integrated control orchard. Canadian Entomologist 94:655-666. Mehinagic, E., G. Royer, R. Symoneaux, F. Jourjon, and C. Prost. 2006. Characterization of odor-active volatiles in apples: influence of cultivars and maturity stage. Journal of Agricultural and Food Chemistry 54:2678-2687. Mensah, R. K. and J. L. Madden. 1991. Resistance and susceptibility of Boronia megastigma cultivars to infestations by the psyllid Ctenarytaina thysanura. Entomologia Experimentalis et Applicata 61:189- 198.

137 Miñarro, M., J. L. Hemptinne, and E. Dapena. 2005. Colonization of apple orchards by predators of Dysaphis plantaginea: sequential arrival, response to prey abundance and consequences for biological control. BioControl 50:403-414. Miyasaka, S. C., J. D. Hansen, T. G. McDonald, and G. K. Fukumoto. 2007. Effects of nitrogen and potassium in kukuyu grass on feeding by yellow sugarcane aphid. Crop Protection 26:511-517. Mody, K., D. Eichenberger, and S. Dorn. 2008. Stress magnitude matters: different intensities of pulsed water stress produce non-monotonic resistance responses of host plants to insect herbivores. Ecological Entomology doi:10.1111/j.1365-2311.2008.01053.x. Mody, K. and K. E. Linsenmair. 2004. Plant-attracted ants affect arthropod community structure but not necessarily herbivory. Ecological Entomology 29:217-225. Mohan, M., S. Nair, A. Bhagwat, T. G. Krishna, M. Yano, C. R. Bhatia, and T. Sasaki. 1997a. Genome mapping, molecular markers and marker-assisted selection in crop plants. Molecular Breeding 3:87- 103. Mohan, M., P. V. Sathyanarayanan, A. Kumar, M. N. Srivastava, and S. Nair. 1997b. Molecular mapping of a resistance-specific PCR-based marker linked to a gall midge resistance gene Gm4t in rice. Theoretical and Applied Genetics 95:777-782. Montgomery, M. E. and H. Arn. 1974. Feeding response of Aphis pomi, Myzus persicae, and Amphorophora agathonica to phloridzin. Journal of Insect Physiology 20:413-421. Moreira, L. A., C. Mollema, and S. van Heusden. 1999. Search for molecular markers linked to Liriomyza trifolii resistance in tomato. Euphytica 109:149-156. Morgan, J. and A. Richards. 1993. The Book of Apples. Ebury Press, London. Morgante, M. and F. Salamini. 2003. From plant genomics to breeding practice. Current Opinion in Biotechnology 14:214-219. Myburg, A. A., M. Cawood, B. D. Wingfield, and A. M. Botha. 1998. Development of RAPD and SCAR markers linked to the Russian wheat aphid resistance gene Dn2 in wheat. Theoretical and Applied Genetics 96:1162-1169. Neilsen, G. H. and D. Neilsen. 2003. Nutritional requirements of apple. Pages 267-302 in D. C. Ferree and I. J. Warrington, editors. Apples: Botany, Production and Uses. CABI Publishing, Wallingford, UK. Newman, J. A. 2005. Climate change and the fate of cereal aphids in Southern Britain. Global Change Biology 11:940-944. O'Kennedy, B. T., M. J. Hennerty, and J. S. Titus. 1975. Changes in the nitrogen reserves of apple shoots during the dormant season. Journal of Horticultural Science 50:321-329. O'Rourke, D. 2003. World production, trade, consumption and economic outlook for apples. Pages 15-29 in D. C. Ferree and I. J. Warrington, editors. Apples: Botany, Production and Uses. CABI Publishing, Wallingford, UK. Oatman, E. R. and E. F. Legner. 1961. Bionomics of the apple aphid (Aphis pomi) on young non-bearing apple trees. Journal of Economic Entomology 54:1034-1037. Oerke, E. C., H. W. Dehne, F. Schonbeck, and A. Weber. 1994. Crop Production and Crop Protection: Estimated Losses in Major Food and Cash Crops. Elsevier, Amsterdan, NL. Omkar, J. and A. Pervez. 2005. Ecology of two-spotted ladybird (Adalia bipunctata): a review. Journal of Applied Entomology 129:465-474. Oppenheimer, C. 1962. Breeding of apples for a sub-tropical climate. Proceedings of the 16th International Horticulture Congress in Brussels 3:18-24. Patocchi, A., M. Walser, S. Tartarini, G. A. L. Broggini, F. Gennari, S. Sansavini, and C. Gessler. 2005. Identification by genome scanning approach (GSA) of a microsatellite tightly associated with the apple scab resistance gene Vm. Genome 48:630-636. Pedigo, L. P. 2006. Entomology and Pest Management. Prentice Hall, Saddle River, NY. Pimentel, D. 1997. Pest management in agriculture. Pages 1-11 in D. Pimentel, editor. Techniques for Reducing Pesticide Use. John Wiley & Sons Ltd., New York. Plourde, D. F., H. F. Goonewardene, and W. F. Kwolek. 1985. Pubescence as a factor in codling moth, oviposition, and fruit entry in five apple selections. Hortscience 20:82-84. Prokopy, R. J. and B. A. Croft. 1994. Apple insect pest management. Pages 543-585 in R. L. Metcalf and W. H. Luckman, editors. Introduction to Insect Pest Management. John Wiley & Sons Ltd., New York. Prokopy, R. J., C. B. With, and T. C. Leskey. 1999. Movement of plum curculio adults toward host trees and traps: flight versus walking. Entomologia Experimentalis et Applicata 91:385-392. Qubbaj, T., A. Reineke, and C. P. W. Zebitz. 2005. Molecular interactions between rosy apple aphids (Dysaphis plantaginea) and resistant and susceptible cultivars of its primary host Malus domestica. Entomologia Experimentalis et Applicata 115:145152.

138 Quiros, C. F., M. A. Stevens, C. M. Rick, and M. L. Kokyokomi. 1977. Resistance in tomato to pink form of potato aphid (Macrosiphum euphorbiae Thomas): role of anatomy, epidermal hairs, and foliage composition. Journal of the American Society for Horticultural Science 102:166-171. Ramaswamy, S. B. 1988. Host finding by moths: sensory modalities and behaviors. Journal of Insect Physiology 34:235-249. Rat-Morris, E. 1993. Development of rosy aphid (Dysaphis plantaginea) on a tolerant apple cultivar 'Florina'. IOBC/wprs Bulletin 16:91-100. Rat-Morris, E., S. Crowther, and M. Guessoum. 1999. Resistance-breaking biotypes of rosy apple aphid (Dysaphis plantaginea) on resistant cultivar 'Florina'. IOBC/wprs Bulletin 22:71-75. Reed, H. C. and P. J. Landolt. 2002. Attraction of mated female codling moths (Lepidoptera: Tortricidae) to apples and apple odor in a flight tunnel. Florida Entomologist 85:324-328. Reissig, W. H., R. W. Weires, and C. G. Forshey. 1982. Effects of gracillariid leafminers on apple tree growth and production. Environmental Entomology 11:958-963. Reyes, M., P. Franck, P. J. Charmillot, C. Ioriatti, J. Olivares, E. Pasqualini, and B. Sauphanor. 2007. Diversity of insecticide resistance mechanisms and spectrum in European populations of the codling moth (Cydia pomonella). Pest Management Science 63:890-902. Risch, S. J. 1987. Agricultural ecology and insect outbreaks. Pages 217-238 in P. Barbosa and J. C. Schultz, editors. Insect Outbreaks. Academic Press, San Diego, CA. Roche, P., F. H. Alston, C. Maliepaard, K. M. Evans, R. Vrielink, F. Dunemann, T. Markussen, S. Tartarini, L. M. Brown, C. Ryder, and G. J. King. 1997. RFLP and RAPD markers linked to the rosy leaf curling aphid resistance gene (Sd1) in apple. Theoretical and Applied Genetics 94:528-533. Rossi, M., F. L. Goggin, M. S. B., I. Kaloshian, D. E. Ullman, and V. Williamson. 1998. The nematode resistance gene Mi of tomato confers resistance against the potato aphid. Proceedings of the National Academy of Sciences of the United States of America 95:9750-9754. Russ, K. 1961. Einfluss wichtiger Witterungsfaktoren auf die Flugtätigkeit des Apfelwicklers (Carpocapsa pomonella L.). Pflanzenschutz Berichte 22:67-82. Sabelis, M., A. Janssen, A. Pallini, M. Venzon, J. Bruin, B. Drukker, and P. Scutareanu. 1999. Behavioral responses of predatory and herbivorous arthropods to induced plant volatiles: from evolutionary ecology to agricultural applications. Pages 269-296 in A. A. Agrawal, S. Tuzun, and E. Bent, editors. Induced Plant Defenses Against Pathogens and Herbivores: Biochemistry, Ecology and Agriculture. APS Press, St. Paul, MN. Sadasivam, S. and B. Thayumanavan. 2003. Molecular Host-Plant Resistance to Pests. Marcel Dekker Inc., New York. Saethre, M. G. and T. Hofsvang. 2002. Effect of temperature on oviposition behavior, fecundity, and fertility in two northern European populations of the codling moth (Lepidoptera: Tortricidae). Environmental Entomology 31:804-815. Schoonhoven, L. M., T. Jermy, and J. J. A. van Loon. 1998. Insect-Plant Biology. Chapman & Hall, London. Schoonhoven, L. M., J. J. A. van Loon, and M. Dicke. 2005. Insect-Plant Biology. Oxford University Press, Oxford, UK. Segura, V., C. Cilas, F. Laurens, and E. Costes. 2006. Phenotyping progenies for complex architectural traits: a strategy for 1-year-old apple trees (Malus x domestica Borkh.). Tree Genetics & Genomes 2:140-151. Sellers, P. J., R. E. Dickinson, D. A. Randall, A. K. Betts, F. G. Hall, J. A. Berry, G. J. Collat, A. S. Denning, H. A. Mooney, C. A. Nobre, N. Sato, C. B. Field, and A. Henderson-Sellers. 1997. Modeling the exchange of energy, water, and carbon between continents and the atmosphere. Science 275:502-509. Selvaraju, K., P. Shanmugasundaram, S. Mohankumar, M. Asaithambi, and R. Balasaraswathi. 2007. Detection of quantitative trait locus for leaffolder (Cnaphalocrocis medinalis Guenee) resistance in rice on linkage group 1 based on damage score and flag leaf width. Euphytica 157:35-43. Servin, B., O. C. Martin, M. Mezard, and F. Hospital. 2004. Toward a theory of marker-assisted gene pyramiding. Genetics 168:513-523. Sharma, H. C., K. K. Sharma, and J. H. Crouch. 2004. Genetic transformation of crops for insect resistance: potential and limitations. Critical Reviews in Plant Sciences 23:47-72. Shockley, F. W., E. A. Backus, M. R. Ellersieck, D. W. Johnson, and M. McCaslin. 2002. Glandular-haired alfalfa resistance to potato leafhopper (Homoptera: Cicadellidae) and hopperburn: development of resistance indices. Journal of Economic Entomology 95:437-447. Silfverberg-Dilworth, E., C. L. Matasci, W. E. van de Weg, M. P. W. van Kaauwen, M. Walser, L. P. Kodde, V. Soglio, L. Gianfranceschi, C.-E. Durel, F. Costa, T. Yamamoto, B. Koller, C. Gessler, and A. Patocchi. 2006. Microsatellite markers spanning the apple (Malus x domestica Borkh.) genome. Tree Genetics & Genomes 2:202-224.

139 Simon, S., P. E. Lauri, L. Brun, H. Defrance, and B. Sauphanor. 2006. Does manipulation of fruit-tree architecture affect the development of pests and pathogens: a case study in an organic apple orchard. Journal of Horticultural Science & Biotechnology 81:765-773. Sircelj, H., M. Tausz, D. Grill, and F. Batic. 2005. Biochemical responses in leaves of two apple tree cultivars subjected to progressing drought. Journal of Plant Physiology 162:1308-1318. Smith, C. M. 2005a. Factors affecting the expression of plant resistance to arthropods. Pages 183-217 in C. M. Smith, editor. Plant Resistance to Arthropods: Molecular and Conventional Approaches. Springer, Dordrecht, NL. Smith, C. M. 2005b. Introduction. Pages 1-18 in C. M. Smith, editor. Plant Resistance to Arthropods: Molecular and Conventional Approaches. Springer, Dordrecht, NL. Smith, C. M. and S. S. Quisenberry. 1994. The value and use of plant resistance to insects in integrated crop management. Journal of Agricultural Entomology 11:189-190. Sokal, R. R. and F. J. Rohlf. 1995. Biometry. Freeman, San Francisco, CA. Sokolov, A. M. 1965. Resistance of apple to green apple and rosy leaf-curling aphids. 4th All-Union Conference on Immunity of Agricultural Plants, Vines and Fruit Crops in Kisinev 1:149-155. Southwood, T. R. E. and P. A. Henderson. 2000. Ecological Methods. Blackwell Scientific Publications, Oxford, UK. Spieser, F., B. Graf, P. Walther, and J. Noesberger. 1998. Impact of apple rust mite (Acari: Eriophyiidae) feeding on apple leaf gas exchange and leaf color associated with changes in leaf tissue. Environmental Entomology 27:1149-1156. Starks, K. J., R. D. Eikenbar, and R. Muniappa. 1972. Interaction between plant resistance and parasitism against greenbug Hemiptera (Homoptera: Aphididae) on barley and sorghum. Annals of the Entomological Society of America 65:650-655. Stekolshchikov, A. V. and A. L. Lobanov. 2004. Dysaphis devecta (Walker) species complex (Homoptera: Aphididae): morphological characteristics. Entomological Review c/c of Entomologicheskoe Obozrenie 83:115-139. Stewart-Jones, A., T. W. Pope, J. D. Fitzgerald, and G. M. Poppy. 2008. The effect of ant attendance on the success of rosy apple aphid populations, natural enemy abundance and apple damage in orchards. Agricultural and Forest Entomology 10:37-43. Stoeckli, S., K. Mody, and S. Dorn. 2008a. Aphis pomi population development, shoot characteristics, and antibiosis resistance in different apple genotypes. Journal of Economic Entomology 101:1341-1348. Stoeckli, S., K. Mody, and S. Dorn. 2008b. Influence of canopy aspect and height on codling moth (Lepidoptera: Tortricidae) larval infestation in apple, and relationship between infestation and fruit size. Journal of Economic Entomology 101:81-89. Stoeckli, S., K. Mody, C. Gessler, A. Patocchi, M. Jermini, and S. Dorn. 2008c. QTL analysis for aphid resistance and growth traits in apple. Tree Genetics & Genomes 4:833-847. Stoeckli, S., K. Mody, A. Patocchi, M. Kellerhals, and S. Dorn. 2008d. Rust mite resistance in apple assessed by quantitative trait loci analysis. Tree Genetics & Genomes doi:10.1007/s11295-11008-10186-11295. Stoeckli, S., K. Mody, C. Gessler, D. Christen, and S. Dorn. 2009. QTL mapping of resistance in apple to Cydia pomonella and Lyonetia clerkella, and of two selected fruit traits. Annals of Applied Biology: in print. Stoner, K. A. 1990. Glossy leaf wax and plant resistance to insects in Brassica oleracea under natural infestation. Environmental Entomology 19:730-739. Strong, D. R., J. H. Lawton, and T. R. E. Southwood. 1984. Insects on Plants: Community Patterns and Mechanisms. Blackwell Scientific Publications, Oxford, UK. Sturm, K., M. Hudina, A. Solar, M. V. Marn, and F. Stampar. 2003. Fruit quality of different 'Gala' clones. European Journal of Horticultural Science 68:169-175. Sutherland, O. R. W., C. H. Wearing, and R. F. N. Hutchins. 1977. Production of α-farnesene, an attractant and oviposition stimulant for codling moth, by developing fruit of 10 varieties of apple. Journal of Chemical Ecology 3:625-631. Takemura, M., R. Nishida, N. Mori, and Y. Kuwahara. 2002. Acylated flavonol glycosides as probing stimulants of a bean aphid (Megoura crassicauda) from Vicia angustifolia. Phytochemistry 61:135-140. Tanksley, S. D., N. D. Young, A. H. Paterson, and M. W. Bonierbale. 1989. RFLP mapping in plant breeding: new tools for an old science. Bio/Technology 7:257-264. Tingey, W. M. and S. R. Singh. 1980. Environmental factors influencing the magnitude and expression of resistance. Pages 87-115 in F. G. Maxwell and P. R. Jennings, editors. Breeding Plants Resistant to Insects. John Wiley & Sons Ltd., New York. Trimble, R. M. and A. M. El-Sayed. 2005. Potential of ethyl (2E, 4Z)-2,4-decadienoate for monitoring activity of codling moth (Lepidoptera: Tortricidae) in eastern North America apple orchards. Canadian Entomologist 137:110-116.

140 Unsicker, S. B. and K. Mody. 2005. Influence of tree species and compass bearing on insect folivory of nine common tree species in the West African savanna. Journal of Tropical Ecology 21:227-231. Vallat, A. and S. Dorn. 2005. Changes in volatile emission from apple trees and associated response of adult female codling moths over the fruit-growing season. Journal of Agricultural and Food Chemistry 53:4083-4090. Vallat, A., H. Gu, and S. Dorn. 2005. How rainfall, relative humidity and temperature influence volatile emissions from apple trees in situ. Phytochemistry 66:1540-1550. van de Weg, W. E., R. E. Voorrips, R. Finkers, L. P. Kodde, J. Jansen, and M. C. A. M. Bink. 2004. Pedigree genotyping a new pedigree-based approach of QTL identification and allele mining. Acta Horticulturae 663:45-50. van Emden, H. F. 1997. Host plant resistance to insect pests. Pages 129-152 in D. Pimentel, editor. Techniques for Reducing Pesticide Use. John Wiley & Sons Ltd., New York. van Loon, J. J. A., C. Z. Wang, J. K. Nielsen, R. Gols, and Y. T. Qiu. 2002. Flavonoids from cabbage are feeding stimulants for diamondback moth larvae additional to glucosinolates: chemoreception and behaviour. Entomologia Experimentalis et Applicata 104:27-34. van Ooijen, J. W. 1999. LOD significance thresholds for QTL analysis in experimental populations of diploid species. Heredity 83:613-624. van Ooijen, J. W., M. P. Boer, R. C. Jansen, and C. Maliepaard. 2002. MapQTL® 4.0: Software for the Calculation of QTL Positions on Genetic Maps. Plant Research International, Wageningen, NL. Vanin, I. I. 1965. Resistance of apple trees to disease and pests. 4th All-Union Conference on Immunity of Agricultural Plants, Vines and Fruit Crops in Kisinev:131-132. Varshney, A., T. Mohapatra, and R. P. Sharma. 2004. Molecular Mapping and Marker Assisted Selection of Traits for Crop Improvement. Anamaya Publishers, New Delhi, IN. Veberic, R., P. Zadravec, and F. Stampar. 2007. Fruit quality of 'Fuji' apple (Malus domestica Borkh.) strains. Journal of the Science of Food and Agriculture 87:593-599. Velten, G., A. S. Rott, C. Cardona, and S. Dorn. 2007. The inhibitory effect of the natural seed storage protein arcelin on the development of Acanthoscelides obtectus. Journal of Stored Products Research 43:550- 557. Venables, W. N. and B. D. Ripley. 2002. Modern Applied Statistics with S-Plus. Springer, New York. Verhoeven, K. J. F., L. Simonsen, and L. M. McIntyre. 2005. Implementing false discovery rate control: increasing your power. Oikos 108:643-647. Vidale, P. L. and R. Stöckli. 2005. Prognostic canopy air space solutions for land surface exchanges. Theoretical and Applied Climatology 80:245-257. Vinatzer, B. A., A. Patocchi, L. Gianfranceschi, S. Tartarini, H. B. Zhang, C. Gessler, and S. Sansavini. 2001. Apple contains receptor-like genes homologous to the Cladosporium fulvum resistance gene family of tomato with a cluster of genes cosegregating with Vf apple scab resistance. Molecular Plant-Microbe Interactions 14:508-515. Volz, R. K., I. B. Ferguson, E. W. Hewett, and D. J. Woolley. 1994. Wood age and leaf area influence fruit size and mineral composition of apple fruit. Journal of Horticultural Science 69:385-395. Volz, R. K., F. R. Harker, and S. Lang. 2003. Firmness decline of 'Gala' apple during fruit development. Journal of the American Society of Horticultural Science 128:797-802. Walde, S. J., J. M. Hardmann, and C. N. Magagula. 1997. Direct and indirect species interactions influencing within season dynamics of apple rust mite (Aculus schlechtendali) (Acari, Eriophyidae). Experimental and Applied Acarology 21:587-614. Wan, Q., Z. S. Zhang, M. Hu, L. Chen, D. J. Liu, X. Chen, W. Wang, and J. Zheng. 2007. T1 locus in cotton is the candidate gene affecting lint percentage, fiber quality and spiny bollworm (Earias spp.) resistance. Euphytica 158:241-247. Wang, J. J. and J. H. Tsai. 2000. Effect of temperature on the biology of Aphis spiraecola (Hom. Aphididae). Annals of the Entomological Society of America 93:874-883. Wearing, C. H., K. Colhoun, G. F. McLaren, B. Attfield, and V. G. M. Bus. 2003. Evidence for single gene resistance in apple to brownheaded leafroller (Ctenopseustis obliquata) and implications for resistance to other New Zealand leafrollers. Entomologia Experimentalis et Applicata 108:1-10. Webster, J. A., C. Inayatullah, M. Hamissou, and K. A. Mirkes. 1994. Leaf pubescence effects in wheat on yellow sugarcane aphids and greenbugs (Homoptera, Aphididae). Journal of Economic Entomology 87:231-240. Westigard, P. H., L. Gentner, and B. A. Butt. 1975. Codling moth: egg and first instar mortality of pear with special reference to varietal susceptibility. Environmental Entomology 5:51-54. Whitaker, P. M., D. L. Mahr, and M. Clayton. 2006. Verification and extension of a sampling plan for apple aphid (Aphis pomi DeGeer) (Hemiptera: Aphididae). Environmental Entomology 35:488-496. White, T. C. R. 1993. The Inadequate Environment: Nitrogen and the Abundance of Animals. Springer, Berlin.

141 Wood, T. G. 1965. Field observations on flight and oviposition of codling moth and mortality of eggs and first- instar larvae in an integrated control orchard. New Zealand Journal of Agricultural Research 8:1043- 1059. Woodside, A. M. 1944. Codling moth infestation at different heights in apple trees. Virginia Agricultural Experimental Station Bulletin 1:360. Wu, Y. Q., Y. H. Huang, D. R. Porter, C. G. Tauer, and L. Hollaway. 2007. Identification of a major quantitative trait locus conditioning resistance to greenbug biotype E in sorghum PI 550610 using simple sequence repeat markers. Journal of Economic Entomology 100:1672-1678. Yan, F. M., M. Bengtsson, and P. Witzgall. 1999. Behavioral response of female codling moths (Cydia pomonella) to apple volatiles. Journal of Chemical Ecology 25:1343-1351. Yencho, G. C., M. B. Cohen, and P. F. Byrne. 2000. Applications of tagging and mapping insect resistance loci in plants. Annual Revue of Entomology 45:393-422. Zar, J. H. 1999. Biostatistical Analysis. Prentice-Hall, Upper Saddle River, NJ. Zehnder, G., G. M. Gurr, S. Kühne, M. R. Walde, S. D. Wratten, and E. Wyss. 2007. Arthropod pest management in organic crops. Annual Review of Entomology 52:57-80.

142 11. Acknowledgements

I would like to thank the many people who made this work possible for their scientific and personal support during the last three years. In particular I wish to mention: Prof. Dr. Silvia Dorn, Professor of Applied Entomology at the Institute of Plant Sciences (ETH Zurich), for the opportunity to work on host-plant resistance and arthropod pests. I highly appreciated the excellent scientific and personal support. The provision of financial contribution and essential infrastructure, and the constructive comments on the manuscripts are of high value within a PhD. Dr. Karsten Mody, Research Associate in the group of Applied Entomology at the Institute of Plant Sciences (ETH Zurich), for supervising my work with overwhelming enthusiasm, patience and competence. I warmly thank him for passing to me his enormous knowledge about field experiments, statistics, and the discussion of scientific results. I awesome look up to him how he manages work-life balance between scientific projects and his young family. It was a great pleasure to learn from his straightforward character. Prof. Dr. Cesare Gessler, Titulary Professor in the group of Plant Pathology at the Institute of Integrative Biology (ETH Zurich), for the collaboration and the support with QTL analysis and manuscript writing. I learned a lot about the genetic basis of host-plant resistance in apple to serious diseases. Dr. Andrea Patocchi (Agroscope Changins-Wadenswil) for the enormous help in QTL analysis and for answering questions about molecular markers in apple. Dr. Markus Kellerhals (Agroscope Changins-Wadenswil) who kindly agreed to be one of my co-examiners and for the opportunity to carry out the field study in the 'Sandhof' apple orchard at the Zurich site. Dr. Mauro Jermini and Dr. Danilo Christen (Agroscope Changins-Wadenswil) for providing apple orchards at the Ticino and Valais site. Heinrich Höhn (Agroscope Changins-Wadenswil), and Joel Meier and Hansjoerg Kull (Syngenta Crop Protection Dielsdorf) for helpful comments on herbivore survey. Dr. Giovanni Broggini and Dr. Awais Khan (ETH Zurich); and Dr. Hans Jansen (Plant Research Wageningen) for their help in QTL analysis. Dr. Hans-Rudolf Roth, Dr. Werner Eugster, Rahel Liesch, and Massimo Merlini (ETH Zurich) for statistical consulting.

143 Prof. Dr. Stephanie Bloem (USDA-APHIS-PPQ-CPHST), Dr. Ute Kührt (Oregon State University), Dr. Kathrin Tschudi-Rein, and Dr. Adriana Najar-Rodriguez and Dr. Maohua Chen (ETH Zurich) for helpful comments on manuscripts and English proofreading. Sarah Fässler (diploma student) and Mark Lendenmann (semester student) for the preliminary results on host-plant resistance to common herbivore pests. I appreciated the opportunity to co-supervise their work. Sarah and Mark did a great job in the field, together with Christophe Rohrer and Michelle Schmocker. I would like to thank Dr. Claude Fornallaz (ETH Zurich) for the great support with technical problems, and solving all kinds of computer difficulties - also in urgent cases. The actual and former group members in Applied Entomology for the great time I had during last years. A special thanks goes to Tanja Christoffel for help with growing apple seedlings, and Claudia Reichle, Conni Frick, Elisabeth Holenstein, and Brigitte Dorn for administrative help. I would like to mention my office mates Antonia Zurbuchen, Bettina Gutbrodt, and Mirco Plath. I highly appreciated our scientific and personal communications. Kathrin Bühler, Thomas Fischer, Iris Spörri, Thomas Werner Alexandre Gouskov, Malgorzata and Jan Taralzcack, Kathrin Grünig, Daniel Margadant, Beat Schuler, and Monika Tobler for strong company during the many 'ETH years'. I thank my parents for their love and their tremendous support. Without their financial aid and believe in me it would not have been possible to carry out a PhD thesis. It was a pleasure to go sometimes to Lucerne after long field days from 6 a.m. to 10 p.m. with a snarling stomach. Finally, I would mention my brother Reto Stöckli and his wife Saskia Bourgeois. It is not possible to describe their contribution to my personal and scientific skills in words. It was highly inspiring to discuss about temperature models, climate change, wind roses, and - energy.

144 12. Curriculum vitae

Sibylle Carmen Stöckli

Born on October 2nd 1976, in Santiago de Chile Citizen of Luthern (LU), Switzerland

2005-2008 Assistant and PhD student at the ETH Zurich (Switzerland) Institute of Plant Sciences / Applied Entomology Doctoral supervisor: Prof. Dr. S. Dorn, Adviser: Dr. K. Mody Thesis title: Host plant resistance in apple (Malus x domestica Borkh.) to common herbivore pests

2003-present Diploma in secondary higher education at the ETH Zurich Department of Biology

2003-present Teaching informal education in ornithology Swiss Association for the Protection of Birds, Zurich

1998-2003 MSc student in Biology the ETH Zurich Department of Biology

Thesis title: Territoriality, nest site selection and breeding biology of Skylarks (Alauda arvensis) in territories with different crop diversity. Supervisors: Prof. Dr. P. Edwards, Dr. M. Jenny, and Dr. R. Spaar

1997-1998 Language courses (Great Britain and Bolivia) 1990-1997 Gymnasium (Lucerne, Switzerland) 1983-1990 Primary School (Brazil and Switzerland)

145