京都大学 / 新制・課程博士 / 博士(理学) / 甲第23744号 / 理博第4834号 / 新制||理||1691(附属図書館) / 京都大学大学院理学研究科生物科学専攻 / (主査)教授 工藤 洋, 准教授 本庄 三恵, 教授 松下 智直 / 学位規則第4条第1項該当 / Doctor of Science / Kyoto University / DGAM

No description available.

Animals and plants in temperate regions must adapt their life cycle to pronounced seasonal variation. The research effort that has gone into studying these cyclical life history events, or phenological traits, has increased greatly in recent decades. As phenological traits are often correlated to temperature, they are relevant to study in terms of understanding the effect of short term environmental variation as well as long term climate change. Because of this, changes in phenology are the most obvious and among the most commonly reported responses to climate change. Moreover, phenological traits are important for fitness as they determine the biotic and abiotic environment an individual encounters. Fine-tuning of phenology allows for synchronisation at a local scale to mates, food resources and appropriate weather conditions. On a between-population scale, variation in phenology may reflect regional variation in climate. Such differences can not only give insights to life cycle adaptation, but also to how populations may respond to environmental change through time. This applies both on an ecological scale through phenotypic plasticity as well as an evolutionary scale through genetic adaptation. In this thesis I have used statistical and experimental methods to investigate both the larger geographical patterns as well as mechanisms of fine-tuning of phenology of several butterfly species. The main focus, however, is on the orange tip butterfly, Anthocharis cardamines, in Sweden and the United Kingdom. I show a contrasting effect of spring temperature and winter condition on spring phenology for three out of the five studied butterfly species. For A. cardamines there are population differences in traits responding to these environmental factors between and within Sweden and the UK that suggest adaptation to local environmental conditions. All populations show a strong negative plastic relationship between spring temperature and spring phenology, while the opposite is true for winter cold duration. Spring phenology is shifted earlier with increasing cold duration. The environmental variables show correlations, for example, during a warm year a short winter delays phenology while a warm spring speeds phenology up. Correlations between the environmental variables also occur through space, as the locations that have long winters also have cold springs. The combined effects of these two environmental variables cause a complex geographical pattern of phenology across the UK and Sweden. When predicting phenology with future climate change or interpreting larger geographical patterns one must therefore have a good enough understanding of how the phenology is controlled and take the relevant environmental factors in to account. In terms of the effect of phenological change, it should be discussed with regards to change in life cycle timing among interacting species. For example, the phenology of the host plants is important for A. cardamines fitness, and it is also the main determining factor for oviposition. In summary, this thesis shows that the broad geographical pattern of phenology of the butterflies is formed by counteracting environmental variables, but that there also are significant population differences that enable fine-tuning of phenology according to the seasonal progression and variation at the local scale. / <p>At the time of the doctoral defense, the following paper was unpublished and had a status as follows: Paper 4: Manuscript.</p>

Increasing temperatures resulting from climate change have within recent years been shown to advance phenological events in a large number of species worldwide. Species can differ in their response to increasing temperatures, and understanding the mechanisms that determine the response is therefore of great importance in order to understand and predict how a warming climate can influence both individual species, but also their interactions with each other and the environment. Understanding the mechanisms behind responses to increasing temperatures are however largely unexplored. The selected study system consisting of host plant species of the Brassicaceae family and their herbivore Anthocharis cardamines, is assumed to be especially vulnerable to climatic variations. Through the use of this study system, the aim of this thesis is to study differences in the effect of temperature on development to start of flowering within host plant species from different latitudinal regions (study I), and among host plant species (study II). We also investigate whether different developmental phases leading up to flowering differ in sensitivity to temperature (study II), and if small-scale climatic variation in spring temperature influence flowering phenology and interactions with A. cardamines (study III). Finally, we investigate if differences in the timing of A. cardamines relative to its host plants influence host species use and the selection of host individuals differing in phenology within populations (study IV). Our results showed that thermal reaction norms differ among regions along a latitudinal gradient, with the host plant species showing a mixture of co-, counter- and mixed gradient patterns (study I). We also showed that observed differences in the host plant species order of flowering among regions and years might be caused by both differences in the distribution of warm days during development and differences in the sensitivity to temperature in different phases of development (study II). In addition, we showed that small-scale variations in temperature led to variation in flowering phenology among and within populations of C. pratensis, impacting the interactions with the butterfly herbivore A. cardamines. Another result was that the less the mean plant development stage of a given plant species in the field deviated from the stage preferred by the butterfly for oviposition, the more used was the species as a host by the butterfly (study IV). Finally, we showed that the later seasonal appearance of the butterflies relative to their host plants, the higher butterfly preference for host plant individuals with a later phenology, corresponding to a preference for host plants in earlier development stages (study IV). For our study system, this thesis suggest that climate change will lead to changes in the interactions between host plants and herbivore, but that differences in phenology among host plants combined with changes in host species use of the herbivore might buffer the herbivore against negative effects of climate change. Our work highlights the need to understand the mechanisms behind differences in the responses of developmental rates to temperature between interacting species, as well as the need to account for differences in temperature response for interacting organisms from different latitudinal origins and during different developmental phases in order to understand and predict the consequences of climate change. / <p>At the time of the doctoral defense, the following papers were unpublished and had a status as follows: Paper 2: Manuscript. Paper 3: Manuscript. Paper 4: Manuscript.</p>

Populations often experience environmental conditions that are variable both in space and in time. Understanding the demographic and evolutionary dynamics of populations in such variable environments has very practical implications for conservation biology, pest and pathogen control, management of antibiotic resistance. This thesis is an attempt to study the ecological and evolutionary implications of spatial and temporal variations of the environment.First, I study how spatially heterogeneous and temporally changing conditions influence the demographic dynamics of a genetically uniform population. The growth of the population is enhanced when individuals preferentially accumulate in high quality habitats. Migration between locations facilitates a good arrangement of individuals such that in general, an intermediate rate of migration maximizes the growth rate.Second, I develop a model where the growth rate of individuals depends on the environment but also on their genetic quality, and possibly on the interaction between the environment and the genotype. If the performance of different genotypes tradeoffs across the environments, several genotypes may be maintained locally in the environment that suit them and a pattern of local adaptation emerges. Moreover, I show that adaptation of populations to environmental fluctuations in the environment generates very dynamic changes in the genetic composition that lag behind the environmental change. Adaptation may be facilitated by the influx of migrants coming from other demes.How can we detect such patterns of adaptation in wild or experimental populations? I develop a formal analysis of several experimental and statistical techniques that are used to detect patterns of local and temporal adaptation. I provide recommendations regarding efficient experimental designs and statistical techniques to detect local adaptation. I also develop a new framework for the analysis of patterns of adaptation in time. I illustrate the potential use of this approach using a data set measuring the adaptation of HIV to the immune response of several recently infected patients. / Populations often experience environmental conditions that are variable both in space and in time. Understanding the demographic and evolutionary dynamics of populations in such variable environments has very practical implications for conservation biology, pest and pathogen control, management of antibiotic resistance. This thesis is an attempt to study the ecological and evolutionary implications of spatial and temporal variations of the environment.First, I study how spatially heterogeneous and temporally changing conditions influence the demographic dynamics of a genetically uniform population. The growth of the population is enhanced when individuals preferentially accumulate in high quality habitats. Migration between locations facilitates a good arrangement of individuals such that in general, an intermediate rate of migration maximizes the growth rate.Second, I develop a model where the growth rate of individuals depends on the environment but also on their genetic quality, and possibly on the interaction between the environment and the genotype. If the performance of different genotypes tradeoffs across the environments, several genotypes may be maintained locally in the environment that suit them and a pattern of local adaptation emerges. Moreover, I show that adaptation of populations to environmental fluctuations in the environment generates very dynamic changes in the genetic composition that lag behind the environmental change. Adaptation may be facilitated by the influx of migrants coming from other demes.How can we detect such patterns of adaptation in wild or experimental populations? I develop a formal analysis of several experimental and statistical techniques that are used to detect patterns of local and temporal adaptation. I provide recommendations regarding efficient experimental designs and statistical techniques to detect local adaptation. I also develop a new framework for the analysis of patterns of adaptation in time. I illustrate the potential use of this approach using a data set measuring the adaptation of HIV to the immune response of several recently infected patients.

Les bases génétiques et l'histoire de la différenciation adaptative ont été étudiées chez les moules du complexe d'espèces Mytilus edulis qui représente un modèle d'étude intéressant pour mieux comprendre comment se propage et se distribue la différenciation adaptative en population structurée. Grâce à la technique AFLP, une approche de génomique des populations (« scan génomique ») a été utilisée pour mesurer la différenciation entre des populations isolées sur des échelles de temps et d'espaces contrastées. Notre objectif était de vérifier si la différenciation génétique n'avait pas une origine plus complexe qu'habituellement proposé. Trois parties s'articulent autour de cette question centrale. La première s'intéresse à la différenciation entre l'Atlantique et la Méditerranée chez l'espèce M. galloprovincialis et a montré que la structure génétique était la conséquence d'un différentiel d'introgression avec l'espèce sœur M. edulis. Dans la deuxième partie de ce travail nous avons mis en évidence qu'une sélection directe, soit balancée soit intermittente, sur un polymorphisme pré-existant expliquait le niveau de différenciation anormalement élevé d'un gène de l'immunité entre populations de la côte européenne chez M. edulis. Enfin, la troisième partie s'est intéressée à revisiter un exemple classique de la littérature de la génétique des populations: le cas des M. edulis du détroit de Long Island et a permis de suggérer que la structure observée à très petite échelle spatiale correspondait probablement à un contact secondaire entre des moules européennes introduites et les moules américaines. D'une manière générale nos résultats démontrent que, quelque soit l'échelle à laquelle nous nous plaçons, la différentiation génétique tire son origine d'une histoire souvent plus complexe qu'attendu. / The genetic basis and history of adaptive differentiation were studied in the species complex M. edulis which is an interesting model system to understand how adaptive differentiation spreads and structure itself in subdivided populations. Using the AFLP technique, a genome scan approach was undertaken to measure differentiation between populations on contrasted spacial and temporal scales. Our objective was to verify wether the origin of genetic differentiation could be more complex than anticipated. This question was addressed in three chapters. The first one focuses on the differentiation between populations of the Atlantic Ocean and Mediterranean Sea in M. galloprovincialis. Our results show that the genetic structure was the result of differential introgression with the sister hybridizing species M. edulis. In the second chapter of this work we demonstrated that direct selection on a pre-existing polymorphism explained the unusually high level of differentiation at a defensin locus between populations of the European coast in M. edulis. Finally, in the third chapter we revisited a classic example of the population genetics literature: the aminopeptidase cline in M. edulis populations of the Long Island Sound. We obtained evidence that the genetic differentiation observed at a very fine spatial scale in the sound was the consequence of a secondary contact between introduced mussels from Europe and indigenous American mussels. Whatever the spatio-temporal scale at which we analyzed genetic differentiation, its origin proved to originate from an unsuspectedly long and complex history.

La diversité génétique est à l’origine de l'évolution et de l'adaptation des pop. et des espèces. Dans les agrosystèmes, la div. gén. intra-population est d'une importance majeure : d'une part, elle peut fournir un effet tampon contre les variations climatiquesinterannuelles et les stress biotiques, et d'autre part cette div. peut permettre l’adaptation locale des pop., du fait de leur évolution sous l’effet des pressions sélectives spécifiques aux conditions locales de la région, particulièrement dans le cas d’uneintrod. dans un nouvel environnement. En raison de son importance socio-économique et de son aire de culture étendue, le blé a été choisi comme espèce modèle dans cette étude, en se focalisant sur l’étude de la précocité de de floraison, un caractère adaptatif majeur qui permet au blé de croître sur une large gamme de conditions écologiques et climatiques. Ce projet de thèse a pour objet l’analyse de l'impact de la diversité int.-pop. sur la réponse adaptative à court terme de pop. soumises à des conditions agro-climatiques contrastées, ce par l'étude des variations génétiques, épigénétiques et phénotypiques.L'absence de marqueurs épig. disponible pendant la thèse a conduit à développer 2 études complémentaires. Dans une 1ère partie, sept var. pays. (pop. conservées à la ferme) et une variété moderne ont été distribuées et cultivées pendant trois ans dans sept fermes localisées dans trois pays d'Europe, puis étudiées pour leur réponse aux différentes conditions agro-climatiques, sous l’angle de leurs variations phénotypiques et génotypiques. Dans une 2nde partie, l'effet de la vernalisation sur le profil de méth. de l'ADN du gène VRN-A1 a été étudié, constituant une 1ère étape vers le développement de marqueurs épig. Les résultats de la 1ère partie de l'étude ont révélé que l'histoire de la conservation des var. pays. a fortement influencé leur div. gén. et leur structure génétique fine. Les var. pays. conservées ex situ montrent une faible div. gén., avec une struct. génétique simple. Les var. pays. et les mélanges conservés in situ révèlent une div. gén. plus élevée, avec une struct. génétique complexe. Une différenc. spatio-temporelle génétique et phénotypique a été observée, en relation avec le niveau de diversité initial et avec la complexité de structure des var. pays. Les variétés traditionnelles se différencient plus nettement que les var. modernes, ce qui plaide en faveur de utilisation dans des systèmes d'agriculture biologique et à bas intrants. De façon intéressante, une différenc. phénotypique significative a été observée pour les var. qui présentaient une div. gén. initiale très faible, ce qui suggère que d’autres facteurs, par exemple épig., pourraient intervenir dans les adaptations misesen évidence. La 2nde partie a permis de mettre en évidence un profil de méth. intéressant de l’ADN de VRN-A1 : sur plantes non-vernalisées, ce gène présente des niveaux élevés de méth. dans la partie centrale du gène, mais pas en début et fin de gène. De plus, une partie du 1er intron montre une augmentation significative du niveau de méth. de l'ADN suite au traitement au froid. Ce changement de méth. est positivement associé au niveau d'expression du gène. Si la compréhension du rôle de cette méth. sur la régulation de VRN-A1 nécessite des analyses complémentaires, cette étude a permis de caractériser les modifications de méth. de VRN-A1 en réponse au froid et constitue une 1ère étape vers l’identification de possibles epiallèles dans nos pop. et fournit une base à la construction de marqueurs permettant de suivre la variabilité épig. dans différentes pop. En conclusion, cette étude apporte des connaissances utiles pour une meilleure compréhension de l’origine et l'évolution de la div. gén. présente dans les var. pays. Elles permettront de dvper des méthodes de conservation et desélection à la ferme, en tenant compte de l'importance de la div. int.-pop., afin de rép. aux contraintes posées par l'agriculture bio. / Biodiversity provides the raw material for evolution and adaptation of populations and species. In agricultural biodiversity, the within-population genetic diversity is of major importance. On one hand, it can provide a buffering effect against the year-to-year variation of climate or biotic pressures and on the other hand diversity serves as a resource for the population to respond to selective pressures due to specific local conditions, thus allowing for local adaptation, particularly in the case where a population is introduced into a new location. Due to its wide geographic distribution indicating a high adaptiveotential and its socio-economic importance, wheat was chosen as model crop in this study. Flowering time is a major adaptive trait which has allows wheat to grow over a wide range of ecological and climatic conditions. This PhD study was designed to gain insights about the influence of within population diversity on the short term response of populations to contrasting agro-climatic conditions by studying the genetic, epigenetic and phenotypic variation. But due to the lack of prior existence of epigenetic markers, this thesis study is divided of two parts: In the first part, European wheat populations coming from a set of seven farmer and one modern varieties, each of which was grown on seven farms (distributed across Europe) for three years, were used to study their short term response to contrasting agro-climatic conditions in Europe by analysing their phenotypic and genotypic variations. For the second part the effect of vernalization on the DNA methylation profile of theVRN-A1 gene in winter wheat was studied as a first step towards the development for the epigenetic marker in this gene.The results from the first part of the study revealed that conservation history of these farmer varieties strongly influenced the genetic diversity and fine genetic structure. Ex situ conserved farmer varieties showed low genetic diversity and simpler structure whereas in situ conserved farmer varieties and mixtures revealed higher level of genetic diversity and complex genetic structure. Genetic and phenotypic spatio-temporal differentiation depending upon the level of diversity and structural complexity of the farmer variety was observed. The traditional varieties tend to become more differentiated than the modern variety arguing in favour of use of these diverse traditional (farmer) varieties in organic and low input agriculture systems. Interestingly, a significant phenotypic differentiation for varieties with very low genetic diversity has also been observed in this study, which gives indication of a possible role of epigenetic variation in the process of evolution.From the second part of the study (effect of vernalization on the DNA methylation profile of the VRN-A1 gene), it was found that in addition to the detection of gene body methylation across the VRN-A1 gene, we identified a region within intron 1 that shows significant increase in DNA methylation in response to vernalization treatment that is positively correlated with the gene expression. Although the role of this shift in gene regulation is still unclear due to time limitations in the thesis and the small number of genotypes analysed, this study will provide a good material towards future identification of new epialleles and the development of epigenetic markers to study the epigenetic variability of these populations.This study at large provides useful knowledge on the understanding of farmers' varieties evolutionary response to be used in the development of different breeding and conservation approaches for organic agriculture, taking into consideration of the importance of within population diversity, to satisfactorily address the problems of organic agriculture.

Cette thèse s'inscrit dans le vaste domaine des recherches sur l'évolution des traits d'histoire de vie. Les traits d'histoire de vie sont directement impliqués dans la reproduction et la survie, et donc dans la valeur sélective, des organismes. L'intérêt pour les traits d'histoire de vie est motivé par l'immense diversité de ces traits dans le monde vivant. En particulier, il existe une grande diversité de patrons de vieillissement dans la nature. Cette thèse traite de l'évolution du vieillissement dans les populations naturelles. Comme pour la plupart des traits d'histoire de vie, les recherches théoriques sur l'évolution du vieillissement ont été menées dans le contexte d'une unique population, stable, et vivant dans un environnement constant. L'objectif de cette thèse est de prédire comment évolue le vieillissement lorsque les populations sont structurées dans l'espace et lorsque l'environnement varie dans le temps et l'espace. Mon approche a été essentiellement théorique, mais j'ai pu explorer certaines questions avec des données récoltées dans des populations naturelles. En introduction, je présente les éléments nécessaires à prendre à compte lorsqu'on s'intéresse à l'évolution des traits d'histoire de vie dans les populations structurées en âge. Je fais le constat que ces populations sont soumises à des variations environnementales et structurées dans l'espace. Dans la première partie, je montre que les variations de l'environnement, dans le temps et dans l'espace, peuvent affecter les patrons de sénescence. J'utilise des méthodes de génétique quantitative où je fais l'hypothèse que les mutations ont un effet spécifique à l'environnement et à l'âge. Dans la deuxième partie, je m'intéresse aux conséquences des dynamiques d'extinction-recolonisation sur la stratégie de partage des ressources entre la survie et la reproduction des individus en fonction de leur âge. Je montre en particulier que la dispersion est une source de variabilité pour cette stratégie et examine comment dispersants et non-dispersants diffèrent pour leur traits d'histoire de vie dans 3 jeux de données. De façon générale, mes résultats montrent que la prise en compte des conditions écologiques et environnementales complexes où vivent les organismes permet de mieux comprendre la diversité des patrons de vieillissement dans la nature. En dernier lieu, je fais un court bilan de cette thèse puis donne des perspectives possibles à mes recherches et de façon plus générale aux recherches sur l'évolution du vieillissement. / This thesis belongs to the wide field of life history evolution. Life history traits are directly involved in reproduction and survival. The interest for life-history evolution arises because organisms have evolved so many different ways of combining these traits to affect fitness. In particular, there is an important diversity of aging patterns in the wild. Aging is defined as the variation of life-history traits, such as survival or fecundity, with age. Senescence describes more precisely the idea that survival or fecundity decline with age. This synthesis deals with the question of aging in natural populations. Like most life-history traits, the evolution of aging has often been studied in the context of a single undisturbed population in a constant environment. My thesis aims at investigating how relaxing this assumption affects the evolution aging. I investigated this question mostly theoretically, but I used data from the field to investigate related questions. In introduction, I first go through evolutionary characteristics of age-structured populations. I also make the observation that age-structured populations are structured in space and live in a varying environment. In the first part, I investigate how environmental variability in space and time influences aging through the key results of Hamilton (1966) that the strength of selection declines with age. I used a quantitative genetics framework where I assumed that mutations that affect survival or fecundity have an age and environment specific effect. In the second part, I examine the role of metapopulation dynamics on the evolution of aging. I used an adaptive dynamics framework where mutations affect how a resource is shared between reproduction and survival at a given age. I show how the optimal age-specific reproductive effort is different in a metapopulation and in a single population. I then investigate how dispersal can be a source of heterogeneity in optimal age-specific reproductive effort. In this part, I went further in testing my predictions (and more generally exploring aging patterns) with data. As a general result, this thesis shows that including environmental and ecological heterogeneities helps understanding variation in aging in the wild. I finally discuss my main findings and give some perspectives to this work.

<p>The national curriculum and the School Act emphasize the importance of establishing partnerships between schools and local businesses. The formation of partnerships was expected to be facilitated by the decentralization reforms that took place in the beginning of the 1990’s. These reforms gave schools and municipalities greater scope for adjusting education to meet local circumstances. However, central government has not made financial resources available for stimulating the establishment of school-business partnerships, nor does it penalize municipalities that do not set up such partnerships.</p><p>The questions in focus in this study are: why do partnerships arise, how are they organised and what factors are important in achieving a partnership which can function in practice? Urban regime theory is applied to explain why collaborations occur and policy network theory is used to analyse how they are organised. A further aim of the study is to examine whether the partnerships can be defined as urban regimes.</p><p>The study was based on case studies of four upper secondary schools in Sweden, each in a different municipality. These municipalities are all industrialized, but the structure of their industrial base varies. Two are dominated by one or a few major companies whereas the others are characterized by the predominance of many small businesses. The study focuses on how three vocational study programmes – industry, electricity and building - collaborate with local businesses.</p><p>The study shows that the main explanation of why a partnership arises is that both partners believe that they can gain something by collaborating. For schools, the main reason for establishing a partnership with business is a belief that this will make the vocational education programmes more attractive for pupils. Through partnerships, schools can get access not only to practical experience for their pupils, but also help from business with developing the content of courses and some financial contributions. However, the economic benefits are of limited importance compared with the perceived gains in terms of the development of the educational content of the courses. This goes against results from studies in other countries.</p><p>For business, it is particularly important to be able to influence decisions about the content of local education. By doing so, they hope to ensure that the pupils, after having completed their education, will enter the local labour market with more relevant qualifications.</p><p>My research shows that several requirements have to be present for a successful partnership to be established. First, there must be a commitment and firm intentions from both partners. Second, size of the businesses involved in the partnerships is important for how they are organized. Larger businesses tend to see the partnership from a more long term perspective. However, it is also possible to organize collaborations with smaller businesses if they can be united under a common organisation. Third, specific actors make a difference in the partnerships. Devoted and interested key actors who are closely associated with the partnership are very important for the continuity of the partnership</p><p>The study suggests that urban regime theory and policy network theory are useful for understanding why and how partnerships between schools and businesses are established and retained. However, it can not be concluded that these collaborations in themselves are urban regimes at a more local level. This would require that their focus was much more long-term.</p>

As individuals and populations diverge ecologically, they become exposed to new parasites and pathogens with potentially harmful fitness consequences. Populations are therefore expected to evolve resistance, possibly at a cost of less resistance to parasites rarely encountered parasites. This trade-off in resistance should generate local adaptation to parasites in different habitats. In chapter one, I show how local adaptation can potentially evolve in response to variation in parasite exposure among eighteen ecologically variable populations of threespine stickleback (Gasterosteus aculeatus). Within populations infection appeared to reflect morphology/diet based exposure differences among individuals. Among populations, however, these patterns were absent or reversed, consistent with the evolution of local adaptation. In chapters two and three I set out to test whether variation major histocompatibility (MHC) genes can underly such local adaptation in stickleback. MHC genes are important components of vertebrate immunity; however, there is little direct empirical support for spatially divergent selection driving local adaptation on MHC loci in the wild. In chapter two I tested for the action of parasite mediated balancing and divergent selection on on MHC loci using naturally infected stickleback in three replicate lake-stream pairs. Despite consistent divergence in parasites and MHC alleles, lakes tended to show decreased parasite burdens with increased allelic richness (consistent with balancing selection), while streams showed some support for divergent selection between lake and stream types. In chapter three I use the same lake-stream pairs to investigate how divergent selection could instead be reflected in variation in the effects of individual MHC alleles among populations. When comparing parapatric populations experiencing gene flow, MHC alleles maintained at relatively high frequency in one population were more likely to be associated with reduced, rather than increased, parasite abundances in that population. Allopatric populations experiencing no gene flow showed no such general relationship between allele frequency and resistance. These results are only consistent with spatially divergent selection, and imply that gene flow and environmental heterogeneity can be important for maintaining MHC diversity. / text