UNIVERSIDADE ESTADUAL DE CAMPINAS
INSTITUTO DE BIOLOGIA
Juliana José
“Evolução da variabilidade genética: separando os
fatores que influenciam a variabilidade das espécies”
Tese apresentada ao Instituto de Biologia para obtenção do Título de Doutor em Genética e Biologia Molecular, na área de Genética Animal e Evolução.
Orientador: Prof. Dr. Sérgio Furtado dos Reis
Co-Orientador: Prof. Dr. José Alexandre Felizola Diniz-Filho
Campinas, 15 de Janeiro de 2009.
BANCA EXAMINADORA
Prof. Dr. Sérgio Furtado dos Reis (Orientador)
Assinatura
Profa. Dra. Vera Nisaka Solferini
Assinatura
Prof. Dr. André Lucci Freitas
Assinatura
Prof. Dr. Evandro Marsola de Moraes
Assinatura
Prof. Dr. Wesley Augusto Conde Godoy
Assinatura
Prof. Dr. Fernando José Von Zuben
Assinatura
Prof. Dr. Louis Bernard Klaczko
Assinatura
Profa. Dra. Lúcia Garcez Lohmann
We may often falsely attribute to correlation of growth, structures which are common to whole groups of species, and which in truth are simply due to inheritance; For an ardent progenitor may have acquired through natural selection some one modification in structure, and, after thousands of generations, some other and independent modification; and these two modifications, having been transmitted to a whole group of descendants with diverse habits, would naturally be thought to be correlated in some necessary manner. – Charles Darwin (1859, p.185)
Dedico este trabalho,
à minha mãe, Marisa, e ao meu amor, Marcelo, por sempre estarem ao meu lado. à memória do Ascon, Xis e Funny,
Agradeço,
a minha mãe, Marisa, e aos meus avós Wanda e Carlos Augusto, todos os mimos; a infinita disposição em me ajudar e me dar suporte de todas as formas; e o interesse em meu trabalho, mesmo sem entendê-lo.
ao meu amor, Marcelo, a vontade de me mimar ainda mais; o exemplo de perseverança, me incentivando a seguir em frente, mesmo quando tudo parecia não ter saída; a paciência nas fases difíceis do trabalho; o grande amor.
a Selma e ao Rodrigo, o esforço para me manter "em forma".
ao Prof. Sérgio, meu orientador, o apoio e a atenção dada em todos os momentos, os os inúmeros ensinamentos, e a calma e paciência durante as complicações do trabalho.
ao Prof. José Alexandre, cuja pesquisa me motivou a fazer essa tese, a ajuda em entender o mundo das análises filogenéticas comparativas e as infinitas idéias para melhorar nossos trabalhos.
ao Prof. Fernando Von-Zuben e ao Wilfredo, a grande disposição em nos ajudar a solucionar os problemas computacionais necessários para o desenvolvimento dos trabalhos.
aos membros da pré-banca, Profa. Dra. Vera Nisaka Solferini e Prof. Dr. Wesley Augusto Conde Godoy, os valiosos comentários.
a Profa. Vera, os ensinamentos que ainda me acompanham.
aos meus queridos amigos, o carinho, apoio, ajuda, consolos, e alegrias.
a todos vocês, e a todos os demais que, de alguma forma, contribuíram com este trabalho. Agradeço por fim a Capes e a Fapesp, a bolsa de doutorado concedida (Fapesp, processo no. 04/13080-3).
ÍNDICE
Resumo ix
Abstract X
Introdução 1
Variabilidade genética como conceito central para a biologia evolutiva 1 Diferentes níveis de organização influenciam a variabilidade genética 2
O problema e a nova perspectiva filogenética 4
As evidências na literatura 5
Nossa abordagem para o problema 6
Capítulo 1
Phylogenetic inheritance of genetic variability produced by neutral models of evolution
9
Introduction 11
Methods 15
Evolutionary models 15
Parameter values for simulations 16
Computational basis for simulations 17
Topology 19
Phylogenetic comparative methods 19
Results 21
Genetic drift 22
Genetic drift vs mutation 23
Genetic drift vs migration 24
The importance of species lifetime 26
Discussion 28
Inheriting genetic variability 28
Predictions for natural species 29
Phylogenetic signal on genetic variability 31 Acknowledgments 33
References 34
Capítulo 2
Phylogenetic signal affects the life history importance for genetic diversity
37
Introduction 39
Methods 43
Estimative of genetic diversity 43
Dataset 43
Phylogenetic signal 45
Analysis of correlated evolution 47
Results 48
Phylogenetic signal 48
Analysis of correlated evolution 59
Discussion 52
There is a phylogenetic signal 52
What changes from previous analysis? 53
Life history traits influencing genetic diversity 55
Conclusions 58
Acknowledgments 58
Literature cited 59
Capítulo 3
Evolutionary history partially explains modularity patterns in networks describing species genetic diversity
65
Introduction 67
Methods 69
Results and Discussion 71
Literature cited 75
Discussão 78
Sinal filogenético na variabilidade genética 78
Conseqüências teóricas do sinal filogenético 80
Mecanismo de herança filogenética 81
Conseqüências do sinal filogenético nas análises da variabilidade 81
Conclusões e Direções Futuras 83
Resumo
A quantidade de variabilidade genética presente nas espécies pode ser influenciada por diversos fatores que atuam em diferentes níveis de organização biológica. Dentre esses fatores, os que afetam a dinâmica populacional têm sido extensivamente estudados. No entanto, a influência da história evolutiva tem sido negligenciada ao se estudar a variabilidade genética das espécies. Nós investigamos pela primeira vez a influência da história evolutiva das espécies sobre sua variabilidade genética, e como a história evolutiva compartilhada afeta as relações já estabelecidas entre a variabilidade genética e outros traços, através dos métodos filogenéticos comparativos e de métodos de análise de redes complexas. Simulações computacionais de modelos neutros de evolução indicaram influência da história evolutiva, e nos deram previsões acerca do sinal filogenético presente na variabilidade genética. Nós de fato observamos o sinal filogenético previsto nas simulações em grupos animais variados que compõe um banco de dados de 1521 espécies amostradas para a diversidade genética de aloenzimas. Detectamos também a influência da história evolutiva das espécies sobre sua variabilidade no padrão modular de redes que representam a similaridade na diversidade genética das espécies. Quando consideramos a história evolutiva na análise das relações entre a variabilidade genética e outros traços das espécies, observamos relações mais fracas do que as que foram previamente estabelecidas na literatura.
Abstract
The amount of genetic variability on species can be influenced by factors acting in different levels of biological organization, and the ones related to population dynamics have been extensively studied. However, past studies neglected the influence of evolutionary history on genetic variability. We studied for the first time the influence of evolutionary history on species genetic variability and how the influence of evolutionary history changes pre-established relationships between variability and other species traits. For our investigations we used phylogenetic comparative methods and complex network analysis. Computer simulations on neutral models of evolution showed influence of evolutionary history and also provide us expectations for a phylogenetic signal on genetic variability. We in fact observed the previously expected phylogenetic signal in a wide variety of animal groups, which compose a database of 1521 species sampled for allozymic genetic diversity. We also detected the influence of species evolutionary history on its genetic variability in the modularity patterns of networks representing genetic diversity similarities between species. When considering evolutionary history on the analysis of genetic variability relationships with other species traits, we observed weaker relationships than those previously established on literature.
Introdução
Variabilidade genética como conceito central para a biologia evolutiva
A Teoria Neutra da Evolução Molecular proposta por Kimura (1983) fez da quantidade de variabilidade genética presente nas espécies uma questão central para a Biologia Evolutiva. Essa teoria veio questionar os mecanismos antes propostos para explicar a manutenção do polimorfismo em populações naturais, baseados em seleção natural. Os geneticistas de populações na época reagiram fortemente à proposição de neutralidade dos polimorfismos e buscaram quantificar a variabilidade genética no maior número de caracteres em no maior número de espécies possível. A idéia era testar se a quantidade de variabilidade presente nas populações naturais seria realmente mais alta que a predita por modelos de seleção natural (Greenough e Harvey 1987).
A busca pela quantidade de variabilidade genética existente em populações naturais foi impulsionada pelos trabalhos de Hubby e Lewontin (1966) e Harris (1966) que introduziram a técnica de eletroforese de aloenzimas nos estudos de genética de populações. A partir desses trabalhos, inúmeros outros se propuseram a estimar a quantidade de variabilidade genética em populações naturais através da eletroforese de aloenzimas (Nevo 1978, Hedrick 1986, Solé-Cava e Thorpe 1991).
Na década de 80, os marcadores moleculares de DNA surgiram e passaram a ser considerados mais adequados aos estudos populacionais (Cavalli-Sforza 1998, Sunnucks 2000), apesar das aloenzimas ainda serem adequadas e bastante úteis (Burton 1996, Altukhov e Salmenkova 2002). Entretanto, mesmo com o sucesso dos novos marcadores, as aloenzimas ainda são os marcadores que foram mais usados e, seus mecanismos de evolução, melhor estudados (Skibinski e Ward 1998, Altukhov e Salmenkova 2002).
Com o desenvolvimento dos métodos de análise e o acúmulo de trabalhos na literatura, as perguntas relacionadas à variabilidade genética se diversificaram. Atualmente, as estimativas de variabilidade genética têm grande importância em diferentes áreas: i) ainda para o debate entre as escolas selecionista e neutralista de evolução (Skibinski e Ward 1998), ii) para a genética de populações, que tenta relacioná-la às forças microevolutivas que estão atuando nas populações naturais (Altukhov e Salmenkova 2002, Lewontin 2002), iii) para as estimativas de biodiversidade, como uma estimativa de diversidade genética (Ryskov 1999, Primack e Rodrigues 2001) e iv) para a biologia da conservação, que requer a estimativa de diversidade genética para analisar sua distribuição no espaço e como está sendo alterada por fatores externos à biologia da espécie (Primack e Rodrigues 2001). A grande importância da variabilidade genética para essas áreas está no pressuposto de que esta é um importante pré-requisito para a mudança evolutiva (Hedrick 1986, Lewontin 2002).
Diferentes níveis de organização influenciam a variabilidade genética
A quantidade de variabilidade genética presente em marcadores moleculares resulta de diversos fatores em diferentes níveis de organização biológica: no nível molecular, no nível populacional e no nível das espécies. Dentre eles, os dois primeiros têm sido extensivamente estudados enquanto os fatores atuantes no nível das espécies têm sido negligenciados.
No nível molecular, a quantidade de variabilidade em aloenzimas é influenciada por fatores como especificidade de substrato (Gillespie e Kojima 1968), seus mecanismos de regulação e função (Johnson 1974), o número e tamanho das subunidades e estrutura quaternária (Ward 1977, Ward et al. 1992). Fatores como especificidade de substrato e
função influenciam a quantidade de variabilidade genética principalmente por seleção natural direcional e outros fatores como tamanho das subunidades e estrutura quaternária parecem atuar de modo neutro (Ward et al. 1992, Skibinski e Ward 1998). Dentre estes fatores, o tamanho das subunidades e a estrutura quaternária estão correlacionados com a heterozigosidade em locos alo enzimáticos (Ward et al. 1992).
Também no nível molecular, a quantidade de variabilidade presente em marcadores de DNA é determinada por fatores como tipo de seqüência, posição no cromossomo, mecanismos de regulação (Goldstein e Pollock 1997, Estoup et al. 2002). Nesse caso, todos os fatores citados atuariam de modo neutro. Os fatores que atuam no nível molecular de modo neutro determinam a quantidade de variabilidade genética alterando a taxa de mutação de cada loco gênico. Essa taxa de mutação é a característica do nível molecular que irá influenciar o balanço das forças microevolutivas na determinação da quantidade de variabilidade genética no nível populacional.
No nível populacional, além da mutação, o fluxo-gênico, a deriva genética e a seleção natural são fatores que podem determinar a quantidade de variabilidade genética em quaisquer marcadores. Os geneticistas de populações têm formulado hipóteses de processos microevolutivos para explicar e predizer a quantidade de variabilidade presente nas espécies (Hedrick et al. 1976, Hedrick 1986, Bohonak 1999, Richman 2000, Johnson et al. 2001, Altukhov e Salmenkova 2002, Lewontin 2002). Essas hipóteses estão baseadas na relação observada entre as características genéticas e as de história natural dessas espécies. Desse modo, a maioria dos padrões de variabilidade preditos está baseada naquelas características de história natural que determinariam a intensidade de ação de cada força microevolutiva.
Algumas revisões se dedicaram à relação entre características de história natural das espécies e a variabilidade genética observada. Nevo (1978, e Nevo et al. 1984) fez grande revisão da literatura existente sobre variabilidade estimada em marcadores aloenzimáticos. Nevo usou 243 espécies de plantas e animais no trabalho de 1978, e 1015 espécies no trabalho de 1984. Nesse trabalho foram encontradas diferenças de variabilidade genética entre espécies de diferentes “zonas de vida”, biotas, habitas, níveis de especialização e também de diferentes grupos taxonômicos. A partir das relações observadas, Nevo (1978 e 1984) concluiu que a variabilidade genética poderia ser vista como adaptação ao ambiente, no espaço e no tempo.
O problema e a nova perspectiva filogenética
Independente da universalidade das associações propostas entre variabilidade genética e características da história natural, essas proposições ignoram a influência da história evolutiva e das características compartilhadas no nível das espécies, devido às relações filogenéticas. Todas as espécies atuais estão conectadas, em maior ou menor grau, por um ancestral comum como resultado do processo genealógico de ancestralidade e descendência entre indivíduos, linhagens e espécies (Felsenstein 1985, Diniz-Filho 2000).
Além do problema teórico de ignorar essas conexões, surge também um sério problema metodológico ligado ao fato de que a história evolutiva de uma espécie tem um papel importante moldando a direção e taxas de evolução dos atributos que as espécies apresentam (Cheverud et al. 1985). Como as espécies possuem ancestrais comuns no tempo, elas compartilham semelhanças em muitas de suas características por causa das características presentes no próprio ancestral (Felsenstein 1985). Essas semelhanças
geradas pelo processo de ancestralidade e descendência resultam em uma dependência filogenética nas características das espécies atuais.
O fato biológico de que as espécies estão conectadas em uma filogenia estruturada hierarquicamente, gera um problema estatístico de que as espécies não podem ser analisadas como se amostradas independentemente de uma mesma distribuição (Felsenstein 1985). Na tentativa de lidar com o problema estatístico para detectar a dependência filogenética, e para isolar o efeito da filogenia em análises comparativas, inúmeros métodos filogenéticos comparativos têm sido propostos (Cheverud et al. 1985, Felsenstein 1985, Gittleman e Kot 1990, Diniz-Filho et al. 1998, Blomberg et al. 2003).
A partir dos métodos filogenéticos comparativos, a dependência filogenética dos atributos das espécies pode ser definida mais precisamente pelo termo sinal filogenético. Blomberg e Garland (2002) definiram o sinal filogenético como a tendência de espécies filogeneticamente relacionadas serem mais semelhantes entre si do que são com outras espécies sorteadas ao acaso na filogenia.
As evidências na literatura
Apesar de nenhum trabalho ter investigado a influência das relações filogenéticas na variabilidade genética presente nas espécies, alguns indícios dessa influência podem ser encontrados na literatura. Um indício é a correlação entre a heterozigosizidade das espécies e a respectiva taxa de evolução prevista pela teoria neutra de evolução molecular (Nei 1972, Skibinski e Ward 1982). Se a heterozigosidade se mantém ao longo da evolução das espécies e varia de acordo com a distância genética entre elas, deve ter deixado um sinal dessas relações ao longo da filogenia.
A correlação entre heterozigosidade e distância genética chega a 70% em espécies de vertebrados e se manteve para diferentes grupos taxonômicos (vertebrados, invertebrados e plantas), tanto no nível intra quanto interespecífico (Ward e Skibinski 1985). Essa correlação também se mostrou robusta à variações no modo de reprodução, no tamanho populacional, na intensidade de fluxo gênico, nas taxas de mutação e nos tempos de divergência (Mukherjee et al. 1987). Através de simulações dos modelos observou-se que correlação positiva entre a heterozigosidade e a distância genética pode também ser explicada por alguns modelos de seleção, mas a mutação neutra ainda seria a melhor explicação para essa correlação (Skibinski e Ward 1998).
Outro forte indício da influência das relações filogenéticas na quantidade de variabilidade genética das espécies pode ser encontrado nas revisões do próprio Nevo (1978, et al. 1984) e de Ward et al. (1992). Esses autores dividiram as espécies estudadas em grupos taxonômicos procurando relações entre as características de história natural e a variabilidade genética observada nas espécies. Apesar da correspondência entre a taxonomia e as relações filogenéticas não ser completa, a divisão em grandes grupos taxonômicos apresenta grande correspondência com as relações filogenéticas inferidas até o momento. Conduzindo as análises separadamente para os táxons amostrados, Nevo (1978, et al. 1984) e Ward et al. (1992) encontraram diferentes padrões de variabilidade nos diferentes grupos e diferenças de heterozigosidade entre esses grupos.
Nossa abordagem para o problema
A partir da expectativa teórica de dependência filogenética na variabilidade genética das espécies e das evidências que aparecem na literatura, nós inferimos que existe um sinal filogenético na variabilidade genética das espécies. Para estudar a influência da
história evolutiva na variabilidade genética nós abordamos o problema de três formas distintas, que resultaram em três trabalhos apresentados na forma de capítulos da tese. Como estimativa de variabilidade genética nós usamos a heterozigosidade esperada (He)
pela lei de Hardy-Weinberg, um parâmetro que reflete a diversidade, pois considera tanto o número de alelos quanto suas freqüências relativas, e tem sido estimado nos estudos comparativos da variabilidade genética.
Nossa primeira abordagem foi entender como o sinal filogenético aparece na evolução neutra da variabilidade genética. Para isso nós usamos simulações computacionais da evolução da variabilidade por três modelos neutros: deriva genética e os equilíbrios deriva-mutação e deriva-migração. Os resultados obtidos nas simulações estão apresentados no capítulo 1.
Após analisar a previsão do sinal filogenético na evolução neutra da variabilidade, nós estimamos esse mesmo sinal na variabilidade genética de uma grande variedade de grupos animais, somando 1521 espécies. Para essas estimativas, nós inferimos as relações filogenéticas entre elas usando hierarquias taxonômicas e filogenias moleculares. Além disso, nós re-avaliamos algumas das relações propostas anteriormente entre a variabilidade genética e traços de história natural das espécies com uma perspectiva filogenética. As estimativas do sinal filogenético e as relações estudadas entre a variabilidade e outros traços das espécies estão apresentadas no capítulo 2.
Com os dados obtidos para o capítulo 2, nós abordamos o problema com a perspectiva de sistemas complexos descrevendo as relações entre a variabilidade genética das espécies na forma de redes de diversidade genética. Nós analisamos aspectos estruturais das redes e a relação de sua modularidade com a história evolutiva das espécies que a compõe. Essa abordagem é nova na biologia comparada e se mostrou
promissora para investigar a influência da história evolutiva nos atributos das espécies em conjunto com as análises filogenéticas. As análises de redes complexas estão apresentadas no capítulo 3.
Capítulo 1
P
HYLOGENETIC
I
NHERITANCE
O
F
G
ENETIC
V
ARIABILITY
Phylogenetic inheritance of genetic variability
produced by neutral models of evolution
J José1; W.J. Puma-Villanueva2; F.J. Von Zuben2; J.A.F. Diniz-Filho3
1Instituto de Biologia, Universidade Estadual de Campinas, Cx.P.6109, CEP
13083-970, Campinas, São Paulo, Brasil. Fone: (19)3521-6279. Fax: (19)3307-7761.
2Laboratório de Bioinformática e Computação Bio-inspirada, Departamento de
Engenharia da Computação e Automação Industrial, Faculdade de Elétrica e Engenharia da Computação, Universidade Estadual de Campinas, Campinas, São Paulo, Brasil.
3Laboratório de Ecologia Evolutiva, Departamento de Biologia Geral,
Universidade Federal de Goiás, Goiânia, Goiás, Brasil.
Corresponding author: J. José E-mail: juliana.jose@gmail.com
Genet. Mol. Res. 7 (4): 1327-1343 (2008) Received August 27, 2008
Accepted September 9, 2008 Published November 25, 2008
Abstract. The amount of genetic variability in species and populations
has been mainly related to microevolutionary forces operating in natural populations and the influence of phylogenetic processes for the distribution of genetic variability has been neglected. To investigate how the current genetic variability distribution depends on the genetic variability of ancestral species, we simulated the evolution of heterozy-gosity on a pre-determined phylogeny under three neutral models of evolution: genetic drift, drift vs mutation and drift vs migration. The distribution of genetic variability resulting from the simulations was used to estimate the phylogenetic signal by the phylogenetic compara-tive method of autocorrelation. Phylogenetic signal in genetic variabil-ity was observed for each of the three models, and its intensvariabil-ity was generally higher and persisted longer when forces of drift, mutation and migration were reduced. The prediction of a phylogenetic signal in genetic variability has consequences for: population genetics, which must consider biological processes acting at the species level influencing the amount and distribution of genetic variability; the macroevolutionary
Key words: Population genetics; Heterozygosity; Neutral evolution; Phylogenetic comparative method; Phylogenetic autocorrelation; Macroevolution.
INTRODUCTION
The amount of genetic variability in natural populations has been an important focus in the study of population genetics, especially its relationship with the microevolutionary forces acting in these populations (Altukhov and Salmenkova, 2002; Lewontin, 2002). The study of genetic variability increased in importance in the 1960s (Lewontin, 1985), with the debate between the relative importance of selection or neutral models of evolution, a debate that persists to this day (Skibinski and Ward, 1998). Genetic variability has also been considered to be an important trait at the species level, resting on the assumption that the reservoir of genetic variability acts as a hedge against extinction (Dobzhansky, 1939) and is thus an important pre-requisite for evolutionary change (Hedrick, 1986; Lewontin, 2002). Genetic variability has been studied at the species level as a good trait to show species selection (Lloyd and Gould, 1993; Lieberman and Vrba, 2005) and as a marker to study extinction risk (Frankham, 1996; Spielman et al., 2004).
Population geneticists usually consider the amount of genetic variability present in natural populations as a result of a balance among microevolutionary forces (Hedrick et al., 1976; Nevo, 1978; Hedrick, 1986). Genetic variability estimates by molecular markers are usually explained with a genetic drift vs. mutation, or genetic drift vs. gene flow balance. This causal relationship assumes that the effective population size and mutation or migration rate determine the amount of genetic variability (Crow and Kimura, 1970; Frankham, 1996; Frankham, 1998; Garner et al., 2005). For allozymes and SNPs, natural
selection has also been considered to determine the amount of genetic variability (Nevo, 1978; Hedrick, 1986; Kreitman and Akashi 1995; Panova and Johannesson, 2004; Nielsen 2005).
Some papers and reviews have tried to establish relationships between species genetic variability and species life-history characteristics (Nevo, 1978; Mukherjee et al., 1987; Ward et al., 1992; Frankham, 1996; Frankham, 1998; Spielman et al., 2004; Garner et al., 2005). Nevo (1978) found different degrees of allozyme genetic variability in species of different life zones, biotas, habitats, specialization levels and also taxonomic groups and concluded that genetic variability can be seen as an adaptation to environment, in space and time. Ward et al. (1992) not only found relationships between allozyme genetic variability and some life-history traits but also relationships with the biochemical properties of each enzyme. More recent works (Frankham, 1996; Frankham, 1998; Spielman et al., 2004; Garner et al., 2005) showed relationships between species genetic diversity and population size, inbreeding rate and extinction risk.
In spite of the generality of the proposed associations between genetic variability and species bionomic characteristics, those propositions ignored the influence of species evolutionary history and the characteristics shared at the species level as a result of ancestral-decent relationships. Species are connected, at a higher or lower degree, by a common ancestor because of the genealogical process of ancestry and decent between individuals, lineages and species (Felsenstein, 1985).
In ignoring these connections, a serious theoretical problem emerges. The evolutionary history of a species has an important role in constraining its characteristics and in the directions and rates of evolution of its traits (Cheverud et al., 1985). Since species share common ancestors, many species characteristics are shared as a consequence
of common ancestral characteristics resulting in a phylogenetic dependence on the characteristics of contemporaneous species (Cheverud et al., 1985; Felsenstein, 1985).
Besides the theoretical problem, there is also a methodological one, if we assume independence when analyzing a trait in a group of species. To overcome these errors, phylogenetic comparative methods have been proposed (Cheverud et al., 1985; Felsenstein, 1985; Gittleman and Kot, 1990; Diniz-Filho et al., 1998; Blomberg et al., 2003) either to estimate the phylogenetic signal retained in a trait or to isolate the phylogenetic signal and study the correlation of two traits among species. The term phylogenetic signal has been widely used to represent the phylogenetic dependency or effect and was recently defined by Blomberg and Garland (2002) as the tendency of related species to resemble each other more than they resemble species drawn at random from a tree.
Some clues for a phylogenetic signal in species genetic variability can be found in previous works. One clue is the correlation between intraspecific polymorphism and interespecific evolutionary rate, a prediction of the neutral theory of molecular evolution (Kimura, 1983). This correlation can be considered as evidence of the phylogenetic signal because it relates the genetic variability to the evolutionary time of species isolation, which represents the degree of phylogenetic relatedness.
A series of studies by Skibinski and Ward explored these correlations (Skibinski and Ward, 1982; Ward and Skibinski, 1985; Mukherjee et al., 1987; Skibinski and Ward, 1998). They used heterozygosity as a genetic variability estimate and genetic distance (Nei, 1972) as an evolutionary rate estimate. For vertebrates, they found a correlation of 70% (Skibinski and Ward, 1982), and high correlations were also found for invertebrates and plants (Ward and Skibinski, 1985). These correlations were not sensitive to variations in
reproductive mode, population sizes, gene-flow intensity, mutation rate or time of divergence (Mukherjee et al., 1987).
Simulations showed that some selective models can also produce these correlations, but neutral mutation is still the most plausible explanation (Skibinski and Ward, 1998). Skibinski, Ward and their co-authors do not invoke phylogenetic effects to explain their data, although their results provide good evidence for the existence of a phylogenetic signal in genetic variability.
Another clue can be found in the review of Nevo (1978). This author divided species into taxonomic groups for analysis and found differences in the amount of genetic variability between groups. In the review by Ward et al. (1992), the care in considering the correlation between genetic variability and genetic distance in comparing taxonomic groups produced results that differ from Nevo (1978). If we assume that there is some correspondence between taxonomy and phylogeny for the species studied, this difference suggests that part of the associations found by Nevo (1978) can be due to a phylogenetic dependence on genetic variability and on the life-history characteristics that he used.
Considering the theoretical expectation of phylogenetic dependence between species and the clues we have from the literature, we infer that there is a phylogenetic signal in species genetic variability. To understand how this phylogenetic signal can emerge from the evolution of genetic variability, we used simulation procedures to evolve the genetic variability over a predefined topology under neutral evolutionary models. Our results are discussed in the context of phylogenetic comparative theory, macroevolution theory and population genetics theory and data, and a mechanism of phylogenetic inheritance is proposed. The patterns we observed from the simulated models can be further compared to available data on species genetic variability.
The study of a population genetic trait in a phylogenetic comparative way is a novelty, and there is just one report of phylogenetic comparative analysis for a population genetic trait, namely population structure (Duminil et al., 2007). This is the first study that analyzes the theoretical expectations for the existence of a phylogenetic signal in a population genetic trait, and also the first study that examines species genetic variability as a phylogenetic character.
METHODS
Evolutionary models
We simulated the evolution of genetic variability under three stochastic models of evolution. The first model is evolution by genetic drift (Wright, 1931), given by
(Equation 1)
where, Ht is the heterozygosity at generation t, Ht-1 is the heterozygosity at generation t1
and Ne is the effective population size.
The second model is evolution by drift and mutation (Kimura and Crow, 1964), given by
(Equation 2)
where, is the mutation rate. The term (1μ)2 then is the probability that both alleles do not mutate.
The third model is evolution by drift and migration (Crow and Kimura, 1970), given by
(Equation 3)
where, M is the migration rate. This model assumes the island model of migration, and the term (1M)2 gives the probability that neither allele is a migrant. Since we simulated heterozygosity changes throughout species, in the three models we considered Ht as the
heterozygosity after t generations at the species n, H0 as the initial heterozygosity at the
lineage n1 and Ne as the effective population size averaged for the lineage.
Parameter values for simulations
The parameter values for the three models of evolution are presented in Table 1. The three models had 4 parameters varying in simulations; for each parameter we chose three mean values resulting in 81 simulations for each model. In simulations of genetic drift, we also varied the parameter VarNe, which represents variance of the distribution of
Ne, from which the values of effective population size for each lineage was randomly
sampled throughout the simulations. Since the variation of this parameter did not influence the results, in simulations of the other two models we fixed the variance of Ne at
0.5.
The choice of values was intended to cover all the range of variation in heterozygosity over time according to the distribution determined by the models. For the genetic drift model, the greater variation in heterozygosity occurs for Ne < 10,000 and for
t<1000. In this model, we also took care to maintain t between t = Ne/10 and t = Ne (Crow
lineage. For the drift-mutation model, the maximum value for time was reduced because, for the defined mutation values, there is almost no variation in heterozygosity after 3000 generations. The mutation rate values were defined setting 4Nμ = 0.4, used by Skibinski and Ward (1998). For the drift-migration model, the values of Ne and t had to be even more
reduced because the equilibrium was attained too fast in the values we used in the other two models. This should not be a problem, because if we take the value of Ne and t very
close to equilibrium, we will have results that can be extended to all higher values.
Table 1. Parameter values for the three models used in simulations.
Genetic drift
H0 Ne VarNe t
0.1 100 0.01 100
0.3 1000 0.03 500 0.5 10000 0.05 1000
Genetic drift x mutation
H0 Ne μ t
0.1 100 0.0001 100
0.3 1000 0.00001 1000
0.5 10000 0.000001 3000
Genetic drift x migration
H0 Ne M t
0.1 50 0.01 50
0.3 100 0.001 100
0.5 1000 0.0001 500
H0 is the inicial heterozygosity, Ne is the effective size of the lineages
simulated, VarNe is the variance of a normal distribution allowed for the
effective size, μ is the mutation rate, M is the migration rate, and t is time measured in number of generations.
Computational basis for simulations
To run simulations on the models of evolution described above, we programmed a routine in the MatLab 6.5 software. The input was: i) a topology, ii) the value of H0, iii) the
mean and variance of the Ne and t distributions in which the values of these parameters
will be randomly sampled, and iv) the number of times this simulation had to be replicated.
The algorithm is:
1. Enter a topology in Newick notation. The next steps will refer to the topology exemplified in Figure 1;
2. Define the initial value of H0 and the mean and variance of Ne and t distribution;
3. Calculate the value of H1 and H2 using a random value of Ne at each iteration;
4. Take the value of H1 as H0 to calculate H3 and H5, with a new random value of Ne at
each iteration;
5. Take the value of H2 as H0 to calculate H4 and H6, with a new random value of Ne at
each iteration.
After step 5, the simulation proceeds repeating steps 4 and 5 using each new heterozygosity value already defined for a node as the ancestor heterozygosity to calculate the heterozygosity of its two descents. The simulation stops when the heterozygosity value is calculated for all internal and external nodes. The simulation was replicated the number of times requested. The output gives two files: i) the values of H to each internal and external node of the topology to each replicate; and ii) a mean value to each internal and external node across all repeats.
Figure 1. Symmetrical topology only with bifurcations and 8
terminal nodes.
Topology
We defined a topology of 128 OTUs, with bifurcations only, and totally symmetrical. This topology has 7 hierarchical levels (Fig. 2) that were analyzed, either in total, or isolated to construct correlograms. The branch lengths were determined by the value of t randomly sampled in each simulation.
Comparative phylogenetic analysis
We analyzed 1,000 replications of each simulation, i.e., for each combination of parameter values for each of the three models. We used Moran’s I, estimated by the phylogenetic comparative analysis of phylogenetic autocorrelation (Gittleman and Kot, 1990; Diniz-Filho, 2001). Moran’s I is a good estimate of the phylogenetic signal retained (Gittleman and Kot 1990; Martins et al., 2002), and it allowed us to estimate the phylogenetic signal to each simulation, and also for each of the 7 hierarchical levels of the
topology, generating a phylogenetic correlogram. These analyses were executed using the Autophy 5.1 program (Diniz-Filho, 2001).
Figure 2. Comb-like topology of 128 terminal nodes used in
simulations. Vertical lines define the classes of phylogenetic distance used for autocorrelation analysis.
The mean value for Moran´s I obtained for each combination of parameters was used in a principal components analysis to summarize correlogram profiles along a single dimension. The first axis corresponded to 77.6% of the total variance, and its scores described a continuous change in correlogram profile, from an abrupt decay curve at higher scores to a gradual decay curve at lower values (Fig. 3). Correlograms with an
abrupt decay lose phylogenetic signal faster, whereas those with a gradual decay retain more phylogenetic signal (Fig. 3). Differences between average correlogram profiles produced by alternative models and by different parameter combinations were tested by an analysis of variance on PC1 scores with a Bonferroni post hoc test.
Figure 3. A. Average correlograms for the three simulated models of
evolution. B. Scores obtained for factor 1, represented by the PC1 axis, of principal components analysis for each model studied.
RESULTS
All three simulated models showed some amount of phylogenetic signal for genetic variability. This signal increases in strength from the ancestral to the derivate nodes as
expected (Diniz-Filho, 2001). For the first class of distance the phylogenetic signal measured with Moran´s I was always equal to 1 (Fig. 3A).
Genetic drift
Just the variation in two parameters, Ne and t, influenced the phylogenetic signal in
genetic drift simulations (respectively, F=70.991, p=0.00) (Fig. 4). The variation in the initial heterozygosity and in the variance of Ne had no effect for the estimates of
phylogenetic signal (F=0.00 and p=1 for both). The phylogenetic signal is retained more with an increase in the values of Ne and with a decrease in the values of t (Fig. 4).
Figure 4. PC1 scores obtained for each parameter value used in
simulations of genetic drift model. The different letters over the PC1 data on scores distribution represent significant differences at the 0.05 level (ANOVA). For abbreviations, see legend to Table 1.
For the phylogenetic signal, the influence of the highest Ne and the lowest t is
similar, whatever the values are for the other parameters, and the correlogram profile is linear until it crosses zero.
Genetic drift vs. mutation
When mutation was introduced into the genetic drift model (equation 2), this new parameter influenced the observed phylogenetic signal (F=3.324, p=0.041) and changed the shapes of the correlogram profiles for Ne and t (F=21.211, p=0.00 and F=23.445, p=0.00),
mainly for high mutation rate values (Fig. 5).
Figure 5. PC1 scores obtained for each parameter value used in
simulations of genetic drift vs mutation model. The different letters over the PC1 data on scores distribution represent significant differences at the 0.05 level (ANOVA). For abbreviations, see legend to Table 1.
We also tested for an interaction between mutation and effective population size, but found that it was not significant (F=1.354, p=0.259) (Fig.6).
Figure 6. Interaction effect of effective size (Ne) with mutation rate (μ) in
simulations of genetic drift vs mutation model.
Genetic drift vs. migration
The role of migration counterbalancing genetic drift is similar to mutation because of the way the model was defined. However, migration rates are usually higher than mutation rates and the influence of migration rates in heterozygosity evolution is stronger. To study the effect of migration we had to increase the intensity of drift. All the correlograms resulting from genetic drift vs. migration simulations exhibited an abrupt
decay of phylogenetic signal (F=21.412, p=0.00), at least for the higher migration rate (Fig. 7). The effect of Ne and t was also high (F=8.207, p=0.001 and F=6.693, p=0.002). The
interaction between migration and effective population size was significant (F=7.514, p=0.00) (Fig.8). In a scenario where drift is very strong (Fig. 8, Ne = 50), the effect of
migration in phylogenetic signal is very low. As we reduce the action of drift, the effect of migration increases (Fig. 8, Ne = 100 and Ne = 1000) and the loss of phylogenetic signal is
more abrupt through phylogenetic distance classes at higher migration rates.
Figure 7. PC1 scores obtained for each parameter value used in simulations
of genetic drift vs migration model. The different letters over the PC1 data on scores distribution represent significant differences at the 0.05 level (ANOVA). For abbreviations, see legend to Table 1.
Figure 8. Interaction effect of effective population size (Ne) with migration
rate (M) in simulations of genetic drift vs migration model.
The importance of species lifetime
In simulations of the three different models, one pattern was common to the parameters μ, M and t : the larger the parameter value, the faster the phylogenetic signal was lost. This reduction of phylogenetic signal is a simple consequence of the fact that higher values for parameters drive the system to equilibrium heterozygosity within species in a few generations and almost all OTUs of a branch end with very similar heterozygosity values.
As an example, we can see the correlogram and the distribution of heterozygosity among phylogenetic distance classes in Figure 9 that corresponds to a drift vs. mutation simulation with Ne = 1000 and t = 1000 and H0 = 0.1. For μ = 10-3, we had a rapid increase
in heterozygosity from 0.1 to 0.8 in the first tree bifurcation. Since the maximum value for heterozygosity in this case is 0.83 (the equilibrium point), the evolution in subsequent lineages had just a small possibility of varying between lineages, and groups became very homogeneous very rapidly, leading to the fixation of the value of heterozygosity in the distance 4 class. In this class, the phylogenetic signal disappears. For μ = 10-3, the initial increase in heterozygosity is high, from 0.1 to 0.46, but it still leaves more space for variation until the maximum value, 0.52, is reached. In this case, there was variation in means between nodes of all classes of distance, and the phylogenetic signal just disappears only in class five.
Figure 9. Results obtained in simulation of drift vs. mutation model with parameter values of H = 0.1, Ne = 100 and t = 1000 . (A) Correlogram and (B) Distribution of mean heterozygosity by phylogenetic distance class.
Discussion
Phylogenetic comparative analysis is usually applied to phenotypic traits that have a quantitative genetic base (such as Brownian motion and the O-U process) and its evolution is usually approximated by general models of evolution (Felsenstein, 1985; Felsenstein, 1988, Hansen and Martins, 1996). The use of evolutionary models specifically constructed for the genetic basis of the evolution of genetic variability certainly gave us more powerful and precise results.
In contrast to that observed for statistical models such as Brownian motion (Gittleman and Kot, 1990; Diniz-Filho, 2001; Martins et al., 2002), our models based on real biological properties of the trait do not lead to linear decreases in phylogenetic signal. In our simulations, it depends on the parameter values. This non-linear decrease was also observed for some neutral models of phenotypic traits evolution studied by Hansen and Martins (1996).
Inheriting genetic variability
The process that retains phylogenetic signal in genetic variability can be called a “phylogenetic inheritance” of this variability because it is consequence of the ancestor-descent relationships between individuals, populations and species. The phylogenetic inheritance of genetic variability could occur by at least three main processes: i) during speciation a part of genetic variability can be maintained from ancestor to descent species; ii) similar mutation rates in phylogenetically close species can help to maintain the same level of variability throughout the species life; and iii) effects of phylogenetic signal in life history traits (Blomberg et al., 2003) (such as dispersion and environmental tolerance) that
influence populations demographically and, as a consequence, the amount of variation can also create a phylogenetic signal in genetic variability.
Predictions for natural species
Population genetic traits have never been used as a phylogenetic character and analyzed in a phylogenetic comparative way since the recent paper of Duminil et al. (2007). Duminil et al. reevaluated the relationships between genetic structure and species traits in seed plants. The phylogenetic signal in the genetic structure was estimated by using taxonomy as an approximation of the phylogenetic relationships, and the estimates were up to 79% of variance in genetic structure, explained by the taxonomic hierarchy. Although the phylogenetic signal may be overestimated by the use of taxonomy instead of phylogeny, the results of Duminil et al. (2007) point to the need to study and analyze population genetic traits in a phylogenetic comparative way.
Considering the results we obtained, we can make some considerations and predictions of the phylogenetic signal that can be observed in natural species. For the simulated models of evolution and for the parameter values we used, the heterozygosity in a species will almost reach the equilibrium value in a species lifetime. However, we allowed variation just in effective population size, which easily varies between species and seem to be related to heterozygosity in nature (Frankham, 1996). In natural populations and species, mutation rate, migration rate and lifetime must also vary between species, increasing heterogeneity between species. The presence of other forces such as natural selection and demographic events will also help to increase the dynamics of heterozygosity change in time (Crow and Kimura, 1970).
Besides the main importance population geneticists have attributed to the stochastic process for the distribution of genetic variability, many population genetic works, including some reviews, gave natural selection large importance on the maintenance of genetic variability, either in microevolution (Nevo, 1978; Hilbish and Koehn, 1985; Hedrick, 1986; Kreitman and Akashi, 1995; Eanes, 1999; Panova and Johannesson, 2004) or in macroevolution (Lloyd and Gould, 1993; Lieberman and Vrba, 2005). Simulations of the O-U process for phenotypic traits are used to represent evolution by stabilizing selection and usually retain lower phylogenetic signal than stochastic models (Felsenstein, 1988). As we reported here for mutation and migration, the higher the strength of selection, the lower the retained phylogenetic signal was (Hansen and Martins, 1996; Diniz-Filho, 2001; Martins et al., 2002).
The dynamics of microevolutionary forces will define the change of heterozygosity in time and its equilibrium point, which can be the loss of heterozygosity within the species lifetime, as we observed in some simulations. In conditions where heterozygosity reaches an equilibrium point within the species lifetime, we can predict the estimate of lower phylogenetic signal simply because there is just a little or no variation between species to test. However, in a natural scenario the action of forces other than the ones we tested here must occur and an equilibrium point for heterozygosity must be improbable to be reached. Therefore, we predict that in natural species the phylogenetic signal estimates may be generally higher than the estimates for the simulated species.
The pattern produced by the three models we used can be used further to compare the expected phylogenetic signal for a given phylogeny and what is observed from data of real species. This procedure is common for phenotypic traits using the Brownian motion as evolutionary neutral model (Felsenstein, 1981; Martins, 1994). For genetic variability, we
are now able to make the same comparisons using precise models and to obtain more powerful results.
Phylogenetic signal in genetic variability
In spite of the variation in phylogenetic signal retained under different evolutionary models, we have to emphasize that it must exist for genetic variability and must not be ignored. There are some consequences for the existence of a phylogenetic signal in genetic variability, for different levels of biological organization, the level of populations and the level of species, and also for statistic analysis of genetic variability data. At the population level, genetic variability must evolve from a starting point determined by the biological processes acting at the species level. It means that not all the variability values we observe within a species can be attributed to its life–history traits and that between populations of the same species we could also have some non-independence for genetic variability.
At the species level, the existence of a phylogenetic signal in genetic variability points to a heritability of genetic variability between species. This heritability would be a basis for a species selection process. Lloyd and Gould (1983) and Gould (2002) treat variability as a case of species selection in the broad sense but a genuine case under the emergent fitness approach. This is a controversial case, as Liberman and Vrba (2005) argued that genetic variability could be a case of species selection in the narrow sense if there is heritability at the species level. However, they conclude that “most differences in variability between clades are evanescent and either are not passed on to descent species or have no causal bearing on species sorting.” It has been argued that the nonrandom
species sort of genetic variability would be a by-product of other traits heritable at the species level (Lieberman and Vrba, 2005).
Our results give a theoretical basis for the existence of an ancestral to descent maintenance of genetic variability, producing a heritability of genetic variability itself, making possible the species selection guided by this trait in a narrow sense as an emergent species-level character. The causal bearing on species sorting is still considered controversial by Lieberman and Vrba (2005) in spite of the evidence for influence of genetic variability on species extinction (Spielman et al., 2004).
In analyzing data statistically, the fact that actual species are connected by an underlying genealogical process increases the type I error. Species are not independent units as supposed by classic statistical methods; they have a degree of dependence that directly reflects their phylogenetic relationships (Felsenstein, 1985). In comparing characteristics between species, the use of any phylogenetic comparative method gives more precise results than the use of traditional statistics (Martins et al., 2002).
In previous analyses of the distribution of genetic variability among species, statistical errors could have occurred not only as a consequence of the phylogenetic signal in genetic variability but also as a consequence of phylogenetic signal in the life-history traits used in comparisons (Blomberg, Garland and Ives, 2003). The effect of phylogeny on life-history traits related to genetic variability would be one of the factors for the phylogenetic inheritance of genetic variability itself, and can overestimate these relationships in traditional analyses (Blomberg, Garland and Ives, 2003, Garland, Bennett and Rezende, 2005).
Part of the associations between genetic variability and life-history traits found in the studies of Nevo (1978 and 1983), Mukherjee et al. (1987), Ward et al. (1992), Frankham
(1996), Frankham (1998), Garner et al. (2005), and Spielman et al. (2004) could be attributed to the phylogenetic relationships connecting the species analyzed. Even in works that had taken care to divide species into taxonomic groups for analysis we can identify two problems: taxonomy does not always correspond to species phylogeny, and the highest phylogenetic signal must be between close-related species for which taxonomy is poorly informative (Felsenstein, 1985).
Our results presented evidence not only for a phylogenetic signal in genetic variability but also for differences in correlogram profiles produced by different evolutionary models and by different combinations of their genetic and demographic parameters. Considering all the studies in the literature that have reported on phylogenetic comparative analysis and the results we obtained here, we suggest that previous analyses on the distribution of genetic variability among species and its association with other ecological and life-history traits must be viewed with caution. Future analysis could use the results we obtained to compare the correlogram profile for genetic variability data of a specific group of organisms with those we obtained, by using similar statistical analysis.
ACKNOWLEDGMENTS
We thank SF Reis for help in the choice of models, for valuable advice on our analysis and comments on the manuscript. We also thank FF Jesus for help in the comprehension of evolutionary models and anonymous referees for valuable comments on the paper. J. José’s research is supported by Fundação de Amparo a Pesquisa do Estado de São Paulo (FAPESP), grant no. 04/13080-3.
REFERENCES
Altukhov YP and Salmenkova EA (2002). DNA polymorphism in population genetics.
Russ J Genet. 38: 989-1008.
Blomberg SP and Garland T (2002). Tempo and mode in evolution: Phylogenetic inertia, adaptation and comparative methods. J Evolution Biol. 15: 899-910.
Blomberg SP, Garland T and Ives AR (2003). Testing for phylogenetic signal in comparative data: Behavioral traits are more labile. Evolution. 57: 717-745. Cheverud JM, Dow MM and Leutenegger W (1985). The quantitative assessment of
phylogenetic constraints in comparative analyses - sexual dimorphism in body-weight among primates. Evolution. 39: 1335-1351.
Crow JF and Kimura M (1970). An introduction to population genetics theory. Harper and Row, New York.
Diniz-Filho JAF (2001). Phylogenetic autocorrelation under distinct evolutionary processes. Evolution. 55: 1104-1109.
Diniz-Filho JAF, Sant'Ana CER and Bini LM (1998). An eigenvector method for estimating phylogenetic inertia. Evolution. 52: 1247-1262.
Dobzhansky T (1939). Genetics and the origin of species. Columbia University Press, New York.
Duminil J, Fineschi S, Hampe A, Jordano P, et al. (2007). Can population genetic structure be predicted from life-history traits? Am Nat. 169: 662-672.
Eanes WF (1999). Analysis of selection on enzyme polymorphisms. Annu Rev Ecol Syst. 30: 301-326.
Felsenstein J (1981). Evolutionary trees from gene frequencies and quantitative characters: Finding maximum likelihood estimates. Evolution. 35: 1229-1242.
Felsenstein J (1985). Phylogenies and the comparative method. Am Nat. 125: 1-15.
Felsenstein J (1988). Phylogenies and quantitative characters. Annu Rev Ecol Syst. 19: 445-471.
Frankham R (1996). Relationship of genetic variation to population size in wildlife. Conserv
Biol. 10: 1500-1508.
Frankham R (1998). Inbreeding and extinction: Island populations. Conserv Biol. 12: 665-675.
Garland T, Bennett AF and Rezende EL (2005). Phylogenetic approaches in comparative physiology. J Exp Biol. 208: 3015-3035.
Garner A, Rachlow JL and F. KJ (2005). Patterns of genetic diversity and its loss in mammalian populations. Conserv Biol. 19: 1215-1221.
Gittleman JL and Kot M (1990). Adaptation: Statistics and a null model for estimating phylogenetic effects. Syst Zool. 39: 227-241.
Gould SJ (2002). The structure of evolutionary thought. Harvard University Press, Cambridge.
Hansen TF and Martins EP (1996). Translating between microevolutionary process and macroevolutionary patterns: The correlation structure of interspecific data.
Evolution. 50: 1404-1417.
Hedrick PW (1986). Genetic polymorphism in heterogeneous environments: A decade later. Annu Rev Ecol Syst. 17: 535-566.
Hedrick PW, Ginevan ME and Ewing EP (1976). Genetic polymorphism in heterogeneous environments. Annu Rev Ecol Systematics. 7: 1-32.
Hilbish TJ and Koehn RK (1985). The physiological basis of natural selection at the lap locus. Evolution. 39: 1302-1317.
Kimura M (1983). The neutral theory of molecular evolution. Cambridge University Press, Cambridge.
Kimura M and Crow JF (1964). The number of alleles that can be mainteined in a finite population. Genetics. 49: 725-738.
Kreitman M and Akashi H (1995). Molecular evidence for natural selection. Annu Rev Ecol
Syst. 26: 403-422.
Lewontin RC (1985). Population genetics. Annu Rev Genet. 19: 81-102.
Lewontin RC (2002). Directions in evolutionary biology. Annu Rev Genet. 36: 1-18. Lieberman BS and Vrba ES (2005). Stephen Jay Gould on species selection: 30 years of
insignt. Paleobiology. 31: 113-121.
Lloyd EA and Gould SJ (1993). Species selection on variability. Proc Natl Acad Sci USA. 90: 595-599.
Martins EP (1994). Estimating rates of character change from comparative data. Am Nat. 144: 193-209.
Martins EP, Diniz JAF and Housworth EA (2002). Adaptive constraints and the
phylogenetic comparative method: A computer simulation test. Evolution. 56: 1-13. Mukherjee M, Skibinski DOF and Ward RD (1987). A simulation study of the neutral
evolution of heterozygosity and genetic distance. Heredity. 58: 413-423. Nei M (1972). Genetic distance between populations. Am Nat. 106: 283-&.
Nevo E (1978). Genetic variation in natural populations - patterns and theory. Theor Popul
Biol. 13: 121-177.
Nevo E, Beiles A and Ben-Shlomo R (1983). The evolutionary significance of genetic diversity: Ecological, demographic and life-history correlates. In: Evolutionary dynamics of genetic diversity (Mani GS, eds.). Springer-Verlag, Manchester, 13-213. Nielsen R (2005). Molecular signatures of natural selection. Annu Rev Genet. 39: 197-218. Panova M and Johannesson K (2004). Microscale variation in aat (aspartate
aminotransferase) is supported by activity differences between upper and lower shore allozymes of littorina saxatilis. Mar Biol. 144: 1157-1164.
Skibinski DOF and Ward RD (1982). Correlation between heterozygosity and evolutionary rate of proteins. Nature. 298: 490-492.
Skibinski DOF and Ward RD (1998). Are polymorphism and evolutionary rate of allozyme proteins limited by mutation or selection? Heredity. 81: 692-702.
Spielman D, Brook BW and Frankham R (2004). Most species are not driven to extinction before genetic factors impact them. Proc Natl Acad Sci USA. 101: 15261-15264. Ward RD and Skibinski DOF (1985). Observed relationships between protein
heterozygosity and protein genetic distance and comparisons with neutral expectations. Genet Res. 45: 315-340.
Ward RD, Skibinski DOF and Woodwark M (1992). Protein heterozygosity, protein structure, and taxonomic differentiation. Evol Biol. 26: 73-159.
Capítulo 2
P
HYLOGENETIC
S
IGNAL
A
FFECTS
T
HE
L
IFE
H
ISTORY
Phylogenetic signal affects the life history
importance for genetic diversity
Juliana José1, José Alexandre F. Diniz-Filho2
1Doutrado em Genética e Biologia Molecular, Instituto de Biologia,
Universidade Estadual de Campinas, Campinas, São Paulo, Brasil. E-mail: jj@unicamp.br.
2Laboratório de Ecologia Evolutiva, Departamento de Biologia
Geral, Universidade Federal de Goiás, Goiânia, Goiás, Brasil.
Abstract. The biological fact that species are connected by a shared evolutionary history results in phylogenetic signal in a species trait and yields a statistical dependency between species traits. Genetic diversity is an important trait for evolutionary biology and has been extensively studied comparatively, but species evolutionary history had not been accounted for. We revisited the studies on genetic variability and analyzed the factors that influence genetic variability including species evolutionary history through the use of phylognetic comparative methods. Phylogenetic signal on genetic diversity was estimated for sub-samples of an allozymic database we construct with 1521 animal species of a wide variety of taxonomic groups, using either the hierarchical taxonomic rank or molecular phylogenies available in the literature. Different groups of animals showed a significant phylogenetic signal retained in their evolution and when accounting for the phylogenetic signal and the statistical dependency, the correlated evolution between genetic diversity and population and life history traits were reduced. In the same way, when we analyzed the dataset of a previous work accounting for the species evolutionary history, almost all the previous correlations observed between genetic diversity and life history traits disappear. The evolutionary history revealed to be an important factor influencing genetic diversity and phylogenetic comparative methods revealed to be a need even in studies of population genetic traits.
INTRODUCTION
A species’ evolutionary history plays an important role shaping the attributes it currently displays and the directions and rates of evolution open to it (Gould and Lewontin 1979, Felsenstein 1985, Cheverud 1985, Garland et al. 2005). Species should not be viewed as adaptive automatons but rather as entities with history that are connected, in a higher or lower degree, by a common ancestor because of the genealogical process of ancestry and decent between individuals, lineages and species (Gould 1980, Cheverud et al. 1985). The biological fact that species are connected, as part of a hierarchically structured phylogeny, yields the statistical problem that species cannot be regarded for statistical purposes, as if drown independently from the same distribution (Felsenstein 1985).
To overcome the statistical problem of non-independence many phylogenetic comparative methods have been proposed both to estimate the correlation of two traits among species (Cheverud et al. 1985, Felsenstein 1985, Gittleman and Kot 1990, Diniz-Filho et al. 1998, Blomberg et al. 2003) and to estimate the phylogenetic signal retained in a trait. The term phylogenetic signal have been widely used to represent the phylogenetic dependency or effect and was recently defined by Blomberg and Garland (2002) as the tendency of related species to resemble each other more than they resemble species drown at random from a tree.
Many species’ traits have been studied in a phylogenetic comparative way, as body mass in birds, rodents, carnivores and primates (Garland et al. 1993, Ebensperger and Cofre 2001, Rezende et al. 2002, Smith and Cheverud 2002), stride frequency on shore birds (Gibert and Huey 2001), enzyme activity in fishes (Pierce and Crawford 1997) and population substructure in seed plants (Duminil et al. 2007). Unfortunately, species’
genetic traits are not common on the complete list of comparative studies. In spite of the large amount of meta-analysis involving genetic traits, there is just one very recent report on the comparative phylogenetic analysis of the population substructure on seed plants (Duminil et al. 2007). Duminil et al. detected phylogenetic signal in the genetic substructure that revealed the need to study and analyze population genetic traits also in a phylogenetic comparative way.
A very important trait derived from the population genetics is the genetic diversity. Genetic diversity is a central concept of evolutionary biology that has been
considered to be an important trait at the species level, resting on the assumption that the reservoir of genetic variability acts as hedge against extinction (Dobzhansky 1939) and thus is an important pre-requisite for evolutionary change (Hedrick 1986, Lewontin 2002). Species’ genetic diversity has been linked to many features as organism complexity (Nevo et al. 1984, Lynch and Conery 2003) and species ability to respond to environmental changes (Nevo et al. 1984, O’brien 1994). A lack of diversity is typically considered as evidence for a small or declining, potentially endangered population (Frankham 1998, 2002, Amos and Balmford 2001, Spielman et al. 2004).
Other features of species populations and natural history can also affect the genetic polymorphism, including population structure (Nevo et al. 1984, Ward et al. 1992, Frankham 1996, Reed et al 2003, Cherry 2003), population bottlenecks (O’brien 1994),
natural selection (Maynard-Smith 1974, Lieberman and Vrba 2005), life cycle (Caballero
and Hill 1992), and mating systems (Nevo et al. 1984, Ward et al. 1992, Hedrick 2005).
The population genetics historically considered the genetic diversity of populations a trait that micro-evolve, as a result of a balance between the microevolutionary forces (Kimura 1983, Hedrick et al. 1976, Nevo 1978, Hedrick 1986). Nevertheless, the
macroevolutionary patterns of genetic diversity have been considered an average result for the combined effects of microevolutionary processes. Meta-analyses of genetic diversity were extensively done already, searching for causal relationships between the species’ genetic diversity and species’ life-history characteristics (Nevo 1978, Mukherjee e al. 1987, Ward et al. 1992, Frankham 1996, Frankham 1998, Spielman et al. 2004, Garner et al. 2005, Bazin et al. 2006).
Nevo (1978, et al. 1984) found different degrees of allozyme genetic variability in species of different life zones, biota, habitats, specialization levels and also different taxonomic groups and concluded that genetic variability can be seen as an adaptation to environment, in space and time. Ward et al. (1992) not only found relationships between allozyme genetic variability and some life-history traits but also relationships with the biochemical properties of each enzyme. More recent works showed relationships between species genetic diversity and population size, inbreeding rate, fitness and extinction risk (Frankham 1996, Frankham 1998, Reed and Frankham 2001, Reed et al. 2003, Spielman et al. 2004, Garner et al. 2005, Bazin et al. 2006).
Almost all of the meta-analysis of genetic diversity do not took in account the evolutionary history that connect species in a hierarchical way and the following statistical problem that arises in genetic diversity and life-history covariance. The caution with species’ evolutionary history can be seen in a few meta-analysis of genetic diversity (Ward et al. 1992, Skibinski and Ward 1992, 1998, Reed et al 2003, Spielman et al. 2004), but those analyses still lack the statistical framework of phylogenetic comparative methods.
The influence of evolutionary history in the genetic diversity of current species may appear as a phylogenetic signal. José et al. (in press) have proposed that neutral models of the evolution of genetic diversity may leave a phylogenetic signal during their
