Review Hybrid fitness across time and habitats Michael L. Arnold1 and Noland H. Martin2 1 Department of Genetics, University of Georgia, Athens, GA 30602, USA 2 Department of Biology, Texas State University–San Marcos, San Marcos, TX 78666, USA There has been considerable debate about the role of vation has been used to argue that genetic exchange via hybrids in the evolutionary process. One question has horizontal gene transfer (involving prokaryotic lineages) or involved the relative fitness of hybrid versus non-hybrid viral recombination can result in similar outcomes to those genotypes. For some, the assumption of lower hybrid obtained through sexual reproduction (i.e. via natural fitness continues to be integral to their concept of hybridization [7]). Specifically, novel evolutionary lineages species and speciation. In contrast, numerous workers might arise that possess elevated fitness, relative to the have suggested that hybrid genotypes might demon- parental lineages, in novel environments. It would there- strate higher relative fitness under various environmen- fore seem important to broaden the discussion to include tal settings. Of particular importance in deciding viral and prokaryotic lineages. Second, in addition to the between these opposing hypotheses are long-term consideration of only eukaryotes, when the previous review analyses coupling ecological and genetic information. was published there was a lack of information that would Although currently rare, such analyses have provided a allow a fine-scale genotypic description of different test of the fitness of hybrid genotypes across gener- hybrids. The technical revolution of the last 15 years ations and habitats and their role in adaptation and has alleviated this deficiency by making possible the sim- speciation. Here we discuss examples of these analyses ultaneous analysis of numerous loci. Finally, fitness esti- applied to viruses, prokaryotes, plants and Darwin’s mates were generally based upon only one component (e.g. Finches. survivorship), and these estimates were almost never derived from analyses across generations or habitats Hybrid fitness as a testable hypothesis (but see [8]). In their 1995 Trends in Ecology and Evolution review, In this article, we review recent examples from viral, Arnold and Hodges [1] argued that plant and animal prokaryotic and eukaryotic groups that have provided infor- hybrids could show reduced, equivalent or higher fitness mation on the relative fitness of hybrid genotypes. We will than their parents depending upon both their genotypic focus specifically on examples that highlight the fundamen- makeup and the environment in which they occurred. tally important role played by long-term investigations of These authors concluded that genotype by environment hybrid fitness that span multiple generations and habitats, interactions would affect the fitness of hybrids just as they and which have a detailed genetic component. Together do nonhybrids. Though not a new idea [e.g. see 2–4], the these examples provide a definition of the genetic architec- suggestion that hybrids could possess higher fitness than ture of hybrid fitness and allow a test of models that predict ‘pure species’ genotypes ran counter to the prevailing uniformly low or, alternatively, variable hybrid fitness. conceptual framework. This framework rested firmly on This, in turn, provides the opportunity to highlight the the related hypotheses that speciation must occur under potential role of genetic exchange in adaptive evolution. allopatric conditions, and that lineages were not ‘good In particular, the findings discussed provide tests for adap- biological species’ if they exchanged genes with other tive trait introgression and adaptive lineage formation. species (see [5] for a review). These assumptions resulted in the conclusion that hybridization would almost never Fitness of recombinant viruses and bacteria occur during speciation (because species had to arise in Because of the implications for human health, pathogenic allopatry), but when it did rarely occur, genetic exchange viral and bacterial lineages are some of the best-studied would only lead to hybrids that were unfit relative to their organisms in terms of their evolutionary origins and geno- progenitors. mic constitutions. The evolution of some of these disease- The review by Arnold and Hodges presented a challenge causing lineages has been traced across time, across differ- to the prevailing paradigm by providing examples of hybrid ent portions of the human population and/or in different genotypes that did indeed run the gamut from less to more niches within the bodies of humans. These analyses have fit relative to their non-hybrid parents [1]. However, there not only provided tests for recombination between diver- were important deficiencies in the types of data considered gent lineages during the evolution of pathogens, but have in this review. First, Arnold and Hodges [1] only considered also highlighted the role such genetic exchange events eukaryotic groups, whereas it has been recognized for might have played in creating ‘hybrid’ organisms with decades that horizontal transfer between prokaryotic increased fitness. lineages can result in adaptive evolution (e.g. the acqui- sition of antibiotic resistance; [6]). Recently, this obser- Influenza By comparing yearly isolates of avian influenza viruses Corresponding author: Arnold, M.L. ([email protected]). from 1999 through 2004, Li et al. [9] were drawn to the 530 0169-5347/$ – see front matter ß 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.tree.2010.06.005 Trends in Ecology and Evolution 25 (2010) 530–536 Review Trends in Ecology and Evolution Vol.25 No.9 conclusion that the ‘H5N1’ viral lineage was endemic in exchange. Like influenza, this species evolves at an extre- Asian poultry populations and that it presented a threat to mely high rate, with a significant proportion of its adaptive human populations through reassortment with human- change due to the exchange of genes between divergent associated H5N1 viruses. Indeed, such viral reassortants lineages. Also like the influenza studies discussed above, caused the influenza pandemics of 1918, 1957 and 1968; the analysis of samples collected across time and ecological pandemics that claimed hundreds of millions of lives [10]. setting provided the information necessary to identify Specifically, the 1918, 1957 and 1968 influenza strains better-adapted ‘hybrids.’ Ranging from the 1980’s to the originated from recombination events giving rise to gen- 2000’s, and from healthy adults to children suffering from a omes consisting of genes from viruses found in humans, variety of pathologies, the temporal and niche breadth other mammals and birds [10]. The unique hybrid combi- encompassed in this study was extensive [12]. By sequen- nations of genes gave rise to viruses with increased patho- cing the genomes of these diverse isolates, it was concluded genicity that reflected an increased fitness relative to that horizontal transfers of various genes allowed the previous lineages. diversification of metabolic functions associated with niche Recently, the origin of a novel recombinant influenza adaptation in these E. coli strains [12]. lineage resulted in extensive, worldwide efforts by various Ogura et al. [13], studying a series of pathogenic, enter- governments and health agencies to reduce the possibility ohemorrhagic E. coli isolates, drew very similar con- of another pandemic. The so-called ‘swine-flu’ of 2009–2010 clusions to those of Touchon et al. [12] with regard to was, like all other influenza variants that infect humans, a the importance of gene acquisition to the adaptation of complex reassortant viral lineage (Figure 1) [11]. This bacterial species. Specifically, although the four enterohe- particular ‘H1N1’ virus is comprised of genes derived from morrhagic strains analyzed were not closely related phy- avian-, swine- and human-associated viral populations. By logenetically, they were found to share the same toxin examining samples across time and space, it was possible producing genes along with other genetic elements import- for Smith et al. [11] to identify the sources of all eight viral ant for their particular pathogenic niche [13]. The obser- genes (Figure 1) and to date the various divergence events vation that much of the remainder of the genomes of these of the genes. In particular, they concluded that: (i) PB2 and pathogens reflected divergent evolutionary histories, PA were of avian origin; (ii) PB1 came from a human virus; whereas the toxin genes indicated a recent common ances- (iii) HA, NP and NS were from one clade of swine influenza try, supported the hypothesis of the horizontal acquisition viruses; and (iv) NA and M originated from avian-like of the toxin loci. viruses found in Eurasian swine populations [11]. The mean divergence dates of these various components of Staphylococcus aureus the 2009 virus from other similar gene sequences were Horizontal gene transfer among bacterial species is well 1992 (N), 1997 (M) and 1998 (PB1, PB2, PA, H, NP, NS) established as a mechanism by which adaptations to new [11]. The divergence times were calculated using known niches are transferred or established [6]. As discussed dates of origin for some 20th Century viral lineages as above for E. coli, this is true for both commensal and parameters for a Bayesian analysis [11]. Thus, divergence pathogenic strains. Often, host shifts associated with and genetic exchange among viral lineages produced yet