Pronounced inter- and intrachromosomal variation in linkage disequilibrium across the zebra finch genome

The extent of nonrandom association of alleles at two or more loci, termed linkage disequilibrium (LD), can reveal much about population demography, selection, and recombination rate, and is a key consideration when designing association mapping studies. Here, we describe a genome-wide analysis of LD in the zebra finch (Taeniopygia guttata) using 838 single nucleotide polymorphisms and present LD maps for all assembled chromosomes. We found that LD declined with physical distance approximately five times faster on the microchromosomes compared to macrochromosomes. The distribution of LD across individual macrochromosomes also varied in a distinct pattern. In the center of the macrochromosomes there were large blocks of markers, sometimes spanning tens of mega bases, in strong LD whereas on the ends of macrochromosomes LD declined more rapidly. Regions of high LD were not simply the result of suppressed recombination around the centromere and this pattern has not been observed previously in other taxa. We also found evidence that this pattern of LD has remained stable across many generations. The variability in LD between and within chromosomes has important implications for genome wide association studies in birds and for our understanding of the distribution of recombination events and the processes that govern them.Linkage disequilibrium (LD), which refers to the nonrandom association of alleles at two or more loci, plays an important role in evolutionary biology and gene mapping (Coop et al. 2008; Slatkin 2008). LD can reveal much about population demography and, because the extent of LD is approximately inversely proportional to the recombination rate, LD can also uncover variability in recombination rates across genomes and chromosomes (Hedrick 1988; Miyashita and Langley 1988; Daly et al. 2001; Jeffreys et al. 2001; Gabriel et al. 2002; Arnheim et al. 2007). Importantly, LD determines the power and precision of association mapping studies, directly influencing our ability to localize genes and/or loci responsible for traits and diseases (Kruglyak 1999; Weiss and Clark 2002).Studies of LD are dependent on the availability of genomic resources, and across vertebrates comprehensive genome-wide studies have been restricted to model species, in particular mammals. Studies in other organisms, however, are likely to reveal extensive variation in patterns of LD not seen in humans (Slatkin 2008), and organisms with different genomic architecture to mammals may provide novel insight into patterns of LD and genome evolution. Birds are interesting in this respect as most bird genomes are composed of many small (micro) chromosomes and relatively few large (macro) chromosomes. Linkage mapping studies have shown that microchromosomes have a higher recombination rate than their larger counterparts (International Chicken Genome Sequencing Consortium 2004; Stapley et al. 2008). Increased recombination rate is expected to reduce the amount of background LD; however, to date there has been no comprehensive analysis of how LD varies with chromosome length and across macrochromosomes and microchromosomes. With respect to LD in birds, the chicken Gallus gallus has received the most attention; studies have quantified LD in several breeds of domestic chicken but have only focused on relatively few chromosomes (Heifetz et al. 2005; Aerts et al. 2007; Andreescu et al. 2007; Wahlberg et al. 2007; Rao et al. 2008; Abasht et al. 2009). In chickens, LD extends very short distances, which reflects the high recombination rate and relatively large effective population size of domestic fowl relative to other livestock species. Passerines, which make up around half of all birds species, and diverged from chickens at least 80 million yr ago, are beginning to receive attention; however, all studies to date have either investigated single chromosomes (Backström et al. 2006), a few genomic regions (Balakrishnan and Edwards 2008), or have low marker coverage (Li and Merilä 2009).The zebra finch, the second bird to have its genome sequenced, can provide a useful target for a comprehensive LD study. In addition to having the two types of chromosomes (macro- and microchromosomes) characteristic of most birds, the zebra finch genome exhibits an unusual pattern of crossing over. Cytogenetic studies have revealed that the location of meiotic crossover events on the macrochromosomes is highly nonrandom, mostly occurring at the ends of the macrochromosomes (Pigozzi and Solari 2005; Calderón and Pigozzi 2006). This suggests there is a large recombination desert on all macrochromosomes, corresponding with the center of the metacentric chromosomes and the center of the long arm of the acrocentric chromosomes (Calderón and Pigozzi 2006). The suppressed recombination in the middle of the chromosomes is not necessarily related to the position of the centromere, nor to the presence of heterochromatin (Calderón and Pigozzi 2006), which is thought to suppress recombination. It is unknown how stable this pattern of recombination is or if this is characteristic of other zebra finch populations. If it is a persistent phenomenon of zebra finch chromosomes, it is likely to affect the extent of LD across macrochromosomes and generate extensive heterogeneity across and between chromosomes. Interestingly, this pattern has not been reported in chickens (Calderón and Pigozzi 2006; Groenen et al. 2009).Patterns of LD may also co-vary with other sequence features that are correlated with recombination rate, such as heterozygosity, GC content, and the number of genes. Recombination rate is expected to increase heterozygosity (Begun and Aquadro 1992; Nachman 2001), although this pattern may be obscured by the action of other forces such as selection and biased gene conversion (Maynard Smith and Haigh 1974; Ohta 1999). Recombination rate is also positively related to GC content and other sequence features known to co-vary with GC content, e.g., gene density, intron length, CpG motifs (Kong et al. 2002; Meunier and Duret 2004; Groenen et al. 2009). It is unknown how LD and these sequence features will co-vary in the presence of highly nonrandom recombination.The aim of this study was to construct genome-wide LD maps of the zebra finch in order to examine (1) whether LD extended further on macro- than microchromosomes, as predicted by previously described differences in recombination rate; (2) whether LD varied across macrochromosomes, corresponding to the biased location of crossing-over events on chromosome spreads; and (3) whether patterns of LD co-vary with heterozygosity, GC content, and the number of genes; and (4) to assess the stability of patterns in LD, by examining whether contemporary genome-wide variation in recombination rates, as detected by linkage mapping, is consistent with historical recombination rate variation, as inferred from LD maps.