Introduction
Smallholder communal farmers in Africa mainly raise indigenous village chickens. The chickens play a major role in the socio-economic aspects of smallholder communities and are an integral part of their nutrition. Generally, households within communities raise these chickens with minimal management interventions (FAO, 1995) that varies depending on the demographic and socio-economic statuses (Ekue et al., Reference EKUE, PONÉ, MAFENI, NFI and NJOYA2002). The chickens scavenge for food from their surrounding environment, which they utilise to meet their growth, reproduction and maintenance needs (Ekue et al., Reference EKUE, PONÉ, MAFENI, NFI and NJOYA2002). There is normally little supplementary feed provided.
In addition, the chickens face daily challenges such as exposure to parasitic and bacterial pathogens as well as harsh environmental conditions. Pathogenic diseases are a major constraint accounting for large mortalities (FAO, 1995). Poor shelter results in increased vulnerability to predators and extreme environmental conditions. This all contributes towards low productivity and survivability of village chickens under the extensive system (Ekue et al., Reference EKUE, PONE, NFI and NJOYA2006; Abdelqader et al., Reference ABDELQADER, WOLLNY and GAULY2007). Numerous studies on village chickens have yet to address the question of the probable underlying mechanisms that shape the phenotypic and genetic diversity observed in these populations and make them survive the harsh and extreme production environments.
Understanding of the genetic architecture of village chickens is a vital step in setting up their conservation and genetic improvement programmes. Studies on population genetic structures were previously limited by the sparse genome coverage of the microsatellite and other autosomal markers available that time and to the use of mtDNA D-loop sequences which were mostly limited to the use of few base pairs of biallelic markers (Cuc et al., Reference CUC, MUCHADEYI, BAULAIN, EDING, WEIGEND and WOLLNY2006; Muchadeyi et al., Reference MUCHADEYI, EDING, WOLLNY, GROENEVELD, MAKUZA, SHAMSELDIN, SIMIANER and WEIGEND2007; Granevitze et al., Reference GRANEVITZE, HILLEL, CHEN, CUC, FELDMAN, EDING and WEIGEND2007; Muchadeyi et al., Reference MUCHADEYI, EDING, SIMIANER, WOLLNY, GROENEVELD and WEIGEND2008; Mtileni et al., Reference MTILENI, MUCHADEYI, MAIWASHE, CHIMONYO, GROENEVELD, WEIGEND and DZAMA2011a; Reference MTILENI, MUCHADEYI, MAIWASHE, GROENEVELD, GROENEVELD, DZAMA and WEIGEND2011b; Mwacharo et al., Reference MWACHARO, BJØRNSTAD, MOBEGI, NOMURA, HANADA, AMANO, JIANLIN and HANOTTE2011).
The first draft of the chicken genome paved way to the use of genome wide high-density single nucleotide polymorphism (SNP) markers and powerful statistical tools (Hillier et al., Reference HILLIER, MILLER, BIRNEY, WARREN, HARDISON, PONTING, BORK, BURT, GROENEN and DELANY2004). Whole genome SNP chips have the potential to unravel the genetic composition of extensively raised chicken populations. The availability of whole genome SNP and sequence data is useful in understanding the evolutionary history of populations and in estimating factors such as linkage disequilibrium (LD) and effective population size (Ne) in the absence of records and pedigree information.
This review examines the challenges associated with communal village chicken production and their potential role in shaping genetic diversity of the chicken populations. Measures of, and factors affecting, effective population size (Ne) are explained. The concept of linkage disequilibrium and the factors that influence it as well as its use in the estimation of population demographic parameters will also be discussed.
Contribution of farming environment and production performance to the genetic composition of village chickens
Rural communities rear chickens mainly under the extensive or scavenging system where they are free to move around and search for feed in the surrounding environment (Ekue et al., Reference EKUE, PONE, NFI and NJOYA2006; Abubakar et al., Reference ABUBAKAR, AMBALI and TAMJDO2007). In village poultry production, the environmental conditions determine feed availability for the chickens and the disease pathogens and climatic stress they are exposed to. The genetic composition of the chickens determines how their populations cope with production challenges. Only genotypes that are adapted to the environment are able to utilise the available feed and those that can sustain, tolerate or resist diseases can survive in such environments (Hall and Bradly, 1995). Environmental conditions play a selective role (natural selection) in the village chicken population by removing animals that cannot utilise poor quality feeds and those that are susceptible to local diseases (Ekue et al., Reference EKUE, PONE, NFI and NJOYA2006; Mtileni et al., Reference MTILENI, MUCHADEYI, MAIWASHE, PHITSANE, HALIMANI, CHIMONYO and DZAMA2009). Factors that affect production in rural conditions vary across the seasons of the year and between villages and include sheltering, feeding, health and flock dynamics.
Sheltering, feeding and health
Poor housing contributes to the low productivity in chicken production systems (Abdelqader et al., Reference ABDELQADER, WOLLNY and GAULY2007; Abubakar et al., Reference ABUBAKAR, AMBALI and TAMJDO2007). The extensively raised chickens are constantly exposed to predators (Moges et al., Reference MOGES, MELLESSE and DESSIE2010). Sub-standard shelter therefore poses a selection pressure on village chickens eliminating individuals that fail to survive in the face of predators and the harsh climate. The genetic profiles of village chickens tend to be biased towards those genotypes that can survive and utilise poor quality and limited feed resources. Genotypes associated with slow growth rates and lower mature body sizes are most likely to be supported by this low-input system. The pathogenic and parasitic environment these chickens are exposed to leaves them vulnerable to infections (Mwale et al., Reference MWALE, BHEBHE, CHIMONYO and HALIMANI2005). The village chicken production system has no bio-security systems in place to manage outbreaks (Muchadeyi et al., Reference MUCHADEYI, WOLLNY, EDING, WEIGEND and SIMIANER2009). Regardless of the high exposure to diseases and disease-causing pathogens, generally village chicken flocks tend to survive and are often able to produce optimally. It has been hypothesised by other researchers that production challenges and selection pressures result in unique and advantageous genotypes being selected and promoted in such populations (Magothe and Kahi, Reference MAGOTHE and KAHI2011; Oke, Reference OKE2011; Ajayi et al., Reference AJAYI, AGAVIEZOR and WIHIOKA2013; Egahi et al., Reference EGAHI, DIM and MOMOH2013; Adomako et al., Reference ADOMAKO, OLYMPIO, HAGAN and HAMIDU2014).
Production performance of village chickens
Generally, village chicken growth and reproductive performance is regarded as poor because of sub-optimal management, lack of resources (Tadelle et al., Reference TADELLE, MILLION, ALEMU and PETERS2003; Gondwe, Reference GONDWE2004) and use of unimproved genotypes (Besbes, Reference BESBES2009). Village chickens are kept in low input and extensive production systems where there is very little protection against extreme weather elements, disease pathogens and predators (Mapiye et al., Reference MAPIYE, MWALE, MUPANGWA, CHIMONYO, FOTI and MUTENJE2008). Studies have observed high variability in growth and other production traits of village chicken populations (Msoffe et al., Reference MSOFFE, MTAMBO, MINGA, OLSEN, JUUL-MADSEN, GWAKISA, MUTAYOBA and KATULE2004). This has been attributed to both management differences and vast genetic diversity in flocks (Tadelle et al., Reference TADELLE, ALEMU and PETERS2000). There is therefore scope to improve village chickens production performance through selection and breeding programmes.
Flock sizes, composition and mating systems
Flock sizes in smallholder farming systems range from five to 20 chickens with a mixture of different phenotypic traits and production purposes (Illango et al., Reference ILLANGO, ETOORI, OLUPOT and MABONGA2002; Mtileni et al., Reference MTILENI, MUCHADEYI, MAIWASHE, PHITSANE, HALIMANI, CHIMONYO and DZAMA2009; Moges et al., Reference MOGES, MELLESSE and DESSIE2010). The number of breeders is biased towards hens (Olwande et al., Reference OLWANDE, OGARA, OKUTHE, MUCHEMI, OKOTH, ODINDO and ADHIAMBO2010) and a hen to cock ratio of 3:1 has been reported in most village chicken production systems (Mtileni et al., Reference MTILENI, MUCHADEYI, MAIWASHE, PHITSANE, HALIMANI, CHIMONYO and DZAMA2009; Moges et al., Reference MOGES, MELLESSE and DESSIE2010). Farmers’ preferred phenotypic traits are used to select breeding stock (Muchadeyi et al., Reference MUCHADEYI, WOLLNY, EDING, WEIGEND and SIMIANER2009; Moges et al., Reference MOGES, MELLESSE and DESSIE2010). It has been observed that farmers keep breeding chickens on the farm for long periods of time in order to maintain their flocks (Muchadeyi et al., Reference MUCHADEYI, WOLLNY, EDING, WEIGEND and SIMIANER2009), a practice that will result in overlapping generations and inbreeding (Ekue et al., Reference EKUE, PONÉ, MAFENI, NFI and NJOYA2002; Illango et al., Reference ILLANGO, ETOORI, OLUPOT and MABONGA2002). Breeding stock is generally selected for mature body weight, success in rearing chicks, and fertility (Muchadeyi et al., Reference MUCHADEYI, WOLLNY, EDING, WEIGEND and SIMIANER2009). Indiscriminate mating of chickens from different households within villages typically takes place when they are out scavenging for feed.
Effective population size, inbreeding and sustainability of a population
Effective population size (Ne) is defined as the ideal magnitude of the population that experiences the same amount of inbreeding, allele frequency variation, or loss of heterozygosity as the natural population under consideration (Hedrick, Reference HEDRICK2004; Tenesa et al., Reference TENESA, NAVARRO, HAYES, DUFFY, CLARKE, GODDARD and VISSCHER2007). It is used to determine the rate of loss of genetic variation, inbreeding and rate of fixation of deleterious alleles of populations (Hedrick, Reference HEDRICK2004). Populations with low genetic variation are associated with high levels of homozygosity particularly at deleterious loci that often lead to reduced performance and survival of populations (Hedrick, Reference HEDRICK2004). Ne has been used as an indicator of the risk of populations to extinction (Lee et al., Reference LEE, CHO, LIM, KIM, CHOI, PARK, KIM, KIM, KIM and YOON2011). Knowing and managing the levels of effective population size is important in ensuring sustainability of genetic diversity of a breed or population over time.
Inbreeding is defined as the mating of related individuals resulting in increased homozygosity (Wang, Reference WANG2005). The small flock sizes of village chickens, with limited gene flow and overlapping generations, can contribute to mating of relatives in the same flock. The mating system is difficult to monitor and manage properly due to the communal scavenging system and absence of pedigree records (Moges et al., Reference MOGES, MELLESSE and DESSIE2010; Uimari and Tapio, 2011). Inbreeding in a population with effective population size of less than 100 is most likely to increases the number of individuals sharing genotypes from common ancestry resulting in loss of population fitness (Keller and Waller, Reference KELLER and WALLER2002).
Methods of measuring effective population size
Different methods have been proposed of estimating Ne and include the use of demographic and genetic data, as well as population census and pedigree data (Wang, Reference WANG2005). Demographic, pedigree and population census methods are, however, challenging in extensively-raised populations because of absence of pedigree data and the intermixing of flocks. For extensively-raised and wild populations, molecular methods are the most reliable as they do not depend on the availability of pedigree data (Wang, Reference WANG2005). Molecular methods include temporal and LD-based approaches (Wang, Reference WANG2005; Lee et al., Reference LEE, CHO, LIM, KIM, CHOI, PARK, KIM, KIM, KIM and YOON2011) that can assist in detecting the changes in effective population size associated with allele frequency variance and loss of heterozygosity in populations (Wang, Reference WANG2005).
Temporal methods
Temporal methods measure changes in allele frequency (F) over time (Hedrick, Reference HEDRICK2004). The changes in allele frequency would be as a result of genetic drift, assuming that all other forces such as mutation, migration and selection remain constant (Wang, Reference WANG2005). Changes in allele frequency over time will therefore be a function of Ne, which can be estimated by taking random samples from the same population at different time intervals. The sampling assumes that the population has discrete non-overlapping generations (Waples, Reference WAPLES1989), an assumption that is often violated in free-range chicken populations.
In natural populations, the estimation of Ne using a temporal method can be challenging since it requires a sampling structure that accounts for age and sex factors and discreet generations. Gene flow or admixture of the population cannot be easily eliminated so accountability of migration in a temporal method is also necessary (Wang, Reference WANG2005). The lack of pedigree records in village chickens makes it difficult to understand the age and relatedness of individuals and implication of temporal methods for the estimation of Ne.
Linkage disequilibrium based methods
Linkage disequilibrium (LD) is the non-random association of alleles at two or more loci (Hedrick, Reference HEDRICK2004). LD is used to determine the evolutionary forces that contribute to the genetic variation of a population under consideration and is now well implemented as a tool for many population genetics studies (Hedrick, Reference HEDRICK2004; Andreescu et al., Reference ANDREESCU, AVENDANO, BROWN, HASSEN, LAMONT and DEKKERS2007; Slatkin, Reference SLATKIN2008). LD is also used for gene mapping (Slatkin, Reference SLATKIN2008). LD provides information on evolutionary history and population demography. The distribution and extent of LD across a genome has been used to study the current and ancestral genetic forces that structure human populations (Tenesa et al., Reference TENESA, NAVARRO, HAYES, DUFFY, CLARKE, GODDARD and VISSCHER2007; McEvoy et al., Reference MCEVOY, POWELL, GODDARD and VISSCHER2011). LD is one of the most common methods used to estimate Ne in the absent of population pedigree (Uimari and Tapio, 2011). It is the most convenient method as it requires only a single sampling compared to the temporal methods (Hedrick, Reference HEDRICK2004). LD, as a squared correlation coefficient of marker pair between two loci, is a function of distance between makers that is influenced by recombination rates. Theoretically, recombination events will be higher between markers that are far apart than those markers that are close to each other. Generally, the LD of marker pairs at a close distance reflects more recent Ne whilst LD of marker pairs at two different loci far apart reflect ancient Ne of population of interest (Hayes et al., Reference HAYES, VISSCHER, MCPARTLAN and GODDARD2003).
There are two commonly used measures of LD (Slatkin, Reference SLATKIN2008). The D´ measure is a standardised LD value which is dependent on allele frequencies (Slatkin, Reference SLATKIN2008). D´ is calculated from D which is the difference between the frequency of gametes carrying allele A and B (pAB) at two Loci and the product of the frequencies of those alleles (pA and pB), DAB = pAB-pApB (Hedrick, Reference HEDRICK2004). The alternative (r2) LD measure is defined as the squared correlation co-efficient between two loci (Hedrick, Reference HEDRICK2004; Slatkin, Reference SLATKIN2008). It can be used for multiple pairs of loci and is less dependent on allele frequencies (Lee et al., Reference LEE, CHO, LIM, KIM, CHOI, PARK, KIM, KIM, KIM and YOON2011).
As a statistical measure, LD is computed under certain assumptions that are applicable to different populations (Slatkin, Reference SLATKIN2008). Under the assumption of infinite size with random mating, no genetic drift and no selection, LD should approach zero with increased marker distance (Corbin et al., Reference CORBIN, BLOTT, SWINBURNE, VAUDIN, BISHOP and WOOLLIAMS2010). The consideration of these factors when calculating LD is vital as the estimation can be biased if validity of assumptions is not investigated (Corbin et al., Reference CORBIN, BLOTT, SWINBURNE, VAUDIN, BISHOP and WOOLLIAMS2010). Livestock populations are confined to the boundaries defined by farmers and are therefore sub-structured with some level of gene flow between sub-populations. New alleles introduced by mutation are generally considered to have a very low impact on LD in most livestock species since the rate of mutation is very low (Ardlie et al., Reference ARDLIE, KRUGLYAK and SEIELSTAD2002). Genetic drift in sub-structured populations and selection can either keep or eliminate new mutations (alleles) in a population that would result in their frequencies being either elevated or reduced in a population. A reduction of LD in a population can be observed if the new gametes are maintained and their frequency increases (Hedrick, Reference HEDRICK2004).
The decay of LD is affected by a degree of inbreeding in a finite population (Rao et al., Reference RAO, LIANG, NA XIA, SHEN, JUN DU, GLONG LUO, HUA NIE, ZENG and QUAN ZHANG2008). LD values can therefore be used to make inferences about the mating system of village chicken populations (Corbin et al., Reference CORBIN, BLOTT, SWINBURNE, VAUDIN, BISHOP and WOOLLIAMS2010). Genetic drift can favour increased homozygosity of which its effects will be similar to inbreeding (Hayes et al., Reference HAYES, VISSCHER, MCPARTLAN and GODDARD2003). It has been observed that village chicken population have high genetic diversity and are generally not sub-structured to promote inbreeding (Cuc et al., Reference CUC, MUCHADEYI, BAULAIN, EDING, WEIGEND and WOLLNY2006; Hassen et al., Reference HASSEN, NESER, DE KOCK and VAN MARLE-KÖSTER2009).
New alleles can be introduced into a population through gene flow (Falconer and Mackay, Reference FALCONER and MACKAY1996). Gene flow or mixing of gametes from different population also has an effect on LD (Li and Merilä, Reference LI and MERILÄ2009). The mixing up of two populations carrying different allele at different frequencies will usually result in a new gene pool characterised by higher heterozygosity levels over time. This increased heterozygosity level will result in reduced levels of LD. The extent of reduction in LD will depend on the origin of the chicken subpopulation in such a way that greatly diverse subpopulations will result in high genetic variation and lower LD values than when genes are exchanged between closely related populations (Li and Merilä, Reference LI and MERILÄ2009).
Linkage disequilibrium has been used to reveal the demographic history including trends in effective population sizes of egg laying commercial chicken populations (Qanbari et al., Reference QANBARI, PIMENTEL, TETENS, THALLER, LICHTNER, SHARIFI and SIMIANER2010a) and extensively raised village chicken populations (Khanyile et al., Reference KHANYILE, DZOMBA and MUCHADEYI2015). In other studies in a number of species, LD have been used to construct maps (Meadows et al., Reference MEADOWS, CHAN and KIJAS2008; Lee et al., Reference LEE, CHO, LIM, KIM, CHOI, PARK, KIM, KIM, KIM and YOON2011), in fine mapping quantitative trait loci (QTL) and population parameters (Tenesa et al., Reference TENESA, NAVARRO, HAYES, DUFFY, CLARKE, GODDARD and VISSCHER2007), analysis of haplotype diversity (Amaral et al., Reference AMARAL, MEGENS, CROOIJMANS, HEUVEN and GROENEN2008) and mating systems (Uimari and Tapio, Reference UIMARI and TAPIO2011; Lee et al., Reference LEE, CHO, LIM, KIM, CHOI, PARK, KIM, KIM, KIM and YOON2011). LD based analysis have found use in population genetics studies where the pedigree data is not readily available. The availability of whole genome high density SNP data has even increased the use of LD based estimates of Ne and other population parameters particularly in humans and domestic livestock that lack pedigree data.
Linkage disequilibrium estimation in the genomics era
Earlier studies measured LD in chicken using microsatellite markers that are sparsely distributed across the genome (Heifetz et al., Reference HEIFETZ, FULTON, O'SULLIVAN, ZHAO, DEKKERS and SOLLER2005) for specific chromosomal regions of interest. After the completion of the first chicken genome in 2004 (Hillier et al., Reference HILLIER, MILLER, BIRNEY, WARREN, HARDISON, PONTING, BORK, BURT, GROENEN and DELANY2004), studies have been conducted to reveal polymorphisms in the chicken genome, which consequently led to the development of genome-wide high-density SNPs arrays. The investigation of genome wide LD became feasible after the development of high-density genome wide SNP chips in most livestock species (Corbin et al., Reference CORBIN, BLOTT, SWINBURNE, VAUDIN, BISHOP and WOOLLIAMS2010; Qanbari et al., Reference QANBARI, PIMENTEL, TETENS, THALLER, LICHTNER, SHARIFI and SIMIANER2010b; Uimari and Tapio, Reference UIMARI and TAPIO2011; Lu et al., Reference LU, SARGOLZAEI, KELLY, CHANGXI, VANDER VOORT, WANG, PLASTOW, MOORE and MILLER2012). The Illumina iselect chicken SNP 60k bead chip (Groenen et al., Reference GROENEN, MEGENS, ZARE, WARREN, HILLIER, CROOIJMANS, VEREIJKEN, OKIMOTO, MUIR and CHENG2011) has been used to understand the population parameters underlying the evolutionary factors in chickens. In commercial chicken lines, LD extended over long distances compared to that of wild fowl (Wragg et al., Reference WRAGG, MWACHARO, ALCALDE, HOCKING and HANOTTE2012). In Southern African chickens, LD was found to be lower in village chickens (average of 0.06) compared to conserved chicken flocks that averaged 0.15 (Khanyile et al., Reference KHANYILE, DZOMBA and MUCHADEYI2015). The estimation of genome wide LD using these high density SNP chips has made it possible to infer effective population size and other the population demography in chickens (Andreescu et al., Reference ANDREESCU, AVENDANO, BROWN, HASSEN, LAMONT and DEKKERS2007; Qanbari et al., Reference QANBARI, PIMENTEL, TETENS, THALLER, LICHTNER, SHARIFI and SIMIANER2010a; Khanyile et al., Reference KHANYILE, DZOMBA and MUCHADEYI2015), dairy cattle (Sargolzaei et al., Reference SARGOLZAEI, SCHENKEL, JANSEN and SCHAEFFER2008) and pigs (Uimari and Tapio, Reference UIMARI and TAPIO2011).
The recently launched 600 K Affymetrix® Axiom® HD genotyping array (Kranis et al., Reference KRANIS, GHEYAS, BOSCHIERO, TURNER, YU, SMITH, TALBOT, PIRANI, BREW and KAISER2013) for chickens contains 10 times more SNPs than in the 60K. Such increased SNP density could see improvements in fine mapping of QTLs and genes as well as estimation of inbreeding, effective population size and other parameters. The plummeting cost in whole genome sequencing could lead to an increase in re-sequencing of livestock genomes including chickens. The 1000 Bull genome project for cattle (Hayes, Reference HAYES2013) is an example where large sequence data is being generated thereby allowing estimation of LD and other population genomic parameters.
Conclusions
The extensive smallholder production systems present some challenges on the survivability and productivity of village chicken populations in Southern Africa and developing countries. Such challenges have a negative impact on biodiversity and sustainability of these important animal genetic resources. The small flock sizes, reduced gene flow between geographically isolated populations and lack of performance records could lead to isolated genetic change resulting in localised inbred sub-populations. The estimation of trends in effective population size will provide insight on the number of breeding individuals and loss of genetic variation and facilitate the assessment of the risk to extinction for village chickens. Such information is important for developing future effective breeding and conservation programmes.
Genome-wide SNP data and LD analysis present opportunities for understanding effective population sizes and other demographic parameters in populations such as village chickens were pedigree records are missing. The Illumina chicken iSelect 60K SNP chip have proven to be useful in analysing background LD, and estimate trends in effective population sizes as well as in identifying regions under selection and with influence on Mendelian traits. Genome-wide high-density SNP data and LD analysis have not been widely applied in African and other village chicken populations. These genomics strategies are however expected to add new knowledge on the genetic composition and evolutionary factors influencing these populations. Whilst QTL and genome-wide association studies are difficult to implement under smallholder extensive systems of production, LD is a relatively new and exciting tool that can be used to determine QTLs and genes of interest in such populations.
