SECTION A: CONCEPTUAL FRAMEWORK
1. What is phenotypic characterization?
In these guidelines, the term “phenotypic characterization of AnGR” is used to refer to the process of identifying distinct breed populations and describing their characteristics and those of their production environments. In this context, the term “production environment” is taken to include not only the “natural” environment but also management practices and the common uses to which the animals are put, as well as social and economic factors such as market orientation, niche marketing opportunities and gender issues. Recording the geographical distribution of breed populations is here considered to be an integral part of phenotypic characterization. Complementary procedures used to unravel the genetic basis of the phenotypes of AnGR, their patterns of inheritance from one generation to the next, and to establish relationships between breeds are referred to as molecular genetic characterization (FAO, 2010b). In essence, phenotypic and molecular genetic characterization of AnGR are used to measure and describe genetic diversity in these resources as a basis for understanding them and utilizing them sustainably.
The guidelines distinguish between two phases or levels of characterization. The term “primary characterization” is used to refer to activities that can be carried out in a single visit to the field (e.g. measurement of animals’ morphological features, interviews with livestock keepers, observation and measurement of some aspects of the production environment, mapping of geographical distribution).
The term “advanced characterization” is used to describe activities that require repeated visits. This includes the measurement of productive (e.g. growth rate, milk production) and adaptive (e.g. resistance or tolerance to specific diseases) capacities of breeds in specified production environments. 2. Non-descript populations
Because of a lack of comprehensive information on breeds’ identities and geographical distributions, many animal populations in the developing regions of the world are commonly referred to as “nondescript” or “traditional”. The inventory of local breeds in these regions is thought not to be exhaustive, and new breeds continue to be identified (e.g. Köhler-Rollefson and the LIFE Network,
2007; Wuletaw et al., 2008). It is primarily in these regions of the world that phenotypic characterization studies on AnGR are needed.
Simplified and coherent procedures for phenotypic characterization are needed in order to support countries in obtaining more complete inventories of their AnGR. These procedures need to be standardized globally to facilitate valid enumeration, analysis and reporting of breeds nationally and internationally. 3. The breed concept
The term “breed” is used in phenotypic characterization to identify distinct AnGR populations as units of reference and measurement. Diversity in AnGR populations is measured in three forms: interpopulation diversity (between breeds), intrapopulation diversity (within breeds), and interrelationships between populations. Phenotypic characterization is used to identify and document diversity within and between distinct breeds based on their observable attributes. The measurement of genetic relationships between breeds and genetic heterozygosity within breeds is the task of molecular characterization (FAO, 2010b).
The breed concept originated in Europe and was linked to the existence of breeders’ organizations.
The term is now applied widely in developing countries, but it tends to refer to a sociocultural concept rather than a distinct physical entity. Hence, the use of the term in developing countries, where most of the world’s traditional and local livestock populations are located, is different from its use in developed countries. Whereas in developed countries breeds are defined in terms of a set of phenotypic standards, breed herd books and formalized breed societies that are often supported by
CGRFA/WG-AnGR-6/10/Inf.5 7 legislation, in developing countries livestock-keeping communities and governments apply the term more loosely and identify breeds more with geographic localities, ethnic identities and the local traditions of their owners than with their phenotypic attributes. In some cases, the term is used interchangeably with “population”, “variety”, “strain” or “line”, within nationally recognized breeds.
Definitions of related terms are provided in Box 1.
Box 1: Definitions of breeds and related terms
Traditional populations: mainly local; often exhibit large phenotypic diversity; are managed by farmers and pastoralists at low selection intensity, but may be subject to high natural selection pressure; pedigree may be partially known; genetic structures are mainly influenced by migration events and mutations; population size is generally large (unless subject to erosion).
Standardized breeds: derived from traditional populations by a community of breeders based on a recognized list of “standard” breed descriptors; exhibit less phenotypic diversity as they are selected to meet minimum standards of phenotype; pedigree is partially known; their genetic structure may be influenced by important founder effects; population size may be large or small.
Selected breeds or commercial lines: derived from standardized breeds or from traditional populations through application of an economic selection objective and use of quantitative genetic methods; breeders are organized for pedigree and performance recording and selected animals are used across flocks or herds; inbreeding increases as a consequence of a high selection intensity; molecular markers may be used, for instance for parentage testing and/or for the identification of genes controlling performance; population size is generally large.
Derived lines: arise from the use of specific breeding methods such as close inbreeding; highly specialized inbred lines exhibit low genetic variability; synthetic lines are derived from crossing standardized breeds or selected lines, and exhibit a high level of genetic variability; transgenic and experimental selected lines fall under this category; population size is generally limited, except for synthetic lines.
These different types of population can be identified easily in highly commercialized populations, such as cattle, pigs and chickens in Europe. The classification may be less relevant to other species such as camelids or geese. However, it may be used as a general framework covering all types of domesticated populations.
Source: adapted from Tixier-Boichard et al. (2007).
FAO uses the following broad definition of the breed concept, which accounts for social, cultural and economic differences between animal populations, and which can therefore be applied globally in the measurement of livestock diversity:
“either a sub-specific group of domestic livestock with definable and identifiable external characteristics that enable it to be separated by visual appraisal from other similarly defined groups within the same species or a group for which geographical and/or cultural separation from phenotypically similar groups has led to acceptance of its separate identity” (FAO,
1999).
These guidelines use the same generic definition.
In addition to the task of characterizing recognized breeds, the guidelines address the task of identifying and characterizing previously unrecognized breeds from among traditional and nondescript populations. This can be done by taking into account the opinions of the livestock keepers themselves along with the use of numerical taxonomic procedures.
One essential characteristic of a breed is near complete reproductive isolation for many generations, during which mating with animals from outside the breed has been very restricted, so that the population evolves with distinctly different appearance and functions from those of other breeds
8 CGRFA/WG-AnGR-6/10/Inf.6
(FAO, 1992; 1998). In traditional livestock-keeping communities, local indigenous knowledge provides perhaps the best preliminary information available about breed identity; i.e. a particular community may claim to maintain a distinct AnGR population in a specific environment and subject to a common pattern of breeding and utilization. Köhler-Rollefson and the LIFE Network (1997) provide the following description of how the breed concept can be applied in traditional communities: “A domestic animal population may be regarded as a breed, if the animals fulfill the criteria of (i) being subjected to a common utilization pattern, (ii) sharing a common habitat/distribution area, (iii) representing largely a closed gene pool, and (iv) being regarded as distinct by their breeders”.
What is common in both traditional and industrialized communities is that breed populations are developed, maintained and influenced by humans and hence become the unit of reference for improvement and conservation. It is, therefore, appropriate that AnGR populations are identified by breed and that phenotypic characterization studies address both indigenous knowledge and active quantitative classification. Molecular tools can be used to corroborate the classification of populations into breeds.
4. Approaches to characterization
In statistical terms, phenotypic characterization can involve either of the following two approaches, depending on the type of background information available:
Exploratory approach – undertaken in situations in which no reliable background information on the existence of recognized distinct breeds in the study area is available; in such circumstances, the objective of phenotypic characterization is to investigate the existence of distinct breeds in the study area.
Confirmatory approach – undertaken in situations in which some basic information on breed identity and distribution is available; in such circumstances, the objective phenotypic characterization is to validate breed identity and provide systematic descriptions of the breeds. In situations where the available secondary information is insufficient to prepare plans for phenotypic characterization, preliminary field data will need to be collected on the identity, geographical distribution, and relative significance of AnGR populations (nationally or locally recognized breeds, non-descript populations, etc.) in the study area and hence to determine whether exploratory or confirmatory approach is required. Preliminary data collection activities may include “mapping expeditions” – journeys within the study area that serve as a means of approximating the geographical distribution of different breed types – and “rapid appraisals” – the use of a range of field-based techniques (complemented where relevant with information from secondary sources) to obtain information from local people. Rapid appraisals may include discussions in group meetings and focus groups, semi-structured interviews with individual livestock keepers and other knowledgeable “key informants” and direct observation on the part of the surveyors. A range of specific techniques have been developed for use in rapid appraisals (mapping exercises, seasonal calendars, ranking and scoring exercises, transect walks, progeny histories, etc.) and can be used to discuss the local production system with groups or individuals. Triangulation – the use of several complementary sources of information – is a key characteristic of the approach. Further information on mapping expeditions and rapid appraisals can be found in the complementary guidelines on surveying and monitoring of AnGR (FAO, 2010a).
Exploratory approach
Once the study area for characterization has been designated, the study team should seek to develop a sampling frame, i.e. set of criteria to be used to identify a sample of households and animals for data
CGRFA/WG-AnGR-6/10/Inf.5 9 collection. If the area is large, it may be necessary to stratify it into more homogenous subunits based on one or more of the following criteria:
geographical isolation of AnGR populations and their patterns of movement or migration;
known patterns of morphological and production characteristics in the AnGR populations or the existence of common breeding practices; and
historical information and indigenous knowledge on the origin of the AnGR.
The exploratory approach to phenotypic characterization also requires estimation of the total livestock population in the study area, as well as the number of livestock keepers who maintain these animals (see section C). Secondary information on the livestock populations in the study area should be sought in published and grey literature. The Domestic Animal Diversity Information System
(DAD-IS – http://dad.fao.org/) may be a useful source of background information on breed inventory and the distribution, national population sizes and risk statuses of recognized breeds.
The exploratory approach hypothesizes that the target AnGR population is homogenous and does not have phenotypically distinct subpopulations. It seeks to test this hypothesis by measuring and analysing the pattern of phenotypic diversity within the target population. Standard phenotypic data
(see Annexes 1 to 4) are collected from sample animals at the study sites.
Primary characterization (i.e. the collection of data through single field visits) falls within the exploratory approach and, for the sake simplicity, the former term is used in these guidelines when referring to this approach.
Confirmatory approach
The confirmatory approach to phenotypic characterization aims to validate information on breed identities as recorded in national AnGR inventories, literature and/or local knowledge. It presumes that known breeds have a defined geographical distribution and some common phenotypic characteristics and pattern of utilization. The standard breed descriptors set out in Annexes 1 to 4 provide a framework for collecting detailed phenotypic data. Statistical tools can be used to test whether there are significant multivariate differences among the populations in the study area and hence to validate their identities as distinct breed types. Additional genetic studies are always required to corroborate such identities.
The confirmatory approach also involves an objective assessment of documented indigenous knowledge and other indicative information. This can bring out important AnGR management issues for closer investigation (e.g. the risk status of existing breeds, emergence of new composite populations and the perceptions of communities about breed identities). This approach can be used to look more closely at differences among breed types identified during primary characterization, with a view to validating the classifications and describing how the distinct groups differ from each other.
The study team may find that additional or more up-to-date information is needed in order to draw up the sampling frame. In such cases, preparatory field work (mapping expeditions and/or rapid appraisals – see above) may need to be conducted in the study area.
Breed evaluation and comparison under on-station or on-farm management conditions (i.e. advanced characterization) are part of the confirmatory approach. Such studies focus on breeds that have already been identified and aim to provide a more comprehensive evaluation of their performance and adaptation. For the sake of simplicity, the term advanced characterization is used in these guidelines when referring to this approach.
5. Quantitative procedures for breed identification
Principles of classification
Based on the premise that livestock breeds are distinguishable by differences in appearance, conformation and dimension, quantitative procedures are used to explore breed types from traditional
10 CGRFA/WG-AnGR-6/10/Inf.6 populations by systematically assessing aggregate morphological characteristics in groups of animals, in exactly the same way taxonomists classify organisms into hierarchical groupings. Known as numerical taxonomy, these procedures explore aggregate morphological resemblances among groups of organisms to develop hierarchical groupings, assuming that the groupings may (but not necessarily) represent historical evolutionary processes associated with gross structural diversity
(Dobzhansky, 1951). When, in addition to morphological characteristics, sociocultural attributes, such as historical association with particular livestock-keeping communities in well-defined production environments, are used to define such animal groups, distinct breed types that are expected to share clearly defined heritable traits and definite areas of distribution, may be identified in line with the broad definition of breed given above. This approach has been applied, for example, among traditional goat populations in Ethiopia (FARM Africa and ILRI, 1996; Ayalew et al., 2000) and corroborated by molecular genetic studies (Ticho, 2004). Similar genetic evidence to support phenotypic breed identities has been obtained in sheep (GebreMichael, 2008), cattle (Dadi et al.,
2008) and chickens (Halima-Hassen, 2007).
Numerical taxonomic procedures that involve multivariate analysis of variance consider large numbers of observable characteristics of equal value (i.e. not weighted) in a large number of individuals and seek to classify the individuals based on their aggregate similarity. The premise behind this method of classification is that morphological variation among individual organisms is typically discontinuous and forms distinctly separate arrays, with each array comprising a cluster of individuals that possess some common characteristics. The discrete clusters are designated as races
(varieties), breeds, species, genera and so forth. The classification arrived at by using this approach is to some extent an artificial one, but the clusters themselves and the discontinuities observed between them are not abstractions on the part of the classifier (Dobzhansky, 1951; pp 3–18). In effect, the patterns of morphological variation within species can be used to identify homogenous subgroups of animals, and these subgroups can be considered breed types or varieties within breeds.
Methodology
Cluster and discriminant analyses. In this type of analysis, the units of reference (taxonomic units) are referred to as operational taxonomic units (OTU). Depending on the perceived pattern of morphological variation at the population level, the OTUs may be individual animals or sample groups of homogenous animals. In situations where there is high flock/herd-level morphological resemblance, as in the case of pastoral livestock populations, average values of sample animals – otherwise known as centroids, – are taken as OTUs. In the absence of such resemblance and in particular when breed-type identities are less clear, individual sample animals are used as OTUs.
Estimation of the degree of phenotypic resemblance (morphological, physiological, behavioural) among OTUs is a fundamental step in the analysis. Multivariate cluster analysis is then used to reorganize a heterogeneous set of taxonomic units into more homogenous groups or clusters with respect to the variables (characters) under consideration (Aldenderfer and Blashfield, 1984). If, within the sample population, there are any distinct categories that are to be considered as initial classes – for instance if local names are given to different animal populations, breed types or varieties
– discriminant analysis can be used to validate these differences (Klecka, 1980). Both cluster and discriminant analyses assume that the aggregate morphological variation is a linear combination of the individual variables, i.e. the character states or phenotypic measurements recorded from OTUs
(individual animals or centroids).
Cluster analysis is used to classify the OTUs by quantifying aggregate similarity relationships of pairs of OTUs with respect to the characters under consideration (Sneath and Sokal, 1973; p 116).
These relationships can be expressed as relative distance (i.e. resemblance) in a multidimensional
Euclidean space, with each character variable defining an axis. In the mathematical sense, the relative distance is, rather, a measure of aggregate differences – the larger the value of this distance the greater is the dissimilarity between the OTUs. Based on the computed values for all the possible pairs of OTUs a hierarchical (classification) tree can be produced (ibid.).
CGRFA/WG-AnGR-6/10/Inf.5 11
Principal components analyses. The major technical limitation involved in producing clusters directly from morphological variables is that the variables are not independent of each other, because the original variables are recorded on each OTU. The procedure known as principal components analysis (PCA) linearly transforms the original variables into a set of uncorrelated variables referred to as principal components, which explain essentially the same statistical information (variance) as the original set of variables. Each principal component is a linear combination of all the variables and has a mean of zero and variance of unity (Dunteman, 1989). However, depending on the nature of variation in the original data set, the first few principal components may account for most of the total variation. As a result, a substantially smaller set of the first (most important) principal components can explain most of the variance in the original variables, thereby reducing dimensionality (number of axes) in the corresponding hyperspace. Furthermore, independence of the transformed variables will ensure orthogonality between each of the axes. Orthogonality of the axes implies that each one of them makes an independent contribution to discrimination between the OTUs or the groups of OTUs
(i.e. clusters). The computed principal components can then be used to develop a classification tree using cluster analysis.
Method of clustering. There are several contrasting methods of clustering (Sneath and Sokal, 1973; pp 201–244; Pimentel, 1979; p 79; Aldenderfer and Blashfield, 1984), but the most widely applied method of clustering in biological systematics, as well as for subspecies-level classification, is one that is sequential, agglomerative, hierarchic and non-overlapping (abbreviated to SAHN). The method starts with t separate OTUs, agglomerates them into successively fewer than t sets, arriving eventually at a single set containing all t OTUs. The resulting taxa at any level (rank) are mutually exclusive (non-overlapping), i.e. OTUs contained within one taxon are not also members of a second taxon of the same rank. An iterative sequence of clustering is used to partition the OTUs into biologically meaningful clusters. This procedure ultimately produces a hierarchical (or classification) tree from which the desired number of groups of homogenous groupings (clusters) can be derived. As long as the goal of the analysis is to explore the general pattern of relationships between the OTUs as represented by the classification tree, the number of clusters can be determined by heuristic decisions. The choice that has the most plausible biological interpretation is the best. Consequently, this procedure is prone to biases arising from the opinions of the researchers as to what should be regarded as the most meaningful structure for the data. The following validation techniques should minimize the bias (ibid.):
replicating the classification procedure using a separate dataset;
checking the accuracy of the classification using discriminant analysis by the proportion of cases correctly classified, which also confirms indirectly the degree of group separation; and
checking the stability (internal consistency) of the classification after repeated trials, preferably using another data set from the same sample population.
The results thus obtained are satisfactory as long as they meet two principal aims of numerical taxonomy (Sneath and Sokal, 1973, p 11):
repeatability and comparability within an acceptable level of error; and
objectivity and a degree of unbiasedness from personal feelings and prejudice.
6. Constituents of phenotypic characterization
A phenotypic characterization study will involve collecting a number of different kinds of data:
the breeds’ geographical distribution and if possible their population sizes and structures;
phenotypic characteristics of the breed populations, including physical features and appearance, economic traits (e.g. growth, reproduction and product yield/quality) and some measures (e.g. range) of variation in these traits – the focus is generally on the productive and adaptive attributes of the breeds;
12 CGRFA/WG-AnGR-6/10/Inf.6
images of typical adult males and females, as well as herds or flocks in their typical production environments;
origin and development of the breeds;
any known functional and genetic relationships with other breeds within or outside the country; biophysical and management environment in which the breeds are maintained;
responses of the breeds to environmental stressors, such as disease and parasite challenge, extremes of climate and poor feed quality and any other special characteristics of the population in terms of adaptation; and
relevant indigenous knowledge (including but not limited to gender specific knowledge) of traditional management strategies used by communities to utilize the genetic diversity in their livestock. While most of these data elements can be collected directly during field work, valuable information may also be obtained from secondary sources in the published and unpublished literature (including electronic datasets related to aspects of the production environment).
Most of the variables listed above can be obtained through primary characterization studies (single visits to field sites); others require advanced characterization studies (repeated measurements and observations). The latter group includes variables that describe economic performance traits (e.g. growth, milk production, egg production, wool production), adaptation (levels of resistance and tolerance to stressors) and trends (e.g. in population size and structure, and phenotypic performance).
The main focus in the guidelines is on primary characterization studies, but possible follow-up steps including advanced characterization studies are also described.
Describing breeds in terms of their qualitative and quantitative traits
Qualitative traits
This category of traits covers the external physical form, shape, colour and appearance of animals.
These traits are recorded as discrete or categorical variables. Their discrete expression relates to the fact that they are determined by a small set of genes. Relative to the quantitative traits discussed below, some of these traits (e.g. colour of hair coat, feather type, horn shape and ear length) may have less direct relevance to the production and service functions of AnGR. However, they may relate to some of their adaptive attributes. For instance, colour of the skin and hair coat, and size of ears and horns, are known to be relevant to the dissipation of excess body heat. Length of tail or size of switch in cattle is important in areas where there is a heavy burden of biting flies. Other traits may relate to the preferences or tastes of livestock keepers and consumers (e.g. colour of hair coat), and some are used for animal identification in situations where permanent identification of individual animals is otherwise impractical. In this context, they are as important as the quantitative traits and hence they need to be included in phenotypic characterization studies.
Qualitative traits are recorded either as discrete categories of expression (such as colour of hair or feathers) or binary variables (e.g. presence or absence of wattles). Collection, management and analysis of data on qualitative traits are therefore different from the equivalent procedures for quantitative traits. Details of these methods are discussed in section C (Data collection for primary characterization) and section E (Data management and analysis) of these guidelines.
Animal temperament is closely tied to various production and service functions of livestock.
Temperament is recorded as a subjective measure (either categorical or binary) preferably at herd or flock level. Some breeds (e.g. the Fulani cattle of the Sahel of western-central Africa) have typical features of temperament and attachment to their owners that distinguish them from other populations.
CGRFA/WG-AnGR-6/10/Inf.5 13
The commonest qualitative traits used in phenotypic characterization of cattle, sheep, goats, chicken and pigs, along with a suitable set of codes for recording them, are presented in Annexes 1 to 4.
Recording traits such as colour of hair, feathers or shanks, or size of hump involves some level of subjectivity. Steps need to be taken to develop a common understanding of these traits among those collecting such data. Enumerators should be given uniform training on these aspects of data collection. Standardized colour charts can be prepared and taken to the field.
Standardization of the coding of qualitative traits is also essential for broad utility of the data, for instance to compare breeds within or between countries. Meta analysis at regional and global levels requires both standardization breed descriptor data and access to the relevant datasets. It is therefore important that National Coordinators for the Management of AnGR enter data on the characteristics of their countries’ breeds consistently and as fully as possible into DAD-IS. It is also important that phenotypic characterization studies provide the data that National Coordinators need to complete the task. It is recommended that phenotypic characterization studies should aim to collect this core set of data items – listed in the Annexes to these guidelines – as fully as possible, both for international reporting and as a sound basis for national actions to improve AnGR management. The set of data items can be expanded as necessary to address specific objectives and preferences at national or local levels. In Spanish- and French-speaking countries, some phenotypic characterization studies present qualitative traits in three categories – morphological, morphostructural and cutaneous (faneropticos)
– but essentially the same set of qualitative traits as those described above is being discussed.
Quantitative traits
Quantitative traits are measures of the size and dimension of animals’ bodies or body parts and are more directly correlated to production traits than qualitative traits are. For instance, body weight and chest girth are directly related to body size and associated production traits. Typically, these variables have a continuous expression. This is because of the numerous genes that determine or influence their expression. While qualitative traits, such as coat colour, are based on a small number of loci and can be precisely recorded and predicted for defined animal populations, the economically important quantitative traits require considerable recording of direct and indirect indicators of the trait in individual animals. Furthermore, unlike many qualitative traits, most quantitative traits are dependent on the age of the animal and the type of production environment in which they are kept.
Consequently, it is imperative to sample only fully adult animals maintained in their typical production environments. Furthermore, the data collected in a single visit can only provide indicative information on the traits measured. Repeated and more structured data collection is required for systematic characterization of economic traits (see Section D for further discussion).
Because of their strong correlations with production traits such as meat and milk production, traits such as body weight, body length and height at withers are used as proxy indicators of the production traits. Body measurements should always be accompanied by explanatory notes on the plane of nutrition, or season of the year and how this affects the availability of feed. In studies that cover large geographical areas and involve the characterization of grazing animals, the objective should be to collect all field data during seasons of the year when feed supplies are similar. Alternatively, body condition scores of sample animals can be collected and used to account for seasonal differences in the plane of nutrition, but this approach requires that the data collectors have the relevant skills.
Traits such as dewlap width, ear length, height at withers and size of preputial sheath are directly related to adaptive attributes of AnGR and are therefore relevant in their phenotypic characterization.
For instance, AnGR that are well adapted to dry and hot climates such as the Jamunapari goat of
India or the Boran cattle of Ethiopia and Kenya, typically have very long ears and a wide dewlap.
Economically important production traits, such as growth rate, milk yield, egg production and fibre
(e.g. wool, cashmere) yield cannot be adequately assessed by single visit to field sites. They require repeated measurements of performance, which need to be investigated through advanced phenotypic
14 CGRFA/WG-AnGR-6/10/Inf.6 characterization work (discussed in more detail in Section D). However, some indicative data on average performance levels can be collected through direct measurement, interviews with livestock keepers or from available records.
Live body weight at a specific age, combined with available knowledge of meat quality and marketability, can be used as a proxy indicator of suitability for meat production in all the species discussed in these guidelines (cattle, sheep, goats, pigs and chickens). Similarly, average milk offtake records from sample animals on the day of data collection, taking into account the stage of lactation, can indicate milk production capacity in cattle, sheep and goats. Formats for capturing such data are presented in Annexes 1 to 4 for the respective species. A more detailed example how such data can be obtained is presented in Box 2.
Box 2: A rapid method of assessing milk production in cattle breeds
As part of a comparative evaluation of the utility value, as perceived by their owners, of four indigenous cattle breeds in southwestern Ethiopia under smallholder management, a semi-structured questionnaire was used to interview 60 cattle-keeping farmers from the home areas of the four breeds,
Abigar, Gurage, Horro and Sheko. The questionnaire covered, among other aspects of production, reproduction characteristics, breeding practices and milk production. The daily milk off-take was estimated by each farmer for the three trimesters of the lactation period, both for the oldest cow and a cow chosen randomly from among the herd. Off-take did not include the amount of milk suckled by calves. Milk production was estimated as an average quantity per day in each trimester of the lactation. Based on these figures and the reported lactation length, the total lactation yield was calculated. Lactation length was longest in Sheko cows and shortest in Gurage and Horro. Milk production was significantly higher for Abigar and Sheko compared to Gurage and Horro. The lowest milk production was reported in the Gurage breed.
Source: Stein et al. (2009).
If specialized production traits such as wool, cashmere or mohair are considered a priority, direct measurements of fibre quality (e.g. percent wool and hair), length, strength and curliness may be taken during primary characterization studies. Whenever these are necessary, however, detailed data collection through advanced characterization studies (on-farm and on-research station) should be planned. Blood samples can be collected during field work and used for assessing blood parameters, such as haematocrit count or prevalence of blood parasites, or for extracting DNA for molecular genetic analysis. Taking such samples needs careful planning and coordination with laboratories that can do the required analysis. Detailed discussion of molecular genetic characterization can be found in the complementary guideline publication devoted to this topic (FAO, 2010b) which is based on a tested set of recommendations for field and laboratory work (FAO, 2010b). From the perspective of organizing a phenotypic study, the main point to note is that the fieldwork phase of the study is an opportunity to collect blood or tissue samples. Importantly, coordinated approaches allow combined analysis and comparison of phenotypic and genetic data to provide a more comprehensive analysis of
AnGR diversity. Such analysis not only facilitates a more definitive identification of distinct breed types when phenotypic differences appear minor (see Box 3) but can also be used for establishing genetic relationships between breeds, which is very useful for planning breed improvement and conservation programmes.
CGRFA/WG-AnGR-6/10/Inf.5 15
Box 3: How to complement genetic characterization with phenotypic characterization
An integrated phenotypic and genetic characterization study of three meat-type goat breeds developed in South Africa (Boer, Savanna and Kalahari Red) was conducted in 2007 to determine whether the typical characteristics of the breeds were being maintained and ensure that their unique traits were not being lost. The same populations were sampled for phenotypic and genetic characterization. A set of twelve linear measurements were taken for the phenotypic characterization. Eighteen microsatellite markers selected from a panel of markers recommended by the International Society for Animal Genetics (ISAG) and FAO were used in the genetic characterization. The results showed that morphometric variation within the breeds is greater than between the breeds and that morphometric differences between the breeds were fairly insignificant. This highlighted the need for genetic characterization to enable the breeds to be distinguished accurately at genotypic level. Results of the genetic studies showed that the three breeds had relatively high heterozygosity values and that each of the populations was clearly distinguishable as a separate breed based on genotyping results with the selected markers. It was concluded that further genetic studies were needed to ensure sufficient diversity within the breeds for long-term conservation of the unique genetic resources.
Source: Pieters et al. (2009)
Additional data on known resistance to, or tolerance of, biotic (diseases, parasites, etc.) and nonbiotic
(climate, water scarcity, seasonal feed scarcity, etc.) stressors can be collected during phenotypic characterization studies by interviewing individual livestock keepers or through focusgroup discussions. Such data are largely dependent on perceptions of the interviewees and hence need to be interpreted with caution. Closer investigation through repeated measurements may be necessary.
Some guiding questions for capturing information on such traits are presented in Annexes 1 to 5.
The traction services provided by cattle are important for many rural populations in Africa and Asia, and hence need to be considered as part of phenotypic characterization in this species. During primary characterization studies, it is only possible to collect data on trait preferences. If necessary, advanced studies to obtain detailed data on speed and work performed can be implemented.
Limitations of primary characterization for collecting data on traits of economic importance
Despite the high cost and huge effort involved in primary characterization studies, very little can be deduced about important production traits such as growth rate (for meat production), lactation milk yield, egg production, wool production or the quality of these products. Some data collection instruments that can be used to capture indicative information on these traits during single field visits are available, but these are not substitutes for advanced characterization based on repeated visits and controlled measurements (see Section D). Resource limitations may mean that it is necessary to choose between covering a large area through a primary characterization or to conduct advanced characterization in a smaller sample or geographical area.
Investigating feral and wild populations
In some locations and production systems, livestock come into contact and interbreed with wild or feral populations. For example, in the mountainous regions of northern Viet Nam, domesticated chicken populations are in frequent contact with their wild relatives. Similarly, numerous native pig populations in isolated rural communities in Papua New Guinea are known to interbreed freely with feral and wild pig populations. Whenever possible, consideration should be given to collecting some data on these resources during phenotypic characterization studies in such locations. Of particular relevance are estimates of the geographical distributions and population sizes of the wild and feral populations and information on whether, and to what extent, there is interbreeding between these populations and domestic animals. Apart from genetic introgression, feral and wild populations can be important in the transmission of contagious diseases to domestic populations. The data collected may also be important from the perspective of managing the wild or feral populations themselves,
16 CGRFA/WG-AnGR-6/10/Inf.6 either to help conserve them as important elements of local biodiversity or, if they are so-called invasive alien species in the local context, to reduce the problems they cause.
Investigating breed population sizes and threats to AnGR
Up-to-date data on the size and structure of breed populations are essential for effective management of AnGR. The task of obtaining a baseline of population and other data on a country’s breeds and subsequent monitoring of these populations is best handled through the development and implementation a national surveying and monitoring strategy, which is likely to involve sample-based
“household” surveys combined with the use of other data gathering tools (for further details, see
FAO, 2010a). In countries where AnGR populations are not well characterized and particularly where they are not distinguished into recognized breeds, phenotypic characterization will be fundamental to the accumulation of a baseline of data on national AnGR.
Many individual phenotypic characterization studies will be too small in scale to allow them to provide precise figures for the population sizes of the breeds covered, particularly if the breeds are widely distributed throughout the country. Nonetheless, such studies represent an opportunity to obtain approximations of the size of breed populations in the study areas. For example, rapid appraisal techniques can be used to collect local knowledge on breed identity and the local distribution of these populations. By mapping these distributions and relating them to available population figures for the relevant species in the relevant administrative units (e.g. from a livestock census) it may be possible to obtain estimates of the sizes of breed populations within these areas
(e.g. FARM Africa and ILRI, 1996; Blench, 1999). Additional information from focus-group discussions and key informant interviews or from additional secondary sources such as reports of previous livestock studies may be useful in further refining crude population estimates.
Consideration should be given to collecting indicative data on these threats to AnGR during phenotypic characterization studies as part of the description of breeds’ production environments.
Interviews and group discussions with livestock keepers and other informants can be used to obtain information on threats related to socio-economic changes, availability of resources or disease epidemics and other disasters. Mapping breed distributions as part of phenotypic characterization studies (see below) will also contribute to the analysis and management of some threats.
Mapping breeds’ geographical distributions
Data on the geographical distribution of livestock breeds are important to the development of AnGR management plans both directly (e.g. knowledge of the location of the animals may be necessary to plan responses to emergencies such as disease outbreaks) and indirectly (because of the link between location and the “natural” aspects of the production environment – climate, elevation, terrain, disease epidemiology, etc.). Phenotypic characterization studies should always record the locations where measurements are taken, and map as accurately as possible the distribution of breeds within the areas covered by the study.
Breed distribution maps can be sketched based on GPS readings taken at study sites combined with information obtained via interviews or mapping exercises conducted with local people. In extensive livestock systems such as those of the pastoral and agropastoral livestock-keeping communities of sub-Saharan Africa, the Andes and parts of Asia, breed identities often match the ethnic boundaries of livestock-keeping communities. Such links can be corroborated using information gathered via focus group discussions and interviews with key informants. Relevant secondary data may also be used to sketch distribution maps, but caution is needed in developing and interpreting data from secondary sources as they may be incorrect or out of date.
Describing production environments
To understand the production and adaptation attributes of livestock breeds or populations, it is essential to describe their production environments. There are several reasons why this is important.
CGRFA/WG-AnGR-6/10/Inf.5 17
If data on production levels are being collected, it is essential that data are also collected on the conditions in which the animals are kept. Without production environment data, performance data are meaningless. Not only do variations in production environments give rise to variations in performance, breeds may be ranked differently in different production environments; i.e. a breed that is the top performer in one production environment may be a poor choice elsewhere. Adaptation traits are complex and difficult to measure, especially in low to medium input production environments.
However, they can be characterized indirectly by describing the production environments in which the populations in question have been maintained over time. Breeds that have had to survive and reproduce in the presence of particular stressors and combinations of stressors (e.g. high or low temperatures, poor-quality feed, specific diseases or parasites) will have been under selective pressure to develop adaptations to these stressors.
Describing the production environment may also be important as a means of identifying potential development opportunities. For instance, the fact that breeds are kept in specific natural environments may be important in the development of niche markets for their products. Descriptions of breeds’ production environments are also essential for the planning of genetic improvement and conservation programmes. Here in particular, it is necessary not only to describe the physical conditions in which the animals are kept but also to describe features of the socio-economic environment, such as the uses and roles of livestock, market orientation and access, specific products and marketing opportunities, and gender-related aspects of livestock keeping.
Meaningful comparisons among breeds require that their respective production environments be described in a standardized way. To address this requirement, FAO and the World Association for
Animal Production convened an expert workshop, held in 2008, that developed a standard set of production environment descriptors (PEDS) for use in DAD-IS and in phenotypic characterization studies (FAO/WAAP, 2008). Individual phenotypic characterization studies should treat this set of
PEDS as a minimum and collect whatever additional production-environment data is relevant to the objectives of the study and to providing a comprehensive description of the conditions in which the animals are kept.
The PEDs framework is presented in Annex 5. Note that the framework includes average climatic data that cannot be obtained in a single visit to a study site (in fact they require several decades of observations). Such data might be obtained from the records of weather stations situated close to the study site. Moreover, many aspects of the production environment are now recorded electronically in high-resolution maps. If a phenotypic characterization study records the geographical locations of the targeted breeds, it becomes possible to create digitized breed-distribution maps that can be overlaid with any other digitized maps that are available for the respective areas. This approach is being used in the PEDs module of DAD-IS for all aspects of the production environment for which digitized maps are available globally. These global maps include not only climatic data, such as temperature, rainfall and relative humidity, but also aspects of terrain and vegetation such as elevation, slope, landcover type, tree cover and soil pH. Data on aspects of the production environment that are not mapped (e.g. management practices) have to be collected directly during field visits. See Sections C and D for further discussion on collecting PEDs data.
Economic valuation of non-production traits
Phenotypic characterization studies may pave the way for genetic improvement or conservation programmes. In the low external input production environments of developing countries, the reasons for raising particular types of livestock include a range of adaptation traits and non-marketable service functions. In stressful environments, tolerance of feed and water scarcity, disease and parasite burden, occasional drought and extremes of temperature may be prioritized over production traits.
Similarly, mothering ability, fertility, capacity to provide traction services or to meet sociocultural roles may be priority traits in some production systems. Unfortunately, these traits are difficult to record during phenotypic characterization studies. Recent advances in the field of economic valuation of AnGR have developed, adapted and tested new data-collection and analysis tools for assessing
18 CGRFA/WG-AnGR-6/10/Inf.6 such traits in ways that can inform genetic improvement and conservation plans (Drucker et al., 2001;
Drucker et al., 2004). These tools can be applied during phenotypic characterization studies. Drucker et al. (2001) provide a critical evaluation of tools for economic valuation of AnGR. Two basic tools that can be used in phenotypic characterization studies of breeds and individual animals are:
1) determining the economic importance of the breed under consideration by asking key stakeholders specific questions about breed preferences (i.e. relative importance of the breeds taking into account all relevant economic traits); and
2) identifying all the relevant traits and putting them in priority order based on livestock keepers’ trait preferences.
If breeds are being considered for inclusion in genetic improvement or conservation programmes, additional studies that collect detailed data on input and output levels in their management may be necessary. The greater significance of non-production traits in the low external input production environments of developing countries means that in these environments it may be particularly important to develop productivity evaluation criteria that take these traits into account and to apply them in assessing and comparing the merits of different AnGR (Ayalew et al., 2003; see Box 4). It also means that nonincome functions (e.g. manure, savings, insurance) may need to be included in genetic improvement programmes in such production systems.
Box 4. Aggregated productivity model for comparative performance evaluation of AnGR
The multiplicity of important production, service and sociocultural functions performed by livestock in smallholder and subsistence production systems cannot be captured by conventional productivity evaluation criteria that focus on production traits. Evaluations based on such criteria are inadequate for evaluating subsistence livestock production because: 1) they fail to capture non-marketable benefits; and 2) the core concept of a single limiting input is inappropriate to subsistence production, as multiple limiting inputs (livestock, labour, land) are involved in the production process. As many of the livestock functions as possible (physical and socio-economic) should thus be aggregated into monetary values and related to the resources used, irrespective of whether the outputs are marketed, home-consumed or maintained for later use. A broad evaluation model involving three complementary flock-level productivity indices was developed and used to evaluate subsistence goat production in the eastern Ethiopian highlands. The results showed that indigenous goat flocks generated significantly higher net benefits under improved than under traditional management, which challenges the prevailing notion that indigenous livestock do not adequately respond to improvements in the level of management. Furthermore, the study showed that under the subsistence mode of production considered, the premise that indigenous × exotic cross-bred goats are more productive and beneficial than the indigenous goats is wrong. The model thus provides a more realistic platform upon which to propose improvement interventions.
Source: Ayalew et al. (2003)
Another important reason for economic valuation of adaptation, service and other non-production traits is the potential public or social functions of AnGR. As often observed in breeds that are at risk of extinction, these roles attract little market interest. Unique traits such as resistance or tolerance to endemic diseases or parasites, or to seasonal feed and water scarcity need to be identified and valued in economic terms through follow-up studies (Drucker et al., 2001).
A common feature of many methods for economic valuation of non-production traits is documentation of the trait preferences of livestock keepers and valuing them in monetary terms.
Indeed, livestock keepers can be asked to state their breed preferences and the specific reasons underlying these preferences whenever multiple breeds are under consideration. Such data can be collected during primary phenotypic characterization studies (see Annex 5, Part V). Analysis of these data may raise more specific economic questions that need investigation through follow-up studies.
CGRFA/WG-AnGR-6/10/Inf.5 19
Economic valuation studies conducted in conjunction with phenotypic characterization studies can provide useful estimates of the values that society places on particular AnGR. Information on livestock keepers’ preferences and perceptions about breeds and their traits is critically important in the design of genetic improvement and conservation programmes. Specific technical input from competent experts should be brought in to assist with the planning and management of economic valuation studies associated with phenotypic characterization work.
SECTION B: OPERATIONAL FRAMEWORK
This section marks the transition between the previous conceptual section which describes the principles, concepts and elements of phenotypic characterization and the following sections which deal with the practical planning and execution of a phenotypic characterization studies.
This section represents the “getting started” phase of such a study. It provides advice on forming the team who will take charge of the study, and defining its activities. It is a critical section in the sense that it guides the study team in defining the objectives and the scope of the study.
1. Establish an inventory of stakeholders
Ideally, the study will be initiated within the framework of the country’s National Strategy and
Action Plan for AnGR (FAO, 2009) and be part of a national strategy on surveying and monitoring strategy that aims meet the countries needs for AnGR-related data and information. In other cases the initiative may come from individuals or group of persons, who are aware of the need to characterize particular local AnGR populations. Whatever the institutional framework, it is important that phenotypic characterization studies are focused on meeting the priority information needs of AnGR stakeholders locally and/or nationally. If the study is not initiated at national level, it is nonetheless essential that the National Coordinator for the Management of AnGR and National Advisory
Committee on AnGR (or equivalent structure) be informed and consulted.
Once a decision to conduct a study has been taken, it is essential that a balanced and competent team be assembled to plan and implement the study. This task will have to be organized by the individual or small group who are initiating the study or who have been delegated the task by the national authorities. The process may involve a series of consultations with key stakeholders. A first step may be to draw up an inventory of all stakeholders that may contribute to the planning and/or implementation of characterization study, be interested in the results or contribute to follow up activities. Representatives of key institutions and stakeholder groups should be identified. Table 1 provides a checklist that may help identify relevant stakeholders