Proceedings of the New Zealand Society of Animal Production 2016. Vol 76: 159-162 159 Introduction Increasing ewe longevity and reproductive output can improve whole-flock efficiency and reduce environmental impact (Jones et al. 2013). By increasing ewe longevity (the length of productive life), fewer replacements are required, reducing the number of unproductive animals and increasing the number of ewe lambs to sell (Dickerson & Glimp 1975). Longevity research often looks for genetic indicators of this trait; however, heritability is low, or occasionally moderate (Mekkawy et al. 2009). An alternative approach may be to alter how animals are retained for breeding the following year, allowing animals to reach their potential and not reducing longevity by selling animals that are still productive. To achieve this, it is essential to be able to identify ewes that will survive and be productive. Although most research on longevity concentrates on the whole lifetime of an animal, it may be informative to consider factors that affect just the following years’ performance, in order to aid retention decisions, particularly when genetic information is unavailable. On UK hill farms, flock managers often select ewes to keep or sell based on current appearance pre-mating. Ewe age is also often selected on with ewes typically being sold at 5.5 years of age. This aims to avoid both lower returns from selling older ewes and increased incidence of mouth/ teeth problems, resulting in ineffective grazing, but there is little scientific evidence for this concern (McGregor 2011). Ewes have been proven capable of having many more productive years after this age (Dickerson & Glimp 1975), so it should be possible to keep these productive animals longer and thereby improve flock longevity. Categories of traits for consideration that predict ewe survival and performance the following year include: visual assessment, previous performance and Estimated Breeding Values (EBVs). Although selection on EBVs can successfully improve production levels in hill systems in the UK (Conington et al. 2004), these can be challenging to obtain for hill breeds such as the Scottish Blackface. There are often a limited number of EBV-recorded rams for sale and conventionally ewes are mated in multi-sire cohorts and lambed on open hills, making parentage recording difficult. Ewe performance data (e.g. weights, litter sizes) could be collected relatively easily at routine handling events, if coupled with Radio Frequency Identification ear tags (now mandatory for UK BRIEF COMMUNICATION: Which traits best predict ewe performance and survival the following year on a UK hill farm? HM Wishart*, NR Lambe, C Morgan-Davies and A Waterhouse SRUC Hill and Mountain Research Centre, Kirkton Farm, Crianlarich, FK20 8RU, Scotland, UK. *Corresponding author. E-mail: Harriet.Wishart@sruc.ac.uk Keywords: sheep; hill systems; retention; selection; longevity sheep) and associated recording and weighing technology (Brown et al. 2015). However, in UK hill sheep systems, these handling events are limited (Morgan-Davies et al., 2012). Visual assessment pre-mating (for example: size, state of mouth, evidence of disease, etc.), which requires no knowledge of the ewes’ previous performance, is another alternative. This is a current approach for selecting animals to retain in the breeding flock in UK hill sheep systems. The aim of this study was to identify how these different categories of traits are associated with ewe performance the following year, in order to inform recommendations for ewe retention strategies. Materials and methods Scottish Blackface ewe performance data were collected at Scotland’s Rural College (SRUC)’s Hill and Mountain Research Centre, in the West Highlands of Scotland. All work involving animals was approved by SRUC’s Animal Ethics Committee. The research farm is run as a UK commercial hill farm; although the animals are recorded more frequently (monthly compared to 3 to 5 times per year). All data were collected or collated at the point of selection (October, pre-mating) over two years. 794 ewes were recorded, with some animals appearing in both years. Ages ranged from 2.5 to 7.5 years old at selection. Commercially available EBVs were generated by Signet Breeding Services. Data were grouped into three categories: 1) appearance pre-mating; 2) recorded performance, during early life and the previous year; and 3) EBVs (Table 1). Ewes were sold if they met predefined rules. These rules removed animals whose current condition had either already significantly reduced their welfare, or was likely to the following year, or who had repeated reproductive failures. Although it could lead to survival bias, for this study, animals sold were not included in the dataset for analysis. The appearance traits scored (listed in Table 1) include all traits that flock managers take into account when visually judging which animals to retain. Each ewe was scored by the same flock manager throughout the trial. The categories of traits were each compared against performance the following year, as defined by number of lambs (NL) and weight of lambs (WL) weaned (as measures of production) and ewe survival (ES). The best combination of traits, within each category, was selected