Individual quality: tautology or biological reality?

Heterogeneity among individuals is a central aspect of evolutionary ecology. Evolutionary changes of a population depend on selection pressures acting on individual’s phenotypic and genetic variation (Lynch & Walsh 1998). Individuals are rarely equally adapted to their environments, and this inter-individual heterogeneity in performance (e.g. survival and reproductive success) has been historically described with a variety of terms such as heterogeneity in fit (Darwin 1859), frailty (Vaupel, Manton & Stallard 1979), health and vigour (Hamilton & Zuk 1982) or organism’s state (McNamara & Houston 1996). Nowadays, a growing number of studies in animal behaviour, population biology and evolutionary ecology refer to inter-individual variations as differences in ‘quality’ (e.g. Birds: Sedinger et al. 2008; Mammals: Hamel et al. 2009; Insects: Cervo et al. 2008; Fishes: Wong, Candolin & Lindstrom 2007). However, since many studies have sought to measure ‘individual quality’ itself, there are many definitions of this concept. Recently, Wilson & Nussey (2010) proposed the following working definition of quality: ‘an axis of among-individual heterogeneity that is positively correlated with fitness’. This definition summarizes well the general acceptance that quality indices mostly relate an individual’s characteristics (phenotype or genotype) to fitness parameters. Yet, the choice of a quality index is always dictated by the study focus or species rather than by a general acceptance of what traits best describe quality. In fact, using different types of quality traits from a variety of conceptual frameworks may prevent researchers from adopting a single definition of quality. Here, we highlight four key distinctions between the views and uses of quality. We emphasize that relabelling other well-established terms (i.e. fitness) under the concept of quality may only bring further confusion. Instead, we suggest three recommendations to help authors being more explicit about their applications of this concept and to facilitate the comparison among studies using similar conceptual frameworks. A decade ago, Cam & Monnat (2000) pointed out to two prevailing views of individual quality (also described in Moyes et al. 2009). The first view corresponds to latent or static traits, expected to be constant throughout life, or at least during the period under investigation (e.g. adulthood), that represent underlying characteristics of an individual. Genetic traits are typical static traits that are known to affect survival or reproductive success (Keller & Waller 2002; Foerster et al. 2003) and are often referred to as ‘genetic quality’ (Hunt et al. 2004). Static traits can also be shaped by environmental conditions, such as birth weight (Stopher et al. 2008; Moyes et al. 2009), lifetime breeding success (Beauplet et al. 2006; Lewis et al. 2006; Lescroël et al. 2009) or longevity (Hamel et al. 2009) that are measured only once during an individual’s life. The second view corresponds to dynamic traits that can be repeatedly measured and may change during an individual’s life. Examples include condition-dependent traits such as feather colour, antler length or body mass (Dale 2000; Gaillard et al. 2000; Vanpéet al. 2007). Although not always mutually exclusive, these two views correspond to different assessment of an individual’s quality. The view of individual quality is also greatly dependent on the field of research where it is applied. On the one hand, studies often have to account for static heterogeneity in quality to unravel potential life-history trade-offs. On the other hand, there are studies seeking specifically for potential dynamic quality traits by looking at correlations between phenotypes and fitness. Demographic studies generally aim at understanding how inter-individual heterogeneity can affect the estimation of demographic parameters, such as survival and fertility (Metcalf & Pavard 2007), that are central to the evolution of life histories (Stearns 1992). For example, in studies investigating senescence patterns, age-specific survival rates could appear to increase towards the end of life, and this counterintuitive pattern could, at least partially, be explained by selective effects early in life on a potentially larger inter-individual heterogeneity (Vaupel, et al. 1979; Carey et al. 1992): age-specific survival could increase if low-quality individuals are the first to die (Curio 1983), thus masking potential effects of senescence on survival (Vaupel & Yashin 1985; McCleery et al. 2008). Similar problems can also be encountered in studies on the evolution of life histories where inter-individual heterogeneity may prevent the detection of trade-offs expected to occur at the individual level; ‘good quality’ animal can invest in reproduction without decreasing survival (van Noordwijk & De Jong 1986; Reznick, Nunney & Tessier 2000; Cam et al. 2002). In these studies, quality is presented as a special case of inter-individual heterogeneity when the differences between individuals are rather consistent throughout life (Moyes et al. 2009). Hence, individual quality is often performance-based and measured using life-history traits (i.e. longevity, contribution to population growth, age at first reproduction, lifetime reproductive success) of cohorts with complete lifetime data (Stopher et al. 2008; Hamel et al. 2009; Moyes et al. 2009). Wilson & Nussey’s (2010) definition of quality mainly refers to this context. In behavioural ecology, studies on mate choice and sexual selection generally consider condition-dependent secondary sexual characters as honest signals of male quality (Zahavi 1975). For example, ornaments under sexual selection may have evolved as indicators of a male’s quality and should either correlate with genetic benefits (e.g. good-gene hypothesis: Hamilton & Zuk 1982; sexy son hypothesis: Weatherhead & Robertson 1979) or advertise non-genetic benefits for female offspring fitness, such as territory quality (Dijkstra, Van der Zee & Groothuis 2008). The underlying prediction of all these hypotheses is that a female can assess the quality of potential mates based on their attributes relative to other males of the same population (Andersson 1994). In this case, quality is measured from condition-dependent phenotypic traits that female use to predict her future offspring’s performance (Galvan & Sanz 2008; Janicke et al. 2008). Consequently, several behavioural ecology studies have aimed at linking variation in an expected quality signal of males to some fitness components (Gustafsson 1986; Merilä & Sheldon 2000; Mainguy et al. 2009). Although Wilson & Nussey (2010) proposed a new working definition of quality based on a multivariate analysis of phenotypes and fitness, it does not alleviate the need to consider the effects of using different quality traits (e.g. static and dynamic), in different contexts (life history, sexual selection), on our understanding of quality. A set of dynamic traits measured on a given individual may not provide the same assessment of expected fitness values compared to a set of static traits (Moyes et al. 2009). This is especially true given that environmental variations alone can sometime drive selection gradients and predictions on trait – fitness correlations independently of quality (Knops, Koenig & Carmen 2007; Tuljapurkar, Steiner & Orzack 2009). Yet, there are still relatively few studies that consider both static and dynamic traits in their assessment of an individual’s quality (but see Hill et al. 1999; van Dongen & Mulder 2008; Roberts & Gosling 2003). So is it a problem that individual quality is being defined using different traits in different contexts? After all, it is conceivable that one could use quality when discussing life histories but assess it differently in the discussion of sexual selection. Yet, besides the difficulties related to the definition and measure of fitness itself (McGraw & Caswell 1996; Brommer, Merila & Kokko 2002; Coulson et al. 2006; Metcalf & Pavard 2007), we believe that the use of quality is especially problematic in cases where it overlaps with other well-established concepts. Confusion arises, for example, when researchers are measuring traits and are relabeling them as quality indices. In Figure 1, we summarize what we think are the most common contexts in which quality is generally defined: from genetic characteristics (e.g. genome-wide diversity) and phenotypic measurements to their links with fitness. Quality indices calculated on standard measurements of phenotype and genotype are largely similar to those used to perform selection analyses (Lande & Arnold 1983); selection is already a function of phenotypic characters and individual fitness components (Arnold & Wade 1984) and the opportunity for selection represents the among-individual variation in relative fitness in a population, independent of traits (Lande & Arnold 1983). Using quality in such context is thus of little value and restrain the possible comparisons among studies measuring selection in the wild but using different long-established terminologies. The worst-case scenario is of course witnessed in studies where fitness is relabelled as quality, which is a pure tautology because performance-based quality measures such as lifetime reproductive success will inevitably correlate with fitness. It then leads to an obvious circularity in arguments related to quality, like the truism that high-quality individuals are doing better. The classical link between genotype and environment affecting phenotype and its relationships with fitness. The concept of quality potentially overlaps with a series of well-defined concepts. The concept of quality may still stand apart from other terms (e.g. Fig. 1) when it is used to rank individuals along a continuum of combinations of particular quantitative traits and assesses how individuals with different quality scores contribute to the evolutionary trajectory of a population. We have also emphasized above that there are simple categories in which quality can fall into: static versus dynamics traits or controlling for performance using life-history traits versus seeking for phenotypic proxies of fitness. We now draw attention to three recommendations that could help reconcile the different uses of the quality concept. First, authors should explicitly provide the type of category in which they are considering quality. Second, authors should discuss the implications of using terms that offer different a priori expectations of fitness given environmental variability. Third, authors should justify why the use of the term quality was preferred over the well-recognized links between genotype, environment, phenotype and fitness (Fig. 1). We believe these recommendations should help circumscribe the biological relevance of the concept of quality and ease comparisons among studies. We thank T. Coulson, A. Wilson and an anonymous referee for their constructive comments on this manuscript. We also thank M. Festa-Bianchet, J.-M. Gaillard and D. Nussey for their discussions and comments on previous drafts of this manuscript. This work was funded by the Natural Sciences and Engineering Research Council (NSERC) of Canada via a studentship to PB and Discovery Grants to FP, DR and DG. FP is also supported by the Canada Research Chair in Evolutionary Demography and Conservation.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.