Genetic monitoring of trout movement after culvert remediation: family matters

Abstract

We contrasted various genetic analyses to evaluate their utility and constraints for detecting movement of cutthroat trout (Oncorhynchus clarkii) through restored culverts in different field settings: population-level metrics of genetic variability (heterozygosity and allelic richness); Bayesian clustering and assignment of individual genotypes from age 1+ fish; and a novel "sib-split" approach, where movement patterns are extracted from the spatial distribution of young-of-year (YOY) full-sibling groups inferred via pedigree reconstruction. Family structure greatly influenced population-level and individual clustering results in our small headwater populations, even though field sampling was implemented to avoid siblings. Sib-split, which uses family structure to detect movement, uncovered passage of YOY just weeks after emergence. When retrospectively applied to older individuals, it proved essential in interpreting clustering patterns and captured passage in several families of 1-and 2-year-olds. Where family structuring may negatively affect genetic analyses or, alternatively, be prominent enough to allow application of sib-split is difficult to predict a priori; we discuss benefits and limitations of all approaches under different ecological, spatial, and management scenarios.Résumé : Nous avons comparé différentes méthodes d'analyse génétique afin d'évaluer leur utilité et leurs limites pour ce qui est de détecter les déplacements de la truite fardée (Oncorhynchus clarkii) par des ponceaux remis en état dans différents contextes de terrain. Ces analyses comprennent des mesures de la variabilité génétique (hétérozygotie et richesse allélique) au niveau de la population, le regroupement bayésien et l'affectation de génotypes individuels de poissons de plus de 1 an, ainsi qu'une nouvelle approche dite de « sib-split », dans laquelle les motifs de déplacement sont extraits de la répartition spatiale de groupes de fratrie de jeunes de l'année (YOY) inférés à la lumière de la reconstitution de l'ascendance. La structure familiale influençait considérablement les résultats au niveau de la population et ceux du regroupement des individus dans nos petites populations d'amont, même si l'échantillonnage sur le terrain était conçu pour éviter la fratrie. La méthode sib-split, qui repose sur la structure familiale pour détecter les déplacements, a détecté le passage de YOY quelques semaines seulement après l'émergence. Appliquée de manière rétrospective à des individus plus âgés, elle s'est avérée essentielle dans l'interprétation des motifs de regroupement et a relevé les passages dans plusieurs familles d'individus de 1 an et de 2 ans. Il est difficile de prévoir les situations où la structure de la famille pourrait avoir un effet négatif sur les analyses génétiques ou être suffisamment forte pour permettre l'application de l'approche sib-split. Nous abordons les avantages et limites de toutes les approches dans différents scénarios écologiques, spatiaux et de gestion. [Traduit par la Rédaction]

Monitoring Demographic and Genetic Responses of a Threatened Inland Trout to Habitat Reconnection

“…Additionally, though family structure may bias interpretation of genetic analyses, information about siblings and their locations can also be used as an effective "direct" method of capturing movement (Hudy et al 2010). In our study, collection of full siblings in different streams would indicate movement between tributaries, i.e., based on the "sib-split" approach of Whiteley et al (2014; see also Neville and Peterson 2014). Accordingly, for fullsibling families with offspring found in different tributaries, we noted the occurrence of this sib-split as indicative of movement and retained one individual from each tributary in the final data set for further genetic analyses; for example, if three siblings were found in one stream and one in another, we removed only two of the three in the first stream to retain one sibling in each stream.…”

“…Compared with other types of habitat improvements, then, the removal of passage barriers may have the largest effect on population responses (Roni et al 2008) and may be especially helpful for a desert fish like LCT in improving resiliency to a dynamic environment and climate change. However, to date there exist few examples where responses to reconnecting habitat have been monitored for any inland trout species (but see Neville and Peterson 2014;Whiteley et al 2014;Chelgren and Dunham 2015), even though such information would provide helpful validation of the efficacy of barrier remediation to guide future efforts (Rolls et al 2013). Furthermore, as most connectivity projects are implemented and monitored only at local scales (stream reach scales), the broader ecological effects of restoration are still poorly understood Lake et al 2007;Rolls et al 2013).…”

“…Although they have widely been promoted as useful for monitoring population changes over time (e.g., Schwartz et al 2006;Luikart et al 2010;Saura and Faria 2011;Hoban et al 2014), their application to monitoring the responses to restoration activities may prove difficult. Response times are expected to be long (Raeymaekers et al 2009;Langduth et al 2010), and natural populations-especially those impacted by restoration-often inherently violate theoretical assumptions such as equilibrium population dynamics, which can complicate interpretation (Whitlock and McCauley 1999;Osborne et al 2012;Neville and Peterson 2014). Still, we cautiously evaluated two metrics of population genetic diversity as well as genetic differentiation between pairs of populations over time because these metrics are commonly used for monitoring and may provide information complementary to demographic data as well as useful baseline insight about the overall (and possibly changing) "genetic health" of populations (e.g., Dunham et al 1999).…”

“…Before performing these and further genetic analyses, however, we first evaluated family structure in our samples, as inclusion of siblings can bias both population-level and individual-based genetic metrics (Hansen et al 1997;Anderson and Dunham 2008;Whiteley et al 2013;Neville and Peterson 2014). We used the maximum likelihood estimation in the program COLONY (Jones and Wang 2010) to identify fullsibling families in each of our temporal collections; COLONY analyses were implemented using the full-likelihood a.…”

“…We did separate analyses for each collection year, combining individuals from all three streams in each temporal analysis (e.g., Beaver, Little Jack, and Coyote Creek samples from 2002 were all run together but separately from 2007 collections). Because COLONY may not accurately estimate very small full-sibling families (Wang and Santure 2009;Hudy et al 2010;Neville and Peterson 2014), we ignored information from defined "dyads" and made inferences only from fullsibling groups of three or more. Where full-sibling families of three or more were identified within a stream, all but one sibling were removed from the data set.…”