Abstract
Inbreeding depression, i.e., the reduction of health and vigour in individuals with high inbreeding coefficients, is expected to increase with environmental, social, or physiological stress. Differences in the strength of sexual selection are therefore predicted to usually lead to higher inbreeding depression in males than in females. However, sex-specific differences in life history may reverse that pattern during certain developmental stages. In salmonids, for example, female juveniles start developing their gonads earlier than males who instead grow faster during that time. We tested whether the sexes are differently affected by inbreeding during that time. To study the effects of inbreeding coefficients that may be typical for natural populations of brown trout (Salmo trutta), and also to control for potentially confounding maternal or paternal effects, we sampled males and females from the wild, used their gametes in a block-wise breeding design to produce 60 full-sib families, released the offspring as yolk-sac larvae into the wild, caught them back 6 months later, identified their genetic sex, and used microsatellites to assign them to their parents. We calculated the average inbreeding coefficient per family based on a panel of >1 million SNPs. Juvenile growth could be predicted from these inbreeding coefficients and the genetic sex: Females grew slower with increasing inbreeding coefficient, while no such link could be found in males. This sex-specific inbreeding depression led to the overall pattern that females grew on average slower than males during the time of gonad formation.
Introduction
Inbreeding can lead to inbreeding depression, i.e., to a reduction in health and vigour, because of the expression of deleterious recessive alleles and a general reduction of heterozygote advantages (Charlesworth & Willis 2009). Males and females can be differently affected by inbreeding, for example because of sex-specific differences in the strength of sexual selection (Ebel & Phillips 2016; Vega-Trejo et al. 2022). A general prediction is that males suffer more from inbreeding than females because the strength of sexual selection is usually higher for males (Janicke et al. 2013; Noel et al.2019). Heterogamety has been discussed as a possible alternative explanation for sex-specific inbreeding depression, but its relevance is still unclear (Connallon et al. 2022; Vega-Trejo et al.2022). Little is known about other possible reasons for sex-specific effects of inbreeding such as differences in early life history (Vega-Trejo et al. 2022).
Most salmonid fish reach sexual maturity at the age of 2 or later, usually with no obvious sexual dimorphism before. However, the sexes differ in many aspects from very early stages on. Sex-specific gene expression could be observed in grayling embryos (Thymallus thymallus ) (Maitre et al. 2017; Selmoni et al. 2019) and around hatching in rainbow trout (Onchorhynchus mykiss ) (Guiguenet al. 2019). Embryos can even show sex-specific stress tolerance (Moran et al. 2016; Nusbaumer et al. 2021). Sex differences could also be found during the early juvenile stages, i.e., within the first few months when gonad formation starts. In grayling, genetic females start gonad formation earlier than males who instead grow faster during that time (Maitre et al. 2017). These sex differences peak around the first summer, possibly making female juveniles more susceptible to heat stress during summer and thereby explaining the observed link between climate warming and male-biased sex ratios among adults (Wedekind et al. 2013). Analogous sex differences in the timing of gonad formation could be observed in brown trout: females start gonad formation earlier than males, and captive-born males grow larger than captive-born females after their first months in the wild (Palejowski et al. 2022).
Inbreeding in wild populations is often a consequence of adaptive responses to local conditions, especially in salmonids (Wang et al. 2002). It has been shown to influence early life-history traits (Kincaid 1976; Naish et al. 2013), disease resistance (Arkushet al. 2002), and reproductive traits (Naish et al. 2013; Waters et al. 2020; Paul et al. 2021) in diverse salmonid species, but other studies did not find significant and consistent negative effects of inbreeding (Houde et al. 2011; Johnsonet al. 2015). Inbreeding effects in salmonid species can be influenced by environmental context (Gallardo & Neira 2005) or temporal and regional genomic effects (Paul et al. 2021). Not much is known about sex-specific effects of inbreeding in salmonids, but a recent meta-analysis on other taxa (mostly insects) highlighted the potential sex-specific effects of inbreeding and concluded that they may mostly be due to difference in the strength of sexual selection while heterogamety seem to play no significant role (Vega-Trejo et al.2022). The role of sex-specific life histories, however, remains unclear.
Here we focus on juvenile brown trout around a time when the sexes differ in gonad formation and the physiological stress that may be associated to this, i.e., around the end of their first summer. To study ecologically relevant inbreeding coefficients while experimentally controlling for potentially confounding maternal and paternal effects, we sampled adult males and females from the wild and use their gametes for in vitro fertilizations in full-factorial breeding blocks. The resulting 60 full-sib families differed in their mean inbreeding coefficients that were, however, mostly low as expected for the study population. We released the larvae into the wild and sampled them 6 months later. Here we (i) compare these captive bred juveniles that had been stocked into the wild with the wild-born of the same cohort, and (ii) test whether there are sex-specific effects of inbreeding on fitness-relevant traits.
Methods
Adult brown trout were caught from the Rotache stream (a rather pristine tributary of the Aare river, Marques da Cunha et al.2019) shortly before the spawning season. The eggs of 12 females were stripped and fertilized with milt of 10 males in two full-factorial breeding blocks (6 x 5 each) to produce in total 60 full-sib families as described in Wilkins et al. (2017). Fin clips were stored in 70 % ethanol at -20 °C.
After egg hardening and sampling eggs for parallel laboratory studies on embryo stress tolerance (Wilkins et al. 2017; Marques da Cunhaet al. 2018), a total of 1,925 eggs (mean±SD number per full-sib family = 32.1±14.8) were incubated under routine hatchery conditions at the cantonal Fischereistützpunkt Reutigen at constant temperature of 8.5°C and stocked into the Mühlibach streamlet (a tributary to the Rotache ; 46.804459°N, 7.690544°E) at a late yolk-sac stage in early March.
About 6 months after release into the wild (i.e., in late August), electrofishing was used along the Mühlibach streamlet to catch as many brown trout as possible (Ntotal = 518). The fish were narcoticized (0.075 g/L tricaine methanoesulfonate buffered with 0.15 g/L NaHCO3) and photographed on a weighting scale to later extract fork length and body weight. Fin clips were collected and stored in 70 % ethanol. After handling, all fish were returned to the wild.
Fin clips of adult breeders and a random subset of juveniles from the wild (N = 376) were used for microsatellite genotyping and genetic sexing. DNA was extracted using the BioSprint® 96 workstation following the manufacturer’s protocol (Qiagen GmbH, Hilden, Germany). DNA was quantified using a HS dsDNA assay on a Qubit® 2.0 Fluorometer (Life Technologies, Carlsbad, CA, USA) and concentrations of up to 20 ng/µL were sent toEcogenics GmbH (Balgach, Switzerland) for genotyping at 13 microsatellite loci and genetic sex determination using the protocol described in Palejowski et al. (2022). Parental assignment of the juveniles was based on the full-likelihood approach implemented in Colony v2.0.6.5 (Jones and Wang, 2010) with a threshold of 0.98.
For all but two breeders (one dam and one sire) high quality DNA extracts or tissue samples could be used for whole genome resequencing to calculate the kinship coefficients for each breeding pair. Samples were sent to the NGS Platform at the University of Bern (Switzerland) for library construction using the Illumina TruSeq DNA PCR-Free Library Prep Kit (Illumina Inc., San Diego, CA, USA) after mechanical shearing of the DNA. Electrophoresis bases size selection (150bp fragments) was used prior to library quantification, quality control and paired-end sequencing using a NovaSeq 6000 S4 flow cell (Illumina Inc., San Diego, CA, USA). Adult samples from the current study were combined with samples from a parallel study to achieve an estimated coverage of 15x. The quality of raw sequence reads was assessed using FastQC v0.11.9 (Andrews 2010). Trimmomatic v0.39 (Bolger et al. 2014) was subsequently used to remove adaptor sequences and remove low-quality reads (i.e. HEADCROP:6 LEADING:3 TRAILING:3 MINLEN:70 CROP:140). High-quality reads were aligned to the indexed reference genome of brown trout (Hansen et al. 2021) with BWA v0.7.17 (Li, 2013) and the obtained BAM files were further processed using Samtools v1.12 (Liet al. 2009) and Picard version 2.24.0 (“Picard Tools - By Broad Institute,” n.d.). BAM files were cleaned by soft-clipping beyond-end-of reference alignment and setting MAPQ to 0 for unmapped reads, alignments were sorted by leftmost coordinates, mate coordinates were filled and duplicated alignments were marked. The resulting clean, coordinate-sorted BAM files were indexed and ordered along the reference genome and variants were called using the HaplotypeCaller function of GATK v4.2.0.0 (McKenna et al. 2010). Variants were subsequently hard filtered according to GATK best practices recommendations (i.e. QD<2.0, QUAL<30, SOR>3.0, FS>60.0, MQ<40.0, MQRankSum<-12.5, ReadPosRankSum<-8.0) (Depristo et al., 2011; Van der Auwera & O’Connor, 2020). Further filtering was performed to remove indels and SNPs with a sequencing depth <10 and >30, a minor allele frequency <0.01, outside of Hardy-Weinberg equilibrium (p<10-8) and showing signs of strong linkage disequilibrium (r2 > 0.6). Only SNPs present in at least 90% of individuals were retained and the R package Hierfstat (Goudet 2005) was used to obtain marker-based estimates of kinship between all parental breeding pairs using a panel of 1,058,625 SNPs (Goudet et al. 2018).
Statistical analyses were done in JMP Pro17 and R 4.0.2 (R Development Core Team 2015). The Akaike information criterion (AIC) were used to describe the fit of different distribution models on juvenile sizes. Standard F-tests were used to compare means when visual examination of the distributions suggested similar variances. Welch’s F tests were used when this model assumption seemed violated. Likelihood ratio tests were used to compare frequencies. Linear mixed-effect models were used to evaluate the combined effects of sex and kinship on growth indicators. Non-parametric Spearman correlation coefficients rs were used to test for correlations between parental inbreeding coefficients, kinship coefficients, and family sex ratios.