Male sexual signaling and expected effects of hatchery-induced sperm competition vary with water depth at which whitefish are caught
Data files
Jul 12, 2023 version files 17.55 KB
-
Data_Perroud_et_al.csv
-
README.md
Abstract
Salmonids like whitefish (Coregonus spp.) are often propagated in supportive breeding. Spawners are caught from their spawning locations, their gametes mixed, and the resulting offspring reared in a protected environment before being released into the wild. This procedure can affect sexual selection, for example, by enhancing the importance of sperm competition or by reducing the relevance of sexual signals. While it is often unclear how sperm competitiveness is affected by a male’s overall genetic quality, there is accumulating evidence that sexual signals reveal good genes and that mate choice based on such signals can increase offspring viability (Auld et al. 2019). Therefore, supportive breeding may affect the genetic variance and the mean genetic quality of next generations. We sampled whitefish from various locations along a depth gradient to test how male characteristics that are likely to affect sexual selection under natural conditions correlate with characteristics that affect hatchery-induced sperm competition. Whitefish are external fertilizers, and multi-male spawning and hence sperm competition is common under natural conditions. Mate choice is not sufficiently understood but could be based on breeding tubercles. These are small conical structures that grow on scales before the breeding sea- son and fall off shortly afterwards. The size of breeding tubercles varies much among males and has repeatedly been found to correlate positively with offspring viability (Wedekind et al. 2001; Keka ̈la ̈inen et al. 2010). Male dominance is typically depend- ent on body size (Auld et al. 2019) and could also be relevant in whitefish. Body size itself can reflect individual inbreeding coefficients (Su et al. 1996) and be an indicator of heritable genetic quality in small or structured populations (Neff and Pitcher 2008). In another fish with a somewhat comparable mating system, the size of breeding tubercles and male size was not correlated but could both be used to predict male reproductive success under close to natural conditions (Jacob et al. 2009). We study whitefish from Lake Hallwil (Switzerland). This lake has suffered so much from anthropogenic eutrophication that it is being artificially aerated since 1985. Three hatcheries around the lake are likely to have played a key role in maintaining the whitefish population, as concluded also from a recent mark–recapture experiment (Vonlanthen 2015). However, eutrophication combined with possible hybridization in hatcheries can have led to a speciation reversal (Vonlanthen et al. 2012) and may thereby have destroyed any genetic structure linked to water depth. Hatchery protocols now focus on maintaining over-all genetic variance by pooling milt of many males before adding the mix to eggs of multiple females. Milt volume varies among sires, for example, because males often lose milt when being pulled up from deep locations (Figure 1), an effect that likely depends on how much the swim bladder is inflated by the change in pressure. This variance in milt volume is likely to affect the genetic variance that, in combination with the average genetic quality, may then affect the long-term survival of a population. The extent to which hatchery protocols affect genetic quality can be estimated by the correlations between male quality indicators and traits that affect hatchery-induced sperm competition, that is, sperm number, velocity, and longevity (summarized here as “milt potency,” see also Supplementary Material). Many breeding protocols are likely to promote genetic quality if male attractiveness or dominance are positively correlated to milt potency. If there are no such correlations or negative ones because of life-history trade-offs, hatchery-induced sperm competition is likely to reduce the average genetic quality in future generations. We sampled fish from various depths and determined their age, size, breeding ornamentation, and milt potency (see methods in the Supplementary Material) to test whether and how different male characteristics affect reproductive success in supportive breeding in a heavily managed population.
Methods
Adult whitefish (Coregonus suidteri, n=104) were caught from Lake Hallwil (Switzerland; 47.2772° N, 8.2173° E) during their breeding season in January 2020 using gill nets set at 40, 25, 12, or 4 m depth along a line perpendicular to the bank (at decreasing distance to the shore, following the natural depth gradient). The nets were separated by distances of about 230, 50, and 40 m, respectively. At each depth, gill nets of 25, 27, 30, and 35 mm mesh size were used to ensure that adult whitefish of all size could be caught. Fish were immediately killed and kept in ice water until further processing. Standard body length (i.e. excluding the caudal fin) was determined, and scales were sampled above the lateral line from the caudal half of the body to later estimate fish age based on the number of annuli, a method offering same age estimates as methods based on otoliths or fin ray for fish younger than 5 years (Muir et al. 2008). Milt was collected by applying gentle bilateral abdominal pressure (see below). Fish were then scanned with a 3D optical scanner to determine mean breeding tubercles volume (see below).
For age estimation, scales were first cleaned using a cloth and 60% ethanol and then rinsed in water. Three scales by fish were mounted onto microscope slides and photographed under a 1.6x magnification. Fish were then aged based on the number of annuli, except for 1 female caught at 25 m depth for which we had no readable scales.
The milt of 22 males caught at 12 m and 4 m depths was stripped into large Petri dishes (145 × 20 mm, Greiner Bio-one, Frickenhausen, Germany; the discrepancy between this sample size and total males sample size (n=71) was due to organizational problems in the field. Care was taken to strip the milt drop by drop and to separate theses drop on the Petri dish, so that all drops of milt that were non-contaminated with urine or faeces (i.e. usually all except the last drop) could be collected and individually pooled, 20 µL was stored at a dilution ratio of 1:9 in Storfish (IMV Technologies, l’Aigle, France), an isotonic inactivating medium, and kept on ice. The remaining milt was centrifugated to collect seminal plasma. Because the weight of the seminal plasma turned out to be not correlated to the sperm concentration (r = 0.18, d.f. = 18, p = 0.46), it was used to calculate the “milt potency”, here defined as
milt potency = sperm concentration × weight of seminal plasma × average sperm velocity × maximum sperm longevity (1)
The milt stored in Storfish were transported to the laboratory where sperm velocity and concentration were analysed within 48 h with CASA using the Qualisperm software (AKYmed AG, Cheseaux-sur-Lausanne, Switzerland) as in Nusbaumer et al. (2019), except that milt traits could mostly be summarised as a mean over 4 trials rather than over 2 trials. Briefly, 20 µL of each sample were activated in standardized water (OECD, 1992) at a 1:500 dilution ratio and measured at 20x magnification under phase contrast, at 6.5°C, 20s post-activation. Sperm longevity was measured as the time by which no more sperm motion could be observed. Our final sample size consisted of 20 males because the within sample measurement repeatability was too low in 2 samples (i.e. the standard error of sperm velocity and/or concentration exceeded 50% of the mean measurement; in the accepted samples, the standard errors of sperm velocity and concentration were on average 9 and 18 % of the means, respectively).
The average size of breeding tubercles per fish was determined with a 3D optical scanner (VR-5000, Keyence, Itasca, IL, USA). Each fish was first carefully dried before applying a thin layer of baby powder (Millette Baby powder, Migros, Zurich, Switzerland) on the skin to make the otherwise transparent breeding tubercles detectable by the optical scanner. The fish was then placed on a slightly angled plate so that its lateral line was approximately perpendicular to the laser beams in order to facilitate correction of the fish curvature in later analysis. Scans were then analysed using Keyence analysis software 3.1.0.56, with curvature correction strength set at 15, reference plane set as continuous, and measurements extracted using the conplane method to obtain the volume of individual breeding tubercles. All scans were carefully checked for anything that could potentially be wrongly identified as a breeding tubercle by the software.
Only 3 males and 4 females could be caught at 40 m depth. This depth category was therefore excluded from all statistical analyses. Also, because age classes varied from 1+ to 4+, but 89.5% of all fish where either 2+ or 3+, the fish were categorized as “young” (1+ or 2+, n=51) and “older” (3+ or 4+, n=44). Fisher’s exact test were then used for frequency analyses. An ANOVA was used to test for differences in body length between the catches. Because no differences in body length was found among the catches, a Gaussian generalised multiple regression analysis (GLM) was then used on the mean breeding tubercles volume as response variable to test the effects of depth at catch (specified as factor), body length, and the interaction of these two potential predictors (after graphically verifying that the model assumptions were not significantly violated). Analogous models were used to test whether depth at catch, body length, mean breeding tubercles volume, or any interaction between these factors would explain variance in mean sperm velocity and milt potency. The AIC of full models was compared to the AIC of models lacking a variable or an interaction and models with the lowest AIC were retained as final models. Wilcoxon rank sum tests were used on milt potency and components of milt potency to test whether they would vary between samples. The same method was used to test whether milt potency or components would vary between age categories. Analyses were done in RStudio 4.0.2 (R Development Core Team 2015) and JMP14.0.0 (SAS Institute Inc., Cary, NC).
References cited
Muir AM, Sutton TM, Peeters PJ, Claramunt RM, Kinnunen RE, 2008. An evaluation of age estimation structures for lake whitefish in Lake Michigan: selecting an aging method based on precision and a decision analysis. N Am J Fish Manag 28:1928-1940.
Nusbaumer D, Marques da Cunha L, Wedekind C, 2019. Sperm cryopreservation reduces offspring growth. Proc R Soc B Biol Sci 286:20191644.
R Development Core Team. 2015. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing.