Project Topics
Get Complete Project Material File(s) Now! »
Study populations
This study was performed over winter 2015/2016 using four wild arctic charr populations located in different lakes of the alpine or sub-alpine region (Figure 1, Table 1). The arctic charr is native to lakes Geneva and Constance, where the populations remained landlocked following the glaciers retreat around 13.000 years ago, now representing the Southernmost native range limit of the species in Europe. As of today, these populations are still stocked yearly through supportive breeding, i.e. with 0+ individuals that are produced and reared in hatcheries from local wild-caught spawners, and released in the wild after having reached the first-feeding stage (Caudron et al., 2014). In the past, both Geneva and Constance have further been stocked with occasional allochthonous fish input, but previous research showed these punctual introductions have had little to no impact on the genetic diversity (Brunner et al., 2001) or genetic structure of these lakes (see Savary et al. [2017] for detailed study of Lake Geneva).
In lake Pavin, the species has been initially introduced in 1859 using 8000 alevins from the Huningue hatchery (Alsace, France) (Desmolles, 2016, Machino 1991). This hatchery translocated wild fish originating from the Rhine basin (Switzerland, Wurtemberg, and Bavaria, see Coumes, 1862), followed by a second introduction event between 1900 and 1906 using 2500 eyed eggs originating from lake Geneva (Reynouard 1909, Dussard 1955, Desmolles, 2016, Machino 1991). The Pavin population has been managed by supportive breeding since its introduction with no further allochthonous introduction documented. In lake Allos, the arctic charr was introduced in 1922 with 2500 alevins originating from lake Geneva, with the second and last introduction event documented in 1924 (Machino and Rivier, 2002). The Allos population is self-sustaining and has never been subject to management practices since 1924.
The four lakes follow an elevation gradient, constituting as such naturally different thermal conditions (Figure 1B). On charr spawning grounds, the temperature from December to March averages 3-4°C in high elevation lakes Allos and Pavin that are ice-covered during that period. In lake Geneva, winter temperature on spawning grounds was known to be vary between 5.2°C and 7.1°C in the 90s’ (Rubin and Buttiker, 1992), but recent observations showed that it could be even higher (>8°C, Mari et al., [2016]). In Lake Constance, literature indicates temperatures of 5-6°C at depths corresponding to spawning grounds (80m deep, [Lieberherr and Wunderle, 2018]).
Sampling and crossings
At the beginning of spawning season (between mid-November and early December), spawners were sampled on spawning grounds using gill nets. For each population, individuals were crossed using 2 females x 3 males or 3 females x 4 males crossing designs (producing all possible sibgroup combinations), depending on the availability of mature individuals and female fecundity. We produced a total of 132 families (Table 1). All fish were measured, weighed, and pectoral fin clips were collected for genotyping (see below). We collected and dry-weighed 30 ova per female to calculate average egg size per family.
Rearing conditions
For the Geneva, Constance and Pavin populations, ova were fertilized in hatchery facilities near the lakes. Temperature was kept between 1 and 5° C during both fertilization and eggs transport to the INRA facilities in Thonon-les-Bains. For Allos population, spawners were transported for practical reasons to the INRA station where fertilization were performed. Following fertilization, eggs (n total = 8628) were distributed into 24-wells cell culture plates (Nunc MicroWellTM), using three replicates of 12 eggs per family per treatment (one family was split across three plates, with 12 individuals per plate).
Eggs were exposed in thermal chambers to a temperature of either 5 °C (4.996 +/- 0.049 °C), referred to as cold or control treatment, or 8.5 °C (8.493 +/- 0.038 °C), referred to as warm treatment. Temperature was recorded every 30 minutes using HOBO loggers (HOBO Onset, #UX100-001) in both chambers.
Life history traits measurements
Survival at hatching was estimated as the number of successfully hatched individuals among those that reached the eyed stage. Overall survival was estimated as the number of living individuals from eyed stage to the end of the experiment. Incubation period was calculated in accumulated degree days (ADD) as the number of days from fertilization to hatching x average daily temperature. Body size measurements were performed at two timepoints: (i) T1: At hatching, after being transferred into the recirculating water circuit so as to minimize the bias linked to strong differences in the length of incubation period among and within populations (aged between 468-488 ADD at cold, 512-564 ADD at warm) (ii) T2: at the end of the experiment, close to yolk resorption stage (740-752 ADD at cold, 862-866 ADD at warm). At both timepoints, all individuals were pooled per family and photographed (Nikon D5300, Nikon DX 18–105mm lens). Measurements were taken using IMAGEJ software (http://rsbweb.nih.gov/ ij/download.html). Length was measured as the total length (snout-tail length) in mm, and yolk sac volume was calculated after Kamler (2008) considering yolk sac shape as an ellipsoid, using the formula: YSV = pi/6 x L x H², with YSV: yolk sac volume, L: yolk sac length (longest axis), H: yolk sac height/width (shorter axis, perpendicular to L axis).
Because hatching was asynchronous but body size measurements of individuals of a same origin were taken at the same date, we standardized both length and yolk sac volume by the age at the respective timepoint (in ADD). Unbiased length at T1 was calculated as lengthT1/ age at hatching, and at T2 as LengthT2 / ageT2, while unbiased yolk sac volume at T1 was calculated as YSVT1/ age at hatching, and at T2 as YSVT2/ ageT2 (in ADD). For clarity, we will refer from now on to these two unbiased measurements as “length” and “yolk sac volume” only.
Neutral and adaptive population divergence
Levels of pairwise population differentiation based on neutral markers were moderate to high and revealed two pairs of weakly differentiated populations, consistently with the information available on introduction events in study populations (. Overall QST estimates revealed divergent selection on all traits at hatching except survival; in all other traits, QST did not significantly differ from FST, hence divergence for those traits can be ascribed to drift. Divergent selection as seen from QST-FST comparisons in salmonids has been well documented by previous investigators (Perry, Audet and Bernatchez, 2005; Jensen et al., 2008; Rogell et al., 2013; McKinney et al., 2014; Côte et al., 2016), but evidence of adaptive divergence among recently founded populations (e.g. less than 100 years for Allos) has been more rarely reported (Koskinen, Haugen and Primmer, 2002; Whitney, Hinch and Patterson, 2013). In contrast, pairwise QST showed a different pattern with evidence of uniform selection in many cases. Uniform selection for survival can be expected for such a trait highly related to fitness, which should be maximized in every habitat. In contrast, such a pattern for morphological traits like yolk volume at hatching and embryo length is more surprising but could be expected between similar environments like the Geneva and Constance lakes. Other studies have documented QST < FST results (Andersen et al., 2007; Chapuis et al., 2007; Lamy et al., 2011), but there might be a general bias toward a discreet reporting of such patterns compared to QST > FST (Lamy et al., 2012). Trait conservatism can be a consequence of uniform selection that always leads to QST lower than FST. However, QST < FST patterns may emerge even under diversifying selection in small populations with reduced gene flow, where the among-population variance due to drift can become larger than among-population variance due to differences in local optima (Lamy et al., 2012; Moore et al., 2017). In particular, the divergence among recently founded populations (Allos and Pavin) may be low due to a stronger role of drift relative to selection in such small populations. Accordingly, the fact that many overall and pairwise comparisons showed a pattern of QST not differing from FST further supports this hypothesis. Alternatively, canalization could be another explanation to QST < FST, as traits could share similar genetic constraints rather than identical selective optima (Flatt, 2005; Lamy et al., 2012).
Finally, it is important to consider that the different patterns we observed between overall and pairwise QST-FST comparisons, may be explained by statistical reasons. First, it has been acknowledged that it is more difficult to detect significant differences between QST and FST in pairwise comparisons due to a loss of statistical power (Gilbert and Whitlock, 2015). Second, the significance of QST can be estimated with several approaches that can reveal contrasted results (e.g. (Leinonen et al., 2013)), hence it may be interesting to compare QST estimates and their confidence intervals using several methods, e.g. bootstraping versus Bayesian approaches.
Within population diversity
Local adaptation also depends on the adaptive potential of populations, which relates to the amount of genetic variance, itself linked to trait heritability. We observed very low heritability values for survival at T1, and almost null heritability for overall survival. Such fitness related traits are expected to have a lower heritability than morphological traits because they are supposedly under stronger selective pressures that tend to reduce additive genetic variance (Merila and Sheldon, 1999). In addition, we observed lower heritability estimates in the warm treatment for most traits, which is consistent with conclusions from Charmantier and Garant (2005) who reported a trend for lower heritability values for traits measured under stressful conditions in wild populations.
Life-history traits
Survival was estimated as the per cent of hatched embryos from the total initial number of eggs. Incubation period was calculated as a development rate (Koops and Tallman, 2004) in accumulated degree days (ADD) from fertilization to hatching: ADD = mean daily temperature x number of days since fertilization. Each newly hatched fish was anaesthetized with 0.35 ml per litre eugenol (clove oil diluted 1 : 10 into 95% ethanol) and photographed (Nikon D5300 and Nikon DX 18–105 mm lens). Length and yolk sac volume (YSV) were measured using IMAGEJ (http://rsbweb.nih.gov/ij/download.html). Length was measured from the snout to the end of the tail (total length), and YSV was calculated as assuming the shape of an ellipsoid: YSV = pi/6 x L x H² (with L, major axis and H, minor axis [Kamler, 2008]).
Study system and populations
The arctic charr (Salvelinus alpinus L.) is the northernmost freshwater fish species on the planet, distributed across the Northern hemisphere up to 80°N (Klemetsen, 2013). The species has a more limited thermal tolerance range than other salmonids (Sæther, Siikavuopio, and Jobling, 2016; Baroudy and Elliott, 1994; Elliott and Elliott, 2010). Literature mentions that the optimal egg incubation occurs at 4-7°C (see Sæther, Siikavuopio, and Jobling 2016; Janhunen, Piironen and Peuhkuri, 2010), and egg mortality increases dramatically when temperature reaches 8°C (Jungwirth and Winkler, 1984). However, southernmost populations are likely to better resist warmer temperatures as functional spawning grounds are often found in warmer environments than their northern counterparts (7-8°C in lake Geneva, [Rubin & Buttiker 1992], over 8°C in lake Sainte-Croix, Figure 1B). Nevertheless, strong evidence of local adaptation is still lacking (Larsson et al., 2005). We collected spawning individuals in December 2016 and January 2017 from four geographically isolated lakes located around the Southern limit of the species distribution range in Europe (Figure 1A): lake Geneva (France/Switzerland) and lake Constance (Switzerland/Germany/Austria), where arctic charr is native, and lake Pavin (France) and lake Sainte Croix (France), where the species has been introduced in 1859 and 1992, respectively. These four populations inhabit contrasted thermal habitats (Figure 1B).
Table of contents :
1 1. Introduction and context
Biodiversity in a changing environment
Environmental changes in freshwater systems
Multiple stressor interactions
Evolutionary concepts: responses of populations to environmental variations .
Relationship between phenotype and environment
Local adaptation
Study model and research questions
Alpine populations of a cold-water specialist, the arctic charr Salvelinus alpinus in a warming context
Local threats: fine sediment impacts on salmonids and potential for interactions with temperature
Research aims
2 2. Temperature as a selective pressure – investigating local adaptation patterns in native and introduced populations
Foreword
Article information
Abstract
Introduction
Material and Methods
Experimental design
Life history traits measurements
Data analysis
Results
Thermal reaction norms
Neutral genetic diversity
Quantitative trait variation
Discussion
Thermal reaction norms
Neutral and adaptive population divergence
Within population diversity
Impacts of stocking
Conclusion
Supplementary files
3 3. Temperature as a constraint on tolerance to other stressors – the example of fine sediments Foreword
Article information
Abstract
Introduction
Material and methods
Study species
Experimental design
Sediments
Temperature
Life-history traits
Statistical analyses
Results
Discussion
Supplementary file
4 4. Response of wild populations in a multiple stressors context
Foreword
Article information
Abstract
Introduction
Material & Methods
Study system and populations
Rearing experiment
Results
Discussion
Separate effects of temperature and sediment
Interactive effects of temperature and sediment
Conservation implications for arctic charr populations at the Southern edge
Conclusions
Supplementary files
5 5. General discussion
Adaptation and plasticity in response to temperature and considerations on thermal tolerance
Response to sediment exposure and assessment of the sediment risk
Life history trade-offs in early life: growth, development, and maintenance
Implications of this work for Southernmost arctic charr populations
Multiple stressors affecting arctic charr and management of multiple stressors
Genetic structure of populations and influence of stocking practices
Concluding remarks
6 6. References
