Project Topics
Get Complete Project Material File(s) Now! »
Brown trout as a biological model for sexual selection
Darwin (1871) often reflected on salmonid astonishing life histories, either because of their life cycles, or because of their intersexual differences in traits behavior or phenotypical traits. Salmonid fishes are indeed an appropriate system for studying evolution of sexual selection facing environmental conditions. First, they are renowned for their tendency to show a wide range of variable behaviours during reproduction and these behaviours can be now be measured in natural and or experimental environments (Esteve, 2005; Freychet, 2011; E Petersson, Järvi, Olsén, Mayer, & Hedenskog, 1999; Schroder, 1981). Second, in salmonids, the environment can vary spatially and temporally leading to a possible evolution of costs and benefits of reproductive strategies which are closely linked with biotic and abiotic pressures.
The genus Salmo is one of the most studied within the family of Salmonidae, along with Salvelinus and Oncorhynchus. The salmonid subfamily Salmoninae exhibits about 30 species well described in the literature (Klemetsen et al., 2003). In the present manuscript I will describe only Salmo trutta L. (brown trout), because I used it as a case study throughout my thesis project. Brown trout is indigenous to Europe, North Africa and western Asia (Klemetsen et al., 2003). Brown trout is present in many regions of Europe from north of Iceland, Scandinavia and Russia to South of the Mediterranean Sea. After many introductions, brown trout has now reached a world-wide distribution (Elliott, 1994) because of its impressive capacity to spread and colonize new areas with ecological variability (Lecomte, Beall, Chat, Davaine, & Gaudin, 2013). Salmo trutta is defined as an anadromous fish which can have two reproductive strategies: the migratory strategy and the resident strategy. In the former, juveniles migrate to the sea to maturate with a period of smoltification and come back to their birth river or a different river for spawning (respectively “homing” and “straying”), whereas residents trout perform both their development and reproduction period in river: the present manuscript will focus only on resident brown trout. Accordingly, river connectivity can affect dispersal in this species and environmental contrast varies greatly from upstream mountain torrents to lowland plain rivers. Thus, local conditions such as population density, ASR, OSR and phenotypic distribution may be strongly affected by these environmental contrasts.
In brown trout, females compete for spawning sites and spawn on gravel bars where they excavate a series of depressions called “nests” where they lay their eggs (Greeley, 1932). The availability of these spawning sites is structured by the variation of particle size. Particle size can notably condition oxygen availability in the redd (Acolas, 2008) and can provide a good protection for the eggs. To access females, a fierce competition between males occurs with a display of agonistic behaviours, such as chases, bites and lateral display (Keenleyside & Dupuis, 1988). Interactions between males are often hierarchized as a function of their reproductive status, i.e. dominant or peripheral (Blanchfield & Ridgway, 1999; Erik Petersson & Järvi, 2001). Larger males have been described as more advantaged in comparison with smaller males during contest competition in different species of salmonids (Fleming & Gross, 1994; Schroder, 1981). Females have been reported to exhibit preference for adiposis fin size (Petersson et al., 1999) and for relative individual body size (Labonne et al., 2009). As a result of strong preference and competition, sexual selection is expected to be relatively strong in brown trout, and recent analyses confirm this view, while also mentioning the role of environmental uncertainty in the maintenance of plasticity in sexual behaviours (Serbezov et al., 2010). Although this thesis will not focus on the genetic basis of traits involved in sexual selection, it is of interest to note that the salmonid genome underwent a polyploidy event some tens of millions years ago (Allendorf & Thorgaard, 1984; Hoegg, Brinkmann, Taylor, & Meyer, 2004), and that the current genome might be highly influenced by this event: former copies of genes may have evolved to code for different functions, whereas some others may still code for similar functions. Second, this polyploidy event de facto erased the sex chromosome. Recent research suggest that a Sex locus is now present in many salmonid species, but at various stage of degradation, and very little is currently known regarding the genes that might be physically linked to this locus (Yano et al., 2012, 2013).
Environmental change and brown trout reproduction
In addition to changes in land use, water use and river channelization that may affect the brown trout life cycle at various stages and levels, the effects of climate change since the late 19th century (IPCC 2013) also threatens river ecosystems. This is particularly theoretical models predict an increase of the rainfall perturbation in frequence and intensity (Dankers & Feyen, 2008; Milly, Dunne, & Vecchia, 2005; R. J. Stevenson & Sabater, 2010; Vitousek, 1994). Indeed the increase in the frequency of extreme rainfall events is expected to directly influence water discharge in rivers, thereby potentially affecting the suitability of reproduction habitats for brown trout. Stream flow is predicted to increase in the western areas of Europe (Stahl et al., 2010; Stahl, Tallaksen, Hannaford, & van Lanen, 2012) such as in the Pyrénées mountain range. Moreover an increase of water temperature in rivers is also predicted with an increase of air temperature (IPCC 2013) which can affect metabolic rate of individuals and therefore their allocation in biological activities such as reproduction (Charnov & Gillooly, 2004; Gillooly, Brown, West, Savage, & Charnov, 2001). An increase of water discharge may have direct consequences on resource availability especially in freshwater food webs (Perkins, Reiss, Yvon-Durocher, & Woodward, 2010). Therefore energy stores are affected which will in turn modify the allocation of energy to the different functions (e.g. reproduction, survival, maintenance…) and will ultimately modify condition survival. This would in turn shuffle the initial conditions at the onset of reproductive season, by changing density, ASR and OSR of populations. Increased stochasticity in river water flow could also impact the energetic budget of spawners during reproduction, by impacting directly the cost of competition or parental care. Droughts and large floods may also have direct impacts on habitat structure. They may impact significantly survival in redds which, by providing protection against predation for the embryonic stage, are at the center of this species’ reproductive system and life cycle. Because the adaptive value of behaviours associated to sexual selection mechanisms is modulated by offspring survival, the evolution of reproductive system in brown trout is probably linked to variations in selective pressures on offspring viability.
For all these reasons, it is logical in this thesis to investigate the evolution of populations and sexual selection in relationship with environmental change, and specifically with climate change.
Semi-natural conditions: the Lapitxuri spawning channel
Study of reproductive behavior in the wild is informative, although it is a challenge: it is usually performed on open populations with little (and costly) possibility to have access to both individual identification during mating behaviours and reproductive success over a whole reproductive season. Additionally, many factors can be confounded and hard to interpret in a context of changing environmental conditions that may prevent efficient monitoring. Alternatively, reproductive behavior of brown trout can be studied in controlled conditions (Petersson et al., 1999) which howaver oversimplify environmental influences. For example, individuals are generally constrained to a limited number of mates, which may modify intra-sexual competition and inter-sexual preference compared to natural conditions.
Here I targeted a specific objective: to be able to monitor reproductive activity and reproductive success of a whole group of individuals, without interfering with mate choice rules. To do so a first experiment A (fully described in the paragraph II.II.7.a) was conducted in 2010-2011 and constituted a first test of wild brown trout reproductive behaviour in the artificial channel of Lapitxuri. I inform the reader that this experiment (experiment A) was undertaken by the lab just before the beginning of my PhD (Freychet, 2011). Back in 2010, we had no evidence that this approach would succeed, but relying on the lab’s experience of reproductive behavior in wild populations, we had some precise ideas of what to expect in terms of behavioural patterns (Garcia-Vazquez et al., 2001; Labonne et al., 2009; Tentelier, Larrieu, Aymes, & Labonne, 2011). In the following paragraphs (from II.II.2 to II.II.6), I detail common methods and information for experiments A, B1 and B2. The differences between these experiments are explained in paragraph II.II.7.
The Lapitxuri channel is a derivation of the Lapitxuri stream, a tributary to the Nivelle River in south-western France (+43° 16′ 59″, -1° 28′ 54″, Fig. 1). It has already been used for many experiments focused on Atlantic salmon reproduction (A. Hendry & Beall, 2004). Because the experimental channel is a derivation from a natural river, food is readily available by drift from incoming water. The channel (total length = 130 m) consists of 13 communicating and linear sections, each measuring 10 meters long and 2.80 meters wide. Upstream and downstream exit from each section can be prevented with grids, and net traps can be placed downstream of each section to catch drifting individuals. The whole channel is covered by nets to prevent avian predation, as well as to protect from disturbance.
Several environmental features can be manipulated in the experimental channel, making it a quite flexible tool to test predictions about the effect of environment on fish reproduction. Riverbed can be modified by adding, removing and arranging different substratum size. Likewise, water depth can be managed at the scale of each section by placing planks of a chosen height at the downstream limit of the section, and at a finer scale by adding or removing substrate. Moreover, woody debris can be placed anywhere to build hiding places for fish. Hence, one can easily arrange favourable zones for spawning or resting. Water discharge can also be manipulated by controlling the quantity of water derived from the Lapitxuri stream. Thanks to both an outlet and a supply pipe plugged between the seventh and the eighth sections, water discharge can be manipulated independently (to some extent) in the upstream and downstream halves of the channel. Additionally, and importantly for studies on reproduction, the density and sex ratio of groups of fish can be manipulated in each section, since sections can be isolated from each other with grids.
Finally, the artificial channel provides advantages for monitoring reproductive behaviour and reproductive success. The net protecting the channel against predation also serve as a hiding fence for observers, and the power outlets along the bank allow plugging video cameras and spotlights to record behaviour on a long term. The precise estimation of reproductive success is facilitated by the possibility to collect virtually all juveniles at the end of the spawning period, either by electrofishing or in drift nets downstream each section.
How to recognize fish during reproduction
One of the key factors in such experiments is to be able to identify individuals, to relate their behaviour, traits (body size and colourness) and fitness at individual scale. Individuals were thus measured, weighed and photographed for recognition before and after reproduction. No tagging of any sort was used, so to avoid interference with either survival or behavior (trout is thought to use visual cues in both intra-sexual competition and inter-sexual preference (Petersson et al., 1999). Fish recognition was possible before and after reproduction from inter-individual phenotypic variation: the density and the position of both black and red spots vary consistently from one individual to another and do not change over the reproductive season (example in Fig. 3). This phenotypic consistency not only ensured individual recognition on pictures taken at the beginning and the end of the experiment. But also allowed individual recognition on underwater videos sequences shot during reproduction (Fig. 4).
Microsatellite multiplex PCR
Amplification of eight microsatellites was carried out in a 5 µL final volume using Qiagen Type-it Microsatellite kits. Each reaction contained 1X PCR Master Mix, 0.2 µM of each unlabeled reverse (Eurofins MWG Operon) and labeled forward primer (6-FAM: Ssa85, Str73INRA, Ssa410Uos, HEX: Str60INRA, Ssosl417, Ssa408Uos (Eurofins MWG Operon) or NED: SsoSL438, Sssp2216 (Life Technologies)) and approximately 25 ng of template DNA. The amplification reaction was carried out using a Applied Biosystem 2720 thermal cycler (Life Technologies) and consisted first in an initial denaturation at 95 °C for5 min, followed by 30 cycles of denaturing at 95 °C for 30 s, annealing at 57°C for 3 min, extension at 72 °C for 30 s and a final extension step at 60 °C for 30 min.
Eight microsatellites previously developed for salmonids were selected: Str60INRA; Str73INRA (Estoup, Presa, Krieg, Vaiman, & Guyomard, 1993); SsoSL438 (Slettan, Olsaker, & Lie, 1995); Ssa85 (O’Reilly, Hamilton, McConnell, & Wright, 1996); SsoSL417 (Slettan et al., 1995); SSsp2216 (Paterson, Piertney, Knox, Gilbey, & Verspoor, 2004); Ssa410Uos and Ssa408Uos (Cairney, Taggart, & HOyheim, 2000). We used a multiplex protocol allowing amplification of the eight loci in one polymerase chain reaction (multiplex PCR, Fig. 6) following Lerceteau-Köhler & Weiss (2006).
Genotyping
Amplified fragments were sized on a ABI 3100-Avant (Life Technologies) using a GeneScan 500 LIZ internal size standard (Life Technologies), scored twice to check error rateusing STRand software (Toonen & Hughes, 2001) and raw allele sizes were binned into discrete allele classes using MSatAllele package (Alberto, 2009) for R version 2.13.0 (R Development Core Team 2011). Figure 6. Examples of microsatellite electrophoregram profile with the multiplex PCR for one individual (a) and diagram showing allele size range of each microsatellite (b). Triangles indicate alleles at each locus. In (b), rectangles represent the potential allele size range known from the literature.
Parentage analysis
Parentage analysis was performed using Cervus software (version 3.0.3, Kalinowski 2002) to assign parents to each sampled offspring, using allele frequencies computed from the genotypes of the candidate parents. The following simulation parameters were used: 10 000 cycles, a number of candidate mothers and candidate fathers depending on the experiment, a mistyping error rate of 1%, a genotyping error rate of 1%. We used the “parents pair analysis, sexes known” option in Cervus to assign juveniles to parents. All juveniles with more than one locus missing were removed from the analysis. We accepted parentage assignment at confidence level of 80% and only when the juvenile was assigned to two parents. Hardy Weinberg equilibrium and linkage disequilibrium between loci were tested using Genepop 4.2 (Rousset, 2008) with Bonferroni correction for multiple comparisons.
Sampling sites and reproductive activity
In order to monitor the correlation between female habitat choice, egg size and egg survival, three samplings have been realized during three consecutive seasons of reproduction (from November to January 2011-2012; 2012-2013; 2013-2014) on two rivers: the River Bastan (+43° 16′ 2.51″, -1° 22′ 32.46″) and the Lizuniaga brook (43°17’02.9″N 1°37’02.2″W), a tributary of the River Nivelle. These systems have been selected because of their accessibility, and because the reproduction activity in these rivers have been previously observed by the lab team. The Bastan is more torrential, wider, water level is less variable and fish size is more variable. We combined these two sites in order to maximize the range of variation in fish traits and habitat features. The reader should also be aware that the differences in flow regimes partially conditioned my decision to sample reproductive activity. For instance, during the second year of sampling (2012-2013), high flow events constrained me to mainly sample the Lizuniaga brook because of its lesser average discharge than the River Bastan, which was not accessible at that time. An attempt on the river Urumea (from where some of the spawners were sampled for experiments B1 and B2) did not provide enough data due to the high difficulty to observe reproductive activity in situ for this river (only 1 sampled redd during the whole first winter).
In order to obtain the different samples, the first step consisted in detecting reproductive activity on a spawning site between one female and one or several males (female digging and chases between males) and to measure this female. To determine its size, the female was first photographed. Then, one conspicuous object (stone, stick of wood…) present on the picture was measured with a ruler after the end of reproductive activity on the redd. Female size was then deduced relatively to the object actual size with an image processing software (ImageJ, 1.45s). When the precise moment of fertilization was observed, we waited 30 minutes to let the female cover her eggs before processing to further samples. When we observed reproductive activity without being able to stay until fertilization, we came back the day after to confirm that the nest was finished. Initially, I had hoped to relate reproductive activity before fertilization to female habitat choice and egg survival, by first observing reproductive activity on the redd, and then manipulating eggs and tracking their survival. But the odds of observing a whole reproductive sequence, and then finding eggs were very low, so I decided to limit the protocol to measuring female size, habitat characteristics, and egg size and survival.
Habitat variables, egg size and experimental setup
Different variables were measured directly on the redd once the reproduction occurred in order to see their potential effect on offspring survival:
– Particle size of substrate moved by the female for redd construction.
– Depth of burying for eggs.
– Egg volume at the individual scale.
Table of contents :
Product quality and export volatility in international trade: an empirical assessment
1 Introduction
2 Empirical model and data
2.1 Empirical strategy
2.2 Data
2.3 Measure of export volatility
3 Quality estimation
3.1 Demand model
3.2 Identification and the instrumental variable
3.3 Quality estimation results
4 Results
4.1 Baseline model
4.2 Role of income fluctuations
4.3 Robustness
5 Conclusions
II Export quality and wage premia
1 Introduction
2 Theory
2.1 Demand
2.2 Technology
2.3 Relationship between wages and average export quality
3 Data and variables
3.1 Data
3.2 Variables of interest
4 Empirical analysis and results
4.1 Empirical model
4.2 Instrumental analysis
4.3 Additional results
5 Robustness
5.1 Alternative instruments
5.2 Alternative measures of quality
5.3 Excluded countries
6 Conclusion
III How taste proximity affects consumer quality valuation of imported varieties: Evidence from the French agri-food sector
1 Introduction
2 Theoretical framework
2.1 Setup
2.2 Implications for the price of varieties
2.3 Export prices and firm revenues
3 Data
3.1 Data on trade
3.2 Data on restaurants
4 Estimation of taste proximity
5 Empirical analysis
5.1 Empirical model
5.2 Additional controls
5.3 Robustness
6 Conclusion
General Conclusion
