Definition of the migratory patterns for allis shad metapopulation in the Garonne river: Influence on the thermal niche of spawning in a global warming context C. Poulet
To cite this version:
C. Poulet. Definition of the migratory patterns for allis shad metapopulation in the Garonne river: Influence on the thermal niche of spawning in a global warming context. Environmental Sciences. 2018. hal-02609205
HAL Id: hal-02609205 https://hal.inrae.fr/hal-02609205 Submitted on 16 May 2020
HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés.
Academic year 2017 – 2018
Internship report Master’s degree of marine biology and ecology University of Bordeaux
DEFINITION OF THE MIGRATORY PATTERNS FOR ALLIS SHAD METAPOPULATION IN THE GARONNE RIVER
Influence on the thermal niche of spawning in a global warming context
By: POULET Camille
Supervised by: Alexis PAUMIER and Patrick LAMBERT Referring teacher: Hugues BLANCHET
FEVRIER- JUIN 2018 IRSTEA
0
ACKNOWLEDGEMENTS
Le stage touchant à sa fin, je tiens à remercier toutes les personnes qui m’ont apporté leur aide et leur soutien dans la réalisation de ce travail et qui ont contribué à la poursuite de celui- ci en CDD.
Tout d’abord, je tiens à remercier Mr Éric ROCHARD et Mr Fréderic SAUDUBRAY, pour m’avoir accueillie au sein de leur établissement. J’adresse un remerciement plus particulier à Éric, qui a fait les démarches administratives nécessaires pour appuyer ma demande de CDD et ce, en dépit des difficultés rencontrées.
Je tiens à remercier également Jérémy LOBRY, directeur adjoint d’unité, pour m’avoir intégrée au sein de son équipe ; mais aussi pour avoir soutenu ma candidature aux côtés d’Éric. Et parce que, tu le vaux bien, merci pour ta gentillesse, ton sourire, ta bonne humeur permanente et contagieuse, et plus encore !
Un grand merci également à Alexis PAUMIER, et Patrick LAMBERT, mes tuteurs de stage, pour m’avoir accompagnée au cours de ces 5 mois. Pour ne pas être dans l’excès, et partir dans un trop plein d’amouuur, je vais résumer ça en quelques mots. Alexis, merci pour ta patience, ta disponibilité, ton investissement, tes conseils, et ton soutient dans ce travail comme dans les démarches de thèse. Si j’en suis là, et que R n’a pas eu raison de moi, c’est quand même un peu grâce à toi ! (Poète!). Patrick, merci pour ta disponibilité, ton expertise, et la richesse de nos échanges, qui m’ont apporté de nouveaux éclairages dans la manière d’appréhender ce sujet. Merci aussi pour tes précieux conseils et tes encouragements. L’Homme expérimental à beaucoup à apprendre de l’Homme expérimenté !
Je remercie également Maud, pour son aide précieuse sur R. Dans les moments de doutes, nos discussions sur le choix de l’analyse la plus judicieuse et pertinente, m’auront été très utiles ! Je crois que ça mérite bien un mojito framboise ça !
Françoise, directrice du projet SHADEAU-FAUNA, pour m’avoir permis de participer à la réunion de projet, et de m’avoir offert l’opportunité de voir des aloses en vrai (si si !).
David, pour nos discussions mutuelles sur les petites aloses ! Merci également d’avoir joué le jeu avec Alexis, et simuler mon entretien de thèse ! Et de manière générale, merci pour ta joie de vivre, (et ton rire qu’on entend dans tout le couloir !) qui mettent toujours de bonne humeur.
De manière plus générale, je remercie toute l’équipe administrative, pour m’avoir permis de continuer en CDD. En particulier, Mme Elodie PLAUT, qui a pris en charge mon dossier en un temps record ! Merci pour votre disponibilité et votre efficacité !
1
J’adresse également mes remerciements à Elodie, qui a partagé mon bureau au cours de ces 5 mois de stage. Dans les moments de panique, nos échanges mutuels sur le bidouillage du logiciel R auront été salvateurs ! J’espère pouvoir continuer à partager ce bureau avec toi, pendant encore 3 longues années. Constance et Adeline, pour la relecture de ce rapport ; et de manière générale pour leur amitié, et leur soutient infaillible au cours de ces deux années de Master.
Enfin, parce que l’aventure ne s’arrête pas là, je remercie Géraldine LASSALLE et Patrick LAMBERT, pour leur confiance : si j’en crois ce qu’on me dit, la thèse n’est pas une aventure de tout repos, mais je pense que le jeu en vaut la chandelle, que ce soit humainement ou intellectuellement parlant !
Et parce que quelques bières (avec sirop pour les petits joueurs !), ça permet de décompresser, merci à toute l’équipe EABx, pour ces soirées du vendredi soir au PMU, ou ces apéros improvisés ! (Qui ne sont pas prêts d’être terminés !)
2
TABLE OF CONTENTS
1. Introduction ...... 1 2. Material and methods ...... 3 2.1. Study area ...... 3 2.2. Environmental data ...... 4 2.3. Upstream migration assessment ...... 4 2.3.1. Catches Per Unit Effort (CPUE) ...... 4 2.3.2. Acoustic survey ...... 2 2.3.3. Fish passes ...... 2 2.4. Modelling approach...... 2 2.4.1. Definition of migration waves ...... 2 2.4.2. Gaussian mixture models (GMM) ...... 3 2.4.3. Selection of migration waves ...... 4 2.5. Relationship with the thermal niche of spawning ...... 5 3. Results ...... 5 3.1. The population decline in the Gironde watershed ...... 5 3.2. Definition of the upstream migration of Allis shad spawners ...... 2 3.2 1. Signal variability and stability ...... 3 3.2.2. Migration speed ...... 5 3.3. Relationship with the thermal niche of spawning ...... 6 4. Discussion ...... 7 4.1. A new model for migration dynamics ...... 7 4.2. Migratory pattern of the upstream migration in the Garonne River ...... 8 4.2.1 The influence of tide cycle in the downstream area ...... 8 4.2.2. Importance of biological factors in the riverine stage ...... 9 4.2.3. The variability at the fishway ...... 9 4.2.4. The Individual Based Model implementation ...... 10 4.3. Migration cost and swimming capacity ...... 10 4.4. Focus on the linkage between migration and reproduction...... 11 5. Impact of the migration patterns on the breeding success ...... 11
3
TABLE OF FIGURES
Figure 1: Conceptual diagram representing the mechanistic model developed by Paumier et al...... 3 Figure 2: Location of the study area in the Gironde watershed. Numbers correspond to the 3 data types used for the analysis...... 4 Figure 3: Diagram of the definition of a migration wave in the present study: migration wave was approached by a Gaussian curve type. Mean (μi), squared (σi) and density probabilities (pi) corresponds to the occurring days of wave migration wave duration and the proportion of migrating fishes at time t, for wave i...... 3 Figure 4: Diagram of the main steps used to construct our model...... 4 Figure 5: Diagram of the method used in the present study to assess the migration speeds from the Gironde Estuary to the spawning grounds ...... 5 Figure 6: Abundance of fishes per year from 2003 to 2007...... 5 Figure 7: Example of fits obtained in 2006 by the model (red) and the associated abundance of fish (blue) for each sector. a) Estuary, b) Tidal part of the River, c) Fluvial zone, d) Spawning grounds, and e) Fish passes. The solids vertical lines indicate the estimated mean days for each wave, and the vertical broken line the means for waves excluded in this analysis. The red fill rectangle corresponds to the migration period. The dark arrow indicates the direction of migration, from downstream to upstream...... 2 Figure 8: Estimated parameters of Allis shad migration waves in the Gironde watershed from 2003 to 2007. Means (in days) and pi (proportion of fishes which migrate) are estimated from a Gaussian mixture using the package Mixdist. The colourless circles correspond to the waves which were excluded in the present study...... 3 Figure 9 : Boxplot investigating the site effect on the migration wave duration. The results of the wilcoxon test are represented by the letters above the boxplot. The standard deviation (σ), in “GFZ” and “SPG” appear to be lower compared to the other sites...... 4 Figure 10: Cluster dendrogram from the agglomerative hierarchical clustering analysis (HAC) performed on the Gaussian estimated parameters. The dashed horizontal line (dissimilarity threshold) separated 4 groups...... 4 Figure 11: Main waves observed in each sector from 2003 to 2007. Migration speeds were calculated from the Gironde Estuary to the spawning ground, distant from 250 km...... 5 Figure 12: Cluster dendrogram from the agglomerative hierarchical clustering separating the site-year combination on the gaussian estimated parameters. The dashed horizontal line (dissimilarity threshold) separated 2 groups. Group 1 was composed of the SPG 2013 and group 2 by the remains site-year combination...... 6
4
5
1. Introduction
It is widely accepted that we are entering a period of unprecedented climate changes. Since the beginning of the Industrial Revolution, human activities have significantly impacted the Earth climate (Steffen et al. 2015). The global average temperature has increased by approximately 0.85°C, and according to the latest IPCC report, this rise could reach +4.8°C by 2100 (IPCC report, 2014). Although natural systems have already been affected by large natural climate variations, the rapidity of current changes threatens the global biodiversity (Root et al. 2003; Thuiller 2007; Wong and Candolin 2015). In response to these environmental changes, species can disperse, adjust behavior to suit the conditions or adapt through genetic evolution (Wong and Candolin 2015; Williams et al. 2008). Regarding to these observations, migratory behavior can be considered as a first adaptive response in this changing environment (Dingle and Drake 2007).
In the animal kingdom, migration is a ubiquitous biological feature in the life cycle of many species. It can be found in all major taxa, from microscopic crustaceans to the large sea-dwelling mammals and plays a key role in the population dynamics (Hansson and Hylander 2009; Horton et al. 2011; Brönmark et al. 2014). Although the processes underlying migration are still poorly understood and are a matter of some debates, migration is defined as a cyclical phenomenon which occurs between two or more separate habitats and includes a movement of a large part of the population (Begon et al, 1990; Metacalfe et al, 2002; Secor, 2015). The migratory behavior usually leads to crucial processes in the life cycle as growth, maturation and reproduction (Northcote, 1978). As most of biological traits, migration is influenced by environmental changes, especially by temperature which plays a key role in initiating and modulating the migration movement (Baglinière and Élie 2000).
Over the last decades, numerous studies have shown an impact of the global warming on a various levels of biological organization, from the individual to the ecosystem levels (H. O. Pörtner and Peck 2010). It includes physiological changes at the molecular, cellular and organism scale; shift in species distribution, phenology and community (H. Pörtner 2001; H. . Pörtner 2002; Hughes 2000; Walther et al. 2002; Root et al. 2003). For example, many species have modified their life cycle event, especially birds or insect, which flying, reproducing and spawning earlier in the season (Hughes 2000; Walther et al. 2002; Root et al. 2003). Changes in migratory patterns, distribution and abundances have also been observed for a large array of fish species from the polar to tropical regions (Roessig et al. 2004).
During their life cycle, diadromous fishes which migrate up and downstream across long distances undergo considerable pressures, from natural as well as anthropogenic pressures (Baglinière et al. 2003). Dams construction habitat degradation, water pollution deterioration in spawning grounds and unsustainable fisheries have been widely cited as
1
responsible causes of the observed decline in abundance of these species (Thorstad et al. 2008; Maes, Stevens, and Breine 2008; Masters et al. 2006). The negative effects of these anthropogenic pressures have been amplified by the ongoing global warming (de Groot 2002; Limburg and Waldman 2009). Given this outlook, a better understanding of migration processes is essential for the protection of these endangered populations.
Among diadromous fishes, the Allis shad (Alosa alosa, L. 1758), is an anadromous clupeid characterized by two main migrations. Juveniles first migrate from the river to the sea to complete their growth stage, and then return to their natal river at the age of 5 years to spawn between August and December (Appendix 1). Historically, the populations ranged along the Atlantic coast from Norway to Morocco. Since the beginning of the 19th century many shads populations have declined, leading to a contraction in the species distribution (Lassalle et al. 2008). Consequently, the species has been classified as “vulnerable” by the French committee of the International Union for the Conservation Nature (IUCN) The Garonne population, known as the largest Allis shad population in Europe, is not spared by this global decline. Indeed, the population recently collapsed, despite the drastic fishing ban implemented in 2008 to reverse this downward trend (Rougier et al. 2012).
For a few years, several studies were carried out in the Gironde watershed to investigate the impact of temperature on the reproduction, embryos and larvae survival. These approaches allowed an estimation of the optimal thermal range of the Allis shad and defined the thermal niche of reproduction (Jatteau et al. 2017; Lambert et al. 2018; Paumier et al, submit). Likewise, knowledge on the factors affecting upstream migration of Allis shad has been provided in several occasions (Rochard 2001; Acolas et al. 2004, 2006; Boisneau et al. 2008). However, no studies have characterised yet, the dynamic of this process along the main water course and have linked reproduction and migration processes. The patterns of migration determine the number and the arrival periods of spawners on the spawning grounds. A desynchronization between migration and reproduction may lead to compromise breeding success and therefore survival rate of the young stages (Lucas and Baras 2001).
The present study aimed at (i) characterizing migratory patterns of Allis shad spawners to provide a global definition of this process (ii) understanding and evaluating how it influences the expression of the thermal niche of spawning previously established (Paumier et al, submit). Firstly, migration waves along the Garonne River were defined using a MULTIFAN-approach (D. A. Fournier et al. 1990), and this typology was then regarded throughout the environmental parameters. Secondly, the variability of this process and its impact on the thermal niche of spawning was assessed throughout a clustering. Finally, the impact of migration variability on the breeding success was discussed.
In a broadest sense, this present work is carried out as part of the Alexis Paumier’s PhD, which investigates the adaptation of spawning in relation to the larvae and juvenile survival. The results obtained will be used to implement a mechanistic model (questions in
2
red; Fig. 1), identifying the optimal spawning strategy in order to ensure survival of the young stages in a global warming context.
Figure 1: Conceptual diagram representing the mechanistic model developed by Paumier et al.
2. Material and methods 2.1. Study area
Migration waves were defined for the Allis shad metapopulation in the Gironde watershed (Fig. 2). The analysis was conducted from the Gironde Estuary to the spawning grounds, on the Garonne River. The Gironde estuary has a total surface area of 625 km2 (measure at high tide) and extends over 80 km from the sea to the Bec d’Ambès, confluence point of the Garonne and Dordogne Rivers (Girardin and al.,2007; Castelnaud, Rochard, and Le Gat 2001). The tide is semi-diurnal and has a 14-day period.
Data have been collected on five sections: Estuary (“EST”), tidal part of the Garonne River (“GMZ”), Garonne Fluvial Zone (“GFZ”), spawning grounds (“SPG”), and Golfech dam (“FP”) on the Garonne River. Two sections (“EST” and “GMZ”) are under the influence of the dynamic tide and extend over 150 km. The remains three sections (“GFZ”, “SPG” and “FP”) are under fluvial regulation. In the tidal part of the river (“GMZ”), hydrodynamic is complex and results of the tide and flow effect combination (Castelnaud and Cauvin., 2002; Girardin and al., 2007). The Garonne River is characterized by seven main spawning grounds (Carry and Jo, 2012). The Golfech dam, located at 270 km from the sea, is the first barrier on this river (Fig. 2). In 1987, fish passes with videos counting equipment were installed at the dam location, to restore diadromous species free movement and ensure reproduction (Travade et al, 1992).
3
Despite this effort, most of fishes reproduce below the dam due to the difficulty to cross this barrier (Groux et al, 2015; Paumier et al, submit).
Figure 2: Location of the study area in the Gironde watershed. Numbers correspond to the 3 data types used for the analysis.
2.2. Environmental data
Daily estimates of water temperature (°C) and flows (m3. s-1) on the Garonne River were used to related the environmental parameters and the upstream migration. Data have been collected from 2003 to 2007, at the Golfech hydroelectric power plant. The average temperature and flow were calculated for each year.
2.3. Upstream migration assessment
The migration of Allis shad spawners was estimated using three data types: (i) Catch per unit effort (CPUE) (ii) acoustic monitoring on the spawning grounds (“SPG”), (iii) observations from the fish passes. Data were available from 2003 to 2007 (moratorium, 2008) except on the tidal part of the river, where no data has been collected between 2003 and 2004.
2.3.1. Catches Per Unit Effort (CPUE)
CPUE (Catch per Unit Effort) is a relative abundance index, commonly defined as the ratio between the number of target fishes caught and the sampling effort (days of fishing for a trammel net in our case) (FAO). CPUE data were provided by both professional and non- professional fishermen in three sections: Estuary (“EST”, only professional), tidal part of the Garonne River (“GMZ”) and Garonne Fluvial zone (“GFZ”). Fishes were caught during
4
their upstream migration between March and June, during the lamprey and shad fishery (Castelnaud and Cauvin, 2002; Castelnaud, Rochard, and Le Gat., 2001).
2.3.2. Acoustic survey
Population reproduction was estimated by counting the number of spawning acts for five consecutive reproductive seasons. Spawning occurs during the night following a succession of characteristic behavioral sequences characterized by a fast, circular and noisy movement at the surface (Menesson et Boisneau, 1990; Baglinière and Elie., 2000). Splashing emissions were recorded by direct or indirect observations to assess the spawning activity. To estimate the number of individuals on the spawning ground, we assumed that (i) spawners spread eggs in only one spawning grounds (ii) only one female is involved on a spawning act, (iii) sex ratio is balanced, (iv) each female spawn in average of 7.5 times. (Cassou-Leins and Cassou- Leins., 1981 ; Menesson et Boisneau, 1990; Chanseau et al. 2004; Acolas et al, 2006).
2.3.3. Fish passes
Since 1987, fish surveys were ensured by video counting equipment installed at Golfech dam. When fishes reached the crossing device, the video equipment system allowed their recorded (Travade et al,1992). An identification was then made by Migado. This monitoring provided a reliable estimation of the migration process throughout the reproductive season (Bellariva 1998).
2.4. Modelling approach
2.4.1. Definition of migration waves
Migration wave is defined as a dense flux of fish, which migrates together approximately at the same time, in response more or less, to the same environmental cues in the neighbour area (Rochard 2001; Baglinière et al. 2003).
Migration wave can be simulated by a Gaussian curve where the mean μi corresponds to the day of the migration peak, the standard deviation σi, is related to the migration wave duration , and pi (t) is the proportion of migrating fishes at time t, for wave i (Fig. 3).
2
휋 푖 (t)
σ푖
μ푖 Time (days) Figure 3: Diagram of the migration wave definition in the present study: Migration wave was approached by a Gaussian curve type. Mean (μi), squared (σi) and density probabilities (pi) corresponds to the occurring days of wave, migration wave duration, and the proportion of migrat ory fishes at time t, for wave i.
The proportion pi,k of migrating fishes from wave i in a site k at time t is calculated by Eq. 2
푡−μ 2 −( 푘,푖) 1 σ p푖,푘(t) = 푒 푘 (Eq.2) √2휋σ푘
2.4.2. Gaussian mixture models (GMM)
The analysis was based on the MULTIFAN approach, a likelihood based method developed by Fournier, to estimate the growth parameters and the age composition from multiple length frequency data set (D. A. Fournier et al. 1990; Daid A Fournier, Hampton, and Sibert 1998). We considered each time series in each site for a given year as a sum of several Gaussians which can be assessed by a Gaussian mixture model. Gaussian mixture model is parametric
probabilistic model mainly used to estimate parameters (πk,i, μk,i, σk) of Gaussians components using the iterative expectation maximization (EM) algorithm. (Reynolds 2015). A BIC (Bayesian Information criterion) is tied to the model to select the best one. In our study, we can assess our Gaussian mixture by the Eq.3
푛 푃푘 (푡) = ∑푖=1 π푖,푘(푡, µ푖,푘, 휎푘) (Eq.3)
where Pk (t) is the relative proportion of fish in the time series from the Gaussian mixture for
a site-year combination; k is the site, 휋푖,푘the relative importance of migration wave i in the time series from site k, µi,k and σ,k the parameters of each Gaussian i.
The model was performed on Rstudio® software (2.1) using the function “mix” and “mixconstr” of the “Mixdist” packages (Macdonald, 2012). In view of the result of the preliminary analysis, we constrained our model to consider 5 waves and fit the same number of waves for all the time series.
3
Considering the tides influence on the upstream migration (Rochard, 2001), we considered that migration waves, at least at the start of the migration in estuary, are related to the tidal cycle and then spaced by around 14 days. We therefore constrained the waves peaks (µ) to be spaced by a 14-day period. In addition, to avoid too flat waves which did not respect our definition, the standard deviations σ were constrained to be the same for all waves in the same site for given year (see Eq.3).
2.4.3. Selection of migration waves
Among the 5 waves, some had very low Pi values (i.e. number of individuals in the specific migration waves). In order to provide a most reliable definition of the migratory process dynamics, only the wave combination which included 95% of the total individuals has been considered. Beyond this threshold, we assumed that the remaining waves have no ecological meaning.
Figure 4: Diagram of the main steps used to construct our model.
2.4.4. Stability and variability of migratory patterns
To evaluate the variability of migration waves, 3 analysis were performed: (i) the standard deviation (σ) was compared in each site using a Wilcoxon signed rank test. This first analysis allowed to estimate the signal deformation during the upstream migration. (ii) a hierarchical agglomerative clustering (HAC) based on the Euclidean distance was performed on wave’s parameters (μ, σ and π), using the average linkage to assess the inter-annual and inter-site variability. (iii) Migration speeds were calculated on a section of 250 km, using the time lag between the peaks of migration observed in Estuary then, on the spawning grounds (Fig. 5).
4
Figure 5: Diagram of the method used in the present study to assess the migration speeds from the Gironde Estuary to the spawning grounds
2.5. Relationship with the thermal niche of spawning
To investigate the influence of migration waves on the reproduction, the model was operated on the spawning grounds dataset, available from 2003 to 2016. A second hierarchical agglomerative clustering was realised and the dendrogram cluster obtained was compared with the dendrogram cluster of the thermal niches of reproduction (Paumier and al, submit)
3. Results 3.1. The population decline in the Gironde watershed
Since 2005, a decline in fish abundances is observed with minimal abundances recorded in 2007 (Fig. 6). The abundances of fishes were divided by more than five.
Figure 6: Abundance of fishes per year from 2003 to 2007. 5
3.2. Definition of the upstream migration of Allis shad spawners
Upstream migration occurred between the beginning of April and the middle of July, from the Estuary to the spawning grounds. Figure 7 presents an example of the fits of the revised MULTIFAN model according to the fish frequency. The remaining fits are included in Appendix 2. A variable number of migration waves, ranging from 3 to 5, were observed depending on the site-year combination (Fig. 8). On the estuarine section (“EST”) and tidal part of the Garonne River (“GMZ”) waves were more flattening and spaced by an average of 15-18 days (Appendix 3). On the fluvial section migration period was shorter, and waves were more closely spaced in times with an average of 10 days between each wave. A summary of the definition of the upstream migration from 2003 to 2007 was presented in Figure 8.
Figure 7: Example of fits obtained in 2006 by the model (red) and the associated abundance of fish (blue) for each sector. a) Estuary, b) Tidal part of the River, c) Fluvial zone, d) Spawning grounds, and e) Fish passes. The solids vertical lines indicate the estimated mean days for each wave, and the vertical broken line the means for waves excluded in this analysis. The red fill rectangle corresponds to the migration period. The dark arrow indicates the direction of migration, from downstream to upstream.
2
Figure 8: Estimated parameters of Allis shad migration waves in the Gironde watershed from 2003 to 2007. Means (in days) and pi (proportion of fishes which migrate) are estimated from a Gaussian mixture using the package Mixdist. The colourless circles correspond to the waves which were excluded in the present study.
3.2 1. Signal variability and stability
The Wilcoxon test performed on the standard deviation (σ) have shown a significant difference between the tidal part of the river “GMZ” and both sites; spawning grounds “SPG” (p-value=0.036) and fluvial zone “GFZ” (p-value=0.016. “GMZ” and “SPG” were indeed characterized by a lower average σ value and thereby a lesser dispersed migration wave (Fig.9). A greater variability was observed on the estuary (“EST”), tidal part of the river (“GMZ”) and fish pass (“FP”) where σ values ranged from 3 to 8.
The HAC analysis performed on the wave’s parameters (ui, day of the migration peak; σi number of days characterising the wave and pi, proportion of individuals which migrate) , discriminated four groups of site-year combination (Fig.10). The first group was composed by fish passes “FP” 2004 and 2007; the second by downstream area (Estuary “EST” and tidal part of the river “GMZ”), the third one by the fluvial zone “GFZ” and the last one included upstream area (spawning grounds “SPG” and fish passes “FP”). Contrary to the other site, fish pass was classified in three different groups.
3
Figure 9 : Boxplot investigating the site effect on the migration wave duration. The results of the wilcoxon test are represented by the letters above the boxplot. The standard deviation (σ), in “GFZ” and “SPG” appear to be lower compared to the other sites.
Figure 10: Cluster dendrogram from the agglomerative hierarchical clustering analysis (HAC) performed on the Gaussian estimated parameters. The dashed horizontal line (dissimilarity threshold) separated 4 groups.
4
3.2.2. Migration speed
The number of days between the main waves observed at each site varied according to the year. The peak of migration occurred approximately at the same time in the tidal and medium part of the river. It takes place earlier than in the estuary that suggested negative migration speeds. On the upstream area (“SPG” and “FP”), the main wave was occurred at the same time, in downstream area and fish passes.
The number of days between the main wave in the estuary and spawning ground ranged from 11 days in 2006 to 49 days in 2007 (Table 1). The delay allowed to estimate an average speed of 12 km/days.
Annual estimate temperatures were similar and varying from 18,5 °C to 20,2 °C. On the contrary, a high inter-annual variability was observed on annual estimates of flows, which ranged from 258,9 m/s to 490 m/s.
Start End
Figure 11: Main waves observed in each sector from 2003 to 2007. Migration speeds were calculated from the Gironde Estuary to the spawning ground, distant from 250 km.
5
Table 1: Number of days between the main wave observed in the estuarine section and the main wave observed on the spawning ground and estimated average speed (km/days) in relation to the annual estimates of water flows (m3/s-1) and temperature (°C).
3.3. Relationship with the thermal niche of spawning
3.3.1. Global rule and model output
The second HAC analysis performed on the wave’s parameters (µi, σi and pi) on the spawning ground allowed separating 2013 from the other years. The same observation is also noticed in the previous results from Paumier et al.
Figure 12: Cluster dendrogram from the agglomerative hierarchical clustering separating the site-year combination on the gaussian estimated parameters. The dashed horizontal line (dissimilarity threshold) separated 2 groups. Group 1 was composed of the SPG 2013 and group 2 by the remains site-year combination. The arrival on the spawning ground occurs between the 106th (in 2015) and the 178th Julian days (in 2012) which correspond to the 16th of April and the 27th of June, respectively. Despite of the inter-annual variability, 3 or 4 waves were observed on the spawning grounds (Table 2, Annexe 4). The main wave occurs between the 125th day and the 178 th day with a proportion of migratory fishes ranged from 0.41 to 0.79.
6
Table 2: Model output table for the spawning ground data set available from 2003 to 2007. In Grey, the wave’s parameters which have been excluded in our analysis; in bold the estimated parameters of the main wave; in red the atypical year highlighted by the clustering
4. Discussion
4.1. A new model for migration dynamics
The previous works carrying out the migration of Allis shad were hitherto assessed the upstream migration using a weighted moving average method, highlighting therefore the spawners abundance peak in the distribution. In view of these works, the upstream migration of Allis shad has been described as a dynamic of one or two peaks of migration (Mennesson- Boisneau 1990; Bellariva 1998; Rochard 2001; Acolas et al. 2004, 2006). In the present study, the upstream migration of Allis shad was described by a new original approach. Times series were decomposed from a model originally developed for length frequency data set (MULTIFAN) and have allowed fitting migration waves using the entire distribution (D. A. Fournier et al. 1990; Daid A Fournier, Hampton, and Sibert 1998). This
7
modelling approach allowed characterizing migration waves at the temporal and spatial scale. The upstream migration has been investigated along the entire main water course, from the Gironde estuary to the spawning grounds, providing therefore a more reliable assessment of the dynamics of this process. This study brings basic knowledge about the migration process and complete the previous works carrying out on the Gironde Estuary (Rochard 2001), and spawning grounds (Acolas et al. 2004, 2006).
4.2. Migratory pattern of the upstream migration in the Garonne River
Our study highlighted the complexity of the migration process. The number of migration waves ranged from 3 to 5 in each site. Estimated parameters, i.e. migration day occurring (µ), duration of migration wave (σ) and its intensity (i.e. the proportion of migratory fish; pi) were variable according to the site and year combination. These results therefore suggested a high inter-annual and inter-site variability, which was confirmed by the both clustering. The inter-site variability was being expressed by (i) a difference in periodicity between the area under marine and fluvial regulation (ii) a deformation of the signal from the estuary to the spawning ground. Likewise, the inter-annual variability was translated into (i) a different number of waves among years for a same site, (ii) a difference in migratory pattern (e.g. 2013).
The migratory behavior is the outcome of internal and external cues that interact and stimulate the movement (Lucas and Baras 2001). It depends on the internal state of the individuals, abiotic conditions, social interactions or reproductive outcomes (Secor 2015; Nathan et al. 2008). Taking into this framework, we assumed that the differences between migratory patterns at each site might explain by different factors modulating the signal.
4.2.1 The influence of tide cycle in the downstream area
In the Gironde estuary and tidal part of the river (“GMZ”), migration waves periodicity ranged from 13,7 days to 25, 6 days, that could be consistent with the tidal influence. The impact of the tidal cycle on migration has been highlighted in few cases (Menesson Boisneau, Rochard 2001; Sabatié 1993). Rochard (2001) and Dodson (1972) suggested a passive behaviour with the tidal current. Unlike other species (e.g. salmons), shads did not use currents selectively and this oscillating movement would be related to the physical adaptation to freshwater before starting the riverine migration (Dodson, Leggett, and Jones 1972; Rochard, 2001). Nevertheless, the great variability observed on the Gironde Estuary, suggested the existence of other factors that modulate the migration. For example, dissolved oxygen concentrations, salinity or turbidity can be stressful for fishes. It may affect the upstream migration and lead fishes to wait more suitable conditions to start the riverine migration (Tétard et al. 2016; Maes, Stevens, and Breine 2008). An Estuarine Turbidity Maximum (ETM), resulting from a resuspension of particles, often characterizes Macrotidal estuaries, such as the Gironde Estuary. In the Gironde estuary, the ETM is associated with hypoxic conditions under high temperature and low river flow (Katixa, 2016). This
8
assumption proposed by Tétard et al (2016) to explain the longer pattern (18.5 days on average) exhibited by shads in the Loire estuary could be adapted to the Gironde Estuary case. It could explain the migratory pattern in the estuary and tidal part of the river, and therefore the longer periodicity observed some years. However, a further analysis of turbidity and oxygen conditions in the Gironde estuary across the study period will be requisite to confirm this hypothesis.
4.2.2. Importance of biological factors in the riverine stage
From the medium part of the river to the spawning grounds, waves were spaced by an average of 10 days and were less dispersal (decreasing on the standard deviation), compared to the estuary and tidal part of the river. Such migratory patterns could have resulted from (i) the tidal dissipation in the medium and higher part of the river, (ii) a difference in the nature and importance of cues modulating and triggering the movement. On the fluvial section, tidal range has no effect. The internal rhythm of migration is modulated by other external factors such as water temperature and water discharge, which are widely cited as the main factor affecting the upstream migration of Allis shad (Menesson-Boisneau et Boisneau; 1990; Sabatié, 1993; Menesson-Boisneau et al, 2000). The latter interacts with biological factors (e.g. maturity state), whose the importance increases as migration proceeds upstream (Baglinière 2001, Baglinière and Elie, 2003). Some physiological and internal modifications (e.g. modification in liver and muscles structure) took place during the riverine stage to support the energetic expenditures of spawning runs and could modulate the movement at the individual scale (Baglinière, 2001). In addition, shads, like most of the migratory fishes (e.g. Herrings, sardines, salmons..) traveling in schools during the riverine stage, to facilitate social transmission and learning of migration routes; increase propulsion capacities by coordinating their movement, or evade threats (e.g. predators) (Lucas and Baras 2001; Secor 2015). This observation could justify the shorter waves observed in these two sites. It is, however, important to consider the gap in time series on the medium part of the river, between 2003 and 2004, because it could be explained the significant difference observed between the medium and the tidal part of the river.
As we suggested before, the migration process resulted on a combination of several factors. Environmental conditions fluctuate among the year and explain the inter-annual variability.
Schools are dynamic in membership and can dissipate or merge with other aggregation (e.g. Atlantic herring) (Makris et al, 2009; Secor, 2015). Temporal and spatial dynamic scale must be considered to explain the dynamic and intensity of waves.
4.2.3. The variability at the fishway
Our study highlighted a great inter-annual variability on the migratory pattern at the fish passes compared to the other site. The clustering did not classify the site-year combination in the same group (see Fig.10). Such separation may be explained by a great variability in the wave flattening and the occurring day of the migration peak. These results differ from what
9
has been observed in the Mondego River (Portugal), where an important proportion of fishes crossed the device at the same period, when the flow conditions became more suitable (Stratoudakis et al. 2016). According to this study we should observe a lesser spread migration waves with important pi-value. These results therefore suggested a highly variable temporal window within which environmental conditions enable passage of individuals. In addition, weir fishways were documented to be stressful for shads and other fishes which traveling in schools (Groux and al, 2015). The difficulty to cross these barriers may be reflected by the extents meandering movements below the dam, as well as a stationary behavior. However, some other factors should be considered to explain the migratory pattern observed at the dam. A low light intensity, unfavourable hydraulic conditions at the device, or a unsuitable fishway dimensioning has now documented to affect the behavior of shad below the dam (Groux and al, 2015).
4.2.4. The Individual Based Model implementation
This study aimed at providing a global definition of the migratory process, to implement the Individual Based Model developed by Paumier et al. (Fig.1). The variability of the migratory patterns across the different sector emphasized the complexity of the migratory process and the difficulty to grasp it. This observation was besides confirmed by the analysis of migratory patterns on the spawning ground for 14 reproductive seasons. Despite this variability, some stability is observed and allows us to give reliable estimation of the migration process on the spawning ground. Considering our results, we can define the following general rule (i) the arrival of spawnners on the spawning ground mainly takes place in 3 or 4 successive waves, with an average of 10-day intervals. (ii) the first wave is often the smaller one and occurs around the May 4th; (iii) the intensity of migration waves (pi) is the most fluctuating parameters (iv) the standard deviation σ ranged around 4 days.
4.3. Migration cost and swimming capacity
The analysis of migration speeds showed an average speed of 12 km/day. During the observation period differences were observed, with relatively low speeds in 2005 and 2007, and a higher average speed observed in 2006. The migration speed obtained was similar to the previous monitoring in the Garonne River for the same 250 km section (Bellariva 1998). However, our estimation was lower than in the Gironde Estuary (Rochard 2001) and Loire River (Menesson, 1993), which estimated an average speed of 21km/days. Nevertheless, average speeds in these both studies were calculated for smaller sections; 70 km and 150 km respectively. These lower values may be explained by a greater expenditure of energetic stores, regarding to the distance travelled (Bellariva 1998). This assumption has been confirmed by several studies carrying out of Alosa sapidissima in the Connecticut River (Leonard and McCormick 1999; Castro-Santos and Letcher 2010). These studies attested of depletion in energy stores with the migration distance and emphasized the importance of hydro-climatic conditions such as water flow or temperature. Regarding to this observation, the reduction in migration speed in 2007, could be explained by two ways (i) the high-water
10
flow recorded in 2007, may be involved in a greater energy expenditure leading to a reduction in migration speed (ii) the high-water flow interrupted the upstream migration and inducing a lower migration speed.
Nevertheless, the average swimming speed was established for each year, using the time lag between the main waves observed on the estuary, then on the spawning ground. This method, widely used in the literature to estimate swimming speeds (Bellariva, 1998; Rochard, 2001), seems particularly simplistic given the complexity of migration processes. We indeed assumed the main wave remained the same over time, but our study emphasized that it was not evident. An approach considering all waves should be more appropriate and should allow estimating migration speeds more reliably. Due to the difficulty to perform such exercise, and owing to time constraints, we did not implement this method.
4.4. Focus on the linkage between migration and reproduction
On the spawning grounds, migration waves and migratory patterns varied depending on the year and may be related to the hydro-climatic conditions. For example, in 2013, the atypical migration may be explained by extreme-weather conditions in the Gironde watershed (significant higher flow associates with cold temperature). Thus, the atypical migration could partially explain the atypical spawning season observed in 2013 (Paumier et al, submit). For the other atypical reproduction (Garonne 2008; 2014 and 2016), the upstream migration took place in a usual way. This observation can confirm the hypothesis assumed by Paumier, that other factors such as stream velocity blocked the reproduction (Paumier et al, submit), and did not affect the migration. In view of this, migratory process seems to be more flexible regarding to the environmental conditions.
5. Impact of the migration patterns on the breeding success.
As previously highlighted, phenology changes (i.e. the time delay of a process) in migration and reproductions are notably assumed to be an efficient response to environmental changes as temperature and flow fluctuations. Despite these two processes are commonly studied in a context of global warming, the effect of migration on reproduction is poorly documented. The present study demonstrated the high variability of migration waves features among years, and then reflected upon the effect of these migratory patterns on the reproductive success. A valuable way to assess this impact, would be to compute survival of the early stages (Jatteau et al. 2017; Lambert et al. 2018), as a measure of reproductive success according to the observed migration pattern for each year. Different hypothesis of spawning behavior (random i.e. a constant daily probability of spawn; optimal i.e. spawn in the best conditions) can be considered. To perform such analysis, we will compute early stage survival for shifted migration patterns (e.g. thermal conditions of 2007 with the migration pattern of 2015). The difference of survival between observed and shifted patterns will allow evaluating the influence of migration in reproductive success.
11
REFERENCES
Acolas, M, M Begoutanras, V Veron, H Jourdan, M Sabatie, and J Bagliniere. 2004. ‘An Assessment of the Upstream Migration and Reproductive Behaviour of Allis Shad ( L.) Using Acoustic Tracking’. ICES Journal of Marine Science 61 (8): 1291–1304. https://doi.org/10.1016/j.icesjms.2004.07.023. Acolas, M, V Veron, H Jourdan, M Begout, M Sabatie, and J Bagliniere. 2006. ‘Upstream Migration and Reproductive Patterns of a Population of Allis Shad in a Small River (L’Aulne, Brittany, France)’. ICES Journal of Marine Science 63 (3): 476–84. https://doi.org/10.1016/j.icesjms.2005.05.022. Baglinière, Jean-Luc, and Pierre Élie. 2000. Les aloses, Alosa alosa et Alosa fallax spp.: écobiologie et variabilité des populations. Paris: Institut national de la recherche agronomique. Baglinière, Jean-Luc, M.R. Sabatié, Eric Rochard, P Alexandrino, and M.W. Aprahamian. 2003. The Allis Shad Alosa Alosa: Biology, Ecology, Range, and Status of Populations. Vol. 2003. Begon,M.,C. R. Townsend, and L.Harper.1990. Ecology from Individuals to Ecosystems. Blackwell, Malden, MA. Bellariva, Gianluca. 1998. ‘Contribution à l’étude Du Déroulement de La Migration et de La Reproduction de La Grande Alose (Alosa Alosa L. ) En Garonne : Étude Prospective de La Dévalaison Des Juvéniles’. Boisneau, C., F. Moatar, M. Bodin, and Ph. Boisneau. 2008. ‘Does Global Warming Impact on Migration Patterns and Recruitment of Allis Shad (Alosa Alosa L.) Young of the Year in the Loire River, France?’ Hydrobiologia 602 (1): 179–86. https://doi.org/10.1007/s10750-008-9291-6. Brönmark, C., K. Hulthén, P.A. Nilsson, C. Skov, L.-A. Hansson, J. Brodersen, and B.B. Chapman. 2014. ‘There and Back Again: Migration in Freshwater Fishes’. Canadian Journal of Zoology 92 (6): 467–79. https://doi.org/10.1139/cjz-2012-0277. Castelnaud, G., E. Rochard, and Y. Le Gat. 2001. ‘Analyse de la tendance de l’abondance de l’alose alosa alosa en gironde à partir de l’estimation d’indicateurs halieutiques sur la période 1977-1998.’ Bulletin Français de La Pêche et de La Pisciculture, no. 362– 363: 989–1015. https://doi.org/10.1051/kmae:2001032. Castelnaud G. et G. Cauvin, 2002. Site atelier Gironde. In Léauté J.P(2002).Caractéristique des petites pêches côtières et estuariennes de la côte atlantique du Sud de l'Europe . Rapport Final Sites ateliers. EC/DG Fish (DGXIV): Contrat 99/024,58p. Carry,L., et Jo,R.(2012). Suivi de la reproduction de la grande alose sur la Garonne en 2011. MIGADO. Cassou-Leins F., Cassou-Leins J.J., 1981. Recherches sur la biologie et l'halieutique des migrateurs de la Garonne et principalement de l'alose: Alosa alosa L. Thèse de doctorat, Institut National Polytechnique de Toulouse, 382p. Castro-Santos, Theodore, and Benjamin H. Letcher. 2010. ‘Modeling Migratory Energetics of Connecticut River American Shad (Alosa Sapidissima): Implications for the Conservation of an Iteroparous Anadromous Fish’. Edited by Bror Jonsson. Canadian Journal of Fisheries and Aquatic Sciences 67 (5): 806–30. https://doi.org/10.1139/F10-026.
12
Chanseau, M., G. Castelnaud, L. Carry, D. Martin-Vandembulcke, and A. Belaud. 2004. ‘Essai d’évaluation du stock de géniteurs d’alose Alosa alosa du bassin versant Gironde-Garonne-Dordogne sur la période 1987-2001 et comparaison de différents indicateurs d’abondance’. Bulletin Français de La Pêche et de La Pisciculture, no. 374: 1–19. https://doi.org/10.1051/kmae:2004023. Dingle, Hugh, and V. Alistair Drake. 2007. ‘What Is Migration?’ BioScience 57 (2): 113–21. https://doi.org/10.1641/B570206. Dodson, Julian J., William C. Leggett, and Robert A. Jones. 1972. ‘The Behavior of Adult American Shad ( Alosa Sapidissima ) During Migration from Salt to Fresh Water as Observed by Ultrasonic Tracking Techniques’. Journal of the Fisheries Research Board of Canada 29 (10): 1445–49. https://doi.org/10.1139/f72-223. Fournier, D. A., John R. Sibert, Jacek Majkowski, and John Hampton. 1990. ‘MULTIFAN a Likelihood-Based Method for Estimating Growth Parameters and Age Composition from Multiple Length Frequency Data Sets Illustrated Using Data for Southern Bluefin Tuna ( Thunnus Maccoyii )’. Canadian Journal of Fisheries and Aquatic Sciences 47 (2): 301–17. https://doi.org/10.1139/f90-032. Fournier, Daid A, John Hampton, and John R Sibert. 1998. ‘MULTIFAN-CL: A Length- Based, Age-Structured Model for Fisheries Stock Assessment, with Application to South Pacific Albacore, Thunnus Alalunga’. Canadian Journal of Fisheries and Aquatic Sciences 55 (9): 2105–16. https://doi.org/10.1139/f98-100. Girardin M., Castelnaud G., A. Laplaud, 2007. Surveillance halieutique de l'estuaire de la Gironde - Suivi des captures 2005- Etude de la faune circulante 2006. Rapport pour EDF CNPE du Blayais/Etude Cemagref, groupement de Bordeaux. Cestas.N°116,218p. Groot, S.J. de. 2002. ‘A Review of the Past and Present Status of Anadromous Fish Species in the Netherlands: Is Restocking the Rhine Feasible?’ Hydrobiologia 478 (1): 205– 18. https://doi.org/10.1023/A:1021038916271. Groux, F., Therrien. J., Chanseau, M., Courret, D., Tétard, S. 2015. Actualisation des connaissances sur l'éfficacité et la conception des dispositifs de montaison pour l'alose- Projet LIFE09NAT/DE/000008- Conservation and restoration of the Allis shad in the Gironde and Rhine watersheds - Action A1. Rapport de WSP à ONEMA.85p. Hansson, L.-A., and S. Hylander. 2009. ‘Size-Structured Risk Assessments Govern Daphnia Migration’. Proceedings of the Royal Society B: Biological Sciences 276 (1655): 331– 36. https://doi.org/10.1098/rspb.2008.1088. Horton, T. W., R. N. Holdaway, A. N. Zerbini, N. Hauser, C. Garrigue, A. Andriolo, and P. J. Clapham. 2011. ‘Straight as an Arrow: Humpback Whales Swim Constant Course Tracks during Long-Distance Migration’. Biology Letters 7 (5): 674–79. https://doi.org/10.1098/rsbl.2011.0279. Hughes, Lesley. 2000. ‘Biological Consequences of Global Warming: Is the Signal Already Apparent?’ Trends in Ecology & Evolution 15 (2): 56–61. https://doi.org/10.1016/S0169-5347(99)01764-4. IPCC, 2014: Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Core Writing Team, R.K. Pachauri and L.A. Meyer (eds.)]. IPCC, Geneva, Switzerland, 151 Jatteau, Philippe, Hilaire Drouineau, Katia Charles, Laurent Carry, Frédéric Lange, and Patrick Lambert. 2017. ‘Thermal Tolerance of Allis Shad ( Alosa Alosa ) Embryos and Larvae: Modeling and Potential Applications’. Aquatic Living Resources 30: 2. https://doi.org/10.1051/alr/2016033.
13
Katixa Lajaunie-Salla. Modélisation de la dynamique de l’oxygène dissous dans l’estuaire de la Gironde. Sciences de la Terre. Université de Bordeaux, 2016. Français.
Masters, Jerome E.G., M.-H. Jang, K. Ha, P.D. Bird, P.A. Frear, and M.C. Lucas. 2006. ‘The Commercial Exploitation of a Protected Anadromous Species, the River Lamprey (Lampetra Fluviatilis (L.)), in the Tidal River Ouse, North-East England’. Aquatic Conservation: Marine and Freshwater Ecosystems 16 (1): 77–92. https://doi.org/10.1002/aqc.686. Mennesson-Boisneau C., 1990. Migration, répartition, reproduction et caractéristique biologiques des aloses Alosa sp. dans le bassin de la Loire. Thèse de doctorat. Université de Rennes I, 106 p. Metcalfe,J .D., G.P. Arnold, and P.W.McDowall. 2002. Migration. In Handbook of Fish Biology and Fisheries, ed. P.J.B. Hart and J.D. Reynolds, 175-199. Blackwelle Scientific, Oxford. Nathan, R., W. M. Getz, E. Revilla, M. Holyoak, R. Kadmon, D. Saltz, and P. E. Smouse. 2008. ‘A Movement Ecology Paradigm for Unifying Organismal Movement Research’. Proceedings of the National Academy of Sciences 105 (49): 19052–59. https://doi.org/10.1073/pnas.0800375105. Northcote, T.G. 1978. Migratory strategies and production in freshwater fishes. In: Ecology of Freshwater Production (ed. S.D. Gerking), pp. 326-359. Blackwell, Oxford. Paumier,A., Drouineau,H., Laurent,C., Lambert.P., Reproduction and adaptation: a definition of the thermal niche of spawning for a French metapopulation of allis shad in a global warming context., submit.
14
Pörtner, H. 2001. ‘Climate Change and Temperature-Dependent Biogeography: Oxygen Limitation of Thermal Tolerance in Animals’. Naturwissenschaften 88 (4): 137–46. https://doi.org/10.1007/s001140100216. Pörtner, H. O., and M. A. Peck. 2010. ‘Climate Change Effects on Fishes and Fisheries: Towards a Cause-and-Effect Understanding’. Journal of Fish Biology 77 (8): 1745– 79. https://doi.org/10.1111/j.1095-8649.2010.02783.x. Pörtner, H.O. 2002. ‘Climate Variations and the Physiological Basis of Temperature Dependent Biogeography: Systemic to Molecular Hierarchy of Thermal Tolerance in Animals’. Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology 132 (4): 739–61. https://doi.org/10.1016/S1095- 6433(02)00045-4. Reynolds, Douglas. 2015. ‘Gaussian Mixture Models’. In Encyclopedia of Biometrics, edited by Stan Z. Li and Anil K. Jain, 827–32. Boston, MA: Springer US. https://doi.org/10.1007/978-1-4899-7488-4_196. Rochard, E. 2001. ‘Migration anadrome estuarienne des géniteurs de grande alose Alosa alosa, allure du phénomène et influence du rythme des marées.’ Bulletin Français de La Pêche et de La Pisciculture, no. 362–363: 853–67. https://doi.org/10.1051/kmae:2001023. Roessig, Julie M., Christa M. Woodley, Joseph J. Cech, and Lara J. Hansen. 2004. ‘Effects of Global Climate Change on Marine and Estuarine Fishes and Fisheries’. Reviews in Fish Biology and Fisheries 14 (2): 251–75. https://doi.org/10.1007/s11160-004-6749- 0. Root, Terry L., Jeff T. Price, Kimberly R. Hall, Stephen H. Schneider, Cynthia Rosenzweig, and J. Alan Pounds. 2003. ‘Fingerprints of Global Warming on Wild Animals and Plants’. Nature 421 (6918): 57–60. https://doi.org/10.1038/nature01333. Rougier, T., P. Lambert, H. Drouineau, M. Girardin, G. Castelnaud, L. Carry, M. Aprahamian, E. Rivot, and E. Rochard. 2012. ‘Collapse of Allis Shad, Alosa Alosa, in the Gironde System (Southwest France): Environmental Change, Fishing Mortality, or Allee Effect?’ ICES Journal of Marine Science 69 (10): 1802–11. https://doi.org/10.1093/icesjms/fss149. Secor, David H. 2015. Migration Ecology of Marine Fishes. Baltimore: Johns Hopkins University Press. Steffen, Will, Wendy Broadgate, Lisa Deutsch, Owen Gaffney, and Cornelia Ludwig. 2015. ‘The Trajectory of the Anthropocene: The Great Acceleration’. The Anthropocene Review 2 (1): 81–98. https://doi.org/10.1177/2053019614564785. Stratoudakis, Yorgos, Catarina Sofia Mateus, Bernardo Ruivo Quintella, Carlos Antunes, and Pedro Raposo de Almeida. 2016. ‘Exploited Anadromous Fish in Portugal: Suggested Direction for Conservation and Management’. Marine Policy 73 (November): 92–99. https://doi.org/10.1016/j.marpol.2016.07.031. Tétard, Stéphane, Eric Feunteun, Elise Bultel, Romain Gadais, Marie-Laure Bégout, Thomas Trancart, and Emilien Lasne. 2016. ‘Poor Oxic Conditions in a Large Estuary Reduce Connectivity from Marine to Freshwater Habitats of a Diadromous Fish’. Estuarine, Coastal and Shelf Science 169 (February): 216–26. https://doi.org/10.1016/j.ecss.2015.12.010. Thorstad, Eva B., Finn Økland, Kim Aarestrup, and Tor G. Heggberget. 2008. ‘Factors Affecting the Within-River Spawning Migration of Atlantic Salmon, with Emphasis on Human Impacts’. Reviews in Fish Biology and Fisheries 18 (4): 345–71. https://doi.org/10.1007/s11160-007-9076-4. Thuiller, Wilfried. 2007. ‘Climate Change and the Ecologist: Biodiversity’. Nature 448 (7153): 550–52. https://doi.org/10.1038/448550a.
15
Travade, F., M. Larinier, D. Trivellato, and J. Dartiguelongue. 1992. ‘Conception d’un Ascenseur à Poissons Adapté à l’alose (Alosa Alosa) Sur Un Grand Cours d’eau : I’ascenseur de Golfech Sur La Garonne’. Hydroécologie Appliquée 4: 91–119. https://doi.org/10.1051/hydro:1992107.
Walther, Gian-Reto, Eric Post, Peter Convey, Annette Menzel, Camille Parmesan, Trevor J. C. Beebee, Jean-Marc Fromentin, Ove Hoegh-Guldberg, and Franz Bairlein. 2002. ‘Ecological Responses to Recent Climate Change’. Nature 416 (6879): 389–95. https://doi.org/10.1038/416389a. Williams, Stephen E, Luke P Shoo, Joanne L Isaac, Ary A Hoffmann, and Gary Langham. 2008. ‘Towards an Integrated Framework for Assessing the Vulnerability of Species to Climate Change’. Edited by Craig Moritz. PLoS Biology 6 (12): e325. https://doi.org/10.1371/journal.pbio.0060325. Wong, B. B. M., and U. Candolin. 2015. ‘Behavioral Responses to Changing Environments’. Behavioral Ecology 26 (3): 665–73. https://doi.org/10.1093/beheco/aru183.
16
APPENDIX
Figure 11: Life cycle of the Allis shad (Alosa alosa). It characterized by to main migrations:
Appendix 1: Life cycle of the Allis shad (Alosa alosa) characterized by an upstream and downstream migration.
a)
17
b)
c)
18
d)
Appendix 2: Example of fits obtained by the model (red) and the associated abundance of fish (blue) for each year a) 2003, b) 2004, c) 2005, d) 2007. The solids vertical lines indicate the estimated mean days for each wave, and the vertical broken line the means for waves excluded in this analysis. The red fill rectangle corresponds to the migration period. The dark arrow indicates the direction of migration, from downstream to upstream.
Appendix 3: Average number of day between each wave at the five site from 2003 to 2007.
19
Appendix 4: Fits to the abundance frequency data obtained with the Mixdist package on the spawning ground from 2003 to 2016. The solids vertical lines indicate the estimated mean days for each wave and the vertical broken line, the mean for waves excluded in this analysis. The red fill rectangle corresponds to the migration period.
20