1
1 INRA, UMR 0985 ESE, INRA/Agrocampus Ouest, Ecotoxicologie et Qualité des Milieux Aquatiques, 65 rue de Saint-Brieuc, 35042 Rennes, France INRA, UE 1036 U3E, Unité Expérimentale d’Ecologie et Ecotoxicologie Aquatique, 65 rue de Saint-Brieuc, 35042 Rennes, France 3 ISAE, Institut en Santé Agro-Environnement, France [1] Besnard et al. 2013. Molecular Ecology Resources, Permanent Genetic Resource Note 13: 158- Nature Reviews Genetics 14, 179–190. 143, 1795–1805. , 9: e106670 [5] Whitlock. M.C., Guillaume, F. 2009. Genetics 183, 1055–1063. Objectives Þ Investigate the evolutionary potential of pesticide tolerance in populations of a non-target species Þ Estimate the relative influence of neutral versus selective forces on genetic variation in tolerance Context = ecological risk assessment (ERA) of pesticides Biological relevance of standard toxicity testing: importance of intraspecific variation Long-term impact: incorporation of genetic and evolutionary criteria to future risk assessment procedures Questions and global approach •Genetics of copper tolerance? description of within- and between-population variation •Population genetic divergence? comparison of tolerance patterns to neutral genetic divergence •Species vs population level relevance of a standard test? proposal of an assessment method Common-garden experiment (8 Lymnaea stagnalis populations, North-Western Europe) Þ Population neutral divergence , 14 microsatellite loci [1]: Global F ST = 0.388; 95% CI = [0.354;0.430] Þ Estimation of copper tolerance (CuSO 4 , 5H 2 O): global LC 50 estimation from a range finding test based on a balanced pool of each population and family representatives (8 concentrations, 0 - 2.56 mg/L). exposure of 8 families (F1s) per population to global 48h-LC50 ( 3 replicate groups of 10 individuals per familiy) CONCLUSIONS Strong population genetic divergence in copper tolerance, consistent with neutral differentiation Divergence pattern inconsistent with homogenizing selection, i.e., with the condition required to safely extrapolate population-level results to the species level Need to account for intra-specific Q -F approach applicable to this Acknowledgements - Work funded by INRA-ONEMA Action « Phylogeny and Polluosensitivity ». Rearing and experimentations performed at INRA U3E, Rennes. Lymnaea stagnalis C l i c h é M . C o l l i n e t ( I N R A ) Control Marie-Agnès Coutellec 1 , Jessica Côte 1 , Anthony Bouétard 1 , Yannick Pronost 2 , Maïra Coke 2 , Anne-Laure Besnard 1 , Fabien Piquet 3 , Thierry Caquet 1 Genetic variation of Lymnaea stagnalis tolerance to copper: a test of selection hypotheses and its relevance for ecological risk assessment with: genetic variance between (Vb) and within populations (Vw) inbreeding coefficient (f) Þ Genetic variance decomposition: Q ST approach [2-4] Observed pattern Theoretical Evolutionary Expectation Consistency with toxicity assessment at the species level Q ST = F ST Neutral divergence (no selection involved) If F ST significant: NO Q ST > F ST Divergent selection (local adaptation) NO Q ST < F ST Homogenizing selection or trait canalization YES Control Copper exp... 40 50 60 70 80 90 100 AMS SAN BIE PUT EMM BUX HED REN Survival (%) population: Model AIC logLikelihood ratio test “model j vs model i ”: P-value (df) Copper sensitivity M1 = Y ~ treatment + size+ (1| population/family) 784.7 M2 = Y ~ treatment + size (1| population) 789.7 0.008 (M2 vs M1) M3 = Y ~ treatment + size (1| family) 801.3 <0.001 (M3 vs M1) M4 = Y ~treatment + (1|population/family) 793.5 0.001 (M4 vs M1) M5 = Y ~size + (1|population/family) 1047.9 <0.001 (M5 vs M1) Shell size M1 = size ~ treatment + (1| population/family) 474.1 M2 = Y ~ treatment + (1|population) 549.9 <0.001 (M2 vs M1) M3 = Y ~ treatment + (1|family) 486.9 <0.001 (M3 vs M1) M4 = Y ~1 + (1|population/family) 475.6 0.062 (M4 vs M1) Table 2. Summary of statistical tests performed on shell size and observed 96-h mortality. Generalized linear mixed effects models compared with a LogLikelihood ratio test (P-value). AIC = Akaike information criterion (in bold: best model). Trait and statistical model Observed Q ST - F ST Neutral Q ST - F ST 95%CI Left p- value Right p-value Copper sensitivity M 96h ~ treatment+size+(1| pop/fam) 0.024 [- 0.247;0.21 4] 0.631 0.369 Shell size Size ~ treatment + (1| pop/fam) -0.195 [- 0.251;0.22 8] 0.085 0.915 Figure 1. L. stagnalis population reaction norm to copper. Mean percent survival (SE) calculated over 8 families per population. Table 3. Summary of the Q ST -F ST analyses performed on L. stagnalis sensitivity to copper (M 96h = number of dead snails after 96-h exposure) and shell size. Left and right p-values give the probability for the observed difference between Q ST and F ST to fall within the 95% CI of the expected neutral distribution. P< 0.025 is indicative of homogenizing (left value, Q ST <F ST ) or divergent selection (right value, Q ST >F ST ) [5]. Table 1. Summary of theoretical hypotheses testable under the Q ST -F ST approach. Homogenizing selection is a prerequisite for toxicity result extrapolation from population (or single strain) to the species level. w b b ST V V f V f Q 2 ) 1 ( ) 1 (

1 INRA, UMR 0985 ESE, INRA/Agrocampus Ouest, Ecotoxicologie et Qualité des Milieux Aquatiques, 65 rue de Saint-Brieuc, 35042 Rennes, France INRA, UE 1036

Embed Size (px)

Citation preview

Page 1: 1 INRA, UMR 0985 ESE, INRA/Agrocampus Ouest, Ecotoxicologie et Qualité des Milieux Aquatiques, 65 rue de Saint-Brieuc, 35042 Rennes, France INRA, UE 1036

1INRA, UMR 0985 ESE, INRA/Agrocampus Ouest, Ecotoxicologie et Qualité des Milieux Aquatiques, 65 rue de Saint-Brieuc, 35042 Rennes, FranceINRA, UE 1036 U3E, Unité Expérimentale d’Ecologie et Ecotoxicologie Aquatique, 65 rue de Saint-Brieuc, 35042 Rennes, France

3ISAE, Institut en Santé Agro-Environnement, France

References – [1] Besnard et al. 2013. Molecular Ecology Resources, Permanent Genetic Resource Note 13: 158-159.[2] Leinonen et al. 2013. Nature Reviews Genetics 14, 179–190.[3]Bonnin et al. 1996. Genetics 143, 1795–1805.[4] Bouétard et al. 2014. PLoS ONE, 9: e106670[5] Whitlock. M.C., Guillaume, F. 2009. Genetics 183, 1055–1063.

ObjectivesÞ Investigate the evolutionary potential of pesticide tolerance in populations of a non-target speciesÞ Estimate the relative influence of neutral versus selective forces on genetic variation in tolerance

Context = ecological risk assessment (ERA) of pesticides Biological relevance of standard toxicity testing: importance of intraspecific variationLong-term impact: incorporation of genetic and evolutionary criteria to future risk assessment procedures

Questions and global approach•Genetics of copper tolerance? description of within- and between-population variation•Population genetic divergence? comparison of tolerance patterns to neutral genetic divergence•Species vs population level relevance of a standard test? proposal of an assessment method

Common-garden experiment (8 Lymnaea stagnalis populations, North-Western Europe)Þ Population neutral divergence , 14 microsatellite loci [1]: Global FST = 0.388; 95% CI = [0.354;0.430]Þ Estimation of copper tolerance (CuSO4, 5H2O):

• global LC50 estimation from a range finding test based on a balanced pool of each population and family representatives (8 concentrations, 0 - 2.56 mg/L).• exposure of 8 families (F1s) per population to global 48h-LC50 ( 3 replicate groups of 10 individuals per familiy)

CONCLUSIONS

Strong population genetic divergence in copper tolerance, consistent with neutral differentiation

Divergence pattern inconsistent with homogenizing selection, i.e., with the condition required to safely extrapolate population-level results to the species level

Need to account for intra-specific variation in standard toxicity testing: QST-FST approach applicable to this context.

Acknowledgements - Work funded by INRA-ONEMA Action « Phylogeny and Polluosensitivity ». Rearing and experimentations performed at INRA U3E, Rennes.

Lymnaea stagnalis

Cliché M. Collinet (IN

RA)

Control

Marie-Agnès Coutellec1, Jessica Côte1

, Anthony Bouétard1, Yannick Pronost2, Maïra Coke2, Anne-Laure Besnard1, Fabien Piquet3, Thierry Caquet1

Genetic variation of Lymnaea stagnalis tolerance to copper: a test of selection hypotheses and its

relevance for ecological risk assessment

wb

bST VVf

VfQ

2)1(

)1(

with: genetic variance between (Vb) and within populations (Vw) inbreeding coefficient (f)

Þ Genetic variance decomposition: QST approach [2-4]

Observed pattern Theoretical Evolutionary Expectation Consistency with toxicity assessment at the species level

QST = FST Neutral divergence (no selection involved) If FST significant: NO

QST > FST Divergent selection (local adaptation) NO

QST < FST Homogenizing selection or trait canalization YES

Control Copper exposure40

50

60

70

80

90

100

AMS

SAN

BIE

PUT

EMM

BUX

HED

REN

Surv

ival

(%)

population:

Model AIClogLikelihood ratio test

“modelj vs modeli”:P-value (df)

Copper sensitivity M1 = Y ~ treatment + size+ (1|population/family) 784.7 M2 = Y ~ treatment + size (1|population) 789.7 0.008 (M2 vs M1)M3 = Y ~ treatment + size (1|family) 801.3 <0.001 (M3 vs M1)M4 = Y ~treatment + (1|population/family) 793.5 0.001 (M4 vs M1)M5 = Y ~size + (1|population/family) 1047.9 <0.001 (M5 vs M1)

Shell size M1 = size ~ treatment + (1|population/family) 474.1 M2 = Y ~ treatment + (1|population) 549.9 <0.001 (M2 vs M1)M3 = Y ~ treatment + (1|family) 486.9 <0.001 (M3 vs M1)M4 = Y ~1 + (1|population/family) 475.6 0.062 (M4 vs M1)

Table 2. Summary of statistical tests performed on shell size and observed 96-h mortality. Generalized linear mixed effects models compared with a LogLikelihood ratio test (P-value). AIC = Akaike information criterion (in bold: best model).

Trait and statistical model ObservedQST - FST

NeutralQST - FST

95%CI

Leftp-value

Rightp-value

Copper sensitivity

M96h ~ treatment+size+(1|pop/fam) 0.024 [-0.247;0.214] 0.631 0.369

Shell size Size ~ treatment + (1|pop/fam) -0.195 [-0.251;0.228] 0.085 0.915

Figure 1. L. stagnalis population reaction norm to copper. Mean percent survival (SE) calculated over 8 families per population.

Table 3. Summary of the QST-FST analyses performed on L. stagnalis sensitivity to copper (M96h = number of dead snails after 96-h exposure) and shell size. Left and right p-values give the probability for the observed difference between QST and FST to fall within the 95% CI of the expected neutral distribution. P< 0.025 is indicative of homogenizing (left value, QST<FST) or divergent selection (right value, QST>FST) [5].

Table 1. Summary of theoretical hypotheses testable under the QST-FST approach. Homogenizing selection is a prerequisite for toxicity result extrapolation from population (or single strain) to the species level.