Evolutionary responses to crude oil from the Deepwater Horizon oil spill by the copepod Eurytemora affinis

Abstract The BP Deepwater Horizon Oil Disaster was the most catastrophic offshore oil spill in U.S. history, yet we still have a poor understanding of how organisms could evolve in response to the toxic effects of crude oil. This study offers a rare analysis of how fitness‐related traits could evolve rapidly in response to crude oil toxicity. We examined evolutionary responses of populations of the common copepod Eurytemora affinis residing in the Gulf of Mexico, by comparing crude oil tolerance of populations collected before versus after the Deepwater Horizon oil spill of 2010. In addition, we imposed laboratory selection for crude oil tolerance for ~8 generations, using an E. affinis population collected from before the oil spill. We found evolutionary increases in crude oil tolerance in the wild population following the oil spill, relative to the population collected before the oil spill. The post‐oil spill population showed increased survival and rapid development time in the presence of crude oil. In contrast, evolutionary responses following laboratory selection were less clear; though, development time from metamorphosis to adult in the presence of crude oil did become more rapid after selection. We did find that the wild population, used in both experiments, harbored significant genetic variation in crude oil tolerance, upon which selection could act. Thus, our study indicated that crude oil tolerance could evolve, but perhaps not on the relatively short time scale of the laboratory selection experiment. This study contributes novel insights into evolutionary responses to crude oil, in directly examining fitness‐related traits before and after an oil spill, and in observing evolutionary responses following laboratory selection.


| INTRODUCTION
The BP Deepwater Horizon Oil Disaster of 2010 was the most catastrophic offshore spill in U.S. history, with an estimated 4.6 million barrels of oil spilling into the Gulf of Mexico over 86 days (Griffiths, 2012). Following this oil spill, 1.6-2.6 × 10 4 metric tons of hydrocarbons accumulated on the seafloor and persisted long after oil was no longer present in the water column (Montagna, Baguley, Cooksey, & Hyland, 2017;Yan et al., 2016). Carbon isotope tracers revealed that transformed oil products persisted for years, particularly in low energy environments (Pendergraft & Rosenheim, 2014). Additionally, the labile hydrocarbons were transferred to the planktonic food web, particularly to mesozooplankton, consisting mostly of calanoid copepods (Graham et al., 2010).
While the adverse effects of the Deepwater Horizon oil spill endure to the present day, relatively little is known regarding the potential evolutionary responses of organisms to the toxic effects of crude oil. Such information is crucial, as evolutionary responses are critically important in cases where acute stress exceeds the capacity of organisms to respond through phenotypic plasticity. Additionally, many species that dominate the water column, such as mesozooplankton (e.g., copepods), are weak swimmers and are deficient in their ability to disperse. In addition, brackishwater species in the Gulf would likely have limited dispersal capacities due to their restricted salinity tolerances (see below). Given the low dispersal capacity of many marine and brackish organisms, studying their evolutionary responses to crude oil toxicity is an essential component toward understanding the potential for recovery of populations from oil spill disasters.
The few studies that have revealed rapid evolutionary responses of organisms to the toxic effects of crude oil have focused predominantly on microorganisms. For instance, laboratory selection experiments revealed that freshwater and marine phytoplankton collected from pristine uncontaminated sites are able to evolve resistance to crude oil toxicity, presumably through selection on novel mutations (Carrera-Martinez, Mateos-Sanz, Lopez-Rodas, & Costas, 2011;Romero-Lopez, Lopez-Rodas, & Costas, 2012). In bacteria, genes that encode alkane hydroxylases (AHs), responsible for crude oil degradation, are relatively widespread and could potentially be transferred to neighboring bacteria through horizontal gene transfer (Nie et al., 2014).
However, for larger multicellular organisms, adaptation to oil pollution might be far more difficult to achieve, given their longer generation times, lower rates of mutation, lower effective population size, and extremely rare incidence of horizontal gene transfer. Evolutionary adaptation to crude oil toxicity in animals would most likely need to rely on selection on standing genetic variation for crude oil tolerance within the populations. In an older study, imposing crude oil exposure for 2-10 days in marine gastropods revealed allele frequency shifts at two allozyme loci, suggesting an evolutionary response to crude oil, though the functional implications of these allele frequency shifts are unclear (Nevo & Lavie, 1989). In a comparison between populations of the Trinidad guppy Poecilia reticulate in oil-polluted versus unpolluted streams, populations in oil-polluted habitats showed little evidence for adaptation to oil pollution, but were actually maladapted to their local environment (Rolshausen et al., 2015). An intriguing comparative study found that the Gulf killifish Fundulus grandis at an oil-contaminated site in the Gulf of Mexico exhibited changes in genome expression and larval CYP1A enzyme expression (a biomarker for PAH exposure) that was divergent from fish from uncontaminated sites, 1-4 months after the oil spill (Whitehead, Dubansky, et al. 2012). This study measured acclimatory (rather than evolutionary) responses of oiled versus nonoiled controls, as it examined expression changes within the lifespan of the fish. A notable study comparing four pairs of pollution-tolerant and intolerant populations of the killifish Fundulus heteroclitus found signatures of selection at genomic regions containing key genes involved in the aryl hydrocarbon receptor (AHR) signaling pathway, which mediates tolerance of PAHs (Reid et al., 2016).
These prior studies lead to the question on the extent and speed to which fitness-related traits could evolve in response to crude oil toxicity in animals. Thus, the goal of this study was to test for rapid evolution of crude oil tolerance in the copepod Eurytemora affinis, a common resident in the Gulf of Mexico. Populations of E. affinis appeared to be largely absent from the Mississippi delta region of the Gulf of Mexico following the Deepwater Horizon oil spill, but returned to this area more than a year later (see section 2 for details). An extensive survey performed in 2010 found that the E. affinis habitats we sampled in the Gulf of Mexico had been exposed to high concentrations of total petroleum hydrocarbons (TPHs) and PAHs, in seawater, seafood, and sediment (Sammarco et al., 2013) (https://web. archive.org/web/20101126225809/http://leanweb.org:80/news/ latest/testing-results-returning-with-high-levels.html). A rebound of E. affinis populations through dispersal is unlikely, given their limited long-distance swimming capabilities. Moreover, salinity poses a major barrier to dispersal of E. affinis populations, which show heritable differences in their salinity tolerances (Lee, 1999(Lee, , 2016Lee & Petersen, 2003;Lee, Remfert, & Chang, 2007;Lee, Remfert, & Gelembiuk, 2003). This low dispersal is reflected in the highly genetically structured populations of E. affinis, even between proximate sites (Lee, 1999(Lee, , 2000Winkler, Dodson, & Lee, 2008). The critical question then is, how did populations of E. affinis rebound from the catastrophic effects of the Deepwater Horizon oil spill? Did they evolve physiological tolerance of crude oil toxicity?
To address the questions above, this study employed a commongarden reaction norm approach to examine: (i) the presence of genetic variation in tolerance of crude oil toxicity in the populations, upon which natural selection could act, (ii) evolutionary responses following the oil spill in the wild, and finally (iii) evolutionary responses following selection in the laboratory. First, we determined whether the population of E. affinis in the Gulf of Mexico evolved elevated crude oil tolerance, by directly comparing populations collected before versus after the Deepwater Horizon oil spill, in terms of survival and development time at varying levels of crude oil exposure. Next, we attempted to induce selection in the laboratory for crude oil tolerance on the population of E. affinis collected before the Deepwater Horizon oil spill. We examined evolutionary responses of the populations to the "water-accommodated fraction" of crude oil, which harbors the PAHs and other toxic compounds (Almeda et al., 2013;Faksness, Brandvik, & Sydnes, 2008;Forth, Mitchelmore, Morris, & Lipton, 2017;Liu & Kujawinski, 2015;Melbye et al., 2009).
Copepods, with their exceedingly large numbers and relatively short generation times (~15-20 days in E. affinis) are likely to experience significant evolutionary responses to environmental perturbations. In addition, the copepod E. affinis undergoes seasonal diapause on an annual basis, leaving a reservoir of resting eggs in the sediment (Ban & Minoda, 1992;Katajisto, 1996), which could contribute to standing genetic variation in the population. Following the oil spill, the eggs hatching from the sediment might have been subjected to natural selection in response to crude oil toxicity. In prior studies, E. affinis has been shown to rapidly evolve in response to environmental variables, such as temperature (Ketzner and Bradley 1982), salinity (Lee, Kiergaard, Gelembiuk, Eads, & Posavi, 2011;Lee et al., , 2007, and food concentration (Lee et al., 2013).
Understanding evolutionary responses of organisms to crude oil toxicity could provide critical insights into the potential of populations to recover from oil spill disasters. This study is notable for applying a common-garden approach to directly examine evolutionary responses before versus after an oil spill at a given location. Additionally, this study provides the first attempt to perform laboratory selection on animals in response to crude oil toxicity. Moreover, this study examined levels of standing genetic variation for crude oil tolerance within the pre-oil spill population, as a measure of evolutionary potential.
The evolutionary analysis performed in this study has important ecological implications for the functioning of food webs in estuarine and coastal ecosystems. Copepods constitute the main grazers of algae and major food source for many important fisheries throughout the world. The copepod Eurytemora affinis, in particular, is a dominant member of coastal ecosystems throughout the Northern Hemisphere (Lee, 1999(Lee, , 2000 and serves as the major food source for some of world's most important fisheries, such as herring, anchovy, salmon, and flounder (Kimmel, Miller, & Roman, 2006;Livdāne, Putnis, Rubene, Elferts, & Ikauniece, 2016;Shaheen et al., 2001;Viitasalo, Flinkman, & Viherluoto, 2001;Winkler, Dodson, Bertrand, Thivierge, & Vincent, 2003).

| Population sampling
In the Gulf of Mexico region, populations of the copepod E. affinis

| Water-accommodated fraction (WAF) of crude oil used for the oil-exposed treatments
Crude oil (50 L) from the Deepwater Horizon spill, from the Macondo Well in the Mississippi Canyon Block 252 well ("MC 252"), was obtained from BP in August, 2012. Standard methods were used to produce the water-accommodated fraction (WAF) of crude oil (Bejarano et al., 2006;Girling, 1989;Singer et al., 2001;Tsvetnenko & Evans, 2002). The WAF was made by making a 10:1 mixture of 5 PSU water to Macondo oil (3,600 mL of 5 PSU water + 400 ml of oil). The 5 PSU water was made from 0.2 μm filtered water from Lake Michigan with Instant Ocean ® added to achieve 5 PSU. The 10:1 mixture was gently mixed in an aspiration bottle for 22 hr under complete dark conditions. The newly formed WAF was pulled from the bottom of the aspiration bottle without disturbing the oil on the top and was stored in a loosely covered container at 12°C. See Almeda et al. (2013) and Cohen et al. (2014) for the composition of PAHs contained within the WAF of Deepwater Horizon Macondo Well oil.
Based on these results, 75% (72.5% in the second experiment) and 100% WAF were chosen as the treatment oil concentrations, as they showed substantial impacts on survival.

| Common-garden reaction norm experiment to compare response to crude oil of pre-and post-oil spill populations
A common-garden experiment was performed to compare responses to crude oil between the "pre-oil spill" (collected before the oil spill) and "post-oil spill" (collected after the oil spill) populations of E. affinis collected from the wild (see section 2.1 above). A commongarden approach is used to assess the heritable genetic differences between populations by rearing and observing different populations under common conditions, and thus removing the effects of acclimation to different environments (i.e., removing the effects phenotypic plasticity). Thus, the difference in response observed between the populations under common-garden conditions would represent the evolutionary differences between the populations, namely, the evolutionary changes in tolerance of crude oil from before the oil spill to after the oil spill.
Eight ovigerous females from each population were chosen at random and the eggs sacs were excised and split six ways across three WAF concentrations (0% [control], 75% and 100% WAF) with replicate vials for each treatment. Each full-sibling clutch represented a distinct genotype. Between 10 to 12 eggs were split between two replicate vials per clutch for each oil concentration and population. With eight clutches per treatment, there were a total of 88-92 individual copepods per oil concentration for each population. Each 20 ml scintillation vial contained 10 ml of 5 PSU water. The vials were placed in racks and reared in environmental chambers at 12°C with a light cycle of 15L:9D and fed the alga Rhodomonas salina in excess. Approximately 75% of the water was removed and new water was added to each vial during weekly water changes. New batches of WAF mixtures were made within 24 hr prior to every water change.
Numbers of individuals at each life stage were counted every other day in the treatments, until death or reaching adulthood. Survival was analyzed as binary data (dead or alive), for hatching (hatched or not hatched for each individual egg), survival from hatching to metamorphosis (dead or alive for each individual nauplius), and survival from metamorphosis to adult (dead or alive for each individual copepodid).
Development time was measured in terms of number of days from hatching to metamorphosis and number of days from metamorphosis to adult. Because developmental stage was recorded every other day, development time was averaged as the midpoint between observations. Metamorphosis was recorded at the transition from the life stages of nauplius VI to copepodid I. Maturation to adult was recorded when the metasomal T5 wings first appeared on females and with the development of a thick right antenna on the males.

| Laboratory selection experiment to measure evolutionary response to crude oil toxicity
The selection experiment attempted to induce an evolutionary response to crude oil toxicity in the "pre-oil spill" population, collected from the Gulf of Mexico region prior to the Deepwater Horizon oil spill (see section 2.1). Selection was imposed by transferring animals to increasingly higher oil concentrations (of the water-accommodated fraction, WAF) across ~8 generations in June 2014. Approximately 500 animals from laboratory cultures were transferred to 50% WAF (diluted with 5 PSU water) and then maintained at that concentration for at least two generations. These cultures were subsequently transferred to 65% WAF (43 days, ~2 generations), then to 70% WAF (15 days, ~1 generation), and finally to 72.5% WAF (70 days, ~3 generations). Meanwhile, control cultures were cultured in 5 PSU water in the absence of crude oil. Both the "pre-selection" (control) and "postselection" lines were cultured as described above and fed the cryptophyte alga Rhodomonas salina.
Selection for crude oil tolerance proved to be much more difficult to execute than prior selection experiments, such as selection in response to salinity (Lee et al., 2007(Lee et al., , 2011. For instance, selection in response to low salinity is typically evident after only 2-6 generations (Lee et al., 2007). However, in this experiment more than three attempts were made to create crude oil selected lines, yet the lines frequently went extinct and had difficulty surviving for multiple generations. In addition, multiple replicate selection lines were started, but most went extinct.
Response to selection was quantified by comparing the "preselection" and "post-selection" lines in a common-garden experiment. In order to remove the effects of environmental acclimation prior to the common-garden experiment, both the "pre-selection" and "post-selection" lines were held at the same condition of 0% WAF prior to the experiment. The "pre-selection" line was continuously reared at 0% WAF, whereas the "post-selection" line was moved to 0% WAF for 30 days (~1.5 generations) prior to the experiment.
As in the previous experiment (previous section), eight ovigerous females were selected at random from each population and the fullsib egg sacs, representing distinct genotypes, were excised from the female and split six ways across three treatments of WAF (this time, 0% [control], 72.5%, and 100% WAF), with two replicate vials per treatment. Between 8 to 10 eggs were split between two replicate vials per clutch for each oil concentration and population. With eight clutches per treatment, there were a total of 65-69 individual copepods per oil concentration for each population. Eggs were transferred to 20 ml scintillation vials containing ~10 ml of 5 PSU water.
The vials were cultured in environmental chambers at 12°C, with a light cycle of 15L:9D, and fed the cryptophyte alga Rhodomonas salina in excess.
Survival and development time were measured in a similar manner as in the "pre-" versus "post-oil spill" common-garden experiment above (previous section). In this experiment, observations of hatching, survival, and development time from hatching to metamorphosis were recorded daily. From metamorphosis to adult, survival and development time were recorded every other day, as in the previous experiment (previous section), and development time was averaged as the midpoint between observations. Animals were determined to reach the adult stage at the first appearance of the metasomal T5 wings in females and the development of a thick right antenna in males.

| Statistical analyses
Survival data were analyzed in a linear mixed model framework using the glmer procedure, whereas development time data were analyzed using the lmer procedure in the lme4 package of R (Bates, Maechler, Bolker, & Walker, 2015;R Core Team 2016). Survival data were treated as binary and models included a logit link function.
Effects of factors on survival and development time were examined (see next paragraph). Fixed effects (factors) included Population, Oil Concentration, and Population × Oil Concentration. Random effects included Clutch (genotype) and the random slope for Oil Concentration (with respect to Clutch). To determine the effects of each factor on the model, likelihood ratio tests were performed to compare goodness of fit between models with each factor included versus reduced models with the factor excluded.
Population comparisons were performed between "pre-oil spill" versus "post-oil spill" populations from the first experiment (section 2.3), and between "pre-selection" versus "post-selection" lines from the second experiment (section 2.4). We measured fitness-related traits in terms of survival (hatch or not hatch, dead or alive from hatching to metamorphosis, dead or alive from metamorphosis to adult) and development time (in days, hatching to metamorphosis, metamorphosis to adult). We compared their traits at three crude oil concentrations (0% WAF control, 75% or 72.5% WAF, 100% WAF).
We also tested for the significance of differences in survival or development time between populations ("pre-oil spill" vs. "post-oil spill" and "pre-selection" vs. "post-selection") at each oil concentration.
These tests used the asymptotic normality of maximum likelihood estimates for differences in log odds, in the case of survival probabilities, and for differences in means, in the case of development time differences.

| Comparisons between wild populations before and after the Deepwater Horizon oil spill
Our results revealed evolutionary shifts in fitness-related traits in a population of the copepod E. affinis in response to crude oil toxicity, following the Deepwater Horizon oil spill of 2010 in the Gulf of Mexico (Figures 1-3; Tables 1-3). In the presence of crude oil, we found significantly greater survival and shorter development times in a population collected after the oil spill ("post-oil spill") relative to a population collected before the oil spill ("pre-oil spill") from same location (Blue Hammock Bayou, LA, USA; Figures 1, 2, Table 3). As our experiment was performed under common-garden conditions, removing the effects of environmental acclimation, our results revealed rapid evolution of crude oil tolerance in a wild population at a particular location, directly following an oil spill.

| Evolution of survival following the Deepwater Horizon oil spill
Our results revealed the evolution of increased survival of the "postoil spill" population, relative to the "pre-oil spill" population, in the presence of the water-accommodated fraction (WAF) of crude oil ( Figure 1, Table 3b,c). This evolutionary shift following the oil spill was consistent with the significant effect of the factor Population on survival (Table 1a) (Table 3b,c, standard error). Hatching rate was higher in the post-than pre-oil spill population in the presence of oil (Figure 1a), but the differences were not significant (Table 3a). Unsurprisingly, and reassuringly, there were no significant differences in hatching or survival between the pre-and post-oil spill populations at the control treatment (0% WAF), in the absence of WAF oil (Table 3a-c).
Given the evolutionary shifts in survival (Table 3b,c), one would expect a significant effect of clutch (genotype) on survival, as genetic variation in oil tolerance in the pre-oil spill population would be required for natural selection to act and cause evolutionary change. We F I G U R E 1 Hatching and survival across oil concentrations for populations of the copepod Eurytemora affinis collected before versus after the Deepwater Horizon oil spill of 2010 in the Gulf of Mexico. Graphs show (a) percentage hatching, (b) percentage survival from hatching to metamorphosis, and (c) percentage survival from metamorphosis to adult, for populations collected before the oil spill (blue, "pre-oil spill") versus after the oil spill (orange, "post-oil spill") observed across three oil concentrations (no oil control, 75% and 100% WAF, "water-accommodated fraction"). Values shown are Mean ± SE for N = 8 clutches, with 10-12 eggs per clutch at each treatment. Values and significance in differences between the populations are listed in Table 3a Table 3d. Development time to metamorphosis (rather than to adult) is depicted here because many clutches did not survive to adulthood (Table 3e) did find the effect of clutch to be significant for survival from metamorphosis to adult, but not for survival from hatching to metamorphosis (Table 1a, Figure 3a). Consistent with selection acting on survival, we found that genetic variation (variance among clutches) for survival at 75% WAF was reduced in the post-oil spill population, relative to the pre-oil spill population, for both survival from hatching to metamorphosis (Table 3b) and metamorphosis to adult (Table 3c).

| Evolution of development time following the Deepwater Horizon oil spill
The "post-oil spill" population tended to exhibit more rapid development relative to the "pre-oil spill" population in the presence of oil  (Table 2a).
Greatly retarded development at higher oil concentrations for both populations (Figure 2) was consistent with the highly significant effects of Oil Concentration on development time (Table 2a)  adult were both significant (Table 2a). This result indicated the presence of ample genetic variation upon which natural selection could act ( Figure 3b). The significant random slope for Oil Concentration (with respect to Clutch, or Clutch × Oil Concentration) (Table 2a) (Table 3e) and for development time from hatching to metamorphosis (Table 3d) and metamorphosis to adult (Table 3e) at 100% WAF.

| Comparisons between populations before and after laboratory selection
In contrast to the comparisons between the "pre-" and "post-oil spill" populations ( Figure 1, Table 1A, 2A), patterns of evolution were less clear following laboratory selection for crude oil tolerance on a Gulf of Mexico (pre-oil spill) population (Tables 1, 2, 4). We did not find greater survival in the "post-selection" line ( Figure 4), but did find a trend of shorter development time from metamorphosis to adult (Figure 5b, Table 4e). Our results suggest that adaptation in response to crude oil might be relatively difficult to achieve and might not proceed very rapidly, within the time frame of the selection experiment (~8 generations).

| Lack of evolution of survival following laboratory selection for crude oil tolerance
We observed no evidence for the evolution of increased tolerance of crude oil toxicity following ~8 generations of laboratory selection T A B L E 2 Effects of multiple factors on development time (using lmer in R) imposed on the pre-oil spill population ( Figure 4). In fact, the "postselection" line exhibited a trend toward reduced survival in response to crude oil toxicity relative to the "pre-selection" line ( Figure 4b,c,  (Figure 4a,b) was consistent with the significant effect of Oil Concentration on these responses (Table 1b). The populations did differ in the pattern of survival from hatching to metamorphosis across oil concentrations (Figure 4b), consistent with the significant Population × Oil Concentration interaction (Table 1b).
Relative to the "pre-selection" line, the "post-selection" line exhibited significantly lower survival, in terms of survival from hatching to metamorphosis at the high oil concentration (100% WAF; Figure 4b, Table 4b) and survival from metamorphosis to adult at the intermediate oil concentration (75% WAF; Figure 4c,  Figure 4a; Table 4a). Interestingly, once individuals managed to survive to metamorphosis, they tended to also survive to the adult stage. That is, while survival from hatching to metamorphosis in the presence of oil was generally quite low (13%-54%; Figure 4b, Table 4b), once animals reached metamorphosis, survival to adulthood was much higher (62%-98%; Figure 4b vs. c, Table 4b vs. c).
Given the lack of an evolutionary response in terms of survival, the lack of significant effect of Clutch (genotype) on survival was not surprising (Table 1b). On the other hand, there were significant effects of Clutch on hatching rate (Table 1b).

| Evolution of development time following laboratory selection for crude oil tolerance
Development from hatching to metamorphosis was retarded in the "post-selection" line at the intermediate oil concentration (72.5% WAF; Figure 5a, For percentage hatching and survival values (a, b, c), means among eight clutches are shown. Development time estimates (d, e) were often based on lower sample size due to mortality. The Z statistic is the difference in log odds of hatching or survival or the difference in mean development time between the populations (after oil spill -before oil spill) divided by the standard error of the difference in log odds. p-values indicate significance of differences between the populations, with p < .05 shown in bold.
T A B L E 3 Survival and development time for Gulf of Mexico populations collected before versus after the Deepwater Horizon oil spill differences between the pre and post-selection lines were not significant due to low power resulting from high mortality (Table 4d,e), the patterns were nevertheless quite clear (Figure 5a,b). In fact, the pre and post-selection lines exhibited highly divergent patterns in development time from metamorphosis to adult across oil concentrations (Figure 5b), reflected by the significant Population × Oil Concentration interaction (Table 2b). As in the previous experiment (Figure 2), development time was generally retarded at the higher oil concentrations for both pre and post-selection lines, indicated by the significant effect of Oil Concentration on development time (Table 2b) Days Before "selection" After "selection" F I G U R E 4 Hatching and survival across oil concentrations for the copepod E. affinis before versus after laboratory selection for crude oil tolerance (at 50%-72.5% WAF for 8 generations). The E. affinis population used in this experiment was collected from the Gulf of Mexico prior to the Deepwater Horizon oil spill of 2010. Graphs show (a) percentage hatching, (b) percentage survival from hatching to metamorphosis, and (c) percentage survival from metamorphosis to adult across three oil concentrations (no oil control, 72.5% and 100% WAF, "wateraccommodated fraction"), for copepod lines before (blue, "pre-selection") versus after (orange, "post-selection") laboratory selection. Values shown are Mean ± SE for eight clutches, with 6-10 eggs per clutch at each treatment. Values and significance of differences between the laboratory lines are listed in Table 4a individuals that did manage to survive to adulthood showed accelerated development relative to the pre-selection line (Figure 5b).
The apparent evolutionary shifts in development time are consistent with the highly significant effects of Clutch (genotype) on development time (Table 2b). In addition, the significant random slope for Oil Concentration (with respect to Clutch, or Clutch × Oil Concentration;  (Figure 5b, orange vs. blue lines).

| Evolution of crude oil tolerance and life history in the wild
Our results revealed evidence for the evolution of crude oil tolerance in a population of the copepod Eurytemora affinis in the Gulf of  (Table 3b,c, Figure 1b,c). The post-oil spill population did show a trend of higher survival than the pre-oil spill population at the high oil concentration, but the difference was not significant (Table 3b,c). Additionally, development time at the high 100% WAF oil concentration was not significantly different between the populations (Table 3d,e). These results suggest that adaptation to the high oil concentration might be difficult to achieve. The high mortality suffered at 100% WAF in E. affinis was comparable to responses of other calanoid copepod species in response to WAF from the Deepwater Horizon Macondo Well crude oil (Almeda et al., 2013;Cohen et al., 2014).
Although, these other studies examined acute or chronic responses of adults or nauplii to WAF, rather than survival during development, and did not investigate evolutionary responses.
Given the evolutionary shifts in survival and development time, one would expect a significant effect of Clutch (genotype) on these traits, as genetic variation in crude oil tolerance in the population would be required for natural selection to act and evolutionary change to occur. We did find the effect of Clutch to be significant for survival from metamorphosis to adult (Table 1a), but not for hatching to metamorphosis, likely due to insufficient power (Table 1a) (Posavi, Larget, Gelembiuk, & Lee, 2014;Turelli & Barton, 2004).
Additionally, the diapause egg bank could serve as a repository of genotypes from past selection regimes and could contribute to the maintenance of genetic variation in the populations (Ban & Minoda, 1992;Hairston, 1996;Posavi et al., 2014). That is, the presence of a diapause egg bank coupled with temporarily varying selection could extend the conditions under which genetic variation is maintained in the wild population (through balancing selection) (Posavi et al., 2014;Turelli & Barton, 2004). This diapause egg bank is replenished each year when the population goes into seasonal diapause (Ban, 1992), and contains eggs that could remain viable for ~10 years (Katajisto, 1996). We do not know the extent to which the diapause egg bank contributes to the effective population size (N e ) of E. affinis populations, which is relatively large and estimated to be on the order of 10 6 -10 7 (Winkler et al., 2008). The contribution of genetic diversity from the diapause egg bank is potentially large, given that the input of recruits from the egg bank can be considerable. For instance, ~50,000 E. affinis nauplii hatch from the egg bank per m 3 per month in the Seine estuary (Glippa, Denis, Lesourd, & Souissi, 2014). Moreover, heritable phenotypic variation in the diapause egg bank has been found to differ from that of the water column population (Derry, Arnott, & Boag, 2010;Hairston, Kearns, & Ellner, 1996).

| Lower survival and rapid development following laboratory selection
The shifts in survival and development following laboratory selection (Figures 4, 5) differed markedly from the evolutionary changes experienced by the wild population following the oil spill (Figures 1,   2). Relative to the "pre-selection" line, the lower survival in the "postselection" line (Table 4b, (Lee et al., , 2007, as well as to other forms of stress, such as starvation and desiccation stress (Barrera & Medialdea, 1996;Chippindale et al., 1998;Harshman, Hoffmann, & Clark, 1999).
In contrast, the relatively few individuals that survived to metamorphosis in the post-selection line (only 13%, Table 4b) tended to have rapid development from metamorphosis to adulthood (Figure 5b). This pattern was consistent with the pattern of evolution of the development in the wild population, where development became accelerated following the oil spill (Table 3, Figure 2). This decrease in development rate was conspicuous (Figure 5b), even though the mean differences were not significant due to the small sample size arising from exceedingly high mortality after selection ( Figure 4b, Table 4b, N = 5 and N = 3). The effect of Clutch (genotype) was highly significant for development time (Table 2b) Table 2b), likely due to low power resulting from the small numbers of survivors (Table 4).
For laboratory selection, the period from hatching to metamorphosis appeared to be the critical life history stage, where survival was the lowest (Figure 4b, Table 4b). This was far more true in the laboratory selection lines than in the wild populations ( Figure 1b, Table 3b).
Interestingly, we found that once individuals survived to metamorphosis during laboratory selection, they tended to also survive to the adult stage (Table 4b vs. c). The high mortality from hatching to metamorphosis ( Figure 4b, Table 4b) suggests that selection acts most intensively during this life history stage. This pattern is similar to the case for selection on low salinity tolerance in E. affinis (Lee et al., , 2007, where selection acts predominantly on the life history stage prior to metamorphosis. Selection acting more intensely during the early life-history stages is commonly found in many different animal species (Garrido et al., 2015;Plough, Shin, & Hedgecock, 2016; Prasad, Shakarad, Anitha, Rajamani, & Joshi, 2001).

| Rapid evolution of crude oil tolerance in the wild and the challenges of inducing an evolutionary response in the lab
This study provided novel insights into the speed and process of evolutionary adaptation during recovery from a catastrophic oil spill. In particular, this study provided a rare examination of the evolution of tolerance and performance of an animal population in response to crude oil, by directly comparing populations at a given location before versus after an oil spill. Given the relatively rapid recovery of the population of E. affinis in the Gulf of Mexico following the Deepwater Horizon 2010 oil spill (2-3 years, ~12-20 generations), we had expected that crude oil tolerance would be relatively easy to induce by imposing selection in the laboratory. On the contrary, we found that ~8 generations of laboratory selection (Figures 4, 5) was insufficient to replicate the evolutionary patterns found in the wild population (Figures 1-3). Namely, we did not find evolution of increased survival in the post-selection line ( Figure 4, Table 4b,c). Although, given that significant evolutionary shifts did occur in development time (Figure 5; significant Population × Oil, Table 2b), it is possible that the post-selection line was in the process of responding to selection, but required longer periods of time for selection to increase survival in crude oil.
Alternatively, the difficulties of inducing a selection response for crude oil tolerance in the laboratory might have stemmed from insufficient standing genetic variation for the relevant traits in the laboratory. Although, the subsampling and measurement of clutch traits from the pre-oil spill population did reveal a considerable amount of genetic variation in crude oil tolerance, upon which selection could act ( Figure 3a). It is possible that even with the presence of significant genetic variation for crude oil tolerance in the pre-selection line (=pre-oil spill population, Figure 3, blue lines), the starting population size might have still been insufficiently large to capture the relevant alleles necessary for a selection response to crude oil in the laboratory. Unlike the laboratory selection lines, wild E. affinis populations receive recruits from a diapause egg bank, which is likely to harbor additional genetic diversity (see previous section) (Derry et al., 2010;Glippa et al., 2014;Hairston et al., 1996).
Our results suggest that recovery from an oil spill would be difficult and slow in absolute terms, even for a copepod that has short generation time (~15-20 days). We found that executing laboratory selection in response to crude oil was comparatively more difficult relative to the ease to which laboratory populations of E. affinis could respond to selection in response to salinity or temperature stress (Ketzner & Bradley, 1982;Lee et al., 2007Lee et al., , 2011. The selection experiments in this study used similar approaches as prior studies, with similar starting population size (hundreds of copepods). Difficulties in inducing an evolutionary response to crude oil was also found in populations of Trinidad guppies, where populations from oil-polluted tributaries showed little evidence of local adaptation in survival and growth (Rolshausen et al., 2015).
Perhaps selection on crude oil tolerance is more difficult to perform because of greater numbers of targets of selection than other stressors (such as salinity or temperature), given the large number of chemicals contained within crude oil (Almeda et al., 2013;Cohen et al., 2014). The copepod E. affinis does possess the cytochrome P450 of the CYP3027 family (Posavi, 2015), found to show increased expression in response to crude oil in other copepods (Han et al., 2014(Han et al., , 2015. It would be worth exploring the specific stressors contained within the complex cocktail of crude oil and the specific targets of selection required for crude oil tolerance to evolve, such as specific CYP detoxification enzymes and potentially genes involved in the aryl hydro-carbon receptor (AHR) signaling pathway (Reid et al., 2016;Whitehead, Pilcher, et al. 2012). Much work remains to be accomplished to dissect the specific physiological mechanisms involved in crude oil tolerance and how these mechanisms could rapidly evolve in order to survive catastrophic oil spills.