Experimental evolution of Chlamydomonas reinhardtii under salt stress
Chase Curtis Moser
Department of Biology
McGill University, Montreal
June 2010
A thesis submitted to McGill University in partial fulfilment of the requirements of the
degree of Master of Science.
© Chase Moser 2010
Library and Archives Bibliothèque et Canada Archives Canada
Published Heritage Direction du Branch Patrimoine de l’édition
395 Wellington Street 395, rue Wellington Ottawa ON K1A 0N4 Ottawa ON K1A 0N4 Canada Canada
Your file Votre référence ISBN: 978-0-494-72799-7 Our file Notre référence ISBN: 978-0-494-72799-7
NOTICE: AVIS:
The author has granted a non- L’auteur a accordé une licence non exclusive exclusive license allowing Library and permettant à la Bibliothèque et Archives Archives Canada to reproduce, Canada de reproduire, publier, archiver, publish, archive, preserve, conserve, sauvegarder, conserver, transmettre au public communicate to the public by par télécommunication ou par l’Internet, prêter, telecommunication or on the Internet, distribuer et vendre des thèses partout dans le loan, distribute and sell theses monde, à des fins commerciales ou autres, sur worldwide, for commercial or non- support microforme, papier, électronique et/ou commercial purposes, in microform, autres formats. paper, electronic and/or any other formats. . The author retains copyright L’auteur conserve la propriété du droit d’auteur ownership and moral rights in this et des droits moraux qui protège cette thèse. Ni thesis. Neither the thesis nor la thèse ni des extraits substantiels de celle-ci substantial extracts from it may be ne doivent être imprimés ou autrement printed or otherwise reproduced reproduits sans son autorisation. without the author’s permission.
In compliance with the Canadian Conformément à la loi canadienne sur la Privacy Act some supporting forms protection de la vie privée, quelques may have been removed from this formulaires secondaires ont été enlevés de thesis. cette thèse.
While these forms may be included Bien que ces formulaires aient inclus dans in the document page count, their la pagination, il n’y aura aucun contenu removal does not represent any loss manquant. of content from the thesis.
Contributions of Authors
The following thesis is made up of a review chapter and two manuscripts to be submitted for publication in peer-reviewed journals.
The experimental design in each chapter was provided by Dr. Graham Bell. I performed the experiments and the data analysis. Each manuscript was drafted in collaboration with
Dr. Graham Bell.
2 Abstract
The environment is now changing much faster than in recent geological time, causing increasing population extinctions. Experiments have shown that extinction can be avoided by adaptation through natural selection leading to evolutionary rescue. I first determined the response of Chlamydomonas to stressful environments by growing populations over a range of salinity. The population growth is halved at 5 g/L salt (NaCl), and 8 g/L is lethal. In this experiment, the genetic correlation between environments increases with environmental similarity. I then manipulated the genotypic diversity in experimental populations and cultured them by serial transfer at 5 g/L salt. The outcome of adaptation is not influenced by initial genetic variation. Instead, populations adapted mainly through the spread of new beneficial mutations. These results suggest that populations have a greater chance of adapting when new environments are similar to current conditions and that adaptation is sometimes dominated by the spread of new mutations, even in the presence of a substantial amount of standing genetic variation.
3 Résumé
Notre environnement change maintenant beaucoup plus rapidement que dans le passé
géologique récent, précipitant l’extinction de plus en plus d’espèces. Des chercheurs ont
démontré que, grâce à l’adaptation par la sélection naturelle, des espèces peuvent éviter
l’extinction, un processus nommé sauvetage évolutif. J’ai d’abord étudié la capacité de
Chlamydomonas à croitre dans des environnements dont la salinité augmente. J’ai trouvé que 5 g/L de sel diminue la croissance de moitié tandis que 8 g/L est suffisant pour empêcher toute croissance. Ici, la corrélation génétique entre environnement augmente avec la similarité des environnements comparés. J’ai ensuite soumis des populations contenant différentes quantités de diversité génétique initiale à une salinité de 5 g/L. La diversité génétique initiale ne semble pas influencer la capacité d’adaptation. Cependant, les populations semblent plutôt s’adapter en utilisant de nouvelles mutations dont l’effet est bénéfique. Ces résultats suggèrent que les populations s’adapteront plus facilement à des environnements similaires aux conditions présentes. De plus, ce processus sera dominé par la fixation de nouvelles mutations, même dans des populations contenant de la diversité génétique.
4 Acknowledgements
I was able to complete this thesis only with the help and support of some very fine people to whom I am grateful.
Foremost I owe a debt of gratitude my supervisor Dr. Graham Bell, for initially putting faith in me and having since provided me with guidance, encouragement, as well as patience at every stage of the project. He provided experimental designs, help with analysis, and draft edits along the way. I am very appreciative of his support and encouragement.
Kathy Tallon provided invaluable assistance in the lab in many ways including lending her technical knowledge and support. Thanks for taking care of all of us in the
lab and not letting our (my) messes get out of hand. Thank you also to Sonja and Anne-
Marie for their help, friendly faces, and positive attitudes.
My lab mates, Etienne and Pedram, have provided thought-provoking discussions, friendship, and emotional support over the last 2½ years. Etienne helped me during the writing process, draft edits, and abstract translation. I thank him for not letting me get away with anything and encouraging me to keep looking. Mark Jewell helped me with some demanding lab work for my experiment in Chapter 3; I thank him for his endurance and patience.
The Leung lab welcomed me into their office space for 2 months during the spring of
2010. They helped me with my troubles with the statistical package R and with portions of my statistical analysis. Special thanks to Dr. Brian Leung, Corey, Paul, Erin, and
Sylvia.
5 Michael Pedruski provided helpful comments and feedback on my literature review.
I owe my parents for my interest in the natural world and how it operates; thanks for making sure the apple didn’t fall far from the tree. Thank you Mom, Dad, Shasta, and
Scott, for your continuous love and support.
My group of friends who have put up with me in one way or another, discussing
Chlamydomonas or how to bring back dinosaurs, thanks for being there and being wonderful, especially the “Choose-Day” crew and Erica Strange. Most importantly, Katie, without your constant support, love, and faith in me, I wouldn’t be where I am.
6 Table of contents
Contributions of Authors 2
Abstract 3
Résumé 4
Acknowledgements 5
General Introduction 9
References 12
Chapter 1 Evolutionary response to deteriorating environments 15
References 32
Figures and legends 38
Linking statement between Chapters 1 and 2 41
Chapter 2 Genetic correlation in relation to differences in dosage of a stressor
43
References 55
Figure legends 57
Figures 58
7 Linking statement between Chapters 2 and 3 65
Chapter 3 The contribution of standing genetic variation and novel mutation to adaptation and evolutionary rescue in a simple laboratory system 67
References 86
Figure legends 89
Figures 91
General Conclusions and Summary 100
8 General Introduction
Rates of environmental change are increasing as human activities increase (Solomon,
2007) and as a result populations are responding to these changes to a greater extent than natural changes alone (Hendry et al., 2008). Severe environmental changes can cause extinctions, but in certain circumstances populations may avoid extinction through evolutionary processes (Bradshaw & McNeilly, 1991; Gomulkiewicz & Holt, 1995).
Evolutionary biology provides a framework to understand how populations adapt to severe change and thereby avoid extinction. There has been a recent call for evolutionary biologists to turn the focus of their studies toward evolutionary processes influencing extinction (Bell & Collins, 2008; Hendry et al., 2010).
The theory of evolutionary rescue involves a decline in abundance caused by a severe environmental stress resulting in the death or sterilization of most of the individuals, followed by a population recovery driven by the survival and reproduction of individuals that are able to withstand the stress (Gomulkiewicz & Holt, 1995). Recovery may occur even if there are only a few individuals which survive, although when populations are reduced to small population sizes there is an increased risk of extinction through stochastic processes (Lande, 1993).
The probability of evolutionary rescue depends on several factors: the strength and rate of the environmental change, the population size, and the availability of variation.
Genetic variation in the population can come from two main sources, standing variation that is already present and novel variation caused by beneficial mutations arising after the
9 stress has been applied. Novel mutation and standing genetic variation lead to somewhat different processes of adaptation (Barrett & Schluter, 2008).
Adaptation from standing variation is faster and may supply the population with alleles that have been tested previously in a similar environment. Standing variation is made up of mutations that have already spread through the population, either through selection or genetic drift, and which are present in more individuals when the environment changes than is the case for novel mutations (Barrett & Schluter, 2008).
Selection can act on standing variation immediately to produce adaptation, whereas some waiting time must elapse before novel mutations appear. Standing variation in the population may include mutations that have arisen during previous episodes of selection to a similar stress. Therefore, the alleles that produce tolerance to the environment may already exist within the population.
The work put forward in this thesis uses experimental evolution to investigate evolutionary rescue and how it is influenced by standing variation. Experimental evolution studies are able to provide direct tests of evolutionary theory and are an essential tool in evolutionary biology (Fuller et al., 2005).The benefit of using populations in experimental evolution studies is that they can be easily manipulated and highly replicated to test evolutionary theory directly.
There has been experimental work that explores the relationship between population size and evolutionary rescue (Bell & Gonzalez, 2009) and mutation supply and evolutionary rescue (Samani & Bell, 2010). There are no corresponding experiments, however, which investigate the influence of standing variation on evolutionary rescue.
Theory suggests that population size alone, excluding the influence of variation,
10 influences the chance of rescue because populations will have a greater chance of
extinction through stochasticity as their size is reduced (Lande, 1993). Bell and Gonzalez
(2009) explored the relationship between extinction and population size and found that
smaller stressed populations are more susceptible to extinction. Samani and Bell (2010)
found that there was a consistent relationship between population size and the extent of adaptation that extended over five orders of magnitude of population size. These are the only studies that directly demonstrate evolutionary rescue. Both use population size as a surrogate for the quantity of selectable variation generated by beneficial mutations.
I will take an experimental approach to explore the relationship between standing genetic variation and evolutionary rescue. I first document the lethal level of stress using
NaCl in the culture medium as a stressor and the chlorophyte Chlamydomonas reinhardtii as a model organism. In doing so, I determine the pattern of genetic correlation across a range of salt environments. I then use the established system in an evolutionary rescue experiment. I manipulate the amount of diversity in the population while keeping the initial population size constant in order to explore how the amount of genetic variation directly influences the chance of rescue.
11 References
Barrett, R. & Schluter, D. 2008. Adaptation from standing genetic variation. Trends in
Ecology & Evolution 23: 38-44.
Bell, G. & Collins, S. 2008. Adaptation, extinction and global change. Evolutionary
Applications 1: 3 - 16.
Bell, G. & Gonzalez, A. 2009. Evolutionary rescue can prevent extinction following
environmental change. Ecology letters 12: 942-948.
Bradshaw, A. & McNeilly, T. 1991. Evolutionary response to global climatic change.
Annals of Botany 67: 5 - 14.
Fuller, R., Baer, C. & Travis, J. 2005. How and when selection experiments might
actually be useful. Integrative and Comparative Biology 45: 391 - 404.
Gomulkiewicz, R. & Holt, R. 1995. When does evolution by natural selection prevent
extinction? Evolution 49: 201-207.
Hendry, A., Farrugia, T. & Kinnison, M. 2008. Human influences on rates of phenotypic
change in wild animal populations. Molecular ecology 17: 20-29.
Hendry, A.P., Lohmann, L.G., Conti, E., Cracraft, J., Crandall, K.A., Faith, D.P., Hauser,
C., Joly, C.A., Kogure, K., Larigauderie, A., Magallon, S., Moritz, C., Tillier, S.,
Zardoya, R., Prieur-Richard, A.H., Walther, B.A., Yahara, T. & Donoghue, M.J.
2010. Evolutionary Biology in Biodiversity Science, Conservation, and Policy: A
Call to Action. Evolution 64: 1517-1528.
Lande, R. 1993. Risks of population extinction from demographic and environmental
stochasticity and random catastrophes. American Naturalist 142: 911.
12 Samani, P. & Bell, G. 2010. Adaptation of experimental yeast populations to stressful
conditions in relation to population size. Journal of Evolutionary Biology 23: 791-
796.
Solomon, S. 2007. Climate Change 2007: the physical science basis: contribution of
Working Group I to the Fourth Assessment Report of the Intergovernmental Panel
on Climate Change. Cambridge University Press, Cambridge.
13 14 Chapter 1 Evolutionary response to deteriorating environments
Introduction
Individuals regularly encounter environmental variation in many ways. Moving
through time in seconds or centuries and through space in µm or km, environmental
changes occur across all scales (Pimm & Redfearn, 1988; Bell et al., 1993; Koscielny-
Bunde et al., 1998; Inchausti & Halley, 2002; all cited by Bell & Collins (2008)). Within
a lifetime and across generations, organisms constantly face change. Environmental
change almost always reduces the mean fitness of a population (Bell & Collins, 2008).
For populations that are maladapted to their conditions, environmental change may be
beneficial. However, we generally assume that a population is well suited to its
environment having adapted to conditions through natural selection (Rose & Lauder,
1996). In these cases, change will decrease a population’s fitness. In theory, when an
individual is perfectly adapted to its environment, any changes in its genetic code will be
deleterious, at best it will cause a slower rate of growth, at worst it will be lethal. I
consider fitness here as the growth of a type relative to the mean fitness of the population
(see Bell, 2008 pp 60-63).
It is clear that the scale and rate of change that populations will have to face are
increasing with human development (Solomon, 2007). As examples, the average surface
air temperature has increased by 0.74 °C in the last 100 years and is expected to increase
by ~ 1.1 – 6.4 °C over the next 90-100 years (Solomon, 2007), while atmospheric CO2 concentrations will change more in this century than in the past 300 million years
15 (Caldeira & Wickett, 2003). This makes the study of how populations cope with change one of the focal questions of our time.
When populations are faced with drastic environmental changes many of them will cause severe stress to populations and one of four things can happen, (i) Populations may deal with the stress through a plastic response, a non-heritable change in phenotype.
(ii) They may avoid the stress by modifying their distribution, avoiding the stress geographically or temporally. (iii) They may adapt to the stress, through heritable genetic changes. (iv) Finally, the population may fail to do any of the above and succumb to the stressful conditions and go extinct. The responses to environmental change are not mutually exclusive as more than one response may, and probably does, occur at a time.
Many organisms respond to minor changes in the environment through plastic responses that are not heritable. Phenotypic plasticity has been defined broadly as environmentally induced phenotypic variation (Stearns, 1989) and more recently as a genotype producing different phenotypes in response to environmental differences, as given in Ghalambor et al. (2007). Certain phenotypes can be induced developmentally such as the classic example of Daphnia pulex and its predator induced head spine. When
D. pulex develops in the presence of predators, it develops a large spine on its head but when grown in absence of predators, the head is smooth and round, without a spine
(Spitze, 1992). This is an example where there is potential for two genetically identical individuals to produce two different phenotypes, depending on their environment. Often a single genotype is capable of producing a range of phenotypes that maintains its fitness across a range of environments. When the environment changes to exceed the threshold beyond the range of the plastic response, the fitness of the genotype begins to decline.
16 The ability to supply short term changes in phenotype may shield individuals from the
effects of a varying environment. Plasticity may have short-term benefits but the overall effects of plasticity on adaptation, whether it accelerates it, retards it or has no effect, is
still unestablished (Price et al., 2003; Ghalambor et al., 2007).
When faced with a change in the environment, populations may change their
distribution to remain in the conditions where they are optimally adapted. The change in
distribution may come in the form of individuals physically travelling to a different area
in space or time, or for sessile organisms, such as plants, it may come through dispersal
of their offspring to a new region. Environmental change causing a shift in distribution
has been well documented in populations of insects (Parmesan, 2006) as well as in plants
(Davis & Shaw, 2001). If the environmental change is severe enough, it is less likely that
small populations that cover a narrow geographic distribution will survive through
migration, compared to larger populations covering a broader area (Pease et al., 1989).
Certain populations are unable to change their distribution, for example populations that
are confined to their habitats like isolated lakes, mountain slopes, coral reefs, etc., and
therefore must endure any changes that occur. Other populations are able to change their
distribution but face limits to how far they can go and are therefore subjected to
environmental changes (Sexton et al., 2009). In their review of responses to climate
change, Parmesan (2006) gives examples of populations that have had to endure change
and as a consequence, have become extinct locally due to climate shifts. Examples
include mountain slope populations of Edith’s checkerspot butterfly (E. editha) and pika
(O. princeps) populations, which have been driven to extinction at several locations.
17 Environmental change can be severe enough that it exceeds the range of
phenotypic plasticity of a population. Also, as discussed, some organisms are confined to
their range and must endure new conditions. If a population cannot avoid, through
changes in distribution, the stress that stands beyond their phenotypic capacities, it must
adapt to this new environment or perish (Bradshaw & McNeilly, 1991; Parmesan, 2006).
Plasticity and displacement may play a role in facilitating adaptation to new conditions. If
there is a sufficient supply of individuals that continue to survive and reproduce,
populations may recover and continue to grow.
This review focuses on the capacity of populations to adapt, which relies on the
combination of population characteristics and rate and magnitude of the change
encountered. First, I will discuss how the phenomenon of adaptation through natural
selection can be directly observed and can rescue a population from decline and
extinction. Second, I will discuss the importance of mutations in this process and the
factors that influence the availability of genetic variance for the process of adaptation.
Third, I will explore how the genetic variance made available through mutation is
sculpted to produce a viable population, covering factors influencing this process
including clonal interference and magnitude of selection. Finally I discuss the consequences for populations that undergo evolutionary rescue and where the direction of future research should be focused.
18 Observing adaptation
The way in which populations adapt to their surroundings was initially thought to be a slow and gradual process that is unobservable by most that studied the topic, as summarized by (Bell, 2010). The gradualist view of adaptation is that heritable changes in phenotype occur but these major changes happen so slowly, or that they are imperceptibly small, that they are unable to be seen in one’s lifetime, and this view dominated the field of evolutionary biology for roughly 100 after the theory of natural
selection was described. The emergence of field studies and experimental evolution in the
last 60 years has provided evidence of populations undergoing strong selection on short
time scales, often described as contemporary evolution eg. (Cain & Sheppard, 1954;
Endler, 1986; Kinnison & Hendry, 2001). It is now the view of some that populations are
capable of responding quickly to change with adaptation and that it happens in nature
frequently (Carroll et al., 2007).
Evolutionary rescue. The process of population recovery to a severe stress has been
termed “evolutionary rescue” by Gomulkiewicz and Holt (1995) and is the basis for a budding field of evolutionary biology that attempts to seek out how populations will adapt to strong environmental stresses. When a population becomes stressed by a sufficiently large environmental change, the result will be a decline in abundance. The population collapses as a large number of individuals that are adapted to previous state of the environment die or fail to reproduce. Following the considerable reduction in numbers, the population is left with those individuals with extreme phenotypes from pre-
19 existing mutations that or that have developed new mutations. During this period the population is at greater risk of extinction due to demographic stochasticity because of its small size. If the population does not go extinct during this period, the individuals with mutations that are suited to the new environment will reproduce and may eventually restore the population to it original magnitude. The process of evolutionary rescue proceeds through the population size decreasing, staying low, and then recovering, following a U-shaped curve as shown in Figure 1.
There has been a substantial body of work that explores the factors that influence evolutionary rescue through mathematical models and comparative studies. The major factors that influence the probability of extinction of a population encountering a severe environmental change are: the severity of the change, the amount of variation in the population, and the size of the population. The theory of how these factors influence adaptation in the face of extinction has been described and reviewed by many authors (for example Maynard Smith, 1989; Holt, 1990; Lande, 1993; Lynch & Lande, 1993;
Gomulkiewicz & Holt, 1995; Holt & Gomulkiewicz, 2004; Orr & Unckless, 2008). There has been little experimental work to date that demonstrates the potential of populations to adapt to a lethal stress via evolutionary rescue. Bell and Gonzalez (2009) have used experimental yeast populations to demonstrate evolutionary rescue. In order to view evolutionary rescue, they subjected populations with an initial size of ~105 individuals to
NaCl in the culture environment at severely stressful levels and measured population growth. They were able to measure population decline and recovery in response to environmental stress. Their measurements show the characteristic U-shaped curve, discussed previously, of initial population decline in response to the stress, then a
20 lowered density until the population begins to recover, and an increase in growth once more. This experiment demonstrates that evolutionary rescue in laboratory populations is consistent with theoretical predictions (Gomulkiewicz & Holt, 1995). Whether populations undergo evolutionary rescue is dependent on the availability of extreme types and their fixation in the population.
Influences on variation
Adaptation occurs through selection acting on genetic variation in a population. R.
A. Fisher’s fundamental theorem of natural selection states that “The rate of increase in fitness of any organism at any time is equal to its genetic variance in fitness at that time.”
(Fisher, 1930). The amount of adaptation that can occur therefore depends on the amount of genetic variation available in the population. Without genetic variance to act upon, there can be no selection and hence, no adaptation. Variation between the growth rates of individuals is introduced fundamentally through mutations. When DNA is replicated there are proofreading mechanisms that detect and fix copy errors but they are not 100% effective (Sadava et al., 2006). Mutations will enter the DNA of offspring because of this imprecision and generates heritable variation on which selection can act to produce adaptation.
If an individual were perfectly adapted to its environment, any mutation that occurs will result in a decrease of fitness. If its environment changes, however, the perfectly adapted individual will no longer be well adapted to the environment and now some mutations that occur in its offspring may have a beneficial effect. If most
21 populations are well adapted to their environment, mutations that occur within the coding
part of the genome are generally neutral or deleterious but occasionally beneficial.
Natural populations that encounter fluctuating environments require mutation for
adaptation. The amount of mutations that occur is kept in check by selection in what is
known as mutation-selection balance. If the supply of mutations is very high population
mean fitness will be reduced, whereas if it is very low there will be too few beneficial
mutations to support adaptation. The distribution of effects of beneficial mutations has
been estimated to fit an exponential distribution (Orr, 1998; Rozen et al., 2002; Wilke,
2004; Kassen & Bataillon, 2006; Eyre-Walker & Keightley, 2007), or possibly a gamma
distribution (Barrett et al., 2006). Kassen and Bataillon (2006) predict the distribution of
effects of beneficial mutations to be exponential and find that measured values fit the
model (Figure 2). Although there are an abundance of small effect mutations, Barrett et al.
(2006) find that the realized distribution of mutations is a Weibull distribution. They show that there are many small-effect mutations but few spread because they are lost by stochastic processes. Many intermediate mutations spread because of their higher selection coefficient. Few large-effect mutations appear but are more likely to spread because they of their large selection coefficients.
Rates of mutation. Mutation rates vary depending on the replicator or organism (Drake et al., 1998). It is important to understand rates of mutation because they determine how much variation is introduced into the population each replication. There are two main ways to describe mutation rates; mutations per site per replication (u) and mutations per genome per generation (U). The rate of mutations per site u is the rate at which errors
22 occur per site replicated and is referred to as the fundamental rate of mutation (Bell,
2008). The rate at which mutations appear per generation per genome U is the fundamental rate of mutation u multiplied by the genome size G. The mutation rate U is useful in describing the average number of mutations that occur in each offspring.
Without giving an exhaustive review on mutation rates, I will acknowledge that many attempts have been made to estimate mutation rates for several groups of organisms.
Drake et al. (1998) review the rates of mutation per generation per genome and find the following range of mutation rates: RNA based organisms U ≈ 0.1 – 2, DNA based microbes (eg. bacteria) U ≈ 0.0034, and in higher eukaryotes, such as fruit flies, mice, and humans, U ≈ 0.1 – 1. The rate of mutations per genome per generation that have a positive effect on fitness Ub is a small percentage of all mutations and has been measured to be somewhere between 10-9 and 10-4, as reviewed by Sniegowski and Gerrish (2010).
The rate at which mutations are fed into the system is given as a mutation supply rate per
site Nu which is equal to the fundamental rate of mutation u multiplied by the population
size N for a haploid population.
Population size and rescue. Population size dictates the probability of survival. Firstly, a
small population has a greater risk of extinction due to demographic stochasticity because
it will be driven to dangerously low numbers sooner than populations of larger size
encountering the same stress (Lynch & Lande, 1993). The risk of extinction increases
with stronger stresses and with smaller, maladapted populations (Lande, 1993). Secondly,
the supply of variation will be greater in larger populations. As evolutionary rescue
occurs, selection acts upon variation existing in the population or the appearance of new
23 mutations. The total number of mutations, as given by the mutation supply rate, increases
with population size. Thirdly, as the total number of mutations grows, so does the chance
that the mutation will be of larger effect, because the chance of picking a mutation of
large effect when drawing from an exponential distribution of beneficial fitness effects is
greater when there are more mutations are drawn (Figure 2). This does not mean that
small populations will never encounter mutations of large fitness effect, it just means that
there is a smaller probability that it will occur.
The relationship between population size and fitness has been explored
experimentally in simulation and in the laboratory. In silico experiments with digital
organisms by Elena et al. (2007) showed a positive relationship between population size
and average fitness of the population. Larger populations had individuals that, on average,
attained higher fitness. The experimental study of evolutionary rescue done by with lab
populations of yeast by Bell and Gonzalez (2009), as described above, also examined the
relationship between population size and rescue. They manipulated the population size
across a range of ~ 101 to 107 individuals in the starting population and subjected them to
severely stressful levels of NaCl. They looked at relationship between population size
and probability of survival in populations with environmental stress compared to no stress. They found that stressed populations, compared to non-stressed populations, were
at a greater risk of extinction at comparable initial population sizes. This study shows a
direct link between population size and the success of adaptation via evolutionary rescue.
The relationship between adaptation to lethal stress and population size was further
explored by Samani and Bell (2010). They experimentally manipulated yeast population
sizes in the laboratory and exposed them to a deteriorating environment. They found that
24 larger effective population sizes facilitated greater adaptation to the stress. Mutations had a greater chance of appearing and spreading in larger populations, thus increasing the population’s chance of success.
Influences on sorting
Simple sorting. Genotypes whose growth rate exceeds the average of the population will tend to increase in frequency over time. This process of sorting is fundamental to how populations adapt and is the mechanism of how evolutionary rescue occurs. It is therefore essential to understand how sorting occurs and the parameters that influence it in order to understand evolutionary rescue. I use an example with a two-type system to illustrate the simplest form of sorting. Two types which have equal growth rates but which differ genetically will remain in the same proportion so long as the system is not disturbed. If the environment changes so that one of the types has higher relative fitness the frequencies of the types will change. Over time the superior type will increase in frequency until it reaches fixation (relative frequency ~1 or 100%). The rate of increase of the superior type is given as the difference between its growth rate and the rate of growth of the weaker type and can be standardized by the average rate of growth of the population (selection coefficient, s).
Clonal interference. In an isolated population with two types, the lineage with a superior genotype will spread and become fixed in isolation. In a population with many types, there is a chance that the best type and the second-best type may have a very small
25 difference in growth, and therefore a small difference in selection coefficients, thus competing with one another and delaying fixation. The competition between superior types has been explored and modeled by Gerrish and Lenski (1998), who called this process “clonal interference”. The major consequence of clonal interference is that the fixation of the best type will be delayed. When related to population size, larger populations are likely to contain larger amounts of variation. They are therefore more likely to adapt to a severe environmental change but because of clonal interference, it may take longer for the best type to reach fixation (Sniegowski & Gerrish, 2010). Recall that as beneficial mutations appear in the population, their effects are drawn from an exponential distribution (Figure 2) and larger populations have a greater mutation supply rate. Therefore, larger populations have a greater chance of drawing more mutations that have closer fitness effects that compete for fixation and interfere with the sorting process.
Whether the rate of adaptation is influenced by clonal interference has been considered by Gerrish and Lenski (1998) and they conclude that ultimately the rate of increase in fitness will reach a limit as mutation rate and population size increase. Further investigation by Wilke et al. (2004) found that in large asexual populations the rate of adaptation will continue to increase despite clonal interference. Desai et al. (2007) consider the same problem but find that mutations occurring on backgrounds that already have mutations will circumvent the linear path of clonal interference and the rate of adaptation will continue to increase with population size. Ultimately the magnitude of adaptation depends on the range of variation introduced to the population. If most mutations are of large effect, the effect of selection should also be quite large in response.
De Visser and Rozen (2006) point out that clonal interference will cause mutations to fix
26 in order with large mutations fixing first and smaller mutations fixing after large mutations are stable in the population.
Standing variation and sorting. When a mutation spreads through the population, it leaves a signature in future generations (Barrett & Schluter, 2008). Over time, mutations
at many loci will appear, creating a population with individuals that are similar in many
ways but different at given loci. Variation from past mutations may stay in the population
and is referred to as standing genetic variation. Whether standing variation is removed
from the population is determined by the size of the population and the degree of
selection. Populations with a small effective population size, Ne, are prone to genetic drift which randomly causes some mutations in individuals to get passed on and others to perish. Random chance processes like genetic drift deplete the amount of standing variation in a population. When sorting occurs, it also reduces the amount of variation in the population. As stated above, when certain types are favoured and increase in relative frequency to the other types, inferior types are reduced in numbers and may be eliminated from the population. Directional selection causes a type of population bottleneck that
reduces the amount variation in a population. Population bottlenecks reduce the amount
of variation available for selection, as reviewed in Hall et al. (2010). Selection removes
genetic variation but it is renewed by the introduction of variation from mutations.
Whether most populations adapt through variation already present or from new mutations
is still an unresolved question (see De Visser & Rozen, 2005; Hermisson & Pennings,
2005; Orr, 2005; Barrett & Schluter, 2008; Orr & Unckless, 2008).
27 Selection acting on existing variation may act faster to produce adaptation than by acting on mutations for several reasons. When selection acts upon existing variation, it eliminates the waiting time for variation to appear, thereby beginning the passage time of the variant immediately (Barrett & Schluter, 2008). Standing variation may also facilitate faster adaptation because currently neutral alleles may be beneficial under induced stress and exist because they are a result of a previous encounter with the given stress (Barrett
& Schluter, 2008). Lastly, beneficial types in populations with standing genetic variation may have spread due to neutrality, demographic stochasticity or linkage, allowing selection to start with more than a single fit individual. In the model of periodic selection, populations that adapt via mutation go through two periods before adaptation is complete: they have to wait for a mutation to occur and if that mutation is beneficial they have to wait for it to spread in the population. The rate at which mutations occur in the population depends on the mutation supply rate Nu. If a beneficial mutation does occur, there is no guarantee that it will spread. By chance alone the individual carrying the mutation may not survive to reproduce through stochastic processes. If the individual does reproduce and the mutation does not confer a large enough benefit to the lineage, the lineage may go extinct by chance. Thus a population can experience a number of beneficial mutations but not have them spread. The probability that a given mutation will spread in a large population is roughly 2 times the selection coefficient of the given mutation (Haldane, 1927). Therefore, mutations that confer a greater fitness benefit have a greater probability of spread to those that do not. There have been more complex estimations of fixation rate of beneficial mutations as reviewed by Patwa and Whal (2008) that I will not discuss here.
28
Rate of environmental change and sorting. The way the environment changes has an impact on how sorting occurs and therefore how adaptation and evolutionary rescue occurs. The pace and direction of change influences how adaptation will occur. When the environment changes quickly, it results in populations that are less fit compared to the same populations where the environment changes more slowly. Collins and de Meaux
(2009) found that when selection pressures are applied more slowly populations adapt through more mutations of small effect compared to populations adapting under swift change that had few mutations of large effect. Samani and Bell (2010) describe the evolutionary rescue of populations to a changing environment as a “creeping barrage” of adaptation. Mutations that increase growth will spread and as the environment worsens and the outcome of adaptation depends on the correlation between mutations across environments. One serious consideration is that environments will not change in a single direction but will change in many ways simultaneously, introducing multiple stressors and thereby exposing populations to strong directional selection in different directions and from various pressures. When considering adaptation to fluctuating environments, it is necessary to consider not only how populations adapt to strong changes in the local environment, but what happens when the intensity and frequency of change in increased
(Lande, 2007; Bell, 2010). It is also necessary to understand how multiple stressors will act to diminish a population (Brook et al., 2008).
29 Conclusion
I have looked at the factors that influence the adaptation of populations to severe stress in the light of evolutionary rescue and given an overview of how populations adapt and the main factors that influence the probability of adaptation as conditions change. If populations do not respond swiftly enough, or if the given stress is too severe, the population will go extinct (Maynard Smith, 1989). The outcome of stronger than usual environmental changes may be extinction if the population does not have the proper demography or amount of variation to respond with adaptation. Pease et al. (1989) predict that under a severe stress, like a range shift, populations without sufficient variation would fail to adapt and face extinction. The harsh reality is that many populations have gone extinct, are going extinct, and will go extinct, because the severity of change will be too extreme.
One of the most important findings from the last 20 years related to evolutionary rescue is that adaptation can occur when populations are introduced to severe environmental change. There remain unsolved questions of the extent that natural populations will be affected by change and if they will be able to recover via adaptation before they are driven to extinction. We have seen experimental demonstrations of the limits of adaptation imposed by the rate of the stress (Collins & de Meaux, 2009) and population size (Bell & Gonzalez, 2009; Samani & Bell, 2010). In each of the cases there have been populations that adapt and populations that fail to adapt. Populations that are large, well adapted to their environment, and that contain a sufficient supply of standing variation (or a high beneficial mutation rate) are predicted to be more likely to survive
30 severe stress through evolutionary rescue. Orr and Unckless (2008) emphasize the peril
that population reduction causes. As population sizes are reduced, so is the pool of
existing variation and the chances for new variation to appear. Certain interconnected
parameters such as population size, standing genetic variation, and mutation supply rate have positive correlations with the probability of survival via evolutionary rescue (Bell &
Gonzalez, 2009; Samani & Bell, 2010). What remains uncertain is what will happen over time in natural populations when adaptation tests the limits of these population parameters by reducing population size and depleting variation. It also remains unresolved how the sources of variation, standing variation or mutation, will influence the adaptability of the population. In general we can begin to add to the body of
experimental work to get a clearer picture of the effects of large environmental change on
populations. In order to determine the effects of adaptation to strong environmental change, it is important to explore experimentally the roles of standing genetic variation and mutation as sources of variation.
Although it seems an overwhelming task to predict how populations will grow and shrink in the near future, it is necessary in order to understand the conditions which present risk and those that do not. Determining the limits to which populations can adapt is essential in knowing where conservationists should focus efforts. If populations are going to survive abrupt and severe changes to their environment, either through the above mentioned processes, migration, plasticity, or adaptation, uncovering the limits to these
processes should be a high priority for research for ecologists and evolutionary biologists
(Bell & Collins, 2008; Hendry et al., 2008; Hendry et al., 2010).
31 References
Barrett, R. & Schluter, D. 2008. Adaptation from standing genetic variation. Trends in
Ecology & Evolution 23: 38-44.
Barrett, R.D.H., MacLean, R.C. & Bell, G. 2006. Mutations of intermediate effect are
responsible for adaptation in evolving Pseudomonas fluorescens populations.
Biology letters 2: 236 - 238.
Bell, G., Lechowicz, M., Appenzeller, A., Chandler, M., DeBlois, E., Jackson, L.,
Mackenzie, B., Preziosi, R., Schallenberg, M. & Tinker, N. 1993. The spatial
structure of the physical environment. Oecologia 96: 114-121.
Bell, G. 2008. Selection: the mechanism of evolution. Oxford University Press, USA,
New York.
Bell, G. & Collins, S. 2008. Adaptation, extinction and global change. Evolutionary
Applications 1: 3 - 16.
Bell, G. & Gonzalez, A. 2009. Evolutionary rescue can prevent extinction following
environmental change. Ecology letters 12: 942-948.
Bell, G. 2010. Fluctuating selection: the perpetual renewal of adaptation in variable
environments. Philosophical transactions of the Royal Society of London. Series
B, Biological sciences 365: 87 - 97.
Bradshaw, A. & McNeilly, T. 1991. Evolutionary response to global climatic change.
Annals of Botany 67: 5 - 14.
Brook, B., Sodhi, N. & Bradshaw, C. 2008. Synergies among extinction drivers under
global change. Trends in Ecology & Evolution 23: 453-460.
Cain, A. & Sheppard, P. 1954. Natural selection in Cepaea. Genetics 39: 89 - 116.
32 Caldeira, K. & Wickett, M. 2003. Anthropogenic carbon and ocean pH. Nature 425: 365.
Carroll, S., Hendry, A., Reznick, D. & Fox, C. 2007. Evolution on ecological time-scales.
Ecology 21: 387-393.
Collins, S. & de Meaux, J. 2009. Adaptation to different rates of environmental change in
Chlamydomonas. Evolution 63: 2952-2965.
Davis, M. & Shaw, R. 2001. Range shifts and adaptive responses to Quaternary climate
change. Science 292: 673 - 679.
De Visser, J. & Rozen, D. 2005. Limits to adaptation in asexual populations. Journal of
Evolutionary Biology 18: 779-788.
De Visser, J. & Rozen, D. 2006. Clonal interference and the periodic selection of new
beneficial mutations in Escherichia coli. Genetics 172: 2093 - 2100.
Desai, M., Fisher, D. & Murray, A. 2007. The speed of evolution and maintenance of
variation in asexual populations. Current Biology 17: 385-394.
Drake, J.W., Charlesworth, B., Charlesworth, D. & Crow, J.F. 1998. Rates of
spontaneous mutation. Genetics 148: 1667-1686.
Elena, S., Wilke, C., Ofria, C. & Lenski, R. 2007. Effects of population size and mutation
rate on the evolution of mutational robustness. Evolution 61: 666-674.
Endler, J. 1986. Natural selection in the wild. Princeton Univ Press, Princeton.
Eyre-Walker, A. & Keightley, P. 2007. The distribution of fitness effects of new
mutations. Nature Reviews Genetics 8: 610-618.
Fisher, R.A. 1930. The Genetical Theory of Natural Selection. Clarendon Press, Oxford.
Gerrish, P. & Lenski, R. 1998. The fate of competing beneficial mutations in an asexual
population. Genetica 102: 127-144.
33 Ghalambor, C., McKay, J., Carroll, S. & Reznick, D. 2007. Adaptive versus non-adaptive
phenotypic plasticity and the potential for contemporary adaptation in new
environments. Ecology 21: 394-407.
Gomulkiewicz, R. & Holt, R. 1995. When does evolution by natural selection prevent
extinction? Evolution 49: 201-207.
Haldane, J.B.S. 1927. A mathematical theory of natural and artificial selection, Part V:
Selection and mutation. Proceedings of the Cambridge Philosophical Society 23:
838-844.
Hall, A.R., Griffiths, V.F., MacLean, R.C. & Colegrave, N. 2010. Mutational
neighbourhood and mutation supply rate constrain adaptation in Pseudomonas
aeruginosa. Proceedings of the Royal Society B-Biological Sciences 277: 643-650.
Hendry, A., Farrugia, T. & Kinnison, M. 2008. Human influences on rates of phenotypic
change in wild animal populations. Molecular ecology 17: 20-29.
Hendry, A.P., Lohmann, L.G., Conti, E., Cracraft, J., Crandall, K.A., Faith, D.P., Hauser,
C., Joly, C.A., Kogure, K., Larigauderie, A., Magallon, S., Moritz, C., Tillier, S.,
Zardoya, R., Prieur-Richard, A.H., Walther, B.A., Yahara, T. & Donoghue, M.J.
2010. Evolutionary Biology in Biodiversity Science, Conservation, and Policy: A
Call to Action. Evolution 64: 1517-1528.
Hermisson, J. & Pennings, P. 2005. Soft sweeps: molecular population genetics of
adaptation from standing genetic variation. Genetics 169: 2335 - 2352.
Holt, R. 1990. The microevolutionary consequences of climate change. Trends in
Ecology & Evolution 5: 311-315.
34 Holt, R. & Gomulkiewicz, R. 2004. Conservation implications of niche conservatism and
evolution in heterogeneous environments. In: Evolutionary conservation biology
Ferrière R., Dieckmann, U., and Couvet, D., ed., pp. 244–264. Cambridge
University Press, Cambridge.
Inchausti, P. & Halley, J. 2002. The long-term temporal variability and spectral colour of
animal populations. Evolutionary Ecology Research 4: 1033-1048.
Kassen, R. & Bataillon, T. 2006. Distribution of fitness effects among beneficial
mutations before selection in experimental populations of bacteria. Nature
genetics 38: 484-488.
Kinnison, M. & Hendry, A. 2001. The pace of modern life II: from rates of contemporary
microevolution to pattern and process. Genetica 112: 145-164.
Koscielny-Bunde, E., Bunde, A., Havlin, S., Roman, H., Goldreich, Y. & Schellnhuber,
H. 1998. Indication of a universal persistence law governing atmospheric
variability. Physical Review Letters 81: 729-732.
Lande, R. 1993. Risks of population extinction from demographic and environmental
stochasticity and random catastrophes. American Naturalist 142: 911.
Lande, R. 2007. Expected relative fitness and the adaptive topography of fluctuating
selection. Evolution 61: 1835-1846.
Lynch, M. & Lande, R. 1993. Evolution and extinction in response to environmental
change. In: Biotic interactions and global change J. K. P. Kareiva, and R. Huey,
ed., pp. 234–250. Sinauer, Sunderland.
Maynard Smith, J. 1989. The causes of extinction. Philosophical Transactions of the
Royal Society B: Biological Sciences 325: 241-252.
35 Orr, H. & Unckless, R. 2008. Population extinction and the genetics of adaptation.
American Naturalist 172: 160-169.
Orr, H.A. 1998. The Population Genetics of Adaptation: The Distribution of Factors
Fixed during Adaptive Evolution. Evolution 52: 935-949.
Orr, H.A. 2005. Theories of adaptation: what they do and don't say. Genetica 123: 3-13.
Parmesan, C. 2006. Ecological and evolutionary responses to recent climate change.
Annual Review of Ecology, Evolution, and Systematics 37: 637-669.
Patwa, Z. & Wahl, L. 2008. The fixation probability of beneficial mutations. Journal of
The Royal Society Interface 5: 1279 - 1289.
Pease, C., Lande, R. & Bull, J. 1989. A model of population growth, dispersal and
evolution in a changing environment. Ecology 70: 1657-1664.
Pimm, S. & Redfearn, A. 1988. The variability of population densities. Nature 334: 613-
614.
Price, T., Qvarnström, A. & Irwin, D. 2003. The role of phenotypic plasticity in driving
genetic evolution. Proceedings of the Royal Society of London. Series B:
Biological Sciences 270: 1433 - 1440.
Rose, M. & Lauder, G. 1996. Post-spandrel adaptationism. In: Adaptation (M. Rose,
Lauder, GV, ed., pp. 1–8. Academic Press, London.
Rozen, D., de Visser, J. & Gerrish, P. 2002. Fitness effects of fixed beneficial mutations
in microbial populations. Current Biology 12: 1040-1045.
Sadava, D., Heller, H., Orians, G., Purves, W. & Hillis, D. 2006. Life: The science of
biology. WH Freeman & Co, Sunderland.
36 Samani, P. & Bell, G. 2010. Adaptation of experimental yeast populations to stressful
conditions in relation to population size. Journal of Evolutionary Biology 23: 791-
796.
Sexton, J.P., McIntyre, P.J., Angert, A.L. & Rice, K.J. 2009. Evolution and Ecology of
Species Range Limits. Annual Review of Ecology Evolution and Systematics 40:
415-436.
Sniegowski, P. & Gerrish, P. 2010. Beneficial mutations and the dynamics of adaptation
in asexual populations. Philosophical Transactions of the Royal Society B:
Biological Sciences 365: 1255 - 1263.
Solomon, S. 2007. Climate Change 2007: the physical science basis: contribution of
Working Group I to the Fourth Assessment Report of the Intergovernmental Panel
on Climate Change. Cambridge Univ Press, Cambridge.
Spitze, K. 1992. Predator-mediated plasticity of prey life history and morphology:
Chaoborus americanus predation on Daphnia pulex. American Naturalist 139:
229-247.
Stearns, S. 1989. The evolutionary significance of phenotypic plasticity. BioScience 39:
436-445.
Wilke, C. 2004. The speed of adaptation in large asexual populations. Genetics 167: 2045
- 2053.
37 Figures and Legends
Figure 1.
Population decline for a period of time followed by recovery due to resistant types surviving or appearing. Based on Gomulkiewicz and Holt (1995).
38
Figure 2. The predicted distribution of effects of beneficial mutations as given by an exponential distribution. Based on Kassen and Bataillon (2006)
39
40 Linking statement between Chapters 1 and 2
In order to study adaptation in the laboratory an experimental system must be established.
Previous work with Chlamydomonas makes it an ideal candidate for experimental evolution studies. I use this chapter to establish the effect of an imposed environmental stress on our model organism. There has been little work that establishes the effect of growth of NaCl on Chlamydomonas reinhardtii. By subjecting populations to a range of salt concentrations, we can gain a high resolution of the effect of stress. It also allows us to calculate genetic correlations between environments. By establishing a library of individuals we are also able to determine the genetic variation of the base population in the environmental stress. Through the fundamental exploratory work of the following chapter, we are able to develop a finely tuned experimental system that can be used to study evolutionary rescue.
41 42
Chapter 2
Genetic correlation in relation to differences in dosage of a stressor
Chase Moser
Graham Bell
Biology Department, McGill University, 1205 avenue Docteur Penfield, Montreal,
Quebec H3A 1B1, Canada.
Corresponding author: [email protected]
Tel: 1 (514) 398-6458
Fax: 1 (514) 398-5069
To be submitted to The Journal of Evolutionary Biology
43 Abstract
We exposed strains of Chlamydomonas reinhardtii isolated from an outbred laboratory population to a range of concentrations of salt (NaCl) up to an extirpative level that the base population could not tolerate. The genetic variance of yield increased with stress over the first half of this range before collapsing to nearly zero. The genetic correlation decreased with environmental distance, whether measured as a difference in dosage or as an environmental variance. This result is consistent with previous studies and provides a basis for interpreting adaptation to a deteriorating environment and the process of evolutionary rescue.
44 Introduction
This study is about how populations respond to stress. An individual exposed to a lethal
dose of a stressor will die, or will be irreversibly sterilized. The lethal dose of most
stressors will vary among individuals, some being more resistant than others. If the dose
applied is less than the lethal dose of the most resistant individuals the population is likely to survive if these individuals transmit a high level of resistance to their progeny. If the dose applied exceeds the lethal dose of the most resistant individual the population will be destroyed because none of its members are capable of surviving. If some
individuals are able to survive only by virtue of some condition that cannot be transmitted
to their offspring the population is likely to dwindle to extinction within a few
generations. Hence, the extirpative level of stress is defined in terms of the current range
of genetic variation with response to the stressor.
If an extirpative stress is applied abruptly, the population is necessarily doomed.
If the level of stress increases gradually over many generations, however, the population
may survive because types exceeding the most resistant type in the current population
may arise by beneficial mutation or recombination. They will not spread by virtue of
their supernormal resistance, where “supernormal” is defined as lying beyond the current
range of variation in the population, because they must become established before the
extirpative level of stress is reached, if the population is to survive. Rather, they spread
by virtue of their resistance to some lesser level of stress, which happens to be genetically
correlated with supernormal resistance. This principle can be broadly generalized: in a
deteriorating environment, adaptation is not attributable to direct selection at the current
45 level of stress, but rather to the indirect effect of prior adaptation. Samani and Bell (2010) called this the “creeping barrage” theory of adaptation to a deteriorating environment, and demonstrated experimentally how this process drives the evolution of supernormal resistance to salt stress in laboratory populations of wild yeast.
The basis of this account of adaptation is a notion of continuity: types that flourish in a given set of conditions will tend to flourish in conditions that are only slightly different. If this were not so, adaptation through natural selection could scarcely occur, because any trifling alteration of conditions would require the spread of a new set of arbitrarily different types. Hence, we should find that the genetic correlation of growth should be high when strains are tested in similar conditions, and more broadly that the genetic correlation should fall as conditions become more different. This difference can be expressed in units of the stressor, for example as the concentration of a toxin, although this cannot be used to compare the effects of different sources of stress. An alternative is to use the response of the organisms themselves as the measure of stress by calculating the environmental variance of growth for the two levels of stress for which the genetic correlation has been estimated; the genetic correlation is expected to decline as the environmental variance increases.
The genetic correlation between environments may predispose individuals to endure extreme levels of stress. The probability that a population will undergo evolutionary rescue depends on its size, the amount of genetic variation available and the severity of the stress (Gomulkiewicz & Holt, 1995). It has been well documented in natural populations that the amount of genetic variation in fitness related traits is
46 abundant (Mousseau & Roff, 1987; Roff & Mousseau, 1987; Houle et al.1996). If there is
a strong genetic correlation between environments, it will facilitate evolutionary rescue
through pre-adaptation to lower levels of the stress. To predict the likelihood of evolutionary rescue we need to know how the genetic correlation of growth varies with the difference between environments.
A general principle of transference has long been widely acknowledged and sometimes demonstrated experimentally, for example, (Jinks & Connolly, 1973; 1975) on adaptation to temperature in the basidiomycete Schizophyllum. The approach suggested
here was pioneered in an earlier study of the effects of mineral nutrients on
Chlamydomonas (Bell, 1992). It was extended by Kassen and Bell (2000) who showed in
the same system that genetic correlation fell both with the environmental variance and with the genetic distance between species of Chlamydomonas. Here, we report an experiment that was specifically designed to address the situation of a deteriorating environment by measuring the growth of genotypes isolated from a single outbred population of Chlamydomonas at many different concentrations of a stressor, NaCl, up to extirpative levels. The experimental evolution of this population when exposed to severe salt stress is described in another report (Moser & Bell, in preparation).
Methods
Base population. We used a population of the unicellular chlorophyte Chlamydomonas
reinhardtii established by Clifford Zeyl in 1992 by crossing the two wild-type strains CC-
47 1952 mt- and CC-2343 mt+ and then crossing their offspring with CC-253 nit1-305 mt-;
a nitroguanisine-derived mutant of the wild-type strain CC-1640 mt-. This population has
been maintained since 1997 in our laboratory through regular transfers and matings. We
isolated 60 spores from this population by picking colonies from thin spreads on agar as
material for estimating genetic variances and covariances. Growth in all our trials was
asexual.
Culture medium and growth conditions. All cultures were grown in 150 μL of Bold’s
minimal medium (Harris, 1989) in 96-well microlitre plates. The medium was supplemented with varying concentrations of NaCl (“salt”) (0 - 8 gL-1), added to the
liquid medium as a solid before sterilization. In previous trials we found a difference in
growth of populations in the outer wells of the culture plate so we grew cultures in all 96
wells but used only the inner 60 wells for the experimental cultures. Cultures were
maintained at room temperature (22 - 24 °C), under constant fluorescent light (100 μmol
m-2 s-1 ). Cultures were mixed once during the growth cycle by pumping 10 μL of culture
10 times using a micropipettor. Measurements were taken with a Bio-Tek synergy HT plate reader using absorbance at 665 nm to estimate optical density. At the end of the
growth cycle; hence, our measure of growth is yield. All transfers and manipulations
were done under a laminar flow hood so that cultures remained axenic for the duration of
the experiment.
Effect of salt on growth. We grew populations of Chlamydomonas in a range of salt
concentrations from 0 to 8 gL-1 in 0.5 g increments to determine its effect on growth. To
48 begin each trial, we inoculated fresh sterile unsupplemented medium with 10 μL of
grown culture (~104 individuals) and incubated this population for two growth cycles
without treatment before it was transferred to salt-supplemented medium. Populations were transferred after 4 days’ growth, with 2 growth cycles per trial. We conducted two
trials. The first established the response of the base population to salt stress. We grew 60
replicates at each level of salt concentration. The second used the 60 isolate strains to
estimate genetic effects. We grew two replicates of a plate containing all 60 isolates at
each level of salt concentration.
Statistical analysis. We used non-linear least-squares regression to describe how yield is
affected by salt. We estimated the genetic variance VarG of the population at each salt
level from a single-factor ANOVA as VarG = (MS groups – MS error)/n, where the
number of replicate observations n = 2. The genetic covariance for yield Y at the two salt
levels i and j, CovG(Yi,Yj), was calculated by conducting a single-factor ANOVA for the
summed yield Yi + Yj and then using the relationship Var (Yi + Yj) = Var(Yi) + Var(Yj)
+ 2 Cov(Yi,Yj). The correlation calculated from this covariance is then a ratio of the pure
genetic covariances and variances (see Kempthorne (1957)). All statistical analysis was
performed in R version 2.9.1 (R Development Core Team, 2009).
Results
Effect of salt on growth. Salt consistently reduced the overall yield of the base
population and the mean of the strains isolated from it (Figures 1A and 1B). The salt
49 concentration that reduces yield by 50% was estimated to be 4.74 (se 0.04) g L-1 for the
base population and 5.68 (se 0.09) g L-1 for the average of the strains. This is consistent
with the study by Reynoso and de Gamboa (1982) who found that 0.085 M NaCl (4.96 g
L-1) reduces growth in Chlamydomonas by 48%. Growth was almost completely
suppressed at concentrations of 8 g L-1 or higher.
Genetic variance of the population. Genetic variance increases from low levels in unsupplemented medium to a maximum at about 5 g L-1, the concentration at which yield
is halved (Figure 2). Over this range, the coefficient of variation increases linearly with
salt concentration (CV = 0.094 salt g L-1 + 0.060, r2 = 0.97). For higher concentrations
this breaks down; the genetic variance declines and is nearly zero at the lethal level of 8 g
L-1.
Genetic correlation between environments. We estimated the genetic correlation for
all pairwise combinations of treatment levels (0.5g L-1 increments of salt) for a total of
289 comparisons. The correlation declined with the environmental distance between
treatments in terms of the difference in salt concentration (Figure 3A). The difference
can be expressed as a response by the environmental variance of yield, VarE = ½ (Yi –
2 Yj) , which enables studies using different treatments to be directly compared. The
genetic correlation between treatment levels declines linearly with the environmental
standard deviation (Figure 3B).
50 Genetic correlation with extreme stress. For any given level of stress the genetic
correlation can be estimated for levels which differ by any amount (Figure 4). At 5g L-1 the correlation is nearly zero for much lower levels (4 - 5 g L-1 less) and increases at
higher levels to reach values of nearly +1 at a very similar level (0.5 g L-1 less). A similar pattern was found for 6 g L-1. At 7 g L-1 the pattern was weaker, with the
correlation at 0.5 g L-1 less being markedly less than +1. At extreme levels of stress, 8 g
L-1, the pattern breaks down completely, with no tendency for the correlation to rise as this level is approached, or even to become consistently positive. This figure is necessary
to show how the genetic correlation between growth at different levels of stress is
sufficient to explain adaptation to concentrations of up to 7 g L-1 NaCl as the
consequence of prior adaptation to lower concentrations but that this mechanism fails to
cause prior adaptation to 8 g L-1.
Discussion
Salt is a severe stress for Chlamydomonas reinhardtii: yield is halved at 5 g L-1, and a
dose of 8 g L-1 or more almost completely prevents growth. Genetic variation is low when
little or no salt is added to the medium, representing the normal chemical conditions of
growth that the base population has experienced for the past few thousand generations. It
is not zero: persistent genetic variation in this population has previously been attributed
to the recombination of genotypes expressing similar values of components of fitness (see
(Bell, 2005)). Variation increases with stress up to a concentration of about 5 g L-1 NaCl.
This reflects the lack of prior selection at these levels of salt and the consequent
51 accumulation of conditionally neutral variation. Over this range, genetic variation as expressed by the coefficient of variation increases linearly with the level of stress.
Finally, variation collapses under more extreme stress, simply because most or all strains are barely able to grow at all.
The genetic correlation of yield between media declines with the dosage of salt. This extends a previous result for fertilization (Bell, 1992) to stress. In both cases, the correlation decreases linearly with the environmental standard deviation, such that it falls to zero for the most extreme comparisons, whether of starvation with nutrient repletion, or of benign with toxic conditions. This result has two broad implications.
The first is that it provides a basis for adaptation to a deteriorating environment that may enable populations to adapt eventually to levels of stress they cannot currently tolerate. It explains how evolutionary rescue can occur, even when contemporary studies show that a population is doomed to extinction. It does not show that rescue is inevitable, of course. Based on contemporary variation the base population of this study will readily adapt to concentrations of up to 7 g L-1 but not to 8 g L-1. Any adaptation to extirpative levels will depend on the generation and harvesting of new genetic variation, through beneficial mutation or recombination, which must be investigated through a selection experiment. It does show, however, that if adaptation to intermediate levels of stress does occur in such an experiment it is likely that it will have the indirect effect of producing adaptation to levels that are lethal for the ancestor. Whether or not this outcome is realized will depend on the rapidity with which the environment deteriorates
52 and on the factors such as population size and recombination rate that govern the
generation of new variation.
The second is that over by far the greater part of the range of conditions in which
growth can be sustained the genetic correlation is on average positive. It is only when
optimal and nearly lethal conditions are compared that genetic correlations tend to become negative. It is often argued that genetic diversity is maintained in populations
because the direction of selection varies in time or space. This requires that the
correlation between allelic effects in different conditions is negative. This may well be
the case in particular circumstances, but this and similar studies give little support for the
conclusion that genetic correlation is often negative for environmental differences that lie
well within the range of tolerance of the population. This may be misleading; there are
many correlations that are individually negative, for example, and our study was not
designed with appropriate replication to evaluate whether they are consistently negative.
It might also be objected that the very simple manipulation of conditions in the laboratory
may not adequately represent the interplay of multifarious factors in natural populations.
Moreover, when a population is propagated over many generations natural selection
acting on two fitness components with zero correlation among arbitrary strains will tend
to generate negative correlation through the elimination of strains with low values of both.
Nevertheless, it would seem worthwhile to investigate more systematically the kinds of
environmental difference that lead to negative genetic correlations.
53 Acknowledgments. This research was funded by a Discovery Grant to GB from the
Natural Sciences and Engineering Research Council of Canada.
54 References
Bell, G. 1992. The ecology and genetics of fitness in Chlamydomonas. V. The
relationship between genetic correlation and environmental variance. Evolution
46: 561-566.
Bell, G. 2005. Experimental sexual selection in Chlamydomonas. Journal of Evolutionary
Biology 18: 722-734.
Gomulkiewicz, R. & Holt, R. 1995. When does evolution by natural selection prevent
extinction? Evolution 49: 201-207.
Harris, E. 1989. The Chlamydomonas Sourcebook. Academic Press San Diego.
Houle, D., Morikawa, B. & Lynch, M. 1996. Comparing mutational variabilities.
Genetics 143: 1467-1483.
Jinks, J. & Connolly, V. 1973. Selection for specific and general response to
environmental differences. Heredity 30: 33-40.
Jinks, J. & Connolly, V. 1975. Determination of the environmental sensitivity of
selection lines by the selection environment. Heredity 34: 401-406.
Kassen, R. & Bell, G. 2000. The ecology and genetics of fitness in Chlamydomonas. X.
The relationship between genetic correlation and genetic distance. Evolution 54:
425-432.
Kempthorne, O. 1957. An introduction to genetic statistics. Iowa State University Press
New York.
Mousseau, T.A. & Roff, D.A. 1987. Natural Selection and the Heritability of Fitness
Components. Heredity 59: 181-197
55 R Development Core Team 2009. R: A language and environment for statistical
computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-
900051-07-0, URL http://www.R-project.org.
Reynoso, G. & de Gamboa, B. 1982. Salt tolerance in the freshwater algae
Chlamydomonas reinhardii: Effect of proline and taurine. Comparative
Biochemistry and Physiology Part A: Physiology 73: 95-99.
Roff, D.A. & Mousseau, T.A. 1987. Quantitative Genetics and Fitness – Lessons From
Drosophila. Heredity 58: 103-118.
Samani, P. & Bell, G. 2010. Adaptation of experimental yeast populations to stressful
conditions in relation to population size. Journal of Evolutionary Biology 23: 791-
796.
56 Figure Legends
Figure 1. The effect of salt supplementation on yield. Bars are 95% confidence intervals
of the mean.
A. Overall response of the base population. [regression statistics]
B. Average response of the 60 strains isolated from the base population
[regression statistics]
Figure 2. The effect of salt supplementation on the genetic variance of yield.
Figure 3. The response of the genetic correlation of yield to the difference in conditions
of growth.
A. The genetic correlation in relation to difference in dosage. The regression is y
= 0.82 – 0.13x, with r2 = 0.34. No formal significance is cited because the
points are not mutually independent.
B. The genetic correlation in relation to the environmental standard deviation.
The regression is y = 0.72 – 4.12 x, with r2 = 0.23.
Figure 4. The increase of genetic correlation in relation to the difference from a severe
level of stress. This shows the correlation between a given salt concentration and one of
the four concentrations chosen to exemplify severe stress as a function of the difference in salt concentration in increments of 0.5 g L-1. As the differences in salt concentration
become smaller, the genetic correlation increases. This holds for all severe levels of stress
up to 8 L-1, where the pattern breaks down.
57
Figure 1A.
58
Figure 1B.
59
Figure 2.
60
Figure 3A.
61
Figure 3B.
62
1.25 5 g/L 6 g/L 1 7 g/L 8 g/L 0.75
0.5
0.25
0
-0.25
genetic correlation with severe stress severe with correlation genetic -0.5
-0.75 -5 -4 -3 -2 -1 0 salt concentration relative to severe stress g/L
Figure 4
63
64 Linking statement between Chapters 2 and 3
The previous chapter focuses on establishing the effect of NaCl on growth in
experimental populations of Chlamydomonas. Certain essential features of the system were determined such as the concentration of salt that reduces growth by half and the genetic variance of the base population. We use the established system from Chapter 2 to explore the influence of standing genetic variation on the potential of evolutionary rescue.
Theory suggests that populations with greater amounts of standing genetic variation should have a greater chance of evolutionary rescue. In the following chapter, we manipulate the amount of genotypic diversity in a population to determine the effect of standing variation on rescue. We use Chlamydomonas that have been isolated from the same base population as in Chapter 2 and apply NaCl as an environmental stress at the concentration that halves growth.
65 66
Chapter 3
The contribution of standing genetic variation and novel mutation to
adaptation and evolutionary rescue in a simple laboratory system
Chase Moser
Graham Bell
Biology Department, McGill University, 1205 avenue Docteur Penfield, Montreal,
Quebec H3A 1B1, Canada.
Corresponding author: [email protected]
Tel: 1 (514) 398-6458
Fax: 1 (514) 398-5069
To be submitted to The Journal of Evolutionary Biology
67 Abstract
We constructed mixtures of strains isolated from a variable outbred laboratory population of Chlamydomonas to evaluate the relative contributions to adaptation to stressful conditions of growth made by the sorting of standing genetic variation and by the substitution of novel beneficial mutations. Our experimental selection lines adapted to high salt concentration in the course of 60-65 generations. We found that mixture diversity had no effect on the outcome of selection, single strains adapted as swiftly and as fully as mixtures, and genetic variance was not consistently depleted in mixtures. On inspection we found that lines were frequently swept by beneficial mutations of large effect, resulting in a doubling of fitness. Our results do not impeach the role of sorting in small sexual populations, but they emphasize the role that novel beneficial mutations can play in rescuing populations exposed to severe and potentially lethal stress.
68 Introduction
When severe stress is applied to a population it will decline in abundance and may
eventually become extinct. If there are some types capable of surviving the stress, however, they will spread through natural selection and may restore, in whole or in part,
the original abundance and mean fitness of the population (Gomulkiewicz & Holt, 1995).
This process of evolutionary rescue depends on the range of genetic variation available
during the period of decline. Some of this will be contributed by alleles that are already
segregating in the population when the stress was applied; another part will arise
subsequently through novel beneficial mutations. The balance between standing genetic
variation and novel mutation is an important aspect of adaptation that has seldom been
systematically investigated.
Mutations are the ultimate source of genetic variation and hence of adaptation.
Evolutionary experiments using microbes routinely begin with isogenic cultures that
proceed to adapt as novel beneficial mutations arise and spread. The process is
dominated by three quantities: population size, the mutation rate to beneficial alleles and
the effect of a mutation on fitness. Population size and mutation rate together govern the
mutation supply rate, while the probability of fixation is largely determined by mutational
effect (De Visser & Rozen, 2005). Beneficial mutations are usually thought to be weak
and rare, but microcosm experiments suggest that they are often strong and common
(Barrett et al., 2006; Desai et al., 2007; Perfeito et al., 2007). Large-effect mutations are
likely to be responsible for adaptation because small-effect mutations, while much more
69 abundant, are usually lost by sampling error when still rare. A high rate of beneficial
mutation for some given population size would suggest that the mutational target – the
number of sites in the genome where a beneficial mutation can occur – is much larger
than expected.
Outside the laboratory, most populations will be a mixture of many genotypes. The abundance of genetic variation in life history traits has been well documented in natural populations (Mousseau & Roff, 1987; Roff & Mousseau, 1987; Houle et al.1996). In
asexual populations, quasi-neutral alleles may fluctuate for long periods of time, and even
deleterious mutations will often persist for many generations. Moreover, frequency-
dependent selection arising from nutrient competition, co-evolution or other processes
may maintain alleles at stable high frequencies. Consequently, large asexual populations
are likely to be far from uniform, but rather to consist of a multitude of genotypes
responding to drift, mutation and balancing selection. In sexual populations the situation
is more complex because combinations of alleles can be built up or broken down by
recombination, so that the potential range of variation may vastly exceed the range
expressed at any particular time. This standing genetic variation, actual and potential,
responds immediately to selection because many beneficial alleles are already abundant
enough to be safe from imminent stochastic extinction (Barrett & Schluter, 2008).
Hence, it seems reasonable to conclude that most instances of rapid adaptation in natural
populations will owe much more to standing genetic variation that they will to novel
mutation. In some classic cases, such as shell colour and banding in snails (Cain &
70 Sheppard, 1954) or heavy metal tolerance in plants (Bradshaw, 1991), this is evidently true. In others, such as armour reduction in freshwater sticklebacks (Colosimo et al.,
2005) and albinism in beach mice (Hoekstra et al., 2006), there is convincing indirect evidence. However, there seem to be few, if any, attempts to gauge the relative contributions of standing variation and novel mutation under controlled conditions.
We have previously used a yeast-salt system to investigate the effect of population size on adaptation and rescue in genetically depauperate populations. In this report we use
Chlamydomonas, because we have maintained large outbred populations in our laboratory for many years, and these serve as a supply of genetic variation from populations close to evolutionary equilibrium under laboratory conditions of culture. The basic design of the experiment is to set up mixtures of known genotypic diversity by sampling from a library of isolates without replacement. We begin by isolating haploid spores at random and propagating each clonally. These monoclonal cultures are labeled
A, B, C … We then form pairwise mixtures in which each strain is represented once and once only: AB, CD, EF … These are then used to form 4-wise mixtures, again with each strain represented once and once only: ABCD, EFGH, IJKL … Having measured the growth of each clone, we can then calculate the genetic variance and the genetic range of any mixture. The expected result of selection by simple sorting in any given mixture is that all genotypes except one, that with the highest rate of growth, will be eliminated. All strains are sufficiently abundant in any mixture that stochastic elimination is extremely unlikely to occur. It follows necessarily from the design that growth rate after selection plotted on mixture diversity will form a triangular graph with a fixed horizontal upper
71 limit (the growth of the best strain overall) and a steadily increasing lower bound
(because a larger number of inferior genotypes are eliminated from more complex
mixtures). It therefore also follows that mean growth will increase with mixture diversity.
Finally, the design ensures that a mixture of any degree of complexity (beyond a single
strain) can be compared with two other mixtures, at the level immediately below, which
comprise precisely the same set of strains. Hence, there are many independent opportunities to confirm that a given quantity of standing genetic variation will lead to a predictably higher level of adaptation.
Nevertheless, there is nothing in this design that prevents novel beneficial mutations from
occurring. Indeed, in the monoclonal cultures they will be the only source of variation.
In successively more complex mixtures they will make a successively less important
contribution to adaptation. This is because they need to be larger in order to spread. In a
monoclonal culture any beneficial mutation, however slight its effect, will tend to spread
(although those of larger effect are more likely to do so). In a mixture of two strains, this
remains true for the superior strain, but a mutation occurring in the inferior strain will
only tend to spread if its effect exceeds the difference between superior and inferior
strains. In more complex mixtures the condition for a novel mutation to spread becomes
necessarily more onerous. As a simple illustration, consider a series of strains with
fitnesses wi in which the top-ranked strain has fitness w1. The rate of beneficial mutations ui with effect si = w1 – wi is likely to decline exponentially with si (Kassen & Bataillon,
2006), such that ui = u0 exp(- ksi), where u0 is the rate of beneficial mutations with
minimal effect on fitness and k is the slope of the exponential decline. The probability
72 that a beneficial mutation will occur in a strain such that the fitness of the mutant exceeds
that of the top-ranked strain is the area under the curve to the right of si, which is ui/k.
The overall rate of beneficial mutation uben in a mixture of n equally frequent strains is
the probability that a beneficial mutation will occur in the top-ranked strain, plus the
probability that a mutation of sufficient magnitude will occur in any of the other strains, so that uben = (u0/nk) Σ exp(-ksi), which is a monotonically declining function of n for
values of k greater than zero. The probability that it will become fixed is weighted by Csi, where C is a small constant that depends on the model for reproduction (see Bell 2008, pp. 74-75).
Hence, the contribution made by selection favouring novel mutations, relative to the contribution made by the sorting of the genetic variation initially present, will decrease as mixtures become more complex and we refer to this phenomenon as ‘mutational damping’. This trend can be calibrated by observing the breakdown of the prediction made on the basis of sorting. This may not occur, even at the level of monoclonal cultures, if beneficial mutations are very rare, or it may occur at some higher level, if they are reasonably common. The outcome will enable us to evaluate the contributions of standing variation and mutation to adaptation and rescue in a simplified and controlled system.
73 Materials and Methods
Ancestor. Our experimental organism is Chlamydomonas reinhardtii, a unicellular
chlorophyte that is physiologically and genetically well known, and which has been used
in many evolutionary experiments. The base population descends from the cross (A x [B
x (B x C)]) made by Clifford Zeyl in October 1992, where A is CC-253 nit1-305 mt-, a
nitroguanisine-derived mutant of CC-1640 wild-type mt-, B is CC-1952 wild-type mt-
and C is CC-2343 wild-type mt+. All three are standard laboratory strains of C.
reinhardtii that are completely interfertile. Three replicate populations were set up in
1997 and have since been transferred by regular cycles of growth and mating. We
isolated 120 spores from one of these populations to serve as the library from which the
experimental mixtures were constructed. The lineage descending from a single spore is
called a “strain”.
Media and culture conditions. We grew cultures in Bold’s liquid medium (Harris, 1989)
under constant light (~ 100 μmol m-2 s-1) at room temperature (22 - 24 °C). We added
common salt (NaCl) to the medium as a stressor. A pilot experiment showed that about
half of our isolates showed greatly reduced growth at 5 g L-1 NaCl, and we chose this concentration for the experiment. Reynoso and de Gamboa (1982) found that 0.085 M
NaCl (4.96 g L-1) reduces growth in Chlamydomonas by 48%. Experimental strains and
mixtures were grown in 20 ml medium in glass culture tubes. We measured growth by
recording transmittance on a digital spectrophotometer at 665 nm and converting this to
74 cell density through a calibration curve that we had previously estimated from a dilution
series.
Selection experiment. We constructed mixtures of 2n strains from n = 0 to n = 6 by
randomly sampling a subset of 64 strains from the library without replacement, in such a
way that each strain was represented at each level exactly once, as described above. This
gave a total of 64 single strains, 32 mixtures of 2 strains, 16 mixtures of 4 strains, 8
mixtures of 8 strains, 4 mixtures of 16 strains, 2 mixtures of 32 strains, and a single
mixture of all 64 strains, for a total of 127 selection lines. Each line was replicated 4
times to form 4 independently randomized blocks grown on different light shelves, so
that the experiment comprised 508 lines in all. The strains were expanded in Bold’s medium, without NaCl, immediately before the selection experiment was set up. Each line was then initiated by inoculating a total of 0.5 ml from the expansion cultures. Thus,
the mixtures of 2 strains received 0.25 ml of each component strain, and so forth. The
ancestral strains attain an average of about 500,000 cells ml-1 after 7 days’ growth in
NaCl-supplemented medium, so that the initial inoculum was about 250,000 cells; hence,
the lowest initial abundance was about 4,000 cells for each strain in the 64-strain mixture.
The lines were transferred at intervals of 7 days by inoculating a fresh tube with 0.5 ml of
grown culture. The lines were propagated for 12 growth cycles, corresponding to about
60-65 generations. All open-tube manipulations were done under a laminar-flow hood
using sterile technique, and the cultures remained axenic until the end of the experiment.
75 The transmittance of each line was recorded at the end of each growth cycle, and transformed to cell density as an estimate of yield.
Assay of strains. When the selection experiment had been completed, we compared the ancestral with the derived single-strain cultures. The ancestral strains were first propagated for two growth cycles in experimental conditions, but without NaCl supplementation. Ancestral and derived strains were then propagated for 14 days in
NaCl-supplemented medium, with optical density measured daily to permit estimation of growth parameters. Four replicates of each ancestral and derived strain were grown in four independently randomized blocks on separate light shelves.
Assay of mixtures. The input genetic variation of mixtures can be calculated from the assay of the ancestral strains. The output genetic variation was estimated by isolating 16 spores from each of 4 randomly selected mixtures of 2, 4, 8 or 16 strains. Each of the 4 mixtures at each level of diversity was allocated to a different block, where each mixture was replicated twice. The blocks were independently randomized and growth recorded at daily intervals for 10 days.
Statistical analysis. Growth curves were fitted to a logistic model by non-linear least squares to yield estimates of the limiting rate of increase r and the limiting density K.
Genetic variances were estimated from single-factor analysis of variance. All calculations were performed in R version 2.9.1 (R Development Core Team, 2009).
76 Results
Genetic variation in the ancestor. Cell density after 7 days’ growth in NaCl-
supplemented medium varied by a factor of two among the strains from the ancestral
population. Mean density was 6.41 x 105 cells mL-1, and the among strain genetic
2 9 variance component was σ G = 2.8 x 10 (F = 1.4765, df = (63, 192), P = 0.02344). The
genetic variance component of the limiting rates of increase, for the among strain
2 -5 2 12 comparisons, (r) and density (K) were σ G = 4.25 x 10 and σ G = 1.687 x 10
respectively. The range of growth in each mixture forms the expected upper-triangular
graph (Figure 1).
Outcome of selection. All lines responded in a similar fashion to salt stress. Yield was
maintained for the first two growth cycles; crashed in the third cycle; recovered and improved over the next 5 cycles, eventually exceeding the ancestor; and finally remained constant for the last 4 growth cycles (Figure 2A). Hence, the lines followed a classical
U-shaped rescue curve (Gomulkiewicz & Holt, 1995; Bell & Gonzalez, 2009). The dose of 5 gL-1 salt that we applied is sufficient to extirpate the ancestral populations at the
given dilution rate, but nevertheless every line persisted without extinction.
All treatment levels (mixtures with different numbers of components) responded in the same fashion, and more complex mixtures do not consistently outperform simpler mixtures (Figure 2B). A repeated-measures analysis of variance with treatment and time
77 (growth cycle) as factors returned a significant effect of treatment (F = 8.1, df = (1, 6086),
P = 0.005), but this is not attributable to a consistent effect of diversity. The 64-strain mixture is often the best, but the 32-strain mixtures are often the worst, and the single strains are intermediate, suggesting no effect of diversity on growth. Moreover, there is no treatment x time interaction (F << 1), so mixture diversity does not affect the rate of improvement.
We compared the observed growth of each line at the end of the experiment with the growth predicted from the data in Figure 1 (Figure 3). There is no correspondence between the two. Instead, the observed growth always substantially exceeds the predicted growth.
Response of single strains. The assay of strains showed that the derived single-spore lines grew much faster than the corresponding ancestral lines (Figure 4A). The limiting rate of increase of the derived lines (mean = 0.513) is much greater than that of the ancestral lines (mean = 0.224) (F = 2278.6, df = (1, 439), P < 0.0001). Indeed, the derived and ancestral values do not overlap (Figure 4B). There is also a significant difference in the limiting density (F = 252.6, df = (1,439), P < 0.0001), but in this case the derived lines evolved limiting densities similar to but not exceeding the ancestral lines with the highest values.
78 Response of mixtures. Genetic variance was not uniformly lost from the mixtures, as it
would be expected to be lost through sorting. Instead, the output variance of many of the
mixtures that were assayed exceeded the input variance (Figure 5).
Effect of diversity on adaptation.
We designed the experiment in such a way so that we could compare the growth of any
one population at one level, (ABCD for example), to the two populations beneath it (AB
and CD). We took an average of fitness of each population over the last four transfer
cycles and used it to compare diversity levels. We found that only 44.4% of the more complex populations had higher fitness than the mean fitness of the two populations it was made of. Hence, there is no effect of diversity on the final outcome of adaptation.
We also looked at the effect of diversity during the period of increase in growth (transfer
weeks 3-8). We calculated the slope of yield across these weeks as a single value for each
population and performed a linear regression on the slopes across diversity levels. We
found that there is a weak positive correlation between the rate of increase in growth
(slope) and diversity level (Figure 6) (r2 = 0.044, p-value = 0.0177).
Rate of beneficial mutation. The superiority of the derived single-strain lines and the maintenance of variation in the mixtures are consistent with the spread of beneficial mutations. We inspected the growth trajectories from the assay of strains and found several instances in which one replicate of an ancestral line diverged strongly from the others and grew as rapidly as the corresponding treatment line. An example is shown in
Figure 7. We found that the distribution of the deviation of individual growth rates from
79 the mean of replicates for a given ancestral spore had a long right tail of abnormally high
values. Using the left tail to represent error, we identified abnormally high deviations
and confirmed by examination that these came from replicates which grew much faster
than expected. We found 12 such cases, which we interpret as beneficial mutations.
The number of replications occurring in the single growth cycle of the assay of lines is
the product of the inoculum size, the number of replications per cell and the number of
lines, which amounts to 1.3 x 108. Hence, the rate of beneficial mutations was 12 / (1.3 x
108) = 9 x 10-8 per replication. The mutations were of large effect. The selection coefficient is (r – r’)/r’, where r is the rate of increase of the mutant and r’ is the mean of the remaining non-mutant strains. The average selection coefficient was 0.88 (standard deviation 0.31), showing that these mutations almost doubled fitness on average. This is consistent with the improvement in the rate of increase evolving in the derived lines
(Figure 4A).
Discussion
Our experiment demonstrated that a genetically variable outbred population close to evolutionary equilibrium may succeed in adapting to an abrupt severe stress that halves its rate of growth. The limiting rate of increase of the ancestral lines in salt-supplemented medium was about 0.25 per day (Figure 4B), so they would increase by a factor of 7 or 8 at most over a single growth cycle, which is not sufficient to prevent them from being washed out by the modest rate of dilution (0.025) that we imposed. Nevertheless, all
80 lines were rescued by virtue of evolving rates of increase of around 0.5 per day, thereby increasing fast enough to reach the limiting density before transfer.
The simple sorting theory on which we based the experiment, however, fails to explain our results completely. As expected, we found an effect of diversity treatment on the rate of adaptation but our other results were unexpected. There was no effect of treatment level on the outcome of selection; genetic variation was not removed by sorting; the observed outcome of selection was unrelated to the predicted outcome. Adaptation was instead fuelled by beneficial mutations of large effect. All populations reached the same plateau in rate of growth following adaptation to salt, suggesting that there is some limit of adaptation. It is natural to conclude that the experiment has told us nothing about how natural populations are likely to respond to stresses of comparable magnitude. There are several reasons that this might be so, but none seem conclusive.
In the first place, our base population may have contained much less variation than most natural populations, so that it was unable to provide enough standing genetic variation for the experiment to work. The standardized genetic variance of fitness is the genetic variance divided by the mean squared, and by the Fundamental Theorem of Natural
Selection (Fisher, 1930) expresses the rate at which population mean fitness will increase
9 5 2 under selection. For our base population it was SVG = (2.8 x 10 )/(6.41 x 10 ) = 0.007
for yield after one growth cycle. The very few reliable estimates from natural
populations give comparable values. Seedlings of the annual herb Impatiens pallida
produced from controlled crosses and planted back into their parental environment gave
81 an estimate of SVG = 0.03 (Bell et al., 1991). A similar study of Impatiens gave SVG =
0.008 in a permissive floodplain site and 0.06 on a physically more stressful hillside
(Bennington & McGraw, 1995). The long-term survey of a small passerine bird, Parus major, near Oxford gave SVG = 0.003 for females and 0.025 for males (McCleery et al.,
2004). A similar study of flycatchers Ficedula albicollis gave SVG = 0.085 for females
and 0.029 for males. These estimates are for the most part similar to the state of our base
population. Moreover, the strains that we chose differed about twofold in yield in the stressful medium, providing plenty of opportunity for selection in our reconstituted mixtures.
Secondly, our experimental populations might be much larger than natural populations of animals and plants. The effective population size of each line is the harmonic mean of its census size in successive generations, Ne = 1/[Σ (1/Ng)]/G] ≈ ½ N0 log2 (NG/N0), where
N0 is the initial (inoculum) census size and NG is the final (transfer) size (Samani & Bell,
5 2010). The initial effective population size of each line is thus Ne = 6.6 x 10 , and that
for each strain in a mixture of n strains is Ne/n. Population size increased threefold over
the course of the experiment, so the overall effective population size was roughly 106, which is quite small for microbial experiments. In plants and animals the effective population size is often only 0.1 – 0.2 as large as the census size ((Nei & Graur, 1984;
Frankham, 1995; Palstra & Ruzzante, 2008); summarized by (Bell, 2008) Chapter 1), so that our cultures correspond to abundant species of plants or animals with census population sizes of about 107 individuals. To put this in context, it corresponds roughly to the number of field voles in 500 km2 in a peak year (Lambin et al., 2000). This is an
82 area equivalent to the island of Montreal, which is surely a spatial scale relevant to the
dynamics of adaptation in response to natural or anthropogenic change. Many animals
are much less abundant, of course, but our system is not completely detached from non-
microbial reality.
One could ask how small a population must be before sorting becomes more important
than cumulation. The outcome of sorting in an asexual population is strictly limited by
the range of genotypes available. The maximum value M of a Normally distributed variable in a sample of N individuals from a population with standard deviation σ is
given roughly by M/σ = C + ½ log10N where C is between 4 and 6 (Bell, 2010). This can
be used to calculate the variance required to generate a maximal value equivalent to twice
the mean value, corresponding to large-effect beneficial mutations in our system, for a
given population size. Unfortunately, the log-scaling makes this calculation very
insensitive to N: for N = 103 we require SV = 0.03, and for N = 106 we require SV = 0.01.
The best that can be said is that these values lie within the range of the meagre estimates
of SVG that have been made. In sexual populations recombination can generate variation
well outside the current range and thereby lead to rapid adaptation. The mean value of
quantitative characters can be altered by 10 or more phenotypic standard deviations by
artificial selection applied for 50 generations or so in experimental populations of a few
hundred mice or flies (see Bell, 2008, Table 6.1). This would be equivalent to increasing
fitness in our experiment by at least 10 x √0.007 = 0.84, which is comparable to the
average value of 0.88 for the large-effect mutations that we observed. Hence, the
outcome of sorting combined with the recombinational assembly of multi-locus
83 genotypes is likely to be comparable to the outcome of our experiment in much smaller populations of 102 – 103 individuals.
Thirdly, the estimated mutation rate may be too low to explain our results: with an effective population size of about 106 and a beneficial mutation rate of about 10-7 per replication we require about 10 growth cycles for a line to acquire a beneficial mutation, whereas our entire experiment only spanned 12 cycles. However, our estimate includes only large-effect mutations that can be unequivocally identified, and does not include any mutations of smaller effect that would be more difficult to identify but that would nevertheless contribute to rescue.
Finally, the estimated mutation rate and the effect of the mutations may be too large to be credible. The beneficial mutation rate is not a more or less fixed property of a replication system, however, because it depends of the state of adaptedness of the population and is likely to be higher in more stressful conditions. Moreover, several recent experiments have reported rates of beneficial mutation much greater than previously expected (Desai et al., 2007; Perfeito et al., 2007), sometimes with effects on fitness as great as or greater than those we observed. One explanation of the high rate of mutation is that the mutational target size may be larger than usually thought. In a recent experiment documenting the spread of mutations conferring tolerance to high concentrations of NaCl in yeast populations we estimated that as much as 1% of the yeast genome may be liable to undergo beneficial mutations at high stress levels (Samani & Bell, 2010).
84 In short, we found that the spread of novel beneficial mutations completely overwhelmed
the contribution of standing genetic variation to adaptation and rescue in our experimental system. This result should not be naively extended to natural populations,
because the standing variation is capable of fuelling rapid adaptation in small sexual
populations. Neither should it be ignored: there is a growing body of evidence that
beneficial mutations of large effect occur more frequently than previously expected in
stressful conditions of growth.
85 References
Barrett, R. & Schluter, D. 2008. Adaptation from standing genetic variation. Trends in
Ecology & Evolution 23: 38-44.
Barrett, R.D.H., MacLean, R.C. & Bell, G. 2006. Mutations of intermediate effect are
responsible for adaptation in evolving Pseudomonas fluorescens populations.
Biology letters 2: 236 - 238.
Bell, G., Lechowicz, M. & Schoen, D. 1991. The ecology and genetics of fitness in forest
plants. III. Environmental variance in natural populations of Impatiens pallida.
The Journal of Ecology: 697-713.
Bell, G. 2008. Selection: the mechanism of evolution. Oxford University Press, USA,
New York.
Bell, G. & Gonzalez, A. 2009. Evolutionary rescue can prevent extinction following
environmental change. Ecology letters 12: 942-948.
Bell, G. 2010. The sucession of minima in the abundance of species. Oikos [in Press].
Bennington, C. & McGraw, J. 1995. Natural selection and ecotypic differentiation in
Impatiens pallida. Ecological Monographs 65: 303-323.
Bradshaw, A.D. 1991. The Croonian Lecture, 1991 – Genostasis and the Limits to
Evolution. Philosophical Transactions of the Royal Society of London Series B-
Biological Sciences 333: 289-305.
Cain, A. & Sheppard, P. 1954. Natural selection in Cepaea. Genetics 39: 89 - 116.
Colosimo, P.F., Hosemann, K.E., Balabhadra, S., Villarreal, G., Dickson, M., Grimwood,
J., Schmutz, J., Myers, R.M., Schluter, D. & Kingsley, D.M. 2005. Widespread
86 parallel evolution in sticklebacks by repeated fixation of ectodysplasin alleles.
Science 307: 1928-1933.
De Visser, J. & Rozen, D. 2005. Limits to adaptation in asexual populations. Journal of
Evolutionary Biology 18: 779-788.
Desai, M., Fisher, D. & Murray, A. 2007. The speed of evolution and maintenance of
variation in asexual populations. Current Biology 17: 385-394.
Fisher, R. 1930. The genetical theory of natural selection. Clar, endon Press, Oxford.
Frankham, R. 1995. Conservation genetics. Annual Review of Genetics 29: 305-327.
Gomulkiewicz, R. & Holt, R. 1995. When does evolution by natural selection prevent
extinction? Evolution 49: 201-207.
Harris, E. 1989. The Chlamydomonas Sourcebook. Academic Press San Diego.
Hoekstra, H.E., Hirschmann, R.J., Bundey, R.A., Insel, P.A. & Crossland, J.P. 2006. A
single amino acid mutation contributes to adaptive beach mouse color pattern.
Science 313: 101-104.
Houle, D., Morikawa, B. & Lynch, M. 1996. Comparing mutational variabilities.
Genetics 143: 1467-1483.
Kassen, R. & Bataillon, T. 2006. Distribution of fitness effects among beneficial
mutations before selection in experimental populations of bacteria. Nature
genetics 38: 484-488.
Lambin, X., Petty, S. & Mackinnon, J. 2000. Cyclic dynamics in field vole populations
and generalist predation. Journal of Animal Ecology 69: 106-118.
87 McCleery, R., Pettifor, R., Armbruster, P., Meyer, K., Sheldon, B. & Perrins, C. 2004.
Components of variance underlying fitness in a natural population of the great tit
Parus major. American Naturalist 164: E62-E72.
Mousseau, T.A. & Roff, D.A. 1987. Natural Selection and the Heritability of Fitness
Components. Heredity 59: 181-197
Nei, M. & Graur, D. 1984. Extent of protein polymorphism and the neutral mutation
theory. Evolutionary Biology 17: 73-118.
Palstra, F. & Ruzzante, D. 2008. Genetic estimates of contemporary effective population
size: what can they tell us about the importance of genetic stochasticity for wild
population persistence? Molecular ecology 17: 3428-3447.
Perfeito, L., Fernandes, L., Mota, C. & Gordo, I. 2007. Adaptive mutations in bacteria:
high rate and small effects. Science 317: 813 - 815.
R Development Core Team 2009. R: A language and environment for statistical
computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-
900051-07-0, URL http://www.R-project.org.
Reynoso, G. & de Gamboa, B. 1982. Salt tolerance in the freshwater algae
Chlamydomonas reinhardii: Effect of proline and taurine. Comparative
Biochemistry and Physiology Part A: Physiology 73: 95-99.
Roff, D.A. & Mousseau, T.A. 1987. Quantitative Genetics and Fitness – Lessons From
Drosophila. Heredity 58: 103-118.
Samani, P. & Bell, G. 2010. Adaptation of experimental yeast populations to stressful
conditions in relation to population size. Journal of Evolutionary Biology 23: 791-
796.
88 Figure Legends
Figure 1. The upper-triangular graph showing the component of each mixture with the
highest level of growth.
Figure 2. The response to selection over the course of the experiment.
2A. The trajectory of all lines. The heavy dashed line is the overall mean.
2B. Trajectories of average values for treatment levels.
Figure 3. Predicted and observed yield. Predicted yield is the yield of the highest- yielding component of the mixture.
Figure 4. Response of single strains.
4A. Growth curves over two weeks for the ancestral and derived lines. The mean value + 95% confidence limits of the mean are shown.
4B. The limiting rate of growth (r) and the limiting density (K) in ancestral and
derived lines.
89 Figure 5. Response of mixtures. The plot shows the input and output standardized
genetic variance (variance divided by the square of the mean) in randomly chosen lines
from four treatment levels. The label of each point is the number of strains in the mixture.
Figure 6. Adaptation across diversity levels. Here we use the slope where populations
increased in growth over time during the treatment. We use this as a proxy for adaptation and compare across diversity levels. There is an increase in steepness of slope,
(adaptation), as diversity increases (r squared = 0.044, p value = 0.0177)
Figure 7. An example of a replicate of an ancestral line which diverges from the other
replicates and converges on the average of the corresponding derived lines.
90
Figure 1.
91
Figure 2A.
92
Figure 2B.
93
Figure 3.
94
Figure 4A.
95
Figure 4B.
96
Figure 5.
97
Figure 6.
98
Figure 7.
99 Conclusion and Summary
We determined the growth response of Chlamydomonas reinhardtii across a range of salt
environments. This allowed us to designate a given level of salt as a severe stress
environment for our evolutionary rescue experiment. We found that a NaCl concentration of 5 g L-1 reduced growth of Chlamydomonas by 50% and a concentration of 8 g L-1
reduced growth to imperceptible levels. We also determined the genetic correlation
across environments. We showed that there are strong genetic correlations between
similar environments and weak correlations between dissimilar environments. This
provides a model for how populations are able to adapt to levels of stress that would be
lethal to their ancestor. Populations that are able to adapt to small changes in the
environment have a greater chance of adapting to larger changes in the same direction.
In Chapter 3, we observed adaptation to severe levels of stress by all of the populations,
regardless of the amount of initial diversity. We found no effect of initial diversity on the
overall outcome of adaptation. The populations in our experiment did not go extinct,
instead they adapted to the stressful environment through spontaneous beneficial
mutations. We found an increase of genetic variation after selection, rather than the
decrease suggested by theory. We believe that this was attributable to novel mutation.
This is consistent with a growing body of evidence suggesting that beneficial mutations are more frequent that was once thought.
100 Although we found no effect of variation on the outcome of adaptation, the role of standing variation facilitating short-term adaptation is quite well known. As a result, we must be cautious in generalizing these results to natural populations.
101