Phenotypic Plasticity and Local Adaptation - DiVA portal

Phenotypic Plasticity and Local Adaptation in Island Populations of Rana temporaria Martin I. Lind

Dept. of Ecology & Environmental Science Umeå University 901 87 Umeå Umeå 2009

Copyright©Martin Lind ISBN: 978-91-7264-875-3 Cover: Martin Lind Printed by: Print & Media Umeå, Sweden 2009

"These foul and loathsome animals are abhorrent because of their cold body, pale color, cartilaginous skeleton, filthy skin, fierce aspect, calculating eye, offensive smell, harsh voice, squalid habitation, and terrible venom…and so their Creator has not exerted his powers to make more of them" - Carl Linnaeus (1758)

"It is our choices [...] that show what we truly are, far more than our abilities" - J. K. Rowling




Title Phenotypic Plasticity and Local Adaptation in Island Populations of Rana temporaria

Abstract Phenotypic plasticity is the ability of a genotype to express different phenotypes in different environments. Despite its common occurrence, few have investigated differences in plasticity between populations, the selection pressures responsible for it, and costs and constraints associated with it. In this thesis, I investigated this by studying local adaptation and phenotypic plasticity in populations of the common frog Rana temporaria, inhibiting islands with different pool types (temporary, permanent or both). The tadpoles develop in these pools, and have to finish metamorphosis before the pool dries out. I found that the tadpoles were locally adapted both in development time and in phenotypic plasticity of development time. Tadpoles from islands with temporary pools had a shorter development time than tadpoles from islands with permanent pools, when reared in a common garden. The population differentiation in development time, estimated as QST, was larger than the population differentiation in neutral molecular markers (FST), which suggest that divergent selection among the populations is responsible for the differentiation. Moreover, tadpoles from islands with more variation in pool drying regimes had higher phenotypic plasticity in development time than tadpoles from islands with only one pool type present. Interestingly, increased migration among populations did not select for increased plasticity, rather it was the local environmental variation that was important. This adaptation has occurred over a short time scale, as the populations are less than 300 generations old. In temporary pools, it is adaptive to finish development before the pool dries out. This could be achieved by entering metamorphosis at a smaller size, as a smaller size takes shorter time to reach. However, I found that there is a minimum threshold size below which tadpoles’ cannot enter metamorphosis, and that there had been no evolution of this threshold size in populations living in temporary environments. That suggests that this developmental threshold is tightly linked to physiological constraints in the developmental process. Despite their expected importance as constrains on the evolution of plasticity, costs of plasticity are often not found in nature. However, theories of why they are absent have not been tested empirically. In this thesis, I show that fitness costs of phenotypic plasticity are only found in populations with genotypes expressing high levels of phenotypic plasticity, while in populations with low-plastic genotypes, I find costs of not being plastic. This suggests that costs of plasticity increase with increased level of plasticity in the population, and that might be a reason why costs of plasticity are hard to detect.

Keywords Costs of plasticity, Developmental threshold, FST, Local adaptation, Phenotypic plasticity, Pool drying, QST




5







6










List of Papers This thesis is a summary and discussion of the following papers, which are referred to by their roman numerals. I.

Lind, M. I. & Johansson, F. 2007. The degree of phenotypic plasticity is correlated with the spatial environmental heterogeneity experienced by island populations of Rana temporaria. Journal of Evolutionary Biology 20: 1288-1297.

II.

Lind, M. I., Persbo, F. & Johansson, F. 2008. Pool desiccation and developmental thresholds in the common frog, Rana temporaria. Proceedings of the Royal Society of London Series B-Biological Sciences 275: 1073-1080.

III.

Lind, M. I., Ingvarsson, P. K., Johansson, H., Hall, D. & Johansson, F. 2009. Gene flow and selection on phenotypic plasticity in an island system. Submitted Manuscript

IV.

Lind, M. I. & Johansson, F. 2009. Costs and limits of phenotypic plasticity in island populations of the common frog Rana temporaria under divergent selection pressures. Evolution 63: 1508-1518.




7







TABLE OF CONTENTS INTRODUCTION Phenotypic plasticity Costs and limits of phenotypic plasticity Estimating population differentiation in markers and traits Study system Objectives of the thesis

9 9 11 14 16 17

METHODOLOGY Working on islands The island system Rearing conditions Laboratory treatments Molecular genetic analyses Statistical analyses Estimating costs of plasticity

18 18 18 19 20 21 21 23

RESULTS AND DISCUSSION Identifying the selection pressures on the islands Phenotypic plasticity and local adaptation Growth, growth rates and developmental thresholds Costs and limits of plasticity Rapid local adaptation – implications for conservation Concluding words

25 25 27 32 36 40 42

REFERENCES

42

WORDS OF THANKS

55

8










Phenotypic Plasticity and Local Adaptation in Island Populations of Rana temporaria Martin I. Lind Department of Ecology and Environmental Science, Umeå University, 901 87 Umeå, Sweden E-mail: [email protected]

INTRODUCTION Phenotypic plasticity Phenotypic plasticity is the ability of a genotype to produce more than one phenotype, depending upon the environment (Gotthard, & Nylin, 1995). In other words, it means that an individual can develop quite differently depending upon what environment it currently inhabits, in contrast to specialists, who produce the same phenotype in all environments. Well-known examples of phenotypic plasticity are the different growth forms of plants in shaded vs. sunny patches (Bradshaw, 1965), morphological defence structures, such as spines, expressed by many aquatic organisms in the presence of predators (Tollrian, & Dodson, 1999), and sun tanning in humans. Plasticity is often visualised graphically as a reaction norm, by plotting the different phenotypes expressed as a function of the environment (Figure 1). However, despite its common occurrence in nature (WestEberhard, 2003), it has long been considered less important by evolutionary biologists (Pigliucci, 2005). Although the first work on phenotypic plasticity was performed as early as the beginning of the 20th century (Woltereck, 1909), with the rediscovery of Mendel’s work on genetics (Mendel, 1866) and especially by formulation of the genetic theory of evolution during the modern synthesis (Huxley, 1942), evolutionary theory essentially became a “theory of genes” (Pigliucci, 2007). In this perspective, plasticity was seen as something that was hiding the true genetic response and a problem in experiments on evolution (Falconer, 1952). However, evidence for plasticity was still accumulating (e.g. Bradshaw, 1965), but it was not until the eighties that theoretical evolutionary biologists became interested in phenotypic plasticity and reaction norms, and the first models on 


9







their evolution were published (Via, & Lande, 1985). Since then, phenotypic plasticity has been well studied theoretically (e.g. Gabriel, & Lynch, 1992; Van Tienderen, 1991; Gomulkiewicz, & Kirkpatrick, 1992; Moran, 1992; Sultan, & Spencer, 2002; Ernande et al., 2004), and we now know that phenotypic plasticity is adaptive and often beneficial. The models have identified a number of conditions that explains whether phenotypic plasticity or local specialisation is most likely to evolve. Plasticity should be selected for if there is environmental heterogeneity present (either variation over the landscape or variation over time). In addition, plasticity is more likely if none of the environmental states are more common than any others. The environmental variation has also to be predictable, so that the organisms can accurately identify the environmental conditions and develop the appropriate phenotype. The evolution of plasticity is constrained by lack of genetic variation in plasticity, and by costs of plasticity. (a)

(b)

(c)

Figure 1. Illustration of phenotypic plasticity using reaction norms of two genotypes. Each line represents one genotype and its phenotypic expression along an environmental gradient. (a) The two genotypes have different phenotypes, but the phenotype is insensitive to the environment, thus no phenotypic plasticity is present. (b) The two genotypes produce different phenotypes in different environments, thus plasticity is present. However, the magnitude of plasticity (the slope of the reaction norm) is similar, so there is no genetic variation in plasticity in the population. (c) The two genotypes show different degree of plasticity, thus there is genetic variation for plasticity in the population (i.e. a genotype × environment interaction).

However, despite the number of studies demonstrating the presence of phenotypic plasticity in natural populations systems (reviewed by

10










Bradshaw, 1965; Schlichting, 1986; Gotthard, & Nylin, 1995; WestEberhard, 2003; Pigliucci, 2005), surprisingly few have focused on the variation in phenotypic plasticity among populations, and many of the theoretical expectations have not been well investigated experimentally (Pigliucci, 2005). Few have investigated the selective forces responsible for phenotypic plasticity in nature, or if and why, plasticity may differ among populations (Pigliucci, 2005). Another area were knowledge is lacking is the role of gene flow for the evolution of phenotypic plasticity. Theoretical models (Scheiner, 1998; Sultan, & Spencer, 2002) have identified gene flow between populations as a factor that increase the likelihood that plasticity and not specialisation will evolve. This is because migration results in specialists encountering the wrong environment more frequently (Scheiner, 1998; Sultan, & Spencer, 2002). Moreover, models have also suggested that plasticity by itself can promote dispersal, as plastic individuals are more likely than specialists to survive in novel environments with different selection pressures (Price et al., 2003). These predictions have some empirical support, as marine invertebrates with high dispersal rates also have a high degree of phenotypic plasticity (Hollander, 2008), and invasive species often are more plastic in the invasion (Sexton et al., 2002; Niinemets et al., 2003; Yeh, & Price, 2004). However, if the more plastic populations also live in the more heterogeneous localities, gene flow may actually disrupt this relationship simply by immigration of maladapted specialists (Alpert, & Simms, 2002; Crispo, 2008). This effect of gene flow for the evolution of plasticity has been much less studied, and although a comparison among amphibian species suggests that the species living in the most heterogeneous environments also are more plastic (Richter-Boix et al., 2006a), no one has directly studied the role of gene flow on locally adaptive plasticity. Costs and limits of phenotypic plasticity Despite the numerous examples of phenotypic plasticity in nature (e.g. reviews by Bradshaw, 1965; Gotthard, & Nylin, 1995), phenotypic plasticity is not always found (Delasalle, & Blum, 1994; Pigliucci, 1997), and the degree of phenotypic plasticity often varies between populations (Donohue et al., 2000; Merilä et al., 2004b; Van Buskirk, & Arioli, 2005). As some degree of environmental variation 


11







often is present, this indicates that phenotypic plasticity is not only beneficial, but is also subjected to costs and constrains. Costs of plasticity were included in the models of the evolution of plasticity by Van Tienderen (1991), and later models have shown that costs of plasticity, together with the accuracy of the environmental cues, is a major determinant as to whether or not phenotypic plasticity will evolve in heterogeneous environments (Moran, 1992; Van Tienderen, 1997; Sultan, & Spencer, 2002). There could be many different costs of plasticity such as: maintenance costs, production costs, information acquisition costs, genetic costs or costs by developmental instability (DeWitt et al., 1998). The cost of plasticity is paid as a reduction in fitness that a non-plastic individual does not pay when expressing the same trait value (DeWitt et al., 1998). The last notation is important, as it is the extra cost paid by the plastic genotypes that is of interest. For example, it is likely that the production of an anti-predator phenotype in Daphnia (Tollrian, & Dodson, 1999) is costly, however, such a cost by a nonplastic individual expressing the same antipredator phenotype and therefore not a true cost of plasticity. Instead the cost of plasticity is the extra cost paid by the plastic genotype when producing this defence compared to a non-plastic genotype that always develop this defence morphology. Hence, it is important to keep this distinction in mind when discussing costs of plasticity. Although a method of how to measure costs of plasticity has been available for a number of years (Van Tienderen, 1991), it was not until the publication of the opinion paper by DeWitt et al. (1998) that experimental research on costs of plasticity started. However, despite the importance plasticity costs have in theoretical models (eg. Van Tienderen, 1991; Moran, 1992), they have been surprisingly difficult to find in natural populations. In fact, they are usually only found in a fraction of the traits measured in a study (DeWitt, 1998; Scheiner, & Berrigan, 1998; Donohue et al., 2000; Dorn et al., 2000; Poulton, & Winn, 2002; Relyea, 2002a; Steinger et al., 2003; Callahan et al., 2005; Weijschedé et al., 2006; Weinig et al., 2006; Avramov et al., 2007; Steiner, & Van Buskirk, 2008; Van Buskirk, & Steiner, 2009), or in a few environments and populations (Donohue et al., 2000; Merilä et al., 2004b; Callahan et al., 2005). Moreover, a number of studies have also found a significant positive relationship between plasticity and fitness (Dorn et al., 2000; Relyea, 2002a; Weijschedé et al., 2006; Weinig et al., 2006; Avramov et al., 2007; Steiner, & Van Buskirk,

12










2008), which suggests that plasticity is sometimes not costly. Instead, this suggests that but that the genotypes with a canalized (i.e. nonplastic) development sometimes pay a higher cost (Poulton, & Winn, 2002). The failure of most studies to find costs of plasticity had often been explained by the action of natural selection, which might already have removed plastic genotypes with high costs (DeWitt et al., 1998; Sultan, & Spencer, 2002; Callahan et al., 2005; Weinig et al., 2006). This hypothesis however is hard to test, as the costs of plasticity (which by definition is expressed even when the plasticity itself is not) should be continuously removed by natural selection, and therefore not present in natural populations. One could argue that a possible way to test this would be to compare costs of plasticity in populations of different age, but as the individuals from these populations have also evolved prior to when they were isolated from other populations, it is not likely that this approach would give any insight in the process. Another possible explanation for the lack of plasticity costs found in most studies is that the costs might differ among populations. If costs of plasticity increases with the degree of plasticity expressed, as suggested by Van Tienderen (1991), then costs of plasticity are more likely to be found in populations where phenotypic plasticity is under selection. On the other hand, there would be no difference among the populations if the magnitudes of the costs were independent of the degree of plasticity (as in DeWitt et al., 1998). This question is investigated in Paper IV. In contrast to plasticity costs, limits of plasticity means that the plastic genotype cannot produce the full range of phenotypes that the different specialists are able to, i.e. the plastic genotype are not able to express the optimal phenotype (DeWitt et al., 1998; Pigliucci, 2005). This is conceptually very different from costs of plasticity, where the plastic genotype always pays a fitness cost, whether or not plasticity is expressed (Pigliucci, 2005). There have been few experiments that have investigated plasticity limits, but none of the experiments have found any support for the concept (DeWitt, 1998; Relyea, 2002a). It has been suggested that there is a lack of support for limits of plasticity ecause alternating selection pressures may select against nonplastic specialists (Relyea, 2002a). Therefore, limits of plasticity




13







should only be present in populations where plasticity is not under selection, but this has never been tested. Estimating population differentiation in markers and traits There is a general observation that different populations of a species often differ from each other in one or more traits (Hendry, & Taylor, 2004). This can be seen as a result of natural selection favouring different trait optima in the different populations (due to different environments). If the populations are large enough and separated for enough time, this could be a first step in a speciation process (Schluter, 2001; Gavrilets, 2003). However, different selection pressures in the different populations are not the only factors that can explain phenotypic differences between populations; neutral genetic drift and novel mutations will over time also lead to accumulated differences between the populations (Merilä, & Crnokrak, 2001; Hendry, & Taylor, 2004). To distinguish between these two processes, one must first be able to estimate the degree of difference between the populations on a common scale. With the introduction of molecular methods in ecological research, population diversification in neutral genetic markers has become this null-model for population differentiation, against which differentiation in quantitative characters can be compared. The genetic difference in a neutral trait between two populations of a species is determined by their degree of genetic isolation, which often is a result of some degree of physical isolation. Genetic differentiation between populations arises through mutations and random genetic drift, but is counterbalanced by gene flow (migration) between the populations, which mixes the gene pools and reduces the differences. Wright (1951) formalised this relationship by using the term FST, which is the proportion of the total variance in a neutral marker loci that is due to difference between populations:

(1) As seen, the larger the genetic variation between the populations is (Vb), compared to the variation within the populations (Vw), the higher the value of FST. Since Vb is found both in the numerator and the denominator, the value of FST can vary between 0 (all variation is

14










found within the populations) and 1 (all variation found between the populations; the populations are isolated without migration). While much of the theory concerning genetic changes in populations is developed using Mendelian genetics with two-locus models, variation in phenotypic traits (for example length) is typically not discrete but continuous (Futuyma, 1998). This continuous variation is due to both environmental effects and the influence of not just one locus; quantitative traits are determined by many interacting loci, each with a number of possible alleles (Reed, & Frankham, 2001). Thus, the traits are determined by additively acting Quantitative Trait Loci (QTLs) (Latta, 2004a, b). In a diploid organism, Wright (1951) showed that, for a quantitative trait with purely additive gene effects, the neutral expectation of mean genetic variance within populations ( ) is: (2) The between population variance (

) would then be: (3)

The term represents the variation that would exist if the populations would be a single unit. Since the total genetic variance in a trait ( ) equals , an estimation of the divergence in quantitative traits can be obtained (Spitze, 1993): (4) This formula is very similar to that of FST (1); the only difference (except for the fact that one is estimating the variance of quantitative traits rather than neutral markers) is that the within-population variance (σ2QW) in the denominator is multiplied by a factor 2. This is because QST is calculated comparing genotypes, while FST is comparing genes (Merilä, & Crnokrak, 2001). Thus, FST and QST are compa


15







rable, and QST has become the standard measure of genetic divergence in quantitative traits. As the population differentiation in neutral markers is purely the effect of reproductive isolation and drift, other factors than genetic drift must be invoked to explain significant differences between FST and QST, and natural selection is the most common explanation. If QST > FST, then divergent or disruptive selection, favouring different trait values in different populations, must be invoked. On the other hand, if QST < FST, the populations are less differentiated than expected based upon the level of genetic drift, and stabilizing selection favouring the same trait value in all populations is the most likely explanation (Spitze, 1993; Whitlock, 1999; Leinonen et al., 2008). Study system The common frog Rana temporaria L. 1758 is distributed over most of Europe, except the most southern parts, and extends it distribution range far north of the Arctic Circle (Gasc et al., 1997). Together with the Moor frog R. arvalis, it is the most common frog species in Scandinavia (Ahlén et al., 1995). R. temporaria reaches sexual maturation after 2-3 years (Miaud et al., 1999). R. temporaria breeds in most types of pools, although northern populations show a preference for fishless pools in favourable microclimatic conditions (Ahlén et al., 1995). It is common on islands in archipelagos of the Baltic Sea (Seppä, & Laurila, 1999), where it often breeds in rock pools of varying duration (Johansson et al., 2005). In the Umeå area, the breeding season of R. temporaria starts soon after the ice has melted, and spawning starts at a water temperature of only 5°C (Elmberg, 1990). Although the females are physically capable of laying two egg clumps per season, one from each ovary, experiments where female R. temporaria from this area were allowed to breed in buckets showed that they only lay one egg clump (Elmberg, 1991), and that no eggs were left in the reproductive system (Elmberg 2007, pers. comm.). If the pools in which the tadpoles of R. temporaria are developing are of short duration, the tadpoles are able to speed up the development and metamorphose at a smaller size (Laurila, & Kujasalo, 1999). Moreover, R. temporaria show genetic differentiation in development

16










time (Merilä et al., 2000a; Laugen et al., 2002; Palo et al., 2003) and plasticity in development time (Laurila et al., 2002; Merilä et al., 2004b) when populations from different parts of Sweden are compared. Therefore it is likely that local adaptation will be found even on a smaller scale. By investigating adaptation in populations separated by only a few kilometres, it is possible to identify the selection pressures behind any population differences in development time and plasticity in development time, as there will be no confounding factors such as latitude, temperature or the length of season. Therefore, I investigated plasticity and local adaptation in R. temporaria from islands at the Baltic coast, which differed in pool drying regimes (Johansson et al., 2005). Objectives of the thesis By characterisation of the island environments, common garden studies with different treatments and molecular analyses of population subdivision and gene flow, my aim was to investigate the following questions: (1) Examine if the different selection pressures on the islands can cause divergent selection in life history traits. (2) Investigate the role of environmental heterogeneity and gene flow for the evolution of phenotypic plasticity. (3) Test if there is a minimum size, which individuals need to pass before metamorphosis is possible, and if this threshold size can evolve even over short time scales in natural populations connected by gene flow. (4) Investigate if the presence or absence of costs of plasticity can be understood if using an approach with populations under different selection pressures.




17







METHODOLOGY Working on islands Since Darwin’s pioneering work on the fauna of the Galapagos Islands (Darwin, 1859), island systems have been of major importance for experimental evolutionary biologists, and many of these studies have increased our understanding of speciation (Carson, & Kaneshiro, 1976; Grant, & Grant, 2002; Savolainen et al., 2006) as well as contemporary evolution (Kruuk, & Hill, 2008). Island systems are in general well suited to study the effect of local adaptation, for many reasons (Emerson, 2002). Islands themselves are well-defined units with clear boundaries, which in this case also is the boundary of the populations. Identifying population boundaries is therefore relatively straightforward, in contrast to the same procedure a continuous landscape with no clearly visible boundaries. This makes it possible to quantify the environmental conditions that the island populations are subjected to in a much simpler way than it is to quantify the environment for a population living in a continuous landscape.

Figure 2. Map of the island system in the archipelagos of Umeå and Skeppsvik along the Baltic coast of Västerbotten, northern Sweden. The 15 study populations are marked in black colour. The isolated population in the top-right corner was not used in the study of gene flow.

The island system The studies have been carried out in the archipelago of Holmsund and Skeppsvik, south and east of the city of Umeå, Sweden (Figure 2). Soon after the ice covering the pools had melted, during the first

18










or second week of May, eggs were collected from the islands. We sampled 40-50 eggs from up to ten (Paper I, III) or twelve (Paper II, IV) egg clumps per island, if that number was present. If the frogs were breeding in multiple pools, eggs were collected from all pools. By collecting egg clumps, we sample full-sibs (individuals sharing the same mother) as the common frog only breeds once a year in this area (Elmberg, 1991), and only lays one egg clump per season (Elmberg 2007, pers comm.). As no individuals from different egg clumps had the same genotype (Lind, unpublished results), every egg clump represents a unique family. Rearing conditions The collected eggs was transported to the laboratory, and kept cool to prevent further development, until all eggs were collected. In order to control for maternal effects expressed through egg size (Laugen et al., 2002), 10 eggs from each female (Gosner stage 10, Gosner, 1960) were placed into a petri-dish, covered with water and photographed together with a scale. From these digital images, egg size was measured using the image analysis software ImageJ (http://rsbweb.nih.gov/ij/). The experiments were carried out in thermo-constant laboratory conditions. At hatching, at Gosner stage 23 (active swimming), a number of tadpoles (the actual number depending upon the experimental design) were randomly chosen and placed individually in plastic experimental containers (9.5 × 9.5 cm, height 10 cm), filled with 750 mL of oxygenated tap water. The water was taken from an indoor cattle-tank, oxygenated in the presence of dead deciduous leaves, in order to get semi-natural water chemistry. The temperature of the thermo-constant laboratories were set to 22°C, with a light : dark cycle of 18 h : 6 h, which corresponds to the natural day : night cycle in the area of egg collection. In order to supply the tadpoles with oxygenated water, and to remove detritus, the water of the experimental containers was replaced every fourth day, just before feeding. The larvae were fed either a 1:1 (Paper II, IV) or a 1:2 (Paper I, III) mixture of fish food and rabbit chow. The food mixture was grounded to a fine powder before feeding. The difference in food composition between the experiments was unintentional (typo in the lab protocol), but did not affect the 


19







patterns found (the same pattern of local adaptation was, for example, found in Paper I and II, which had different composition of the food). The experiment continued until the tadpoles reached Gosner stage 42 (Gosner, 1960). At this stage, the front legs are visible, and we defined this as our estimate of metamorphosis. Time to reach this developmental stage was recorded as development time, and the wet weight was correspondingly recorded as metamorphic weight. Admittedly, the metamorphosis is not complete until the tail is complete reabsorbed (at Gosner stage 46), but Gosner stage 42 is a well defined stage (either the front legs are present or they are not), while it is difficult to exactly determine when the tail is completely reabsorbed. However, the two measures are correlated (Paper I), and the metamorphic weight at Gosner stage 42 is also correlated to the metamorphic weight as a tadpole at Gosner stage 37 (Johansson, Lederer & Lind, manuscript). Laboratory treatments The tadpoles were subjected to different treatments in the lab, either a water level treatment (Paper I, III, IV) or a food level treatment (Paper II). The water level treatment was used to estimate the degree of phenotypic plasticity in development time as a response to simulated pool drying. In order to induce a plastic acceleration of development time, the initial water volume (750 mL) in the artificial drying treatment (D) was lowered by 33% every fourth day, until 66 mL remained (see Paper I for a detailed description of the artificial drying treatment). This procedure is often used when studying adaptive plasticity to pool desiccation (Merilä et al., 2004b; Gervasi, & Foufopoulos, 2008), and the environmental cue seems to be the decrease in water volume per se (Denver et al., 1998). The degree of plasticity in development time was then recorded as the mean development time of a family in the constant water level treatment, minus the development time in the artificial drying treatment. The food level treatment was used in order to estimate the developmental threshold of metamorphosis, following Plaistow et al. (2004). We used three different food levels, to represent unlimited food and two degrees of starvation. The high and medium food levels were the

20










same food levels as have been used in a comparison of the size at metamorphosis of different population of R. temporaria across Sweden (Merilä et al., 2004a). The low food level was chosen to be half of the food used in the medium food level, to ensure that the tadpoles were severely food limited and would metamorphose at the developmental threshold. The exact amount of food given over time in the different food levels is to be found in Paper II. Molecular genetic analyses The markers used for studying neutral differentiation must measure variation that is not thought to be subjected to selection, thus behaving neutral. They can either be different electrophoretically distinguishable forms of a protein (allozymes) or DNA fragments (microsatellites, RAPD, AFLP) (Storfer, 1996), and sometimes RFLPs from the mitochondrial genome (Lynch et al., 1999). During the recent years, microsatellite markers have been the marker of choice (REF). To obtain estimates of population differentiation in neutral genetic markers, allelic variation was assessed using six previously published microsatellite loci: Rt2Ca2-22 (Lesbarrères et al., 2005), RRD590 (Vos et al., 2001), RtµH (Pidancier et al., 2002), RtU4 (Berlin et al., 2000), Rtempµ4 and Rtempµ7 (Rowe, & Beebee, 2001). RtU4 was later excluded from all analyses due to strong deviations from HardyWeinberg Equilibrium (due to a deficit in heterozygotes). Statistical analyses In contrast to many studies of phenotypic plasticity, which have only studied plasticity within a single population (DeWitt, 1998; Van Buskirk, & Relyea, 1998), we have consistently used a multipopulation approach. Therefore, when investigating the effect of the natural selection pressures (Paper I, II, III), general island characteristics (Paper I) and gene flow (Paper III), the population has been our unit of replication. Therefore, the different families have mostly been used to estimate the population mean. The exception have been when we have been interested in the individual response, as when estimating costs and limits of plasticity (Paper IV). In these analyses, the population has been treated as a random factor. In paper III, we compared the population differentiation in neutral genetic markers with the differentiation in life history traits, in order 


21







to investigate if the differentiation was larger than expected based only upon neutral genetic drift (i.e. if divergent selection has been involved). Population differentiation in neutral markers (FST) was compared to the population differentiation in development time, plasticity in development time and metamorphic weight (QST). QST was estimated using equation 4, and the within- and between population variance components were estimated from a linear model with the quantitative trait as response variable, family and population as random effects and, when estimating development time and metamorphic weight, treatment as a fixed effect. However, when calculating QST of plasticity, the individual variation in plasticity within a family was needed. This posed a problem, as the same individual cannot be subjected to the two water level treatments at the same time. Therefore, we estimated the plasticity in development time in all four possible pair-wise combinations of full-sibs, and then randomly picked two plasticity measures from each family, making sure that no individual had been used to estimate both plasticity measures. QST was estimated using a Bayesian approach (O'Hara, & Merilä, 2005; Hall et al., 2007). Recently, the practise of comparing FST and QST has been criticised on several ground (Pujol et al., 2008; Whitlock, 2008). Most importantly, individuals must be raised in a common garden in order to estimate the genetic component of the phenotypic differences (Pujol et al., 2008) and maternal effects need also to be controlled for (Whitlock, 2008). This was done in the analyses, by using a common garden with two water level treatments and controlling for maternal effects transferred through egg size. As the condition, age and life history strategy of the mother in many cases can influence the life history of the offspring through nongenetic maternal effects (Mousseau, & Fox, 1998; Räsänen, & Kruuk, 2007; Marshall, & Uller, 2007), we had to investigate the role of maternal effects in our analyses. This was done using a North Carolina II crossbreed half-sib design (Lynch, & Walsh, 1998) in one population, where we investigated the magnitude of maternal effects in our system (Paper I). As maternal effects in R. temporaria mainly are transferred through the maternal investment in the size of the eggs (Laugen et al., 2002), we investigated and controlled for the role of egg size in all analyses.

22










Estimating costs of plasticity Costs of plasticity were not included in the models concerning the evolution of phenotypic plasticity until the work of Van Tienderen, (1991), who also was the first to outline an empirical approach to search for costs of plasticity in natural populations. The underlying assumption was that, if costs of plasticity are present, the plastic individual has a fitness cost when expressing the same trait value as a nonplastic individual (DeWitt et al., 1998). However, it took several more years until Scheiner, & Berrigan (1998) formulated the following general model for testing if costs of plasticity were present in a population, which then has been the basis for all later empirical tests: W = X + X2 + plX + X × plX + X2 × plX Where W is the fitness estimate, X is the trait value in the tested environment and plX is the plasticity of the trait, measured between the two environments. X and X2 estimate the linear and nonlinear selection components (Scheiner, & Berrigan, 1998).

Figure 3. Illustration of a cost of plasticity. A cost of plasticity is a negative relationship between the plasticity of a trait and fitness, when controlling for the trait value (i.e. a regression between plasticity and the fitness residuals).

We were interested in estimating costs for plasticity in development time, as this trait was likely to be under diversifying selection in the island system (Paper I, III). Here we used metamorphic weight as a fitness estimate. This is not a direct estimate of fitness, but as R. tem


23







poraria is a long-lived species with 2-3 years to reach sexual maturity in this area (Miaud et al., 1999), this was our only option. Large size is selected for under natural conditions (Altwegg, & Reyer, 2003), and individuals with high metamorphic weight usually maintain or increase this size advantage (Smith, 1987; Morey, & Reznick, 2001; Altwegg, & Reyer, 2003). As metamorphic weight also is known to increase survival during the first terrestrial year (Berven, 1990; Morey, & Reznick, 2001; Altwegg, & Reyer, 2003), and is positively related to physical performance such as jump length (Ficetola, & de Bernardi, 2006; Richter-Boix et al., 2006b; Johansson, Lederer & Lind, in prep), we believe that the use of metamorphic weight as a fitness estimate is justified. In order to investigate if the costs of plasticity in a population are related to the degree of plasticity present, we analysed costs of plasticity first for three populations with high plasticity in development time, and then for three populations with low levels of plasticity. Population was included as a random effect in the models. Plasticity was measured in constant water and under an artificial drying treatment (see section above for rearing conditions) and the analyses were performed separately for the two treatments. Limits of plasticity (investigating if the plastic individuals were not able to express the extreme traits) were investigated by linear regressions of family level plasticity in development time on the development time for each treatment.

24










RESULTS AND DISCUSSION Identifying the selection pressures on the islands In temperate regions, lenthic freshwater habitats are distributed along a gradient from ephemeral pools to permanent water bodies, and the community composition depends upon the interaction between physical and biotic factors (Wellborn et al., 1996). Along this gradient, three major types of water bodies are identified: temporary habitats, permanent fishless habitats with invertebrate top predators, and permanent water bodies with fish as the top predator. Thus, there are two major transitions; a permanence transition and a transition where there is no oxygen stress during the winter, which enables fish to survive (Wellborn et al., 1996). In our system, no ponds are large and deep enough to enable fish presence, and fish presence has never been recorded in any of the island pools. Therefore, the two major selection pressures are pool permanence and predator regime. As life history traits of tadpoles are affected both by pool drying (Johansson et al., 2005) and by the local predation pressure (Skelly, & Werner, 1990; Relyea, 2002b), these two factors can be confounded when studying local adaptation in nature. However, as will be discussed below, in this island system, the only selection pressure that differs between the island populations is the degree of pool drying. Pool drying The mean degree of pool drying, estimated as the decrease in water depth over the summer for all pools in which breeding had occurred, differed between the islands, as did the coefficient of variation in pool drying regimes (Paper I). The variation in pool drying regimes reflects the spatial heterogeneity in selection pressures a population is subjected to, i.e. the environmental variation. The environment can also vary on a temporal scale, so that some pools vary more than other in water level between years. As temporary pools have a higher between-year variation in the degree of drying than permanent pools (Newman, 1992; Brooks, & Hayashi, 2002; Loman, & Claesson, 2003) the island differences in mean pool drying might also reflect island differences in temporal environmental heterogeneity in drying regimes. The degree of, and variation in pool drying regimes was not correlated with the age or size of the islands, the distance to the nearest population or to the mainland, or to the number of pools present 


25







(Paper I). Thus, the islands can be seen as randomly distributed replicated of the different pool drying regimes. Predators As invertebrate predators often are more common in permanent water bodies (Wellborn et al., 1996), we investigated the difference in predator abundance between islands with permanent or temporary pools. Predators were sampled on May 4 and May 6 in 2007 on three islands with permanent pools (Gåshällan, Lillhaddingen and Storhaddingen) and three with temporary pools (Ålgrundet, Bredskär and Sävar-Tärnögern). A plastic cylinder (diameter 25 cm and height 30 cm) was submerged into the bottom sediment and all invertebrates inside the cylinder were collected with a small hand net (mesh size 0.5 mm) and preserved in ethanol for later identification in the laboratory. Six cylinder samples were taken in each pool and islands were used as the replicate unit for the analysis. Dytiscidae (predaceous diving beetles) were the only invertebrate predators that were found in the samples and they occurred in very low numbers: about one individual per sample. There was no difference in mean number of Dytiscidae between temporary (mean = 6.33, SE = 3.84) and permanent pools (6.33, SE = 2.96; T-test unequal variance: p = 1.00 and Kruskall Wallis test: p = 0.50). Therefore, predators are equally uncommon in permanent and temporary pools, meaning that any local adaptation to predator presence was not confounded by local adaptation to the pool-drying regime. In addition, a recent meta-analysis has shown that abiotic factors, such as pool permanence, are more important for structuring amphibian communities than is biotic factors (Werner et al., 2007). Temperature In addition to the degree of pool drying and predator abundance, tadpoles can also be locally adapted to the water temperature (Ståhlberg et al., 2001; Laugen et al., 2003). Given that temporary pools are shallower than permanent pools, and do dry out faster, it might be reasonable to believe that the temporary pools are warmer and that the populations therefore can be adapted to different thermal optima. Therefore, we recorded the water temperature in pools on five islands from early May to late June 2007 (Weinreich, 2007). The pools spanned the whole gradient from temporary to permanent pools, and surprisingly the temporary pools were not warmer than the permanent ones. In fact, the deep permanent pools had a higher daily mean

26










temperature, mainly because of less cooling during the night. In addition, we found no local adaptation to temperature in a common garden trial with two temperature treatments for any population (Weinreich, 2007). Phenotypic plasticity and local adaptation A model of the local adaptation in the island system Colonization The common frog colonized Scandinavia from two directions: microsatellite based inferences suggest that southern Scandinavia was colonized from Denmark, while northern Scandinavia was colonized from eastern Europe through Finland (Palo et al., 2004). Therefore, the island system presented here belongs to the eastern clade. Since the last glaciation, the coast of north Sweden is affected by an extensive isostatic land uplift, which makes it possible to date the age of the islands simply by measuring their height. Based on this relationship, we know that the first frogs dispersed to the islands from the mainland less than 300 generations ago (Johansson et al., 2005). The colonization of the islands could in principle follow two routes (Paper III). According to an isolation model, the islands were colonized from the mainland when they emerged out of the sea, but have then been isolated since the colonization event. The other model emphasis that migration did not stop when the islands were colonized, but that there have been continuous migration since then. These two models give very different predictions regarding the genetic relatedness among the populations. According to the isolation model, the populations should be more genetically isolated if they have experienced a longer time of independent evolution and drift. This is in complete contrast to the predictions from the migration model, where the age of the populations are not thought to influence the genetic structure of the population, but rather its geographical location and its proximity to other islands and the mainland. As the archipelago is situated on a land uplift coast, the age of an island can be determined by its height above the current sea level, and therefore we could test these models.




27







We tested the isolation model by calculating all pairwise isolation times between all island pairs, by summing their height over the current sea level and investigated if this was correlated to the pairwise differences in FST. These factors were not correlated, and we could therefore falsify the isolation model as an explanation of the population differentiation among the islands (Paper III). Moreover, as within population allelic richness and gene diversity were perfectly correlated, there seems not to have been any bottleneck events (as would have been identified by a drop in allelic richness) (Comps et al., 2001; Widmer, & Lexer, 2001). Instead, we found support for a model with continuous migration from the mainland, as the islands were more genetically distinct from inland populations the longer distance of seawater that isolated them from the mainland (Paper III). With random dispersal, we also expected that the populations would be more genetically isolated with increasing distances (Wright, 1943). However, we found no isolation by distance among the island populations, which might be surprising as isolation by distance is commonly found among R. temporaria populations in Sweden (Palo et al., 2003), (Johansson et al., 2006). This could be explained by very high or very low gene flow, or by other factor than distance affecting gene flow among islands (Keyghobadi et al., 2005; Bergek, & Björklund, 2009). As we find no evidence of unlimited gene flow (we found significant population structure) or any evidence of complete isolation (isolation model not supported), we can rule out these two models as explanations to the lack of isolation by distance. However, as the islands are distributed along a line, following the coast, and the major gene flow seems to come from the mainland, the islands are located perpendicular to the direction of the gene flow, and therefore no isolation by distance among islands is likely to be found. Instead, distance from mainland, measured in distance seawater, seems to determine the genetic structure. Another evidence for restricted dispersal is the observation that the population on the island Grisslögern went extinct during 2006 (by unknown reason, the breeding pool was scattered with dead adults) and the island was not recolonised during the following two years. Interestingly, the individuals dispersing to the islands are not a random sample of the individuals present on the mainland. Instead, we found that the individuals on the islands had other proportions of behavioural types than in the mainland populations. The individuals that had migrated to the islands were of a bolder personality type

28










(Lind, Brodin, Wiberg, and Johansson, in prep). This pattern is consistent with experiments where individuals with known personalities have been released (Fraser et al. 2001). However, personalities of dispersers in natural populations has only been investigated once before (Dingemanse et al. 2003), and our study is the first to find that bold individuals are not only most likely to disperse, but also to colonize new environments. Adaptation We know that phenotypic plasticity in development time, comparable in magnitude to what we can see in the island populations, is present in mainland populations (Almfelt, 2005). Therefore, it is likely that the individuals that migrate to the islands have some degree of plasticity in development time present. To understand the local adaptation and phenotypic plasticity in development time, it is useful to see them as two interacting traits that are allowed to evolve separately. Recently, such a model has been developed by Leimar et al. (2006). They suggested a framework in which development is idealized as a switching device, with genetic (specialisation) and environmental (plasticity) cues as inputs. As an illustrative example is the regulation of development time among species of Spadefoot toads and Parsley frogs, where the mean development time is determined by the tissue sensitivity to thyroid hormone, while the plasticity in development time is governed by adjustments of the concentration of thyroid hormone in the blood (Gomez-Mestre, & Buchholz, 2006). The independence of the trait mean and the plasticity of the trait have been debated, as some have suggested that they should be connected (Via, 1993) while others have argued for their independence as separate traits (Scheiner, 1993; Schlichting, & Pigliucci, 1993). In our system, as well as in the model of Leimar et al (2006), they are not connected, as we find no correlation between the development time and the plasticity in development time among the island populations (Paper I). As genes for plasticity have been found (Pigliucci, & Schmitt, 1999), I think it is useful to see them as separate traits that can respond independently to different selection pressures, as will be shown below.




29







We found local adaptation both in development time (Paper I, II, III) and in phenotypic plasticity of development time (Paper I, III). The populations were more differentiated in development time than they were expected to be, based upon their differentiation in neutral genetic markers (Paper III), which is a strong indication of divergent natural selection (Spitze, 1993; Whitlock, 1999; Merilä, & Crnokrak, 2001; McKay, & Latta, 2002; Leinonen et al., 2008). The selection pressure responsible for this diversifying selection is the mean degree of pool drying, as we found that the mean development time of tadpoles from an island was explained by the degree of pool drying present on that island (Paper I, II, III). Local adaptation in development time in amphibians is known both at the population level (Merilä et al., 2000b; Laugen et al., 2003; Palo et al., 2003) and at the species level (Morey, & Reznick, 2000, 2004; Richter-Boix et al., 2006a). However, here we have shown that the local selection pressures can lead to divergent selection even over small geographic scales with relatively high levels of gene flow. Moreover, we also found local adaptation in phenotypic plasticity in development time, so that tadpoles from islands with high spatial variation in pool drying regimes (i.e. islands with both temporary and permanent pools) had a higher plasticity in development time than tadpoles from islands with only one pool-drying regime present (temporary or permanent) (Paper I, III). This relationship is predicted by theory (Moran, 1992; Sultan, & Spencer, 2002; Ernande, & Dieckmann, 2004) and is supported by between species comparisons (Richter-Boix et al., 2006a; Hollander, 2008) but has not, in spite of several attempts (eg. Loman, & Claesson, 2003; Huber et al., 2004), previously been found within species. Our studies, however, can demonstrate this relationship due to the fact that we are using an island system and are working on a small geographical scale. Thus, we can easily determine the boundaries of a population (Paper III) and control for other environmental variables that potentially could confound any relationship (Paper I). In contrast to predictions (Sultan, & Spencer, 2002; Hollander, 2008), we found no positive effect of gene flow between the islands on the level of plasticity (Paper III). We note that our analysis of QST cannot rule out that the population divergence seen is a result of neutral processes and not by divergent selection, however, the QST analysis for plasticity has a low power (see Paper III) and the overall picture based upon matching with the local

30










selection pressures (Paper I, III) and no positive effect of gene flow on plasticity (Paper III) strongly suggests that there is adaptive population divergence in phenotypic plasticity between the populations, and that the driver is the variation in pool drying regimes present on the islands. As we found that, within all populations, the high-plastic individuals are always able to express similar or more extreme phenotypes than the less plastic individuals (with a more canalized development) (Paper IV), it is likely that the presence of plasticity enabled colonization even of the most temporary pools, where the optimal strategy is fast development (Paper I). If plasticity enables survival in these extreme environments (Price et al., 2003; Lande, 2009), standing genetic variation in development time may then accumulate so that evolution could continue in the adaptive direction, i.e. through a change in the mean development time via the Baldwin effect (Baldwin, 1896; WestEberhard, 2005; Crispo, 2007). Such scenario is not unlikely, but has been observed among species of spadefoot toads and parsley frogs. The ancestral state in this lineage has been long development time and plasticity in development time, which has allowed species to adapt to the more ephemeral conditions in North America (GomezMestre, & Buchholz, 2006). The lack of environmental variation on the islands with only temporary pools would suggest that canalisation and not plasticity is the optimal developmental path (Via, & Lande, 1985; Van Tienderen, 1991; Moran, 1994; Sultan, & Spencer, 2002; Leimar et al., 2006) and that plasticity should be reduced due to genetic assimilation (Lande, 2009). However, phenotypic plasticity is still present in these populations (Paper IV), despite the fact that they have a genetically shorter mean development time than tadpoles from islands with permanent pools (Paper I, II, III). As temporary pools are considered not only to be a time stressed environment, but also an environment with high year-to-year variation in drying regimes (Newman, 1992; Brooks, & Hayashi, 2002; Loman, & Claesson, 2003), it is reasonable to think that this temporal variation could select for plasticity in development time. However, we found no relationship between the degree of plasticity in development time and the mean pool drying on the islands (Paper I). That does not mean that selection due to temporal variation in drying regimes is not there, only that it does not vary with pool drying regimes, as predicted (Loman, & Claes


31







son, 2003). Moreover, we found that, even in individuals from the most temporary pools, it is the individuals with high plasticity in development time that can express the fastest development times (Paper IV). This may also be a reason why plasticity is retained in the populations from islands with temporary pools, in spite of the fact that they in general have a small difference in development time among families, compared to families from islands with permanent pools (Paper II). Additionally, because the costs of plasticity are lower in the low-plasticity populations (Paper IV), some plasticity may have been retained, as it is not strongly selected against. Growth, growth rates and developmental thresholds Growth and metamorphic weight During development, the tadpoles are not only developing morphologically and physiologically, they are also increasing in weight. However, the weight and the development of the tadpoles are not independent. By speeding up development via phenotypic plasticity in the artificial drying treatment, the tadpoles are smaller at metamorphosis than the tadpoles from the constant water level treatment, where they remain tadpoles for a longer period of time (Paper I). Thus, there is a trade-off between metamorphic weight and development time, at least on the individual level (Laurila, & Kujasalo, 1999; Laurila et al., 2002; Richter-Boix et al., 2006b). Together with the local adaptation in development time, we also found evidence of divergent selection in metamorphic weight (Paper III). This might be surprising given the mixed evidence for any relationship between pool drying regime and metamorphic weight (Johansson et al., 2005; Paper I, II). However, given the trade-off between development time and metamorphic weight (Laurila, & Kujasalo, 1999), which implies that organisms that speed up their development pays the price of metamorphosing at a smaller size, I interpret the indication of divergent selection on metamorphic weight as a correlated response to selective pressures on development time. This interpretation is further supported by the wider QST confidence interval of metamorphic weight, compared to the confidence interval for development time (Paper III). As metamorphic weight is closely connected to general fitness (Smith, 1987; Berven, 1990; Morey, & Reznick, 2001; Altwegg, & Reyer, 2003) and froglet performance

32










(Ficetola, & de Bernardi, 2006; Richter-Boix et al., 2006b; Johansson, Lederer & Lind, manuscript), it is likely that it also is affected by other factors, independent of development time. An intriguing result is that, although individuals under the artificial drying treatment sped up their development rate (Paper I, II), they had a lower growth rate (increase in mass over time) (Paper I). This result seems to indicate an underlying trade-off between development rate and growth rate when an individual is under time-constrains, but as far as I know, the nature of this trade-off has never been investigated. How small can you be? Given that short development time is an advantage in time stressed environments, one might wonder how fast it is possible to develop. Due to the trade-off between age and size at metamorphosis (Laurila, & Kujasalo, 1999), the question could also be reformulated; how small can one be to successfully complete metamorphosis? A naïve answer, and an answer that is surprisingly common in evolutionary studies, is that, as long as there is genetic variation associated with size at metamorphosis, we would see evolution of this minimum size at metamorphosis. However, many traits are not free to evolve, even with genetic variation, if they are subjected to constraints, and life history transitions are commonly found to be constrained (Leips, & Travis, 1994; Day, & Rowe, 2002; Plaistow et al., 2004). For example, the most well known constrain is probably the minimum birth weight in humans, where children under a certain weight have low survival without the aid of modern healthcare (Whitmore, & Su, 2007). Thus, there is a developmental threshold that the individual has to grow beyond for a successful life history transition to be possible (Wilbur, & Collins, 1973; Day, & Rowe, 2002). As the threshold is determined by strong physiological constrains and not simple life history strategies (Day, & Rowe, 2002), differences in thresholds between lineages must be very uncommon. Therefore, it is surprising that Morey & Reznick (2000, 2004) found substantial variation in the developmental threshold of metamorphosis among three closely related species of Spadefoot toads in North America. The variation was not random, but could be explained by the adaptation of the species to pools of different duration, where the species living in the most ephemeral pools had the fastest development time as well as the lowest developmental threshold (Morey, & Reznick, 2000, 2004). In one 


33







way, this is what would be expected since ephemeral pools are timeconstrained environments and a physiologically lower developmental threshold is reached more quickly. However, if the threshold represents a physiological constraint in development, as assumed (Wilbur, & Collins, 1973; Day, & Rowe, 2002) then the finding that threshold size can indeed evolve challenges this assumption. If developmental thresholds can evolve, it is unclear as to what keeps the developmental threshold high in populations from permanent environments, or why thresholds are present at all. Inspired by the findings of Morey & Reznick (2004), we investigated the potential evolution of developmental thresholds in the island system. This system is well suited to study this, as we have the same selection pressures as in the study of Morey & Reznick (2004), but instead of species we have populations, locally adapted to the degree of pool drying (Paper I, II, III). Thus, we were able to investigate if the developmental threshold is as flexible as development time to divergent selection, or if there indeed is a deep physiological constrain which will not readily evolve in relatively young populations connected by gene flow (Paper III). As there is phenotypic plasticity present in development time (Newman, 1992), (Laurila et al., 2002), just recording the size of an organism when a life history transition occur (as in Wesselingh et al., 1997) is not a valid approach to investigate thresholds, as individuals are free to make the life history transitions at different times after they have passed the threshold, depending upon their life history strategy (Plaistow et al., 2004). The key therefore is to investigate the developmental threshold under conditions where there is only one life history strategy that is possible, and that is achieved by using very low food availability (Day, & Rowe, 2002; Plaistow et al., 2004). When the food level is so low that almost no growth is possible, delaying metamorphosis after the threshold is reached is not an adaptive strategy, as there will be no increase in size at metamorphosis by delaying it. Therefore, the organisms will go through the life history transition at the developmental threshold (Plaistow et al., 2004). Using this methodology, we investigated if there had been evolution of the developmental threshold in our island system, by comparing the threshold in two populations from islands with temporary pools with two islands where the pools are permanent. Unless the threshold is a strong constrain (as suggested by Day, & Rowe, 2002), we predicted that the threshold should have evolved to lower size in populations from islands with temporary pools, following the among-species results of

34










(Morey, & Reznick, 2004). However, if the threshold is tightly linked to physiological constraints in the developmental process, we predicted no such relationship. In accordance with the findings of Morey & Reznick (2004), as well as earlier studies in the island system (Johansson et al., 2005), we found that tadpoles from islands with temporary pools had a genetically shorter development time than tadpoles from islands with permanent pools (Paper II), as a result of divergent natural selection (Paper III). However, we found no difference in the developmental threshold of metamorphosis (Paper II), which contrasts the adaptive evolution of the developmental threshold found among species of Spadefoot toads (Morey, & Reznick, 2004). This finding could actually be interpreted in two ways. Either, there is a benefit of being able to metamorphose at a size below the developmental threshold, but the threshold is such a constraint that an evolutionary change of the threshold is not possible. Alternatively, there is no selection on the threshold as all individuals are easily larger than the threshold size because of higher food availability in nature than in the low-food experimental treatment. When sufficient food is available under laboratory conditions, tadpoles from even the most temporary pools metamorphose at a far higher weight than the threshold weight (Paper I, II, III), suggesting that they are not genetically programmed to metamorphose as soon as it is physiologically possible, even under the artificial pool drying treatment (Paper I, III). Interestingly, in another system I have found population differentiation in the developmental threshold of maturation. In isolated long-term lab populations of soil mites Sancassania berlesei, living in environments with different variation in food availability, I found a lower developmental threshold in a population that often experienced episodes of starvation (Lind, 2005). Thus, it seems that the developmental threshold can indeed evolve among populations, but that strong selection under many generations, and an absence of gene flow is needed.




35







Costs and limits of plasticity Costs of plasticity imply that plastic genotypes pay a fitness cost (DeWitt et al., 1998; Callahan et al., 2008), and are major determinant of whether plasticity or local specialisation will evolve in theoretic models (e.g. Moran, 1992). However, plasticity costs are often not found (Van Buskirk, & Steiner, 2009), but theories of why they are hard to find have never been tested empirically. We found costs of plasticity in populations with high mean plasticity, while populations with low mean plasticity had a cost of canalization (Paper IV), i.e. a cost of being a specialist. This indicates that costs of plasticity increase with the magnitude of plasticity present (as suggested by Van Tienderen, 1991), and may be the reason for why costs of plasticity have been so hard to find in natural populations (e.g. Scheiner, & Berrigan, 1998; Donohue et al., 2000; Weijschedé et al., 2006; Steiner, & Van Buskirk, 2008). This interpretation is also supported by a study where populations of R. temporaria from south and north Sweden were compared, and it was found that the more plastic southern population also had a cost of plasticity, which was not found in the less plastic northern population (Merilä et al., 2004b), which gives no support for the hypothesis that selection removes all costly genotypes in all populations (as suggested by DeWitt et al., 1998; Sultan, & Spencer, 2002; Callahan et al., 2005; Weinig et al., 2006). Instead, our finding suggests that costs of plasticity are not likely to be found in all populations and traits, only in the populations and traits where increased plasticity is selected for. We also found that the fitness cost of plasticity was treatment dependent, and only expressed in the stressful artificial drying treatment (Paper IV). This is not surprising, given that individuals under stress are possibly unable to allocate enough resources both to development and maintenance (Van Tienderen, 1991). Recently, a meta-analysis suggested that this pattern is common in most animal studies (Van Buskirk, & Steiner, 2009). Recently, it has been suggested (Crispo, 2008) that the existence of plasticity costs is a result of gene flow. According to this hypothesis, finding costs of plasticity in some populations would indicate that specialisation and not plasticity would be the optimal strategy in these populations, but that gene flow is working against local specialisation. However, in the few systems where costs of plasticity have been

36










compared among populations, we (Paper IV) and other (Merilä et al., 2004b) have found that costs of plasticity are more commonly present in populations experiencing a high degree of phenotypic plasticity. As we have shown that phenotypic plasticity is locally adaptive in this island system, and that local selection for plasticity overrides the effect of gene flow, I do not find support for the idea that high costs of plasticity indicates that local specialisation is the optimal outcome. Instead, our data confirms with the models of Van Tienderen (1991), in which costs of plasticity are increasing with increased levels of plasticity (Paper IV). In contrast to the concept of plasticity costs, where the plastic genotype can express all phenotypes but has a cost of being plastic, limits of plasticity implies that the plastic genotype is not able to express the extreme traits that a specialist genotype is able to, i.e. a developmental range limit (DeWitt et al., 1998). Although limits of plasticity are not as well studied as costs of plasticity, none of the published works have found any evidence for the hypothesis that plastic individuals cannot express the most extreme traits (DeWitt, 1998; Relyea, 2002a). However, costs of plasticity were not the only cost present in our system, as we also found a significant cost of canalization (i.e. a cost of not being plastic) in the low plasticity populations (Paper IV). This is not an uncommon finding, canalization costs have been found in a number of systems (Dorn et al., 2000; Relyea, 2002a; Weijschedé et al., 2006; Weinig et al., 2006; Avramov et al., 2007; Steiner, & Van Buskirk 2008), and are found in almost the same number of studies and traits as plasticity costs (Van Buskirk, & Steiner, 2009). As will be discussed below, this can have major implications for our understanding of the evolution of phenotypic plasticity and specialisation. Integrating plasticity and canalization costs I argue that costs of canalization must be included in models of the evolution of phenotypic plasticity. These costs are likely to have major implications for the evolution of phenotypic plasticity: whether plasticity as opposed to a canalized development is to evolve depends not only on the cost of plasticity but also on the opposing cost of canalization. First, including a cost of canalization makes plasticity a more likely outcome under a larger range of parameter values in models of the 


37







evolution of phenotypic plasticity (Figure 4), and second, this might be the reason that we often not find costs of plasticity in natural populations: if the cost of plasticity and the cost of canalization are equal, we will not find any significant cost of plasticity, since we are measuring the cost of plasticity relative to the cost of canalization.

Figure 3. Combinations of a (the accuracy of the environmental matching) and r (the frequency of environment 1) for which plasticity or canalization is favoured under spatial environmental heterogeneity with two different environments. The bold lines represent the parameter space above which plasticity is favoured and below which canalization is favoured under different cost of plasticity / cost of canalization (Cp/Cc) ratios, in an extension of the optimality model of Moran (1992) (Lind, unpublished).

If the evolution of phenotypic plasticity is influenced not only by costs of plasticity (e.g. Moran, 1992) but also by the costs of maintaining a canalized phenotype over a range of environmental conditions (Poulton, & Winn, 2002), this has major implications for how we are measuring costs of plasticity in natural populations. The traditional multiple regression of fitness as a function of the trait mean and plasticity (DeWitt et al., 1998) was developed assuming that plastic individuals could only perform worse than non plastic ones, the opposite pattern was never considered. Thus, the regression was

38










designed to estimate if cost of plasticity were present, assuming that no other balancing costs were paid by the non-plastic genotypes. However, if the more plastic individuals pays a cost of being plastic and the more canalized individuals pays a cost of canalization, a non significant regression coefficient of the plasticity term should not be interpreted as a failure to find plasticity costs (as in DeWitt et al., 1998). Instead, a non-significant coefficient of the plasticity term means that the costs of plasticity and the costs of homeostasis were of similar size in the trait and environment combination. Moreover, the analysis gives no hint of the magnitude of these costs, they could ether be small or of substantial magnitude, all we know is that they are not significantly unbalanced. Thus, the standard analysis for estimating costs of plasticity (DeWitt et al., 1998) has a low power of detecting costs, since many costs of plasticity will be hidden under the opposing costs of canalization. Moreover, the analysis cannot be used to determine the magnitude of the costs of phenotypic plasticity, since the regression coefficient of the plasticity term only estimates how much one type of cost is influencing fitness relative to the influence of the other type of cost. Therefore, I argue that the standard methodology for detecting plasticity costs is of limited use when studying natural populations. Environmental cues In addition to plasticity costs and limits, another factor that heavily influences whether plasticity will evolve or not is the accuracy of the environmental cues (Moran, 1992; Sultan, & Spencer, 2002). If the cues that the organism uses for identifying which environment it is located in is uninformative, a specialist or generalist genotype is selected for. Moreover, I would argue that being able to match the phenotype to the environment also means that there is scope for making mistakes, which has nothing to do with the accuracy of the cues. Those mistakes are only present in plastic genotypes, as specialists, by having a canalized development, are not free to make mistakes. In a constant environment, individuals with canalized developmental paths will always hit the optimum phenotype, while the plastic genotypes will now and then miss it, simply due to erogenous interpretation of the cues.




39







To my knowledge, no one has yet studied the accuracy of the plastic response experimentally, and I realise that this might not be a simple task. As the response variable is not the within family mean, but the within family variation in the trait as a function of the plasticity of the trait, care must be taken to eliminate all causes of within family variation. Ideally, a crossbreeding design would be the ultimate experimental setup. However, despite the complex experiments needed, the accuracy of the plastic response is identified as a potentially very important factor influencing the evolution of phenotypic plasticity (Moran, 1992; Sultan, & Spencer, 2002) and experimental studies are needed in order to examine its role for shaping the variation in plasticity we observe in nature. Rapid local adaptation – implications for conservation The finding that populations often are more different in quantitative traits than in neutral markers (Leinonen et al., 2008; Paper III) has important implications for conservation biology. Populations that have separate evolutionary histories have a high conservation value because of their distinctiveness and are known as Evolutionary Significant Units (ESUs) (Crandall et al., 2000). With the development of new molecular methods and their incorporation in conservation genetics, molecular methods have become the most used method for assessing ESUs (DeSalle, & Amato, 2004), as well as in conservation genetics in general (Haig, 1998). The incorporation of molecular methods in the toolbox of conservation biologists has in many ways revolutionised the field. Molecular methods are now widely used tools in assessing population structures and finding ESUs, because it, in comparison to quantitative genetic methods, is fast (DeSalle, & Amato, 2004) and does not need extensive sampling (Storfer, 1996). This is often very important, since conservation biology is a “crisis discipline” (DeSalle, & Amato, 2004), where decisions have to be rapid and often based upon currently available data Thus, especially since DNA material now can be analysed from very small amounts of tissue, even from feathers and faeces, an experimental procedure based upon adaptive differences between populations seems unnecessary complicated if molecular methods can an-

40










swer the proposed questions in an equally good manner. However, there are uprising doubts that molecular methods give the best answer to the questions conservation biologists want to answer. Because neutral molecular markers, by definition, are selectively neutral, they have no intrinsic value, which calls into question if it can be used to assess whether the differences among populations contribute to increased persistence of the network of populations that the species distribution range consists of (Pearman, 2001). One important question is whether differentiation in neutral markers can serve as a shortcut that indicates population subdivision in morphological and life history traits under selection (Storfer, 1996; Pearman, 2001; Bekessy et al., 2003), in other words, if FST can be used to predict QST. As discussed above, the value of QST often exceeds the value of FST in natural populations, and even though there is a positive relationship between the two measurements (Leinonen et al., 2008) the relationship has low predictive power. Especially for low FST values, QST can obtain almost any value (Leinonen et al., 2008). Moreover, QST can even differ among traits within a single species (e.g. Edmands, & Harrison, 2003). Sometimes, neutral genetic markers might not just fail to detect a pattern, but can also give a complete opposite population structuring compared to when quantitative genetic differentiation is assessed. When surveying populations of the vulnerable monkey puzzle tree Araucaria araucana across its entire distribution range in South America (Bekessy et al., 2003), neutral genetic markers failed to detect the quantitative genetic divergence in drought tolerance across the Andean mountain range. Using neutral markers as tools to aid reforestation projects would then not be optimal, since the plants clustered together would differ in their response to drought. Crandall et al. (2000) suggested not basing the ESUs on monophyly but on exchangeability - both ecological and genetic. For two populations to be ecologically exchangeable, individuals can be moved between the populations and still occupy the same niche or selective regime. Crandall argues that it is only rejection of the null hypothesis of ecological exchangeability that merits a population conservation status. In other words, ESUs shall be based upon adaptive differences in phenotypic traits, not upon genetic distinctiveness.




41







Concluding words I found local adaptation in development time and plasticity in development time in island populations of R. temporaria. The selection pressures driving this adaptation were found to be the degree and variation in pool drying regimes of the pools in which the frogs were breeding. Given the young age of the populations (less than 300 generations), and the fact that they are connected by gene flow, the speed of this local adaptation is quite remarkable. Moreover, I have also shown that fitness costs of phenotypic plasticity are only found in populations with genotypes expressing high levels of phenotypic plasticity, while in populations with low-plastic genotypes, I find costs of canalisation. This suggests that costs of plasticity increase with increased level of plasticity in the population, and that might be a reason why costs of plasticity are hard to detect. My studies show the uniqueness of the island populations, both in terms of genetic distinctness and especially in their different adaptations to the prevailing environmental conditions. This is important for conservation, as population differentiation in the traits under selection can be many times larger than the differentiation inferred from genetic markers.

ACKNOWLEDGEMENTS I thank Frank Johansson, Melanie Monroe, Carolina Tegström and Viktoria Tengel for valuable comments and suggestions on earlier versions of this manuscript.

REFERENCES Ahlén, I., Andrén, C. & Nilson, G. 1995. Sveriges grodor, ödlor och ormar: fältguide och faktasamling, . Naturskyddsföreningen. Stockholm Almfelt, J. E. M. (2005) Plastic acceleration of metamorphosis in the common frog (Rana temporaria) in response to pool drying: the effect of geographical isolation. M. Sc. Thesis. Umeå University Alpert, P. & Simms, E. L. 2002. The relative advantages of plasticity and fixity in different environments: when is it good for a plant to adjust? Evolutionary Ecology 16: 285-297.

42










Altwegg, R. & Reyer, H. U. 2003. Patterns of natural selection on size at metamorphosis in water frogs. Evolution 57: 872-882. Avramov, S., Pemac, D. & Tucić, B. 2007. Phenotypic plasticity in response to an irradiance gradient in Iris pumila: adaptive value and evolutionary constraints. Plant Ecology 190: 275-290. Baldwin, J. M. 1896. A new factor in evolution. American Naturalist 30: 441-451. Bekessy, S. A., Ennos, R. A., Burgman, M. A., Newton, A. C. & Ades, P. K. 2003. Neutral DNA markers fail to detect genetic divergence in an ecologically important trait. Biological Conservation 110: 267-275. Bergek, S. & Björklund, M. 2009. Genetic and morphological divergence reveals local subdivision of perch (Perca fluviatilis L.). Biological Journal of the Linnean Society 96: 746-758. Berlin, S., Merilä, J. & Ellegren, H. 2000. Isolation and characterization of polymorphic microsatellite loci in the common frog, Rana temporaria. Molecular Ecology 9: 1938-1939. Berven, K. A. 1990. Factors affecting population fluctuations in larval and adult stages of the wood frog (Rana sylvatica). Ecology 71: 1599-1608. Bradshaw, A. 1965. Evolutionary significance of phenotypic plasticity in plants. Advances in Genetics 13: 115-155. Brooks, R. T. & Hayashi, M. 2002. Depth-area-volume and hydroperiod relationships of ephemeral (vernal) forest pools in southern New England. Wetlands 22: 247-255. Callahan, H. S., Dhanoolal, N. & Ungerer, M. C. 2005. Plasticity genes and plasticity costs: a new approach using an Arabidopsis recombinant inbred population. New Phytologist 166: 129140. Callahan, H. S., Maughan, H. & Steiner, U. K. 2008. Phenotypic plasticity, costs of phenotypes, and costs of plasticity. Annals of the New York Academy of Sciences 1133: 44-66. Carson, H. L. & Kaneshiro, K. Y. 1976. Drosophila of Hawaii: systematics and ecological genetics. Annual Review of Ecology and Systematics 7: 311-345. Comps, B., Gomory, D., Letouzey, J., Thiebaut, B. & Petit, R. J. 2001. Diverging trends between heterozygosity and allelic richness during postglacial colonization in the European beech. Genetics 157: 389-397.




43







Crandall, K. A., Bininda-Emonds, O. R. P., Mace, G. M. & Wayne, R. K. 2000. Considering evolutionary processes in conservation biology. Trends in Ecology & Evolution 15: 290-295. Crispo, E. 2008. Modifying effects of phenotypic plasticity on interactions among natural selection, adaptation and gene flow. Journal of Evolutionary Biology 21: 1460-1469. Crispo, E. 2007. The Baldwin effect and genetic assimilation: revisiting two mechanisms of evolutionary change mediated by phenotypic plasticity. Evolution 61: 2469-2479. Darwin, C. 1859. On the origin of species by means of natural selection, or the preservation of favoured races in the struggle for life, 1st ed. John Murray, London. Day, T. & Rowe, L. 2002. Developmental thresholds and the evolution of reaction norms for age and size at life-history transitions. American Naturalist 159: 338-350. Delasalle, V. A. & Blum, S. 1994. Variation in germination and survival among families of Sagittaria latifolia in response to salinity and temperature. Internation Journal of Plant Sciences 155: 187-195. Denver, R. J., Mirhadi, N. & Phillips, M. 1998. Adaptive plasticity in amphibian metamorphosis: Response of Scaphiopus hammondii tadpoles to habitat desiccation. Ecology 79: 1859-1872. DeSalle, R. & Amato, G. 2004. The expansion of conservation genetics. Nature Reviews Genetics 5: 702-712. DeWitt, T. J. 1998. Costs and limits of phenotypic plasticity: tests with predator-induced morphology and life history in a freshwater snail. Journal of Evolutionary Biology 11: 465-480. DeWitt, T. J., Sih, A. & Wilson D S 1998. Costs and limits of phenotypic plasticity. Trends in Ecology & Evolution 13: 77-81. Dingemanse, N. J., Both, C., van Noordwijk, A. J., Rutten, A. L. & Drent, P. J. 2003. Natal dispersal and personalities in great tits (Parus major). Proceedings of the Royal Society of London. Series B: Biological Sciences 270: 741-747. Donohue, K., Messiqua, D., Pyle, E. H., Heschel, M. S. & Schmitt, J. 2000. Evidence of adaptive divergence in plasticity: densityand site-dependent selection on shade-avoidance responses in Impatiens capenses. Evolution 54: 1956-1968. Dorn, L. A., Pyle, E. H. & Schmitt, J. 2000. Plasticity to light cues and resources in Arabidopsis thaliana: testing for adaptive value and costs. Evolution 54: 1982-1994.

44










Edmands, S. & Harrison, J. S. 2003. Molecular and quantitative trait variation within and among populations of the intertidal copepod Tigriopus californicus. Evolution 57: 2277-2285. Elmberg, J. 1990. Long-term survival, length of breeding season and operational sex ratio in a boreal population of common frogs Rana temporaria. Canadian Journal of Zoology 68: 121-127. Elmberg, J. 1991. Ovarian cyclicity and fecundity in boreal common frogs Rana temporaria L. along a climatic gradient. Functional Ecology 5: 340-350. Emerson, B. C. 2002. Evolution on oceanic islands: molecular phylogenetic approaches to understanding pattern and process. Molecular Ecology 11: 951-966. Ernande, B. & Dieckmann, U. 2004. The evolution of phenotypic plasticity in spatially structured environments: implications of intraspecific competition, plasticity costs and environmental characteristics. Journal of Evolutionary Biology 17: 613-628. Ernande, B., Dieckmann, U. & Heino, M. 2004. Adaptive changes in harvested populations: plasticity and evolution of age and size at maturation. Proceedings of the Royal Society of London Series B-Biological Sciences 271: 415-423. Falconer, D. S. 1952. The problem of environment and selection. The American Naturalist 86: 293. Ficetola, F. & de Bernardi, F. 2006. Trade-off between larval development rate and post-metamorphic traits in the frog Rana latastei. Evolutionary Ecology 20: 143-158. Fraser, D. F., Gilliam, J. F., Daley, M. J., Le, A. N. & Skalski, G. T. 2001. Explaining leptokurtic movement distributions: intrapopulation variation in boldness and exploration. The American Naturalist 158: 124-135. Futuyma, D. J. 1998. Evolutionary Biology, . Sinauer Associates, Massachusetts, U.S.A, . Gabriel, W. & Lynch, M. 1992. The selective advantage of reaction norms for environmental tolerance. Journal of Evolutionary Biology 5: 41-59. Gasc, J., Cabela, A., Crnobrnja-Isailovic, J., Dolmen, D., Grossenbacher, K., Haffner, P., Lescure, J., Martens, H., MartÝnez Rica, J. P., Maurin, H., Oliveira, M. E., Sofianidou, T. S., Veith, M. & Zuiderwijk, A. 1997. Atlas of amphibians and reptiles in Europe, Muséum National d'Histoire Naturelle & Service du Petrimone Naturel, Paris




45







Gavrilets, S. 2003. Perspective: Models of speciation: What have we learned in 40 years? Evolution 57: 2197-2215. Gervasi, S. S. & Foufopoulos, J. 2008. Costs of plasticity: responses to desiccation decrease post-metamorphic immune function in a pond-breeding amphibian. Functional Ecology 22: 100-108. Gomez-Mestre, I. & Buchholz, D. R. 2006. Developmental plasticity mirrors differences among taxa in spadefoot toads linking plasticity and diversity. Proceedings of the National Academy of Sciences of the United States of America 103: 19021-19026. Gomulkiewicz, R. & Kirkpatrick, M. 1992. Quantitative genetics and the evolution of reaction norms. Evolution 46: 390-411. Gosner, K. L. 1960. A simplified table for staging anuran embryos and larvae with notes on identification. Herpetologica 16: 183190. Gotthard, K. & Nylin, S. 1995. Adaptive plasticity and plasticity as an adaptation: a selective review of plasticity in animal morphology and life history. Oikos 74: 3-17. Grant, P. R. & Grant, B. R. 2002. Unpredictable evolution in a 30year study of Darwin's finches. Science 296: 707-711. Haig, S. M. 1998. Molecular contributions to conservation. Ecology 79: 413-425. Hall, D., Luquez, V., Garcia, V. M., St Onge, K. R., Jansson, S. & Ingvarsson, P. K. 2007. Adaptive population differentiation in phenology across a latitudinal gradient in european aspen (Populus tremula, L.): a comparison of neutral markers, candidate genes and phenotypic traits. Evolution 61: 2849-2860. Hendry, A. P. & Taylor, E. B. 2004. How much of the variation in adaptive divergence can be explained by gene flow? - An evaluation using lake-stream stickleback pairs. Evolution 58: 2319-2331. Hollander, J. 2008. Testing the grain-size model for the evolution of phenotypic plasticity. Evolution 62: 1381-1389. Huber, H., Kane, N. C., Heschel, M. S., von Wettberg, E. J. & Banta, J. 2004. Frequency and microevolutional pattern of selection on plastic shade-avoidance traits in a natural population of Impatiens capensis. American Naturalist 163: 548-563. Huxley, J. 1942. Evolution: The Modern Synthesis, . Allen & Unwin, London. Johansson, F., Hjelm, J. & Giles, B. E. 2005. Life history and morphology of Rana temporaria in response to pool permanence. Evolutionary Ecology Research 7: 1025-1038.

46










Johansson, M., Primmer, C. R. & Merilä, J. 2006. History vs. current demography: explaining the genetic population structure of the common frog (Rana temporaria). Molecular Ecology 15: 975983. Keyghobadi, N., Roland, J. & Strobeck, C. 2005. Genetic differentiation and gene flow among populations of the alpine butterfly, Parnassius smintheus, vary with landscape connectivity. Molecular Ecology 14: 1897-1909. Kruuk, L. E. B. & Hill, W. G. 2008. Evolutionary dynamics of wild populations: the use of long-term pedigree data. Proceedings of the Royal Society of London Series B-Biological Sciences . Lande, R. 2009. Adaptation to an extraordinary environment by evolution of phenotypic plasticity and genetic assimilation. Journal of Evolutionary Biology 22: 1435-1446. Latta, R. G. 2004a. Gene flow, adaptive population divergence and comparative population structure across loci. New Phytologist 161: 51-58. Latta, R. G. 2004b. Relating processes to patterns of genetic variation across landscapes. Forest Ecology and Management 197: 91-102. Laugen, A. T., Laurila, A. & Merilä, J. 2002. Maternal and genetic contributions to geographical variation in Rana temporaria larval life-history traits. Biological Journal of the Linnean Society 76: 61-70. Laugen, A. T., Laurila, A., Räsänen, K. & Merilä, J. 2003. Latitudinal countergradient variation in the common frog (Rana temporaria) development rates - evidence for local adaptation. Journal of Evolutionary Biology 16: 996-1005. Laurila, A., Karttunen, S. & Merilä, J. 2002. Adaptive phenotypic plasticity and genetics of larval life histories in two Rana temporaria populations. Evolution 56: 617-627. Laurila, A. & Kujasalo, J. 1999. Habitat duration, predation risk and phenotypic plasticity in common frog (Rana temporaria) tadpoles. Journal of Animal Ecology 68: 1123-1132. Leimar, O., Hammerstein, P. & Van Dooren, T. J. M. 2006. A new perspective on developmental plasticity and the principles of adaptive morph determination. American Naturalist 167: 367376. Leinonen, T., O’Hara, R. B., Cano, J. M. & Merilä, J. 2008. Comparative studies of quantitative trait and neutral marker divergence: a meta-analysis. Journal of Evolutionary Biology 21: 1-17.




47







Leips, J. & Travis, J. 1994. Metamorphic responses to changing food levels in two species of hylid frogs. Ecology 75: 1345-1356. Lesbarrères, D., Primmer, C. R., Laurila, A. & Merilä, J. 2005. Environmental and population dependency of genetic variabilityfitness correlations in Rana temporaria. Molecular Ecology 14: 311-323. Lind, M. I. (2005) Environmental conditions influencing the phenotypic plasticity in age and size at maturity in an experimental system. Master Thesis, University of Aberdeen Loman, J. & Claesson, D. 2003. Plastic response to pond drying in tadpoles Rana temporaria: tests of cost models. Evolutionary Ecology Research 5: 179-194. Lynch, M., Pfrender, M., Spitze, K., Lehman, N., Hicks, J., Allen, D., Latta, L., Ottene, M., Bogue, F. & Colbourne, J. 1999. The quantitative and molecular genetic architecture of a subdivided species. Evolution 53: 100-110. Lynch, M. & Walsh, B. 1998. Genetics and analysis of quantitative traits, Sinauer Associates, Inc, Sunderland Marshall, D. J. & Uller, T. 2007. When is a maternal effect adaptive. Oikos 116: 1957-1963. McKay, J. K. & Latta, R. G. 2002. Adaptive population divergence: markers, QTL and traits. Trends in Ecology & Evolution 17: 285-291. Mendel, J. G. 1866. Versuche über Pflanzenhybriden. Verhandlungen des naturforschenden Vereines in Brünn, 3-47. Merilä, J. & Crnokrak, P. 2001. Comparison of genetic differentiation at marker loci and quantitative traits. Journal of Evolutionary Biology 14: 892-903. Merilä, J., Laurila, A., Laugen, A. T. & Rõsõnen, K. 2004a. Heads or tails? - Variation in tadpole body proportions in response to temperature and food stress. Evolutionary Ecology Research 6: 727-738. Merilä, J., Laurila, A. & Lindgren, B. 2004b. Variation in the degree and costs of adaptive phenotypic plasticity among Rana temporaria populations. Journal of Evolutionary Biology 17: 11321140. Merilä, J., Laurila, A., Pahkala, M., Räsänen, K. & Laugen, A. T. 2000a. Adaptive phenotypic plasticity in timing of metamorphosis in the common frog Rana temporaria. Ecoscience 7: 1824.

48










Merilä, J., Laurila, A., Laugen, A. T., Räsänen, K. & Pahkala, M. 2000b. Plasticity in age and size at metamorphosis in Rana temporaria - comparison of high and low latitude populations. Ecography 23: 457-465. Miaud, C., Guyétant, R. & Elmberg, J. 1999. Variations in life-history traits in the common frog Rana temporaria (Amphibia: Anura): a literature review and new data from the French Alps. Journal of Zoology 249: 61-73. Moran, N. A. 1994. Adaptation and constraint in the complex life cycles of animals. Annual Review of Ecology, Evolution, and Systematics 25: 573-600. Moran, N. A. 1992. The evolutionary maintenance of alternative phenotypes. American Naturalist 139: 971-989. Morey, S. & Reznick, D. 2001. Effects of larval density on postmetamorphic spadefoot toads (Spea hammondii). Ecology 82: 510-522. Morey, S. & Reznick, D. N. 2000. A comparative analysis of plasticity in larval development in three species of spadefoot toads. Ecology 81: 1736-1749. Morey, S. R. & Reznick, D. N. 2004. The relationship between habitat permanence and larval development in California spadefoot toads: field and laboratory comparisons of developmental plasticity. Oikos 104: 172-190. Mousseau, T. A. & Fox, C. W. 1998. The adaptive significance of maternal effects. Trends in Ecology & Evolution 13: 403-407. Newman, R. A. 1992. Adaptive plasticity in amphibian metamorphosis: what type of phenotypic variation is adaptive, and what are the costs of such plasticity. BioScience 42: 671-678. Niinemets, Ü., Valladares, F. & Ceulemans, R. 2003. Leaf-level phenotypic variability and plasticity of invasive Rhododendron ponticum and non-invasive Ilex aquifolium co-occurring at two contrasting European sites. Plant, Cell & Environment 26: 941956. O'Hara, R. B. & Merilä, J. 2005. Bias and precision in QST estimates: problems and some solutions. Genetics 171: 1331-1339. Palo, J. U., O'Hara, R. B., Laugen, A. T., Laurila, A., Primmer, C. R. & Merilä, J. 2003. Latitudinal divergence of common frog (Rana temporaria) life history traits by natural selection: evidence from a comparison of molecular and quantitative genetic data. Molecular Ecology 12: 1963-1978.




49







Palo, J. U., Schmeller, D. S., Laurila, A., Primmer, C. R., Kuzmin, S. L. & Merilä, J. 2004. High degree of population subdivision in a widespread amphibian. Molecular Ecology 13: 2631-2644. Pearman, P. B. 2001. Conservation value of independently evolving units: Sacred cow or testable hypothesis? Conservation Biology 15: 780-783. Pidancier, N., Gauthier, P., Miquel, C. & Pompanon, F. 2002. Polymorphic microsatellite DNA loci identified in the common frog (Rana temporaria, Amphibia, Ranidae). Molecular Ecology Notes 2: 304-305. Pigliucci, M. 2005. Evolution of phenotypic plasticity: where are we going now? Trends in Ecology & Evolution 20: 481-486. Pigliucci, M. 1997. Ontogenetic phenotypic plasticity during the reproductive phase in Arabidopsis thaliana (Brassicaceae). American Journal of Botany 84: 887-895. Pigliucci, M. & Schmitt, J. 1999. Genes affecting phenotypic plasticity in Arabidopsis: pleiotropic effects and reproductive fitness of photomorphogenic mutants. Journal of Evolutionary Biology 12: 551-562. Pigliucci, M. 2007. Do we need an extended evolutionary synthesis? Evolution 61: 2743-2749. Plaistow, S. J., Lapsley, C. T., Beckerman, A. P. & Benton, T. G. 2004. Age and size at maturity: sex, environmental variability and developmental thresholds. Proceedings of the Royal Society of London Series B-Biological Sciences 271: 919-924. Poulton, J. & Winn, A. A. 2002. Costs of canalization and plasticity in response to neighbors in Brassica rapa. Plant Species Biology 17: 109-118. Price, T. D., Qvarnström, A. & Irwin, D. E. 2003. The role of phenotypic plasticity in driving genetic evolution. Proceedings of the Royal Society of London Series B-Biological Sciences 270: 1433-1440. Pujol, B., Wilson, A. J., Ross, R. I. C. & Pannell, J. R. 2008. Are QstFst comparisons for natural populations meaningful? Molecular Ecology 17: 4782-4785. Räsänen, K. & Kruuk, L. E. B. 2007. Maternal effects and evolution at ecological time-scales. Functional Ecology 21: 408-421. Reed, D. H. & Frankham, R. 2001. How closely correlated are molecular and quantitative measures of genetic variation? A meta-analysis. Evolution 55: 1095-1103. Relyea, R. A. 2002a. Costs of phenotypic plasticity. American Naturalist 159: 272-282.

50










Relyea, R. A. 2002b. Local population differences in phenotypic plasticity: predator-induced changes in wood frog tadpoles. Ecological Monographs 72: 77-93. Richter-Boix, A., Llorente, G. A. & Montori, A. 2006a. A comparative analysis of the adaptive developmental plasticity hypothesis in six Mediterranean anuran species along a pond permanency gradient. Evolutionary Ecology Research 8: 1139-1154. Richter-Boix, A., Llorente, G. A. & Montori, A. 2006b. Effects of phenotypic plasticity on post-metamorphic traits during premetamorphic stages in the anuran Pelodytes punctatus. Evolutionary Ecology Research 8: 309-320. Rowe, G. & Beebee, T. J. C. 2001. Polymerase chain reaction primers for microsatellite loci in the common frog Rana temporaria. Molecular Ecology Notes 1: 6-7. Savolainen, V., Anstett, M., Lexar, C., Hutton, I., Clarkson, J., Norup, M., Powell, M., Springate, D., Salamin, N. & Baker, W. 2006. Sympatric speciation in palms on an oceanic island. Nature 441: 210-213. Scheiner, S. M. 1993. Plasticity as a selectable trait: reply to Via. American Naturalist 142: 371-373. Scheiner, S. M. & Berrigan, D. 1998. The genetics of phenotypic plasticity. VIII. The cost of plasticity in Daphnia pulex. Evolution 52: 368-378. Scheiner, S. M. 1998. The genetics of phenotypic plasticity. VII. Evolution in a spatially-structured environment. Journal of Evolutionary Biology 11: 303-320. Schlichting, C. D. 1986. The evolution of phenotypic plasticity in plants. Annual Review of Ecology and Systematics 17: 667-693. Schlichting, C. D. & Pigliucci, M. 1993. Control of phenotypic plasticity via regulatory genes. American Naturalist 142: 336-370. Schluter, D. 2001. Ecology and the origin of species. Trends in Ecology & Evolution 16: 372-380. Seppä, P. & Laurila, A. 1999. Genetic structure of island populations of the anurans Rana temporaria and Bufo bufo. Heredity 82: 309317. Sexton, J. P., McKay, J. K. & Sala, A. 2002. Plasticity and genetic diversity may allow saltcedar to invade cold climates in North America. Ecological Applications 12: 1652-1660.




51







Skelly, D. K. & Werner, E. E. 1990. Behavioral and life-historical responses of larval american toads to an odonate predator. Ecology 71: 2313-2322. Smith, D. C. 1987. Adult recruitment in chorus frogs: effects of size and date at metamorphosis. Ecology 68: 344-350. Spitze, K. 1993. Population structure in Daphnia obtusa: quantitative genetic and allozymic variation. Genetics 135: 367-374. Ståhlberg, F., Olsson, M. & Uller, T. 2001. Population divergence of developmental thermal optima in Swedish common frogs, Rana temporaria. Journal of Evolutionary Biology 14: 755-762. Steiner, U. K. & Van Buskirk, J. 2008. Environmental stress and the costs of whole-organism phenotypic plasticity in tadpoles. Journal of Evolutionary Biology 21: 97-103. Steinger, T., Roy, B. A. & Stanton, M. L. 2003. Evolution in stressful environments II: adaptive value and costs of plasticity in response to low light in Sinapis arvensis. Journal of Evolutionary Biology 16: 313-323. Storfer, A. 1996. Quantitative genetics: a promising approach for the assessment of genetic variation in endangered species. Trends in Ecology & Evolution 11: 343-348. Sultan, S. E. & Spencer, H. G. 2002. Metapopulation structure favours plasticity over local adaptation. American Naturalist 160: 271-283. Tollrian, R. & Dodson, S. I. (1999) The ecology and evolution of inducable defenses. In: The ecology and evolution of inducable defenses, (Tollrian, R. & Harvell, C. D., eds). Princeton University Press, Princeton, New Jersey. Van Buskirk, J. & Arioli, M. 2005. Habitat specialization and adaptive phenotypic divergence of anuran populations. Journal of Evolutionary Biology 18: 596-608. Van Buskirk, J. & Relyea, R. A. 1998. Selection for phenotypic plasticity in Rana sylvatica tadpoles. Biological Journal of the Linnean Society 65: 301-328. Van Buskirk, J. & Steiner, U. K. 2009. The fitness costs of developmental canalization and plasticity. Journal of Evolutionary Biology 22: 852-860. Van Tienderen, P. H. 1997. Generalists, specialists, and the evolution of phenotypic plasticity in sympatric populations of distinct species. Evolution 51: 1372-1380.

52










Van Tienderen, P. H. 1991. Evolution of generalists and specialists in spatially heterogeneous environments. Evolution 45: 13171331. Via, S. 1993. Adaptive phenotypic plasticity: target or by-product of selection in a variable environment? American Naturalist 142: 352-365. Via, S. & Lande, R. 1985. Genotype-environment interaction and the evolution of phenotypic plasticity. Evolution 39: 505-522. Vos, C. C., Antonisse-De Jong, A. G., Goedhart, P. W. & Smulders, M. J. M. 2001. Genetic similarity as a measure for connectivity between fragmented populations of the moor frog (Rana arvalis). Heredity 86: 598-608. Weijschedé, J., Martínková, J., de Kroon, H. & Huber, H. 2006. Shade avoidance in Trifolium repens: costs and benefits of plasticity in petiole length and leaf size. New Phytologist 172: 655666. Weinig, C., Johnston, J., German, Z. M. & Demink, L. M. 2006. Local and global costs of adaptive plasticity to density in Arabidopsis thaliana. American Naturalist 167: 826-836. Weinreich, F. (2007) Local adaptation in tadpole development at different pool drying regimes. Master Thesis. Umeå University Wellborn, G. A., Skelly, D. K. & Werner, E. E. 1996. Mechanims creating community structure across a freshwater habitat gradient. Annual Review of Ecology, Evolution, and Systematics 27: 337-363. Werner, E. E., Skelly, D. K., Relyea, R. A. & Yurewicz, K. L. 2007. Amphibian species richness across environmental gradients. Oikos 116: 1697-1712. Wesselingh, R. E., Klinkhamer, P. G. L. & Boorman, L. A. 1997. Threshold size for flowering in different habitats: effects of size-dependent growth and survival. Ecology 78: 2118-2132. West-Eberhard, M. J. 2005. Developmental plasticity and the origin of species differences. Proceedings of the National Academy of Sciences of the United States of America 102: 6543-6549. West-Eberhard, M. J. 2003. Developmental Plasticity and Evolution, . Oxford University Press Inc, USA Whitlock, M. C. 2008. Evolutionary inference from Qst. Molecular Ecology 17: 1885-1896. Whitlock, M. C. 1999. Neutral additive genetic variance in a metapopulation. Genetical Research 74: 215-221. 


53







Whitmore, G. A. & Su, Y. 2007. Modeling low birth weights using threshold regression: results for U. S. birth data. Lifetime Data Analysis 13: 161-190. Widmer, A. & Lexer, C. 2001. Glacial refugia: sanctuaries for allelic richness, but not for gene diversity. Trends in Ecology & Evolution 16: 267-269. Wilbur, H. M. & Collins, J. P. 1973. Ecological aspects of amphibian metamorphosis. Science 182: 1305-1314. Woltereck, R. 1909. Weitere experimentelle untersüchungen über artveränderung, speziell über das wesen quantitativer artunderschiede bei daphniden. Verhandlungen der Deutschen Zoologischen Gesellschaft 19: 110-172. Wright, S. 1943. Isolation by distance. Genetics 28: 114-138. Wright, S. 1951. The genetical structure of populations. Annual Eugenics 15: 323-354. Yeh, P. J. & Price, T. D. 2004. Adaptive phenotypic plasticity and the successful colonization of a novel environment. American Naturalist 164: 531-542.

54










Words of Thanks The First of all, I’ll take the opportunity to thank my supervisor Frank Johansson. We’ve had a great collaboration during these years, we tend to see science in the same way and all our discussions are very prestige-free, which makes working and writing together easy and very enjoyable! I was very hesitant to move up to the dark and cold northern Sweden, but after the interview and our discussions I realised that this was the place to be. I have never regretted it. Pelle Ingvarsson has been my statistical and genetics-mentor, not to mention my guide into the world of geeky computer-related stuff. Barbara Giles, thank for all time you’ve spend reading through my manuscripts and discuss science, you’ve been like a third supervisor for me! Brad Anholt, Locke Rowe and Anssi Laurila have all given helpful comments on my manuscripts, and Anssi, Olle Leimar and Ivan Gomez-Mestre have given me a lot of inspiration and new ideas when discussing science on various conferences. Research here in Umeå would have been impossible without all those who have helped me out in the lab. My Master-students Josephine, Frida, Friederike and Baptiste have all been great company in the lab, and the best of luck with your future careers! David and Carin have been my guides in the DNA-lab, and have helped me solve all kind of problems. Helena, you’ve been the best of room-mates, and although your advices often are sort of elvish, you are a never-ceasing source of knowledge and especially opinions! Tomas; thanks for all the fun and for the cool research we’ve done! Tadpole personalities are a bit sexier than the stuff in this thesis. Viktor, I wish I had just a fraction of your knowledge about species of all kinds! Thanks Folmer, for all your interesting ideas and Nina, Xiao-Ru, Xiao-Fei and all others for all more or less scientific discussions in the journal club. Sometimes, I wonder how I ended up here, and it’s a long and winding story. Thanks to Mom and Dad, I’ve always been out in nature, but I think my real interest in biology started when birdwatching with Lennart as a kid. I remember being so impressed by your knowledge and I wanted to learn more about nature. My interest in evolutionary 


55







biology started with the lectures and discussion by Bo Widerberg at Lundby Gymnasium, you are probably the one to ”blame” that I ended up in this field. At Göteborg University, Regina Fritsche, Hans Blanck, Staffan Andersson, Conny Liljenberg and Ulf Molau made me interested in research. Anna Stina Sandelius and Mats Andersson let me in to the lab, and taught me how to carry out independent research. I learned a lot during that half-year, even though I never became a membrane physiologist. I thank Ian Patterson for enrolling me in the MSc in Ecology at University of Aberdeen, it was the best year I’ve had at Uni, and I also found the missing part of my soul in Scotland. Tim Benton introduced me to life history evolution (and to the mitey world), which was the very reason why I ended up in Umeå (which should not be interpreted as if I was trying to get as far away from the mites as possible…). The university life is, however, not only research, but also teaching. Thank you, Ulla and Bent, you have always been more than willing to discuss teaching, running and cameras. Melanie; thanks a lot for all the fun during teaching and travelling! May you always have a Gelato around! But what would teaching be without the students? I like to thank you all for your curiosity and enthusiasm, and especially Miria, Caspar, Filip, Linnea, Sandra and Sven for your superb tadpoleproject, and Carex-Frida, Polarulls-Sofia, Mattias, Niklas and Cissi & Truls for all the fun during the alpine botany course! My research might be about life, but life is not just about research, although there’s still quite a few researchers involved in it… Thank you EMG, for being the most social, professional and open minded department I’ve ever visited, and thanks Kicki, for all your efforts to keep the department like that and to continue to improve it, I am proud of being a part of this department. Life at the department would be so dull without the birdwatchers, the geeky discussions with Pelle, David and Mårten, the wild ideas you get when discussing with Åsa G and Göran (brilliant minds with the gift of thinking outside the box), the weird discussions with Carro and Viktoria, the interesting arguments with Magnus, Karin and Johan Olofsson, the weird stories of Carolyn and Melanie, and of course every odd subject that comes up when the German Maffia

56










(chocolate-Anja, Anja, Anna, Arne, lederhosen-Christopf, Esther and Kathrin) is around! What would life be without birds? Well, half of the department would spend more time in office during the spring… Thanks a lot Magnus, Lars, Peter, Åsa, Jögga, Ante, Jonas, Micke, Tomas, Darius, Katarina, Frank, Stefan, Viktor and Ronja for all company in the field, although you never have understand that you cannot give a poor woodpecker that is injured and missing a toe species status! Next year, team”Gäddfräsarna” will win the BirdRace! Emg Band I (Johanna, Henrietta, Ullis, Johan) and II (Fredrik, Anna, Magnus, Christopf, Henrik) have made me pick up the guitar (and occasionally other instruments) and having a lot of fun! Melanie, Karin, Lotta, Carolyn and Magnus have also shared my love of painting; we must have an exhibition soon! Ecology United might not be the best football team in Umeå, but we have most fun! Even if we have realised that our score is actually negatively affected by us turning up to the game. But next year, we’ll score! What would a department be without a Friday Pub? Thanks Pieter and Carolyn for being the best of pub masters, and offering the widest selection of ale and bear in town! Mia, Patrik and Ylva, thank you for being the best companions you can have in the land of wine and cheese! Thanks Åsa and David for all nice dinners! The players in UIK have given me many enjoyable hours watching football, and Umeå Jazzklubb has been an oasis for music. Annelie and Malin have made me realise that Skåne is one of the nicest parts of Sweden! Linda, Lina and Fredrik, thanks for always being there whenever I pop by Stockholm. I also have to thank my hometown Göteborg for its rain, mist and gales that always greets me, and all the nice people (”goa gubbar”) that always makes me return to the best coast! Edde ”Rob Winks” 


57







and Erik the Captain, you’ve always been there, and you always will! Friends you’ve had for over twenty years are friends that’ll be there for ever. When it’s time for football, Peter, Daniel and Martina are always willing to join! I never joined the ecologists at Zoologen in Göteborg, but it’s always nice to pop by (or pop by Bishop’s) and hear the recent news from Andreas, Rasmus and Bart (and your glögg was not 20 percent! No way!). Andreas, you must be the most enthusiastic person I know of, whether it is a new gore-tex jacket, research area, gps, whisky or book, you’ve found. Remember when both of us wanted to be Ebola-researchers… I also have to thank Scotland for your hospitality, your culture, whisky, haggis, mist, pipes, rain, highlands, salt-spray, tartans and accents! A part of my soul will always be with you. Anja och Åsa, ett jättetack för all kul vi har, även om ni kanske tycker att grodforskning är en lite skum sysselsättning! Mamma och Pappa, jag har alltid kännt ett jättestort stöd hemifrån för allt jag gjort! Tack vare er har jag fått mitt naturintresse, intresset för att läsa och skriva, och en väldig trygghet i allt jag gjort. Pappa, ditt lugn i alla situationer, ditt positiva tänkande och din otroliga nyfikenhet är så inspirerande! Mamma, med ditt logiska tänkande, din förmåga med pennan och din kunskap och argumentationsförmåga i alla diskussioner så är du en förebild i min forskningsroll! Mormor och Morfar, tack för att ni passat mig som liten, alltid stöttat mig, alla era berättelser och att ni visat och fått mig att älska Bohuslän, det vackraste landskapet som finns!

58