The incidental response to uniform natural selection

Abstract

When populations are exposed to novel conditions of growth, they often become adapted to a similar extent, and at the same time, evolve some degree of impairment in their original environment. They may also come to vary widely with respect to characters which are uncorrelated with fitness, as the result of chance genetic associations among the founders, when these are a small sample from a large and variable ancestral population. I report an experiment in which 240 replicate lines of the unicellular chlorophyte Chlamydomonas were derived from primarily photoautotrophic ancestors and cultured as heterotrophs in the dark. All adapted to the dark and were impaired in the light after several hundred generations of culture. They also displayed a wide range of colony morphologies that were uncorrelated with fitness. This incidental response to selection probably arose through random variation in the initial composition of the lines. The differences between closely related species or varieties may likewise arise, in similar circumstances, by sampling error rather than natural selection.

1. Introduction

A population that is exposed to a novel source of selection may respond in several ways. The most obvious is the direct response to the specific agent of selection, manifested by increased fitness, which constitutes evolutionary adaptation. Replicate populations are very generally convergent with respect to the direct response, always in the direction of change (positive) and often in magnitude. Any other change is usually treated as an indirect, or correlated, response that is related to the direct response in proportion to its genetic correlation with fitness [1,2]. Underlying the direct response is a change in the average states of those characters responsible for enhancing fitness under the new conditions of life, caused directly by natural selection by virtue of the mechanistic link between character state and fitness. The differences between related species may constitute divergent responses to somewhat different ways of life. The diagnostic characters that separate species may appear trivial, but might reflect precise adaptation to ways of life that are only slightly different. A special kind of correlated response will be the change of fitness in the original conditions of life, before exposure to the novel agent of selection. This is very generally negative, and may be called the antagonistic response.

All of these responses are connected with adaptation. However, there may also be changes in characters that have no connection with enhanced fitness. In a formal sense, this means that they have zero genetic correlation with fitness in the new conditions. In practice, we cannot usually estimate this correlation, but rather we mean that there is no evident link between such characters and those involved in the causal response, and that any such link seems unlikely on physiological or morphological or other a priori grounds. I shall call these changes the incidental response to selection. Because they do not arise through selection as a causal response, nor to the necessary indirect effect of selection as an antagonistic response, they must be attributable to variation arising through ancestry (and existing in some form before the change of conditions) or chance (and arising during the course of adaptation). Animal and plant breeders have long been familiar with the dynamics of subdivided populations [3], where differences between subpopulations or breeds may arise either through specific adaptation or through drift [4]. In domestic animals and crop plants, bottlenecks lead to inbreeding and thence to variation among replicate selection lines descending from the same ancestral population [5]. Quantitative morphological differences attributable to genetic drift arose in bottlenecked experimental populations of the butterfly Bicyclus [6], and similar observations have been made in other experimental systems [7]. Hence, the incidental response may lead to distinctive differences between breeds, varieties or species adapted to the same conditions of life that might be mistaken for adaptive divergence.

In this report, I describe the experimental evolution of heterotrophic growth in the dark of a normally photoautotrophic green alga. The direct, correlated and antagonistic responses were described in a previous report [8]. Here, I shall describe the incidental responses of hundreds of replicate populations exposed to a uniform agent of natural selection.

2. Material and methods

Experimental design and procedures are described in detail by Bell [8]. Briefly, three outcrossed laboratory populations of Chlamydomonas reinhardtii were cultured on minimal agar in the light for hundreds of sexual cycles and thousands of vegetative generations. Small samples of about 2000 cells from these base populations were used to found 2188 replicate selection lines that were subsequently propagated vegetatively in the dark on soft agarose medium supplemented with sodium acetate and transferred at intervals of a few weeks. More than 90 per cent became extinct within eight transfers. The remaining 240 lines were cultured for a further 50 cycles and evolved a high degree of adaptation to dark growth at a given acetate concentration (direct response) that was maintained over a range of acetate concentrations (correlated response), coupled with a general regress of fitness in the light (antagonistic response). The lines were then thin-spread on regular agar to examine the appearance of colonies in August 2010 and again in August 2011. The visual appearance of the lines is described in the electronic supplementary material, table S1.

The fitness of each line was estimated by competition with a tester as described by Bell [3]. A phenotypically distinct tester that formed yellow colonies in the dark was isolated from one of the selection lines. This tester was mixed with each dark line and the mixture cultivated for six cycles in the dark in medium supplemented with 1.2 gl⁻¹ sodium acetate. The frequency of the tester was estimated at the end of each cycle by plating on hard agar and observing colony morphology after growth in the dark. The selection coefficient was estimated as the rate of change of the logarithm of the ratio of tester to dark line cells over time. The tester was then competed in a similar way against each ancestor, to allow the selection coefficient of each line to be estimated relative to the ancestor. Selection coefficients are listed in relation to the colony morphology of each line in the electronic supplementary material, table S2. Data deposited in the Dryad repository: doi:10.5061/dryad.0tt5k.

3. Results

(a). Colony morphology

Thin spreads of the ancestral populations on hard agar displayed a range of colony phenotypes when cultured in the dark. Most colonies were small, pale and inviable, whereas a minority (less than 1%, 4% and 30% in the three ancestral populations) were viable with a broad range of shape and colour. In most cases, each experimental line displayed a single colony phenotype, although about 40/240 lines were dimorphic with two clearly distinct types present in the population (figure 1). Complete data for all lines are available as visual scores in two successive years (see the electronic supplementary material, table S1) and images of whole plates (see the electronic supplementary material, figure S1) and individual colonies (see the electronic supplementary material, figure S2). The greater part of variation in morphology was generated by three attributes of the growing colony.

• — Shape. In most strains, each colony expands isometrically to form a circular colony. Variant types develop a degree of concavity, resulting in irregularly shaped colonies.

• — Relief. The most usual form of the colony is a dome, resulting under transmitted light in a darker centre shading into a paler margin. This can be modified in either direction. A steeper, pillar-like colony is intensely green with only a narrow margin, if any. A flattened dome, on the other hand, gives a ‘fried-egg’ appearance with a broad margin, and at the extreme a ‘pancake’ colony consisting of a flat sheet of cells.

• — Margin. The standard circular dome colony has a smooth edge. Variants have a rough or ragged edge produced by the local proliferation of marginal cells.

Figure 1.

Colony morphology in the dark lines. The first three rows show variation among lines derived from the same base population with respect to margin, relief and shape. The fourth row shows further colony phenotypes from the other two ancestors.

(b). Colour

Most colonies are green in the dark. They vary in colour from pale watery green to intensely dark green, partly through morphology but also through age and size. The shade of green was not always a reliable and repeatable character in itself. Non-green colours were repeatable on respreading, however: pure white and pure yellow are clear and unambiguous phenotypes.

(c). Fitness in relation to morphology

There were no significant correlations between fitness and colony morphology or colour (table 1).

Table 1.

The fitness of strains in relation to colony morphology. Lines descending from each of the three base populations (Anc 1, Anc 2 and Anc 3) are analysed separately. The comparisons made were: shape: circular (0), not circular (1); margin: smooth (0), not smooth (1); relief: not flat (0), flat (1); white: not present (0), present (1); yellow: not present (0), present (1). The mean, standard deviation and sample size for the selection coefficient (per generation) are given for each category, with the estimate of t-value for each comparison. None of the values of t are significant testwise at p < 0.05.

		Anc 1			Anc 2			Anc 3
		mean	s.d.	n	mean	s.d.	n	mean	s.d.	n
shape	0	0.121	0.077	66	0.112	0.131	44	0.081	0.124	46
	1	0.125	0.115	26	0.116	0.101	24	0.133	0.101	34
	t	−0.19			−0.13			−1.99
margin	0	0.116	0.094	67	0.107	0.1236	41	0.112	0.119	37
	1	0.136	0.073	25	0.123	0.1179	27	0.096	0.116	43
	t	−0.93			−0.50			0.64
relief	0	0.116	0.090	75	0.113	0.125	59	0.093	0.125	57
	1	0.145	0.083	17	0.117	0.091	9	0.130	0.092	23
	t	−1.20			−0.09			−1.29
white	0	0.128	0.089	82	0.114	0.121	58	0.103	0.117	77
	1	0.072	0.078	10	0.110	0.123	10	0.125	0.128	3
	t	1.88			0.09			−0.32
yellow	0	0.126	0.087	72	0.117	0.116	40	0.105	0.118	71
	1	0.105	0.093	20	0.108	0.129	28	0.094	0.115	9
	t	0.97			0.33			0.25

4. Discussion

The incidental response to selection in this experiment was unexpectedly conspicuous and variable. Colony morphology is not equivalent to the morphology of multicellular organisms, because it does not reflect integrated complexity, but with this in mind, the differences among lines are as clear and conspicuous (so far as human judgement goes) as those among closely related species of animals and plants. The incidental response probably arose from sampling error (founder effects) causing genetic drift. More than 90 per cent of the dark selection lines became extinct soon after they were first inoculated, showing that most of the small samples from the base populations used to found the lines contained no viable cells. The thin spreads of the ancestral populations confirmed that viable cells were in a minority, sometimes a very small minority in the ancestral lines. Hence, most successful populations descended from only a few viable cells, or a single cell, bearing alleles responsible for dark growth in association with a random set of alleles at other variable loci. Independent replicate lines are therefore likely to differ from the outset with respect to a range of characters unconnected with the ability to grow in the dark. If there are several viable founders, the fixation of the lineage with highest fitness in the dark will also fix the states of these unconnected characters, because reproduction was mostly or entirely asexual after the isolation of the lines. Thus, the variability of the incidental response to selection observed in this experiment is readily explained by sampling error.

The origin of the dark lines as small independent samples from much larger source populations is akin to peripatric speciation on archipelagos [9], such as the classic case of Hawaiian Drosophila [10]. Malay & Paulay [11] have recently described peripatric speciation in the radiation of the hermit crab Calcinus in the Indo-west Pacific. Sister species of Calcinus are often distinguished by strikingly different colour patterns, and the authors remark that ‘coloration is so conspicuous and varied that it can be reasonably assumed to serve a purpose and thus be acted upon by natural selection’ [11, p. 651]. This may very well be the case, but the divergence of laboratory populations in this experiment shows that it is not necessarily the case. When large outbred populations diverge in allopatry, it is unlikely that sampling error will give rise to consistent differences between them, but small peripheral isolates may readily retain initial differences that have arisen by chance.

Closely related species with similar ways of life are often distinguished by some readily apparent difference in character state. The wood warblers (Parulidae) of eastern North America, for example, have distinctive streaks and patches of colour on the plumage of body and wing; ladybird beetles (Coccinellidae) differ in the ground colour and spotting pattern of the elytra; violets (Viola, Violaceae) vary in flower colour and the proportions of the leaves. Such examples appear in every handbook and identification key. The characters concerned may be either plain or subtle (to human senses) but in any case they are diagnostic and serve to distinguish reliably a particular species from its closest relatives. There are two strongly contrasted views of the biological significance of these distinguishing features. Cain [12] argued that even the smallest details of morphology can have a crucial effect on function, and gave, as an example, the diagnostic Y-shaped bar of chitin on the limb base of the centipede Polyxenus. This structure is inconsequential at first sight, but in fact enables the animal to run upside down on ceilings by providing a solid base for muscles that produce a large angle of swing and a powerful grip on the surface [13]. Gould & Lewontin [14], on the other hand, pointed out that some features, such as the colour and ornamentation of the shells of clams that live permanently buried in sediment, cannot influence the fitness of the individuals that express them, and must, therefore, be attributable to history or chance. There is no doubt that examples of both views can be found. What is at issue is whether or not the great majority of diagnostic differences between species are adaptive.

This experiment reported here shows how striking differences among related lineages can arise solely by chance, in circumstances where divergent selection can be ruled out. It does not help us to interpret any particular natural radiation, of course, in which natural selection may vary to any extent. Rather, it provides a caution that even strongly marked patterns may not be reliable guides to processes.