The Quarterly Review of Biology

Home , Judith Hooper, List of polymorphisms, Peppered moth

Volume 78, No. 4 December 2003 The Quarterly Review of Biology

THE RISE AND FALL OF THE CARBONARIA FORM OF THE PEPPERED MOTH

Laurence M. Cook The Manchester Museum, University of Manchester Manchester M13 9PL, United Kingdom e-mail: [email protected]

keywords peppered moth, Biston betularia, melanism, selection, predation, Kettlewell

abstract The evidence for change in frequency of the melanic carbonaria morph in the peppered moth Biston betularia (L.) (Lepidoptera: Geometridae) in England and Wales is reviewed. At mid-20th century a steep cline of melanic phenotype frequency running from the north of Wales to the southern coast of England separated a region of 5% or less to west from 90% or more to northeast. By the 1980s the plateau of 90% frequency had contracted to northern England. The frequency has since continued to drop so that the maximum is now less than 50% and in most places below 10%. There have been similar declines in Europe and North America. Evidence from surveys and from two-point records shows the change to require 5% to 20% selection against the melanic. The melanic is more disadvantageous in regions where its frequency was initially high than in regions where it was low. Experiments to investigate predation by birds show a net advantage to carbonaria morphs in regions where typical frequencies were low at the time of the experiment, and a disadvantage where typical frequencies were high. This would be expected if environment and frequency were associated, and selective predation played a part in generating the association. The cryptic advantage of carbonaria was large in areas of heavy pollution where typical frequencies were 20% or less. The moth usually has a low density but is relatively highly mobile. The ability of present information to explain the patterns has been tested in simulations. They indicate a system under strong selection that has always been in a dynamic state without equilibria.

ISTON BETULARIA (L.) has been called marks on a white background. Harris noted B the peppered moth at least since the that the larvae come in a variety of colors time of Moses Harris (1766), because typically from black to olive and that the moths are the wings and body are sprinkled with black “sometimes found sitting against the bark.”

399 400 THE QUARTERLY REVIEW OF BIOLOGY Volume 78

Since he mentions larval but not adult varia- urban environment. Pollutant gases declined tion it is unlikely that black forms were known too, but less dramatically and they were more to lepidopterists in his time. There were one widely dispersed. or two black individuals in collections by the The change in frequency of melanic forms beginning of the 19th century (Ford 1975; of moths in the 19th century was brought to Berry 1990), but they were not reported in the attention of the scientific community by the literature until the second half of the cen- William Bateson in relation to his dispute tury (Edelston 1864; Chappell 1886). The with the biometricians over the evolutionary darkest melanic was named doubledayaria by significance of distinct variants (see e.g., Prov- Millie`re (1872), later replaced in common ine 1971). Bateson (1900) published a “mem- usage by carbonaria ( Jordan 1869). The sev- orandum” from the Royal Society’s Evolution eral intermediate forms received the varietal Committee, asking entomologists for infor- name insularia (Thierry-Mieg 1886). The mation. The results were published (Barrett material for these descriptions came from 1901; Doncaster 1906) and discussed by Bate- northern England (see Cook and Muggleton son (1913). Melanism in the peppered moth 2003). was soon shown to be genetic and dominant Increase in melanics among moths fol- in effect (e.g., Bowater 1914). First sightings lowed the onset of industrialization, with its of carbonaria were from industrial Lancashire accompanying pollutants. London had suf- and Yorkshire (1848 to 1860), and progres- fered serious atmospheric pollution since the sively later from points north and south, Middle Ages, but developments in the indus- reaching London in 1897. There was much trial north of England were on a new scale. contemporary interest in the phenomenon, In the 17th century, spinning and weaving of and the pattern of spread suggests a single linen, cotton, and wool took place in hill val- source of origin (Steward 1977a). leys of Lancashire and Yorkshire, first as cot- A massive and widespread increase in the tage occupations and then in water-powered frequency of melanics had evidently taken mills. Cities such as Manchester and Leeds place, but the evidence was fragmentary and were manufacturing, finishing, and distribu- anecdotal. Haldane (1924, 1932) used it as an tion centers. In time the whole complex example of strong selection and examined became driven by coal power. The population mean selection per generation for periods of of Manchester was 24,000 in 1770. In 1800 it several generations (in Europe there is one was three times larger, and by 1850 over 10 generation per year). Work by H B D Kettle- times larger. Coal was used in the home as well showed that by the 1950s a large area of well as to drive machines, so smoke and sulfur England had carbonaria frequencies over and nitrogen gases became appallingly 90%. The frequency started to drop again in intense. Contemporary accounts record the the early 1970s. darkness of the skies and blackness of the It is intended here to present information buildings; attempts to control pollution were on this decline in melanic frequency and to largely ignored. Towns were not completely discuss the probable magnitude and type of devoid of plants and animals, however. Trans- selection involved. One aim is to draw port required horses; cows for milk were kept together and assess a range of data, some of in cities. Quantities of hay and grain were it from rather inaccessible sources. The story brought in, which would support populations was interesting at the start of the 20th century of small birds. Although dense, pollution was and is still so now. However, another reason highly localized and surrounding areas still for a review is the recent tendency in several retained their rural character. Smoke pollu- quarters to level charges of ignorance or tion continued in the growing conurbations deceit at researchers in the field. Part of the of the 20th century, and it was only in the attack comes from the antievolution lobby, 1960s, when coal as a fuel was replaced by oil e.g., Wells (2000), who writes of moves to “rid and electricity, that a significant drop textbooks of this fraud” (for a specific rebut- occurred. Smoke abatement legislation and tal of his claims see Rudge 2002). A book extensive rebuilding again transformed the review by Coyne (1998) gave much satisfac- December 2003CARBONARIA FORM OF THE PEPPERED MOTH 401 tion to creationists, and is frequently quoted 1a represents the data from 1952 to 1970. The on the web in their material on evolution and most notable feature is the cline running education (e.g., by J D Morris of the Institute from the north of Wales to the southeast of for Creation Research and by the Biblical England, which separates a region of low car- Creation Society; but see Rudge 2000). Other bonaria frequency west from the high fre- attacks are more mysterious. Although they quency areas of the east. Northern industrial are “certain that this phenomenon is a prod- England, including parts that are not urban- uct of selection,” Sargent, Millar, and Lam- ized, had a uniform frequency over 90%. This bert (1998) claim that almost everything continued southwards, dropping to 70% to about the melanism studies carried out so far 80% in the London area but remaining high is, possibly, wrong. In advance publicity for in rural eastern parts. The cline was steepest her book, Hooper (2002) contended that in the north, close to the Welsh/English bor- Kettlewell, “a deluded bumbling scientist,” der. A more detailed map given by Bishop et fraudulently arranged his evidence to satisfy al. (1978a) has the contours displaced some- an equally dissimulating E B Ford (but see what to the east because it is based on a more Coyne 2002; Grant 2002; Rudge 2003). No restricted set of values (the full data for this one determined to find fault will be swayed region are in Clarke and Sheppard 1963, by the presentation of evidence here, but 1966; Bishop 1972; Bailey et al. 1973; Whittle hopefully it will be useful in discussion of the et al. 1976; and the microfiche in Bishop et subject when a calmer atmosphere prevails. al. 1978a). During this period, melanic morph frequencies were positively associated Surveys in Britain with atmospheric sulfur dioxide concentration rather than with smoke (Kettlewell 1973; H B D Kettlewell coordinated three surveys Lees et al. 1973; Bishop et al. 1975). of morph frequencies over the country, from A survey carried out in the 1980s used 1952 to 1956, from 1957 to 1964, and from information provided by students of the 1965 to 1970 (Kettlewell 1958, 1965, 1973). Open University (Cook et al. 1986). Each had They are usually taken to represent the stable a standard collecting trap to sample from pattern at the middle of the century (e.g., in their home location; all accepted records pie diagrams by Kettlewell 1973; Lees 1981; being verified as the correct species and Majerus 1998). In Biston betularia there are morphs. Small individual samples were several alleles intermediate in expression and obtained (often fewer than 10 individuals), dominance between the carbonaria and typi- but they were widely dispersed. The results cal forms, which provide the insularia phe- (Figure 1b) show a decline by the mid-1980s. notypes. They are quite rare except in south- Frequencies of over 90% carbonaria were still ern Wales and adjacent areas of England encountered, but in a contracted area of where they sometimes exceeded the fre- northern industrial England. quency of carbonaria (e.g., Steward 1977a). The Rothamsted Insect Survey has moni- The main pattern exemplifying industrial tored peppered moth morphs since 1974, melanism is contributed by the carbonaria using standard light traps at over 350 sites allele, and insularia will not be considered (Woiwod and Harrington 1984). Again, indi- further here. Recent changes in insularia and vidual samples are often small. Data were the selection acting on it are discussed by grouped into 5-year periods in three regions: Cook and Grant (2000). (a) northern industrial England; (b) rural Figure 1 shows carbonaria frequencies in western England, Wales, and Scotland; and England and Wales at three successive times, (c) an environmentally intermediate region as contours derived from point data (Adams in southern England (Cook et al. 2002). In 1970). Although samples are unevenly distrib- the old industrial region, carbonaria declined uted and variable in size, this method pro- from 66% in the first 5-year period to 18% by vides a fair picture of regional pattern for the end, and in the intermediate region from comparison between dates. The literature 39% to 9%. In the rural region, carbonaria was cited contains original sample values. Figure less than 1% throughout. To provide an illus- 402 THE QUARTERLY REVIEW OF BIOLOGY Volume 78

Figure 1. Frequency of Carbonaria in England and Wales at at Three Successive Periods (a) Mid-20th century, derived from data of Kettlewell 1973, Lees and Creed 1975, Steward 1977a, Bishop et al. 1978a, and Lees 1981; (b) 1983–1984, from the Open University survey (Cook et al. 1986); (c) 1987–1999, from Rothamsted Insect Survey (Cook et al. 2002) and other recent records. The grid shows Ordnance Survey 100 km squares. tration for a later date, the 76 sites with data tain that the form seen was true carbonaria from 1987 to 1999 have been combined with rather than a very dark insularia, but either records from longitudinal studies to produce way the populations were changing. Sightings Figure 1c. The area of high frequency is were then made progressively to the north nowhere higher than about 50%. and east. There were melanics in Berlin, The most complete survey of change in Prague, and northern Bohemia in the ﬁrst melanic frequency in a single area (taken decade of the 20th century and extending from Caldy, on the Wirral peninsula between north to the Baltic coast and Copenhagen. Liverpool and northern Wales) was collected Frequencies rose rapidly in industrialized and by C A and F M M Clarke (Clarke et al. 1985; urban areas. In Berlin, melanics were “pres- Clarke et al. 1990; Clarke et al. 1994; Grant et ent” in 1903; in 1933 they were at 25%, about al. 1996) between 1959 and 2001. Other time 50% in 1939, and 85% by 1955 (Cleve 1970). series are given by Mani and Majerus (1993) Dresden had over 90% carbonaria (Schummer for Cambridge, the London area, and Staf- 1976). Nova´k and Spitzer (1986) showed car- fordshire; by Cook et al. (1999) for the Man- bonaria at 50% to 90% in Bohemia and Mora- chester area; and by West (1994, 2003) for a via from 1976 to 1981. The insularia forms location east of London in Kent. Grant et al. were up to 30%, while less urbanized Slovakia (1998) collected 18 sets of data, allowing mid- had lower carbonaria frequencies, sometimes century estimates to be compared with 1996. overtaken by those of insularia. Together these surveys show very large drops In the Netherlands from 1969 to 1973, car- in frequency in places where the frequency of bonaria was 70% to 73% in the urbanized carbonaria was initially high. Three examples south, dropping to 30% to 40% in the north are given in Figure 2. (Brakeﬁeld 1990). In Scandinavia, carbonaria has always been at a low frequency (Douwes Other Parts of the Range et al. 1973, 1976) and appears to have The earliest melanic record for continental increased in range northwards in Norway Europe is from Breda, the Netherlands, in (Andersen and Bengtson 1980) and east- 1867 (Heylaerts 1870). Ule (1925) recorded wards to Finland and Latvia (Mikkola 1984a), the melanics spreading from the southern even up to the 1980s, while insularia forms Netherlands into Germany by 1880 and to the became well established and reached fre- industrialized Ruhr by 1882 (illustrated by quencies over 50% in Danish Zealand and Rensch 1959; Cleve 1970). It is not always cer- southern Sweden. Frequencies subsequently December 2003CARBONARIA FORM OF THE PEPPERED MOTH 403

Figure 2. Decline in Carbonaria Frequency in the 20th Century Solid circles: Caldy, Wirral, northwestern England near the border with northern Wales (Clarke et al. 1985; Clarke et al. 1990; Clarke et al. 1994; Grant et al. 1996); squares: Manchester area, northwestern England (Bishop et al. 1978a; Cook et al. 1999); crosses: Kent, east of London (West 1994, 2003). Vertical lines are standard errors. declined in the Netherlands, as in Britain show that melanism has not evolved in Japan. (Brakefield and Liebert 2000). There, heavy industry tends to be in coastal These results provide a picture similar to regions where the moth is rare, and has not the British one. In the Ruhr there was a more blackened the countryside the way it did in than 20-fold increase in coal production Britain and Europe. between 1860 and 1960 (Cleve 1970), after Melanism developed in B. betularia cogna- which its use declined. An initial spread of taria in North America, where the black form carbonaria was followed by development of swettaria (Barnes and McDunnough) is indis- foci of high melanic frequency in intensively tinguishable from carbonaria. There are also industrialized regions. As in Britain, the dis- intermediate phenotypes similar to insularia. tribution of insularia maps less closely to Three morphs in the New World and Old industrialization than carbonaria and results World subspecies are illustrated by Grant, in high frequencies even in quite rural areas. Owen, and Clarke (1996), and the pattern of Melanics have decreased in frequency over change is reviewed by Grant and Wiseman the last three decades. We have little infor- (2002). The first observations of melanics mation on what happens further east or in were collated in an important study by Owen the southern parts of Europe. The moth (1961, 1962a, 1962b). The earliest is from occurs across the northern Palearctic from southeastern Pennsylvania (1906), then near Russia to Sakhalin as several subspecies, Pittsburgh (1910) and Detroit (1929)—all sometimes increasing to pest density in decid- centers of heavy industry. Within Pennsylva- uous forests in western Siberia and northern nia, outward movement occurred through Kazakhstan (Gninenko 2002). At the other the 1920s to the 1950s. Melanics spread extreme of Eurasia, Asami and Grant (1995) between Detroit and Pennsylvania, up the 404 THE QUARTERLY REVIEW OF BIOLOGY Volume 78 east coast to New England, and around Lake sqqq=−()()/ 1 − q2 − pq Michigan (Owen 1961, 1962a). Scudder 11  (1972) records them from British Columbia. Selection generates sigmoid curves of fre- Owen (1961) reports frequencies of 80% to quency in time. Haldane used the definite 97% melanic in Michigan in 1958 to 1960, 3% integral of the finite difference equation to to 11% in Massachusetts, and 36% in Ontario. estimate s over several generations. Another Moving south, 1% to 4% swetteria was method, easy with modern computers, is to recorded from wooded rural southwestern run the recurrence relation over the n gen- Virginia (West 1977). As in Europe, the erations and find a value of s, which provides exploitation of coal decreased, effective a good match to the final frequency. Both smoke control was introduced, and later methods are sensitive to accidental fluctua- observers noted a decline in melanic frequencies from the peaks of the 1960s. At a locality tion in frequency at high or low frequencies, in Pennsylvania, melanic frequency was 52% but give reasonable matches. (Equally, esti- from 1971 to 1978, 35% from 1981 to 1986, mates of selection over long periods from low and 18% in 1996 (Manley 1981, 1988; Grant frequencies are highly sensitive to initial fre- et al. 1998). Michigan frequencies dropped quency. An increase to 98% melanics in 47 from over 80% to less than 20% at about the years, sometimes quoted for Manchester, same time as the similar drop in carbonaria at requires an advantage of 50% if gene fre- 6מ Caldy in northwestern England (Grant et al. quency at the start was 10 but 25% if it was 2מ 1996). Michigan and Pennsylvania mirror 10 .) Standard errors may be calculated by each other in dropping from 90% in 1959 to the integration method, which takes account less than 10% in 2001, while in Virginia of sample size. melanics have almost disappeared (Grant Kettlewell’s data span 20 years, and should and Wiseman 2002). first be examined for consistency. There were 26 sites sampled both from 1952 to 1956 and Selection Estimated from Morph at a later date (usually 1957 to 1964; see Ket- Frequency Change tlewell 1973, Appendix C). Only two changes The strength of selection may be estimated between these dates are significant (in one from frequency changes. To provide a start- carbonaria increased, in the other it declined). ing point it will be assumed that selection The overall change in carbonaria frequency is ,(1.07 ס affects melanic homozygotes and heterozy- not significant (normal deviate, d gotes in the same way. Wing color is con- and is not associated with initial typical fre- ,0.98 ס t ,0.197מס trolled by alleles Wc for carbonaria and Wt for quency (Spearman’s r typical. For convenience we may call these c n.s.). There is, however, significant hetero- df, P 25 ,69.7 ס and t, and assign them frequencies p and q. geneity between sites (v2 The simple pattern of selection may then be Ͻ0.001). Lees and Creed (1975) pointed out represented by: that there were increases in melanic frequency in southern and western parts of visual phenotype carbonaria typical England, often associated with relatively high frequencies of insularia. The variations are genotype cc ct tt frequency p 2 2pq q 2 quite small, and it seems reasonable to do as s l others have done and treat this whole data setמs lמfitness l as the mid-century baseline from which frequencies later deviated. The typical gene frequency one generation Cook et al. (1986) compared the contour later is: surfaces generated for Kettlewell’s data with =− −−2  the Open University data to provide a coeffi- qq1 ()111 sps/ () q cient s of selection against carbonaria averag-

The new melanic gene frequency is 1–q1, and ing 12% per generation. This method is subj- 2 the morph frequency 1–q1 . Rearrangement ect to bias introduced by contouring and provides: possible aberrant samples that distort the sur- December 2003CARBONARIA FORM OF THE PEPPERED MOTH 405 face. Nevertheless, the result is consistent TABLE 1 with the contraction in the extent of the high Disadvantage to carbonaria when data from the frequency area, and suggests the magnitude beginning of the decline are compared with 1990s of the selection involved. The Rothamsted Initial data include 31 sites in which carbonaria fre- Location Typical Frequency S SE quency could be compared over an average Chesterfield 0.01 0.195 0.046 elapsed time of 12 generations. The decline Bradford 0.04 0.118 0.017 in frequency from the early period (average Manchester 0.02 0.208 0.025 38.3%) to the late (average 15.7%) provides Liverpool 0.03 0.199 0.045 an estimate of selection against carbonaria at Chester 0.07 0.187 0.020 9.5%. Caldy 0.09 0.152 0.007 Figure 2 shows the complete series for Bishops Stortford 0.11 0.135 0.015 Caldy, Wirral, compared with the Manchester Grimsby 0.11 0.109 0.029 area and with Kent. All three drop after 1970 Northwood 0.12 0.114 – Birmingham 0.13 0.139 0.020 to converge by the end of the period, and Bradwell 0.14 0.137 0.010 there is a suggestion that the drop is steeper Cambridge 0.18 0.105 – when initial frequency was higher. Grant et al. Hawarden 0.18 0.114 0.013 (1998) give figures for places throughout Kent 0.22 0.130 0.015 England where estimates of melanic fre- Fritton 0.23 0.116 0.018 quency from 1952 to 1956 can be compared Rhosemor 0.29 0.148 0.023 with those for similar localities in 1996. Using Cromer 0.35 0.071 0.023 these and some additional data from within Mostyn 0.39 0.060 0.016 the same period, 26 values of s can be calcu- Lixwm 0.50 0.081 0.013 lated and related to the initial typical fre- Loggerheads 0.57 0.085 0.013 Oxford 0.66 0.087 0.019 quency (Table 1). There is a significant East Sussex 0.83 0.105 0.023 decline in s as initial typical frequency Isle of Man 0.87 0.028 0.023 ס מס increases (Spearman’s r 0.860, t 8.60, West Sussex 0.88 0.028 0.031 P Ͻ0.001). When the initial typical frequency Marlborough 0.91 0.018 0.025 is at less than 20%, the mean disadvantage of Carrog 0.94 0.058 0.051 carbonaria is 16%. It is 11% between 20% and Sites are broadly distributed over England and Wales. Data 80% typical, and 5% above that (a slightly dif- from Clarke et al. 1985; Mani and Majerus 1993; West 1994; ferent set is illustrated in Cook 2000). Grant et al. 1998; Pratt 1999. The long series of data collected by the Clarkes (Figure 2) comes from a location at in 3מthe edge of the patch of high carbonaria fre- tration dropped from about 300 lgm in 1975 and then 3מquency in Liverpool and industrial Lanca- 1960 to about 50 lgm shire, but close to a region of much lower fre- remained fairly constant. If selection is linearly quency. It therefore records the fall as the correlated with concentration a better fit is cline moved progressively northeast. From achieved. 1959 to 2001, the mean selection against The existence of this complete sequence melanics was 14.4%. However, the resulting invites us to consider what other influences fitted curve drops too steeply at the outset may have acted. For example, selection could and only tends to coincide with the observa- be to some extent frequency dependent tions after about 1986 (cf. Grant et al. 1996). because predators tend to overlook forms The pattern suggests low selection until 1977 that are numerically rare. If so, the response stronger selection from 1977 to would be modified at both extremes to bring ,(0.09 ס s) -and lower selection subse- the curve closer to the data points. In addi ,(0.28 ס s) 1986 -It is most probable that tion, we do not know for certain that the fit .(0.18 ס quently (s selection did vary, and the pattern can be ness of the two melanic genotypes is identical; related closely to change in atmospheric sulfur a large set of breeding data examined by dioxide concentration near the site (Clarke et Creed, Lees, and Bulmer (1980) suggested al. 1985; Mani 1990). Sulfur dioxide concen- that melanic homozygotes have higher pread- 406 THE QUARTERLY REVIEW OF BIOLOGY Volume 78 ult viability than heterozygotes. With that affected by experience; moths may not be assumption a significantly better fit to the full attacked when they are rare. sequence can be obtained. Migration could The next sequence of observations was also be involved. There is abundant evidence made in a polluted area of woodland with that moths can fly from central Liverpool to high carbonaria frequency near Birmingham. the collecting site, as they can from northern Observations of attacks by birds on released Wales. If the net influx was from Liverpool, moths were made, and the list of bird species where initial frequency was higher and seen taking moths was enlarged to nine. Ket- decline occurred later, then the drop in fre- tlewell noted that the background was vari- quency would have been delayed. If incomers able so that morph visibility would be affected were mostly from the west, in Wales, the fit by choice of settling site. Releases were made would be worsened. at different times of day to see how predation varied. Moths of other resident species were Selection from Predation also released so that peppered moths did not Experiments dominate the fauna available to birds. Ket- tlewell satisfied himself that the traps did not Kettlewell established that industrial areas catch the morphs differentially, that the had smoke-blackened trees and buildings, morphs had the same average life spans, and and that large amounts of particulate smoke that they dispersed to the same extents. Dif- could be collected from air extractions or sur- ferential return would therefore indicate faces. Plant growth is reduced in these con- some other mechanism at work, of which the ditions, and lichens are almost absent from most obvious is differential predation. Three tree and stone surfaces (Kettlewell 1973). In trials there permitted estimates of relative sur- fact, lichen cover was only rich and abundant vival to be made. Melanics were recaptured in extreme western parts of England and with greater frequency than typicals, as Wales, as a result of both low atmospheric pol- expected if the result arose from difference lution and high average humidity. Given that in crypsis, providing carbonaria with a substan- to the human observer carbonaria was the tial advantage. most cryptic form in polluted areas while the Kettlewell then tested the situation in an typical morph was the most cryptic in western uncontaminated woodland where there was unpolluted ones, it was natural to try to dis- low atmospheric pollution and rich lichen cover whether predators experienced them cover, and melanics were not present. Typi- in the same way and exerted selection that cals were extremely difficult to detect in these would change gene frequency. This was not conditions, carbonaria was conspicuous. necessarily the case. Little evidence existed to Again, he noted qualifications about the suggest that birds were important predators nature of the habitat, resting positions, and on Lepidoptera, and birds may have suffi- bird behavior. In this case, the release and ciently good vision to see both types under recapture experiment indicated that carbon- either condition. aria was strongly disadvantageous. Since Kettlewell (1955a, 1956, 1973) carried out melanics were at an advantage where melan- three sets of experiments to investigate pre- ics were common and at a disadvantage where dation. In his publications the enthusiasm of the population was typical, Kettlewell con- a field naturalist sifting through the succes- cluded that visual selection strongly influ- sive steps of the argument is very obvious. enced local frequency. The idea of selective First, moths were released in an aviary with predation as the chief determinant was estab- great tits, Parus major. When they had taken lished. These experiments represent a thor- up resting positions, Kettlewell scored them ough attempt to assess the likelihood of bird for degree of conspicuousness to himself. predation and its selective power, with more After a period of familiarization the birds details and qualifications than are usually sup- took the insects, the majority removed being plied. the ones he had scored as conspicuous. He Since then a number of similar experi- noted that response by the predators will be ments have been carried out. Because of the December 2003CARBONARIA FORM OF THE PEPPERED MOTH 407 but there is (0.014 ס difficulties of arranging a trial with living ani- not differ from 1 (v2 ,df 35 ,61.85 ס mals, moths have usually been raised, killed, heterogeneity between sets (v2 frozen in lifelike resting attitudes, and P Ͻ0.005). attached to the tree surfaces. There are dif- The data in Figure 3 suggest that carbonaria ferences in method among trials and a variety fitness is high when typical frequency in the of types of estimation of fitness have been area is low, and vice versa. In view of the vari- used, some discussed by Bishop et al. (1978b). ability of estimates it is worth testing this sta- When comparing results the most context- tistically. Linear regressions are significantly free approach is to find the relation of fre- negative whether weighted or unweighted by quency of morphs presented to the frequency variance. We have no expectation of linearity, remaining after predation. If starting num- however, and the local frequencies are them- bers of carbonaria and typical are c1 and t1, and selves estimates. In addition the frequency of numbers remaining are c2 and t2, then the insularia, which is not represented here, is higher at intermediate typical frequencies ס fitness of carbonaria may be measured as w c2t1/t2c1. The variance of log(w) may be cal- than at extremes. A parameter-free method culated (Manly 1985). The normal deviate of testing the trend, which minimizes these (measuring deviation from proportional sur- problems, is to compare the results for the vival) is log(w)/SE and its square is a v2 with third of the trials from sites with the lowest one degree of freedom. Figure 3 shows results local morph frequency with the third from taken from the literature for 36 trials from sites with the highest. In the 12 fitnesses at ס v2) 2.95 ס various parts of England and Wales. Estimates lowest typical frequencies, d have been plotted on estimates of the typical 8.72, P Ͻ0.01). At the other extreme with the frequency at the location when the experi- 12 fitnesses at the highest typical frequencies, ס v2) -2.74 ס ment was carried out. Most of these figures the overall normal deviate is d were illustrated by Cook (2000), where fitness 7.50, P Ͻ0.01). The summed v2 is 16.23 with This .(0.0003 ס of typical rather than carbonaria was pre- two degrees of freedom (P sented. figure measures the significance of the evi- Several investigators have examined the dence that carbonaria is, on average, at an effect on the estimate of differences in resting advantage where there is low typical fre- site position. Clarke and Sheppard (1966) quency but at a disadvantage where there is classified parts of the tree surface as pale or high typical frequency. It is not dependent on dark, and tested selection with the four pos- the accuracy of estimation of local frequency sible combinations of pale and dark moths on or assumptions about functional relation. Fig- pale and dark surfaces. Except where both ure 3 suggests that carbonaria only gained a morphs were on pale backgrounds, typicals substantial advantage in areas where typical were at a disadvantage at the site used, where frequencies were 20% or less. How great the typical frequency was low. Lees and Creed selection will be in practice depends on nat- (1975) compared results when tree surfaces were dry and pale or wet and dark. The car- ural backgrounds, resting positions, and pre- bonaria form was at a disadvantage on the dation rates. paler dry surface, and at an advantage in dark wet conditions. Howlett and Majerus (1987) Criticisms of Predation Experiments put moths in the positions they would expect Although they show clear selective differ- to find them on trees and in the exposed posi- ences the experiments have been subject to tions usually used by other investigators. They several types of criticism (see Grant 1999 for found no difference in relative removal. review). It has been objected that many of the The standard errors for individual tests are experiments involved dead frozen insects. understandably large, but taken overall the Since the moths remain immobile during the results show a correspondence between local day, when selective predation is likely to take frequency and estimated fitness, and indicate place, live and dead moths do not differ in that selective pressure can be great. Taking appearance. The use of dead moths allows the data sets as a whole, the mean fitness does the investigator to control density and site on 408 THE QUARTERLY REVIEW OF BIOLOGY Volume 78

Figure 3. Fitness of Carbonaria from Predation Experiments Relation of fitness w of melanic to frequency of typical at site at the time of the experiment. Small adjustments have been made to frequencies to separate points. Vertical lines are standard errors. Open squares: Kettlewell (1955a, 1956); closed squares: Clarke and Sheppard (1966); open circles: Howlett and Majerus (1987); crosses: wet and dry from Lees and Creed (1975). Other values from Bishop (1972), Lees and Creed (1975), Whittle et al. (1976), Steward (1977b), Bishop et al. (1978b), and Murray et al. (1980). the trees, and potential differences of physi- exposed trunks. Among 203 individuals seen ology or dispersal behavior in living insects near traps, 23.6% were on trunks. Neverthe- are avoided. less, given the opportunity the moths often A related criticism is that the moths have settle on branches high in tree tops or under been placed in the “wrong” positions, namely branches at the point where they join the on tree bark surfaces. Peppered moths cer- trunk (Mikkola 1979, 1984b; Howlett and tainly rest on tree trunks; this has been Majerus 1987; Liebert and Brakefield 1987). reported since the earliest observers and they Studies of settling enlarge our understanding can frequently be seen there on a morning of the behavior of the species. They may well following the use of a light trap. The best data modify our view of how selection operates, for are those of Howlett and Majerus (1987; example in identifying particular sets of likely Majerus 1998). Of 47 observations of settled bird predators or pointing to associations moths away from traps, they found 12.8% on with particular lichens or surfaces. However, December 2003CARBONARIA FORM OF THE PEPPERED MOTH 409 it is quite untrue to say that they cast doubt if insects were sufficiently cryptic. The tests on experiments already carried out. At the show that selection can occur, and is different time of the tests there was a gross difference in different places. between the appearance of trees in polluted and unpolluted locations. In industrial Numbers and Movement regions there were virtually no epiphytes on In order to understand selection in prac- trees or other surfaces, which were usually tice it is necessary to get a picture of popu- heavily coated with smoke. In rural areas trees lation structure. To do so we often have to were paler, more polychrome, and sometimes rely on indirect information. It is difficult well covered in epiphytes. In both types of even to define the characteristic habitat. The habitat there was a strong correlation specific name suggests an association with between trunks and branches. In urban areas birch trees (Betula spp.). In fact, the species trees were relatively scarce, walls common, appears to be polyphagous and to feed on a and there were abundant light sources to dis- wide variety of trees and lower-growing plants orient moths (see Frank 1988). The distinc- (Chappell 1886; Seppa¨nen 1954; Chalmers- tion between “natural” and “unnatural” set- Hunt 1976). Early larvae produce silk threads tling positions is inappropriate. which assist in wind dispersal, so that individ- Kettlewell (1955b) placed moths in con- uals from a single brood may feed on differ- tainers with black and white striped surfaces ent plant species and a single individual may to see whether the different morphs chose eat more than one species during its devel- similar or contrasting backgrounds. He found opment. that they settled preferentially, typical on Two mark, release, and recapture experi- white and carbonaria on black, the surfaces they most resembled to the human eye. Some ments have been carried out to estimate later investigators obtained similar results population parameters. The first of these while others did not. Howlett and Majerus (Bishop 1972) involved a central release (1987) point out that the human eye is not point and more or less equal trapping prob- the best instrument for judgments of crypsis. ability up to a distance of about 5 km over a 2 Because of the structure of their wings, typi- total area of 78.5 km . Bishop released 1433 cals may actually be better protected on dark moths over a 15-day period. Seventeen per- surfaces. Several tests show such a preference, cent of 173 recaptures had flown 3 km or which is partially genetically determined and more. Previous observations had shown that may vary between individuals (Grant and the moths can fly 2 km per night and that Howlett 1988). Furthermore, the spectral considerably longer flights may occur (Whit- sensitivity of birds differs from our own and tle et al. 1976). Daily estimates of male popu- relative crypsis of different morphs on partic- lation were between 250 and 500, about half ular types of lichens may not be as we see it the average number of moths. Survival rate of (Majerus et al. 2000). Birds can see well into the males was estimated as between 0.3 and the ultraviolet range, although there are dif- 0.4 per day, some of the loss being due to emi- ferences between species (Cuthill et al. 2000). gration. Bearing in mind the length of the Background choice and relative visibility season and the daily survival rate, there were need further investigation. The present posi- about 400 individuals per km2 per season, and tion is reviewed by Majerus (1998), who 3 to 6 males per km2 per day. The second makes the interesting suggestion that features population estimate (Bishop et al. 1978b) was of resting site preference may explain appar- roughly three times that of the first. Moths ent anomalies where frequencies of melanics were caught up to 2 km from release. At both are higher or lower than we would expect at sites most recaptures were caught within one first (human) sight. Once again, however, day of release but the fraction at large for information on behavior and visual range more than one day was higher in the more and acuity does not negate results which have distant traps, indicating progressive dispersal. already been obtained. Selective predation, In Britain the moth is relatively rare com- and indeed predation itself, would not occur pared with many other species. Woiwod and 410 THE QUARTERLY REVIEW OF BIOLOGY Volume 78

Taylor (1984) studied catch size of moths over to be density dependent. If many females are a 5-year period in traps distributed through- present the males fly little before encounter- out the country. The peppered moth ranked ing one; when they are rare, mating journeys 216th in abundance out of 263. Cook and are longer. Females do not fly if they mate Graham (1996) examined numbers in over shortly after emergence, but will do so if no 300 samples made over 24 years in a single male arrives. Moths were abundant at the wooded valley in southern England. More Clarkes’ site; over 35 years they caught than 92,000 individuals comprised 305 taxa, roughly 5 to 10 moths per night. Of these, less the two most abundant being 23% and 15% than 0.5% were females. The Open University of the total. The peppered moth was 18th at students trapped in average sites. The mean 0.9%. Peppered moth abundance correlated catch was 0.3 per night but there were 3.8% with total sample size, which varied by an females among checked specimens. The stu- order of magnitude over the period. dents also classified their site subjectively as Other more circumstantial data have a rural or urban. Of 502 sites considered to be bearing on movement. Bishop (1972) esti- rural, 9.2% had 6 or more captures. The 482 mated selection along the cline in frequency urban sites had a significantly lower 2.7% with of carbonaria from Liverpool into northern 6 or more captures. Cook et al. (2001) Wales. Given the differences in frequency, divided the country into industrial, rural, and moths would have to move on average 2.4 km intermediate regions. The means per site per per generation to set up a cline of the season were 2.7 for the industrial, 3.2 for the observed width (May et al. 1975). Since aver- intermediate, and 5.3 for the rural region. age life expectancy is little more than a day, Taken together these estimates show that that indicates average daily movement of in Britain the peppered moth is present in a some 2 km. In the late 1960s and early 1970s low density network, sometimes with only a the Cheshire plain and southern Lancashire few individuals per km2 at any time, while they had a uniformly high frequency of carbonaria can fly 2 km or more per night. It is probable (Bishop et al. 1978a). Morph frequencies that densities are greater in rural and wooded were invariant for some years just before the areas than urban ones, and mobility increases decline began. Selection was probably similar with decrease of density. At these densities over the whole region, and because an earlier predators are unlikely to become familiarized advantage was being translated into a disad- with the moths as a food source. Individual vantage at the time, may have been more or demes of small size will tend to show high less neutral. Under these circumstances the inbreeding and drift effects, resulting in dif- relation of movement to variability may be ferentiation. High migration has a homoge- examined using Wright’s island model. There nizing effect, however, and in a network over- were 56 sampling sites with an average pair- all effective population size can become wise distance of 15.4 km and nearest neigh- larger than actual numbers (Cherry 2003; bor distance of 3.2 km. The mean carbonaria Wang and Whitlock 2003). Selection over a allele frequency (weighted by square root of wide geographical range can be effectively sample size) was 0.824, with a variance of determinate despite locally variable condi- 0.011. The number of migrating individuals, tions. Nm, may be estimated as [q(1–q)–v]/(4v), where q is the allele frequency and v the var- Discussion iance. The model assumes that any individual The results presented bring together infor- could reach any site; if they only reach nearest mation on melanic morph frequency over neighbors, Nm must be doubled (Crow 1986). England and Wales from the middle to the The calculated Nm is 3.1, suggesting that 3 end of the 20th century. At the start of this moths per generation move to another loca- period, more than half of the country had tion if the whole area operates as a single unit melanic frequencies over 80%, and there was (average distance 15 km) or correspondingly a steep cline between the melanic region and more for nearest-neighbor movement. the low frequencies of the west and southwest. Directed movement by the moths is likely The southern end of the cline began to decay. December 2003CARBONARIA FORM OF THE PEPPERED MOTH 411

At first frequencies to the northeast remained sition of the bird fauna, enormously increase almost at their original level while the the concentration of sulfur and nitrogen pol- melanic patch contracted. Later the high pla- lutants, modify local climate, and sharpen the teau dropped to less than half its original fre- contrast between cities and rural environ- quency and in many places lower. If the trend ments. With the disappearance of coal as a continues carbonaria will become very rare fuel, and reduction and change in the com- throughout the country in the next two position and dispersal of other atmospheric decades. pollutants, conditions have converged on, but The estimates of selection, made by differ- are in numerous respects different from, ent methods and with different degrees of those of preindustrial times. security, provide a consistent picture. During The changes in morph frequency are the period of decline the disadvantage of car- undoubtedly a response to these environmen- bonaria was between 5% and 20%. This is simi- tal changes. Kettlewell’s demonstration of lar to the advantage the form possessed dur- selective predation focused attention on that ing its increase, as indicated by early records agency, but the question of the relative impor- (Kettlewell 1973: Table 9.1). There is evi- tance of visual and nonvisual selection is as dence that selection against carbonaria has old as the study itself (e.g., Tutt 1891). Dis- been about twice as severe where it was ini- crepancies between ecology and expected fre- tially high than where it was low. This may be quency if visual selection was all-important because environmental amelioration in more were noted at an early date (e.g., Leigh 1911) polluted regions has a more profound effect and continued to be discussed (Kettlewell than in less polluted regions. 1973). There are differences in response Estimates of selection from predation stud- between B. betularia and other species with ies are larger than those obtained from melanics (Bishop et al. 1978a; Cook et al. morph frequency changes. This is to be 2001), which point to the operation of other expected. The tests are carried out at higher factors in addition to crypsis. densities and in more exposed locations than The idea of heterozygote advantage was normally present. The effectiveness of pre- widespread in the mid-20th century. It was dation as a selective agent depends on both inferred for carbonaria from patterns of the discriminatory power of the predators morph frequency (Haldane 1956; Kettlewell and predation rate. We do not know enough 1958; Clarke and Sheppard 1963, 1966). about the ecology of predation to relate the Unfortunately, direct investigation of physi- two types of measurement in detail. Selective ology and metabolism of the morphs has not predation can undoubtedly be substantial, been attempted or is fragmentary, and infer- and those with field experience are per- ences from frequency patterns are weakened suaded that it is an important, perhaps the because the patterns are not stable. Thus, in most important, factor in determining the case of heterozygote advantage, Haldane changes in morph frequency (e.g., Majerus suggested that the failure of carbonaria to 1998; Grant 1999). The predation experi- reach 100% in Manchester may indicate a dis- ments show carbonaria to be at an advantage advantage in homozygotes, but his argument in regions with typical frequency less than does not take account of the high levels of 20% and disadvantageous where typical fre- mobility. Clarke and Sheppard pointed out quency is high. The changes over the last that where frequencies were known for three quarter century require carbonaria to be dis- dates during the increase, the fit to the points advantageous throughout the range. The was improved by assuming heterozygote experiments come from before the recent advantage. Again, migration was assumed to downturn in frequency, however, and relate be insignificant and selection unchanging to different environmental conditions. over the period. The effect of the industrial revolution on What more can be deduced from the data cities was to darken the trees and cover the about the factors influencing morph fre- environment with smoke, remove epiphytes quency? Any estimation or curve fitting and general vegetation, change the compo- involves an underlying model. Examination 412 THE QUARTERLY REVIEW OF BIOLOGY Volume 78 of the Caldy data shows that when part of the selection of the order given by the predation information is taken at a time we have little experiments, varying regionally and chang- except parsimony to guide us in choosing ing with time during the increase; (b) non- between explanations. The next logical step visual selection of about 20% favoring melan- is to weigh all available information together, ics of both genotypes; (c) migration levels of in order to determine what hypotheses about 2.5 km per day; and (d) weak frequency emerge and what must be excluded. This was dependent selection protecting morphs the intention of the very thorough modeling when very rare (Mani 1990). exercise carried out by G S Mani. As a first If selection is both visual and nonvisual at step the pattern from the 19th century the rates suggested then selection coefficients through 1950 to 1970 was examined (Cook derived from change in time must be com- and Mani 1980). Selection was assumed to be posites of the two forces. Observed s values of dominant in effect, as it would be if wholly 5% to 10% translate into visual selection coef- due to bird predation. It was also assumed to ficients of 20% to 25%. Visual selection be different in different regions of the coun- against melanics would have to have been try, the values being based on the predation very large in preindustrial Britain to keep experiments. Migration was introduced and melanics at mutation frequencies, but given the resulting simulation compared with avail- the input data, the conclusion that there is able observations to test the requirement for strong nonvisual selection is unavoidable. heterozygote advantage. It was found that a There are, however, two ways in which the satisfactory fit could be obtained for the tran- data may mislead. One possibility is that all sect from northern Wales to northwestern selection is visual but that over most of the England and for southern Wales to London. range carbonaria is less conspicuous than the The model failed to represent the pattern predation experiments suggest. Selective from central Wales through central England. regions may be mosaics of subregions having It would have produced a better fit in this large variation in selective value (Bishop and region if the melanics had a nonvisual advan- Cook 1975; Mani 1990). Settling behavior on tage which counteracted the visual selection trees or other substrates may give the moths against them in the more rural parts. The fit relative selective values of which the preda- would also be improved if frequency depen- tion experiments are unrepresentative (Lees dent selection protected the morphs when and Creed 1975; Majerus 1998 and references they were very rare. There was, however, no therein). reason to include heterozygote advantage. Another possibility is that we still underes- Mani (1980) then examined the cline from timate the reactivity of the system to changing Liverpool to northern Wales, first discussed conditions. When Mani examined the north- by Bishop (1972). In 1966 and 1967, Bishop ern Wales cline he found the simulations to estimated relative survival at points along the show a retreat toward Liverpool unless an cline. Using these figures and a range of likely advantage to carbonaria was included. The migration rates, a simulation was tested experiments that produced the selection esti- against the 15 frequency values that defined mates were carried out in 1966 and 1967. the cline and came mostly from 1963 to 1968. Many of the frequency estimates being mod- Again, a fit could not be obtained unless eled came from a few years earlier. Given what there were nonvisual selection that favored we now know, a retreat toward Liverpool is carbonaria. The best estimate suggested a non- what we should expect. The 1963 frequency visual fitness value of 1.2, applying equally to of carbonaria at one site used in the simulation heterozygous and homozygous melanics. was 49.0% (Clarke and Sheppard 1966), but Having obtained this value for nonvisual by 1984 it had become 12.9% (Clarke et al. selection, Mani (1982) showed that an accept- 1985). The same changes occurred across able fit could be obtained throughout England. Many of the records being modeled England and Wales, including previously came from the 1950s. The plateau of high fre- unpublished data (Mani and Majerus 1993). quency started to contract shortly after, and The parameters required were: (a) visual if a nonvisual advantage to melanics was December 2003CARBONARIA FORM OF THE PEPPERED MOTH 413 excluded the simulation tended toward what Kettlewell surveyed frequency and mea- was coming, rather than matching the pub- sured predation, but he also examined the lished records. ideas developed by Ford. He bred moths to investigate phenotypic expression and sur- Conclusion veyed other species for geographic, relict, or The peppered moth has had a varied role ancient melanism. He wished to understand as a model in evolutionary studies. William more fully why some species adapted so rap- Bateson (1913) chose it while contesting the idly to industrialization, why melanics tended proposition that evolution proceeds by infin- to be dominant and why there were many itesimal steps under the direction of natural cases of melanic polymorphism. Kettlewell selection. He argued that variation is usually spent much time on these issues, but the tolerated (that is, neutral). Internal forces explanatory power of selective predation and due to genetic constitution ensure that spe- the correlation between melanic frequency cies remain constantly different from each and background came to dominate peoples’ view of the subject. As industrial melanism other, while from time to time polymor- achieved its textbook status it came to exem- phisms of large effect occur. These may con- plify the power of selective predation, medi- tribute to the step from one species to ated by changes in crypsis brought about by another, a supposition that fitted well at the atmospheric pollution. time with accumulating evidence of discrete Although neat, the textbook description Mendelian inheritance. was simplistic. A reaction developed, both E B Ford (1937) devoted much of his because of the general prevalence of more career to natural balanced polymorphisms. neutralist explanations for genetic variability These often have a complex basis, being and because simple gene frequency change determined by groups of linked genes segre- tells us little about species formation or the gating together to form a switch mechanism. higher reaches of evolution. The decline in Why should that be so? A possibility, drawing carbonaria frequency since Kettlewell’s time on R A Fisher, is that there is dominance mod- has amply confirmed the strength of selec- ification. Genes are pleiotropic, with multiple tion, and later results reinforce his predation effects on phenotypic expression, some studies. It now seems, however, that carbonaria advantageous and others deleterious. Selec- is unusual among melanic alleles affected by tion tends to make the favorable effects of the industrialization. In Britain, at least, it seems gene dominant and the unfavorable ones to have originated from a single center and recessive. Given sufficient time the heterozy- in postindustrial conditions will probably be gote will exhibit the advantageous traits while lost. In contrast, industrial melanics of many homozygotes have a mixture of both. Evolu- other moth species occur as polymorphisms tion can therefore trap mutant alleles in pop- in rural regions and became more common ulations through selection acting on genes as conditions required. This appears to be that affect phenotypic expression. Chromo- true of 19% of 262 British species with somes eventually build up sets of linked loci melanic variants in Kettlewell’s (1973) with complementary effects. Much popula- Appendix B. It may apply to the paler insu- tion genetics since Ford’s time has addressed laria alleles at the carbonaria locus. We know different aspects of this scenario, whether little more now than in Bateson’s day about expressed in his terms or not. Most modern why related species in similar conditions population geneticists would argue that near respond in different ways. The peppered neutral genetic variability is much more prob- moth may yet have a further role. able, and coadaptation less prevalent, than he acknowledgments supposed (e.g., Crow 1986). Nevertheless, the I am grateful to Bruce Grant for helpful comment, ecological genetic theory, as set out by Ford, discussion and suggestions. This paper is dedicated to had a significant effect on the way work on the memory of Jim Bishop, Cyril Clarke, Bernard Ket- the peppered moth was conducted and tlewell, and Philip Sheppard, who contributed so received in the period that followed. much to the study of the peppered moth. 414 THE QUARTERLY REVIEW OF BIOLOGY Volume 78

REFERENCES

Adams R P. 1970. Contour mapping and differential Chalmers-Hunt J M. 1976. Lepidoptera of Kent. Ento- systematics of geographical variation. Systematic mologist’s Record 88 Supplement:156–157. Zoology 19:385–390. Chappell J. 1886. Varieties of Amphidasys betularia. Andersen T, Bengtson S-A. 1980. Melanism in the pep- Entomologist 19:253–254. pered moth Biston betularia (L.) in western Norway Cherry J L. 2003. Selection in a subdivided population (Lepidoptera: Geometridae). Entomologica Scandi- with dominance or local frequency dependence. navica 11:245–248. Genetics 163:1511–1518. Asami T, Grant B. 1995. Melanism has not evolved in Clarke C A, Clarke F M M, Dawkins H C. 1990. Biston Japanese Biston betularia (Geometridae). Journal of betularia (the peppered moth) in West Kirby, Wir- the Lepidopterists’ Society 49:88–91. ral, 1959–1989: updating the decline in f. carbon- Bailey W J, Clarke C A, Sheppard P M, Whittle D J. aria. Biological Journal of the Linnean Society 39:323– 1973. The status of the industrial melanics of Biston 326. betularia (L.) in the Isle of Man. Entomologist Clarke C A, Grant B, Clarke F M M, Asami T. 1994. A 106:210–214. long term assessment of Biston betularia (L.) in one Barrett C G. 1901. Lepidoptera of the British Isles. Volume UK locality (Caldy Common near West Kirby, Wir- VII. London: Reeve. ral), 1959–1993, and glimpses elsewhere. Linnean Bateson W. 1900. Collective enquiry as to progressive 10(2):18–26 melanism in moths—memorandum from the Evo- Clarke C A, Mani G S, Wynne G. 1985. Evolution in lution Committee of the Royal Society. Entomolo- reverse: clean air and the peppered moth. Biologi- gist’s Record 12:140. cal Journal of the Linnean Society 26:189–199. Bateson W. 1913. Problems of Genetics. New Haven (CT): Clarke C A, Sheppard P M. 1963. Frequencies of the Yale University Press. melanic forms of the moth Biston betularia (L.) on Berry R J. 1990. Industrial melanism and peppered Deeside and in adjacent areas. Nature 198:1279– moths (Biston betularia (L.)). Biological Journal of the 1282. Linnean Society 39:301–322. Clarke C A, Sheppard P M. 1966. A local survey of the Bishop J A. 1972. An experimental study of the cline distribution of the industrial melanic forms in the of industrial melanism in Biston betularia (L.) (Lep- moth Biston betularia and estimates of the selective idoptera) between urban Liverpool and rural values of these in an industrial environment. Pro- north Wales. Journal of Animal Ecology 41:209–243. ceedings of the Royal Society of London B 165:424–439. Bishop J A, Cook L M. 1975. Moths, melanism and Cleve K. 1970. Die Erforschung der Ursachen fu¨r das clean air. Scientific American 232(1):90–99. Auftreten melanistischer Schmetterlingformen im Bishop J A, Cook L M, Muggleton J. 1978a. The Laufe der letzten hundert Jahre. Zeitschrift fu¨r Ange- response of two species of moths to industrializa- wandte Entomologie 65:371–378. tion in northwest England. I. Polymorphisms for Cook L M. 2000. Changing views on melanic moths. melanism. Philosophical Transactions of the Royal Soci- Biological Journal of the Linnean Society 69:431–441. ety of London B 281:491–515. Cook L M, Dennis R L H, Mani G S. 1999. Melanic Bishop J A, Cook L M, Muggleton J. 1978b. The morph frequency in the peppered moth in the response of two species of moths to industrializa- Manchester area. Proceedings of the Royal Society of tion in northwest England. II. Relative fitness of London B 266:293–297. morphs and population size. Philosophical Transac- Cook L M, Graham C S. 1996. Evenness and species tions of the Royal Society of London B 281:517–542. number in some moth populations. Biological Jour- Bishop J A, Cook L M, Muggleton J, Seaward M R D. nal of the Linnean Society 58:75–84. 1975. Moths, lichens and air pollution along a tran- Cook L M, Grant B S. 2000. Frequency of insularia sect from Manchester to north Wales. Journal of during the decline in melanics in the peppered Applied Ecology 12:83–98. moth Biston betularia in Britain. Heredity 85:580– Bowater W. 1914. Heredity of melanism in Lepidop- 585. tera. Journal of Genetics 3:299–315. Cook L M, Mani G S. 1980. A migration-selection Brakefield P M. 1990. A decline of melanism in the model for the morph frequency variation in the peppered moth Biston betularia in the Netherlands. peppered moth over England and Wales. Biological Biological Journal of the Linnean Society 39:327–334. Journal of the Linnean Society 13:179–198. Brakefield P M, Liebert T G. 2000. Evolutionary Cook L M, Mani G S, Varley M E. 1986. Postindustrial dynamics of declining melanism in the peppered melanism in the peppered moth. Science 231:611– moth in the Netherlands. Proceedings of the Royal 613. Society of London B 267:1953–1957. Cook L M, Muggleton J. 2003. The peppered moth, December 2003CARBONARIA FORM OF THE PEPPERED MOTH 415

Biston betularia (Linnaeus, 1758) (Lepidoptera: the peppered moth (Biston betularia Linn.): indi- Geometridae), a matter of names. Entomologists’ vidual differences. Biological Journal of the Linnean Gazette. Society 33:217–232. Cook L M, Riley A M, Woiwod I P. 2002. Melanic fre- Grant B S, Owen D F, Clarke C A. 1996. Parallel rise quencies in three species of moths in post indus- and fall of melanic peppered moths in America trial Britain. Biological Journal of the Linnean Society and Britain. Journal of Heredity 87:351–357. 75:475–482. Grant B S, Wiseman L L. 2002. Recent history of mel- Coyne J A. 1998. Not black and white. Review of Mel- anism in American peppered moths. Journal of anism: Evolution in Action, by M E N Majerus. Nature Heredity 93:86–90. 396:35–36. Haldane J B S. 1924. A mathematical theory of natural Coyne J A. 2002. Evolution under pressure: a look at and artificial selection. Transactions of the Cambridge the controversy about industrial melanism in the Philosophical Society 23:19–41. peppered moth. Review of Of Moths and Men: Haldane J B S. 1932. The Causes of Evolution. London: Intrigue, Tragedy and the Peppered Moth, by J Hooper. Longmans. Nature 418:19–20. Haldane J B S. 1956. The theory of selection for mel- Creed E R, Lees D R, Bulmer M G. 1980. Pre-adult anism in Lepidoptera. Proceedings of the Royal Society viability difference on melanic Biston betularia (L.) of London B 145:303–306. (Lepidoptera). Biological Journal of the Linnean Soci- Harris M. 1766. The Aurelian or Natural History of ety 13:251–262. English Insects; Namely, Moths and Butterflies. [1986 Crow J F. 1986. Basic Concepts in Population, Quantitative reprint.] Twickenham (UK): Newnes. and Evolutionary Genetics. New York: Freeman. Heylaerts F J M. 1870. Les macrole´pidopte`res des envi- Cuthill I C, Partridge J C, Bennett A T D, Church S C, rons de Breda. Tijdschrift voor Entomologie 13:142– Hart N S, Hunt S. 2000. Ultraviolet vision in birds. 157. Advances in the Study of Behaviour 29:159–214. Hooper J. 2002. Of Moths and Men: Intrigue, Tragedy and Doncaster L. 1906. Collective enquiry as to progressive the Peppered Moth. London: Fourth Estate. melanism in Lepidoptera. Entomologist’s Record Howlett R J, Majerus M E N. 1987. The understanding 18:165–168. of industrial melanism in the peppered moth (Bis- Douwes P, Mikkola K, Petersen B, Vestergren A. 1976. ton betularia) (Lepidoptera: Geometridae). Biologi- Melanism in Biston betularius from north-west cal Journal of the Linnean Society 30:31–44. Europe (Lepidoptera: Geometridae). Entomologica Jordan R C R. 1869. Note on the black variety of Amphi- Scandinavica 7:261–266. dasis betularia. Entomologist’s Monthly Magazine 6:40. Douwes P, Petersen B, Vestergren A. 1973. Melanism Kettlewell H B D. 1955a. Selection experiments on hos Biston betularius L. i Danmark och so¨dra Sverige industrial melanism in the Lepidoptera. Heredity (Lepidoptera). Entomologen 2:15–20. 9:323–342. Edelston R S. 1864. Amphydasis betularia. Entomologist Kettlewell H B D. 1955b. Recognition of appropriate 2:150. backgrounds by the pale and black phases of the Ford E B. 1937. Problems of heredity in the Lepidop- Lepidoptera. Nature 175:943–944. tera. Biological Reviews 12:461–503. Kettlewell H B D. 1956. Further selection experiments Ford E B. 1975. Ecological Genetics. Fourth Edition. Lon- on industrial melanism in the Lepidoptera. Hered- don: Chapman & Hall. ity 10:287–301. Frank K D. 1988. Impact of outdoor lighting on moths: Kettlewell H B D. 1958. A survey of the frequencies of an assessment. Journal of the Lepidopterists’ Society Biston betularia L. (Lep.) and its melanic forms in 42:63–93. Britain. Heredity 12:51–72. Gninenko Yu I. 2002. Specific features of Biston betu- Kettlewell H B D. 1965. A 12-year survey of the fre- laria (Lepidoptera, Geometridae) biology in east- quencies of Biston betularia L. and its melanic forms ern part of its area. Zoologicheskii Zhurnal 81:375– in Great Britain. Entomologist’s Record 77:195–218. 378. Kettlewell B. 1973. The Evolution of Melanism. Oxford: Grant B S. 1999. Fine tuning the peppered moth par- Clarendon. adigm. Evolution 53:980–984. Lees D R. 1981. Industrial melanism: genetic adapta- Grant B S. 2002. Sour grapes of wrath. Science 297:940– tion of animals to air pollution. Pages 129–176 in 941. Genetic Consequences of Man Made Change, edited by Grant B S, Cook A D, Clarke C A, Owen D F. 1998. J A Bishop and L M Cook. London: Academic Geographic and temporal variation in the inci- Press. dence of melanism in peppered moth populations Lees D R, Creed E R. 1975. Industrial melanism in in America and Britain. Journal of Heredity 89:465– Biston betularia: the role of selective predation. Jour- 471. nal of Animal Ecology 44:67–83. Grant B, Howlett R J. 1988. Background selection by Lees D R, Creed E R, Duckett J G. 1973. Atmospheric 416 THE QUARTERLY REVIEW OF BIOLOGY Volume 78

pollution and industrial melanism. Heredity 30: and predation by birds in the moths Biston betularia 227–232. and Phigalia pilosaria. Proceedings of the Royal Society Leigh H S. 1911. A biological enquiry into the nature of London B 210:277–283. of melanism in Amphidasys betularia Linn. Entomol- Nova´k I, Spitzer K. 1986. Industrial melanism in Biston ogist 44:162–165. betularia (Lepidoptera, Geometridae) in Czecho- Liebert T G, Brakefield P M. 1987. Behavioural studies slovakia. Acta Entomologica Bohemoslovaca 83:185– on the peppered moth Biston betularia and a dis- 191. cussion of the role of pollution and lichens in Owen D F. 1961. Industrial melanism in North Amer- industrial melanism. Biological Journal of the Linnean ican moths. American Naturalist 95:227–233. Society 31:129–150. Owen D F. 1962a. The evolution of melanism in six Majerus M E N. 1998. Melanism: Evolution in Action. species of North American geometrid moths. Oxford: Oxford University Press. Annals of the Entomological Society of America 55:695– Majerus M E N, Brunton C F A, Stalker J. 2000. A bird’s 703. eye view of the peppered moth. Journal of Evolu- Owen D F. 1962b. Parallel evolution of European and tionary Biology 13:155–159. North American populations of a geometrid moth. Mani G S. 1980. A theoretical study of morph ratio Nature 195:830. clines with special reference to melanism in moths. Pratt C R. 1999. A Revised History of the Butterflies and Proceedings of the Royal Society of London B 210:299– Moths of Sussex. Brighton (United Kingdom): 316. Booth Museum of Natural History. Mani G S. 1982. A theoretical analysis of the morph Provine W B. 1971. The Origins of Theoretical Population frequency variation in the peppered moth over Genetics. Chicago (IL): University of Chicago Press. England and Wales. Biological Journal of the Linnean Rensch B. 1959. Evolution Above the Species Level. Lon- Society 17:259–267. don: Methuen. Mani G S. 1990. Theoretical models of melanism in Rudge D W. 2000. Does being wrong make Kettlewell Biston betularia—a review. Biological Journal of the wrong for science teaching? Journal of Biological Linnean Society 39:355–371. Education 35:5–11. Mani G S, Majerus M E N. 1993. Peppered moth revis- Rudge D W. 2002. Cryptic designs on the peppered ited: analysis of recent decreases in melanic fre- moth. International Journal of Tropical Biology and quency and predictions for the future. Biological Conservation (Revista de Biologiá Tropical) 50:1–7. Journal of the Linnean Society 48:157–165. Rudge D W. 2003. Judith Hooper’s Of Moths and Men: Manley T R. 1981. Frequencies of the melanic morph Intrigue, Tragedy and the Peppered Moth. Journal of the of Biston cognataria (Geometridae) in a low pollu- History of Biology 36:207–209. tion area of Pennsylvania from 1971 to 1978. Jour- Sargent T D, Millar C D, Lambert D M. 1998. The nal of the Lepidopterists’ Society 35:257–265. ‘classical’ explanation of industrial melanism: assessing the evidence. Evolutionary Biology 30:299– Manley T R. 1988. Temporal trends in frequencies of 322. melanic morphs in cryptic moths of rural Pennsyl- Schummer R. 1976. Zum Melanismusproblem bei Bis- vania. Journal of the Lepidopterists’ Society 42:213–217. ton betularia und B. strataria im Gebiet der DDR. Manly B F J. 1985. The Statistics of Natural Selection. Lon- Deutche Entomologische Zeitschrift 23:281–294. don: Chapman & Hall. Scudder G C E. 1972. Industrial melanism: a possibility May R M, Endler J A, McMurtrie R E. 1975. Gene fre- in British Columbia. Journal of the Entomological Soci- quency clines in the presence of selection opposed ety of British Columbia 69:46–48. by gene flow. American Naturalist 109:659–676. Seppa¨nen E J. 1954. Suomen suurperhostoukkien ravin- Mikkola K. 1979. Resting site selection by Oligia and tokasvit (Die Futterpflanzen der Grossschmetterlingsrau- Biston moths (Lepidoptera: Noctuidae and Geo- pen Finnlands). Helsinki (Finland): So¨derstro¨m. metridae). Acta Entomologia Fennici 45:81–87. Steward R C. 1977a. Industrial and non-industrial mel- Mikkola K. 1984a. Dominance relations among the anism in the peppered moth, Biston betularia (L.). melanic forms of Biston betularius and Odontoptera Ecological Entomology 2:231–243. bidentata (Lepidoptera, Geometridae). Heredity Steward R C. 1977b. Melanism and selective predation 52:9–16. in three species of moths. Journal of Animal Ecology Mikkola K. 1984b. On the selective forces acting in the 46:483–496. industrial melanism of Biston and Oligia moths Thierry-Mieg P. 1886. Aberrations nouvelles de Le´pi- (Lepidoptera: Geometridae and Noctuidae). Bio- dopte`res europeéns. Le Naturaliste 3:236–237. logical Journal of the Linnean Society 21:409–421. Tutt J W. 1891. Melanism and melanochroism in Brit- Millie`re P. 1872. Iconographie et de´scription de che- ish Lepidoptera. Entomologist’s Record 1:5–7. nilles et le´pidopte`res ine´dits. Annales de la Socie´te´ Ule C H. 1925. U¨ ber Auftreten und Verbreitung von Linneénne de Lyon 19:1–90. Amphidasis betularia L. f. carbonaria Jord. auf dem Murray D D, Bishop J A, Macnair M R. 1980. Melanism Kontinent. Entomologische Zeitschrift 18:257–263. December 2003CARBONARIA FORM OF THE PEPPERED MOTH 417

Wang J, Whitlock M C. 2003. Estimation of effective 1976. Further studies on the industrial melanic population size and migration rates from genetic moth Biston betularia (L.) in the north west of the samples over space and time. Genetics 163:429–446. British Isles. Procedings of the Royal Society of London Wells J. 2000. Icons of Evolution: Science or Myth? Wash- B 194:467–480. ington (DC): Regnery. Woiwod I P, Harrington R. 1984. Flying in the face of West B K. 1994. The continued decline in melanism change: the Rothamsted Insect Survey. Pages 321– in Biston betularia L. (Lep.: Geometridae) in NW 342 in Long-Term Experiments in Agricultural and Eco- Kent. Entomologist’s Record 106:229–232. logical Sciences, edited by R A Leigh and A E John- West B K. 2003. Biston betularia L. (Lep. Geometridae): ston. Wallingford (UK): CAB International. continued decline in industrial melanism in north- Woiwod I P, Taylor L R. 1984. Synoptic monitoring for west Kent. Entomologist’s Record 115:13–16. migrant insects pests in Great Britain and Western West D A. 1977. Melanism in Biston (Lepidoptera: Geo- Europe. V. Analytical tables for the spatial and tem- metridae) in the rural central Appalachians. Hered- poral population parameters of aphids and moths. ity 39:75–81. Rothamsted Experimental Station, Report for 1983 Part Whittle P D J, Clarke C A, Sheppard P M, Bishop J A. 2:261–293.