Nachr. entomol. Ver. Apollo, N.F. 19 (1): 69-88 (1998) 69 Similarities between British and north-west European Aricia "subspecies"1 (Lepidoptera: Lycaenidae) Bill Smyllie Dr Bill Smyllie, 164 Dobcroft Road, Sheffield Sil 9LH, U.K. Abstract: The following main Aricia “subspecies” are referred to in this pa­ per: Aricia artaxerxes allous (Geyer, 1836), Aricia artaxerxes artaxerxes (Fabri- cius, 1793), Aricia agestis ([Denis & Schiffermüller], 1775), and Aricia cra- mera (Eschscholtz, 1821). They are referred to as “subspecies” because all are capable of cross-breeding with one another. These butterflies have bet­ ween 0 and 6 submarginal lunules — orange spots — on each upper wing. One puzzling aspect is that it is possible to find individuals which show the com­ plete range of variation within one colony. In this paper data on upper fore- wing lunulation are presented, together with information on several other as­ pects: mendelism, white scales in upper forewing discal spots, phased emer­ gence, photoperiod and climate. The main emphasis is on upper forewing lu­ nulation and statistical checks are applied, but it is shown that points from all the above aspects reinforce one another and indicate interpenetration of “subspecies” with a broad hybrid zone between Latitude 54° and 59° N. Any “subspecies” in this zone are identified in the text, and colonies at Durham northern England and Skäne southern Sweden are shown to be hybrids. Ear­ lier similarities noted between British and Scandinavian colonies are extend­ ed southward to northern and central Germany. Subsequent discussion co­ vers interpenetration, variation of colonies in the hybrid zone, and the rela­ tionship of Aricia “subspecies” to other butterflies, including the possibility that present forms have spread from a common ancestor. Ähnlichkeiten zwischen britischen und anderen nordwesteuropäischen Ar/c/a-,/Unterarten" (Lepidoptera: Lycaenidae) Zusammenfassung: In dieser Arbeit werden die folgenden wichtigsten „Un­ terarten“ von Aricia aus Nordwesteuropa behandelt: Aricia artaxerxes allous (Geyer, 1836), Aricia artaxerxes artaxerxes (Fabricius, 1793), Aricia agestis ([Denis & Schiffermüller], 1775) und Aricia cramera (Eschscholtz, 1821). Sie werden hier als „Unterarten“ behandelt, weil sie sich offenbar unterein­ ander fertil paaren können. Diese Schmetterlinge haben je zwischen 0 und 6 submarginale Halbmonde (orangefarbene Flecken) auf der Flügeloberseite. Auffällig ist, daß man jeweils innerhalb einer Teilpopulation Individuen fin- The term “subspecies” is used here not to denote a systematic unity or taxonomic rank (e.g., as de­ fined by the International Code of Zoological Nomenclature), but to describe a taxon which, in my opi­ nion, is not a full species (because the different populations are interfertile), without deciding on their status here. 70 den kann, die die komplette Variationsbreite zeigen. In der vorliegenden Ar­ beit werden die Daten der Halbmondflecke auf der Vorderflügeloberseite präsentiert, danaben Informationen über folgende weitere Aspekte gegeben: Mendelianischer Erbgang, weiße Schuppen in den Zellschlußflecken der Vor­ derflügekoberseite, schubweiser Falterschlupf, Photoperiodizität und Klima. Das Hauptgewicht liegt auf den orangen Flecken der Vorderflügeloberseiten; dazu werden statistische Verfahren angewendet. Insgesamt zeigen alle ange­ führten Punkte Hinweise auf gegenseitige Vermischung der „Unterarten“ mit offenbar einer breiten Hybridzone zwischen 54° und 59° nördlicher Länge in NW-Europa. Alle „Unterarten“ innerhalb dieser Zone werden identifiziert; die Kolonien von Durham in Nordengland und Skäne in Südschweden sind Hybriden. Früher durchgeführte Vergleiche zwischen britischen und skandi­ navischen Populationen werden ausgedehnt nach Nord- und ins südwestli­ che Deutschland. In der weiteren Diskussion werden die Durchmischungszo­ nen, die Variabilität der verschiedenen Teilpopulationen in der Hybridzone und die Vergleichsbeziehungen der Arida-„Unterarten“ mit anderen Tagfal­ tern sowie ihre mögliche Ableitung von einem gemeinsamen Vorfahren be­ handelt. Introduction The generally accepted position of the taxa of Aricia in Britain today is from south to north (Heath et al. 1984): • bivoltine agestis from the south coast up to Derbyshire; • univoltine colonies designated Aricia artaxerxes salmacis (Stephens, 1831) (Derbyshire and further north in England); • univoltine A. a. artaxerxes in Scotland. A query to the Biological Records Centre re the status of the univoltine Peak district colonies in Derbyshire revealed (Heath, pers. comm. 1971) that the distribution map for Aricia was due to Jarvis. A. agestis colonies (Jarvis 1969) were described as uniformly well-lunulated compared with A. a. salmacis colonies at Durham where the Emulation was very variable. Checks on Peak district specimens via museum collections and photographs led to the conclusion that the Peak district race was well Emulated. If this was so, where did lunulation become less? Eventually over 3000 specimens were examined, mainly from 23 different museums with some field records and photos. Upper forewing lunulation was quantified in order to provide a positive comparison. Specimens were examined without magnification and lunules were counted, traces or more counting as one lunule. They were then classified into Cr (crassilunulata) or Pv (parvilunulata forms). These terms were used (Jarvis 1969) to describe well and poorly Emulated 71 butterflies respectively. Males were less well lunulated than females, so Cr males were set at 5 or 6, females at 6 lunules. It follows that Pv males have 0-4, Pv females 0-5 lunules. Numbers for any one colony/area were built up randomly mainly covering England and Scotland with some from the continent. The data were manipulated mathematically to provide a 50/50 combination of males and females which produced a CC:CP ratio — combined Cr : combined Pv. From a statistical examination a minimum ra­ tio of 5.0 was required for A. agestis and this was easily met by all bivoltine and three univoltine colonies, one of which was the Peak district (Smyllie 1992 a). Further north the CC:CP ratio first dipped quite distinctly below 5.0 at two colonies, one at Pickering in north Yorkshire. It was the male lu- nulation which had dropped — the female had not changed. Table 1 gives selected British and continental examples: it also includes subsequent data from Scandinavia and Germany. The CC:CP ratio has the great advantage of describing the degree of lunula- tion of any colony by one simple figure which can then be compared with any other. The lower the ratio, the poorer the lunulation. Table 1: Lunulation comparison between selected areas. Abbreviations: (J male; $ female; Cr crassilunulata, Pv parvilunulata; T total; CC combined crassilunulata, CP combined parvilunu- lata; S south; SW south west; N north; C central. Area â Cr c?Pv 6 T $ Cr $Pv <j>T CC CP CC:CP 1 Sandhammaren to 10/8 8 38 46 17 11 28 21.9 34.1 0.64 2 Skäne ex 10 10 20 12 3 15 19.5 10.5 1.86 Sandhammaren 3 S England 232 49 281 184 11 195 345.0 45.0 7.67 4 Peak district 71 16 87 42 0 42 76.3 7.7 9.91 5 Pickering 18 28 44 14 0 14 19.7 8.3 2.37 6 Durham coast 10 75 85 39 23 62 46.3 77.7 0.60 7 Inland Durham 16 13 29 28 11 39 36.8 21.2 1.74 8 Solway SW Scotland 7 28 35 8 7 15 10.3 19.7 0.52 9 Perthshire C Scotland 2 115 117 13 42 55 13.9 96.1 0.14 10 Rügen N Germany 9 16 25 20 3 23 28.3 17.7 1.60 11 Mainzer Sand 10 2 12 6 0 6 11.0 1.0 11.00 12 Continental A. agestis 39 11 50 52 1 53 88.1 11.9 7.41 13 Continental A. a. 1 80 81 1 40 41 1.5 80.5 0.02 allons 72 Scandinavian data A substantial paper (Hoegh-Guldberg 1966) includes data on lunulation from several colonies in Scandinavia. Some could be converted to give CC:CP ratios in Table 1, but most required more manipulation to give a comparison for both males and females which had 4-6 upper forewing lunules. This was expressed as a percentage of the total (Smyllie 1995), and Table 2 again compares selected colonies. There are slight modifica­ tions from that originally published plus the German data. The map in Fig. 1 indicates localities either by name or number in this Table. Latitudes are included so that sites in different countries can be compar­ ed easily. Also the number of specimens from which the male and female percentages have been obtained are shown. Higher figures will be more stable. Both male and female percentages are given to the nearest whole number; the combined % column gives the average, i.e. a 50/50 contribu­ tion of S<S and ÇÇ. The right-hand classification column shows one of 3 categories. It is easiest to start with agestis, and here all male figures are Fig. 1 : Map of the location of sites in Table 2 by name or number. 73 96 °o or more. These are significantly higher than for any other colonies and females in agestis colonies are all 100 % giving a combined figure of 98 °o minimum. Intermediates follow at 30 to 98 %. Finally allous colonies are less than 30 % The demarcation line between intermediates and allous is arbitrary and identifies a point below which lunulation is considered poor. Note that no British colonies are in 3. The table again shows a similarity between Rügen and Pickering. Male upper forewing lunulation Males are less well lunulated than females, and this is advantageous when comparing lunulation characteristics of different colonies. Table 3, pre­ viously unpublished, gives the complete range of male lunulation for selected colonies and is used as the basis for column charts Figs. 2 and 3. Fig. 2 also is used to provide statistical information (see next section). Fig. 2: Percentage of 5 & 6 lunules from Table 3 with 5 % significance limits. 74 Table 2: 4-6 upper forewing lunules in Britain, Scandinavia and Germany. Additional abbre­ viations: NC north central, DEN Denmark, SWE Sweden, NOR Norway, ENG England, EUR continental Europe, SCO Scotland; Lat. northern Latitude in 0 N; L lunules; Comb. % combin­ ed percentage (as in a 50/50 contribution of <S<S and $$); CL classification into three categor­ ies: 1 = agestis (98 % or more), 2 = intermediates (30-98 %), 3 = allous (less than 30 %). Locality Lat. °N <S 4-6 L/T % $ 4-6 L/T % Comb. % CL 1 Sandhammaren to 10.8 55.2 30/61 48 49/54 91 69.5 2 2 Sandhammaren 2nJ brood 55.2 18/20 90 12/13 92 91.0 2 3 Skäne less Sandhammaren 55.3 44/53 83 26/26 100 91.5 2 4 DEN agestis 56.0 93/122 76 40/40 100 88.0 2 5 SNOR Jomfruland 59.0 6/42 14 26/29 90 52.0 2 6 SWE Gotland 57.5 7/24 29 14/14 100 64.5 2 7 SWE Öland 56.5 9/24 37 6/9 67 52.0 2 8 SWE Angermanland 63.5 0/8 0 5/9 56 28.0 3 9 NDEN Hirsthals 57.5 5/241 2 25/96 26 14.0 3 10 NNOR Lyngenfjord 69.5 0/9 0 1/5 20 10.0 3 11 EUR allous - 7/65 11 14/35 40 25.5 3 12 SENG agestis 51.0 214/233 96 137/137 100 98.0 1 13 CENG Peak district 53.0 85/87 98 42/42 100 99.0 1 14 CGER Mainzer Sand 50.0 12/12 100 6/6 100 100.0 1 15 EUR agestis - 48/50 96 53/53 100 98.0 1 16 NGER Rügen 54.2 19/25 76 23/23 100 88.0 2 17 NCENG Pickering 54.2 36/44 82 14/14 100 91.0 2 18 NENG Durham coast 54.8 25/92 27 30/39 77 52.0 2 19 NENG Inland Durham 54.6 21/29 72 29/29 100 86.0 2 20 SSCO Solway district 54.9 15/35 43 13/15 87 65.0 2 21 CSCO Perthshire 57.0 16/117 14 34/55 62 38.0 2 22 NCSCO Invernesshire 57.5 11/39 28 14/19 74 51.0 ? Points to be noted from the right-hand column in Table 3 or from Fig. 2 are the high and consistent figures for 5 and 6 lunules for agestis colonies nos. 1-4, the low consistent figures for the northern artaxerxes colonies nos. 12- 14, and the considerable variation in the remainder, nos. 5-11. The 0 lunules column shows the same general effect but in reverse — see Fig. 3. 75 Table 3: Male upper forewing lunulation for selected colonies. Abbreviations see above dtf lunules [n] lunules % total 0 2 3 4 5 6 0 2-4 5-6 1 SEXG counties 177 - 1 9 22 67 78 0.0 18.1 81.9 1 CEXG Peak District 138 1 - 2 21 64 50 0.7 16.7 82.6 3 CGER Mainzer Sand 12 - - - 2 7 3 0.0 16.7 83.3 4 EUR agestis 50 - - 2 9 18 21 0.0 22.0 78.0 5 XCEXG Pickering 44 - 1 7 18 8 10 0.0 59.1 40.9 6 XGER Rügen 25 1 1 4 10 8 1 4.0 60.0 36.0 - XEXG Durham coast 172 56 26 40 25 12 3 32.6 58.7 8.7 8 XEXG Durham inland 29 3 1 4 5 10 6 10.3 34.5 55.2 9 SSWE Sandhammaren 62 9 37 16 14.5 59.7 25.8 10 SSWE Skäne ex Sandh’n 20 1 9 10 5.0 45.0 50.0 11 SWSCO Solway district 35 7 2 11 8 4 3 20.0 60.0 20.0 12 CSCO Perthshire 117 72 8 21 14 2 0 61.5 36.8 1.7 13 XCSCO Invemesshire 39 20 4 4 9 1 1 51.3 43.6 5.1 14 EUR allous 65 37 9 12 6 1 0 56.9 41.6 1.5 . ■ . . - :: . • 1---- 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Fig. 3: Percentage of 0 lunules from Table 3. 76 Statistical checks In Britain the greatest number of colonies and resulting specimens is in the south and reduces steadily further north. Museum collections contain more examples of agestis than other “subspecies”, and it is therefore best to start with agestis. A statistical check on the effectiveness of lunulation data can be obtained via the formula: n p ± k yjp q n where n is the number of specimens in the sample, p is the proportion with a given characteristic, for example male 5 and 6 lunules, q is the remainder, and k is a constant which depends on the degree of signifi­ cance selected. For 5% significance k = 1.96. This is the standard choice and means that limits can be drawn up for a sample of any number n, and that statistically similar groups will be within the limits: there is only a 5% chance (1 in 20) of a similar group being outside the limits. All male specimens from colonies in each of the three classes mentioned have been checked to find the average figures for their 5 and 6 lunule content, and the data are condensed into Table 4. Table 4: Statistical comparisons for male upper forewing lunulation data. Total n 5&6L P (n=n1p00) kjpqn High % Low % T3 all agestis 850 705 0.829 82.9 7.4 90.3 75.5 4/4 21/21 hybrids 721 170 0.236 23.6 8.3 31.9 15.3 2/7 4/14 all./art. 384 12 0.031 3.1 3.4 6.5 0 3/3 8/8 The information in Table 4 needs expanding. The left-hand column 1 gives the total number of male specimens followed by 2 which gives the number with 5 and 6 upper forewing lunules. Then 3 gives p = column 2 divided by column 1. The 5 % significance limits vary with sample size, the smaller the sample the wider the limits. This is why column 4 has been set at n = 100, a constant but reasonably high figure. Column 5 gives the ± figure, producing the High and Low 5 % significance limits in columns 6 and 7. Column 8 (T3) shows the number of colonies/areas in Table 3 which are inside the limits compared with the totals, and the right-hand column 9 gives the picture for all colonies/areas for which there is data. In 9 some of the Table 3 data has been split into smaller units, e.g. individual old English counties for agestis. Other old counties 77 and the other two univoltine areas have been added. The middle column labelled hybrids is split into 1 km squares in Durham and north Lanca­ shire. In Britain it goes from Pickering to the north of Scotland. The bot­ tom line labelled allous/artaxerxes contains specimens north of 56° N in Scotland in 6 old counties groups and also includes a series of Aricia arta­ xerxes vandalica (Kaaber &; Hoegh-Guldberg, 1961) from Hirsthals, north Denmark which are in Jarvis’ collection at the Natural History Museum in London, plus the European allous figures. The 5 % significance limits from Table 4 are shown in Fig. 2 for the 3 clas­ ses to which they relate. They show that nos. 1-4 are all well within their limits, as are nos. 12-14. The same applies to the larger picture in column 9. In these overall checks individual calculations have been made to pro­ duce 5 % significance limits for each colony/area. However, the hybrids in the intermediate zone show a very different picture, and it must be re- memered that the highest and lowest figures in Durham and south Swe­ den are from colonies very close together. Provided any specimen had its locality labelled it was accepted. Thus the data bank takes no account of climatic variation, whether the specimen was first, second or only brood, what the foodplant was, whether it was caught in the wild or bred and if bred whether this was via normal or forced heat and light conditions. The approach is rugged enough and stable enough to produce consistent re­ sults for males. The same goes for agestis females where 487 out of 512 have 6 lunules giving a higher figure for p of 0.951. Some of these are in Tables 2 and 3: again when split up into old counties or identified areas all are within the 5 % significance limits. The above figures for agestis are impressive and provide a stable position against which to judge the lunu- lation of any other colony. Returning to the bottom line, while the male lunulation is consistently low, female lunulation is much higher and variable as Table 2 shows, thus leaving all Scotland in the hybrid zone. A check on cramera lunulation showed that out of a total of 37 specimens, 20 males and 17 females, all had the full complement of lunules although males did show some varia­ tion in lunule size. As far as specimens with 0 lunules are concerned, the agestis colonies have virtually none, while in the group with Scotland north of Lat. 56° N there is a total of 222 males out of the 384, making p = 0.578, and the me­ dian line at 57.8 %. All the colonies are within the 5 % significance limits. 78 The percentage of 0 lunule males will rise further in colonies in north Sweden and Norway; Hoegh-Guldberg records that at Lyngenfjord, the most northerly colony, no males were seen with any lunules. As far as males are concerned, there may be two stable lunulation zones, one with p near 1 and another at p = 0.578. However, more work would be needed to investigate this aspect. The p = 0.578 figure is useful when looking into the constitution of colonies at Durham and Sandhammaren. Pickering and Rügen The first distinct drop in male lunulation compared with agestis occurs at Pickering in north Yorkshire which is only 18 km westnorthwest from the nearest agestis colony in the Yorkshire Wolds. At Pickering the 5 and 6 lu­ nule content at 40.9 % is roughly half that of the agestis colonies, a large difference well outside the 5 % significance limits. At the same time the 0 lunule content has not increased, while at other colonies further west and north the 0 lunule content is significant. Pickering does therefore have characteristics which represent the start of lunulation change just outside the agestis limit. Since there are similarities between the colonies at Dur­ ham and south Sweden which both contain a significant 0 lunule content, it seemed worthwhile seeking a possible parallel to Pickering in north Germany. Thanks to the interest and expertise of Dr. Klaus Schurian both bred and wild specimens were obtained from Inseln Rügen and their lunulation characteristics are included in the Tables. It is difficult to be precise in any comparison, and in this case the difficul­ ty lies in the single 0 lunule specimen shown in Table 3. Many more spe­ cimens would need to be checked to decide what the 0 lunule content finally is. In the case of the Peak district the 0 lunule specimen occurred early, but has not recurred in a total of 138. There is a distinct similarity between the lunulation patterns for Pickering and Rügen, and even if a low 0 lunule content were to be confirmed, it is still near the change from agestis. The lunulation approach has identified a gap in the north-south progression east of Britain in Scandinavia which has been looked for and confirmed by the presence of the colony at Rügen.