89

On the taxonomy of Dipriodonta sericea Warren (Lepidoptera : Drepanidae)

By CHRISTOPHER WILKINSON Department of Biological Sciences, Portsmouth Polytechnic*

SYNOPSIS Details are given of a computer-aided assessment of the taxonomic position of the

species Dipriodonta sericea Warren. A complete taxonomic account of sericea is given for the first time, and its affinities with related species are discussed.

INTRODUCTION A REVISION of the major genera of white Drepanids from the Indo-Australian Region has been completed by both classical and numerical methods (Wilkinson, 1967a, b; 1968). Since then I have examined an additional species from Assam, which has not been properly described and whose taxonomic position requires clarification. As discussed on p. 96 this work has resulted in the production of a new numerical method that enables a species to be added to a principal co-ordinates analysis.

Dipriodonta sericea Warren comb. rev. Dipriodonta sericea Warren, 1897 : 14. Dipriodonta sericia Warren; Hampson, 1897 : 286 (incorrect subsequent spelling). Dipriodonra sericea Warren; Warren in Seitz, 1922 : 448, pl. 48c. Macrocilix sericea (Warren) Gaede, 1931 : 6. Macrocilix sericea (Warren); Watson, 1968 : 131.

Male and Female Head: palps reach labrum, dark brown distally, white proximally; lower part of head brown,

upper part white; vertex white with some brown scales; collar white; antennae widely lamellate in males, closely lamellate and narrower in females, white with brown scales distally. Wings: venation as for Ditrigona, with quadrate areole and R1 of fore wing arising from it. Sc -F RI of hind wing anastomoses with Rs. Ground colour, costa and fringe white; highly lustrous, semi-transparent patch in centre of fore wing. Fore wing marked as in plate, with indistinct or reduced double sub- basal, single antemedial, double postmedial and single terminal fasciae, brown. Distal postmedial and terminal fasciae straight, others lunulate, often reduced to spots. A spot immediately distal to end of cell and three apical brown “smudges” present. Hind wing with weak sub-basal, double postmedial and single terminal fasciae; terminal sometimes reduced to spots on veins; additional brown marks distal to postmedial fascia; minute disco-cellular spots present. Ventral surface of all wings white but irrorate with black between cell and costal margin. Legs: front of fore leg brown, remainder white, other legs white; epiphyses present; mid-leg with one pair of terminal spurs on tibia; hind leg with two pairs of tibia1 spurs, one pair terminally placed. Thorax white and buff. Abdomen white with few buff scales.

Genitalia.-Male as in figures 1 4 ; uncus single; socii small, curved posteriorly; valves broad, half length of genitalia, not reaching uncus; saccus small; gnathos absent; juxta not projecting; aedeagus lanceolate. cornutus without spines; eighth tergite weakly sclerotised as shown; eighth stcrnite with a pair of horn-like octavals extending anteriorly to form apodomes. Female as in figure 5 ; ovipositor lobes simple; ductus bursae short; bursa copulatrix small; accessory sac and apophyses apparently absent.

Wing measurements (taken from centre of mesothorax to fore wing apex).-Lectotype, 10 mm., male, 10-12 mm. (18); female, 10-12 mm. (10).

Diagnosis This species is distinguished from those of the allied genera Ditrigona and Teldenia

by the wing markings and male genitalia. The fore wings have a hyaline patch in the cell and the area posterior to it (see Plate I ) . In the other genera this character occurs only in Ditrigona species-group derocina, in which the wings are entirely hyaline. The presence of a large spot on each fore wing distal to the end of the cell is another diagnostic feature. I n the male genitalia the absence of posterior extensions to the * At present on leave from Portsmouth, at Entomology Research Institute, Ottawa.

Proc. R. ent. Soc. LonJ. (B). 39 (7-8). Pp. 89-98, 9 figs., I plate. 1970. 4

90 C. Wilkinson on the taxonomy of Dipriodonta sericea

2

4 1 FIGS. 14-Dipriodonro sericea, male genitalia: ( I ) genitalia; (2) aedeagus; (3) eighth tergite;

(4) eighth sternite.

L FIG. S.-Dipriodonru sericea, female genitalia.

valves separates sericea from the Teldenia species and, together with the single uncus, separates it from species in the mytylata and triangularia groups of Ditrigona. The protruding valves and shape of the aedeagus distinguish sericea from others in the quinaria and derocina species-groups of Ditrigonia. The wing span of this species is much less than in the majority of Ditrigona species. D. sericea is separated from the Macrocilix species not only by external appearance but also by the venation and genitalia. In the fore wing of the presently known species of Macrocili,r, R2 arises from R I . but in sericea it arises from Rs; Sc + R1 of the hind wing of Macrocilix do not anastomose with Rs as in sericea. For genitalia and other characters of Macroci1i.v species see Watson (1968 : 131).

C. Wilkinson on the raxonomy of Dipriodonta sericea 91

Discussion This species was first described by Warren (1897 : 14) in the monotypic genus

Dipriodonta, which was then considered to belong to the Geometridae. Hampson (1 897 : 286) referred to this species incorrectly as sericia. Gaede (I93 1 : 6) recognised that sericea had been described in the wrong family and transferred it to the genus Macrocilix (Butler, 1886 : 18) in the Drepanidae. Although sericea shows superficial external similarity to the Macrocilix species, it has, in fact, few characters in common with them. Therefore, sericea is here replaced in Dipriodonta gen. rev., which is still monotypic but is recognised as belonging to the Drepanidae.

At the beginning of this re-examination of the species, it was placed in the quinaria species-group of Ditrigona, in a systematic position between the species quinaria (Moore) and obliquilinea (Hampson). This decision was based only on classical taxonomic methods, which included examination of the male and female genitalia for the first time. Although sericea has unique characters, the genitalia show features that indicate affinity with all the species-groups of Ditrigona. For example, the aedeagus is very similar to that of titana, the ostium and ostial pocket resemble those of berres, and the male genitalia as a whole have similarities with those of the quinaria species-group. The main difficulty was to decide with which species-group sericea had the greatest affinity.

The numerical tests described on p. 92 were then applied, and less affinity was found to occur than had been expected; primarily on the basis of numerical taxonomy, Dipriodonta was thought to represent the correct genus for sericea. Later considera- tion of the findings of both classical and numerical methods suggested that, in this instance, the numerical result was, in fact, the correct one.

Distribution Only known from the Khasia Hills, Assam.

Specimens examined Warren described the species from three males. Although one specimen is

labelled by him as “type”, it has not been designated in print. This specimen is in very poor condition, and therefore I designate as lectotype one of the other syntypes with the following data: lectotype S, ASSAM: Khasis, vii. 1896 (Nat. Coll.), Roth- schild Bequest. B.M. 1939-1, Drepanidae Genitalia Slide No. 1784, in British Museum (Natural History) ; paralectotypes: 2 6, data as lectotype.

Other specimens examined: 16 S, 10 0, locality as type, vi-vii. 1896 (Nat. Coll.), in British Museum (Natural History).

NUMERICAL TAXONOMY A numerical analysis of Dipriodonta sericea was carried out to find what phenetic

affinity it had with the species in the genera Ditrigona, Teldenia and Argodrepana. Programmes for cluster analysis and principal co-ordinates analysis had been previously used for the above genera without sericea (Wilkinson, 1967a, b ; 1968; 1970b), and the following new analyses incorporating sericea were therefore carried out :- (1) A complete new cluster analysis including sericea together with the 87 species from the above genera. (2) A complete new principal co-ordinates analysis including sericea together with the same 87 species. (3) An analysis to superimpose sericea on to the original principal co-ordinates analysis of the three genera. The accuracy of this method was tested by comparing the results with those of the previous analysis ($2).

The numerical taxonomy involved the use of two of J. C. Gower’s programmes on the Orion computer at Rothamsted Experimental Station. They operated on 97

Proc. R . ent. SOC. Lond. (B) 39 (7-8) Pp. 89-98, 9 figs, 1 plate. 1970. 49

92

character tests from 88 species and subspecies in three genera. The first, a cluster analysis (CLASP), is basically a hierarchical sorting programme, which was revised by Ross from an earlier programme of Gower. It produced a similarity coefficient between each pair of species, which was defined as a number in the range (0, l), so that species theoretically identical in all respects would have similarity 1. For a given phenon level L, a taxometric grouping G(L) of the species was defined so that every species belonged to a group and no species belonged to more than one group. CLASP arranges “singly-linked” groups, in which every member of a group GI is similar to some other member of the group at a phenon level greater than L, but similar to all non-members of GI at a phenon level less than or equal to L. These groups cannot be further subdivided. L was initially set to a maximum phenon level and then decreased by 2.5 per cent. at eachstage until only a single group re- mained (in this instance when L = 62.5 per cent).

TABLE I.-Results of 97 character tests carried out on the species Dipriodonta

C. Wilkinson on the taxonomy of Dipriodonta sericea

Test 1 2 3 4 5 6 7 8 9

10 I 1 12 13 14 15 16 17 18 19 20 21 22 23 24

sericea for numerical analysis Result Test

2 25 E 26 A 27 A 28 C 29 A 30 C 31 A 32 A 33 1 34 I 35 D 36 A 37 1 38 0 39 0 40 A 41 A 42 B 43 3 44 A 45 0 46 A 47 0 48

Result Test 120 49 100 50 I20 51 100 52 6 53 0 54 C 55 C 56 0 57 C 58 0 59 E 60 0 61 C 62 C 63 0 64

C/D 65 0 66 E 67 B 68 B 69 A 70 I 71 0 72

Result Test 0 73 C 74 C 75 A 76 1 77 1 78 2 79 1 80 I 81 A 82 0 83 I 84 1 85 A 86 0 87 0 88 0 89 0 90 0 91 3 92 1 93 I 94 2 95 0 96

97

Result 0 0 1 I 0 0 I 1 1 I 2 I 0 0 0 0 0 0 0 1 1 0 0 0 0

The second programme, a principal co-ordinates analysis (LRVSRCZ), starts with the similarity matrix and derives a new matrix in which the elements measure the “distance” between the species and subspecies according to the formula dij = d(2( 1 - 1 sf;)). This new distance matrix is transformed by the rotation of the co-ordinate axes so that the new axes are aligned with the directions of maximum variation. This is basically a latent root problem. New co-ordinate directions, in which the variation is statistically negligible (corresponding to a latent root of less than 1) are ignored, thus reducing the dimensions of the problem. The aim is to preserve, as accurately as possible, in the reduced number of dimensions, all the distances in the original many-dimensional object-space. The effect of the analysis in this instance was to produce ten latent roots giving co-ordinate directions for each species and subspecies. With these ten directions the overall affinities can be deter- mined. If analysis within a smaller group (e.g. a genus) is required, strictly a re- analysis of this subgroup is desirable. However, if the residual that measures the effect of ignoring the remaining directions is small, the ten directions previously selected will still contain the principal variations within the subgroup. If a genus

C. Wilkinson on the taxonomy of Dipriodonta sericea 93

TABLE I I.-Species in the genera Ditrigona, Teldenia, Argodrepana and Dipriodonta The number preceding the species name corres-

The percentage following the name is the used in the numerical taxonomy. ponds with that used in the figures. similarity coeficient of that species compared with Dipriodonta sericea.

Ditrigona Species-group 1 derocina 1. derocina 2. diana 3. pruinosa

4. spilota 5. inconspicua 6. spatulato 7. furvicosta 8. jardunaria 9. media 10. innotata I I. sericea ( Ditrigona) 12. quinaria 13. nivea (Ditrigona) 14. erminea 15. spodia 16. leucophaea 17. obliquilinea 18. thibetaria 19. idaeoides

Species-group 111 triangularia 20. triangubria 2 1. regularis 22, Iitana 23. ponienaria 24. typhodes 25. polyobofaria 26. sciara 21. fasciata

Species-group 1V mytylata 28. lineata 29. tephroides 30. legnichrysa 3 I. policharia 32. artema 33. marmorea 34. candida 3 5. conflexaria 36. micronioides 3 7. cerodeta 38. quinquelineata 39. mytylata 40. margarita 41. aphya 42. berres 43. chama 44. chionea 45. platytes 46. Virgo 47. cirruncata 48. komarovi

Species-group I 1 qirinaria

Percentage

49.1 48.8 43.9

61.3 62.6 64.6 62.7 60.7 62.3 62,4 58.9 59.9 61.6 62. I 61.7 61.4 62.8 61.2 56.0

51.4 47.6 56.0 52.2 50.9 56.0 53.3 31.0

58.9 57.5 55.4 52.9 52.0 61.9 53.3 40.6 54.1 53.5 60.0 52.7 60.3 62.6 63.6 58.4 63.3 53.0 69.0 69.0 64.8

Teldenia Species-group I niveata 49. niveata 50. obsoleta 5 1. specca 52. alba 53. vestigiata

Species-group I1 nigrinotata 54. latilinea 55. psara 56. nigrinotata 57. melanosticta 58. celidographia 59. a 60. geminata 6 I . aurilinea 62. sparsata 63. illunata 64. desma 65. inanis 66. apata 67. monilhta 68. helena

Species-group 111 pura 69. unistriaota 70. nivea (Teldenia) 71. b 72. pura 73. cathara 74. argeta 75. d 76. c

Species-group IV strigosa 77. subpura 78. ruficosta (Teldenia) 79. seriata 80. strigosa

A rgodrepana 8 1. verticata 82. galbana 83. auratifrons 84. denticulata 85. rujicosta (Argodrepana) 86. tenebra 87. umbrosa

Dipriodonta 88. sericea (Dipriodonia)

Percentage

56.4 53.8 55.1 53.5 52.2

49.3 55.6 51.4 56.2 56.2 55.9 544.9 60.8 54.2 52,l 51.6 56.5 59.5 58.5 49.1

48.2 53.5 61.6 59.9 55.0 53.1 60.8 58.9

49.9 45.9 49.6 48.7

54.7 57 .O 59.1 54.5 58.9 41.4 48.1

-

94 C . Wilkinson on the taxonomy of Dipriodonfa sericea

1 4 1 5 1 6 1 7 1 8

I

Fro. 6.-Dendrogram showing the phenetic relationships between the species of Ditrigona, Teldenia, Argodrepana and Dipriodonta.

k 5 c B

I I

t r I

I

96 C. Wilkinson on the taxonomy of Dipriodonta sericea

is examined within the overall group, its basic similarities will be reflected in having very small variation in some of the ten co-ordinate directions; hence the number of co-ordinate directions can be reduced still further. In figure 9, therefore, the direc- tions (A2 and A4) selected for Ditrigona are those for which the variation within the genus was greatest, and not necessarily those which corresponded with the non- numerical results.

Details of the 97 character tests applied and definitions of the terms used have already been cited (Wilkinson, 19673; 1968). The same characters were observed for sericea, and their values are given in Table I. In the figures the species are represented by numbers that correspond to those given in Table 11, in which the species names are also given.

The third analysis ($3) involved adding sericea to the original principal co-ordinates analysis, and comparing the results with those of a new analysis ($2) which included sericea initially. The purpose of this analysis is a side issue, but worthy of mention here. After a taxonomic revision has been completed, further species may be dis- covered which perhaps should be included. When computer methods have been used as an aid in the revision it will also be desirable to find the phenetic affinities of the additional species; such was the case for Dipriodonta sericea. It may be incon- venient to carry out a complete new analysis each time an additional species is found. This analysis was therefore devised (Gower, 1969) to find if an additional species could be added to a previous principal co-ordinates analysis with a good degree of accuracy, and this is the first time the method has been used. Full details of the method and mathematical principles involved have been published elsewhere (Wilkinson, 1970a), and only the taxonomic implication need concern us here. The characters of sericea were used together with those of the original 87 species, which were calcu- lated in the cluster analysis. The co-ordinates obtained by adding sericea were

-. . . I L L L N A 78

3 a E R A V/lF I POINT A53ED

FIG. 8.-Two-dimensional vector diagram with minimum spanning tree of the genera Difrigona, Teldenia, Argodrepana and Dipriodonta.

C. Wilkinson on the taxonomy of Dipriodonta sericea 97

compared with those obtained by a new analysis (§2), and it was found that the two sets of values agree very well. Figure 9 is an example of a sectional diagram showing the change in position of each point arising from the difference in method. It is a pity that sericea is such an out-lier, but the experiment shows that adding a point will produce a satisfactory result without the need to do a new analysis each time a fresh sample is found.

Once the programmes and modifications exist for adding a point, it is a simple matter to test new examples; this method is thus ideal for suggesting whether or not a new species might belong to an existing taxon.

RESULTS Cluster Analysis

Reference to the similarity coefficients (Table 11) shows sericea to have highest affin- ity with Ditrigona Virgo (Butler) and D. cirruncata Wilk. (69.0 per cent.), a relationship not suspected in the classical taxonomy. However, looking at the species-groups as a whole, Ditrigona species-group 11 quinaria has the highest affinity with sericea; it was in this group, if any, that classical methods placed the species, and a method based on average linkage clustering rather than single linkage clustering might well support this association. In figure 6, however, the dendrogram shows that sericea (88) is really an outlying species, and that perhaps it (andfasciafa (27)) does not belong

-4 A - Q D AF4ALYSS m+ I RINTS 35:

31

b N E W A W V Y S 08FOPolHn

98

to any of the three main genera. The position of regularis (21) has been discussed elsewhere (Wilkinson, 19706). Although the similarity coefficients may seem high for an outlying species, they are not so when examined in relation to the entire S-matrix.

Principal co-ordinates analysis The three dimensional diagram ( h l , 2, 3) (fig. 7) illustrates the distinctness of

Ditrigona, Teldenia and Argodrepana. It can also be seen that sericea (88) occurs amongst the Ditrigona species, betweenplatytes Wilk. (45) and sciara Wilk. (26). Again, sericea appears to be quite central when plotted in two dimensions, either with the three genera (as in fig. 8) or with Ditrigona alone (fig. 9). However, when information from the cluster analysis is superimposed (fig. 8) by drawing lines between the most closely related species and giving the similarity coefficient, it is clear that the distance from 47-88 corresponds with one of the smallest similarities (69 per cent.) and shows that sericea (88) is very much of an outlier. It will also be seen from the analysis that its distance even from the ten-dimensional fit (p. 92) is quite enormous (0.645), which would not be expected between closely related species. In fact, for the original analysis without sericea, the proportion of variance accounted for by ten dimensions is 63 per cent., whereas for sericea it is only 124 per cent. Thus, in spite of first conclusions, the indications are that sericea is not closely related to the species in the other genera.

SUMMARY

C. Wilkinson on the taxonomy of Dipriodonta sericea

A redescription of Dipriodonta sericea Warren is given, including details and figures of the male and female genitalia. The diagnosis separates the species from those in the allied genera Ditrigona, Teldenia and Macrocilix, and it is replaced in the monotypic genus Dipriodonta (Drepanidae). A lectotype and paralectotypes have been desig- nated.

A cluster analysis and a principal co-ordinates analysis, used to assist in ascer- taining the affinities of the species, indicate that D. sericea is not closely related to the species in Ditrigoa, Teldenia and Argodrepana

I should like to acknowledge my gratitude to Mr. John Gower (Rothamsted Experimental Station)

I am also indebted to Portsmouth Polytechnic and the Trustees of the British Museum (Natural for all his help with the numerical taxonomy.

History) for providing facilities to enable this work to be carried out.

REFERENCFS BIJIXER, A. G., 1886, Illustrations of typical specimens of Lepidoptera Heterocera in the British

Museum 6, xv + 89pp., 20 pls. GAEDE, M., 1931, in Lepidopterorum catalogus. GOWER, J. C., 1966, Some distance properties of latent root and vector methods used in multivariate

analysis. Biometrika 53 : 325-338. - 1968, Adding a point to vector diagrams in multivariate analysis. HAMPSON, G. F., 1897, The Moths of India. Supplement. J . Bombay. nut. Hist. SOC. 11 : 277-297,

WARREN, W., 1897, New genera and species of moths from the Old World regions in the Tring Museum. Novit. 2001. 4 : 12-130.

- 1922, in Seitz, Die Gross-Schmerterlinge der Erde 10. ix + 909 pp. WATSON, A., 1968, The taxonomy of the Drepaninae represented in China, with an account of the

world distribution (Lepidoptera : Drepanidae). Bull. Br. Mus. nut. Hist. (Ent.) Supp. 12,151 pp., 293 figs., 14 pls.

WILKiNsoN, C., 1967a, A taxonomic study of a new genus of Drepanidae (Lepidoptera) from New Guinea. Proc. R . enf. SOC. Lo&. (B) 36 : 17-29, 35 figs., I plate, 1 map. - 19676, A taxonomic revision of the genus Teldenia Moore (Lepidoptera : Drepanidae,

Drepaninae). Trans. R . ent. SOC. Lond. 119 : 303-362 pp.. 123 figs., 4 plates, 3 tables, 6 maps. - 1968, A taxonomic revision of the genus Ditrigona (Lepidoptera : Drepanidae : Drepaninae).

Trans. zool. SOC. Lond. 31 : 407-517, 96 figs., 8 pls., 3 tables. - 197Oa, Adding a point to a principal co-ordinates analysis. - 1970b, Numerical taxonomic methods applied to some Indo-Australian Drepanidae. J . nut.

Hist. 4 : 269-88, 16 figs.

London. Drepanidae. 49 : 1-59. Berlin.

Biometrika 55 : 582-585.

1 PI., 3 figs.

Stuttgart.

Syst. Zoo!.

(Manuscript received 22nd August, 1969)

Proc. R. eiit. SOC. Loiid. ( B ) Vol. 39.

PLATE I Dipriorlonru sericeu. external features ( x 4)

PLATE 1

C . Wilkinsoii