Amenia longicornis (Malloch)

Colless, D. H., 1998, Morphometrics in the genus Amenia and revisionary notes on the Australian Ameniinae (Diptera: Calliphoridae), with the description of eight new species, Records of the Australian Museum 50 (1), pp. 85-123 : 87-91

publication ID	https://doi.org/ 10.3853/j.0067-1975.50.1998.1275
DOI	https://doi.org/10.5281/zenodo.4657266
persistent identifier	https://treatment.plazi.org/id/9A5987F8-163F-FFF6-FA1F-FE63F88CFCC9
treatment provided by	Felipe
scientific name	Amenia longicornis (Malloch)
status

Treatment

Amenia longicornis (Malloch) View in CoL

Crosskey (1965) saw only 12 males and four females of this species, but marked variation in several characters was clearly evident. To quote Crosskey: "Male specimens from South Australia have short antennae, a short facial carina which is regularly elongate in facial view, a broader vertex, rather strongly developed outervertical setae and a strongly marked forward bowing of the costal margin of the wing; most males from Western Australia, on the other hand, have elongate antennae and a long rather lanceolate facial carina, a narrower vertex, very weak or hair-like outer vertical setae and no well marked bowing forward of the costa". On the few specimens available these characters suggested a gradual transition from one form to the other, and Crosskey could only speculate on the possible existence of an East-West cline .

The abundant material now available shows that male (but not female) characters provide a clear geographic division into eastern and western populations ( Fig. 1 View Figure 1 ). The former is, as far as known, isolated in the Flinders Ranges and around the head of Spencer Gulf in South Australia, extending north to the vicinity of Alice Springs. The latter occurs across the Nullarbor Plain and has been taken as far east as Nunjikompita (longitude 134°20'E); but is rare or absent between there and Iron Knob in South Australia (longitude 137°09'E). For instance, it was not taken during intensive collecting at Poochera (longitude l 34°50'E), whereas numerous specimens were taken a few days later around the head of the Great Australian Bight (longitude ca 131°OO'E). To the west, there is a more or less continuous distribution as far as the vicinity of Norseman (longitude 121°50'E), extending south to the coast at Esperance and north to Widgiemooltha ; but only 5 specimens are known from areas further north or west. These include an enigmatic specimen from Payne's Find, 500 km northwest of Norseman (see below).

The analyses that follow concentrate, therefore, on possible differences between the isolated populations, and on possible clinal variation on an east-west axis.

In all 54 males were studied. The eastern popUlation was represented by only 8 (3 from Wilpena Pound and 5 from the Whyalla-Iron Knob area )-a meagre sample, but (I believe) sufficient to support the conclusions reached. The western specimens included the holotype (from an unknown locality in Western Australia) and samples of from one to ten specimens from 15 localities covering the known range. In general, sample sizes were so small that MDA was applied only to the East-West contrast .

PCA yielded 3 components that together explained 89% of variation. Examination of scores on component 3 showed that specimens were scattered more or less at random on that axis, so it was considered no further. Component 1, explaining 70% of variance, was an "overall size" component, with most dimensional characters showing substantialloadings of the same sign. Component 2 explained a further 13% of variance, contributed principally by Ant and FeZ (positive), with 88 Records of the Australian Museum (1998) Vol. 50 secondary contributions from Uob and Cwm (negative). In the ordination shown in Fig. 2, which plots scores on component 1 against those on component 2, it can be seen that eastern and western populations are clearly differentiated on component 2. An exception is a specimen from near Madura, W.A., with an unusually short (perhaps aberrant) antenna, but even it falls outside the envelope of points for the eastern specimens. On component 1, although the eastern specimens tend to be larger, 4 out of 8 fall within the range of the western population. The specimen from Payne's Find, however, is clearly differentiated from all others by its size.

Thus, eastern and western populations are clearly separated by a complex that included 5 of the 6 characters noted by Crosskey; significant variation was not confirmed for vertex width, but was found in one extra character (development of upper orbital bristle). While important, the bristle characters were not crucial. PCA of dimensional charactersalone yieldedan ordination (notshown) very similar to Fig. 2, with (as expected) component 2 dominated by FeZ and Ant . Interestingly, the two populations are simply and completely separated by "shape" of the carina, as indexed by the ratio CwblCwm ( Table 1 View Table 1 ; note that apparent overlap of ranges is due to rounding).

Finally, MDA showed eastern and western populations to be completely discriminated by a function for which scores are summarised in Table 1 View Table 1 . The major loadings were on Frw and Csb, with smaller but substantial contributions from AMi, Eyh, and Fe ! (the last negative, the others positive). MDA thus confirms the significance of the one remaining character (Frw) from Crosskey's set. A crude but effective version of the discriminant function is the ratio (Frw + Eyh Fe!) IHdw ( Table 1 View Table 1 ).

A specimen that has recently come to hand (Palm Valley, MacDonnell Ranges , Northern Territory, 21-22 May 1983, G.A. Rolloway; AM) allows prospective application of the MDA function. As might be expected, its measurements conform well with those of the Eastern form .

The Payne's Find specimen is exceptional, in that MDA allocates it to the eastern form. Although in many ways similar to the eastern specimens, it is distinguished by its larger size, intermediate scores on principal component 2 and the discriminant function, and, of course, its immense geographic separation. Its status remains obscure--especially since its two nearest neighbours (from Walyahmoning Rock) are quite normal "western" specimens.

Figure 2 includes a suggestion of differentiation, perhaps clinal, within the eastern form (there is an apparent tendency of specimens from the same locality to cluster on component 2). However, there is clearly no suggestion whatever of such a phenomenon in the western form. Note the wide scatter of specimens in the samples from east and west of longitude 125°00. PCA of western specimens alone (not shown) releases them from within-group correlation effects (the linear clustering seen in Fig. 2); and, as expected, they scatter more or less at random on components 1 and 2. That is to say, the variation observed by Crosskey turns out to involve, not a simple cline, but two distinct geographic forms-with, perhaps, clinal variation within one of them. We may add to that the intriguing hint of a north-western population (Payne's Find) that resembles the eastern, rather than the western form ( Table 1 View Table 1 ).

Amenia imperialis Robineau-Desvoidy

Crosskey (1965) recognised that A. imperialis could be readily distinguished from the very similar A. dubitalis Malloch ; but, in view of the closeness of that similarity and an apparent separation in their geographic ranges, he preferred to relegate the latter as a subspecies of the former. Accumulated material now shows that the degree of geographic separation is negligible, the range of A. dubitalis being almost completely included within that of A. imperialis ( Figs 3-6 View Figure 3 View Figure 6 ). The consequent likelihood that the two are in fact perfectly "good" biological species calls for more detailed study of their morphological distinctness.

Crosskey's characterisation of the two forms now requires some amendment (see Part 2, below), but his sole dimensional character-the width of frons or vertexclearly applies. Preliminary tests pointed to correlated variation in other parameters ofhead shape, although other bodily dimensions showed no sign of significant variation.

The morphomettics of head shape were, therefore, investigated further in a series of 68 males and 63 females, chosen to cover the full geographic range of each taxon as well as a representative range of body sizes. They comprised 20 males and 20 females each of A. dubitalis and the "classical" A. imperialis (from coast and adjacent mountains), identified by the characters recommended by 90 Records of the Australian Museum (1998) Vol. 50 Crosskey (1965). The remainder were identified as A. imperialis , but came from outlying areas or seemed for other reasons worthy of special scrutiny. In the light of the preliminary tests, 6 dimensions were chosen for measurement: frons width (Frw), frons length (Frl), gena width (Gnw), face length (FeZ), eye height (Eyh), and head width (Hdw). All were measured as described above for A. longieornis ; this implies that Frw was measured towards the middle of the frons in males, but across the posterior ocelli in females.

The close similarity of A. imperialis and A. dubitalis and the propriety of analysing them as a single groupwas first tested by separate PCA for the 20 dubitalis and 20 "classical" imperialis males. The 6 loadings on each of the first 3 components from the two analyses (18 paired observations) were almost perfectly correlated (r=0.96). PCA of data forthe full set of 68 males then yieldeda first, "general size" component explaining 73% of variation. Component 2 explained 18%, with dominant, positive loadings on Frw and FeZ and negative on Eyh. Component 3 explained a further 8%, bufprovided no obvious taxonomic information and was not considered further.

Figure 7 shows a plot of component 1 against component 2. Not only is A. dubitalis clearly separated from A. imperialis , but an unexpected pattern of variation is evident within the latter taxon. Two distinct geographic segregates are separated from the main cluster: one comprising six specimens from localities near Cooktown and two from near Cairns, the other including two specimens from the Northern Territory, eight from the Kimbedeys (northwesternAustralia), and seven from several arid, inland localities in New South Wales and Queensland. I am callingthe formerthe "Cooktown form" and the latter the "Western form" for the present. Other plots (not shown) show quite clearly that the distinction between A. dubitalis , the Cooktown form, and other A. imperialis is governed largely by the regression of Frw on overall size; a similar regression of FeZ on size separates the Western form of A. imperialis from all the others. For practical purposes, these distinctions are well reflected in the ratios summarised in Table 2 View Table 2 .

Accepting the foregoing groupings, MDA (not plotted) confirms that the 4 groups are indeed clearly identifiable. The first discriminant axis explains 80% of variance and, as expected, is dominated by Frw; Gnw and Fel also have lesser but substantialloadings. The second explains 19%, with its principal loading on FrI. There were lesser but substantialloadings on Frw, FeZ, and Eyh (the first two of these were also prominent on the third principal component). All 4 groups are discriminated on the first axis alone, the second serving mainly to accentuate the distinction between populations of A. imperialis from the coastal regions and the "Western form".

PCA of data for females showed rather similar results, but less clear-cut. The first, "size" component explained 91% of variation; the second explained 4%, with principal loadings on Frw and Fcl. An ordination on components 1 and 2 (not shown) has the "Cooktown form" mingled with other A. imperialis , but specimens of the "Western form" are clearly split off. The remainder form a loose cluster showing a partial separation into A. dubitalis and A. imperialis .

Further study of the geographic forms of A. imperialis revealed that all are distinguishable on additional, qualitative attributes. These are discussed in Part 2 of this paper.

To sum up, A. dubitalis and A. imperialis are clearly separated, not only by qualitative characters, but also by quantitative parameters of head-shape. Also, morphometrics of " imperialis " point to the existence of at least 3 distinct populations. Two are geographically separated, one of them restricted largely to the Northern Territory and Kimberleys, the other to eastern Australia, but with some slight overlap in the interior of New South Wales and Queensland. A third population is everywhere sympatric with the eastern Australian form. The taxonomic implications of these findings are given below in Part 2.