Fig. 16—Petroica macrocephala. Mean wing and tail length of males of the five geographic subspecies plotted against mean temperature of habitat. See text for further explanation.

pairs). Current interpretation of the mechanism governing such intra-specific clines correlated with geographic gradients is that they are selectively determined. There is a prima facie case for believing that response to a geographic rule may be fairly rapidly attained by races of a far-flung species, granted that the species is initially obedient to it: should members of a species already exemplifying a geographic rule invade a new thermal environment, selection pressure will immediately tend to modify the invaders in terms of that rule. The adaptational control of size in accordance with Bergmann's rule is significant, because absolute size may determine allomorphic differences in proportions, differences that appear important functionally and which are doubtless under selective control, but which are due primarily to the correlation of body size with climate.

In Miro there is no consistent obedience to the Bergmann principle: on the contrary, while the North Island longipes is smaller than the South Island australis, the evidence points to intra-racial clines with slope opposed to the Bergmann rule, with an abrupt discontinuity in the character gradient at Cook Strait. It must be noted that the differences between the means on which figures 15 and 18 are based are not in all eases statistically significant (the single aberrant skin from Wellington district is omitted): nevertheless the phenomenon seems worth recording, if only as a stimulus for the acquisition of more complete data. Mayr (1942, p. 90) notes that “the reasons for exceptions” (to the Bergmann rule) “are seldom apparent”: no guess is hazarded for this case, but the discrepancy between intra-group and inter-group slopes is not without parallel (see Falla, 1940, p. 229, for cases in seabirds).

Tail/Wing Ratio: In Petroica there is a tendency, illustrated in Fig. 17, for the tail to be shorter, in relation to the wing, in tropical races and longer in sub-Antarctic races. Lack of information about body weight hinders interpretation.

The question that arises is: which, if any, of the linear dimensions is correlated with general size? Amadon (1943) has made a cogent plea for the recording of weights in the routine examination of bird specimens, because “we cannot fully evaluate the biological significance of geographical variation in measurements of appendages without first relating these measurements to general size.” An attempt to use the dimensions “body length” obtained by subtracting “tail length” from “length of skin” was fruitless because of the variation in methods of taxidermy, and because of the smallness of the differences involved. When mean values for each dimension are plotted graphically against each race (the method used by Miller, 1941) some

Fig. 18—Clines in wing length in races of Petroica (Miro) australis (Spar.).

general correlations are evident, but it is impossible to tell by inspection whether such correlations are isometric or allometric. The most satisfactory method of demonstrating the relationship between the size of one organ and that of another in a related group of races is by plotting the information on a double logarithmic graph when, if the points for several different forms fall on a straight line “curve,” a constant growth rate (in this case phylogenetic and not ontogenetic) may be inferred. If the curve slopes at an angle of 45° with abscissa and ordinate, the relationship between the two organs is isometric—i.e., change in one is in direct proportion to change in the other, but if the slope of the curve is at any other angle, the relationship is allometric—i.e., changes are differential. Allometry between adults of different races, implying phylogenetic, rather than ontogenetic changes, is known as allomorphy (Huxley, Needham and Lerner, 1941) or heteragony.

The wing and tail measurements^* of races of P. macrocephala, plotted on a log-log grid, fall on a reasonably straight line “curve” indicating allomorphic relationship. (Fig. 19C.) Now, when the tarsus/tail and tarsus/wing relationships are plotted in the same way (Fig. 19, A, F) they also indicate, with an exception to be noted, definite relationships leading to the following argument:

1. The tarsus and tail bear an isomorphic relationship to each other—i.e., the curve slopes at 45° (exception: chathamensis).

2. The slopes of the tarsus/wing and tail/wing curves are parallel and allomorphic, sloping at an angle of about 52°.

3. There seems no reason why changes in tarsus and tail lengths should be isomorphic unless both are isometric with body size, although this statement can hardly be proved without body weights.

4. If tarsus and tail are isometric with body size as they are with each other, then the wing in Petroica macrocephala exhibits negative allometry. In the formula for the allometry involved (y = bx^k) the constant k = 0·81 and b = 1·85.

The wing and tail measurements of the Australian, Norfolk Island and Pacific races of P. multicolor have not been found to bear any simple relationship to each other. In the subgenus Miro there is an

Figure 19

increase in tail/wing ratio from 70 in the North Island longipes to 74 in the South Island australis and 75·7 in the Stewart Island rakiura: as in P. macrocephala the tail is relatively longer (or the wing relatively shorter) in the south, but in this case there is no simple allomorphy since the size-latitude relationship is irregular. This suggests that the tail/wing ratio may be selectively controlled by environment independent of the allomorphic relationship demonstrated in Petroica. New Zealand Petroica are exceptions to the generalisation of Rensch and others that wings of races that live in a cold climate are relatively longer than those of races that live in a warm climate.

When the mean dimensions for wing, tail and tarsus in Miro are plotted on a double logarithmic grid^* there are approaches to straight line curves in some cases, but notable departures in others (Fig. 19, B, D, B). The tarsus/tail curve (B) approaches the isometric slope (45°) but longipes is off the line suggesting an excessively short tail, and traversi is also aberrant with a tarsus shorter than the rest of the series (ante p. 146). The wing/tarsus curve (E) is allometric (slope 38°) but traversi is divergent, again because its tarsus is not long enough to put it into the series. The tail/wing ratio curve (D) is also allometric, with a slope not greatly different from that of wing/tarsus; longipes is again aberrant in the direction of shorter tail. There are almost as many exceptions as there are points for the drawing of these “curves” so that the situation is capable of more than one interpretation. The inferences affecting wing and tail are:

1. Wing tends to be positively allomorphic in Miro—i.e., larger forms have longer wings. Since the wing of traversi is believed degenerate for other reasons than its shortness, the generalisation might be better expressed as a tendency to reduction of relative wing length in smaller forms. That the allomorphy is in the opposite sense from that in P. macrocephala is inevitable since the size gradients in the two groups slope in opposing directions, whereas the gradients in tail/wing ratio have the same direction of slope.

2. Petroica (Miro) australis longipes has acquired a short tail by departure from the general rule (in New Zealand Petroica) that tail tends to vary isomorphically.

3. The high tail/wing ratio of P. (M.) traversi could have been acquired by reduction of tail and wing in accordance with the same allomorphic formula that pertains among the robin populations of the South Island and Stewart Island.

Tarsus Length: The tarsus/wing ratios in P. multicolor and P. macrocephala at first sight show no regular geographic correlation. However, the highest values for multicolor all occur in insular races; macrocephala has a higher value than continental multicolor and for the insular races of macrocephala the values are higher still, that of chathamensis being highest of all. The plotted data (Fig. 19, A, E) suggests that the discrepantly long tarsus of chathamensis is due to phylogenetic positive allometry of that bone; the alternative that wing

and tail have undergone correlated negative allometry is not in accord with the sum total of evidence.

In the subgenus Miro the tarsus is relatively longer than in other Petroica (indeed than in other Muscicapidae). Not only are tarsus/wing and tarsus/tail ratios high, but the mean toe/tarsus ratio, which ranges from 76 to 83 in races of P. macrocephala, is 74 in P. traversi (supporting other evidence that the high tarsus/wing ratio of that species is in part the result of wing degeneration) and only 65–68 in australis. It is apparent that the long tarsus of Miro is the result of positive allometry in the history of the group just as in P. m. chathamensis. If the environment on islands is in some way selective for long tarsus, that character of Miro may be the result of the same adaptive trend whereby P. m. chathamensis and P. multicolor multicolor (for instance) acquired longer tarsi than their nearest relatives. The forest environment of New Zealand had in common with such oceanic islands as Norfolk and the Chathams an almost complete freedom from mammalian and other predators; and, in P. australis at least, the habit of forest-floor feeding gives a plausible adaptive significance to the long tarsus.

Wing Shape: Wing formula in Petroica varies from 2 = 6/7 to 2 = 10: in the following table the figure in parentheses is the number of skins examined for this character.

In the “continental” races, the tip of the wing is relatively long; in the insular races shortening of the tip reduces the second primary below the 7th, and, in macrocephala, below the 8th. In Miro the process has gone further. The shortening and roundening of the wing in Neozelandic Petroica may also be illustrated by comparison of the actual and relative lengths of the rudimentary first and the second primaries.

It seems likely that the vestigial first primary has not taken part in the reduction of the wing in insular forms, and there is incomplete evidence that it has a roughly isometric relationship with general body size: at any rate it is a fair generalisation that shortening and roundening of the wing in insular races has resulted in a relatively