WIDTH OF GAPE AS A DETERMINANT OF SIZE OF PREY EATEN BY TERNS School of Australian Environmental Studies, Griffith University, Nathan, Q 4111. Received 17 August 1979; accepted 22 May 1980. SUMMARY HULSMAN, K. 1981. Width of gape as a determinant of size of prey eaten by terns. Emu 81: 29-32. Partial correlations were used to analyse the relations between four morphological characteristics and length of prey for fourteen species of tern. The horizontal width of gape had the greatest effect in reducing the correlations of weight, lengths of bill and wing with length of prey to nonsignificant or negatively significant levels. The horizontal width of gape may be used to predict the optimal size of prey of a predator that eats its prey whole. Comparisons of the frequency-distributions of preferred widths of prey with those of actual prey eaten by various predators may provide information about interactions between potential competitors and availability of prey. INTRODUCTION Mor~holoeical data have often been used to test or mak; ecoiogical generalizations (review Hespenheide 1973). In this paper I concentrate on the relation between size of predator and size of its prey. In studies of birds, length of bill is sometimes used to illustrate the ecological generalization that larger predators eat larger prey than do smaller predators (references in Hespenheide 1973). I aim to show that the horizontal width of gape (relaxed) or simply gape could better predict the size of prey for predators that eat their prey whole than does length of bill and that it directly affects the size of prey that a predator eats. METHODS The lengths of bill, horizontal widths of gape, lengths of wing and weights of fourteen species of tern are given in Table I. Measurements of Black Noddies and Bridled Terns were from live birds at One Tree Island (23"31fS, 152"05'E). Measurements of Blue-grey Noddies, except the width of the gape, were taken from Ashmole and Ashmole (1967), who measured specimens held in the Museum of Natural History at Yale. Measurements of other species (including width of gape of Blue-grey Noddy) were from specimens in museums. Specimens of Crested Tern were at the Queensland Museum, Brisbane, and those of the remaining species were at the Museum of Natural History, Leiden, The Netherlands. I measured length of bill and width of gape to the nearest 0.5 millimetre with vernier calipers. The horizontal width of gape was measured from underneath to avoid damaging the bird's eyes (Fig. 1). A ruler was used to measure length of wing to the nearest millimetre. The tip of the left wing was held against the ruler to allow me to read the measurement. Weights of birds were taken from the literature (the sources are given in Table I), except those of Black Noddies and Bridled Terns, which were weighed with a Pesola balance (300 grams) to the nearest gram. The total lengths of prey were measured with calipers or taken from the literature as indicated in Table I. JustiJication of methods The length of prey eaten by a species of tern varies seasonally (Ashmole and Ashmole 1967; Hulsman 1978) and those fed to dependants and those eaten by the hunters may,differ significantly (Hulsman 1978). I pooled these data to give an estimate of overall length of prey eaten by a particular population of a species of tern. This provides a more practical comparison of length of prey between species than comparing prey fed to dependants of one species with prey eaten by adults of another species. Measurements from skins are suitable for the analysis because relative measures (not only absolute ones) reveal trends in data, i.e., lengths of wings in larger species may change more than those in smaller species in an absolute sense but the relative position of the measurements remain the same. The length of Dunlins' bills Calidris alpinus did not change for at least eighteen months after the birds' death (Greenwood 1979). Therefore the gape, which is made of the same material as the bill, of a live bird probably does not change or differ significantly from that of its skin. Three morphological variables that limit the size of prey that can be eaten are the horizontal and vertical opening of the beak as well as the length of the oesophagus. The width or depth of body of prey often limits the size of prey eaten before its length does, because a seabird sometimes swallows prey that is longer than the bird's oesophagus and so the prey's tail pokes out of the bird's mouth (Swennen and Duiven 1977; pers. obs.). Some coral-reef fishes, e.g. pomacentrids and acanthurids, have very deep bodies and are compressed laterally. A tern cannot swallow fish that are too deep-bodied even though the lengths of the fish are less than the length of the bird's oesophagus. The horizontal width rather than the vertical width of gape limits the 30 K. HULSMAN: SIZE OF PREY OF TERNS Figure 1. Method of measuring width of gape. size of prey that can be eaten because seabirds when swallowing orientate the prey so that its widest dimension (i.e, lateral width or depth of body depending on the species of prey) is aligned along the horizontal plane, not the vertical one. Therefore, the maximum width to which the gape can be stretched limits the size of prey that a predator that eats its prey whole can eat. The relaxed gape would be positively correlated with the maximum width to which it could be stretched, provided the elasticity of the surrounding tissue is similar from species to species. In fact, the relaxed gape is probably directly proportional to the maximum width to which it can be stretched. Thus the relaxed gape may be used as a relative measure of the maximum width of the gape. Data from live birds fitted the trends obtained from data from skins. I therefore combined measurements from live birds for two species with those from skins of EMU81 other species because this did not affect the result of the analysis. To reduce weight to a linear measure, as are the other four variables, I used the cube root of weight in the analysis. A linear model underlies Pearson's Correlation and a linear relation is more likely to occur between two like dimensions than between two different ones, e.g. linear and cubic. Partial correlation analysis was used to explore which of four morphological characters had the highest correlation with length of prey eaten by terns and the greatest effect in reducing correlations in higher order partial correlations. From these results one may infer which of the four variables potentially best predicts the size of prey eaten by predators such as terns that eat their prey whole. Partial correlations were used instead of multiple regression because the variables are highly correlated with each other (Table 11; Snedecor and Cochran 1967). RESULTS Width of gape (G) has the highest correlation with length of prey (P) of the four morphological characters used in the analysis (Table 11). There are indications that there is some relation between width of gape and length of prey. First, width of gape when held constant reduced first and second order partial correlations between length of prey and the other three variables to nonsignificant or negatively significant levels (Table HI). Secondly, the relation between width of gape and length of prey remained statistically significant in five out of six cases where the values of the other variables were held constant (Table 111). In the second order and third order partial correlations, the combined effect of length of bill (B) and cube root of weight ( ~ t % reduced ) the correlation between TABLE I Mean measurements (mm) of lengths of bill and wing, width of gape, weight (g) and length of prey of fourteen species of tern used in the analysis. Species Bill Gape Wing Weight Prey Blue-grey Noddy Procelsterna cerulea Black Noddy Anous minutus Black-naped Tern Sterna sumatrana White Tern tiygis alba Roseate Tern S. dougallii Arctic Tern S. paradisaea Bridled Tern S. anaethetus Sandwich Tern S. sandvicensis Common Noddy A. stolidus Sooty Tern S. fuscata Common Tern S. hirundo Lesser Crested Tern S. bengalensis Crested Tern S. bergii Caspian Tern Hydroprogne caspia 1. Ashmole and Ashmole 1967; 2. Serventy et al. 1971; 3. Witherby et al. in Serventy et al. 1971; 4. Lernmetyinen 1973; 5. Pearson 1968; 6. Veen 1977; 7. Brown 1975; 8. Koli and Soikkeli 1974. 1981 K. TABLE I1 Correlations between lengths ,of pre and weight, lengths of bill and wmg and wdth oPgape of terns. Prey Gape Bill Wing 31 HULSMAN: SIZE OF PREY OF TERNS Gape Bill Wing Wtg +0.971 +0.824 +0.892 +0.945 +0.946 +0.891 + 0.955 +0.987 +0.943 + 0.959 1% level of significance, 12 df = 0.661 length of prey and width of gape to non-significant levels. The cube root of weight and width of gape remained highly correlated even when lengths of prey, bill and wing (W) were held constant ( r w t % , G,P,B,W = +0.766, df 9, P<0.01). This can be seen in the first order partial correlation between ~ t and % P, which is reduced to almost zero, when G is held constant (Table 111). Therefore, the combined effect of cube root of weight and width of gape is not as great as their individual effects in some cases, e.g. correlations between length of wing and prey. DISCUSSION Sometimes spurious correlations are shown by partial correlations, e.g. length of bill increases with weight of body but the force exerted at the tip of the bill decreases as the length of the bill increases, unless the strength of the adductor muscles that close the bill is increased to compensate (Ashmole 1968). Therefore, there may be a negative correlation between lengths of bill and prey. This negative correlation was unmasked by second and third order partial correlations. Width of gape affects, to some extent, the size of prey that terns may eat as indicated by its reducing first and second order partial correlations between length of prey and the other three variables to non-significant or negatively significant levels. The maintenance of significant correlations between length of prey and width of gape when the values of other variables were held constant may indicate that there is some underlying functional relation between them. Width of gape may be positively correlated with the amount of adductor muscle (presumably its strength). In second and third order partial correlations the combined effect of length of bill and cube root of weight reduced the correlation between length of prey and width of gape to non-significant levels. This is feasible because, when the strength of the muscles operating the bill is no longer increased, a wider gape will not enable a bird to catch larger prey. A wide gape enables a bird to swallow large prey, not catch it. Weight seemed to have a slightly greater effect in reducing correlations than did length of wing. Weight will be correlated with length of prey because the weight of the adductor muscles forms part of the total weight of the bird. Length of wing is correlated with length of prey indirectly. Length of prey may be correlated with wing chord (wing area/wing span), which affects the amount of lift that can be generated by a wing. The larger the prey, the greater the lift that is needed for the bird to carry it. Length of wing is directly correlated with both area and span of the wing and these determine wing chord. Width of gape is but one of many variables that affects the size of the prey eaten by predators that eat their prey whole. Of the four characters tested, width of gape was potentially the best predictor of size of prey eaten by terns. Swennen and Duiven (1977) showed that the preferred size of prey eaten by alcids was about half the maximum width of the bird's gape. Kislalioglu and, Gibson (1976) showed that the optimal size of prey of the Fifteen-spined Stickleback Spinachia spinachia was about half the width of the largest prey that it could swallow. Here optimal size is defined as the size that yields the greatest amount of energy ingested per unit of handling time. Therefore, if the maximum width of the TABLE I11 First, Second and Third Order Partial Correlation Coefficients. The variable correlated with length of prey is given under the heading Correlates and the variable(s) which has (have) been held constant is (are) given under the heading Constant. Constant Correlates G:P Constant Correlates B:P Constant Correlates W:P Constant (1) P<O.Ol; (2) P <O.O5 Length of bill (B); length of prey (P); width of gape (G); length of wing (W); and cube root of weight ( W t s ) . Correlates Wt % :P 32 K. HULSMAN: SIZE OF PREY OF TERNS gape is known, then the predator's preferred size of prey can be calculated. The relaxed gape is easily measured on live birds and therefore is a useful field measure that provides an estimate of their preferred size of prey. If the relaxed gape (G) is proportional to its maximum width (G,x), i.e. G,, = kG, then kG/2 would be the preferred width of prey for that particular individual. On the other hand, if the relaxed gape was half the maximum width that the gape could be stretched (i.e. k=2), then the relaxed gape would be the same width as that of the preferred width of prey. However, the relation between the relaxed width of gape and the preferred width of prey is still to be examined. Once the relation between relaxed gape and preferred width of prey is established, one may calculate the preferred widths of prey and so sizes of prey, for a sample of a population. This could be particularly useful in the study of coexisting species. Comparisons of the frequency-distributions of preferred widths of prey with those of the actual prey may provide information about which species were at a competitive disadvantage. Such comparisons would provide information about the availability of prey. In conclusion, the horizontal width of the gape may be used to predict the preferred width of prey eaten by predators that eat their prey whole. Because this is valuable for making predictions and the gape can be measured easily in the field, it ought to become a standard measure for all birds that eat their prey whole. The usefulness of the gape as a source of ecological data is not restricted to birds; it could be used for any animal that eats it prey whole. ACKNOWLEDGEMENTS I thank Mr D. P Vernon and Mr G. Ingram of the Queensland Museum and Dr G. F. Mees of the Museum of Natural History, Leiden, for allowing me access to their collections. My thanks to Drs L. Zwarts, who kindly analysed the data for me, and Profs J. Kikkawa and J. M. Cullen, Dr R. L. Kitching and Dr J. N. M. Smith EMU81 for their helpful suggestions on an early draft of the manuscript. I am grateful to Mr S. Parker for information about shrinkage of skins. I thank Ms L. Shinkarenko for her helpful comments on a later draft of the manuscript and for drawing Figure 1. Mrs K. Belolevu kindly typed the drafts of the manuscript. REFERENCES ASHMOLE, N. P . 1968. Body size, prey size and ecological segregation in five sympatric tropical terns (Aves: Laridae). Syst. Zool. 17:292-304. -, and M. J. ASHMOLE. 1967. Comparative feeding ecology of seabirds of a tropical oceanic-island. peabody Mus. nat. Hist. Bull. 24:l-132. BROWN, W. Y. 1975. Parental feeding of young Sooty Terns (Sternafuscata [L.]) and Brown Noddies (Anous stolidus [L.]) in Hawaii. J . Anim. Ecol. 44:731-748. GREENWOOD, J. G. 1979. Post-mortem shrinkage of Dunlin Calidris alpina skins. Bull. Br. Orn. Club. 99: 143-145. HESPENHEIDE, H. A. 1973. Ecological inferences from morphological data. Ann. Rev. Ecol. Syst. 4:213-229. HULSMAN, K. 1978. Feeding and breeding biology of six sympatric species of tern (Laridae) at the One Tree Island, Great Barrier Reef. Abstract of PhD thesis, University of Queensland, 1977. Aust. J. Ecol. 3:240-241. KISLALIOGLU, M., and R. M. GIBSON. 1976. Prey 'handling time' and its importance in food selection by the 15-spined stickleback Spinachia spinachia (L.). J. E q . mar. Biol. Ecol. 25: 151-158. KOLI, L., and M. SOIKKELI. 1974. Fish prey of breeding Caspian Terns in Finland. Ann. zool. Fenn. 11:304-308. LEMMETYINEN, R. 1973. Feeding ecology of Sterna paradisaea Pontopp. and Sterna hirundo L. in the archipelago of southwestern Finland. Ann. zool. Fenn. lO.5n7-525 - - .- - - -- . PEARSON, T. H . 1968. The feeding biology of sea-birds breeding on the Farne Islands, Northumberland. J. Anim. Ecol. 37:521-552. SERVENTY, D. L . , V. SERVENTY and J. WARHAM. 1971. The Handbook of Australian Sea-Birds. Sydney: Reed. SNEDECOR, G. W., and W. G . COCHRAN. 1967. Statistical Methods. 6th ed. Iowa: Iowa St. Univ. Press. SWENNEN, C., and R. DUIVEN. 1977. Size of food objects of three fish-eating seabird species: Uria aalge, Alca torda and Fratercula arctica (Aves, Alcidae). Netherlands J. Sea Res. 11:92-98. VEEN, J. 1977. Functional and causal aspects of nest distribution in colonies of Sandwich Tern (Sterna s. sandvicensis Lath.). Behaviour Suppl. 20.