Present address: Department of Zoology, Division of Wildlife Biology, AVC College, Mannampandal 609305, Tamilnadu, India.
Mussel valve discrimination and strategies used in valve discrimination by the Oystercatcher, Haematopus ostralegus
Article first published online: 19 JUN 2002
Volume 16, Issue 3, pages 339–345, June 2002
How to Cite
Nagarajan, R., Lea, S. E. G. and Goss-Custard, J. D. (2002), Mussel valve discrimination and strategies used in valve discrimination by the Oystercatcher, Haematopus ostralegus. Functional Ecology, 16: 339–345. doi: 10.1046/j.1365-2435.2002.00635.x
- Issue published online: 19 JUN 2002
- Article first published online: 19 JUN 2002
- Received 3 July 2001;revised 31 October 2001;accepted 15 November 2001
- shell thickness
1. In 72% of a sample of common mussels Mytilus edulis L. that had been attacked by ventrally hammering Oystercatchers, Haematopus ostralegus L., the mussel had been opened through the right valve rather than the left.
2. In a matched sample of unopened shells, 56·5% of mussels were ventrally thinner on the right. Mussel thickness differences were unrelated to overall mussel thickness, so this proportion should apply also to opened shells even though they are on average thinner than unopened shells.
3. Since Oystercatchers attack a higher proportion of mussels by the right valve than are actually thinner on the right, their preference cannot depend solely on detecting measurable differences in thickness.
4. The pattern of preference can best be explained if Oystercatchers detect and attack the thinner valve in the mussel when the thickness difference between the two valves is more than a threshold of 0·036 mm, and otherwise always attack the right valve. This strategy would save 15·5% of the blows the birds would make if they attacked shells without regard to differences in valve thickness.
5. The improvement in the overall net energy intake rate that could be achieved by such a valve thickness discrimination strategy was 3·6%.
Both foraging theory and practical conservation science depend crucially on measurements of the densities of prey available to animals in particular environments. However, accurate measurements of prey availability depend on a thorough understanding of the predator’s behaviour. Many individuals of an apparently abundant prey species may be inaccessible to predators because of various morphological and behavioural antipredator adaptations of the prey. Furthermore, predators use different cues and techniques to detect the most vulnerable prey, and thus improve their chances of survival. Without an understanding of these preferences, and the ecological and perceptual factors underlying them, we may gain an entirely false impression of the impact of environmental changes on effective prey availability and likely predator survival.
Oystercatchers, Haematopus ostralegus L., continuously make decisions on how and where to search for prey and which of the prey of differing value and accessibility that they encounter they should take. They are extreme specialists in their feeding behaviour (Sutherland et al. 1996). Wintering Oystercatchers on the Exe estuary, SW England, mainly feed on mussels, Mytilus edulis L., and open them by hammering a hole through the shell either on the ventral or dorsal side or by stabbing between the intact valves (Hulscher 1996). Oystercatchers are highly selective in the mussels that they attack. In particular, ventrally hammering Oystercatchers select ventrally thin mussels that support few barnacles and have brown-coloured and ventrally flat shells (Durell & Goss-Custard 1984; Nagarajan et al. 2002). However, there has been much less investigation of how selective Oystercatchers might be in the way they attack a mussel. It is known that the birds may preferentially attack the right valve of mussels (Sutherland & Ens 1987), but the reason for this preference has yet to be discovered. Hence in this paper we explore why Oystercatchers seem to prefer to open one side of the mussel. Since the obvious reason for a preference would be a difference in thickness between the two valves, our method involved thickness measurements on both valves. Clearly, if Oystercatchers are able to select the location or mode of attack for a particular mussel so as to exploit its individual weaknesses, assessments of effective prey availability will need to take that into account. It is in principle possible that the variations in resistance to attack between possible sites on an individual mussel could be comparable with the interindividual differences as they are usually measured.
Materials and methods
The exe estuary and mussel bed
The study area was the river Exe estuary in SW England. The estuary, the dynamics of its Oystercatcher population, and the location of mussel beds within it are presented in detail by Goss-Custard et al. (1982). For this study the data were collected from mussel bed 4, which is located on the western side of the estuary, between Starcross and Cockwood (50°37′ N, 03°27′ W). This bed supports a good population of ventrally hammering Oystercatchers, and is free from human disturbances and easy to access.
Mussel collection and characteristics
Once a fortnight, towards the end of the low tidal cycle, 50 fresh mussel shells opened by ventrally hammering Oystercatchers were collected. Most of the Oystercatchers arrive in September and remain until March, hence samples were collected from September 1996 to March 1997. Freshly opened mussel shells were identified from the fresh flesh remains inside the shell near the attachment of the adductor mussel (Durell & Goss-Custard 1984). For each opened mussel, an unopened mussel of the same length was collected from under nearby weed (Oystercatchers selected 85% mussels from under weed); these are referred to as ‘comparator mussels’. The following measurements were made on all mussels, using the procedure of Durell & Goss-Custard (1984); see also Fig. 1.
The length of the mussel from anterior tip (umbo) to posterior tip was measured using a vernier calliper accurate to 0·05 mm.
The shell thickness at the mid-point of the edge of the ventral surface was measured using a Mitutoyo digimatic micrometer accurate to 0·001 mm. Values expressed in the text are mean ± SE.
The shell thickness was measured at the centre of the scar where the adductor mussel attaches to the valve.
The valve that was attacked was noted as either right or left side. In the lateral view of a mussel, keeping the umbo on the right and facing the ventral side so the hinge will be at the back side, the lower valve is the ‘right valve’ and the upper is the ‘left valve’.
Effort required for opening
One measure of effort is the number of blows needed to break into the mussel (Meire 1996). To find the number of blows needed to break into mussel shells of various thickness, an artificial Oystercatcher bill was designed (Nagarajan 2000). An iron rod was shaped in accordance with the measurements given by Durell, Goss-Custard & Caldow (1993) for the bills of ventrally hammering Oystercatchers of the Exe estuary and was dropped in a controlled manner on to sample mussel shells; shell strength was measured by the number of blows, from a fixed height, required to break the shell. Between mussel size classes, it was necessary to vary the dropping height in order to maintain the number of blows required for breakage within a reasonable range. By varying dropping height within one size class, it was found that the number of blows required to break the shell was proportional to the square of dropping height. A constructed variable, equal to (dropping-height2 × blows required for breakage), was therefore used as a measure of the effort required to open a mussel. In total 366 mussels were collected and cracked across the season, equally distributed between small (34–36 mm), medium (44–46 mm) and large (54–56 mm) size classes.
Oystercatchers had opened 392 mussels (72%) on the right valve and the remaining 153 (28%) on the left valve; none was attacked on both valves. This preference was significantly different from chance (χ12 = 104·40; P < 0·0001), and did not vary according to mussel length or through the season.
Differences in thickness between valves
Figure 2 shows that the ventral thickness of right valves of comparator mussels’ valves were, on average, thinner than the left valves (it was not possible to make this comparison with opened mussels because the relevant part of the opened valve had been broken away). However, the mean difference was only 0·006 ± 0·0048 mm, which is not significantly different from zero (t = 1·34; P = 0·18; N = 545). Nagarajan (2000) repeated the test using a larger sample of mussels collected from the same area of the Exe estuary in a subsequent year, and the difference remained similar at 0·008 ± 0·0026 mm, but the difference from zero was highly significant (t = 3·34; P = 0·0008; N = 2265). No further analysis was carried out on this sample, however, since they were not collected under conditions that were as closely matched to the opened mussel sample as was the case for the original comparator sample.
Although the difference in shell thickness between valves was quite small on average across the whole population, the actual difference facing a bird attacking a particular mussel might be either larger or smaller. The comparator mussels were grouped into three groups, ventrally right-thinner (N = 308; 56·5%), left-thinner (N = 227; 41·7%) and both sides equal within the error of our measurements (N = 10; 1·8%). The difference in ventral thickness between the valves was 0·075 ± 0·0043 mm (range 0·001–0·512 mm) for the right-side thinner group and 0·086 ± 0·0057 mm (mean ± SE; range 0·001–0·630 mm) for the left-side thinner group (Fig. 2); it is not meaningful to ask whether these differences are significantly different from zero, since both groups of mussels were selected for non-zero differences. The extent of thickness asymmetry did not differ significantly between the two groups (t = 1·58, P = 0·11, df = 448).
Although it is not possible to compare thicknesses of the two valves of an opened mussel, it is possible to compare the thickness of the unopened valve between right-opened and left-opened mussels. The ventral thickness for the left valves of right-side opened mussels was 0·903 ± 0·008 mm (N = 389) and for the right valves of left-side opened mussels it was 0·883 ± 0·013 (N = 167). There was no difference in the ventral thickness (F1,554 = 1·57; P = 0·211) of the unopened valve between the right- and left-side opened mussels.
The effort required to break mussels of different thickness, using the artificial bill, was examined by a stepwise regression analysis. Mussel shells are structurally dynamic with thinning and thickening both occurring during winter (Nagarajan 2000). Hence the number of days elapsed since 1 August (DAY) was included in the analysis of effort, with linear, quadratic and cubic terms. The dependent variable was the logHeight2 × Blows measures described in the Materials and methods section, above; the independent variables selected by the stepwise procedure were mussel ventral thickness, length, width and DAY. The multiple regression equation is:
- logHeight2 × Blows = 4·16 + 0·416 width + 0·837Ventral left thickness + 0·022 length − 0·082DAY + 0·0005DAY2 − 0·0000009DAY3, eqn 1
N = 366; R2adj. = 67·6%; P < 0·0001.
Oystercatchers in the Exe estuary opened a significantly higher proportion of mussels on the right valve. This preference is consistent with previous data. For example, Oystercatchers preferentially attacked the right valve of Macoma (Hulscher 1982) and, also, four out of seven captive, mussel feeding Oystercatchers showed preference for the right (Sutherland & Ens 1987). This preference appears to be adaptive since our data show that the right valves of mussels do tend to be thinner than their left valves. Obviously, we can only demonstrate this difference, and investigate its quantitative extent, in comparator mussels, not opened mussels. The two samples are not identical, because Oystercatchers do not select mussels at random but choose those with thinner shells, and this raises the question of whether the comparator mussel sample is an adequate proxy for the opened sample in investigating valve asymmetry. Regression analysis of the mean of left and right ventral thickness in the comparator sample against shell length and direction of asymmetry (right-thicker or left-thicker) shows that mean thickness is unrelated to direction of asymmetry even with length taken into account (regression coefficient for direction of asymmetry 0·0046 ± 0·0163 mm, t = 0·28, P = 0·776, df = 532). Furthermore, although there is a difference in mean thickness of opened and comparator shells, their thickness distributions overlap substantially, as is evident from Fig. 2, and this holds even within length classes: for example, in the modal length class of 50–60 mm, the ventral left thicknesses of comparator mussels range from 0·62 to 1·81 mm, while those of right-opened mussels range from 0·53 to 1·40 mm, and similar results are found for other length classes and for right valves). We conclude that there is no reason to suppose that, in mussels of the length and thickness typically attacked by Oystercatchers, valve asymmetry in the opened and comparator mussels would differ in either distribution or extent.
We now consider a number of hypotheses concerning the mechanism that might account for the observed, adaptive, preference for attacking mussels by the right valve.
(1) Oystercatchers respond to the population difference in thickness between right and left valves
In its purest form, this hypothesis would imply that only right valves should be opened. However, if we relax the hypothesis a little, to allow for the possibility that only some Oystercatchers show a right preference, or that those that do only follow it on some occasions, it can explain any level of right preference. While we cannot dismiss this hypothesis conclusively, it does make one prediction that is not borne out by the data. The hypothesis predicts that the difference between the right valves of left-opened mussels and the left valves of right-opened mussels would be the same as in the population as a whole. In fact, as Fig. 2 shows, the difference is considerably greater, suggesting that some other selection process is going on.
Furthermore, there are two theoretical objections to this hypothesis. Presumably, the right preference would have to derive from the effort saved, on the average, by attacking right rather than left valves. But if birds only respond to the population difference, this saving of effort is very small. From the experiments with the artificial bill, and the regression equation derived from them, the effort saved by exploiting a given thickness difference can be calculated, and compared with the effort under the null hypothesis that Oystercatchers attack either valve at random. From the regression equation reported above, it can be calculated that on average an Oystercatcher pursuing a pure long-term learning strategy would save at most 1% of daily effort expenditure. It seems unlikely that this alone could drive the long-term learning required to form the preference. Secondly, as will shortly be shown, other strategies yield a much greater energy saving.
(2) Oystercatchers detect the thinner valves of individual mussels and attack it
Since a majority of the mussel population have thinner right than left valves, the Oystercatchers’ preference could in principle be explained if they detected the difference and attacked the thinner shell. Figure 2 shows that, for an Oystercatcher attacking a particular mussel in which the thickness of the left and right valves differed, the difference between the thickness of the left and right shell is much greater than is suggested by the comparison of population means. This hypothesis therefore requires much less sensitive thickness detection than the hypothesis that the Oystercatchers respond to the population difference. However, this hypothesis is not consistent with the quantitative data on preference. Among the comparator mussels, only 56·5% of mussels were thinner on the right valve, yet 72% of the mussels opened by the Oystercatchers had been attacked on the right valve. A 2 × 2 χ2 contingency test between the number of right- and left-side thinner comparator mussels (population) and number of mussels opened on right and left valves ( χ12 = 24·44; P < 0·001) confirmed that the Oystercatchers opened more mussels on the right side than would be expected if they simply attacked the thinner valve of each mussel. This result implies that Oystercatchers sometimes opened mussels on the right valve when this was thicker than the left valve. How might this be understood? One possibility is that the Oystercatcher population contains a mixture of birds, some of whom can discriminate between valves and some of whom cannot, and the non-discriminating birds have a right-opening preference based on the population valve thickness differences. Although some individual differences in thresholds are to be expected, this hypothesis is unsatisfactory, because it is effectively saying that Hypothesis 1 above is correct for some birds, and we have already seen that Hypothesis 1 is unsatisfactory. Accordingly, we look instead at the possibility that there is a mixture of strategies within an individual bird.
(3) Discrimination combined with long-term learning
The decision chart in Fig. 3 describes a simple model that incorporates a strategy for opening mussels that would lead to the observed discrepancy between the proportion of mussels that are thinner on the right and the larger proportion that are attacked on the right. The model makes three assumptions. The first is that there is some threshold, below which Oystercatchers cannot detect a thickness difference. Secondly, if an Oystercatcher is able to detect which valve is thinner when it attacks a particular mussel, it opens the mussel through that valve. The final assumption is that, when the difference between left- and right-side shell thickness is too small for the bird to detect, then it always attacks the mussel through the right valve (because of the population tendency for more mussels to be thinner on the right valve than on the left). That is, the Oystercatchers choose the right valve as the ‘best bet’ in doubtful situations. This mechanism would result in the proportion of mussels that is opened on the right side being higher than the proportion of right-side thinner mussels in the population.
According to the model, whenever an Oystercatcher opened a mussel on the left valve it had detected that the left valve was thinner and accordingly attacked that side. Suppose that when the mussel is actually thinner on the left valve, the Oystercatcher correctly detects this with probability ‘P’, which is unknown. Since about 42% of mussels were thinner on the left, the probability of opening on left side is 0·42 × P. From the available data, we know that this equals 0·28 (28% opened on left valve), so P = 0·28/0·42 = 66·7%. So, in total about two-thirds of the mussels that were thinner on the left were correctly detected as such and opened through the left valve. In the remaining 33% of attempts, the Oystercatchers wrongly attacked the right side, the difference between valves being so small that detecting a difference was apparently not possible.
So the maximum thickness difference among this wrongly attacked 33% of mussels should be close to the threshold in the difference in thickness that an Oystercatcher can just detect in a mussel. To find this value, the left valve thinner mussels (N = 227) were selected, and the distribution of their mussel ventral thickness difference was examined, to find the 33rd percentile of the distribution. This was found to be 0·036 mm. Hence according to the model, Oystercatchers could detect the thinner valve in a mussel if the difference in the ventral thickness between the two valves was more than 0·036 mm. It is expected that there would be some individual variation around this value, but this does not alter the principles of the model.
This model does partly rely on long-term learning to explain why the Oystercatchers would choose the right valve when they cannot detect the thickness difference in a particular mussel. But unlike the first hypothesis, it provides a plausible means by which such learning could take place. From the calculations above, we see that in two-thirds of mussels the birds could detect the thickness difference. From the population statistics (Fig. 4), we know that in 55% of these cases the birds’ perceptual discrimination would lead them to attack the right valve. Accordingly, a right-valve choice would be reinforced more often than a left-valve choice. This difference should be enough to set up a right-valve preference that would still operate when there was no perceptible valve difference.
Impact of strategies for valve discrimination
We now ask whether the birds would have saved biologically significant amounts of effort by adopting the proposed strategy for selecting which valve of a mussel to attack, of discriminating all thickness differences greater than 0·036 mm and otherwise always attacking on the right.
According to our observations, Oystercatchers open 72% of their mussels on the right valve, and according to Hypothesis 3, this would include all mussels except those whose left valves were discernably thinner than the right. The mean ventral thickness difference of the 392 mussels matching this specification in the comparator sample was −0·056 mm. The remaining 28% (153 mussels), which would be opened on the left valve, had a mean ventral thickness difference of 0·120 mm (Fig. 4). Using the multiple regression equation above (equation 1), it can be shown that, by adopting this strategy the Oystercatchers would save 15·5% of blows relative to opening either valve at random.
As remarked above, an Oystercatcher that simply had a long-term preference for attacking mussels from the right would save only 1% of daily effort. On the other hand, an Oystercatcher that discriminated exactly, without any threshold (an inconceivable ability in practice, but worth considering in theory), would save only 16%, barely more than what it would save by using the strategy outlined in Hypothesis 3. Finally, attacking the undiscriminated mussels at random causes no loss relative to the hypothesized strategy (in fact, in our sample, a slight gain, to 15·7%).
Thus by selecting the thinner valve where they could, Oystercatchers would save a useful proportion of blows, and thus of hammering time: Oystercatchers hammer mussels at a roughly constant rate, so under Hypothesis 3 the handling time saved should also equal 15·5%. Any saving in hammering time would decrease the overall handling time and therefore the overall foraging time required to find and consume a mussel, and therefore increase the food intake rate. For the whole winter, the overall proportion of handling time that was spent in hammering was 0·54 and the proportion of the time spent foraging (foraging time) that comprised handling time was 0·42 (R. Nagarajan unpublished data). Hence the proportion of foraging time consisting of hammering time was (0·54 × 0·42), or 0·23, and the improvement in the overall intake rate that would be achieved by the proposed valve thickness discrimination strategy was 0·155 × 0·23, or 3·6%.
Cues used to discriminate between valves
Our preferred Hypothesis 3 requires the Oystercatchers to have an extraordinarily fine capacity to discriminate mussel valve thickness difference, down to a threshold of 0·036 mm. Hypothesis 1 (discrimination on the basis of long-term leaning alone) would require an even finer discrimination, of a mean difference of 0·008 mm. How could Oystercatchers achieve these remarkable discriminations?
It is of course possible that the discrimination is achieved indirectly, through some more obvious cue such as colour or the presence of barnacles. However neither of these properties, nor any other that we have measured, shows any valve asymmetry in our samples. It is much more likely that, if Oystercatchers are able to detect the thinner valve of a single mussel, they do so by the same means that they use to select thinner shelled mussels. According to Hulscher (1996) the cues Oystercatchers probably use to select mussels are vision and touch.
In this study, we know that the Oystercatchers did not often use visual cues to select the thin-shelled mussels, because they selected 85% of mussels from under weed. However, Ens & Alting (1996) have shown that Oystercatchers were able to identify thin mussels before they started to hammer them. Hulscher (1996) suspected that the Oystercatchers were able to touch-locate prey in darkness, and in the daytime in the absence of visual cues, by means of the Herbst corpuscles in the bill tip. We observed in the field that the Oystercatchers pecked at the mussels before selecting them and tapped them a few times before starting to hammer them. Often the Oystercatchers placed the mussel firmly on the anvil, but then, after having tapped the mussel valve, moved from one end to the other end in an apparent attempt to change the valve they were hammering. This behaviour suggests that the Oystercatchers identify the weak point in the shell and then hammer there in order to break it. However detection by bill contact could involve either of two mechanisms: touch as proposed by Hulscher, or the sound made when the shell is tapped by the bill, as proposed by Meire (1996). The acoustic sensitivity required to discriminate the sounds of thin and thick shells is unlikely to be finer than that needed by, for example, the Barn Owl, Tyto alba, which catches mice in total darkness by detecting their highly pitched squeaks (Payne 1962), or the fruit-eating, cave-living Oilbird, Steatornis caripensis, which echolocates with sharp clicks between 15 and 20 milliseconds long over a broad frequency spectrum (Konishi & Knudsen 1979). In birds, acoustical information is processed primarily by auditory nuclei in the hindbrain (Carr 1992), and birds that rely on sound have an exceptional number of ganglionic cells in the medulla. For example, the Barn Owl has 47 600 cells whereas a diurnal carrion crow, Corvus sp., only has about 13 600 (Winter 1963). It is not yet known where the Oystercatcher and other waders fit on this continuum, and it would be straightforward to investigate this anatomically.
Benefits of thickness discrimination
Our calculations suggest that Oystercatchers increase their intake rate by 3·6% by discriminating between the valves of the mussel. At first sight, this may seem a very small increase in efficiency but other data already demonstrate that Oystercatchers show remarkable efficiency in discriminating tiny differences in the prey and its environment and exploiting them to increase intake rate. For example, Goss-Custard et al. (1995) showed that Oystercatchers are able to detect differences of only 3% in intake rate between alternative feeding areas. Also, Nagarajan et al. (2002) have reported that Oystercatchers showed a strong preference towards brown coloured morphs of mussels, and he argued that this is because of the lower water content of these mussels. By making this selection, the Oystercatchers increased their intake in a feeding bout by between 2% and 20%. Although Oystercatchers can be seen as strict specialists in several ways (Sutherland et al. 1996), by adopting a range of foraging techniques they learn to differentiate tiny differences in the prey and environments and adapt and adjust their behaviours to increase the intake rate efficiently.
R.N. gratefully acknowledges the Commonwealth Scholarships Commission, UK, and Ministry of Human Resources Development, Government of India for a PhD fellowship. We are grateful to Robin Gill and Arie van der Lugt for making the figures, Catriona Ryan, Ian Hocking and Avril Mewse for comments, Rachel Kirby for computer assistance, Britta Osthaus and Jacqueline Hill for some field assistance and David Taylor and Peter Goodes for making the artificial bill.
- 1992) Evolution of the central auditory system in reptiles and birds. Evolutionary Biology of Hearing (eds D. R.Webster, R. R.Fry & A. N.Popper), pp. 511–544. Springer-Verlag, New York. (
- 1984) Prey selection within a size-class of mussels, Mytilus edulis, by oystercatchers, Haematopus ostralegus. Animal Behaviour 32, 1197–1203. & (
- 1993) Sex-related differences in diet and feeding method in the oystercatcher Haematopus ostralegus. Journal of Animal Ecology 62, 205–215. , & (
- 1996) Prey selection of a captive oystercatcher Haematopus ostralegus hammering mussels Mytilus edulis from the ventral side. Ardea 84A, 215–220. & (
- 1982) Use of mussel bed Mytilus edulis beds by oystercatchers Haematopus ostralegus according to age and population size. Journal of Animal Ecology 51, 543–554. , , & (
- 1995) Deriving population parameters from individual variations in foraging behavior. I. Empirical game theory distribution model of oystercatchers, Haematopus ostralegus feeding on mussels Mytilus edulis. Journal of Animal Ecology 64, 265–276. , , , , (
- 1982) The oystercatcher Haematopus ostralegus as a predator of the bivalve Macoma balthica. the Dutch Wadden Sea. Ardea 70, 89–152. (
- 1996) Food and feeding behaviour. The Oystercatchers from Individuals to Population (ed. J. D.Goss-Custard), pp. 7–29. Oxford University Press, Oxford. (
- 1979) The oilbird: hearing and echolocation. Science 204, 425–427. & (
- 1996) Using optimal foraging theory to determine the density of mussels Mytilus edulis that can be harvested by hammering oystercatchers Haematopus ostralegus. Ardea 84A, 141–152. (
- 2000) The foraging behaviour of oystercatcher (Haematopus ostralegus) in relation to food depletion during winter on the river Exe estuary, England. PhD Thesis, University of Exeter, Exeter. (
- 2002) Oystercatchers use colour preference to achieve longer term optimality. Proceedings of the Royal Society of London, Series B 269, 523–528. , & (
- 1962) How the barn owl locates prey by hearing. Living Bird 1, 151–159. (
- 1987) The criteria determining the selection of mussels Mytilus edulis by oystercatchers, Haematopus ostralegus. Behaviour 103, 187–202. & (
- 1996) Specialization. The Oystercatcher from Individuals to Populations (ed. J. D.Goss-Custard), pp. 56–76. Oxford University Press, Oxford. , , & (
- 1963) Vergleichende qualitative und quantitative untersuchungen an der Hörbar von Vogeln. Zeitschrift für Morphologie und Oekologie der Tiere 52, 365–400. (