The kinematics of feeding and drinking in palaeognathous birds in relation to cranial morphology

Since Merrem (1813)separated the Palaeognathae from all other birds, the taxon Palaeognathae has caused many disputes among ornithologists. One of the most important characters that separates the Palaeognathae from all other modern birds(Neognathae) is the dromaeognathous (= palaeognathous) palate, first described by Huxley (1867). It is not only the `palate' that is different from that found in Neognathae, but a whole complex of morphological characters, which includes structures such as the pterygoid, quadrate and vomer (see McDowell, 1948; Bock, 1963; Gussekloo and Zweers, 1999). The character set of the jaw mechanism that discriminates the Palaeognathae from the Neognathae will be referred to as the palaeognathous Pterygoid–Palate Complex (palaeognathous PPC; Gussekloo and Zweers, 1999). The mechanical function of this PPC in neognathous birds is well known. The PPC participates in the movement of the upper bill(Bock, 1964). Upper bill movement is induced by rostrad rotation of the quadrate, which pushes both the lateral jugal bars and the medial pterygoid–palate bar forward. Each bar transfers its forces and movement onto the premaxilla. The forward movement of the premaxilla results in an upward rotation of the upper bill around a hinge,either in the nasal–frontal area (prokinesis) or in the rostral part of the bill (rhynchokinesis), depending on the position of a flexible zone. In the palaeognathous birds it is assumed that a large flexible zone is present in the centre of the upper bill (Fig. 1, Zusi, 1984). The pterygoid–palate bar and the quadrate are of great importance for cranial kinesis since these bars transfer the forces, and the muscles for the movement of the upper bill attach to these elements.

Although many authors have used the palaeognathous PPC for systematic purposes (Fürbringer,1888; Gadow, 1892; Beddard, 1898; McDowell, 1948; de Beer, 1956; Bock, 1963), studies on the function of the system are very limited. Most previous analyses assumed that the special morphology of the PPC in Palaeognathae is related to rhynchokinesis (Hofer, 1954; Simonetta, 1960; Bock, 1963), mainly because of the osteology of the PPC, the flexibility of the dorsal and ventral bars of the upper bill, and the incomplete ossification of the lateral bar(Zusi, 1984). The movement pattern of the PPC during bill opening has been measured, and showed very little difference between Palaeognathae and Neognathae(Gussekloo et al., 2001). These studies, however, were done on either osteological specimens or head preparations and it is currently unknown if movement of the PPC and rhynchokinesis actually occurs in living palaeognathous birds.

In the present study we did a functional analysis of the feeding behaviour of the greater rhea Rhea americana (L.) in order to elucidate the function and origin of the palaeognathous PPC. Several hypotheses can be postulated about the evolution of the special PPC morphology in the Palaeognathae. Our main hypothesis is that the specific palaeognathous morphology of the PPC is an adaptation to selective forces that act on the PPC in palaeognathous birds, but not in neognathous birds. Since the function of the PPC is the transfer of forces and movements during upper bill movement, it is assumed that these selective forces must be related to bill movement. Feeding behaviour is considered the strongest selection force acting on bill movement, and therefore on the PPC. Other behaviours such as vocalisation,preening and social behaviour are considered to have little effect on the osteology of the bills. To investigate whether differences in selection forces on bill movement are present, bill movement of a typical palaeognathous bird during feeding will be described and compared with a previously described general neognathous-feeding pattern(Zweers et al., 1994). If differences are found in the feeding behaviour it may be possible to infer which selective forces resulted in the differences in PPC morphology. If no differences in feeding pattern can be found between Neognathae and Palaeognathae it must be concluded that no different selective forces act on the PPC during feeding and an alternative hypothesis about the origin of the difference in morphology between Neognathae and Palaeognathae must be postulated.

The greater rhea Rhea americana (L.), a middle-sized palaeognathous bird from South-America, was chosen as representative for the Palaeognathae. This species has a general palaeognathous PPC configuration(McDowell, 1948) and its natural history and behaviour are well known (Raikow, 1968, 1969; Bruning, 1974; Martella et al., 1995, 1996; Reboreda and Fernandez, 1997). For the analysis two animals, one male and one female, were trained to feed on several food types within the experimental set-up. For this analysis, feeding will include only the behavioural elements from picking-up the food item until swallowing. All phases prior to the picking-up for intra-oral transport are considered a part of food-acquisition.

The feeding behaviour of the birds was recorded using video imaging (25 frames s^–1). The recordings were made in an experimental set-up in which a lateral view and a frontal view of the bird were obtained in the same frame using a mirror situated in front of the bird at an angle of 45°. The birds had to approach the feeding arena through a small corridor,ensuring a good lateral position of the bird with respect to the camera. Behind the bird, from the camera's viewpoint, a grid (2 cm×2 cm squares)was placed to make scaling possible. The films were analysed, frame-by-frame,by digitising the position of several points on the upper and lower bill relative to the standard grid (Fig. 2). Prior to the feeding analysis the position of the bending zones was determined through manipulating osteological specimens. The positions found were compared to previous descriptions(Hofer, 1954; Simonetta, 1960; Bock, 1963; Zusi, 1984) and used to determine the position of points for digitising. In addition to these points on the bills, some reference points on the skull of the bird were also digitised (Fig. 2). From the complete set of digitised points a number of distances and angles was calculated (Table 1). The accuracy of the calculated distances and angles was determined on the basis of the variation in a standard measurement calculated as the distance between two digitised points of the reference grid. The standard error in this distance measurement was approximately 0.08 mm. The standard error in digitising a point was therefore approximately 0.04 mm in each direction. The errors for points were used to calculate the error for angles, which was dependent on the distance between the points and the angle between lines. The standard error of the mean angle of two parallel lines both of 4 cm (a typical length used in our analyses) is just under 0.5°. For each time point the mean value and 95% confidence intervals of the behavioural parameters were calculated and used to describe the mean behavioural pattern. The same data were also used to test if cranial kinesis is present in palaeognathous birds.

The data on head displacement were used to determine maximum velocities and accelerations of the head during feeding. The complete trajectory of the head was determined by interpolation to 250 points s^–1 using a cubic-spline interpolation technique. The spline interpolation technique was used under the assumption that head movements follow a gradual and symmetric path around the points of change of direction. Behavioural observations confirm these assumptions. The acceleration data, in combination with the mass of the head (estimated from the head mass in other individuals) were used to determine the forces acting on the head.

A range of food types was offered (Table 2), varying in size between 4 mm and 35 mm in length. At least five items of each food type were analysed for each bird. Large apples were only eaten by the male and only on three occasions. Drinking cycles were observed in both individuals, but only seven cycles could be analysed.

To investigate the diversity and variability of the feeding behaviour, a Principal Component Analysis (PCA) was used to describe the variation in feeding behaviour due to different food-types. The PCA, with Varimax rotation,was based on the correlation matrix of characters. The characters were obtained from the movement patterns of the different head elements involved in feeding (Table 3; see also Figs 5, 6). Differences in principal component scores were determined using an analysis of variance (ANOVA).

In addition to an analysis of the structure of the general feeding and drinking pattern, emphasis was laid on the presence of cranial kinesis, since many authors have coupled the morphology of the palaeognathous PPC directly to it. During all the observed feeding cycles both movement near the nasal–frontal area and movement in the upper bill were monitored to determine if cranial kinesis is present during normal feeding behaviour. Changes in angles were compared statistically using ANOVA to determine if bending actually occurs.

When describing the feeding behaviour of the rhea, the same elements are used as in the general description for neognathous feeding used by Zweers et al. (1994). The first elements of the neognathous feeding behaviour are preliminary head fixation,preliminary head approach and final head fixation. None of these elements are observed in the rhea. After final fixation neognathous feeding behaviour continues with the following elements: (1) final head approach, (2) catch at jaw tips, (3) stationing and repositioning, (4) catch at jaw tips, (5)intra-oral transport (`Catch and Throw'), (6) intra-pharyngeal transport.

The general feeding sequence of the rhea (Figs 3A, 4) resembles this pattern: the bird approaches the food item while opening the bills and the food item is picked up. When the head hits the ground the acceleration of the head is approximately 11.30 m s^–2 (a=11.30±6.57 m s^–2, N=41). With an estimated head mass of 0.25 kg,the mean calculated impact force was 2.83 N and the maximum did not exceed 7.54 N (a_max=30.17 m s^–2).

Grasping the food is sometimes followed by repositioning behaviour. Repositioning occurs in the rhea more often when large rather than small food items were eaten.

When the food item was correctly positioned, a single `Catch and Throw'movement is used to transport the food particle into, or near to, the entrance of the oesophagus. A `Catch and Throw' movement starts when the food is fixed between the bills, the head is accelerated upward and slightly backward. Then the bills open and the head is suddenly moved forward. The accelerated food item continues to move upward while the head of the bird moves downward, which results in the transport of the food item. The palaeognathous single `Catch and Throw' movement is accompanied by a large gape and a large depression of the tongue. This depression results in an enlargement of the buccal cavity,which facilitates transport of the food item into the caudal part of the oropharynx. No tongue movement was observed other than the one resulting in the depression of the mouth floor. The limited functions of the tongue in feeding are reflected in the morphology of the tongue, which is relatively small with no remarkable features (Fig. 5). The oropharynx itself also shows very little remarkable characters that might indicate feeding patterns other than `Catch and Throw'behaviour.

The rhea uses two different types of drinking behaviour, depending on the area of water available to drink from. The preferred method of drinking can be described as scoop drinking followed by a low-amplitude tip-up phase(Fig. 3B). During drinking the bird opens the bill, inserts it into the water, and with a forward scooping motion of the head the lower bill is filled with water. The bill is then closed and the head is elevated until the neck is almost completely stretched,while the head itself is in a horizontal position. Finally, the water is transported into the oesophagus by a slight elevation of the bill tips and a retraction of the tongue. In some cases small horizontal `Catch and Throw'movements are used to transport the water more caudally in the oropharynx just prior to swallowing.

When the size of the water surface limits the scooping movement, the rhea uses a drinking technique that is very similar to pecking behaviour. The bill is opened and inserted almost vertically into the water, the bill is then closed and in a single head jerk the water is accelerated vertically, the bill is opened and the water is transported to the back of the oropharynx. Since this behaviour strongly resembles pecking, and is not the basic drinking behaviour, it was not included in this analysis.

To characterise the movement patterns quantitatively 36 parameters were chosen (Table 3, Figs 4, 6) and analysed using a Principal Component Analysis (PCA). The first three principal components of the PCA (PC1–3), based on the characters of the feeding and drinking behaviours described 63% of the total variance. An analysis of variance(ANOVA) over the principal component scores was used to determine the main differences between individuals/sexes and food types. None of the first three principal components showed a difference between individuals/sexes (d.f.=47,PC1: F=0.264, P=NS; PC2: F=0.198, P=NS;PC3: F=0.240, P=NS, where NS=not significant) and therefore the data from both individuals were combined. It is clear from the plot of the first principal component (PC1) against the second principal component (PC2)that drinking behaviour is remarkably different from feeding behaviour(Fig. 7, Table 4). The first principal component describes the absence of the second gape movement (`Catch and Throw'movement), differences in neck movement (duration of the neck cycle) and the duration of the total feeding cycle (Table 5). The second principal component describes differences in food manipulation by the bills, such as position of the food item between the bills, depression of the lower bill and upper bill kinesis. To investigate the differences between food types without the large distorting effect of drinking, the PCA was repeated using the four types of feeding behaviour only. In this analysis 65% of the variance was explained by the first three principal components. To test whether there are significant differences between food types, a one-way ANOVA over the first three principal components scores was used. Differences between the food types were tested using a t-test with Bonferroni correction. There are significant differences between food types on both the first and third principal component (d.f.=40,PC1: F=28.678, P<0.001; PC2: F=0.365, P=NS; PC3: F=3.628, P<0.05). It is clear that PC1 describes the effect of food size (Fig. 8). The differences on the first principal component represent mainly the effect of the duration of the movement for each food type (e.g. gape period, head elevation period, food period, lower bill period), the size of the first gape (Gape 1) and the elevation of the head (difference head Y; Table 6). All these parameters increase with an increase of the size of the food type, which indicates that the movement pattern of food uptake is relatively constant and that only the duration, mainly the effect of repositioning, and amplitude of the movement differ between different food sizes. The change in movement described by PC1 becomes smaller when the size of the food items increases. A difference on PC1 is only found between seeds and all other food types (food type 1 vs 2, 3 and 4, t-test, Bonferroni correction, P<0.001).

The third principal component mainly describes the handling of the food item, which affects the amount of depression of the lower bill (lower bill at Gape 1), position of the food item between the bills during the upward movement of the head (food level, food level s.d.), and amplitude of cranial kinesis (e.g. prokinesis at Gape 1 and 2, rhynchokinesis at Gape 2; Table 6). Differences between food types on PC3 are only found between large apples and seeds (food type 1 vs 4, t-test, Bonferroni correction, P<0.05). However, no clear trends can be determined with a change in size of the food types.

To test the presence of kinesis in the skull of the rhea, several measurements were taken. The movement between the cranium and the upper bill around the point where the nasal–frontal hinge would be in prokinetic birds was measured, and will be further referred to as prokinetic movement(Fig. 1A). A second measure of kinesis was the movement between the rostral and caudal part of the upper bill with the border of the two parts in the bending region of the upper bill. Movement of the rostral part relative to the caudal part of the upper bill will be referred to as the rhynchokinetic movement(Fig. 1B). Since food types are different in size, the kinesis of the upper bill was determined for each food type separately. The large apple was not used for this analysis due to the small number of repeated measurements.

It is assumed that maximal kinesis is observed during the large amplitude gapes when the food item is picked up or swallowed. Velocities of the head are very large during the second phase of the `Catch and Throw' movement, which makes it very difficult to determine points accurately. Because of this low accuracy and the relatively small movements in the upper bill, cranial kinesis could only be analysed accurately in the grasping phase. From repeated experiments the average pick-up cycle was calculated and plotted with the standard error. The plot of gape vs time shows a clear pattern(Fig. 9A) similar to a single food uptake cycle, and differences between the time segments are significant(ANOVA, small apple: d.f.=85, P<0.001; pellets: d.f.=90, P<0.001; seeds: d.f.=168, P<0.001; water: P<0.001).

A similar analysis was made for the lower bill movement, expressed as the depression angle (Fig. 9D). For all food types the same pattern was found, and only for the largest food type analysed (small apple) were the differences between successive time segments not significant, due to large variation (ANOVA, small apple: d.f.=86, P=NS; pellets: d.f.=89, P<0.05; seeds: d.f.=168, P<0.05; water: d.f.=100, P<0.05).

Angular measurements were also used to test the response of prokinetic and rhynchokinetic movement during the feeding cycle. Prokinetic movement showed a pattern similar to the lower bill movement but with a much smaller amplitude(Fig. 9B). However, the prokinetic movement pattern is not significant for any food type or drinking(ANOVA; small apple, d.f.=87, P=NS; pellets: d.f.=89, P=NS,seeds: d.f.=167, P=NS; water: d.f.=102, P=NS). The rhynchokinetic movement patterns can be clearly recognised except in the drinking behaviour (Fig. 9C),but are only significant for the two largest food types, small apple and pellets (ANOVA, small apple: d.f.=84, P<0.05; pellets: d.f.=88, P<0.05; seeds: d.f.=166, P=NS; water: d.f.=98, P=NS). In Table 7 the maximal changes in the mean angles of the different types of kinesis are given, showing an increase in cranial kinesis with an increase in food size.

Our study showed clearly that the feeding behaviour of the palaeognathous birds strongly resembles the feeding patterns of neognathous birds. Some differences are found in the approach phase, the intra-oral transport and intra-pharyngeal transport. In general the tongue plays a more important role in neognathous feeding behaviour than in palaeognathous feeding behaviour. The complete intra-oropharyngeal transport phase in palaeognathous birds is achieved by a single `Catch and Throw' movement, which is in large contrast with the complicated `Slide and Glue' mechanism and complex transport of food through the oropharynx often used by neognathous birds. However, for large food items neognathous birds may also use a single `Catch and Throw' movement,but still show complex intra-pharyngeal transport(Zweers et al., 1994). The difference between the single `Catch and Throw' movement of neognathous and palaeognathous birds is that the former use this movement to transport the food item onto the lingual base, while the latter use it to transport the food item to the area caudal to the lingual base. Tomlinson(2000) showed that in palaeognathous feeding behaviour the final transport of the food item into the oesophagus is achieved by a single retraction of the tongue and larynx. The protraction/depression of the tongue during the `Catch and Throw' movement and the retraction during swallowing are the main functions of the tongue during feeding.

Similarly, the drinking behaviour of the rhea lacks the tongue movement present in neognathous drinking. The general drinking pattern of neognathous birds (Zweers, 1992) consists of (1) a fixation phase, in which the bird orientates its head, (2) the downstroke, in which the head is lowered towards the water, (3) the immersion phase, during which the actual water intake takes place, and (4) the upstroke,in which the head is positioned in such a way that gravitational forces facilitate transport of the water from the oropharynx into the oesophagus(swallowing). All these phases are also represented in palaeognathous drinking behaviour and similar to neognathous phases. Large differences, however, are found in the immersion phase. In the rhea there is no stationary immersion phase but a scooping motion, the bill remains widely opened, and no tongue movement is observed during this phase. Intra-oral transport during immersion in Neognathes always includes pro- and retraction of the tongue. Only during the head upstroke does a single protraction of the tongue in rhea facilitate the movement of the water into the oesophagus.

Although feeding and drinking behaviours were analysed under controlled conditions, field data show that the observed feeding behaviours are present in the natural behaviour of the rhea as well. The natural feeding and food-acquisition behaviour of all Palaeognathae, except the kiwi(Apteryx sp.), can be described as browsing, which means eating a wide variety of plant material with some occasional carnivorous food. The food preferences of the greater rhea in the wild(Martella et al., 1996)suggest that no fundamentally different feeding behaviours are required, other than the ones analysed in our study. The diet consists of a wide variety of food items, but is mainly vegetarian(Mosa, 1993; Martella et al., 1996; Quin, 1996). The assumption that the feeding behaviour is characteristic for all Palaeognathae is confimed by a number of observations. The single `Catch and Throw' feeding behaviour and both the scooping and `Catch and Throw' drinking behaviour have been observed in the greater rhea in the wild. We also observed the single `Catch and Throw' feeding behaviour in wild and captive ostriches Struthio camelus L., captive emus Dromaius novaehollandiae Latham and captive cassowaries Casuarius casuarius (L.). Although there are some differences between the diets of the various palaeognathous species, these seem due to local food availability, and not to preference or performance.

In order to determine the importance of cranial kinesis in the feeding behaviour of the palaeognathous species, we determined bending both in the area of the nasal–frontal articulation and in the upper bill itself. Our study showed that during feeding behaviour kinesis is found between the rostral and caudal part of the upper bill in rhea. No, or only very limited,bending occurs in the area of the nasal–frontal articulation, the position where in many neognathous species the naso–frontal hinge is situated. The elevation amplitude of the bill tip relative to the cranium in the rhea is similar to the elevation of the upper bill found in prokinetic neognathous birds (approximately 5–10°; Kooloos and Zweers, 1989; Heidweiller and Zweers, 1990; van den Heuvel, 1992).

One hypothesis about the role of rhynchokinesis states that it reinforces the grip on food items by simultaneously depressing the upper bill tip and elevating the lower bill tip, as found in certain Charadriiformes(Zusi, 1984). No upper bill depression is observed in the rhea, which indicates that rhynchokenesis is not used in this way in Palaeognathae.

Our video recordings of beak movement suggested that there may be a difference between neognathous rhynchokinesis and paleognathous rhynchokinesis. While a clear bending point is present in Neognathae, the upper beak in rhea seems flexible over its full length. The elevation angle of the upper bill gradually declines more caudally in the upper bill. This strongly suggests that a single hinge or narrow bending zone is not present in Palaeognathae. This conforms with the description of Zusi(1984), who named this flexibility over the full-length `central' rhynchokinesis. The relation between rhynchokinesis and the detailed anatomy of the beak is explored in an accompanying paper (Gussekloo and Bout,2005).

To determine whether the feeding behaviour of the Palaeognathae is derived or primitive within modern birds, a comparison can be made with the general feeding patterns found in other tetrapods. The method of feeding in tetrapods depends on the presence of a well-developed lingual apparatus. If a well-developed lingual apparatus is absent two main types of non-lingual feeding are present within the tetrapods: inertial feeding and the feeding pattern observed in snakes (de Vree and Gans, 1994). Comparison of the feeding behaviour of the rhea with the nearest living sister group of birds, the crocodilians, shows that the feeding behaviour of the rhea is more similar to reptilian inertial feeding than the general feeding pattern of neognathous birds(Zweers et al., 1994; Cleuren and de Vree, 1992). In crocodilian intra-oral transport the tongue elevates the food item until it presses against the palate. Then gape is rapidly increased and the cranium moved forward (the avian `Catch and Throw'), while the tongue is depressed to enlarge the buccal cavity and to facilitate the transport of the food item. In the rhea the final transport of a food item into the oesophagus is achieved by a retraction of the hyolingual apparatus(Tomlinson, 2000), similar to transport in crocodilians. The fact that feeding behaviour of the rhea resembles feeding behaviour of crocodilians, and lacks certain elements found in the general feeding pattern of neognathous birds, seems to suggest that inertial feeding behaviour is basal within birds. This would agree with the widely accepted hypothesis that the Palaeognathae are the oldest offshoot in the phylogeny of modern birds (Bock,1963; Meise, 1963; Parkes and Clark, 1966; Cracraft, 1974; de Boer, 1980; Prager and Wilson, 1980; Sibley and Ahlquist, 1981; McGowan, 1984; Feduccia, 1985; Handford and Mares, 1985;Elzanowski, 1986; Houde, 1986; Bledsoe, 1988; Caspers et al., 1994; Lee et al., 1997; van Tuinen et al., 1998; Simon et al., 2004). However,lingual feeding is found in the more primitive amphibians(de Vree and Gans, 1994), and present in many reptilians (Bramble and Wake, 1985; Reilly and Lauder,1990; Herrel et al.,1996). As crocodiles are a very distant sister group of birds the possibility remains that crocodiles and Paleaognathae are specialized inertial feeders and that lingual feeding is the most primitive avian feeding mechanism. Tetrapod inertial feeding is believed to have evolved many times independently within vertebrates (de Vree and Gans, 1994). The simple movement patterns of the tongue in rhea may be the consequence of a reduction in size related to efficient `Catch and Throw' feeding behaviour. Since it cannot unambiguously be determined whether the feeding pattern of the palaeognathous birds is primitive or a specific adaptation, these data cannot be used to determine the phylogenetic position of the Palaeognathae.

From the comparison of feeding patterns we conclude that the feeding and drinking behaviours of the rhea resemble those of neognaths, but lack certain elements found in the general feeding pattern of neognathous birds, especially with respect to tongue movements. We found no elements in the feeding behaviour that might impose additional functional demands on the PPC, nor are any of the behavioural elements investigated more demanding than in neognathous feeding. This indicates that the specific morphology of the PPC is not the result of specific functional demands from palaeognathous feeding behaviour. Also the hypothesised role of rhynchokinesis in relation to the cranial morphology could not be confirmed. Central rhynchokinesis is present in the upper bill, but does not play an important role in improving grip on the food item or in increasing the gape. It must therefore be concluded that the kinetic feature of the bill is not the factor that determined the morphology of the PPC.

Alternative explanations for the presence of the characteristic PPC complex have been suggested. Bock(1963) proposed that the special morphology of the PPC might be an adaptation to the high impact forces on the bill during pecking. Our movement analysis showed that the rhea is capable of controlling the impact force of pecking. The head hits the ground at approximately 11.30 m s^–2. The mean calculated impact force was 2.83 N and the maximum did not exceed 7.54 N. Using a compressive strength of 170×10⁶ Pa for bone we can calculate that a cross-sectional area of the bones of just 0.05 mm² is sufficient to withstand these forces. It seems clear that this area is many times smaller than the actual cross-sectional area in the skull of the rhea.

Another explanation is that the morphology of the palaeognathous PPC is the result of selection forces that are not directly related to feeding, but do affect the morphology of the PPC indirectly. While Paleognathae are often believed to be basal to Neognathae, an alternative hypothesis states that Palaeognathae have actually a derived phylogenetic position and have evolved through neoteny from a flying ancestor (de Beer, 1956). The hypothesis on the neotenous origin of the Palaeognathae was recently revived by physiological/ontogenetic data(Dawson et al., 1994) and molecular systematics (Mindell et al.,1997; Härlid and Arnason,1999). The physiological/ontogenetic experiments showed that induced neoteny in neognathous birds results in a morphology of the PPC that was similar to that of the Palaeognathae, while the molecular systematic data show a derived position of the Palaeognathae within the Neognathae and not a basal position of the group. A comparison of the cranial morphology of the Palaeognathae with different developmental stages of neognatous birds showed,however, that not a single developmental stage resembles the palaeognathous configuration (Gussekloo and Bout,2002). This is a clear indication that the palaeognathous PPC is not the result of neoteny.

A third hypothesis on the origin of the special morphology of the palaeognathous PPC proposes that the morphology of the extant palaeognathous PPC is the result of the continuous reduction of bony and ligamentous elements in the lateral aspect of the skull(Gussekloo and Zweers, 1999). Although birds in general have less bony and ligamentous elements in the lateral aspect of the skull than closely related groups such as dinosaurs and other reptiles, Palaeognathae have even less than most birds. Compared to Neognathae, Palaeognathae lack a clear Ligamentum postorbitale and the lateral bar of the upper bill (Bock,1964; Zusi, 1984). The reduction of these elements might have resulted in a relatively unstable configuration of the upper bill, especially under conditions of external loading. The only type of food acquisition that is not covered by our study is pulling leaves off plants. It is possible that this way of feeding imposes special functional demands on the construction of the upper beak. The removal of leaves is mainly achieved by neck motion, and generates external forces on the upper bill. While preliminary observations showed that the transport of the food items used in the present study is very similar to the transport of grass or leaves that are removed from the plant, it is possible that the morphology of the upper bill of the Palaeognathae is adapted to oppose the reaction forces during pulling (see also Gussekloo and Bout, 2005). Although it seems that such a force regime is not unique for Paleognathae(e.g. grazing geese, pulling off berries by passerines), it is unique when combined with the absence of lateral bars in the skull. Since the Palaeognathae could not counteract the external forces by reinforcing these lateral elements, it may have been necessary to reinforce the unstable upper bill configuration in the ventral elements, resulting in a more rigid PPC. As a consequence of this reinforcement of the ventral plane of the upper bill,active kinesis of the upper bill may also become limited. Additional experiments to test this hypothesis are described in the accompanying paper(Gussekloo and Bout,2005).