The gastrointestinal tract may be considered one of the major interfaces between an organism and its environment, mediating their interactions (Karasov 1990). Several interspecific studies on other bird groups have shown that the size of the gut is related to diet and that the morphology of the gastrointestinal tract often reflects the birds' feeding strategies (Ricklefs 1996; Battley and Piersma 2005; Caviedes-Vidal et al. 2007; Lavin and Karasov 2008). Moreover, a direct influence of feeding strategies on the structures, functionality and physiology of the digestive tract has been shown in other vertebrates such as mammals (e.g., Schieck and Millar 1985; Korn 1992; Lovegrove 2010), amphibians and reptiles (e.g., Stevens and Hume 1995; O'Grady et al. 2005), and fish (e.g., German and Horn 2006; Wagner et al. 2009). However, it has to be considered that biological structures are not only fine tuned to their functional demands by natural selection but are also influenced by phylogenetic history and biochemical and mechanical constraints (Raia et al. 2010).
Within parrots, we have shown here that nectarivory is associated with reduced extension of the glands in the lower part of the esophagus below the crop (Pars thoracica). These glands produce a mucous secretion which helps hard ingesta to glide through the glandular stomach (proventriculus) and reduces the risk of mechanical damage to the latter (Güntert 1981). It can be expected that parrot species eating exclusively soft or liquid food evolve reductions of these glands within the esophagus, and our data corroborate this. On the other hand, we could not find any indication that the nectarivorous parrots have a longer esophagus as was proposed for the lories by Güntert (1981).
Between the glandular stomach (proventriculus) and the gizzard (muscular stomach, ventriculus), there is an intermediate zone characterized by the absence of compound glands of the former and absence of the koilin layer of the latter (Ziswiler and Farner 1972). This intermediate zone has the function of a storage space, where the proteolytic enzyme pepsin from the proventriculus can react with ingesta (Güntert 1981). We found the nectarivorous parrots to have a longer intermediate zone compared to the remaining parrots, even though the trait values of one nectarivorous taxon, Brotogeris, were more similar to the non-nectarivorous species. The intermediate zone appears to play an important role in the digestion of pollen (Güntert 1981) and a prolonged intermediate zone may be an adaptation to optimize the extraction of amino acids from pollen grains. Pollen grains have a high protein content, with their interior (protoplast) consisting of diverse amino acids (van Tets and Hulbert 1999; Gartrell and Jones 2001). Acidifications of pollen grains in the proventriculus may be important so that their contents can be extruded and digested, whereas mechanical break-up of pollen grains in the gizzard does not seem to be important (Gartrell and Jones 2001). The amount of energy extracted from meals can be enhanced by increasing the retention time (McWhorter et al. 2009). A prolongation of the intermediate zone can thus increase the rate of protein digestion as it increases the retention time of pollen grains, which seems to be the case in nectarivorous parrots. Pollen ingestion may require less energy than feeding on insects as an additional amino acid supply, because nectarivorous birds will encounter pollen while feeding on nectar (Nicolson and Fleming 2003).
Besides being a storage organ, the gizzard functions as an organ of mechanical digestion, the site of preliminary acid proteolytic digestion, and a filter for indigestible material (Ziswiler and Farner 1972). During contraction of the gizzard, the thick muscles close up, narrowing the lumen to a thin cleft and forcing the contents into two pouches (cranial and caudal sac, cf. Fig. 2) that lie under the thin muscles (McLelland 1979). Species feeding on a soft diet do not need the grinding function to break down their food, and can be expected to evolve reduced gizzard musculature (Steinbacher 1934; McLelland 1979). Indeed, the nectarivorous parrots differed from the remaining parrots by having less developed gizzard muscles. In contrast, a simple allometric relationship best explained the width of the whole gizzard and its lumen as well as its width at the caudoventral thin muscle and the maximum lumen at the thin muscles. The thin muscles act antagonistically to the main muscles and have no grinding function. Therefore, they are not expected to be developed more strongly in species relying on the grinding function of the gizzard. In congruence with our results, Richardson and Wooller (1990) also found two species of lories (Glossopsitta porphyrocephala, Trichoglossus haematodus) to have smaller and less muscular gizzards than four nonnectarivorous parrot species of Australia (Melopsittacus undulatus, Barnardius zonarius, Neopsephotus bourkii, Platycercus icterotis). Interestingly, we found that the reportedly mainly frugivorous Pesquet's parrot Psittrichas fulgidus (Collar 1997) shared a similarly reduced muscularity with the nectarivorous parrots. On the other hand, the blue-crowned hanging-parrot Loriculus galgulus did not show an overall reduced muscularity. This may correspond to the higher amount of seeds in its diet compared to L. philippensis (Homberger 1980). Similarly, gizzard measurements of Lathamus discolor and Brotogeris jugularis clustered with the non-nectarivorous parrots, and these two species were not found to have overall reduced gizzard muscularity either. Other studies also found Lathamus discolor to have retained the muscular gizzard of a granivorous species (Güntert and Ziswiler 1972; Gartrell et al. 2000). This may allow this species to feed on harder food when nectar and pollen are rare (Gartrell et al. 2000). Gizzard dimensions also vary in other passerines according to diet, with longer gizzards in seed- than in fruit- and insect-eaters and thicker muscular and glandular layers in insect- compared to fruit- and seed-eaters (Ricklefs 1996). Smaller gizzards with a reduced muscularity were also found in the nectarivorous honeyeaters (Meliphagidae) compared to similar-sized passerines (Richardson and Wooller 1986); however, phylogenetic nonindependence was not controlled for in that study.
Chemical digestion of food principally takes place in the intestine (Ziswiler and Farner 1972). Richardson and Wooller (1990) found two species of lories (Glossopsitta porphyrocephala, Trichoglossus haematodus) to have shorter intestines compared with four non-nectarivorous parrot species of Australia (Melopsittacus undulatus, Barnardius zonarius, Neopsephotus bourkii, Platycercus icterotis). This was explained as a consequence of sugars in nectar needing less processing in the intestine than other food. In contrast, we could not find any indication for shorter intestines of nectarivorous parrots. In a broad comparative study of birds, Lavin et al. (2008) did not find any significant effect of diet on small intestine length either (Lavin et al. 2008). However, this result is only partly comparable with ours, since we measured the whole lengths of the intestine owing to the difficulty of clearly distinguishing the small and the large intestine in parrots due to lack of caeca (Ziswiler and Farner 1972; Güntert 1981). In addition to lengths, intestine function depends inter alia on volume, surface area, villi, and microvilli area as well as enzymatic activity (Ricklefs 1996; Lavin et al. 2008; McWhorter et al. 2009). Moreover, the efficiency of digestion in the intestine may be influenced by the passive absorption of hydrosoluble compounds through the paracellular pathway. This is prominent in birds and may be especially important for nectarivores because they have to deal with large amounts of sugar in their diet (Karasov and Cork 1994; Napier et al. 2008; McWhorter et al. 2009). In general, birds have a lower nominal surface area of the intestine and a shorter small intestine as well as shorter digestive retention times than mammals; however, their higher passive absorption compared to mammals may compensate for this (McWhorter et al. 2009). This may render predictions about the intestine dimensions in relation to diet more difficult.
In conclusion, our analyses showed that nectarivorous parrots differ, after correction for phylogenetic nonindependence, from the remaining parrots in several traits of the digestive tract. Hence, we uncovered significant phenotype–environment correlations for the prolongation of the intermediate zone, the reduction of gizzard muscularity and the reduction of glands in the esophagus. The similarity in these trait features among some of the different nectarivorous groups is an indication of parallel evolution under the same or similar environmental conditions, that is, the shift to a nectarivorous diet, and implies that natural selection was the main driving force (cf. Losos et al. 1998; Schluter et al. 2004; Colosimo et al. 2005). Moreover, functional considerations suggest that the adaptations in the intermediate zone of nectarivorous parrots (probably except Brotogeris) allow them to rely effectively on nectar as a food source, and thus implying evidence for trait utility (Schluter 2000).
Phenotypic flexibility of the gastrointestinal tract
The size, structure, and functional characteristics of the gastrointestinal tract of birds can reversibly change within the lifetime of a bird (phenotypic flexibility, sensu Piersma and Drent (2003)) as a fast adaptive response to current functionality demands caused by environmental changes or circannual endogenous control (Starck 1999a,b; Piersma and Drent 2003; Starck and Rahmaan 2003; Battley and Piersma 2005; McWhorter et al. 2009). As pointed out by Lavin et al. (2008), comparative studies like ours have the limitation that species were not analyzed under common-garden conditions and thus it is not possible to assess to what extent the variation found among species is influenced by phenotypic flexibility and plasticity at the individual level. However, the individuals analyzed in this study all stem from captivity, where more stable conditions than in nature can be expected, thus mirroring a common-garden experiment. Furthermore, the inclusion of several individuals for most species and the wide range of body sizes among species considered certainly minimized the effect of intraspecific variation. There is additionally some evidence that phenotypic flexibility of the gastrointestinal tract is limited in parrots (cf. Güntert 1981). All individuals of Lathamus for example analyzed in this study were fed with a nectar-alternative, but they retained the partly muscular gizzard similar to that of a granivorous species, and the features of their gastrointestinal tract did not appear to differ from wild specimens analyzed by Gartrell et al. (2000). Nevertheless, further studies, preferably on wild birds, are needed to document the interplay between natural selection, plasticity and potential flexibility in features of the digestive tract in different parrot species.
Evolutionary trait shifts and species proliferation
We found that in the lories, the diet shift to nectarivory was associated with a significant shift in morphological evolution, chiefly of several gizzard traits, at the base of their radiation, implying a trait shift or evolutionary jump.
The lories have diversified into an exceptionally species-rich clade (Schweizer et al. 2011) and their diet shift might thus be considered to be an evolutionary key innovation. Nectar may have provided a spatially widespread underutilized niche and this may have allowed the lories to expand their ranges and to colonize even remote oceanic islands, which may have fostered allopatric speciation. Even today, congeneric species of the lories generally do not overlap geographically (Collar 1997). Sympatry within genera is found in eastern Australia, New Guinea and Wallacea, regions with a complex and composite environmental and geological history with several potential vicariance opportunities in the past (Hall 2002; Esselstyn et al. 2009; Byrne et al. 2011; Deiner et al. 2011). However, this ecological expansion was not followed by further significant ecological specializations within the radiation of the lories. Similar to honeyeaters, the other highly nectarivorous and species-rich bird-group of Australasia (Newton 2003), lories are generalized flower visitors and their ecological relationships with plants are not as specialized as those of hummingbirds or sunbirds (Fleming and Muchhala 2008). Avian pollinator assemblages differ regionally and the evolutionary specializations between nectar-feeding birds and their food-plants are strongest in the Neotropics, decreasing through Africa and South Asia to Southeast Asia and Australasia (Fleming and Muchhala 2008). The co-evolution of specialized plant-pollinator relationships may take time. While the hummingbirds split from their closest relatives in the Eocene (about 50 Ma) or even earlier (Brown et al. 2008; Pratt et al. 2009) with a major radiation after 20 Ma (Jetz et al. 2012), the lories split from their common ancestor with Melopsittacus only in the middle Miocene (about 15 Ma, Schweizer et al. 2011). Hummingbirds certainly had more time to co-evolve with plants than did lories. However, the similarly specialized sunbirds are likely to be younger than hummingbirds, and the evolutionary diversification of the generalist honeyeaters started at a similar age (Eocene) (Barker et al. 2004). Thus, explanations other than time may account for the low ecological specialization of Australasian nectarivorous birds compared with sunbirds and hummingbirds. Specific interactions between plants and pollinators are only likely to evolve when floral resources are spatially and temporally predictable (e.g., Waser et al. 1996). While this seems to be the case in the Neotropics, flowering of eucalypts in Australia varies in space and time, and trees in lowland and montane Papua New Guinea commonly have non-annual flowering patterns (Fleming and Muchhala 2008). Birds feeding on them may hence not be able to afford to specialize. This may account for the low specialization in plant–pollinator relationships of Australasian nectarivorous birds and may explain the lack of evolution of plant-specific specializations among the lories after their shift to a nectarivorous diet.
In conclusion, the key innovation of the lories allowed an expansion into a new adaptive zone and we hypothesize that the subsequent species proliferation may have essentially been nonadaptive through allopatric speciation. The lories may thus be considered an example of a nonadaptive radiation (Rundell and Price 2009). It is possible that the ecological opportunity provided by their key innovation did not trigger an adaptive radiation because of the unpredictable nature of the new resource. The key innovation nevertheless promoted significant lineage diversification through allopatric partitioning of the same broad new niche. Although other parrot groups switched to a nectarivorous diet, this did not increase their diversification rates and species richness compared to other parrots (Schweizer et al. 2011). Various factors may have inhibited an increased rate of cladogenesis in Lathamus, Brotogeris, and Loriculus following their change to a nectarivorous diet. Such factors can include developmental or genetic constraints, but also ecological circumstances like interspecific competition or the lack of opportunities for allopatric speciation. Hence, an evolutionary innovation does not necessarily lead to increased diversification (Vermeij 2001; Price et al. 2010). The question as to which factors hampered increased species proliferation in other nectarivorous parrots will be an interesting avenue for future research.