Phylogenetic relationships of the paraphyletic ‘caprimulgiform’ birds (nightjars and allies)

Introduction

Phylogenetic analyses of both morphological and molecular data have congruently shown that the traditional avian taxon ‘Caprimulgiformes’ (nightjars and allies) is not monophyletic. A sister group relationship between the ‘caprimulgiform’ Aegothelidae (owlet-nightjars) and Apodiformes (swifts and hummingbirds) is well-established by multiple lines of independent evidence. This hypothesis was first proposed from an analysis of 25 morphological characters (Mayr 2002), and is supported by all subsequent molecular studies in which these taxa were included (Mayr et al. 2003; Cracraft et al. 2004; Barrowclough et al. 2006; Ericson et al. 2006; Hackett et al. 2008; Pratt et al. 2009).

In addition to the Aegothelidae, the ‘Caprimulgiformes’ include four other taxa of crepuscular or nocturnal birds, i.e. the Steatornithidae (oilbird), Podargidae (frogmouths), Caprimulgidae (nightjars) and Nyctibiidae (potoos). Mayr (2002) established a clade including Caprimulgidae, Nyctibiidae, Aegothelidae and apodiform birds, and considered the relationships of Podargidae and Steatornithidae uncertain. Subsequent analyses of morphological data supported a sister group relationship between Steatornithidae and the clade including Caprimulgidae, Nyctibiidae, Aegothelidae and apodiform birds, but did not resolve the position of the Podargidae (Mayr 2005a,b). All ‘caprimulgiform’ and apodiform birds share an elongated crus longum of the os carpi ulnare (Fig. 1), but this feature is also present in some other avian taxa (e.g. trogons, Trogonidae), and no unambiguous morphological apomorphies of a clade including these birds have as yet been identified. Although a clade including all ‘caprimulgiform’ and apodiform taxa was obtained in an analysis of 2954 morphological characters (Livezey and Zusi 2007), none of the three putative apomorphies (Livezey and Zusi 2007: tab. 2) is present in all of its representatives.¹

Meanwhile, however, there exists congruent molecular support for a clade including Apodiformes and all taxa of the paraphyletic ‘Caprimulgiformes’, which resulted from analyses of multiple individual gene loci (Ericson et al. 2006; Mayr 2008; Hackett et al. 2008; Fig. 2a,b), and received high bootstrap support in Hackett et al.'s (2008) comprehensive ‘phylogenomic’ study (Fig. 2c). Equal molecular support does not exist for alternative phylogenies, and the fact that this clade was obtained in independent analyses of different kind of molecular data constitutes strong evidence that the included taxa indeed constitute a monophyletic group (see also Mayr 2008).

The relationships between Steatornithidae, Podargidae, Caprimulgidae and Nyctibiidae, however, remain controversial, and the results of the above studies and other molecular analyses (Mariaux and Braun 1996; Cracraft et al. 2004; Barrowclough et al. 2006; Harshman 2007; Larsen et al. 2007; Brown et al. 2008) do not agree concerning their position. Whereas, for example, Cracraft et al. (2004) assumed that Steatornithidae, Nyctibiidae and Caprimulgidae form a clade, Livezey and Zusi (2007) hypothesized the traditional ‘Caprimulgiformes’ to be monophyletic, and the analysis of Hackett et al. (2008) resulted in a sister group relationship between Steatornithidae and Nyctibiidae, with Podargidae and Caprimulgidae being successive sister taxa of the Aegothelidae/Apodiformes clade (Fig. 2).

Despite a superficially similar external appearance, ‘caprimulgiform’ birds distinctly differ in anatomical features, and the aim of the present study is to show that morphological data contribute to a resolution of the relationships between these taxa. Sixty-nine morphological characters are included in the analysis, and the implications of the resulting phylogeny for the evolution of dark-activity are discussed.

Material and Methods

Skeletons of the following species were examined (all in the collection of Forschungsinstitut Senckenberg): Steatornithidae: Steatornis caripensis; Podargidae: Batrachostomus septimus, Podargus strigoides; Aegothelidae: Aegotheles cristatus; Nyctibiidae: Nyctibius griseus; Caprimulgidae: Caprimulgus europaeus, Caprimulgus pectoralis, Chordeiles minor, Eurostopodus mystacalis, Hydropsalis torquata (‘brasiliana’), Lurocalis semitorquatus, Macrodipteryx longipennis (skull), Nyctidromus albicollis, Phalaenoptilus nuttalii; Apodidae: Apus apus, Apus melba, Chaetura vauxi, Collocalia salangana, Collocalia vanikorensis, Cypseloides senex, Streptoprocne zonaris; Hemiprocnidae: Hemiprocne comata; Trochilidae: Amazilia versicolor, Archilochus colubris, Chrysolampis mosqitus, Glaucis hirsuta, Melanotrochilus fuscus, Phaethornis pretrei, Thalurania furcata, Trochilus polytmus. Osteological terminology follows Baumel and Witmer (1993), if not indicated otherwise.

The phylogenetic analysis was performed with the heuristic search modus of nona 2.0 (Goloboff 1993) through the winclada 1.00.08 interface (Nixon 2002), using the commands hold 10 000, hold/10, mult*1000 and max*. Sixty-nine characters were scored for eight ingroup taxa, based on the revised and emended matrix of Mayr (2005a) (see Appendix 2). Outgroup comparisons were made with Tinamidae (tinamous) and Anseriformes (waterfowl). The consistency index (CI) and retention index (RI) were calculated. Bootstrap support values were calculated with 1000 replicates, three searches holding one tree per replicate, and TBR branch swapping without max*.

Results of phylogenetic analysis

Analysis of the character matrix in Appendix 2 resulted in a single most parsimonious tree, which is depicted in Fig. 3. The following characters were optimized as apomorphies of a clade including Podargidae, Caprimulgidae, Nyctibiidae, Aegothelidae and apodiform birds to the exclusion of the Steatornithidae:

Two homoplastic characters were further shown to be apomorphies of this clade. One of these (tarsometatarsus without canalis interosseus distalis; character 55) occurs in many neornithine birds and is absent in the Nyctibiidae. The other (vinculum between tendons of musculus flexor perforans et perforatus digiti III and m. perforatus digiti III absent; character 64) is variable in the Caprimulgidae (see Appendix 1).

A clade including Caprimulgidae, Nyctibiidae, Aegothelidae and apodiform birds is supported by six apomorphies:

Characters (1)–(3) were already mentioned as apomorphies of this clade by Mayr (2002), who listed two further apomorphies. Both were included in the present analysis, but were not unambiguously optimized as apomorphies of any of the nodes (palatinum with well-developed processus rostromedialis; character 6 [Fig. 4], and fossa dorsalis of phalanx proximalis digiti majoris divided into two depressions by distinctly raised oblique bulge; character 43 [Fig. 6]). Characters (5) and (6) are newly identified in the present study.

A sister group relationship between Caprimulgidae and Nyctibiidae is supported by seven synapomorphies:

Characters (1)–(5) were discussed by Mayr (2002), characters (6) and (7) are newly identified as apomorphies of this clade in the present study.

The sister group relationship between Aegothelidae and Apodiformes is supported by seven synapomorphies:

Characters (1), (4), (5) and (7) were identified as apomorphies of the Aegothelidae/Apodiformes clade by Mayr (2002), who discussed two further apomorphies of this clade. One of these (os palatinum with strongly protruding angulus caudolateralis; character 7 in Appendix 2) is optimized as an apomorphy of a more inclusive clade in the present study (see beginning of this chapter). The polarity of the other character (absence of processus basipterygoidei; character 10 in Appendix 2) is uncertain. Steatornithidae, Caprimulgidae and Nyctibiidae exhibit well-developed processus basipterygoidei, whereas these processes are absent in Podargidae and the Aegothelidae/Apodiformes clade. It is equally parsimonious to assume that they were reduced in the latter taxa, or that they were regained in Steatornithidae and the clade including Caprimulgidae and Nyctibiidae. Processus basipterygoidei are found in palaeognathous birds and Galloanseres, i.e. all non-neoavian Neornithes, and are part of the groundplan of Neornithes, to which ‘Caprimulgiformes’ and Apodiformes belong. In order to show whether they also belong to the groundplan of the ‘Caprimulgiformes’/Apodiformes clade, it would be necessary to either identify the sister taxon of this clade, or to examine whether or not these processes occur in the early ontogeny of those taxa, in which they are absent in the adult representatives.

Classification

Based on the results of the phylogenetic analysis, the following classification is proposed:

Strisores Baird, 1858.

Sangster (2005) introduced the term ‘Daedalornithes’ for the clade including Aegothelidae and Apodiformes, considering the name Apodimorphae inappropriate, because Sibley et al. (1988) restricted it to apodiform birds. However, taxon names above the ‘family-level’ are not regulated by the International Code of Zoological Nomenclature, and I prefer the latter term, which more appropriately reflects the actual content of this clade.

Worthy et al. (2007: 25) stated that Aegothelidae were first elevated into ‘ordinal’ rank by Simonetta (1967). However, this assumption is erroneous, as Simonetta (1967) classified owlet-nightjars within Caprimulgiformes and just assigned them to a new ‘suborder’ Aegothelae, which was listed in his linear sequence between Podargi (including Podargidae) and Caprimulgi (including Caprimulgidae and Nyctibiidae). The term Aegotheliformes was thus actually introduced by Worthy et al. (2007).

Discussion

Except for the position of the Nyctibiidae, the results of the present analysis are in concordance with the phylogeny of Hackett et al. (2008). Potoos resulted as the sister taxon of the Steatornithidae in the Hackett et al. (2008) analysis, but neither the Steatornithidae/Nyctibiidae clade nor the clade including Caprimulgidae, Aegothelidae and apodiform birds received significant bootstrap support (Fig. 2c). By contrast, a sister group relationship between Nyctibiidae and Caprimulgidae is strongly supported in the present study, both in terms of bootstrap robustness and the number of shared apomorphies. Likewise well-corroborated is a clade including Nyctibiidae, Caprimulgidae, Aegothelidae and apodiform birds, and a sister group relationship between Steatornithidae and Nyctibiidae would imply homoplasy of at least 13 morphological features, some of which are unparalleled within Aves (see chapter ‘Results of phylogenetic analysis’).

A sister group relationship between Steatornithidae and Nyctibiidae is not recovered in other analyses of sequence data (Mariaux and Braun 1996; Barrowclough et al. 2006; Ericson et al. 2006; Brown et al. 2008). Although a Steatornithidae/Nyctibiidae clade resulted from the DNA-DNA hybridization studies of Sibley and Ahlquist (1990), these have been criticized for methodological reasons (Houde 1987; Lanyon 1992; Harshman 1994), and Nyctibiidae were not included in the analyses of ‘caprimulgiform’ birds using the Fitch algorithm (Sibley and Ahlquist 1990: fig. 334), which according to Harshman (1994) should be preferred to the trees based on the UPGMA algorithm.

Except for the greatly abbreviated tarsometatarsus and the possibly functionally correlated absence of an ossified pons supratendineus on the tibiotarsus (Fig. 8), Nyctibiidae and Steatornithidae do not share derived morphological features that indicate a sister group relationship, and a clade consisting of Nyctibiidae and Caprimulgidae is certainly much better supported by morphological data. Because molecular evidence for a Steatornithidae/Nyctibiidae clade is weak, a sister group relationship between Nyctibiidae and Caprimulgidae is more likely to reflect the true phylogenetic affinities of these birds. The present study may thus serve as an example where morphological features contribute to phylogenetic issues that are not conclusively resolved by molecular data.

In contrast to all other recent studies, Livezey and Zusi’s (2007) analysis resulted in monophyly of the traditional ‘Caprimulgiformes’, with Aegothelidae as the sister group of the other taxa, and Nyctibiidae as sister taxon of a clade including Steatornithidae and Podargidae. Apomorphies that support ‘caprimulgiform’ relationships were not listed, and as I have detailed elsewhere (Mayr 2008), the results of the analysis are called into question by disputable character scorings, which in several instances are based on unverified assumptions.

Steatornithidae, Caprimulgidae, Nyctibiidae, Podargidae and Aegothelidae are crepuscular or nocturnal birds, whereas all Apodiformes are diurnal like most other birds. Irrespective of the exact affinities of the Nyctibiidae, the most parsimonious interpretation of the distribution of dark-activity in Strisores is a single origin in the stem lineage of these birds, and a reversal to a diurnal way of living in the stem lineage of the Apodiformes (Fig. 9a). Such an evolutionary scenario was first proposed by Mayr (2002), and has also been assumed by Cracraft et al. (2004) and Hackett et al. (2008). So far, however, no anatomical or physiological features were identified that indicate a nocturnal stem species of apodiform birds. Steatornithidae and Caprimulgidae further show distinct differences in retina morphology, with oilbirds having much more rods and lacking a tapetum lucidum (Rojas et al. 2004; to the best of my knowledge, the retina of other ‘caprimulgiform’ birds has not been studied so far). Fidler et al. (2004) also reported differences between Aegothelidae, Podargidae and Caprimulgi in the nucleotide sequences of the arylalkylamine N-acetyltransferase gene, which regulates melatonin biosynthesis and is assumed to play a role in nocturnal activity (Steatornithidae and Apodiformes were not included in this study). Although these differences per se do not prove an independent origin of dark-activity in ‘caprimulgiform’ birds, it is not straightforward to explain their existence under the assumption that the stem species of the Strisores was crepuscular or nocturnal. None of the extant ‘caprimulgiform’ taxa includes secondarily diurnal representatives, and Holyoak (2001: 12) hypothesized that the ‘successful return to diurnal foraging may be prevented by combinations of factors, such as loss of colour vision, susceptibility to hawks and other diurnal predators, and potential competitive interactions with the many diurnal insectivorous bird species’. According to Holyoak (2001: 10) there are further ‘huge evolutionary incentives for insectivorous birds to become nocturnal’, such as the great abundance of insects and the fewer number of predators. Despite being a less parsimonious scenario, a fourfold origin of dark-activity in the stem lineages of the Steatornithidae, Podargidae, Caprimulgi and Aegothelidae (Fig. 9b) is thus not implausible, and remains an alternative to be seriously considered in future studies. Possible reasons for a multiple independent origin of dark-activity may have been competition with diurnal insectivorous birds, or access to a new ecological zone due to an early Paleogene radiation of nocturnal insects (see Mayr 2005c, 2009).

Many ‘caprimulgiform’ and apodiform birds are capable of torpor, i.e. facultative hypothermic responses to environmental conditions, which has been reported for Podargidae, Caprimulgidae, Aegothelidae, Apodidae and Trochilidae (Bartholomew 1985; Brigham et al. 2001; Körtner et al. 2001; McKechnie and Lovegrove 2002; Lane et al. 2004). The distribution of this trait in the extant taxa indicates that it was already present in the stem species of the Podargocypseli. Kitto and Wilson (1966) hinted at the possibility that occurrence of torpor in ‘caprimulgiform’ and apodiform birds may be related to an unusually slow motility of the enzyme S-malate dehydrogenase in the studied representatives of these taxa. By reducing oxaloacetate to malate, this enzyme indeed plays a role in gluconeogenesis after the hibernation of mammals (Epperson et al. 2004). A correlation between torpor and the properties of the S-malate dehydrogenase may also be indicated by the fact that its motility is particularly slow (only 27% of that of the chicken enzyme) in the Common Poorwill, Phalaenoptilus nuttalii, which is the only bird known to be capable of extended periods of torpor over several weeks (e.g. Bartholomew 1985). However, Kitto and Wilson (1966), showed that the S-malate dehydrogenase is also unusually slow in Charadriiformes, which are not known to be capable of torpor, and torpor regulation is likely to be a multifactorial process that is dependent on more than a single enzyme. P. nuttalii was the only representative of the ‘Caprimulgiformes’ included in the study of Kitto and Wilson (1966), and studies of the motility of the S-malate dehydrogenase in other ‘caprimulgiform’ birds are warranted.