A New Species of Large Duck (Aves: Anatidae) from the Miocene of New Zealand "2279

We describe a new species of extinct duck, Miotadorna catrionae sp. nov. (Anatidae, Tadornini, Tadorninae), based on a right humerus from the Miocene lacustrine deposits of St Bathans, Otago, New Zealand. Principal component analysis reveals that the new taxon is distinguished by its large size and relative proportions. This is the eighth and largest species of duck described from the St Bathans fossil assemblage and further underscores the global importance of this site for understanding anatid evolution.


Introduction
In the St (Saint) Bathans area of Central Otago, New Zealand, outcrops of the Bannockburn Formation, a late early-earliest middle Miocene lacustrine sequence, include a rich fossil biota from the ancient Lake Manuherikia [1][2][3]. The deposits have yielded the most diverse terrestrial vertebrate assemblage described from pre-Pleistocene Zealandia: numerous fish [4] and water birds, as well as rarer land birds, bats, lizards, frogs, crocodilians, a rhynchocephalian, a turtle and other mammal remains of uncertain affinity [2].
St Bathans is one of the richest localities for Miocene anatids worldwide, with seven species described to date [2]. Two geese may also have been present but remains thus far have been too fragmentary to allow their formal description [5]. Here, we describe a large anatid humerus and demonstrate that it represents a new species of large duck.

Material and Methods
Until now, most of the large anatid fossils from the St Bathans area were referred to the extinct shelduck Miotadorna sanctibathansi Worthy, Tennyson, Jones, McNamara and Douglas, 2007 [6], while a few bones were referred to undescribed forms of geese [5,6]. We tested the hypothesis that all the large anatid humeri from this fossil assemblage represent one species. This was conducted by assessing morphological variation, using both morphometrics and anatomy.
We measured all large anatid humeri (those about the size of M. sanctibathansi) from the St Bathans assemblage housed at the Museum of New Zealand Te Papa Tongarewa (Wellington, New Zealand; NMNZ) and one specimen at the Geology Museum, Department of Geology, University of Otago (Dunedin, New Zealand; OU) that preserve at least one of the following: total length, proximal width, proximal depth, shaft width, distal width and depth of dorsal condyle (Appendix A). We then compared these data to humeral measurements (taken at NMNZ) of the extant paradise shelduck (Tadorna variegata) to assess whether the St Bathans humeri fall into more than one size class.
All fossils were measured with callipers to the nearest 0.1 mm. We defined total length as the maximum length parallel to the main axis of the shaft; proximal, distal and shaft width as the maximum transverse widths perpendicular to the main axis of the shaft, with the proximal width measured to the dorsal tubercle (excluding the deltopectoral crest where the latter projected further dorsally) and the shaft width measured at the distal end of the deltopectoral crest; proximal depth as the maximum cranio-caudal diameter of the humeral head when viewed proximally; and distal depth as the maximum cranio-caudal diameter of the dorsal condyle when viewed distally.
To test for different humeral proportions between taxa, we performed a correlationbased principal component analysis in Past 4.08 [7] (Figure 1C,D). Only humeri with all 6 measurements were included (see Appendix A). Additionally, we used frequency histograms of proximal depth and distal width to visualise sexual dimorphism (bimodal distribution) or the lack thereof (unimodal distribution) for both the fossils and Tadorna variegata ( Figure 2).
Taxonomy 2022, 2, FOR PEER REVIEW 2 width and depth of dorsal condyle (Appendix A). We then compared these data to humeral measurements (taken at NMNZ) of the extant paradise shelduck (Tadorna variegata) to assess whether the St Bathans humeri fall into more than one size class. All fossils were measured with callipers to the nearest 0.1 mm. We defined total length as the maximum length parallel to the main axis of the shaft; proximal, distal and shaft width as the maximum transverse widths perpendicular to the main axis of the shaft, with the proximal width measured to the dorsal tubercle (excluding the deltopectoral crest where the latter projected further dorsally) and the shaft width measured at the distal end of the deltopectoral crest; proximal depth as the maximum cranio-caudal diameter of the humeral head when viewed proximally; and distal depth as the maximum craniocaudal diameter of the dorsal condyle when viewed distally.
To test for different humeral proportions between taxa, we performed a correlationbased principal component analysis in Past 4.08 [7] (Figure 1C,D). Only humeri with all 6 measurements were included (see Appendix A). Additionally, we used frequency histograms of proximal depth and distal width to visualise sexual dimorphism (bimodal distribution) or the lack thereof (unimodal distribution) for both the fossils and Tadorna variegata ( Figure 2).

Results
Our PCA revealed a pattern in which the vast majority of anatomical variation was captured by the first two principal components ( Figure 1C,D). PC1 explains most of the variance (82.9%) and correlates with an even increase in all measurements. PC2 explains only 5.6% of the variance and describes a relative shape change where total length, proximal depth and proximal width increased while shaft width, distal width and distal depth decreased ( Figure 1D). PC1 clearly separates male from female T. variegata and furthermore sets apart one of the St Bathans humeri (NMNZ S.47273) from all other St Bathans anatids. Below, we describe and compare this specimen in more detail and argue that it is sufficiently distinct to warrant the recognition of a new species, despite probable sexual dimorphism in Miotadorna (see Comments section).

Results
Our PCA revealed a pattern in which the vast majority of anatomical variation was captured by the first two principal components ( Figure 1C,D). PC1 explains most of the variance (82.9%) and correlates with an even increase in all measurements. PC2 explains only 5.6% of the variance and describes a relative shape change where total length, proximal depth and proximal width increased while shaft width, distal width and distal depth decreased ( Figure 1D). PC1 clearly separates male from female T. variegata and furthermore sets apart one of the St Bathans humeri (NMNZ S.47273) from all other St Bathans anatids. Below, we describe and compare this specimen in more detail and argue that it is sufficiently distinct to warrant the recognition of a new species, despite probable sexual dimorphism in Miotadorna (see Comments section). Suggested English vernacular name: Catriona's shelduck. Diagnosis: Shelduck characterized by a robust humerus (142.5 mm long) with a wide proximal and a narrow distal end. Differs from all other anatid humeri from St Bathans in its larger size, which is comparable to that of a small goose, such as the Egyptian goose (Alopochen aegyptiaca) (e.g., NMNZ OR.29470: humerus length 142.2 mm). It specifically differs from Miotadorna sanctibathansi-the largest previously described anatid from St Bathans-in having a longer, more tapered humerus with a wider proximal end but a similar-sized distal end ( Figure 2B, Appendix A). Resembles M. sanctibathansi in having an elongate shaft, a prominent capital shaft ridge directed towards the ventral side of a markedly elevated dorsal tubercle, a relatively narrow dorsal pneumotricipital fossa that does not undercut the humeral head, an elongate deltoid crest that extends well distal of the bicipital crest, and in having the attachment of the superficial pronator muscle located cranial of centre on the ventral face.

SYSTEMATICS Class Aves
Differs from anserines, which have also been reported at St Bathans [5], in lacking a prominent capital shaft ridge directed towards the humeral head and in having a capital shaft groove that undercuts the humeral head [11]; from Johnstones' duck Dunstanetta johnstoneorum Worthy, Tennyson, Jones, McNamara and Douglas, 2007 [6] in having a deeper brachial fossa that does not extend as close to the ventral margin of the bone and a shallower pit for the attachment of the superficial pronator muscle; from Enright's duck Matanas enrighti Worthy, Tennyson, Jones, McNamara and Douglas, 2007 [6] in having a narrower dorsal extension of the pneumotricipital fossa, a shaft that tapers distally, a larger and more excavated brachial fossa, a more cranially located pit for the attachment of the superficial pronator muscle on the ventral face, and a dorsal condyle that does not ventrally overlap the ventral condyle in cranial view [6]; and from Manuherikia spp. in having a narrower dorsal pneumotricipital fossa that does not undercut the humeral head, a broader medial ridge bounding the brachial fossa, the pit for the attachment of the superficial pronator muscle is more proximal relative to the ventral supracondylar tubercle, a distinct dorsal epicondyle and a dorsal condyle that is shorter distally than both the flexor process and the ventral condyle [2,6]. Use of 'medial ridge' (=shaft between brachial depression and ventral margin) follows Worthy et al. [6].
Description: The humerus is largely complete but distorted post-mortem (as indicated by the lack of any sign of healed fractures), with the proximal end twisted caudally relative to the shaft and the middle of the shaft displaced dorso-ventrally by about 10 mm. Its total length was calculated by adding together measurements of each fractured section (see Figure 1B). Note that some of this bone's distortion is in the cranio-caudal plane and, therefore, is not obvious in the figure. The head is prominent and excavated by a welldeveloped pneumotricipital fossa. The shaft is long and bears a prominent captial shaft ridge. The distal end is robust but notably narrower than the proximal end. The dorsal and ventral condyles are prominent, with the former being separated from the flexor process by a deep caudal depression.
Comments: Given that M. catrionae sp. nov. and M. sanctibathansi are phenotypically very similar, their differing sizes could plausibly reflect sexual dimorphism instead of two species. As T. variegata is characterized by having larger males [12], sexual dimorphism could also explain the size range observed in Miotadorna, as previously suggested [6]. Frequency histograms of proximal depth and distal width suggest bimodal (i.e., dimorphic) patterns in both the extant and extinct tadornines, however NMNZ S.47273 does not consistently follow this pattern (Figure 2). NMNZ S.47273 matches the larger, putatively male fossil subset in its distal width, but it is deeper than all fossils proximally. This change in relative position suggests that size differences related to sexual dimorphism alone are insufficient to explain its distinct anatomy. We consider that all the other fossil humeri we inspected (including three other humeri from Mata Creek Site 9) most likely belong to M. sanctibathansi.

Discussion
Miotadorna catrionae sp. nov. is the eighth anatid named from St Bathans, underscoring the global importance of this area for understanding the evolution of the family. Anatomically, M. catrionae sp. nov. closely resembles the previously described St Bathans shelduck M. sanctibathansi, to the extent that all the characters distinguishing the latter from other tadornines are also shared (where preserved) by M. catrionae sp. nov.: capital shaft ridge very well developed, separated from dorsal tubercle by a deep, distally-flaring groove; ventral pneumotricipital fossa relatively smaller, shallower, and more occluded internally by pneumatic bone; brachial fossa deepest in ventro-distal portion; shaft between brachial fossa and ventral margin rounded [13]. Only its larger size and proportional differences distinguish NMNZ S.47273 from the humeri of M. sanctibathansi, leading us to include both in the same genus.
The functional reasons for the distinctive shape of the humerus of Miotadorna catrionae sp. nov. remain to be investigated but a relatively narrow distal humeral end is found in a diverse group of Anatidae, including swans and Manuherikia ducks [6,11].
Our results show that Miotadorna catrionae sp. nov. co-occurs with a smaller form of tadornine at Mata Creek Site 9, that is probably M. sanctibathansi or possibly the female of the new species, though none of the smaller Mata Creek Site 9 humeri possess definitive features characteristic of the new species. Other, much smaller, anatids also occur in this layer: the minute Manuherikia duck Manuherikia minuta Worthy, Tennyson, Jones, McNamara and Douglas, 2007 [6] and Manuherikia primadividua Worthy, Scofield, Salisbury, Hand, De Pietri, Blokland and Archer [2]. A previous study suggested that two geese, approximately the size of M. catrionae sp. nov., also occurred within the wider St Bathans area [5]. No humeri have yet been identified that may belong to these species, thus we remain unable to resolve their status.

Acknowledgments:
We thank Tony and the late Jack Enright, as well as Euan and Ann Johnstone for access to their land. Thanks also to the many people who helped with the St Bathans excavations over the years, particularly Ewan Fordyce, Sam Tennyson, Alex Brown, Allison Miller, Kieren Mitchell, Holly Heiniger, Kerry Walton, Lachie Scarsbrook and Cailie Ward; and Trevor Worthy (Flinders University) and Paul Scofield (Canterbury Museum) who were part of our team for many years. Finally, we thank Al Mannering for preparation of the holotype and assistance in the field; Jean-Claude Stahl for photography; Murray Williams for supplying many specimens of Tadorna variegata to Te Papa; Luke Easton for statistical advice; and three referees for their helpful comments on the manuscript.

Conflicts of Interest:
The authors declare no conflict of interest.