Tigrisomatinae

Scofield, R. Paul, Worthy, Trevor H. & Tennyson, Alan J. D., 2010, A Heron (Aves: Ardeidae) from the Early Miocene St Bathans Fauna of Southern New Zealand, Records of the Australian Museum 62 (1), pp. 89-104 : 96-99

Subfamily Tigrisomatinae (Tiger herons)

The tarsometatarsi of tiger herons differs from Matuku otagoense in the more proximal position of dorsal opening of the distal foramen relative to the proximal end of trochlea metatarsi III (equal in Matuku ), the significant medial protrusion of trochlea metatarsi II from the shaft (weak in Matuku ), and the weak crista plantaris lateralis (strong in Matuku ).

Subfamily Agaminae (Agami Heron) Although not examined by us, the skeleton of Agamia is said to differ little in morphology from the Ardeini and Egrettini (Payne & Risley, 1976).

Coracoid ( Fig. 2 ). The fossil coracoid most closely matches that of Cochlearius . The three autapomorphies on the coracoid (see diagnosis of Matuku above) justify the placement of Matuku otagoense outside the five traditional subfamilies of Ardeidae .

Quadrates (S.50852, Fig. 2 ; S.50854, not illustrated)

The two quadrates are typical ardeid quadrates. In overall proportions and morphology they are most similar to Cochlearius but differ in lacking a large foramen pneumaticum mediale. These quadrates, that we refer to Matuku , are distinguished by one autapomorphy from all living species of the Ardeidae examined: in ventral aspect the vallecula intercondylaris is a deep “U” shaped fossa, opening into a wide sulcus on the rostral surface.

Axis ( Fig. 1 ). The axis does not differ significantly from some extant members of the Egrettini and Ardeini , although the knob-like protruding hypapophysis (processus ventralis corporis) is more similar to Cochlearius . However, Matuku differs from Cochlearius in relative proportions. The axis of Matuku is not as elongate as it is in Ardea and Egretta , nor so short and robust as it is in Nycticorax .

Mandible tip ( Fig. 3 ). The mandible tip referred to Matuku is most similar in proportions to that of Nycticorax , with the symphysis much shorter than it is in Egretta or Ardea . A mandible of this type suggests a bill designed for the stabbing capture of prey.

Are the elements from the same species?

Comparison of available measurements from Matuku ( Fig. 4 ) with those of living species of heron using Simpson Ratio diagrams ( Simpson, 1941), suggests that these elements are probably from the same species as they co-vary from the standard taxon in a fashion consistent to that seen in other taxa.

Phylogenetic analyses

We defined 50 multistate characters and coded these for 18 ingroup and six outgroup taxa. The non-ardeid Ciconiiformes , the pelecaniform and the phoenicopteriform were defined as the outgroup. The remaining 18 terminal taxa, including the four extant Recent New Zealand herons and the fossil taxon, were unconstrained.

We tried including Scopus in the data set and found that it tended to have a sister relationship to herons in preliminary analyses, but in the present dataset its inclusion introduced conflict in the data causing a polytomy of it and the main heron clades. For this reason we present the analyses with it excluded. Similarly, preliminary analyses with all characters unordered resulted in 562 shortest trees, length 146. But when 19 characters (characters 1, 2, 4, 6–14, 18, 19, 23, 25, 35, 36, 47) that clearly formed morphoclines, were treated as ordered, the analysis retrieved just 12 shortest trees, length 153, for which a strict consensus tree was completely resolved. For this our preferred tree, we show bootstrap support (1000 bootstrap replicates of the Heuristic search) and the numbers of unambiguous apomorphies supporting each node ( Fig. 5 ).

Our strict consensus tree of the 12 shortest trees retrieved three major clades of herons with significant bootstrap support: 1 a bittern clade (bootstrap 83%), egrets and day herons (bootstrap 71%), and a clade of night herons (bootstrap 76%) ( Fig. 5 ). Although we used few elements and, in general, only scored characters that were present in our poorly preserved fossil specimens, our analysis does not support the monophyly of the true day herons (Tribe Ardeini ; Kushlan & Hancock, 2005), despite this group being only represented by Ardea in our analyses, but does support the monophyly of the bittern subfamily Botaurinae . Like many other authors, we find the genera Ardea and Egretta to both be polyphyletic. The boat-billed heron was found to be sister to the bittern clade, but this relationship was weak (bootstrap <50%; 4 unambiguous apomorphies).

Matuku otagoense is strongly supported as a member of the Ardeidae (Bootstrap 99%), but there is no support for it being a member of the Ardea-Egretta clade (Bootstrap = 71%, 2 unambiguous apomorphies), the bittern clade (Bootstrap = 83%, 3 unambiguous apomorphies), or the night herons Nycticorax, Nyctinassa and Pilherodius (Bootstrap = 76%, 3 unambiguous apomorphies). However, 4 unambiguous apomorphies support Tigrosoma joining the night herons as a clade, and 2 unambiguous apomorphies support Syrigma joining this enlarged night heron clade to the exclusion of the St Bathan’s heron. There is no support (no unambiguous apomorphies, Bootstrap <50%) for Matuku otagoense forming a clade with egrets and bitterns sister to night herons. and there is only limited support for a clade of egrets and bitterns to the exclusion of SB heron (1 unambiguous apomorphy, Bootstrap = 0.250).

A Bayesian consensus tree derived from 4,001 trees sampled after the burnin period has credibility values (percentage posterior probabilities) exceeding 0.50 as shown in Fig. 6 . The probability plot in Tracer plateaued by 250,000 generations, so the burnin discard of 1,000,000 generations was more than adequate. The two runs achieved stationarity and convergence: log likelihood statistics (LnL) after Burnin = 4001 for Run 1 (mean = –546.302, s.d. = 0.099, Effective Sample Size = 3208.424); Run 2 (–546.223, 0.113, 2388.97), and after 5,000,000 generations the average standard deviation of split frequencies was 0.03750, well within the recommended cutoff value of <0.1.

The Bayesian consensus tree ( Fig. 6 ), retrieved the same three well supported clades as the parsimony analysis and differed in topology from the parsimony consensus tree ( Fig. 5 ) only in the branching order of the three main clades and in which nodes all had weak Bootstrap support. This is unlikely to be the result of missing data in Matuku as only four characters are missing thus we consider that this data set simply is not adequate to resolve these deeper relationships.

So in summary these analyses show that Matuku otagoense is a member of Ardeidae but is there is no support for a sister group relationship with any of the main groups of herons.

Figures

Figure 1. Specimens of the fossil heron Matuku otagoense. Scale bar is 1 cm. Right coracoid, extremitas omalis, paratype of Matuku otagoense, S.50004, HH1a, in medial (A) and lateral (B) aspects; left tarsometatarsus, holotype of Matuku otagoense, S.50003, HH4, in plantar (C) and dorsal (D) aspects; and referred axis vertebra S.50853, HH4, in cranial (E) and left lateral (F) aspects.Abbreviations:fac, clavicular facet, facies articularis clavicularis; smc, supracoracoidal sulcus, sulcus musculi supracoracoidei; ridge, character 23 “Slight ridge running sternally, barely dividing sulcus into two segments”; pa, acrocoracoid, processus acrocoracoideus; fah, humeral facet, facies artic. humeralis; ila, impression for the acrocoracohumeralis ligament, impressio ligamentum acrocoracohumeralis; bt, brachial tuberosity; p, procoracoid, proc. procoracoideus; cpl, crista plantaris lateralis; fm1, fossa metatarsal I; imm, Incisura intertrochlearis medialis; tm3, trochlea metatarsi III; po, dens, processus odontoideus; fcr, cranial facies of articulation, facies articularis cranialis; zcr, prezygapophysis, zygapophysis cranialis; fca, caudal facies of articulation, facies artic. caudalis; zca, postzygapophysis, zygapophysis caudalis; pv, hypapophysis, proc. ventralis corporis; fp, pneumatic foramen, foramen pneumaticum; ps, spinous process, proc. spinosus; ft, transverse foramen, foramen tranversarium.

Figure 2. Referred specimens of the fossil heron Matuku otagoense. Scale bar is 1 cm. Left quadrate (S. 50852; HH4): (A) caudal aspect; (B) rostral aspect; (C) lateral aspect; (D) medial aspect; (E) ventral aspect. Abbreviations: o, capitulum oticum; ic, vallecula intercapitularis; s, capitulum squamosum; ct, crista tympanica; c, condylus caudalis; pt, condylus pterygoideus; pm, processus medialis; m, condylus medialis; l, condylus lateralis; pl, processus lateralis; cm, crista medialis; cl, crista lateralis; or, processus orbitalis; fb, fossa basiorbitalis; pf, facies pterygoidea; dp, depression praecondylaris; qj, cotyla quadratojugalis; fm, foramen pneumaticum mediale; in, vallecula intercondylaris; t on dp, tubercle on depressio praecondylaris.

Figure 3. Mandible tips of referred specimen of the fossil heron Matuku otagoense n.sp., (S.51174, HH1a), (A, C, E) and Recent Nankeen Night Heron Nycticorax caledonicus, SAM B.48523, South Australia, (B, D, F). Scale bar is 1 cm. Rostrum mandibulae: in medial aspect (A, B); in dorsal aspect (C, D), and in ventral aspect (E, F) ventral aspect. Abbreviation: ps, pars symphysialis.

Figure 4. Ratio diagram (after Simpson, 1941) using measurements of Nycticorax caledonicus as the origin (i.e. log of measurement of Nycticorax caledonicus is equivalent to 0). This diagram indicates that the relative proportions of the fossil heron bones vary in a similar way to those of other herons and so it is likely that the fossil bones come from a single species. Abbreviations: (A), tarsometatarsus, maximum distal width; (B), tarsometatarsus, maximum distal depth; (C), tarsometatarsus, width of trochlea metatarsi III; (D), Tarsometatarsus, estimated total length; (E), coracoid, maximum cranial width (from hum.facet to brachial tub.); (F), coracoid,maximum shaft width (below procoracoid); (G), coracoid, length of humeral facet; (H), coracoid, depth of humeral facet; (I), quadrate, mean of 2 individuals, depth from capit. squam. to cond. lateralis; (J), quadrate, mean of 2 individuals, depth from capit. squam. to cond. medialis; (K), axis, width of facies articularis cranialis; (L), axis, inter-condyle distance; (M), axis, width of single zygapophyses caudalis; (N), axis, depth single zygapophyses caudalis; (O), axis, greatest width (across facies artic. caudalis).

Figure 5. A strict consensus tree of the 12 shortest trees (length = 153, CI = 0.4444, HI = 0.5556, RI = 0.7222) in which the topology was constrained with Ardeidae as sister group to an outgroup comprised of the non ardeid Ciconiiformes and the pelecaniform—Phalacrocorax carbo. Bootstrap support values (> 0.50) are shown above and the number of significant (> 0.50) unambiguous apomorphies are shown below the corresponding node.

Figure 6. A Bayesian consensus tree derived from 4,001 trees sampled: Run 1 (mean = –546.302, s.d. = 0.099, Effective Sample Size = 3208.424); Run 2 (–546.223, 0.113, 2388.97). Support values are shown above the corresponding node.