Palaeogeography, Palaeoclimatology, Palaeoecologylithornis.nmsu.edu/~phoude/Bell et al 2018 Morphometric comparison of the... · aDinosaur Institute, Natural History Museum of Los

Contents lists available at ScienceDirect

Palaeogeography, Palaeoclimatology, Palaeoecology

journal homepage: www.elsevier.com/locate/palaeo

Morphometric comparison of the Hesperornithiformes and modern divingbirds☆

Alyssa Bella,⁎, Yun-Hsin Wua,b, Luis M. Chiappea

a Dinosaur Institute, Natural History Museum of Los Angeles County, 900 Expedition Blvd., Los Angeles, CA 90007, USAbDept. of Earth Sciences, University of Southern California, Los Angeles, CA 90089, USA

A R T I C L E I N F O

Keywords:MesozoicEvolutionHesperornithiformesFoot-propelled diving birds

A B S T R A C T

The Cretaceous Hesperornithiformes are an extinct group of aquatic birds long recognized to have practiced foot-propelled diving. This specialization is seen today in a number of modern avian families (loons, grebes, cor-morants, and some ducks) that have convergently derived a diving lifestyle. Historically, hesperornithiformshave usually been compared to modern loons and grebes as analogues. However, these comparisons are based onqualitative observations of skeletal morphology and have never been tested using morphometric data. Studyingthe morphological details underlying this functional convergence provides an opportunity to better understandthe lifestyle of the extinct Hesperornithiformes, particularly in regard to adaptations for foot-propelled divingand ecological niches these birds may have occupied.

This study presents a morphometric analysis of the hindlimb of hesperornithiforms and modern foot-pro-pelled diving birds. Measurements were collected from the tarsometatarsus, tibiotarsus, and femur of eightgenera of modern birds and six genera of hesperornithiforms. In order to explore variation within the data, anumber of different combinations of measurements (for example, tarsometatarsus and tibiotarsus, or tibiotarsusand femur) were subjected to principal components analyses (PCA). The primary reason for investigating thedata in different subsets was to increase taxonomic inclusion among the fossil specimens, where missing datawas the largest impediment to the study.

Results of this study show that hesperornithiforms rarely shared morphospace with loons or grebes, but moreoften did share morphospace with cormorants and diving ducks. Therefore, loons and grebes may not serve asthe most appropriate analogue for the Hesperornithiformes. Use of these sorts of analyses, in conjunction withdetailed morphological work, may enhance our understanding of the evolution of complex ecological strategiesand niche partitioning among birds and other organisms.

1. Introduction

The use of modern analogues to better understand extinct taxa haslong been a cornerstone of paleontological research (e.g., Osi andBarrett, 2011; Schonberg and Tapanila, 2006; Uchman et al., 2011). Inhis initial description of the Cretaceous toothed diving bird Hesperornis,Marsh said of the newly discovered fossils, “they represent a giganticswimming bird, having its nearest living allies probably in the Co-lymbidae” (Marsh, 1872; p. 361; Colymbidae is no longer an acceptedtaxonomic group, but at the time included modern loons and grebes).While some later workers followed Marsh's work in considering theHesperornithiformes as basal members of modern grebes (Brodkorb,1963, 1971), loons (Simpson, 1980), or both (Cracraft, 1982), modernphylogenetic analyses have established the Hesperornithiformes as

basal members of the Ornithurae, and outside the crown clade Neor-nithes (Chiappe, 2002; Clarke, 2004; Bell and Chiappe, 2015a).Throughout Marsh's seminal monograph on the Hesperornithiformes,he identified similarities in the skeletons of Hesperornis and its smallerclose relative Baptornis with those of loons and grebes, thus establishingthese two groups of modern birds as analogues for the extinct Hesper-ornithiformes (Marsh, 1880). Marsh's work was extremely influentialand subsequent researchers further promoted loons and grebes asmodern analogues of hesperornithiforms (Mayr and Anderson, 1951;Stolpe, 1935; Storer, 1958; Olson, 1992; Martin et al., 2012). Yet theevidence in support of this view was primarily qualitative, most typi-cally references to general morphology (primarily a reduced femur andcompressed tarsometatarsus), and lacked justification for the exclusiveuse of those taxa as analogues (e.g., Stolpe, 1935; Storer, 1958; Olson,

https://doi.org/10.1016/j.palaeo.2017.12.010Received 2 July 2017; Received in revised form 8 November 2017; Accepted 14 December 2017

☆ This work was supported by generous donations from the Gretchen Augustin family to the Dinosaur Institute of the Natural History Museum of Los Angeles County.⁎ Corresponding author.E-mail address: [email protected] (A. Bell).

Palaeogeography, Palaeoclimatology, Palaeoecology xxx (xxxx) xxx–xxx

0031-0182/ © 2018 Elsevier B.V. All rights reserved.

Please cite this article as: Bell, A., Palaeogeography, Palaeoclimatology, Palaeoecology (2018), https://doi.org/10.1016/j.palaeo.2017.12.010

1992; Acosta Hospitaleche and Gelfo, 2015). While recent work withthe newly erected Vegaviidae further builds upon the relationship be-tween these and other osteological features as indicators of foot-pro-pelled diving (e.g. thickened cortical bone, bowed femoral shaft, ex-panded tibial condyle, etc.) (Agnolin et al., 2017), work has yet to bedone to quantify proposed diving-associated skeletal features.

Undoubtedly, the modifications of the hindlimb of Hesperornis andits relatives strongly support early interpretations of a diving lifestyle,which are compounded by the reduced size of the forelimb and therecovery of the vast majority of fossils from off-shore deposits. Suchecological specializations help explaining the broad geographic dis-tribution of the group, as most species (if not all) would have beenflightless. The Hesperornithiformes are represented by twenty-threespecies in eleven genera, distributed throughout the northern hemi-sphere and sharing adaptations of the hindlimb characteristic of foot-propelled divers (e.g., shortened femur, flattened tarsometatarsus,

expanded cnemial crest on tibiotarsus) (Bell and Chiappe, 2015a). Thus,the ancient Hesperornithiformes continue to be regularly considered asa classic example of morphological convergence with loons and grebes(Reynaud, 2005; Bell and Chiappe, 2015a, 2015b). Some researchershave indeed noted differences between these birds and modern loonsand grebes but these differences have never been quantified and theunderlying assumption that these modern taxa are appropriate analo-gues has not been seriously questioned (Olson, 1992; AcostaHospitaleche and Gelfo, 2015). However, foot-propelled diving is notlimited to modern loons and grebes. There are a number of other generaof modern birds that are aquatic and swim via foot-propelled diving,primarily the cormorants (Phalacrocorax; Watanuki et al., 2005; Katoet al., 2006) and certain diving ducks (Aythya – Lovvorn et al., 2001;Mergus – Kovacs and Meyers, 2000). None of these main groups of di-vers are particularly closely related to each other (Jarvis et al., 2014;Prum et al., 2015), indicating convergent evolution leading to the foot-

Fig. 1. Measurements collected from the tarsometatarsus(A), tibiotarsus (B), and femur (C). Measurements are in-dicated by numbers as follows: 1) length, 2) proximalwidth, 3) mediolateral midshaft width, 4) craniocaudal(dorsoplantar in the tarsometatarsus) midshaft width, 5)distal width, 6) craniocaudal width of lateral condyle, 7)craniocaudal width of head, 8) length of cnemial process.

A. Bell et al. Palaeogeography, Palaeoclimatology, Palaeoecology xxx (xxxx) xxx–xxx

2

propelled diving lifestyle. As such, all of these groups might be rea-sonable analogues for the extinct Hesperornithiformes, even if onlyloons and grebes have previously been considered (e.g., Stolpe, 1935;Olson, 1992; Acosta Hospitaleche and Gelfo, 2015).

Therefore, the purpose of this study is to examine the moderngroups of foot-propelled divers and determine which group or groupsare most appropriate as modern analogues for the diving properties ofthe extinct Hesperornithiformes. The approach of this study is two-fold:1) a quantitative, multivariate analysis of measurements of the hin-dlimbs of these birds and 2) a qualitative discussion of some specificfeatures that could not be included in the quantitative analysis. Modernfoot-propelled diving birds differ widely in their habitat and foragingstyles. Better understanding of the morphological differences of thesebirds, as well as how they compare to hesperornithiforms, will shedlight on the evolution of this specialized lifestyle and provide insightsinto the biology of the extinct Hesperornithiformes.

2. Materials and methods

2.1. Morphometric data

Measurements of modern taxa were collected from the skeletalcollections of the Ornithology Department at the Natural HistoryMuseum of Los Angeles County. All foot-propelled diving speciesavailable in the collection were included in the database. Measurementswere collected from modern taxa [59 cormorants (Phalacrocorax), 66diving ducks (Aythya, Mergus), 69 grebes (Aechmophorous, Podiceps,Podilymbus, Rollandia), and 30 loons (Gavia)] and 74 hesperornithiformspecimens from collections across North America and Europe [Baptornis(5), Brodavis (2), Fumicollis (1), Hesperornis (58), Parahesperornis (3),Pasquiaornis (5)]. Of the hesperornithiform specimens included, all buttwo were measured directly, with measurements for YPM 1474 (H.crassipes; Marsh, 1880) and ZIN PO 5464 (H. rossicus; Panteleyev et al.,2004) taken from published measurements or photographs. In order toensure these measurements were not biasing the outcomes of the PCA,the tarsometatarsus analysis (see details below) was performed withand without these two specimens, with no noticeable change to therelative positions of the other specimens in morphospace. Measure-ments were collected from the femur (6 measurements), tibiotarsus (5measurements), and tarsometatarsus (4 measurements) (Fig. 1). Thechoice of measurements used reflects those dimensions that are mostcommonly preserved in hesperornithiforms that also capture the shapeof the element (see Supplementary Fig. 1 for a summary of the abun-dance of hesperornithiform skeletal elements). Appendix 1 presents thecomplete database of modern bird measurements.

2.2. Principal component analysis (PCA)

Principal Component Analyses (PCA) of the log-transformed vari-ables were carried out in PAST (Hammer and Harper, 2005). SevenPCAs were conducted that varied by which elements were included[tarsometatarsus (Tmt), tibiotarsus (Tbt), femur (Fmr), tarsometatarsus+ tibiotarsus (TmtTbt), tarsometatarsus + femur (TmtFmr), tibio-tarsus + femur (TbtFmr), and tarsometatarsus + tibiotarsus + femur(TmtTbtFmr)]. Multiple analyses were conducted in order to maximizethe number of hesperornithiform specimens that could be included, ashesperornithiform fossils are, in general, highly fragmentary and rarelypreserve all three hindlimb elements considered here. Relying on asingle element or combination of elements would have drastically re-duced the number of fossil specimens and taxa that could be analyzed(see Supplementary Table 1 for the number of hesperornithiform spe-cimens that were included in each of the PCAs).

3. Results

The principal component scores from the covariance matrix for allseven PCAs are shown in Fig. 2 and given in Appendix 2. For three ofthe analyses (Tmt, Fmr, TmtFmr), PC1 described over 90% of the var-iance, while in the other four analyses (Tbt, TmtTbt, TbtFmr,TmtTbtFmr) PC1 described 69–72% of the variance. The results of theanalyses are shown graphically in Figs. 3–5, with the loadings availablein Supplementary Table 2.

When comparing PC1 and PC2, hesperornithiforms rarely coincidein morphospace with modern taxa. The exceptions to this are thesmaller hesperornithiforms, Pasquiaornis, Baptornis, Brodavis amer-icanus, and Fumicollis, which overlap with cormorants (particularly P.harrisi, the flightless cormorant) in the Tmt and TmtFmr analyses andwith cormorants and loons in the Fmr analysis. The modern taxa wereall somewhat separated in morphospace, and rarely overlapped.Exceptions to this are the two genera of diving ducks, Aythya andMergus, which partially overlapped in five of the seven analyses (Tbt,Fmr, TmtTbt, TbtFmr, TmtTbtFmr), and two of the grebe genera,Aechmophorus and Podiceps, which also overlapped in those same fiveanalyses. In general, the only analysis in which PC1 vs PC2 showed highdegrees of overlap in morphospace was that of the femur, where thecentral morphospace is shared by cormorants, some grebes(Aechmophorus, Podiceps), loons, and the hesperornithiformsPasquiaornis hardiei and Baptornis advenus.

Comparison of PC2 vs PC3 shows more overlap between modernbirds and hesperornithiforms in morphospace. Hesperornithiformsoverlap most frequently with cormorants (all analyses except TbtFmr)and diving ducks (all analyses except Tbt, Fmr), and with grebes inthree analyses (Tmt and Fmr – Podilymbus and Rollandia, Fmr – allmodern genera). Modern birds show only slightly more overlap amongthemselves, with diving ducks and some grebes (Aechmophorus,Podiceps) overlapping when comparing PC1 and PC2, but with threeanalyses (Tmt, Fmr, TmtFmr) showing high degrees of overlap amongall or most taxa (Supplementary Fig. 3 shows a full comparison of themodern taxa that share morphospace with hesperornithiform taxa foreach analysis).

The discussion section below presents possible interpretations forwhat the principal components of each analysis represent, as indicatedby the magnitude and direction of the loadings for each variable.

4. Discussion

The PCAs presented in this study provide the first quantitativecomparison of the extinct Hesperornithiformes to representatives of allgroups of modern foot-propelled diving birds. These analyses reveal anumber of differences in the proportions of and between the hindlimbelements studied, as discussed below. It should be noted that theloadings from all analyses show that all variables are strongly projected

Fig. 2. Scores for the first three principal components for each of the seven principalcomponent analyses.

A. Bell et al. Palaeogeography, Palaeoclimatology, Palaeoecology xxx (xxxx) xxx–xxx

3

onto PC1, ranging in value from 0.14 to 0.61 (Supplementary Table 2).This is most relevant to the Femur, Tarsometatarsus, andTarsometatarsus + Femur PCAs, where the eigenvalues indicate PC1accounts for over 90% of the variance in the data (Fig. 2). In general,PC1 is commonly thought of as summarizing body size variation inordinations of linear measurements (e.g. Jolicoeur, 1963; Berner,2011). In this study, the individuals in the dataset range widely in size,from Rollandia rolland with a tarsometatarsus length of around 35 mmto Hesperornis rossicus, with a tarsometatarsus length of 170 mm.Therefore, it seems reasonable that PC1 reflects the wide variation inbody size found in the dataset. This is demonstrated for these data inFig. 6, which shows a linear relationship between PC1 and the natural

log of body mass estimates (r2 = 0.856).

4.1. Skeletal adaptations for foot-propelled diving

The following sections integrate the results of the implications foreach of the elements included, as well as relative element lengths, asindicated through the PCAs. These implications are then related toqualitative features and morphological structures that vary among thestudied taxa, and therefore may relate to the differing evolutionaryhistories of these diverse birds.

Fig. 3. PCAs of the: tarsometatarsus (A), tibiotarsus (B), and femur (C) measurements plotting PC1 vs. PC2 (left) and PC2 vs. PC3 (right). Modern taxa are shown in grey clouds andabbreviated as follows: C, Cormorant Phalacrocorax; DM, Diving Duck Mergus; DA, Diving Duck Aythya; GA, Grebe Aechmophorus; GPd, Grebe Podiceps; GPd, Grebe Podilymbus; GR, GrebeRollandia; L, Loon Gavia. Loadings are labelled with a number that corresponds to those shown on the measurements in Fig. 1.

A. Bell et al. Palaeogeography, Palaeoclimatology, Palaeoecology xxx (xxxx) xxx–xxx

4

4.1.1. TarsometatarsusIn a number of analyses the PCA loadings for the dorsoplantar and

mediolateral midshaft widths of the tarsometatarsus are opposite indirection (Tmt PC2, TmtTbt PC3, TmtFmr PC2, TmtTbtFmr PC2, 3).This indicates a high degree of variability in the cross-sectional shape ofthe tarsometatarsus, as increases in one dimension relative to the otherwould result in a more oval cross section. Fig. 7 shows a plot of tar-sometatarsal dorsoplantar vs mediolateral midshaft width. Within themodern taxa, two clear patterns are observed. Loons (Gavia) and thegrebes Podiceps and Aechmophorus have a larger dorsoplantar midshaftwidth while the cormorants, Phalacrocorax, and the diving duck Aythyahave a larger mediolateral width. The grebes Rollandia and Podylimbusand the diving duck Mergus appear to fall intermediate to, and partiallyoverlapping with, these two patterns, showing a more circular

midshaft. Hesperornithiforms range widely in these dimensions, as bestexhibited by Hesperornis, which has no clear pattern in the data. An-other pattern seen in the data is the relationship of the midshaft widthsof the tarsometatarsus to tarsometatarsus length. While the loadings ofthe mediolateral width are opposite to those of the length in all of therelevant analyses (as the mediolateral width increases the relativelength decreases, and vice versus), those of the dorsoplantar width areonly rarely opposite those of the length (Supplementary Table 2). Fig. 8shows both midshaft widths plotted against length. The mediolateralmidshaft width shows a variable relationship with length across all taxa(Fig. 8a), however, the dorsoplantar midshaft width shows a uniformlinear relationship to length in the majority of specimens, across all taxa(Fig. 8b). Supplementary Fig. 3 shows the variation in tarsometatarsusshape of all included genera, with relevant measurements to this

Fig. 4. PCAs of the: tarsometatarsus + tibiotarsus (A), tarsometatarsus + femur (B), and tibiotarsus + femur (C) plotting PC1 vs. PC2 (left) and PC2 vs. PC3 (right). Modern taxa areshown in grey clouds and abbreviated as follows: C, Cormorant Phalacrocorax; DM, Diving DuckMergus; DA, Diving Duck Aythya; GA, Grebe Aechmophorus; GPd, Grebe Podiceps; GPd, GrebePodilymbus; GR, Grebe Rollandia; L, Loon Gavia. Loadings are labelled with a number that corresponds to those shown on the collected measurements in Fig. 1.

A. Bell et al. Palaeogeography, Palaeoclimatology, Palaeoecology xxx (xxxx) xxx–xxx

5

analysis labelled.The variation in mediolateral, but not dorsoplantar, midshaft width

of the tarsometatarsus as shown in Fig. 8 may have implications fordiving mechanics of the foot. While a limited body of research has beendeveloped investigating the effects of element lengths (Zeffer andLindhe Norberg, 2003) or musculature (Wilcox, 1952; Zinoviev, 2010)on diving capabilities, none have examined element shape. Flattening

Fig. 5. PCAs of the combined tarsometatarsus + tibiotarsus + femur plotting PC1 vs.PC2 (A) and PC2 vs. PC3 (B). Modern taxa are shown in grey clouds and abbreviated asfollows: C, Cormorant Phalacrocorax; DM, Diving Duck Mergus; DA, Diving Duck Aythya;GA, Grebe Aechmophorus; GPd, Grebe Podiceps; GPd, Grebe Podilymbus; GR, Grebe Rollandia;L, Loon Gavia. Loadings are labelled with a number that corresponds to those shown onthe collected measurements in Fig. 1.

Fig. 6. PC1 plotted against body mass estimates. Mass estimates were made using theequation of Maloiy et al. (1979) based on femur length. Trend lines calculated in Mi-crosoft Excel.

Fig. 7. Tarsometatarsus mediolateral vs. dorsoplantar midshaft widths. Inset shows themeasurements plotted.

A. Bell et al. Palaeogeography, Palaeoclimatology, Palaeoecology xxx (xxxx) xxx–xxx

6

of the tarsometatarsus can be achieved by reducing the dorsoplantarwidth as compared to the mediolateral width. This might assist innarrowing the profile of the foot while swimming and thereby reducingdrag, particularly if the narrow side of the foot is rotated and orientedagainst the flow of water. This has been noted for loons (Wilcox, 1952),and such specialization would have been even better developed in he-sperornithiforms. Many hesperornithiforms have a unique stacking ar-rangement of the metatarsals not seen in modern birds, such that themedial margin of the tarsometatarsus forms a sharp angle, which wouldhave more easily cut through water during the recovery portion of thefoot stroke.

4.1.2. TibiotarsusWhile few complete hesperornithiform tibiotarsi were available for

analysis, the PCAs including the tibiotarsus show striking variation inthe length of the cnemial process across all taxa. The cnemial process ofthe tibiotarsus is formed from the cranial and lateral cnemial crests. Inmost birds, the cranial cnemial crests extend slightly proximally beyondthe proximal articular surface of the tibia. However, in many aquaticbirds both the cranial and lateral crests extend far more proximally,forming the medial and lateral borders of a marked triangular cnemialprocess (Baumel and Witmer, 1993). Within this dataset, the amount ofvariation in the length of the cnemial process is indicated by themagnitude of the loadings. In the TmtTbtFmr analysis the loading onthis variable is very high, indicating it to be one of the largest sources ofvariation in the total dataset. In all of the loadings for relevant analyses,the length of the cnemial process is opposite in direction from theproximal width of the tibiotarsus, and often for the length as well(Supplementary Table 2). Fig. 9 shows the cnemial process lengthplotted against total length (Fig. 9a) and proximal width (Fig. 9b).Among modern birds three distinct patterns are seen. Loons (Gavia)have a highly elongate cnemial process, much more so than any of the

other taxa. The grebes (Aechmophorus, Podiceps, Podilymbus, Rollandia)follow a different pattern, with a relatively shorter cnemial process thanthat of the loons. The cormorants (Phalacrocorax) and diving ducks(Aythya, Mergus) follow a third pattern, with the smallest relativecnemial process of all taxa. The flightless cormorant, P. harrisi, appearsto be slightly off this trend, with the cnemial process longer than wouldbe anticipated given the relationship observed in other cormorants. Thehesperornithiforms vary widely in these proportions. Unfortunately thelow numbers of specimens included for the majority of genera makeextrapolating patterns difficult, however taken as a single group theyappear to follow a pattern intermediate to that of the cormorants anddiving ducks and the grebes.

The cnemial process functions in concert with the patella as theorigin for several muscles involved in the flexion (m. extensor digitorumlongus, m. tibilais cranialis) and rotation (m. iliotibialis cranialis, m. fe-morotibialis medius,M. tibialis cranialis) of the foot as well as extension ofthe pes and pedal phalanges (m. gastrocnemius, m. extensor digitorumlongus) (Wilcox, 1952; Owre, 1967; Zinoviev, 2010). The size and shapeof the patella varies widely among the birds included in this analysis(Supplementary Fig. 4). Loons (Gavia) have an extremely reduced pa-tella, which is “a small, free flake of bone” imbedded in the patellartendon (formed from the m. femorotibialis externus and medius and them. iliotibialis) (Wilcox, 1952; p. 517). The reduction is so extreme thatthe patella often does not survive skeletal preparation, and so was atone time thought to have been absent (e.g. Shufeldt, 1884; Thompson,1890). In the present study none of the loon specimens measured in-cluded a patella, which precluded patellar measurements from inclu-sion in this analysis. Among the other modern birds, there are two mainforms of patella. Grebes have a thin, triangular patella while cormor-ants and diving ducks have a more robust patella, being pyramidal inthe former and almost round in the latter. Hesperornithiforms show asimilar range of variation in patellar shape and size, with Hesperornisand Parahesperornis having thin, elongate patellae (like grebes) whileBaptornis and Fumicollis have shorter, pyramidal patellae (like cormor-ants). The patella is unknown for other hesperornithiform taxa (Sup-plementary Fig. 4 shows the tibiotarsus and patella of each of thegenera in dataset). The muscle attachments thus vary widely amongfoot-propelled divers depending on the size and shape of the patella.Among modern divers, loons are entirely reliant on the cnemial processfor muscular attachments (Wilcox, 1952) while other taxa divide at-tachments between the patella and cnemial process (Owre, 1967;Zinoviev, 2010). Thus neither the lack of a patella (as in loons) nor areduced cnemial process (as in cormorants and diving ducks) necessa-rily represent a loss in muscle attachment area, as muscle attachmentsmay be divided between the two.

While the general shape of the patella of Hesperornis andParahesperornis is most similar to that of grebes, the size of the patella ofthese hesperornithiforms is remarkably expanded, even given theoverall size discrepancy among these birds (Supplementary Fig. 4). Thepatella of Baptornis and Fumicollis much more closely resembles that ofcormorants, in terms of relative size and shape. This has two mainimplications for the areas of muscle attachment. First, the expandedcranial and caudal faces of the patella where the m. extensor digitorumlongus originates and the m. femorotibialis medialis inserts is shortened inBaptornis, Fumicollis, and cormorants. The m. extensor digitorum longushas an additional origin along the cranial surface of the tibiotarsal shaft,and so reduction of the patellar origination does not represent a largeloss overall. However, the insertion area of the m. femorotibialis medialisis much reduced in cormorants as compared to loons and grebes. Inloons this muscle inserts along the expanded cnemial process (Wilcox,1952) while in grebes, and presumably in Hesperornis, it inserts alongthe expanded patella (Zinoviev, 2010). Cormorants lack an expandedcnemial process, thus representing a reduction of insertion area for them. femorotibialis medialis (and also the m. extensor digitorum longus).Given that Baptornis and Fumicollis also lack the expanded cnemialprocess of loons or the expanded patella of grebes and Hesperornis, they

Fig. 8. Tarsometatarsus length vs. mediolateral midshaft width (A) and dorsoplantarmidshaft width (B). Insets show the measurements plotted.

A. Bell et al. Palaeogeography, Palaeoclimatology, Palaeoecology xxx (xxxx) xxx–xxx

7

may also have had a reduced insertion area for the former muscle. Them. femorotibialis medialis assists in fixing the knee joint in the loon, achange from its role as an extender of the lower leg in other birds(Wilcox, 1952). Differences in the size of the insertion area of thismuscle among hesperornithiforms may be reflecting the different de-grees of foot-propelled diving specializations among these birds (seeBell and Chiappe, 2015a, 2015b).

A second implication for the shortened patella in Baptornis andFumicollis is reduced power in the M. tibialis cranialis. In grebes thismuscle is restricted to the patella (Zinoviev, 2010), while in cormorantsit originates on both the cnemial process and the cranial surface of thepatella (Owre, 1967). The latter seems the most likely arrangement forBaptornis and Fumicollis, given the similarity of their patella to that ofcormorants. Furthermore, a similar division between the cnemial pro-cess and patella has been proposed for Hesperornis (Zinoviev, 2010). Ifcorrect, this would imply a much larger attachment area for Hesperornisalong the entire length of the greatly expanded patella, and thereforepotentially a more powerful muscle than in any of the other birds,modern or extinct, capable of stronger ankle flexion. As this motion isthe primary source of force generation while diving (Wilcox, 1952;Johansson and Lindhe-Norberg, 2001), it appears that Hesperornis andParahesperornis may have been much stronger divers than any of themodern divers or their smaller relatives Baptornis and Fumicollis.

4.1.3. FemurIn a number of the analyses the PCA loadings for femur length were

opposite those of width, particularly the mediolateral midshaft widthand distal width (Fmr PC2,3; TmtFmr PC2,3; TbtFmr PC 3; TmtTbtFmrPC 3). Fig. 10 shows femur length plotted against distal width (Fig. 10a)and mediolateral midshaft width (Fig. 10b). Discernable differencesbetween the modern taxa are more difficult to identify than in thetarsometatarsal or tibiotarsal measurements presented above. However,in Fig. 10b the slopes of the lines for modern taxa do show slight dif-ferences, with those of the diving ducks and loons being steeper(3.5–3.7) than those of cormorants and grebes (1.9–2.4). Baptornis andHesperornis appear to be most similar to cormorants and grebes, withslopes of 2.2 and 1.7, respectively. Both of these plots represent theelongation of the femur, with the diving ducks and loons having slightlymore elongate femora than the other birds. The femur showed lessvariation across measurements for all taxa than did the other elements,as indicated by the generally low magnitude of the loadings in thecombined analyses, particularly in the PCA of all elements (Supple-mentary Table 2). This may indicate stabilizing selection on the femoralproperties measured in this analysis. In foot-propelled diving birds thefemur extends laterally from the pelvis and is held relatively stable

while swimming, with little motion (Wilcox, 1952; Owre, 1967). Footpropelled diving birds propel themselves forward almost entirely withtheir feet, with the controlling muscles primarily originating on thetibiotarsus, but with a few on the distal-most femur (Wilcox, 1952).This might provide strong selective pressures to maintain the orienta-tion of the femur. Supplementary Fig. 5 compares femur shape acrossall genera included in the analysis.

4.1.4. Relative element lengthsThe loadings from all PCAs indicate that while tarsometatarsal and

tibiotarsal length scale linearly across all taxa, femur length is more

Fig. 9. Length of the cnemial process of the tibiotarsus compared to total length (A) and proximal width (B). Insets show the measurements plotted.

Fig. 10. Femur length vs. distal (A) and mediolateral (B) widths. Inset shows the mea-surements plotted.

A. Bell et al. Palaeogeography, Palaeoclimatology, Palaeoecology xxx (xxxx) xxx–xxx

8

variable among taxa (Supplementary Table 2). It has long been re-cognized that diving birds display a reduction in femur length, ascompared to other birds (Owre, 1967; Wilcox, 1952). In diving birds thefemur extends laterally from the pelvis, with the tibiotarsus and

tarsometatarsus extending distally from the knee, perpendicular thefemur. A shorter femur, therefore, enables the bird to align its feet di-rectly behind the body while swimming. In loons (Wilcox, 1952) andgrebes (Johansson and Lindhe-Norberg, 2001) the feet have beendocumented as almost touching behind the body during the last phaseof the power stroke. The PCAs presented here indicate that moderndiving birds vary in the degree of shortening of the femur, but not in therelationship of tarsometatarsal to tibiotarsal length (Fig. 11). Cormor-ants and diving ducks have generally longer femora in proportion toeither lower limb bone, as compared to other modern divers. The trendrelating the size of these elements differs between different groups oftaxa, with slopes for femur vs tarsometatarsus (Fig. 11a) ranging from0.27 to 1.18 and for femur vs. tibiotarsus (Fig. 11b) ranging from 0.21to 0.64. In contrast, the slopes for tarsometatarsus vs. tibiotarsus(Fig. 11c) only range from 0.44 to 0.53 for all taxa. These results mayimply that cormorants and diving ducks are not as specialized for foot-propelled diving as loons and grebes, as the longer femur would reducethe ability for cormorants and diving ducks to get their feet completelybehind their body. It may also represent a trade-off between the abilityto walk upright and diving. Cormorants and diving ducks have not beenreported as having difficulty walking upright, while loons and grebesstruggle to stand on land (del Hoyo et al., 2008). The hesperornithi-forms vary widely in relative femur length. Baptornis and Fumicollis aresimilar to cormorants and diving ducks when comparing femur andtarsometatarsus lengths (Fig. 11a), however they are more similar tothe grebe Podiceps when comparing the femur to the tibiotarsus(Fig. 11b). Hesperornis displays more variability in these measurementsthan modern taxa, particularly in regard to tarsometatarsus length, socomparison to modern birds is difficult. However, the femur vs. tibio-tarsus plot (Fig. 11b) shows Hesperornis and Parahesperornis are mostsimilar to loons. The shortened femur of Hesperornis and Parahesperornisindicates these birds would have been able to align the foot directlybehind the body, thus reducing drag at the end of the power stroke.

4.1.5. Further considerationsThe dataset developed in this study consists of measurements that

represent multiple dimensions of the three main elements of the hin-dlimb. However, the fragmentary hesperornithiform fossil record andskeletal differences among modern taxa (e.g., the lack of a patella inloons) limited the amount of data that could be included. Variation inthe patella was discussed above, however another source of variationamong these diving birds is the articulation and structure of the toes.Most modern diving birds have webbed feet, except for grebes, whichhave asymmetrically lobed toes. Hesperornithiforms do not have anyknown foot impressions to indicate whether they had lobed or webbedfeet. Stolpe (1935) proposed lobed toes for hesperornithiforms on thebasis of the morphology of the articulation between the pedal pha-langes. In grebes and Hesperornis, Stolpe claimed that the lateral tro-chlear ridges on the tarsometatarsus are smaller than the medial, thuswhen the toes are bent they also rotate around this uneven articulation(Stolpe, 1935; see also Martin and Tate, 1976). Furthermore, the pedalphalanges of the fourth toe also have a size discrepancy between thesmaller lateral trochlea and larger medial trochlea (Stolpe, 1935). Closeinspection of the tarsometatarsus of the birds included in this analysisreveals that most taxa have smaller lateral trochlear ridges on thefourth trochlea, not just grebes (Supplementary Fig. 6). The exceptionto this is the loon, Gavia, where the trochlear ridges are similarly sized.Among modern taxa the flightless cormorant (Phalacrocorax harrisi)displays the largest size discrepancy, while among fossil taxa Hesper-ornis and Parahesperornis have the widest discrepancy. However, var-iation between the medial and lateral trochleae of the phalanges of thefourth toe is more variable across taxa, with only grebes, loons, theflightless cormorant, Hesperornis, and Parahesperornis having reducedlateral as compared to medial trochleae. The reduction in the moderntaxa is minimal when compared to the drastic reduction of Hesperornisand Parahesperornis, where the lateral trochlea is reduced to a small

Fig. 11. Comparison of elements lengths: femur vs. tarsometatarsus (A), femur vs. ti-biotarsus (B), and tarsometatarsus vs. tibiotarsus. Insets show the measurements plotted.

A. Bell et al. Palaeogeography, Palaeoclimatology, Palaeoecology xxx (xxxx) xxx–xxx

9

round peg and the medial is expanded to a dramatic arc that wrapsaround the lateral. Among the other birds the medial and lateral tro-chleae are similarly sized.

In grebes the articulations of the toes have been identified as animportant tool for allowing a stacked, or ‘shingled’, arrangement of thetoes when folded (Martin and Tate, 1976), however the degree to whichthis is possible varies widely among species (Storer, 1958). This ar-rangement has been identified as useful for allowing the individuallylobed toes to function as separate hydrofoils, thus increasing the liftgenerated during the power stroke (Johansson and Lindhe-Norberg,2001). The utility of such an arrangement in other diving birds thathave webbed feet, and therefore are incapable of having separate hy-drofoils, has not been explored. It may simply be that by stacking thetoes when folded, birds are able to reduce the surface area of the foot,thus reducing drag during the recovery stroke. It is clear that the dra-matic specializations present in Hesperornis and Parahesperonis are be-yond anything seen in modern birds, and may represent a drastic in-crease in diving capabilities. Further support from this comes from thedeep ventral grooving seen on the plantar surface of the phalanges ofthe fourth toe in Hesperornis and Parahesperornis. This grooving hasbeen interpreted as accommodating thick tendons that provide struc-tural support against the toes collapsing under the resistance force ofwater while swimming (Storer, 1958). The extremely deep grooves seenin Hesperornis and Parahesperornis are beyond anything seen in moderndivers or other hesperornithiforms, indicating an unmatched ability tooppose the resistance of water while generating strong thrust.

It should be noted that this study finds no reason to support lobedtoes over webbed toes for hesperornithiforms. First, the proposed os-teological correlate of lobed toes, a size discrepancy in the lateral andmedial articulations of the tarsometatarsus and toes, is seen to somedegree across a variety of modern divers that have either webbed orlobed feet, as discussed above. Furthermore, not all hesperornithiformshave the proposed osteological correlate - Baptornis and Fumicollis donot show a size discrepancy in the trochlea of phalanx IV. Therefore,even if the proposed osteological correlate is accurate, an interpretationof lobed toes is not supported for all hesperornithiforms. Finally, severalspecies of coot (Fullica) have lobed toes, and do not display a size dis-crepancy between the trochleae on phalanx IV. Therefore, it appearsthere is not a fixed relationship between the proposed skeletal mor-phology and the soft tissue of the toes, thus limiting interpretations ofthe feet of Hesperornis. What is more important is the clear developmentof varying degrees of toe rotation in foot-propelled diving birds, withHesperornis and Parahesperornis showing hyper-rotational abilities.

Another hindlimb region with potentially important implications fordiving capabilities that could not be quantified for the analysis is thearticulation of the femur with the acetabulum of the pelvis. The tro-chanter of the femur articulates against the antitrochanter of the pelvis,while the head of the femur articulates into the acetabulum. The re-lative proximal extents of the head and trochanter of the femur, as wellas the lateral protrusion of the antitrochanter, control the angle atwhich the femur extends from the pelvis. Supplementary Fig. 5 showsthe range of variation among foot-propelled divers. In loons, grebes,and the hesperornithiform Pasquiaornis, the trochanter extends slightlyfarther proximally than the head. In these birds the femur is thereforeangled laterally and ventrally away from the pelvis, extending thelower leg and feet further below the center of the body mass. In theother taxa (cormorants, diving ducks, and most hesperornithiforms),the head and trochanter are even. This allows the femur to splay out-ward and extend almost directly laterally from the pelvis, thus aligningthe feet behind the center of mass. This is the ideal positioning for foot-propelled diving, as the main propulsive force will be generated directlybehind the body.

4.2. Ecological implications

While all modern taxa included in this analysis are foot-propelled

diving birds, the environments they inhabit and the foraging strategiesthey employ vary widely across these birds and play important roles inthe birds' overall ecologies. Many of these fine-scale specializations varymuch more widely than skeletal morphology and therefore the con-clusions that can be applied to hesperornithiforms are limited.Additionally, interpretations of habitat preference for hesperornithi-forms are limited to the depositional environment in which they werediscovered, which may not precisely align with the environment inwhich they lived. However, some general conclusions can be drawn.Hesperornithiforms are predominantly known from marine environ-ments, with some specimens known from continental and transitionalenvironments (Supplementary Fig. 9, Supplementary Table 3). It ispossible that taphonomic processes are responsible for a greater accu-mulation of specimens in marine sediments (and hence an over-estimation of their abundance in these types of habitats), as individualsthat died in coastal areas could have been washed seaward. For ex-ample, the Cambridge Greensand Member preserves reworked Albian-aged fossils, many of which are terrestrial, and were presumably wa-shed out to sea (Benton and Spencer, 1995). It is also worth noting thatthe continental hesperornithiforms are all small taxa, and highly in-complete. This implies that perhaps the largest hesperornithiforms wererestricted to coastal and marine environments, as, if present in con-tinental settings, the larger taxa would be more likely to fossilize thanthe small taxa we know of. Small hesperornithiforms are known frommarine deposits as well, such as the abundant but highly fragmentedPasquiaornis and Enaliornis.

Many of the modern taxa included in this study inhabit lakes andinland wetlands, with seasonal migrations to coastal wetlands and es-tuaries. This pattern of habitat usage is found in loons (Earnst et al.,2006; Remsen and Binford, 1975), grebes (Osnas, 2003; Ulfvens, 1988),Aythya (Cornelius, 1982; Giles, 1994; Torrence and Butler, 2006), andMergus (Kear, 2005; Marquiss and Duncan, 1997). It should be notedthat Mergus also inhabits rivers (Gregory et al., 1997). Cormorants arehighly adaptable in their habitat, with many living near the seashore(Bearhop et al., 2006; Heithaus, 2005), while other species are knownto inhabit inland rivers and lakes (Rudstam et al., 2004; Suter, 1995).The role of flight in habitat utilization must also be considered. Themodern birds included in this study are almost all capable of flight, andmost taxa migrate seasonally. Most hesperornithiforms were un-doubtedly flightless (Marsh, 1880), however flight cannot be ruled outfor some of the smaller or highly incomplete taxa. For example, En-aliornis is a small hesperornithiform that preserves no forelimb material(Galton and Martin, 2002), while Pasquiaornis is a small hesper-ornithiform with some forelimb material that retains characteristics offlying birds (Tokaryk et al., 1997). If capable of flight, these smallerhesperornithiforms may have been capable of migrations to inlandlakes, as many modern divers do. However, the flightless hesper-ornithiforms were unlikely to travel overland due to the extreme spe-cialization of their hindlimbs. Given the restriction of hesperornithi-form fossils to coastal and marine sediments, as well as the inability ofmost taxa to travel overland, it seems reasonable that most hesper-ornithiforms were coastal and/or seagoing birds, much like somemodern cormorants. The few instances of inland hesperornithiformsmay represent species that were actually capable of flight, or perhapsthese birds were flightless and restricted to inland lakes. Amongmodern birds examples of this are found among the three genera offlightless grebes (Livezey, 1989), specimens of which were not availablefor this study.

5. Conclusions

The analyses presented here comprise the first quantitative studycomparing the extinct Hesperornithiformes with all major groups ofmodern foot-propelled diving birds. The previous use of loons andgrebes as traditional analogues of the hesperornithiforms has precludedother foot-propelled divers, such as cormorants and diving ducks, from

A. Bell et al. Palaeogeography, Palaeoclimatology, Palaeoecology xxx (xxxx) xxx–xxx

10

comparison. All of these birds have evolved foot-propelled divingcapabilities convergently, and so only through comparisons across alltaxa can we fully evaluate the development of foot-propelled diving inthe extinct Hesperornithiformes. The results of the PCAs conducted inthis study indicate hesperornithiforms most frequently share morpho-space with cormorants and diving ducks, and much less frequently withloons and grebes (Figs. 3–5). This implies that cormorants and divingducks are, in fact, better analogues for the hesperornithiforms in termsof hindlimb morphometrics, stemming from morphologies likely asso-ciated with foot-propelled diving. Furthermore, as noted by previousstudies (Bell and Chiappe, 2015a, 2015b), there are variations in divingcapabilities among hesperornithiform taxa, which are further high-lighted by the present study.

The analysis of specific variables in the dataset, as indicated throughparticularly high or directionally opposing loadings, can provide detailsof the similarities and differences among foot-propelled divers. First,hesperornithiforms appear to share a mediolaterally broader and dor-soplantarly flattened tarsometatarsus with cormorants (Phalacrocorax)and diving ducks (Aythya and Mergus). Second, the size of the cnemialprocess of the tibia is the most highly variable measurement in thedataset. Loons have the largest cnemial process (relative to tibialength), followed by grebes, with cormorants and diving ducks sharingsimilarly shortened processes. While discerning a trend in hesper-ornithiforms is difficult, they appear to be intermediate to grebes andcormorants/diving ducks. Third, variation in the femur is very lowacross all taxa, as exhibited in the high degree of overlap on the PCplots (Fig. 3–5) as well as the low loadings on these variables (Sup-plementary Table 2). Such minimal variation suggests the existence ofmorphofunctional constraints in the femur that may be related to foot-propelled diving, however expansion of the dataset to include non-diving birds would be necessary to confirm this. While the relativeproximal extent of the trochanter and head could not be incorporatedinto the analysis, visual comparison indicates most hesperornithiformsappear to be similar to cormorants and diving ducks. Additionally,while the patella could not be included in the PCAs due to its absence inloons, in general the shape is highly variable among these birds. Theshape of the patella of Hesperornis and Parahesperornis is most similar tothat of grebes, while Baptornis and Fumicollis are most similar to cor-morants. The enormous patella of Hesperornis and Parahesperornis re-presents an increase in attachment area for key propulsive muscles,indicating what truly powerful swimmers the large hesperornithiformsmust have been.

When considering the results of the PCAs as well as the ecologicalconsiderations presented here, the cormorants are identified as the mostappropriate modern analogue of the extinct Hesperornithiformesoverall. However, hesperornithiforms represent a mosaic of featuresseen in other modern foot-propelled diving birds as well, and so futureanalyses must take care not to exclude these divers. Furthermore, itshould be recognized that hesperornithiforms show a considerable de-gree of variation among themselves in many adaptations to foot-pro-pelled diving. Hesperornis and Parahesperornis are particularly welladapted, to the extent that in many aspects they exceed the speciali-zations seen in modern foot-propelled divers (e.g., dramatically ex-panded patella and extreme discrepancy between the medial and lateraltrochleae of the distal phalanges of the fourth toe).

These data demonstrate that excluding some modern taxa (such ascormorants, that have long been excluded from discussions of hesper-ornithiforms) without appropriate consideration leads to an incompleteunderstanding of the ecological and functional aspects of fossil taxa,which has important implications for paleontological studies that seekto utilize modern analogues.

Supplementary data to this article can be found online at https://doi.org/10.1016/j.palaeo.2017.12.010.

Acknowledgements

The authors would like to thank Chris McDonald and the students ofProyecto Dinosaurios for all their hard work measuring bird skeletons.We would also like to thank the reviewers of the manuscript for theirhelpful comments.

References

Acosta Hospitaleche, C., Gelfo, J.N., 2015. New Antarctic findings of Upper Cretaceousand lower Eocene loons (Aves: Gaviiformes). Ann. Paleontol. 101, 315–324.

Agnolin, F.L., Egli, F.B., Chatterjee, S., Marsa, J.A.G., Novas, F., 2017. Vegaviidae, a newclade of southern diving birds that survived the K/T boundary. Sci. Nat. 104, 87–96.

Baumel, J.J., Witmer, L.M., 1993. Osteologia. In: Baumel, J.J., King, A.S., Breazile, J.E.,Evans, H.E., Vanden Berge, J.C. (Eds.), Nomina Anatomica Avium, 2nd Edition. 23.Publications of the Nuttall Ornithological Club, pp. 45–132.

Bearhop, S., Phillips, R.A., McGill, R., Cherel, Y., Dawson, D.A., Croxall, J.P., 2006. Stableisotopes indicate sex-specific and long-term individual foraging specialisation indiving seabirds. Mar. Ecol. Prog. Ser. 311, 157–164.

Bell, A., Chiappe, L.M., 2015a. A species-level phylogeny of the CretaceousHesperornithiformes (Aves: Ornithuromorpha): implications for body size evolutionamongst the earliest diving birds. J. Syst. Palaeontol. 14, 239–251.

Bell, A., Chiappe, L.M., 2015b. Identification of a new Hesperornithiform from theCretaceous Niobrara Chalk and implications for ecologic diversity among early divingbirds. PLoS ONE 10 (11), e0141690. http://dx.doi.org/10.1371/journal.pone.0141690.

Benton, M.J., Spencer, P.S., 1995. Fossil Reptiles of Great Britain. GeologicalConservation Review Series, No. 10 Chapman and Hall, London (386 p.).

Berner, D., 2011. Size correction in biology: how reliable are approaches based on(common) principal component analysis? Oecologia 166, 961–971.

Brodkorb, P., 1963. Birds from the Upper Cretaceous of Wyoming. In: Sibley, C.G.,Hickey, J.J., Hickey, M.B. (Eds.), Proceedings of the XIII International OrnithologicalCongress. American Ornithologists Union, Baton Rouge, pp. 55–70.

Brodkorb, P., 1971. Origin and evolution of birds. In: Farmer, D., King, J. (Eds.), AvianBiology. Academic Press, New York, pp. 19–55.

Chiappe, L.M., 2002. Basal bird phylogeny: problems and solutions. In: Chiappe, L.M.,Witmer, L.M. (Eds.), Mesozoic Birds: Above the Heads of Dinosaurs. University ofCalifornia Press, Berkeley, California, pp. 448–475.

Clarke, J.A., 2004. Morphology, phylogenetic taxonomy, and systematics of Ichthyornisand Apatornis (Avialae: Ornithurae). Bull. Am. Mus. Nat. Hist. 286 (179 pp.).

Cornelius, S., 1982. Wetland salinity and salt gland size in the redhead Aythya americana.Auk 99, 774–778.

Cracraft, J., 1982. Phylogenetic relationships and monophyly of loons, grebes, and he-sperornithiform birds, with comments on the early history of birds. Syst. Zool. 31,35–56.

del Hoyo, J., Elliot, A., Sargatal, J. (Eds.), 2008. Handbook of the Birds of the World.Vols. 1–13 Lynx Edicions, Barcelona.

Earnst, S., Platte, R., Bond, L., 2006. A landscape-scale model of yellow-billed loon (Gaviaadamsii) habitat preferences in northern alas. Hydrobiologia 567, 227.

Galton, P., Martin, L., 2002. Enaliornis, an early hesperornithiform bird from England,with comments on other Hesperornithiformes. In: Chiappe, L.M., Witmer, L.M. (Eds.),Mesozoic Birds: Above the Heads of Dinosaurs. University of California Press,Berkeley, California, pp. 317–338.

Giles, N., 1994. Tufted Duck (Aythya fuligula) habitat use and brood survival increasesafter fish removal from gravel pit lakes. In: Developments in Hydrobiology: AquaticBirds in the Trophic Web of Lakes. vol. 96. pp. 387–392.

Gregory, R.D., Carter, S.P., Baillie, S.R., 1997. Abundance, distribution and habitat use ofbreeding Goosanders Mergus merganser and Redbreasted Mergansers Mergus serratoron British rivers. Bird Study 44, 1–12.

Hammer, Ø., Harper, D.A.T., 2005. Paleontological data analysis. Blackwell Publishing,Oxford, UK 368 pp.

Heithaus, M.R., 2005. Habitat use and group size of pied cormorants (Phalacrocoraxvarius) in a seagrass ecosystem: possible effects of food abundance and predation risk.Mar. Biol. 147, 27–35.

Jarvis, E.D., Mirarab, S., Aberer, A.J., Li, B., Houde, P., Li, C., Ho, S.Y.W., Faircloth, B.C.,Nabholz, B., Howard, J.T., Suh, A., Weber, C.C., da Fonseca, R.R., Li, J., Zhang, F., Li,H., Zhou, L., Narula, N., Liu, L., Ganapathy, G., Boussau, B., Md, S., Bayzid, V.,Zavidovych, S., Subramanian, T., Gabaldón, S., Capella-Gutiérrez, J., Huerta-Cepas,B., Rekepalli, K., Munch, M., Schierup, B., Lindow, W.C., Warren, D., Ray, R.E.,Green, M.W., Bruford, X., Zhan, A., Dixon, S., Li, N., Li, Y., Huang, E.P., Derryberry,M., Frost Bertelsen, F.H., Sheldon, R.T., Brumfield, C.V., Mello, P.V., Lovell, M.,Wirthlin, M.P., Cruz Schneider, F., Prosdocimi, J.A., Samaniego, A.M.V., Velazquez,A., Alfaro-Núñez, P.F., Campos, B., Petersen, T., Sicheritz-Ponten, A., Pas, T., Bailey,P., Scofield, M., Bunce, D.M., Lambert, Q., Zhou, P., Perelman, A.C., Driskell, B.,Shapiro, Z., Xiong, Y., Zeng, S., Liu, Z., Li, B., Liu, K. Wu, Xiao, J., Yinqi, X., Zheng, Q.,Zhang, Y., Yang, H., Wang, J., Smeds, L., Rheindt, F.E., Braun, M., Fjeldsa, J.,Orlando, L., Barker, F.K., Jønsson, K.A., Johnson, W., Koepfli, K.-P., O'Brien, S.,Haussler, D., Ryder, O.A., Rahbek, C., Willerslev, E., Graves, G.R., Glenn, T.C.,McCormack, J., Burt, D., Ellegren, H., Alström, P., Edwards, S.V., Stamatakis, A.,Mindell, D.P., Cracraft, J., Braun, E.L., Warnow, T., Jun, W., Gilbert, M.T.P., Zhang,G., 2014. Whole-genome analyses resolve early branches in the tree of life of modernbirds. Science 346, 1320–1331.

Johansson, L.C., Lindhe-Norberg, U.M., 2001. Lift-based paddling in diving grebe. J. Exp.

A. Bell et al. Palaeogeography, Palaeoclimatology, Palaeoecology xxx (xxxx) xxx–xxx

11

Biol. 204, 1687–1696.Jolicoeur, P., 1963. The multivariate generalization of the allometry equation. Biometrics

19, 497–499.Kato, A., Ropert-Coudert, Y., Gremillet, D., Cannell, B., 2006. Locomotion and foraging

strategy in foot-propelled and wing-propelled shallow-diving seabirds. Mar. Ecol.Prog. Ser. 308, 293–301.

Kear, J. (Ed.), 2005. Ducks, Geese and Swans: Species Accounts (Cairina to Mergus). vol.2 Oxford University Press (915 pp.).

Kovacs, C.E., Meyers, R.A., 2000. Anatomy and histochemistry of flight muscles in a wing-propelled diving bird, the Atlantic Puffin, Fratercula arctica. J. Morphol. 244,109–125.

Livezey, B.C., 1989. Flightlessness in Grebes (Aves, podicipedidae): its independent evo-lution in three genera. Evolution 43, 29–54.

Lovvorn, J.R., Liggins, G.A., Borstad, M.H., Calisal, S.M., Mikkelsen, J., 2001.Hydrodynamic drag of diving birds: effects of body size, body shape and feathers atsteady speeds. J. Exp. Biol. 204, 1547–1557.

Maloiy, G.M.O., Alexander, R.M., Njau, mR., Jayes, A.S., 1979. Allometry of the legs ofrunning birds. J. Zool. 187, 161–167.

Marquiss, M., Duncan, K., 1997. Seasonal switching between habitats and changes inabundance of Goosanders Mergus merganser within a Scottish river system. Wild 45,198–208.

Marsh, O.C., 1872. Description of Hesperornis regalis, with notices of four other newspecies of Cretaceous birds. Am. J. Sci. Arts 3, 1–7.

Marsh, O.C., 1880. Odontornithes: A Monograph on the Extinct Toothed Birds of NorthAmerica. Government Printing Office, Washington, pp. 312.

Martin, L., Tate, J., 1976. The skeleton of Baptornis advenus (Aves: Hesperornithiformes).Smithson. Contrib. Paleobiol. 27, 35–66.

Martin, L.D., Kurochkin, E.N., Tokaryk, T., 2012. A new evolutionary lineage of divingbirds from the Late Cretaceous of North America and Asia. Palaeoworld 21, 59–63.

Mayr, E., Anderson, D., 1951. A classification of recent birds. Am. Mus. Novit. 1496,1–42.

Olson, S.L., 1992. Neogaeornis wetzeli Lambrecht, a Cretaceous loon from Chile (Aves:Gaviidae). J. Vertebr. Paleontol. 12, 122–124.

Osi, A., Barrett, P., 2011. Dental wear and oral food processing in Caiman latirostris:analogue for fossil crocodylians with crushing teeth. N. Jb. Geol. Paläont. (Abh.) 261,201–207.

Osnas, E.E., 2003. The role of competition and local habitat conditions for determiningoccupancy patterns in Grebes. Waterbirds 26, 209–216.

Owre, O.T., 1967. Adaptations for locomotion and feeding in the Anhinga and theDouble-Crested Cormorant. Ornithol. Monogr. 6, 1–138.

Panteleyev, A.V., Popov, E.V., Averianov, A.O., 2004. New record of Hesperornis rossicus(Aves, Hesperornithiformes) in the Campanian of Saratov Province, Russia.Paleontol. Res. 8, 115–122.

Prum, R.O., Berv, J., Dornburg, A., Field, D.J., Townsend, J.P., Moriarty Lemmon, E.,Lemmon, A.R., 2015. A comprehensive phylogeny of birds (Aves) using targeted next-generation DNA sequencing. Nature 526, 569–578.

Remsen, J.V., Binford, L.C., 1975. Status of the Yellow-billed Loon (Gavia adamsii) in thewestern United States and Mexico. Western Birds 6, 7–20.

Reynaud, F.N., 2005. Functional morphology of the hindlimbs of Hesperornis regalis: acomparison with modern diving birds. Geol. Soc. Am. Abstr. Programs 37, 133.

Rudstam, L., VanDeValk, A., Adams, C., Coleman, J., Forney, J., Richmond, M., 2004.Cormorant predation and the population dynamics of walleye and yellow perch inOneida Lake. Ecol. Appl. 14, 149–163.

Schonberg, C., Tapanila, L., 2006. The bioeroding sponge Aka paratypica, a modern tra-cemaking analogue for the Paleozoic Ichnogenus Entobia devonica. Ichnos 13,147–157.

Shufeldt, R.W., 1884. Concerning some of the forms assumed by the patella in birds. Proc.U.S. Nat. Mus. 7, 324–331.

Simpson, G.G., 1980. Fossil birds and evolution. In: Campbell, K.E. (Ed.), Papers in avianpaleontology honoring Hildegarde Howard. vol. 330. Contributions in Science:Natural History Museum of Los Angeles Co., pp. 3–8.

Stolpe, M., 1935. Colymbus, Hesperornis, Podiceps: ein Vergleich ihrer hinteren extremitat.Journal fur Ornithologie 83, 115–128.

Storer, R.W., 1958. Evolution in the diving birds. In: Proceedings of the 12th InternationalOrnithological Congress. vol. 2. pp. 694–707.

Suter, W., 1995. Are Cormorants Phalacrocorax Carbo wintering in Switzerland ap-proaching carrying capacity? An analysis of increase patterns and habitat choice.Ardea 83, 255–266.

Thompson, D.W., 1890. On the systematic position of Hesperornis. In: Studies from theMuseum of Zoology in University College, Dundee. vol. 1. pp. 97–111.

Tokaryk, T., Cumbaa, S., Storer, J., 1997. Early Late Cretaceous birds from Saskatchewan,Canada: the oldest diverse avifauna known from North America. J. Vertebr.Paleontol. 17, 172–176.

Torrence, S.M., Butler, M.G., 2006. Spatial structure of a diving duck (Aythya, Oxyura)guild: how does habitat structure and competition influence diving duck habitat usewithin northern prairie wetlands? Can. J. Zool. 84, 1358–1367.

Uchman, A., Hu, B., Wang, Y., Song, H., 2011. The trace fossil Diplopodichnus from thelower Jurassic lacustrine sediments of Central China and the isopod Armadillidiumvulgare (Pillbug) Lebensspuren as its recent analogue. Ichnos 18, 147–155.

Ulfvens, J., 1988. Nest characteristics and nest survival in the horned grebe Podicepsauritus and great crested grebe Podiceps cristatus in a Finnish archipelago. Ann. Zool.Fenn. 25, 293–298.

Watanuki, Y., Takahashi, A., Daunt, F., Wanless, S., Harris, M., Sato, K., Naito, Y., 2005.Regulation of stroke and glide in a foot-propelled avian diver. J. Exp. Biol. 208,2207–2216.

Wilcox, H.H., 1952. The pelvic musculature of the loon, Gavia immer. Am. Midl. Nat. 48,513–557.

Zeffer, A., Lindhe Norberg, U., 2003. Leg morphology and locomotion in birds: require-ments for force and speed during ankle flexion. J. Exp. Biol. 206, 1085–1097.

Zinoviev, A.V., 2010. Notes on the hindlimb myology and syndesmology of the Mesozoictoothed bird Hesperornis regalis (Aves: Hesperornithiformes). J. Syst. Palaeontol. 9,65–84.

A. Bell et al. Palaeogeography, Palaeoclimatology, Palaeoecology xxx (xxxx) xxx–xxx

12