growth interruptions in silici¢ed conifer woods from the upper

Growth interruptions in silici¢ed conifer woods from theUpper Cretaceous Two Medicine Formation, Montana, USA:implications for palaeoclimate and dinosaur palaeoecology

Howard J. Falcon-Lang �

Department of Earth Sciences, University of Bristol, Bristol BS8 1RJ, UK

Received 15 February 2002; accepted 23 June 2003

Abstract

Taphonomic studies suggest that most dinosaur bone assemblages in the Upper Cretaceous (Campanian) TwoMedicine Formation of Montana originated from drought-induced mass mortalities on the floodplain. Thishypothesis is tested further by studying intra-annual growth patterns in silicified woods of cupressaceous/taxodiaceousconifers found in association with dinosaur bone material at a palaeolatitude of 48‡N. True growth rings arecompletely absent, but growth interruptions characterised by concentric, variably persistent zones, 1^8 latewood cellswide are common. Interruptions are irregularly spaced so that, for example, four interruptions may occur along aradius of only 25 cells in one specimen, whilst in another specimen a radius more than 650 cells wide may completelylack interruptions. Irregular growth patterns like this occur in Upper Cretaceous woods across a broad latitudinal beltin the US Western Interior demonstrating that they resulted from regional climatic forcing rather than localisedphenomena. Growth interruptions record the occurrence of short-term, aperiodic disturbances to tree growth, underan equable, megathermal climate that was otherwise favourable for continuous, year-round cambial activity. The onlyprobable cause of growth interruption formation were drought-induced fluctuations in the water table and resultantwater stress. Growth patterns suggest that sufficient water for uninhibited growth was continuously available for yearsat a time, but when water stress did occur, it did so repeatedly over a very few months. Comparison with present-dayEast Africa, where trees show very similar growth patterns, implies that the equable, but erratically humid,environment of the Late Cretaceous Western Interior was exactly the kind of setting where mass mortalities amongstmegafauna might be expected. Like East Africa, this environment would have had sufficient carrying capacity topermit large herds of megafauna to develop under normal conditions, but when repeated phases of droughts struck,such populations would have been highly vulnerable.= 2003 Elsevier B.V. All rights reserved.

Keywords: Cretaceous; Campanian; Montana; dinosaur; conifer wood; drought

1. Introduction

The Upper Cretaceous Two Medicine Forma-tion of northwest Montana, USA, ranks amongstthe world’s most productive dinosaur-bearing for-

0031-0182 / 03 / $ ^ see front matter = 2003 Elsevier B.V. All rights reserved.doi:10.1016/S0031-0182(03)00539-X

* Tel. : +44-117-954-5496.E-mail address: [email protected]

(H.J. Falcon-Lang).

PALAEO 3187 18-9-03

Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 299^314

www.elsevier.com/locate/palaeo

mailto:[email protected]

http://www.elsevier.com/locate/palaeo

mations (Rogers, 1997). Dinosaur remains havebeen described from this £uvio^deltaic unit sincethe beginning of the twentieth century (Gilmore,1914, 1917), and at present at least 14 genera ofdinosaurs are known from the Formation (Rog-ers, 1990; Varricchio and Horner, 1993; Samp-son, 1995). The most spectacular discoverieshave been of a number of dinosaur nests whichpreserve both hatched and unhatched eggs con-taining intact embryos, and associated bones ofjuvenile hatchlings of the hadrosaur Maiasaura(Horner and Makela, 1979; Horner, 1982; Hor-ner and Currie, 1994). In addition, a large numberof low-diversity bonebeds dominated either by ha-drosaurs, lambeosaurs or ceratopsians have beenexcavated (Gilmore, 1917; Rogers, 1990; Varric-chio and Horner, 1993), together with one higher-diversity bone assemblage containing both igua-nodontoids and theropods (Varricchio, 1995).Many of these bonebeds are parautochthonousas indicated by the degree of skeletal articulation(Varricchio and Horner, 1993).

These bone assemblages have provided tantalis-ing insights into many aspects of dinosaur ecologyranging from their social structure (Horner andMakela, 1979; Horner, 1982) to their thermoreg-ulation (Barrick et al., 1996). Lately, there hasbeen considerable interest in the taphonomy ofthis material, and the palaeoecological implica-tions of these data. Whilst some of the hadrosaurnesting sites may have been instantaneouslyburied by volcanic ash falls (Horner, 1994) orby clastic sediments from £uvio^deltaic crevassesplay events (Lorenz and Gavin, 1984), the largermajority of the bonebed material is believed torepresent the parautochthonous remains ofdrought-induced mass mortalities on the £ood-plain (Rogers, 1990; Varricchio and Horner,1993; Varricchio, 1995). The evidence fordrought-induced mass death is fourfold (cf. Ship-man, 1975). First, much of the bone material oc-curs in the ¢ne-grained deposits of inactive riverchannels interpreted as waterholes (or oxbowlakes), left following cessation in £uvial discharge.Second, the associated sedimentary sequence con-tains calcrete palaeosols indicating a seasonallydry climate (Goudie, 1983). Third, the vertebrateassemblages contain a preponderance of sub-adult

animals, the age class most vulnerable to drought-mortality according to recent studies (Conybeareand Haynes, 1984; Rogers, 1990). Fourth, mostskeletal assemblages are pauci-speci¢c or mono-speci¢c. This is a feature common to many mod-ern vertebrate assemblages resulting fromdrought-induced mortalities related to the varyingsusceptibility of di¡erent species to adverse con-ditions (Conybeare and Haynes, 1984; Rogers,1990; Schwartz and Gillette, 1994).

Nevertheless, direct evidence for frequent andsustained droughts during the deposition of dino-saur-bearing units in the Two Medicine Forma-tion is still lacking. The best way of analysingpalaeoclimate on a su⁄ciently ¢ne scale to detectdrought events is by studying growth rings in fos-sil woods (Creber, 1977; Creber and Chaloner,1984; Francis, 1984; Falcon-Lang, 1999). Underenvironmentally favourable conditions, wood iscontinuously deposited centripetally by the vascu-lar cambium of trees. However, at the onset ofunfavourable conditions such as low winter tem-perature, drought, or insect defoliation, cambialactivity ceases, resulting in the formation of agrowth ring in the wood marked by a concentriclayer of higher density secondary xylem tissuetermed latewood (Fritts, 1976). Wood thereforerepresents an important palaeoenvironmental ar-chive, providing a continuous intra-annual recordof what growing conditions were like in the vicin-ity of the tree over a period of decades to centu-ries (Creber, 1977; Creber and Francis, 1999). Inthis paper, I describe and analyse the patterns ofgrowth seen in fragments of anatomically pre-served wood from the Two Medicine Formationto analyse the Late Cretaceous climate of north-west Montana, and to further test the hypothesisthat dinosaur skeletal accumulations resultedfrom drought-induced mass mortalities.

2. Geological setting

The Upper Cretaceous Two Medicine Forma-tion is a V600-m-thick sedimentary package, ex-posed for approximately 220 km in highlandplains to the east of Glacier National Park innorthwestern Montana (Fig. 1). It represents the

PALAEO 3187 18-9-03

H.J. Falcon-Lang / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 299^314300

most proximal part of a mollasse-type, non-ma-rine wedge deposited between the actively risingElkhorn Mountain Range to the west and themeridionally orientated Western Interior Seawayto the east, at a palaeolatitude of 48‡N (Lorenzand Gavin, 1984; Golonka et al., 1994; Smith etal., 1994). The formation was deposited in a £u-vio^deltaic setting, and is bounded by the trans-gressive shallow marine units of the underlyingVirgelle Formation and overlying Bearpaw For-mation (Varricchio, 1993; Rogers, 1998). 40Ar/39Ar radiometric dating of plagioclase and biotitein common bentonite layers above and close tothe base of the formation constrain its age be-tween 74 and 83 Ma and place it in the latterpart of the Campanian Stage of the Upper Creta-ceous (Rogers et al., 1993).

The fossil woods described in this paper werecollected from the Seven Mile Hill locality of theOld Trail Museum, 10 km south of Choteau innorthwest Montana, south of US Highway 287(Fig. 1; G. J. Dyke, pers. commun., 2001). Atthis site a 70-m-thick succession is exposed fromnear the base of the Two Medicine Formation.Three main sedimentary facies occur. First, thereare up to 10-m-thick green/grey mudstone unitscontaining root and burrow traces, rare dinosaurbones and abundant aquatic fresh/brackish watergastropods, together with rare impersistent cal-crete layers. This facies is interpreted as the de-posits of freshwater lakes or brackish bays in alower deltaic plain setting (Varricchio, 1993) withcalcrete layers indicating periods of emergenceand pedogenesis under a seasonally dry climate(Goudie, 1983). Second, there are 1^7-m-thick,erosionally-based, ¢ne- to medium-grained sand-stone lenses. These are typically only a few tens ofmetres wide, contain a coarse basal lag of re-worked calcrete nodules, mudstone rip-up clasts,dinosaur bones, permineralised wood, and mol-luscan fragments, and they ¢ne upward. This fa-cies is interpreted as the deposits of small, dis-tributary river channels on the delta plain(Varricchio, 1993). Third, 6 1-m-thick bentonitelayers and one crystal tu¡ layer occur. These rep-resent the deposits of repeated volcanic ash fallsand a single hot pyroclastic £ow, respectively,which probably originated from the nearby Elk-

horn Mountain Volcanic belt (Roberts and Hen-drix, 2000).

The unique crystal tu¡ layer occurs 105 mabove the base of the Two Medicine Formationat Seven Mile Hill and has been 40Ar/39Ar datedto 80.0 MaO0.1 my (Rogers et al., 1993). Thefossil woods discussed here occur in a channellag deposit associated with rare dinosaur bones,approximately 60 m above the crystal tu¡ (Fig. 2).The fossil woods are angular, have long-axes ofup to 5^15 cm, and a few fragments exhibit a highdegree of anatomical preservation due to silici¢-cation. Examination of growth ring boundarycurvature, which is always straight or undulating,indicates that these specimens represent fragmentsof much larger tree trunks, of at least 20^30 cm indiameter. Charred tree trunks within the crystaltu¡ unit nearby range in diameter from 20 cm to1 m (Roberts and Hendrix, 2000), implying thatthe diameter estimates of Seven Mile Hill materialmay be conservative.

Fig. 1. Location details of area studied. Northwest Montanashowing the outcrop of the Two Medicine Formation andthe position of the wood-bearing and dinosaur-bearing sites.Inset: situation of Montana; stippled area depicts region oflarger map.

PALAEO 3187 18-9-03

H.J. Falcon-Lang / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 299^314 301

3. Fossil woods

3.1. Materials and methods

Standard 6U4-cm petrographic thin sectionswere prepared for the six best-preserved woodfragments, orientated along three planes, radiallongitudinal section (RLS), tangential longitudi-nal section (TLS) and transverse section (TS).These specimens were assigned the accession num-bers TMF1^6, and are now stored in the Palae-ontology Research collection of the Departmentof Earth Sciences, Dalhousie University, Canada.The thin sections were described in detail with theaid of an Olympus BH-2 transmission light micro-scope using a combined qualitative and quantita-tive approach (Phillips, 1948; Falcon-Lang andCantrill, 2000, 2001). All specimens consisted ofpycnoxylic conifer wood belonging to a singleform taxon and were classi¢ed to generic levelusing the scheme of Krau«sel (1949).

3.2. Description

In RLS tracheids are characterised by circularbordered pits (15^18 Wm diameter) possessing a

large circular aperture (7 Wm diameter) (Fig.3A). Tracheid pitting is dominantly uniseriate(91%) with biseriate, opposite to sub-opposite pit-ting separated by bars of Sanio (9%) con¢ned tothe ends of the widest tracheids (Fig. 3A,B). Ad-jacent tracheid pits are either contiguous (63% ofinstances), spaced greater than half of one pit di-ameter apart (34%), or squashed together so thatadjacent pit borders are distorted into ovals(3%)(Fig. 3B). Rays are thin-walled, and com-posed of parenchymatous cells, 30 Wm in length,12^17 Wm in height, and 10^12.5 Wm in width.Ray tracheids are absent. Cross-¢eld pitting ischaracterised by 1^4 elliptical (7.5U5 Wm diame-ter), obliquely oriented cupressoid or taxodioidpits with elliptical apertures per ¢eld (Fig. 3C).Checking is weakly developed in some tracheids(cf. Jones, 1993).

In TLS, rays are highly abundant, being spacedonly 1^3 tracheids apart (8^12 per linear mm). Inaddition, they are almost always uniseriate, typi-cally being 1^30 cells high, although rare exam-ples are up to 58 cells high and possess shortbiseriate zones up to 5 cells in length (Fig. 3D).Latewood tracheids commonly possess abundantsmall, circular bordered pits (7.5 Wm diameter) on

Fig. 2. Sedimentary log of wood-bearing section, Seven Miles Hill, near Choteau, Montana. Crystal tu¡ is argon dated at 80 Ma(Rogers et al., 1993).

PALAEO 3187 18-9-03


their tangential walls (Fig. 3E). Axial parenchymais common, being composed of vertically elongatecells 30^40 Wm in height and 5^10 Wm diameter(Fig. 3E).

In TS, common ‘growth features’ with subtle tomore marked ring boundaries occur, and will bedescribed in detail in Section 3.3. Resin ducts areabsent, but resin-¢lled axial parenchyma is com-monly present, and randomly distributed through-out each growth increment (Fig. 3F). Rays are upto 5.1 mm in length. Tracheids are angular andrectangular in cross-section (Fig. 3F). Although

the wood is generally excellently preserved, locallydiamond-shaped cracks (1 mm long) and withlong-axes orientated radially, are present betweenadjacent ¢les of tracheids, occurring in clustersclose to the growth features (Fig. 4B).

3.3. Palaeobotanical identity

Under the classi¢cation scheme of Krau«sel(1949), these woods would be named either Cu-pressinoxylon or Taxodioxylon because both cu-pressoid and taxodioid cross-¢eld pit-types occur.

Fig. 3. Two Medicine Formation wood of Cupressinoxylon/Taxodioxylon-type. (A) Biseriate, opposite tracheid pitting exhibitingbars of Sanio, RLS, TMFc, U150. (B) Uniseriate bordered pits, RLS, TM3c, U200. (C) 3^4 cupressoid and taxodioid cross-¢eldpitting, RLS, TM3c, 375. (D) Tall, uniseriate, closely-spaced rays, RLS, TM3b, U75. (E) Axial parenchyma, TLS, TM3b, U225.(F) Uniform size rows of tracheids and closely spaced rays, TS, TMF3a, U40.

PALAEO 3187 18-9-03


The woods compare most closely with silici¢edspecimens of Taxodioxylon cryptomerioidesScho«nfeld from the Campanian Oldman Forma-tion of nearby Alberta, Canada (Ramanujan,1972), which di¡ers in its shorter rays (1^20 cellshigh). However, as ray height varies with the de-velopmental age of the wood (Falcon-Lang andCantrill, 2000), the samples may nevertheless be-long to this species. The samples are also similarto a Maastrichtian specimen of Taxodioxylondrumhellerense Ramanujan and Stewart (1969)

from Alberta, which di¡ers more signi¢cantly inhaving both taxodioid and glyptostroboid cross-¢eld pits. Another close match is with Cupressi-noxylon coloradense (Knowlton, 1917) from theMaastrichtian of Colorada, but precise compari-son is limited due to the incomplete description ofKnowlton’s specimen.

Comparison of the Two Medicine Formationmaterial with modern conifer woods indicates aclose a⁄nity with either the extant Cupressaceaeor Taxodiaceae families (Greguss, 1955, 1972),

Fig. 4. Examples of growth interruptions in Cupressinoxylon/Taxodioxylon from the Two Medicine Formation in TS. (A) Growthinterruptions are locally absent for a considerable distance, TM1a, U25. (B) Closely spaced zones of variably persistent growthinterruptions, TM6a, U12. (C) Some growth interruptions have ‘double’ boundaries, TM3a, U15. (D) Variably developed inter-ruptions, TM6a, U12. (E) The most marked growth boundary in the Two Medicine wood assemblage, TM6a, U40. (F) A typi-cal weakly developed interruption, TM6a, U15.

PALAEO 3187 18-9-03


but due to the numerous shared anatomical char-acteristics of woods of these groups, it is di⁄cultto ascertain precisely which of the two families thefossil wood specimens belong to (Phillips, 1948).This question may ultimately prove to be academ-ic because a recent revision of the Order Conifer-ales has proposed that Taxodiaceae should be ab-sorbed into Cupressaceae (Farjon, 1998).Attribution of the wood specimens to the Cupres-saceae/Taxodiaceae is also strongly supported bythe presence of taxodiaceous and cupressaceousfoliage in the Two Medicine Formation (Crabtree,1987; Wing and Boucher, 1998), and by abundanttaxodiaceous and cupressaceous pollen in its cor-relative equivalent, the Campanian Belly RiverFormation, southern Alberta (Jerzykiewicz andSweet, 1988).

3.4. Palaeo-ecosystem reconstruction

Studies of the facies distribution of plant partsin the Two Medicine Formation suggest that dis-turbed river channel margin environments werelargely colonised by high diversity communitiesof shrubby angiosperms (Crabtree, 1987), asseen in other Late Cretaceous Northern mid-lat-itude £oodplain assemblages from North Americaand Europe (Wing and Boucher, 1998; Falcon-Lang et al., 2001b). In contrast, more distal in-ter-channel regions of the £oodbasin appear tohave supported open woodlands of taxodia-ceous/cupressaceous conifers as indicated by occa-sional in situ petri¢ed forests overlain by thickmudstone units (Wolfe and Upchurch, 1987;Roberts and Hendrix, 2000). The allochthonouswood specimens described in this paper were pre-sumably sourced from one such £oodbasin wood-land.

Three lines of evidence suggest that dinosaurslived in close association with these conifer wood-lands in the Choteau region of Montana. First,the fossil wood fragments studied here occur infacies association with dinosaur bones. Second, atanother nearby site charred dinosaur bones andCupressaceae/Taxodiaceae conifer logs are foundmixed together in a crystal tu¡ near Choteaiu,recording a single biological community suddenlyoverwhelmed by a hot pyroclastic £ow (Roberts

and Hendrix, 2000). And third, but most compel-ling of all, at a third site permineralised dinosaurcoprolites associated with nesting grounds andbonebeds in the Choteau region contain the re-mains of Cupressaceae/Taxodiaceae wood, indi-cating that herbivorous dinosaurs such as Maia-saura browsed amongst the forests (Chin and Gill,1996).

4. Analysis of wood growth patterns

True growth rings, de¢ned here as being con-tinuous around the tree’s circumference and hav-ing a distinctly asymmetric boundary (Fritts,1976; Creber and Chaloner, 1984), are completelyabsent in the Two Medicine Formation woods.However, the woods do contain a large numberand wide variety of what Schweingruber (1992,1996) has termed ‘growth interruptions’ or‘growth zones’, features most commonly seen inlow latitude trees today. In order to understandthe nature of the climate under which dinosaurslived, growth patterns in the Taxodioxylon/Cu-pressinoxylon specimens were examined in trans-verse thin section. In addition to these qualitativeobservations, detailed quantitative cell-by-cellmeasurements of successive tracheid radial diam-eters (including the cell walls) were made alongthe wood radii with the aid of a graduated scalemounted in the microscope eyepiece. Five adja-cent rows of cells were measured for each speci-men, and then averaged using the method of Fal-con-Lang (1999). Mean Sensitivity statisticalanalysis was also applied to the growth interrup-tion sequences using the equation:

MS ¼ 1n31

Xt¼n31

t¼1

M2ðxtþ13xtÞxtþ1 þ xt

M

where x is the distance between adjacent growthinterruptions, n is the number of growth incre-ments in the sequence analysed, and t is theyear number of each increment. The Mean Sensi-tivity technique was developed to quantify theyear-to-year variability in sequences of truegrowth ring widths (Fritts, 1976). Whilst true an-nual rings are not present and therefore this sta-tistic is used in an unconventional manner, its

PALAEO 3187 18-9-03


application to the growth interruptions does givea good numerical measure of their spacing vari-ability, and is thus very valuable in helping todescribe the phenomenon observed.

4.1. Description of ring boundary types

The growth interruptions consist of concentric,variably persistent zones characterised by trache-ids with smaller than normal radial diameters(analogous to the latewood tracheids of truegrowth rings). Although a complete continuumof growth interruptions exist in these woods, itis useful to informally describe two end-membertypes. At one end of the spectrum are those withmore marked growth boundaries that are contin-uous around the partial circuit of the largestspecimens (15 cm), and are de¢ned by a gradualdecline in cell diameter from 40^45 Wm to 12^20 Wm over approximately 4^8 latewood cells.The subsequent reversion to 40 Wm diameter cellsoccur abruptly over 1^2 cells and gives the growthinterruption zone weakly asymmetric boundaries(Fig. 4D,E). At the other end of the spectrum arezones with much more subtle growth boundaries.In many cases, these cannot be traced laterallyaround the stem for more than about 10 cm be-fore they either fade out or wedge out to mergewith an earlier-produced growth interruption.They are de¢ned by a rapid decline in cell diam-eter from 40 Wm to approximately 20^25 Wm over1^3 latewood cells, followed by an equally rapid

rise back up to the normal tracheid diameter, andtherefore have symmetrical growth boundaries(Fig. 4B^D,F).

Two further general properties of these inter-ruptions are noteworthy. First, the smaller trache-ids that mark the interruptions do not have pref-erentially thickened cell walls (Fig. 4E). In fact,their cell walls are often slightly thinner (4^5 Wm)in comparison with those seen in the normal-sizedtracheids (5^6 Wm). Second, a concentric zone(6 10 cells wide) of unusually large diameter (upto 50^60 Wm) tracheids often occurs immediatelyafter the growth interruptions (Fig. 4B,C).

4.2. Description of ring boundary spacing

The distribution of growth interruptions in thewoody trunks is highly spatially irregular. Onlyshort continuous sequences of growth interrup-tions, ranging from 10 to 19 ring increments, areobservable in the Two Medicine Formation thinsections due to patchy silici¢cation, cellular crush-ing and their irregular occurrence. Individualgrowth interruptions are spaced between 0.12mm and 29.07 mm apart (mean 1.81 mm,n=65). Individual wood fragments possessMean Sensitivity values that range from 0.412 to0.878 with a mean value of 0.639 (n=6). Com-pared with the ring spacing in modern temperatering sequences, these values demonstrate highlyirregular growth patterns; temperate ring are con-sidered irregular (or ‘sensitive’) if year-to-year

Fig. 5. Numerical representation of growth rings in Specimen TM6a. Growth is left to right. Arrows indicate subtle growth inter-ruption boundaries.

PALAEO 3187 18-9-03


ring widths give mean sensitivities s 0.3 (Fritts,1976).

However, these statistics do not communicatethe irregular distribution of growth interruptionsas adequately as a qualitative description. Forexample, one specimen (TMF1a), almost com-pletely lacks growth interruptions altogether(Fig. 4A). It contains zones 4^29 mm wide (i.e.approximately 90^650 cells wide) composed ofrows of tracheids consistently 36^53 Wm in diam-eter, separated by very weakly developed growthinterruptions. Other specimens, e.g. TMF2a-5a,contain a large number of growth interruptionsspaced 0.2^1.3 mm apart, with boundaries oftende¢ned by two very closely adjacent interruptionsthat grade into each other (Fig. 4C). In Fig. 5,cell-by-cell growth patterns are shown for part ofa particularly well-preserved sequence in specimenTMF6a. This specimen contains a highly irregulargrowth pattern with some zones lacking interrup-tions over 150 cells whilst another shows fourinterruptions over a radius of only 25 cells (Fig.4B).

4.3. Review of regional-scaleCampanian^Maastrichtian tree growth patterns

As only a small number of wood specimenswith su⁄cient anatomical preservation to examinegrowth patterns were found at the Seven Mile Hilllocality, additional growth data from approxi-mately contemporaneous deposits across the re-gion were collected through a literature review.This approach not only increases sample sizebut also improves understanding of the extent ofthe geographical area over which similar treegrowth patterns occurred. The palaeogeographiclocation of additional wood sites was ascertainedusing Fred Ziegler’s base maps located at http://pgap.uchicago.edu/movies.html (Fig. 6)

The regional review shows that fossil coniferwoods exhibit qualitatively identical growth pat-terns to those described here in detail from theTwo Medicine Formation over a wide palaeolati-tudinal range from 34^48‡N along the margins ofthe Western Interior Seaway. For example, speci-mens of Cupressinoxylon conifer wood from theMaastrichtian Vermejo Formation of Colorado

(palaeolatitude 34‡N) contain either multiple,closely spaced growth interruptions (1^2 mmapart) de¢ned by a narrow latewood band 3^4(rarely 6) cells wide or completely lacked rings(Knowlton, 1917). Conifer wood from the Cam-panian^Maastrichtian of New Mexico, Montanaand Wyoming (palaeolatitude 35^44‡N) exhibitcontinuous uninhibited growth or with weakly de-lineated growth interruptions (Wolfe and Up-church, 1987). Finally, conifer woods from theTwo Medicine Formation itself (palaeolatitide48‡N), have previously been described brie£y aspossessing growth interruptions of highly variablethickness, closely comparable to those noted here(Crabtree, 1987).

In contrast, at lower palaeolatitudes (6 30‡N)Campanian^Maastrichtian woods from Arizonia,Utah, Idaho, and Texas exhibit continuousgrowth, completely lacking any growth ring fea-tures whatsoever (Spackman, 1948; Thayn et al.,1985; Lehman and Wheeler, 2001), whilst at high-er palaeolatitudes (s 55‡N) specimens of Cupres-sinoxylon (Penhallow, 1908) and Taxodioxylon(Ramanujan and Stewart, 1969; Ramanujan,1972) contain true growth rings, continuousaround the trunk circumference, possessing well-marked, asymmetrical ring boundaries de¢ned bynumerous latewood cells.

Fig. 6. Global palaeogeographic map (http://pgap.uchicago.edu/movies.html) showing position of Montana on the mar-gins of the Western Interior Seaway, USA, and the locationof wood-bearing sites showing growth interruptions

PALAEO 3187 18-9-03


pgap.uchicago.edu/movies.html

pgap.uchicago.edu/movies.html

http://pgap.uchicago.edu/movies.html

http://pgap.uchicago.edu/movies.html

5. Interpretation of growth patterns

In summary, growth features in the Two Med-icine Formation are quite unlike true growth ringsexhibited by most present-day temperate-zoneconifers (cf. Greguss, 1972), but bear a close re-semblance to what Schweingruber (1992) has de-scribed as growth interruptions in subtropical andtropical woods. In addition, these Late Creta-ceous growth interruptions are not geographicallylocalised phenomena but in fact characterise abroad latitudinal belt along the Western InteriorSeaway. The palaeoclimatic signi¢cance of growthinterruptions has generated much confusion anddebate (Ash and Creber, 1992), and as a preambleto interpreting the palaeoclimatic signi¢cance ofthese structures it is worth beginning by examin-ing how their nature and origin di¡ers from thatof true growth rings.

5.1. Formation of growth interruptions

Wood is produced by the vascular cambium(Creber and Chaloner, 1984). As new tracheidspass out of the cambial zone they undergo aphase of radial expansion followed by a phaseof wall thickening (Ford et al., 1978). The factorsthat control the degree to which tracheids radiallyexpand and thicken, and hence the growth pat-terns observed in the wood, are complex. Thethree most important factors in£uencing tracheidmaturation are the supply of photosynthate andgrowth promoting phytohormones, and the tugorof the cambial derivative cell (Schweingruber,1996).

Under temperate climatic conditions, fallingtemperatures and shortening photoperiod duringlate summer/early autumn inhibit the rate of pro-duction and transport of photosynthate andgrowth promoting phytohormones, at a timewhen water supply is also declining. This leadsto a reduction in the number of cells in the cam-bial zone, and causes newly formed tracheids tospend longer in the zone of cell wall thickeningrelative to the zone of cell expansion (Creber andChaloner, 1984; Uggla et al., 1998). A transitionfrom large diameter, thin-walled earlywood tra-cheids to small diameter, thick-walled latewood

tracheids is therefore seen in the latewood. Duringwinter, cambial activity completely ceases, andwhen it reactivates under optimum spring growingconditions, large earlywood tracheids are pro-duced. This process gives rise to a true growthring. True growth rings may be also formed inseasonal tropical regions with a long dry season,where drought induces long-term cambial dor-mancy (e.g. Rao and Rajput, 1999), but in con-trast these do not necessarily represent annual in-crements of growth (Duke et al., 1981; Ash, 1983;Falcon-Lang, 1999).

In environments where cambial activity can po-tentially occur continuously (e.g. in the non-sea-sonal tropics), short-term disturbances to growthmay occur which in£uence patterns of wood de-velopment in a qualitatively di¡erent way to thatmanifesting in true growth rings (Jacoby, 1989).Examples of such disturbances are droughts,£oods, ¢res, wind damage, unusually low temper-atures, synchronous coning/leaf £ushing, or in-sect-attack (Fritts, 1976; Ash, 1983; Dechamps,1984; Schweingruber, 1992, 1996; Young et al.,1993). Each disturbance has the potential to limitthe degree to which tracheids in the cambial zoneundergo expansion and wall thickening. Thegrowth interruptions that result have a symmetri-cal or weakly asymmetrical ring boundary, exhib-iting a gradual decline and subsequent gradualrise in tracheid diameter, and are produced by atemporary slowing down, but not switching o¡, ofcambial activity (Schweingruber, 1992, 1996). Inaddition, they typically fade out around the trunkcircuit because phytohormonal response to weakshort-term disturbances is imprecisely synchron-ised growth throughout an individual tree (Lar-sen, 1956). Growth interruptions are thereforedistinctly di¡erent from true growth rings withrespect to their morphology and origin; the for-mer result from a temporary slowing down ofcambial activity, the latter are produced by acomplete cessation of cambial activity (Creberand Chaloner, 1984). Growth interruptions mayalso occur in temperate woods, which contain truegrowth rings, having resulted from short-term dis-turbances during the growing season. In this con-text growth interruptions have been termed falserings (Fritts, 1976).

PALAEO 3187 18-9-03


5.2. Interpretation of Late Cretaceous growthinterruptions

The complete absence of true growth rings inthe woods of the Two Medicine Formation andadjacent coeval localities between 34 and 48‡ ofpalaeolatitude indicate that these trees grew underconditions that were potentially favourable forcontinuous cambial activity year-round. Two in-dependent strands of data support this interpreta-tion. First, the results of mid-Cretaceous com-puter climate models suggest that mean wintertemperatures for the Western Interior of theUSA were 16^20‡C, and summer mean temper-atures were 20^24‡C (Valdes et al., 1996). Second,leaf physiognomic data suggest that NW Monta-na lay in the transition zone between a non-sea-sonal megathermal climate and a weakly seasonalmesothermal climate, a region with a mean annu-al temperature close to 20‡C, and a mean annualtemperature range of only 8‡C (Wolfe and Up-church, 1987). Summer^winter temperature con-trast was probably reduced by the amelioratinge¡ect of the Western Interior Seaway (Valdes etal., 1996).

However, the presence of growth interruptionsin Western Interior fossil woods indicates thattrees growing in this megathermal environmentwere subject to repeated growth disturbances.The highly irregular spacing of the growth inter-ruptions (6 5 to s 650 cells apart; Mean Sensi-tivity 0.639) indicates that this growth disturbanceprocess entirely lacked any periodicity whatso-ever. Assuming a conservative tracheid produc-tion rate of ca. 1 cell per day (cf. Ford et al.,1978), the four interruptions closely-spaced acrossa ¢le only 25 cells wide in TMF6a may representrepeated disturbances in less than a single month.In contrast, specimen TMF1, which largely lacksinterruptions, likely represent many years of un-disturbed growth. Furthermore, the di¡ering de-grees to which the growth interruptions are devel-oped in these woods, suggest that the intensityand duration of the disturbance events were alsohighly variable.

In the Late Cretaceous Western Interior,two environmental phenomena could potentiallyhave given rise to the repeated, aperiodic, and

variably intense environmental disturbances re-corded by the growth interruptions; these aredrought (e.g. Ash, 1983) and £ooding (e.g.Young et al., 1993). However, drought is theoverwhelmingly most likely of the two causesfor four intrinsic reasons. First, growth inter-ruptions are common to woods over 14 degreesof latitude indicating that they formed in responseto regional climatic phenomena rather than beingfeatures related to the local vagaries of £oodplainhydrology. Second, the fossil woods contain fea-tures like diamond-shaped cracks and checking,similar to features known in modern woods tobe created by the wood drying out during growth(Jones, 1993; Schweingruber, 1996). Third,growth interruptions caused by £ooding are usu-ally accompanied by the development of roundedtracheid, with intercellular spaces (Hook, 1984;Falcon-Lang et al., 2001a), a feature not presentin the Two Medicine woods. Fourth, zones com-posed of unusually large diameter tracheids thatfollow many of the growth interruptions indicatethat disturbance events ended by a rapid returnto optimum growing conditions, a feature thathas been related to an increase in water availabil-ity and cellular tugor after rains (Ford et al.,1978).

Several independent lines of evidence support adrought-induced origin for the growth interrup-tions. First, angiosperm leaves in the Two Medi-cine Formation are coriaceous but lack drip trips,a physiognomy associated with seasonally dry en-vironments (Crabtree, 1987). Second, charcoal iscommon in the Two Medicine Formation, indi-cating the occurrence of ¢res and implyingdrought (Carpenter, 1987). Third, studies of Cam-panian palaeosols at a palaeolatitude of 40‡N inNew Mexico indicate the operation of a humidclimate with an annual precipitation of 750^1000 mm yr31 (Mack, 1992), whilst Maastrichtianpalaeosols at a palaeolatitude of 49‡N in Monta-na are believed to have formed under seasonallyhumid climates with a marked (but not severe)dry season and a mean annual rainfall of 900^1200 mm yr31 (Retallack, 1994; Buck andMack, 1995). Fourth, the results of non-paramet-ric climate models have implied slight rainfall sea-sonality over Late Cretaceous Montana, with

PALAEO 3187 18-9-03


summers being humid but winters being ratherdrier (Golonka et al., 1994).

However, there may not have been a simpledirect linkage between drought and growth inter-ruption formation. Droughts lead to the physio-logical water stress of trees by depressing the localwater table (Little, 1975). Irregular growth inter-ruption patterns therefore probably re£ectdrought-induced £uctuations in water table, andthis mediating process may have introduced agreater degree of aperiodicity into the growth in-terruption patterns observed, compared with theactual aperiodicity of the rainfall.

5.3. Comparison with present-day East Africangrowth patterns

The interpretation of the Two Medicine For-mation wood growth patterns as indicatingdrought-interrupted growth under a megathermalclimate is strongly supported through comparison

with woods from present-day seasonal, megather-mal regimes in East Africa. The woods of a widediversity of conifers from this region are held inthe collections of the Jodrell Laboratory at theRoyal Botanic Gardens, Kew, UK, and exhibitidentical growth patterns to those described inthis paper. For example, trunk wood of Juniperusprocera Hochst ex Endlicher from Somalia andKenya (a cupressaceous conifer related to thosestudied from the Two Medicine Formation) ex-hibits wide radii lacking any growth features(Fig. 7A) or contains irregularly spaced growthinterruptions characterised by symmetrical toweakly asymmetric growth (Fig. 7B). Other pina-ceous conifers such as Pinus oocarpa Schiede andSchltdl from Uganda contain common well-devel-oped symmetrical growth interruptions (Fig. 7C),and asymmetric growth interruptions that oftenexhibit a double boundary composed of twoclosely spaced interruptions (Fig. 7D). This latterfeature is very similar to the double interruptions

Fig. 7. Examples of growth interruptions in conifer woods from East Africa. All specimens stored at Jodrell Laboratory, RoyalBotanic Gardens, Kew. (A,B) Juniperus procera and (C,D) Pinus oocarpa. (A) Growth interruptions are locally absent, U12.(B) Irregularly spaced weak interruptions, U15. (C) Marked symmetrical growth interruption, U12. (D) Marked asymmetrical‘double’ interruptions, U15.

PALAEO 3187 18-9-03


seen in the Two Medicine sample, TM3a (cf. Fig.4C). Still other podocarpaceous conifers such asPodocarpus milanjianus Rendle and Afrocarpusmannii (Hook) Page from Tanzania and Ugandahave even more weakly developed growth featuresexhibiting only the faintest interruptions (not ¢g-ured), and then only rarely. All the East Africanconifers described grew in near-equatorial settingswhere annual temperature variability is minimaland irregular growth interruption patterns are en-tirely the result of erratic rainfall and seasonaldroughts (Jacoby, 1989).

6. Discussion

In summary, growth patterns in Late Creta-ceous woods of the Western Interior suggest theoperation of megathermal climates punctuated byerratic seasonal rainfall. The only other site docu-mented in the fossil record that contains abun-dant woods with very similar patterns of growthto those documented from the Two Medicine For-mation, and surrounding areas, is the Upper Tri-assic Chinle Formation of Arizonia, USA (Ashand Creber, 1992). Growth interruptions at thissite have been interpreted as recording eitherdrought-induced £uctuating water table or per-haps free-running endogenous rhythms (Ash andCreber, 1992). The absence of true growth rings,despite abundant independent evidence for rain-fall seasonality, has been linked to the palaeoval-ley setting of the fossil forests, which may havemaintained an arti¢cially high water table, andrendered the trees less sensitive to regional climate(Demko et al., 1998). Comparison with the ChinleFormation is highly signi¢cant because this unit,like the Two Medicine Formation, contains richmono-speci¢c or pauci-speci¢c bonebeds of dicy-nodonts and early dinosaurs believed to haveoriginated from drought-induced mass mortalities(Schwartz and Gillette, 1994; Fiorillo et al.,2000).

It is interesting that at both the CretaceousTwo Medicine Formation and the Triassic ChinleFormation sites, trees producing weakly de¢nedgrowth interruptions, and drought-induced bone-beds co-occur. I suggest that this is no coinci-

dence, and that such types of geological environ-ment actually possess the optimum conditions forcausing drought-induced mass mortalities. Woodgrowth patterns indicate that in both settings, thetrees could grow continuously for long periods,but when water stress did occur it did so repeat-edly for one to several months. During longfavourable environmental phases when wateravailability was unlimiting for many years, theecosystem carrying capacity of the region wouldhave been high, and large populations of mega-fauna could develop. However, when droughtsbegan to lower the water table, as they did sowith an erratic aperiodicity, these populationswould have become very vulnerable and massmortality would have followed.

Very similar population dynamics are seen inEast Africa amongst elephant herds where as pre-viously noted, climate is megathermal with a sea-sonal rainfall distribution (Owen-Smith, 1988).Under normal conditions rainfall is regular, andwater availability is high. However, due to theerratic vagaries of tropical climate, rainfall occa-sionally fails, the rivers dry up, and water holesbecome rarer. Megafauna that need to drinkevery few hours become restricted in their move-ments to a within a few kilometres of the remain-ing water holes. If the droughts persist, vegetationin the vicinity of the water holes is quickly com-pletely consumed, and the megafauna die of star-vation in very large numbers (Conybeare andHaynes, 1984; Haynes, 1988).

In conclusion, analysis of the growth interrup-tion patterns seen in the conifer woods of the TwoMedicine Formation and surrounding regions in-dicate the existence of a megathermal climate witherratic rainfall. Megafaunal populations wouldhave been highly unstable under such conditions,and the new data presented in this paper thereforestrongly support the hypothesis that para-autoch-thonous dinosaur bonebeds originated fromdrought-induced mass mortalities.

Acknowledgements

I gratefully acknowledge a Killam Post-doctor-al Fellowship from Dalhousie University. I thank

PALAEO 3187 18-9-03


Gareth Dyke (University College Dublin) for col-lecting the fossil woods used in this study, andPeter Gasson at the Jodrell Laboratory of theRoyal Botanical Gardens, Kew, for making avail-able modern wood thin sections for study. Thispaper has greatly bene¢ted from discussions withGeo¡ Creber (University College London) overseveral years, and from the reviews by JennyChapman (Cambridge University) and an anony-mous reviewer.

References

Ash, J., 1983. Growth rings in Agathis robusta and Araucariacunninghami from tropical Australia. Aust. J. Bot. 31, 269^276.

Ash, S.R., Creber, G.T., 1992. Palaeoclimatic interpretation ofthe wood structures of the trees in the Chinle Formation(Upper Triassic), Petri¢ed Forest National Park, Arizona,USA. Palaeogeogr. Palaeoclimatol. Palaeoecol. 96, 299^317.

Barrick, R.E., Showers, W.J., Fischer, A.G., 1996. Compari-son of Thermoregulation of four ornithischian dinosaursand a varanid lizard from the Cretaceous Two MedicineFormation: Evidence from oxygen isotopes. Palaios 11,295^305.

Buck, B.J., Mack, G.H., 1995. Latest Cretaceous (Maastricht-ian) aridity indicated by paleosols in the McRae Formation,south-central New Mexico. Cretac. Res. 16, 559^572.

Carpenter, K., 1987. Palaeoecological signi¢cance of droughtsduring the Late Cretaceous of the Western Interior. Occ.Pap. Tyrell Mus. Palaeontol. 3, 42^47.

Chin, K., Gill, B.D., 1996. Dinosaurs, dung beetles, and coni-fers: Participants in a Cretaceous food web. Palaios 11, 280^285.

Conybeare, A., Haynes, G., 1984. Observations on elephantmortality and bones in water holes. Quat. Res. 22, 189^200.

Crabtree, D.R., 1987. Angiosperms of the northern RockyMountains: Albian to Campanian (Cretaceous) megafossil£oras. Ann. Mo. Bot. Garden 74, 707^747.

Creber, G.T., 1977. Tree rings: A natural data-storage system.Biol. Rev. 52, 349^383.

Creber, G.T., Chaloner, W.G., 1984. In£uence of environmen-tal factors on the wood structure of living and fossil trees.Bot. Rev. 50, 357^448.

Creber, G.T., Francis, J.E., 1999. Fossil tree-ring analysis:Palaeodendrology. In: Jones, T.P., Rowe, N.P. (Eds.), FossilPlants and Spores: Modern Techniques. Geol. Soc. Lond.,pp. 245^250.

Dechamps, R., 1984. Evidence of bush ¢res during the Plio^Pleistocene of Africa (Omo and Sahabi) with aid of fossilwoods. In: Coetzee, J.A., Van Zinderen Bakker, E.M.(Eds.), Palaeoecology of Africa and Surrounding Islands16, Rotterdam.

Demko, T.M., Dubiel, R.F., Parrish, J.T., 1998. Plant taphon-omy in incised valleys: Implications for interpreting paleo-climate from fossil plants. Geology 26, 1119^1122.

Duke, N.C., Birch, W.R., Williams, W.T., 1981. Growth ringsand rainfall correlations in a mangrove tree of the genusDiospyros (Ebenaceae). Aust. J. Bot. 29, 135^142.

Falcon-Lang, H.J., 1999. The Early Carboniferous (Cour-ceyan^Arundian) monsoonal climate of the British Isles:Evidence from growth rings in fossil woods. Geol. Mag.136, 177^187.

Falcon-Lang, H.J., Cantrill, D.J., 2000. Cretaceous (Late Al-bian) Coniferales of Alexander Island, Antarctica, 1. Woodtaxonomy a quantitative approach. Rev. Palaeobot. Palynol.111, 1^17.

Falcon-Lang, H.J., Cantrill, D.J., 2001. Gymnosperm woodsfrom the Cretaceous (mid-Aptian) Cerro Negro Formation,Byers Peninsula, Livingston Island, Antarctica: The arbor-escent vegetation of a volcanic arc. Cretac. Res. 22, 277^293.

Falcon-Lang, H.J., Cantrill, D.J., Nichols, G.J., 2001a. Biodi-versity and terrestrial ecology of a mid-Cretaceous, high-latitude £oodplain, Alexander Island, Antarctica. J. Geol.Soc. Lond. 158, 709^725.

Falcon-Lang, H.J., Kvacek, J., Ulicny, D., 2001b. Fire ecologyand palaeoclimate of a Late Cretaceous £uvial to estuarineenvironment, Pec|¤nov Quarry, Czech Republic. Geol. Mag.138, 563^576.

Farjon, A., 1998. World Checklist and Bibliography of Coni-fers. Royal Botanic Gardens, Kew, 298 pp.

Fiorillo, A.R., Padian, K., Musikasinthorn, C., 2000. Ta-phonomy and depositional setting of the Placerias Quarry(Chinle Formation: Late Triassic, Arizona). Palaios 15, 373^386.

Ford, E.D., Robards, A.W., Piney, M.D., 1978. In£uence ofenvironmental factors on cell production and di¡erentia-tion in the earlywood of Picea stichensis. Ann. Bot. 42,683^692.

Francis, J.E., 1984. The seasonal environment of the Purbeck(Upper Jurassic) fossil forests. Palaeogeogr. Palaeoclimatol.Palaeoecol. 48, 285^307.

Fritts, H.C., 1976. Tree Rings and Climate. Academic Press,London, 576 pp.

Gilmore, C.W., 1914. A new ceratopsian dinosaur from theUpper Cretaceous of Montana, with a note on Hypacrosau-rus. Smiths. Miscell. Coll. 43, 1^10.

Gilmore, C.W., 1917. Brachyceratops, a ceratopsian dinosaurfrom the Two Medicine Formation of Montana, with noteson associated fossil reptiles. U.S. Geol. Surv. Prof. Pap. 103,1^45.

Golonka, J., Ross, M.I., Scotese, C.R., 1994. Phanerozoic pa-laeogeographic and palaeoclimatic modelling maps. In: Em-bry, A.F., Beauchamp, B., Glass, D.J.B (Eds.), Pangea:Global Environments and Resources. Can. Soc. Pet. Geol.Mem. 17, pp. 1^48.

Goudie, A.S., 1983. Calcrete. In: Goudie, A.S., Pye, K. (Eds.),Chemical Sediments and Geomorphology. Academic Press,London, pp. 93^131.

PALAEO 3187 18-9-03


Greguss, P., 1955. Identi¢cation of Living Gymnosperms onthe Basis of Xylotomy. Akademia Kiado, Budapest, 263 pp.

Greguss, P., 1972. The Xylotomy of the Living Conifers. Aka-demiai Kiado, Budapest, 501 pp.

Haynes, G., 1988. Mass deaths and serial predation: Compar-ative taphonomic studies of modern large mammal deathsites. J. Archaeol. Sci. 15, 219^235.

Hook, D.D., 1984. Adaptations to £ooding with fresh water.In: T.T. Kozlowski (Ed.), Flooding and Plant Growth. Aca-demic Press, London, pp. 265^293.

Horner, J.R., 1982. Evidence of colonial nesting and site ¢del-ity among ornithischian dinosaurs. Nature 297, 675^676.

Horner, J.R., 1994. Comparative taphonomy of some dinosaurand extant bird colonial nesting grounds. In: Carpenter, K.,Hirsch, K.F., Horner, J.R. (Eds.), Dinosaur Eggs and Ba-bies. Cambridge University Press, Cambridge, pp. 116^123.

Horner, J.R., Currie, P.J., 1994. Embryonic and neonatal mor-phology and ontogeny of a new species of Hypacrosaurus(Ornithischia, Lambeosauridae) from Montana and Alberta.In: Carpenter, K., Hirsch, K.F., Horner, J.R. (Eds.), Dino-saur Eggs and Babies. Cambridge University Press, Cam-bridge, pp. 312^336.

Horner, J.R., Makela, R., 1979. Nests of juveniles providesevidence of family structure among dinosaurs. Nature 282,296^298.

Jacoby, G.C., 1989. Overview of tree-ring analysis in tropicalregions. IAWA Bull. 10, 99^108.

Jerzykiewicz, T., Sweet, A.R., 1988. Sedimentological andpalynological evidence of regional climatic changes in theCampanian to Paleocene sediments of the Rocky MountainFoothills, Canada. Sediment. Geol. 59, 29^76.

Jones, T.P., 1993. New morphological and anatomical evi-dence for a wild¢re origin for fusain from comparisonwith modern charcoal. Spec. Pap. Palaeontol. 49, 113^123.

Krau«sel, R., 1949. Identi¢cation key for genera and species ofidenti¢able fossil conifer woods. Palaeontographica B 89,150^189.

Knowlton, F.H., 1917. Fossil £ora of the Vermejo and RatonFormations of Colorada and New Mexico. U.S. Geol. Surv.Prof. Pap. 101, 223^455.

Larsen, P.R., 1956. Discontinuous growth rings in suppressedslash pine. Trop. Woods 104, 80^99.

Lehman, T.M., Wheeler, E.A., 2001. A fossil dicotyledonouswoodland/forest from the Upper Cretaceous of Big BendNational Park, Texas. Palaios 16, 102^108.

Little, C.H.A., 1975. Inhibition of cambial activity in Abiesbalsama by internal water stress: Role of abscisic acid.Can. J. Bot. 53, 3041^3050.

Lorenz, J.C., Gavin, W., 1984. Geology of the Two MedicineFormation and the Sedimentology of a Dinosaur NestingGround. Montana Geol. Soc. Field Conf., NorthwesternMontana, pp. 175^186.

Mack, G.H., 1992. Paleosols as an indicator of climatic changeat the Early^Late Cretaceous boundary, southwestern NewMexico. J. Sediment. Petrol. 62, 483^494.

Owen-Smith, R.N., 1988. Megaherbivores: The In£uence ofVery Large Body Size on Ecology. Cambridge Univ. Press,New York, 369 pp.

Penhallow, D.P., 1908. Report on a collection of fossil woodsfrom the Cretaceous of Alberta. Ottawa Nat. 22, 82^88.

Phillips, E.W.J., 1948. Identi¢cation of softwoods by their mi-croscopic structure. For. Res. Bull. 46, 1^56.

Ramanujan, C.G.K., 1972. Fossil coniferous woods from theOldman Formation (Upper Cretaceous) of Alberta. Can. J.Bot. 50, 595^602.

Ramanujan, C.G.K., Stewart, W.N., 1969. Fossil woods ofTaxodiaceae from the Edmonton Formation (Upper Creta-ceous) of Alberta. Can. J. Bot. 47, 115^124.

Rao, K.S., Rajput, K.S., 1999. Seasonal behaviour of vascularcambium in teak (Tectona grandis) growing in moist decid-uous and dry deciduous forests. IAWA J. 20, 85^93.

Retallack, G.J., 1994. A pedotype approach to latest Creta-ceous and earliest Tertiary paleosols in eastern Montana.Geol. Soc. Am. Bull. 106, 1377^1397.

Roberts, E.M., Hendrix, M.S., 2000. Taphonomy of a petri¢edforest in the Two Medicine Formation (Campanian), north-wet Montana: Implications for palinspastic restoration ofthe Boulder Batholith and Elkhorn Mountains Volcanics.Palaios 15, 476^482.

Rogers, R.R., 1990. Taphonomy of three dinosaur bone bedsin the Upper Cretaceous Two Medicine Formation of north-western Montana: Evidence for drought-related mortality.Palaios 5, 394^413.

Rogers, R.R., 1997. Geology and taphonomy of the UpperCretaceous Two Medicine Formation, northwestern Monta-na. In: Currie, P., Padian, K. (Eds.), The Dinosaur Ency-clopedia. Academic Press, pp. 760^765.

Rogers, R.R., 1998. Sequence analysis of the Upper Creta-ceous Two Medicine and Judith River Formations, Monta-na: Nonmarine response to the Claggett and Bearpaw ma-rine cycles. J. Sediment. Res. 68, 615^631.

Rogers, R.R., Swisher, C.C., Horner, J.R., 1993. 40Ar/39Ar ageand correlation of the nonmarine Two Medicine Formation(Upper Cretaceous), northwestern Montana, U.S.A.. Can. J.Earth Sci. 30, 1066^1075.

Sampson, S.D., 1995. Two new horned dinosaurs from theUpper Cretaceous Two Medicine Formation of Montana;with a phylogenetic analysis of the Centrosaurinae (Orthi-schia: Ceratopsidae). J. Vert. Palaeontol. 15, 743^760.

Schwartz, H.L., Gillette, D.D., 1994. Geology and taphonomyof the Coelophysis Quarry, Upper Triassic Chinle Forma-tion, Ghost Ranch, New Mexico. J. Palaeontol. 68, 1118^1130.

Schweingruber, F.H., 1992. Annual growth rings and growthzones in woody plants in southern Australia. IAWA Bull.13, 359^379.

Schweingruber, F.H., 1996. Tree Rings and Environment Den-droecology. Swiss Federal Institute for Forest, Berne, 609pp.

Shipman, P., 1975. Implications of drought for vertebrate fos-sil assemblages. Nature 257, 667^668.

PALAEO 3187 18-9-03


Smith, A.G., Smith, D.G., Funnell, B.M., 1994. Atlas of Me-sozoic and Cenozoic Coastlines. Cambridge Univ. Press,Cambridge, 99 pp.

Spackman, W., 1948. A dicotyledonous wood found associatedwith Tempskya. Ann. Mo. Bot. Gard. 35, 107^116.

Thayn, G.F., Tidwell, W.D., Stokes, W.L., 1985. Flora of theLower Cretaceous Cedar Mountain Formation of Utah andColorado. Part III Icacinoxylon pittiense n. sp. Am. J. Bot.72, 175^180.

Uggla, C., Mellerowicz, E.J., Sundberg, B., 1998. Indole-3-Acetic Acid controls cambial growth in Scots Pine by posi-tional signalling. Plant Physiol. 117, 113^121.

Valdes, P.J., Sellwood, B.W., Price, G.D., 1996. Evaluatingconcepts of Cretaceous equability. Palaeoclimates 2, 139^158.

Varricchio, D.J., 1993. Montana climatic changes associatedwith the Cretaceous Claggett and Bearpaw transgressions.In: Hunter, L.D.V. (Ed.), Energy and Mineral Resourcesof Central Montana. Field Conf. Guide, Billings, pp. 97^102.

Varricchio, D.J., 1995. Taphonomy of Jack’s Birthday Site, adiverse dinosaur bonebed from the Upper Cretaceous TwoMedicine Formation of Montana. Palaeogeogr. Palaeocli-matol. Palaeoecol. 114, 297^323.

Varricchio, D.J., Horner, J.R., 1993. Hadrosaurid and lambeo-saurid bone beds from the Upper Cretaceous Two MedicineFormation of the Montana: Taphonomic and biologic im-plications. Can. J. Earth Sci. 30, 997^1006.

Wing, S.L., Boucher, L.D., 1998. Ecological aspects of the£owering plant radiation. Ann. Rev. Earth Planet. Sci. 26,379^421.

Wolfe, J.A., Upchurch, J.R., 1987. North American nonmar-ine climates and vegetation during the Late Cretaceous. Pa-laeogeogr. Palaeoclimatol. Palaeoecol. 61, 33^77.

Young, P.J., Megonigal, J.P., Sharitz, R.R., Dat, F.P., 1993.False ring formation in Baldcypress (Taxodium distichum)saplings under two £ood regimes. Wetlands 13, 293^298.

PALAEO 3187 18-9-03


growth interruptions in silici¢ed conifer woods from the upper

Documents