Early Angiosperm Ecology: Evidence from the Albian-Cenomanian of Europe

Abstract

• Background and Aims The mid-Cretaceous is a period of sudden turnover from gymnosperm to angiosperm-dominated floras. The aim was to investigate the fossil plant ecology in order to follow the spread of angiosperm taxa.

• Methods Floristic lists and localities from the latest Albian-Cenomanian of Europe are analysed with Wagner's Parsimony Method, a clustering method currently used in phylogeny (cladistics).

• Key Results Wagner's Parsimony Method points out that (a) gymnosperms dominated brackish water-related environments while angiosperms dominated freshwater-related environments (e.g. swamps, floodplains, levees, channels), (b) angiosperms showed the highest diversity in stable, freshwater-related environments, (c) a single angiosperm, ‘Diospyros’ cretacea, is restricted to brackish water-related environments and (d) the families Lauraceae and Platanaceae were exclusive to disturbed, braided river environments, implying a opportunist strategy for early tree angiosperms.

• Conclusions During the Mid-Cretaceous, European floras were characterized by (a) coastal gymnosperms, (b) highly diversified fluvial angiosperms and (c) the first European brackish water-related angiosperm.

INTRODUCTION

The Mid-Cretaceous is a key period in angiosperm evolution, marking the dawn of their rise to dominance. The first angiosperm-dominated floras occurred during the latest Albian and Cenomanian (Lidgard and Crane, 1990), only leaving the record of a ‘hollow’ turnover between the Lower and Upper Cretaceous when successfully competing in environments previously dominated by gymnosperms. Thus, knowledge of Mid-Cretaceous palaeoenvironments and plant biocoenoses are essential for understanding early angiosperm origin(s) and evolution. Many hypotheses concerning early angiosperm ecology have been proposed: (a) woody Magnoliids (Thorne, 1976), (b) weedy xeric shrub or riparian weeds (Stebbins, 1965; Hickey and Doyle, 1977), (c) ruderal (i.e. opportunist) herbs or shrubs (Taylor and Hickey, 1996), (d) aquatic (Sun et al., 2002) or (e) disturbed, understorey shrubs (Feild et al., 2004).

In Europe, numerous localities were prospected and worked out over two centuries. However, only a single or a few localities from the same area were studied at one time (e.g. Uličný et al., 1997; Nguyen Tu et al., 1999, 2002; Gomez et al., 2001, 2002), and no synthesis on a European scale is available for the Mid-Cretaceous. Wagner's Parsimony Method (WPM), a clustering method currently used in phylogeny (cladistics), allows a hierarchical classification of the localities to be processed from floristic lists of fossil plant megaremains. The result is compared with the palaeoenvironmental data in order to infer the relationship between the fossil record (biocoenoses) and the palaeoenvironments (biotopes).

MATERIALS AND METHODS

The database consists of a set of floristic lists of fossil plant impressions and compressions ordered according to a ‘presence/absence’ per locality matrix (see the Appendix). It is compiled from the Albian and Cenomanian literature of Europe, especially from the Czech Republic (Frič and Bayer, 1901; Knobloch, 1971, 1999; Kvaček J., 1992; Uličný et al., 1997; Kvaček and Dilcher, 2000; Nguyen Tu et al., 2002), France (Nguyen Tu et al., 1999; Gomez et al., 2004; Néraudeau et al., 2005), Italy (Pigozzo, 2002; Gomez et al., 2003), Portugal (Teixeira, 1948) and Spain (Gaibar-Puertas, 1962; Alvárez-Ramis and Meléndez, 1971; Alvárez-Ramis et al., 1981; Alvárez-Ramis and de Lorenzo, 1982; Román-Gómez, 1985, 1987; Gomez et al., 1999, 2000, 2001, 2002). A floristic list from a place bearing ill-defined fossil plant beds are not considered in the analysis. Otherwise, each well-defined fossil plant bed in a given place corresponds to a distinct locality (Fig. 1). This is in order to contrast environments both in time and space. Leaf megaremains cannot endure long-distance transport without suffering angular tears along the veins or entrapment by riparian vegetation or banks (Gastaldo et al., 1989; Martín-Closas and Gomez, 2004). No such damage is described in the literature and the fossil assemblages used are generally considered to be autochthonous or parautochthonous deposits. In these circumstances, one may assume that the growing and final depositional environments were very similar on a working landscape ecology scale. The localities were grouped according to the species content with WPM, which previously had only been applied to synecology (for a review, see Coiffard et al., 2004). The early-mid Albian localities are used as outgroup, the floristic changes (total species turnover) between the Albian and Cenomanian being more significant than the coeval locality variations. The resulting consensus tree was subsequently compared with the taphonomical data. Most parsimonious trees and the resulting strict consensus tree were constructed with the software Paup 4·0 b2 for PC (Swofford, 1999).

Fig. 1.

Geographical location of the localities studied.

RESULTS

Groups and subgroups

The consensus tree of localities has a length of 238 steps, a consistency index of 0·5672 and a rescaled consistency index of 0·3132 (Fig. 2). It shows a four-branch polytomy at the base, of which branches are the highest hierarchical level of the distribution of taxa in the localities. Group A comprises three localities, which are characterized by one Bennettite, Nilssoniopteris pecinovensis, and two conifers, Brachyphyllum squamosum and Sphenolepis pecinovensis. Group B includes nine localities, of which the floral content includes one ginkgo, Nehvizdya/Eretmophyllum sp., one conifer, Ceratostrobus sequoiaphyllus, and one angiosperm, Diospyros cretacea. Group C shows five localities, and is characterized by one fern, Raphaelia lobifolia, one conifer, ‘Widdringtonia’ graminea, and one angiosperm, ‘Myrica’ fragiliformis. The fourth main branch is subdivided into three groups, D–F, because the localities within these groups share numerous taxa, whereas a few taxa are shared by the whole branch. Groups D, E and F share Myrtophyllum geinitzii, while D and E are clustered and share Myrtophyllum angustum. Group D consists of seven localities, and displays one fern, Onychiopsis capsulifera and three angiosperms, Hederaephyllum primordiale, Araliphyllum formosum and ‘Magnolia’ amplifolia. Group E represents six localities, and is composed of one angiosperm, Araliphyllum daphnophyllum. The group F comprises four localities each showing two angiosperms, Grevilleophyllum constans and Platanus laevis. Four species are frequent in two unrelated groups: (1) a Taxodiaceae, Cunninghamites lignitum in A and C; (2) a Cheirolepidiaceae, Frenelopsis alata in B and C; (3) an angiosperm, Debeya coriacea in C and E; and (4) a fern, Phlebopteris dunkeri in C and D.

Fig. 2.

Consensus tree of the floristic lists from the Albian-Cenomanian of Europe showing six groups of localities (A, estuary mouths; B, salt-marshes; C, backswamps; D, floodplains; E, levees; F, braided rivers).

DISCUSSION

Mid-Cretaceous environments

From the sedimentological point of view, all the localities of Group A, except ‘Velké Opatovice’, were interpreted as a shallowing-upward infill of a tide-dominated estuary mouth (Uličný et al., 1997). There is no accurate sedimentological study for ‘Velké Opatovice’. So, A may represent the vegetation of more or less sandy estuary mouths, which are characterized by coarse sandstones, frequent disturbance (channel displacements) and brackish conditions.

Group B includes localities characterized by dark clays, marine plankton or glauconite, indicating brackish environments. In the most accurate studies, these localities are considered to be an internal estuary (Gomez et al., 2004), lagoon (Nguyen Tu et al., 1999) or salt-marsh (Uličný et al., 1997; Nguyen Tu et al., 2002). As a consequence, one may assume that group B is salt-marsh vegetation under stable brackish water-related conditions.

Although group C lacks accurate palaeoenvironmental description, Quasisequoia crispa and Cunninghamites lignitum lived in a wet and stable backswamp assemblage (Kvaček, 2001).

Groups D–F correspond to fluvial environments. Groups D and E consist of beds of laminated clays intercalated in fluvial sandstones without brackish water-related environment indicators. Amongst them, localities Brník (Nguyen Tu et al., 2002), Praha Mala Chuchle, Vyšehorovice (Kvaček and Dilcher, 2000) and Puy Puy (Gomez et al., 2004) are considered to be floodplain or levee deposits. The localities of E differ from D by the presence of muscovite and coarser clasts that may reflect the difference between floodplains s.s. (D) and levees (E). Flood events are more frequent in levees E, and these disturbances may limit the vegetation growth. Additionally, in the levees the coarser clastic material may result in higher drainage. Most localities in groups F may reflect coarse-grained, braided river channel infills (Uličný et al., 1997). However, there is no accurate sedimentological study for ‘Hloubetín brown claystones’. Group F appears to correspond to disturbed, frequently flooded, freshwater-related environments.

In addition, the close similarity between the living and final depositional environments is supported by the strong correlation between plant assemblages and sedimentary context. Thus it would be surprising to have so many similar plant communities living in very dissimilar biotopes that might fossilize in similar deposits.

Environmental factors

The relationship between palaeoenvironments and plant assemblages deals with wide space and time scales corresponding to the landscape ecology or ‘eco-complexes’ sensu Dajoz (1996). Thus each leaf assemblage may reflect environmental mosaics juxtaposed in the same geomorphological unit (e.g. braided river environments ranging from disturbed sandy banks to ‘hardwood’ forests). Nowadays, such environmental mosaics take place within a few metres.

The distribution of the vegetation can be explained according to two environmental parameters: (1) the ‘disturbance’, understood herein as an event destroying the vegetation and implying recolonization, and (2) the salinity. Groups A and B correspond to brackish water-related environments and contrast to the remainder. Within the brackish water-related environments, the estuary mouths (A) differ from the salt-marshes (B) by more frequent disturbances due to displacements of the estuary channels. Like the freshwater-related environments, the braided rivers (F) were characterized by very frequent disturbances related to the channel displacements; the floodplains (D and E) underwent less frequent flood disturbance, especially concerning the levees (E). The backswamp environments (C) were only disturbed by the long-term meandering of the streams. These parameters allow for an easy comparison with the classification by Grime (1977, 1979), in which (a) braided rivers correspond to disturbed environments favouring ruderal (opportunist) strategies; (b) salt-marshes are constrained environments characterized by low water availability and require tolerator strategies; and (c) floodplains and backswamps are mesic environments that enhance competitor strategies. Estuary mouths and levees are intermediate environments between constrained and disturbed, and disturbed and mesic environments, respectively.

Ecology and environments

The most diverse assemblages are stable, freshwater-related environments (backswamps and floodplains) including approx. 12–14 taxa (on average per groups), whereas disturbed and/or brackish water-related environments show only approx. four to eight taxa (Fig. 3). Salinity appears to be the most relevant parameter to distinguish the distributions of the gymnosperms s.l. and angiosperms (Fig. 3). The gymnosperms s.l. are the most flourishing in the brackish water-related environments, whereas the angiosperms predominated in the freshwater-related environments. Furthermore, angiosperms and gymnosperms s.l. are nearly exclusive to the disturbed environments (braided rivers and estuary mouths, respectively) probably because of the higher competition during recolonization.

Fig. 3.

Block diagram showing the means of the numbers of species of the main component taxa (e.g. angiosperms) in each of the environments (n = number of localities per environment).

Otherwise, autecological variations occur at a lower taxonomic level. Amongst gymnosperms (Fig. 4), the Bennettitales, Cycadales, Cheirolepidiaceae and Ginkgoales were restricted to the brackish water-related environments, while Taxodiaceae occupied nearly all environments except the braided rivers. According to the classification by Grime (1977, 1979), they might have been tolerator, while Taxodiaceae might have been competitor and tolerator. Living Taxodiaceae still include competitors, such as Sequoia Endl., and tolerators, such as Taxodium L. C. Rich. Angiosperms played three main strategies (Fig. 4): (1) the bulk of the angiosperms occupied the stable, freshwater-related environments in association with the ferns, which may reflect a competitor strategy or, more probably, a shade-tolerant strategy; (2) ‘Diospyros’ cretacea might have been one of the first brackish and tolerator angiosperms in Europe; (3) Lauraceae (sensu Upchurch and Dilcher, 1990; Kvaček, 1992) and Platanaceae highly dominated in disturbed, freshwater-related environments and acted more or less like ruderal (opportunist) trees. Living Platanaceae share a similar ecology, and offer a ‘striking example of habitat fidelity’ (Wing and Boucher, 1998). Cenomanian Lauraceae are also well-developed in stable, freshwater-related environments and, at least in Europe, appears to have experienced a radiation during the Mid-Cretaceous (Kvaček and Ecklund, 2003).

Fig. 4.

Scheme of the main ecology and environment within an ideal floodplain during the Mid-Cretaceous in Europe. Dark blue, main stream and water bodies; red, group A, estuary mouth (Bennettites and conifers); orange, group B, salt-marshes (ginkgoes, conifers and Diospyros cretacea); yellow, group C, backswamps (ferns, conifers and angiosperms); green, groups D and E, meandering river with levees (conifers and angiosperms) and floodplain (ferns, conifers and angiosperms); light blue, group F, braided river (Platanaceae and Lauraceae).

Evolutionary trends

The dominance of Ginkgoales and Cheirolepidiaceae in brackish water-related environments is striking, and they could correspond to refuge niches and/or free angiosperm niches (i.e. absence of salt-tolerator angiosperm trees). Conifers also occurred in stable, freshwater-related environments, in which Taxodiaceae took the part of the overstorey.

The angiosperms showed the highest morphological and taxonomic diversity in stable, freshwater-related environments, where they consisted of probable eudicots bearing compound leaves, Lauraceae, and simple toothed leaves similar to ANITAs (i.e. five groups of basal angiosperms, Amborella, Nymphaeales, Illiciales, Trimaniaceae, Austrobaileyaceae) and Magnoliids. Such a palaeoecological situation fits well with the ‘dark and disturbed early angiosperms’ hypothesis by Feild et al. (2004). Such vegetation also occurred in North America, and was composed of Magnoliids (Liriophyllum and Magnoliaephyllum), Sapindopsis and ferns (Retallack and Dilcher, 1981, 1986). Nevertheless, in contrast to the Northern American floras, Europe lacked Magnoliales such as Liriophyllum (Kvaček and Dilcher, 2000). The angiosperm ‘Diospyros’ cretacea displays thick cuticles and cyclocytic stomata (Kvaček, 1983) and only occurred within gymnospermous, brackish water-related environments (in contrast to other coeval, saline-tolerant angiosperms), suggesting particular water loss adaptations for saline environments and a wide early ecological range. Otherwise, angiosperms are nearly exclusive to disturbed, freshwater-related environments (i.e. braided rivers). This might indicate a ruderal (opportunist) strategy (Stebbins, 1965; Taylor and Hickey, 1996), though the Lauraceae and Platanaceae species that grew there do not belong to angiosperm basal clades (APG, 2003). Similarly, channel environments in North America show platanoid-dominated vegetation (Doyle and Hickey, 1976; Retallack and Dilcher, 1981; Upchurch et al., 1994). Lauraceae and Platanaceae wood (Paraphyllantoxylon and Icacinoxylon, respectively) also occur, and even large trunk fragments were collected (Falcon-Lang et al., 2001). Bond (1989) suggested a very convenient hypothesis to account for the early dominance of woody angiosperms in disturbed environments: seedlings of living angiosperm trees grow faster than those of living conifer trees (Becker, 2000). The appearance of the fast seedling character, in association with the tree habit, may help the Lauraceae and Platanaceae to compete with gymnosperm trees such as the Taxodiaceae in disturbed habitats. In addition, the good representation of the Lauraceae in stable, freshwater-related environments may reflect an ‘early-successional habit’. Such a habit was suggested for Magnoliaephyllum (Retallack and Dilcher, 1981) and Myrtophyllum (Uličný et al., 1997). The radiation of Lauraceae in Cenomanian, which is stated by Kvaček and Ecklund (2003), may be related to rapidity of seedling growth and tree habit, making the family more competitive during regeneration.

APPENDIX

Data matrix

Outgroup: 1, Belas1 (Alvárez-Ramis and Meléndez, 1971); 2, Belas2 (Teixeira, 1948); 3, Buarcos lower (Teixeira, 1948); 4, Buarcos mid (Teixeira, 1948); 5, Buarcos upper (Teixeira, 1948); 6, Caixaria (Teixeira, 1948); 7, Pola de Siero (Alvárez-Ramis and Meléndez, 1971); 8, Préjano (Román-Gómez, 1985, 1987); 9, Rubielos de Mora (Gomez et al., 1999, 2000, 2001, 2002); 10, Runa (Alvárez-Ramis and Meléndez, 1971); 11, Tavarede (Teixeira, 1948).

Ingroup: 12, Font Benon, Les Nouillers (Gomez et al., 2004); 13, Bohdankov (Frič and Bayer, 1901); 14, Brník (Nguyen Tu et al., 2002); 15, Evan (Knobloch, 1971); 16, Hloubetín (Nguyen Tu et al., 2002); 17, Hloubetín brown claystone (Kvaček J, 1992); 18, Horousany (Nguyen Tu et al., 2002); 19, Landsberg (Frič and Bayer, 1901); 20, Le Brouillard, Ecouflant (Nguyen Tu et al., 1999); 21, Lipenec (Knobloch, 1971); 22, Pecínov bei Nove, Pecínov (Knobloch, 1999); 23, Pecínov bei Nove lower, Pecínov (Knobloch, 1971); 24, Pecínov bei Nove middle, Pecínov (Knobloch, 1971); 25, Pecínov unit 1, Pecínov (Uličný et al., 1997); 26, Pecínov unit 2, Pecínov (Uličný et al., 1997); 27, Pecínov unit 2b, Pecínov (Nguyen Tu et al., 2002); 28, Pecínov unit 3, Pecínov (Uličný et al., 1997); 29, Pecínov unit 3A, Pecinov (Uličný et al., 1997); 30, Pecínov unit 3B, Pecínov (Uličný et al., 1997); 31, Pecínov unit 5, Pecínov (Nguyen Tu et al., 2002); 32, Pecínov unit 5b, Pecínov (Nguyen Tu et al., 2002); 33, Praha Klicov, Praha (Knobloch, 1971); 34, Praha Mala Chuchle, Praha (Kvaček and Dilcher, 2000); 35, Praha Slivenec, Praha (Knobloch, 1971); 36, Puy-Puy, Tonnay-Charente (Gomez et al., 2004); 37, Renardière, Tonnay-Charente (Gomez et al., 2004); 38, Rudka (Knobloch, 1971); 39, Rudka2. (Knobloch, 1999); 40, Touchovice (Knobloch, 1971); 41, Velké Opatovice (Knobloch, 1971); 42, Vyšehorovice 1 (Knobloch, 1971); 43, Vyšehorovice 2 (Knobloch, 1971); 44, Zbrasin (Knobloch, 1971); 45, Zbraslavec (Knobloch, 1971).

Table 1.