A LEAFCUTTER BEE TRACE FOSSIL FROM THE MIDDLE EOCENE OF PATAGONIA, ARGENTINA, AND A REVIEW OF MEGACHILID (HYMENOPTERA) ICHNOLOGY

The family Megachilidae comprises about 3000 extant species worldwide, which are included in four subfamilies: Pararhophitinae, Fideliinae, Liturginae and Megachilinae. One of the most common genera, Megachile, contains numerous species, particularly in xeric regions of the Old World (Michener 1979; Engel 1999; Baker and Engel 2006; Engel and Baker 2006). Adults are distinctive because they carry pollen in a metasomal structure called the ventral scopa. Nest building is an extensive activity among megachilids, for which females of some Megachile species use rounded leaf or petal fragments for construction, resulting in the presence of distinctive excision patterns along the margin leaves (Michener 1953, 2000; Horne 1995; Petit 2002) (Text-fig. 2A–B).

Ichnological evidence for fossil leafcutter bees has been reported by several authors, indicating that the use of leaves in nesting behaviour by these insects is old (Cockerell 1910; Berry 1931; Brooks 1955; Johnston 1993; Lewis 1994; Wappler and Engel 2003). These contributions document examples from the Northern Hemisphere that expand the past evolutionary history of the group (Text-fig. 1A). Historically, these trace fossils were described as semicircular excisions along the margin of angiosperm, and rarely Ginkgo, leaves (Lewis 1994; Labandeira 2002a, b), and were attributed to leafcutter bees because of their rounded shape as evaluated by visual inspection.

For South America, the Río Pichileufú (RP) palaeoflora of central Patagonia, from where the trace fossil in this contribution originates (Text-fig. 1B), comprises a diverse assemblage of plants. Early studies on this site were focused on plant diversity and taxonomy (Berry 1925, 1928, 1935a–c, 1938); more recently they have concentrated on geological setting, palaeoenvironment, and palaeoclimate (Romero 1978, 1986; Aragón and Romero 1984; Wilf et al. 2005a).

There have been only a few studies on fossil leaf damage in South America hitherto. Durango De Cabrera and Rodriguez De Sarmiento (1995), Rodriguez De Sarmiento and Durango De Cabrera (1995) and Rodriguez De Sarmiento et al. (1998) identified fungal attack on leaves from the Eocene Laguna del Hunco Flora. From the same locality, Wilf et al. (2005b) analysed the plant diversity and plant-insect associations, and Sarzetti et al. (2007) mentioned elevated levels of plant diversity and plant-insect associations. The only contributions that include an ichnotaxonomic analysis of plant-insect associations in Patagonia have been those of Genise (1995) and Genise and Hazeldine (1995), who identified insect borings in wood and described new ichnogenera and ichnospecies from Jurassic and Late Cretaceous localities.

The present contribution has three principal objectives. The first is to describe a new trace fossil and provide other examples of a distinctive type of plant damage attributed to leaf-cutter bees within an ichnotaxonomic context, thus contributing to our knowledge of plant-insect associations in the fossil record. The second is to provide a method, using an explicit shape analysis of ellipse eccentricity of leaf margins and holes, such that ‘circular’ trace-fossil excisions can be attributed to leaf-cutter bees by comparison with similar excisions from extant bees. The third is to place the documented occurrences in their broader spatiotemporal context of early bee biogeography and phylogeny.

Geological setting

The Río Pichileufú (RP) locality is located in the Río Negro Province of southern Argentina (Text-fig. 1B), and occurs in exposures of volcanic lake deposits of the Ventana Formation (Berry 1938; Aragón and Romero 1984). The palaeoflora comprises many leaves but also includes compressions of insects, flowers, seeds and fruits. Its age was first suggested to be Miocene (Berry 1938), but recently ⁴⁰Ar/³⁹Ar determinations from volcanic ashes (Wilf et al. 2005a) intimately associated with the flora have dated it as 47.46 ± 0.5 Ma (middle Eocene, Lutetian), indicating that the RP flora was within the initial phase of a global cooling interval, post-dating the Paleocene–Eocene Thermal Maximum (Zachos et al. 2001; Wing et al. 2005).

Palaeobotanical research at this locality by Berry (1925, 1935a–c, 1938) revealed a broad spectrum of plant taxa. Berry described approximately 130 species of plants, but his assignments are problematic today because very few of his specimens have been re-examined in the light of current palaeobotanical knowledge (Wilf et al. 2005a). Extensive new collections were made in 2002 and 2005 (Wilf et al. 2005a), especially from a single quarry known as RP3. These are housed at the Museo Paleontológico ‘Asociación Paleontológica Argentina’ in Bariloche (Río Negro Province). Observations on this material and the type collections at the National Museum of Natural History (NMNH), in Washington, DC, indicate a diverse assemblage of angiosperms and to a lesser extent, ferns and gymnosperms (Wilf et al. 2005a; Sarzetti et al. 2007).

Previously, Romero (1978, 1986) considered the RP palaeoflora to be akin to a ‘mixed flora’ of heterogeneous composition and origin, with some species adapted to either tropical or subtropical conditions and others to a more temperate climate typical of a Gondwanan or other Austral realm. Thus, Romero (1978) concluded that the flora represented a stable community lacking a modern analogue.

Observations from Recent studies, both correlative and experimental, have shown that insect diversity is often associated with plant diversity (Murdoch et al. 1972; Siemann et al. 1998; Hawkins and Porter 2003), suggesting that the richness of the RP flora may also have a positive relationship with the diversity of foliar damage types in the ichnofauna.

Material and methods

The extant species Megachile rotundata Fabricius was selected for comparison with the potential producer of the trace fossil described herein. Michener (2000) divided the genus Megachile into three groups, defined by distinctive external morphology and behaviour among subgenera. The species of Group 1 generally use leaf fragments to line cells in soil substrates or in hollows left by wood-boring beetles (Text-fig. 2C). According to Michener (2000) these species use oblong leaf fragments for the cell bases and lateral walls, and circular leaf discs for cell closures and separation between adjacent cells (Text-fig. 2D).

Nests of extant M. rotundata were used in this study to analyse shapes of leaf discs for comparison with the similar fossil excisions from RP and other traces described in the literature. Megachile rotundata was introduced to Argentina to improve the pollination of alfalfa (also known as lucerne, purple medick or trefoil: Fabaceae, Medicago sativa L.), at agricultural experimental stations (Arretz Vergara and Martinez 1988). The nests analysed were obtained from INTA (Instituto Nacional de Tecnología Agropecuaria) where the nesting bees are reared in cavities within woody substrates.

Foliar disc shape was analysed according to the following procedure. Each piece of leaf within the nests was separated using forceps and the more ‘circular’ pieces were selected, which meant selecting discs from the entrance of the cells as well as several from the base (Text-fig. 2D). These discs were subsequently mounted on paper and optically scanned to obtain a digital image analyzable by computer software. Analysis of the shape of these leaf discs, in addition to the trace fossil, was made using Corel Draw. One hundred discs of alfalfa leaves used by M. rotundata for their nests were analysed. An ellipse was redrawn over the perimeter of each disc to obtain the major and minor axes. Twelve potential ellipses were redrawn using the same methodology for the leaf from the RP site as well as previously documented trace fossils from the literature to obtain the average value of the minimum and maximum axial diameters.

These values were then used to evaluate the eccentricity of the leaf discs, using the general formula of an ellipse: a² = b² + c², where a is the major axis, b the minor axis, and c represents the distance between the centre and the focus. The eccentricity is represented by e = c/a and e = √a²–b²/a, whereby e is 0 when it is equivalent to a circle and 1 when it is equivalent to a line. Intermediate values represent ellipsoidal figures ranging from circles to ellipses. The unaided eye can distinguish more accurately between circles and ellipses when the values are, respectively, closer to 0 or 1.

Potential errors inherent to the methodology should be noted. The margin of leaf discs could inadvertently be broken when being separated manually, resulting in the lowest values of eccentricity obtained, between 0.06 and 0.10. Moreover, it was difficult to distinguish in some cells between those discs located at the base from adjacent pieces of the lateral wall that were progressively more oval. This, in turn, may result in the highest values obtained of around 0.67. Additional error may be introduced when redrawing the ellipses over the discs, producing subtle differences between the shapes of the ellipses and discs.

To evaluate these measurement errors, 25 discs were chosen at random from the entire sample ordered in rows and columns. The first specimen selected was placed on the upper-left corner, and subsequently a disc from each of four discs was chosen. Four possible ellipses were redrawn over each of these 25 discs and four eccentricity values obtained for each disc. The minimum value of these was subtracted from the average of the four. Twenty-five values were so obtained and the average of these values resulted in the error.

Systematic palaeontology

Discussion

Phagophytichnus was defined for cuspate excisions on the margin of pinnules and leaves, and included only the ichnospecies P. ekowskii (van Amerom 1966; van Amerom and Boersma 1971). Later, Straus (1977) erected six new ichnospecies: P. nervillo-reliquens, P. circumsecans, P. marginis-folii, P. nervos-mutans, P. nigromarginatus and P. catellarius, which included both margin- and hole-feeding traces. Similarly, Givulescu (1984) included in this ichnogenus margin- and hole-feeding trace fossils, creating the new ichnospecies P. uvaeformis and P. gastropodinus, and identifying two more that were unnamed. However, the difference between margin and hole feeding should be regarded as ichnotaxonomically important considering that it involves two distinct morphological patterns, which reflect clear behavioural differences from potentially different groups of insect producers. Usually insects that feed on leaf margins avoid the interior of the lamina because non-marginal leaf feeders need distinctive anatomical modifications (Edwards and Wratten 1980; Scott et al. 1992). In addition, the fossil evidence described hitherto has indicated that marginal feeding appeared earlier than hole-feeding (Scott et al. 1992; Stephenson and Scott 1992; Labandeira 1998; Labandeira and Allen 2007). Marginal feeding appeared during the Middle/Late Mississippian boundary interval on Triphyllopteris austrina, a tree fern of small stature (Iannuzzi and Labandeira in prep.), and became prominent principally on various types of medullosan tree ferns during the Middle Pennsylvanian (van Amerom and Boersma 1971; Scott et al. 1992). By contrast, hole feeding was rare during the late Middle Pennsylvanian, but became prominent during the Early Permian, principally on gigantopterid seed-fern taxa (Beck and Labandeira 1998; Glasspool et al. 2003).

By the mid Mesozoic, many lineages of non-angiospermous seed plants displayed a wide variety of external foliage feeding, particularly hole and margin feeding (Ash 1999; Rasnitsyn and Krassilov 2000; Scott et al. 2004). During the Early Cretaceous, gymnosperm floras dominated most environments, including that of Baissa in Transbaikalia (Russia), which is of Barremian age. It is from this flora that Vasilenko (2006) established Pinovulnus to include P. regularis, P. serpentiformis and P. erectus from leaves of Pinales and Ginkgoales. According to him, Pinovulnus differs from Phagophytichnus in its plant-host taxonomic affiliation. It is proposed herein to maintain Phagophytichnus only for margin-feeding traces as defined originally. We strongly advocate that plant-insect association ichnotaxa should be defined by their morphology and underlying behaviour, rather than by their plant-host taxonomic assignments. A possible exception concerns highly specific associations between a herbivore and the host-plant, such as gall insects, wherein developmental or structural features of the host-plant are co-opted by the production of physiologically plant-mimicking hormones.

The attribution of P. pseudocircus to leafcutter bees required a particular type of analysis. Earlier authors (Cockerell 1910; Berry 1931; Brooks 1955; Johnston 1993; Lewis 1994; Wappler and Engel 2003) attributed trace fossils in leaves to megachilids based on their apparently circular shape. However, evaluation of the degree of circular expression involved a subjective analysis of the outline of these fossil excisions as well as of leaf fragments cut by extant megachilids. Herein, the eccentricity of the ellipse fitted to the cut margin was used as a test to evaluate objectively the circularity of the trace fossils as well as the extant traces. More than 60 per cent of the values of the entire sample (N = 100) of discs cut by Megachile rotundata had eccentricity values that ranged from 0.35 to 0.63, indicating that, in fact, the discs approximate more to an ellipse than to a circle. The resulting eccentricity average was 0.42 ± 0.10. The eccentricity value of the holotype of P. pseudocircus (0.56) and material reported by Brooks (1955), Cockerell (1910) and Wappler and Engel (2003), fall within the range of the values obtained from excisions of extant megachilids, indicating, with a high degree of confidence, that these trace fossils are the result of the nesting activity of a leafcutter bee. Additionally, Horne (1995) demonstrated that megachilid bees usually cut several discs from the same leaf, as seen in the leaves of P. pseudocircus considered herein and in previous studies (Cockerell 1910; Brooks 1955; Wappler and Engel 2003).

Pre-Cenozoic fossil bee evidence comes from Halictidae nests described by Elliott and Nations (1998) from the Cenomanian Dakota Formation of the USA and by Genise et al. (2002) from the Santonian–Campanian Laguna Palacios Formation of Argentina. Until late 2006, the oldest documented bee body-fossil was Cretotrigona priscaMichener and Grimaldi (1988), originating from Late Cretaceous amber from New Jersey (USA) (Michener and Grimaldi 1988; Engel 2000). However, its derived position within the superfamily Apoidea suggests a substantially earlier origin for the entire group and that significant stages of bee evolution had already occurred by the Late Cretaceous (Engel and Perkovsky 2006). In a report from late 2006 (Poinar and Danforth 2006), a significantly earlier and smaller bee specimen was described from Myanmar (Burmese) amber, dated as late Albian or latest Early Cretaceous. This specimen was assigned to a new family, Melittosphecidae. More recently, however, Ohl and Engel (2007) have indicated that this insect body fossil is not of a bee; moreover, they stated that Melittosphex burmensis is a synonym of Cretospilomena (Hymenoptera: Crabroninae), also described from Burmese amber (Antropov 2000). For the Megachilidae, which lack Cretaceous body fossils, the trace- and body-fossil record only becomes relatively abundant for the mid Palaeogene. The evidence for fossil megachilids in North America comes from the late Eocene Florissant Formation in Colorado (Cockerell 1906, 1908, 1911, 1913, 1914, 1917, 1923, 1925a; Engel 2001; Meyer 2003), and from amber of the early Miocene (Aquitanian) La Toca Formation in the Dominican Republic (Engel 1999). In Europe, by contrast, major fossil occurrences are of an older Palaeogene age. Megachilid body fossils have been documented from amber from the late Eocene (Priabonian) Obukhov Formation in the Rovno Region of the Ukraine (Engel and Perkovsky 2006), the middle Eocene (mid Lutetian) Prussian Formation of the Baltic Region (Cockerell 1909a, b; Engel 2001), the upper Oligocene (Chattian) brown coals of Rott, in Schwaben, Germany (von Heyden 1862; Meunier 1920; Statz 1936), the Miocene (probably Aquitanian) of the Kudia River, near Amagu in the Far East of Russia (Cockerell 1925b), and late Miocene deposits of Oeningen, Germany (Heer 1849, 1865; Cockerell 1909a; Zeuner and Manning 1976). For Asia there is evidence for the presence of megachilid bees from the middle Miocene Shanwang Formation, near Linqu, Shandong, China (Zang 1989a, b).

The subfamily Megachilinae is over-represented among fossil Megachilidae when compared with the Recent bee fauna (Engel and Perkovsky 2006). For South America, body-fossil bee evidence comes only from a fossil bee assigned to a corbiculate apid from Late Paleocene deposits in north-west Argentina (Petrulevicius 2002). The Megachilidae as a whole is regarded as a derived group based on phylogenetic analysis (Roig-Alsina and Michener 1993); it probably originated during the Paleocene or latest Cretaceous (Engel and Perkovsky 2006). Since derived megachilids produce a distinctive leaf-excision pattern along the margin of leaves, the trace-fossil evidence presented herein demonstrates that a more advanced clade of Megachilini was already present in South America by the early Eocene some 47.5 myr ago, thus indicating a significantly earlier origin for the Megachilidae.