Veget Hist Archaeobot (2008) 17 (Suppl 1):S277–S286 DOI 10.1007/s00334-008-0176-4 ORIGINAL ARTICLE Taxonomic identification of dry and carbonized archaeobotanical remains of Cucurbita species through seed coat micromorphology Vero´nica Lema Æ Aylen Capparelli Æ Marı´a Lelia Pochettino Received: 31 October 2007 / Accepted: 7 June 2008 / Published online: 30 July 2008 Ó Springer-Verlag 2008 Abstract Cucurbita seeds are difficult to identify to problem, several authors have made statistical analyses on species level using only their external morphology. In this measurements of seeds, rinds and peduncles of North contribution, we discuss anatomical features of fresh, American Cucurbita (eg. Cucurbita pepo L.) to identify dehydrated and experimentally carbonised specimens that useful diagnostic features (Decker and Wilson 1986; are useful for the identification of archaeological Cucurbita Newsom et al. 1993; Smith 2000, 2006). However, the seeds. Qualitative and quantitative differences in the seed diagnostic criteria of North American Cucurbita seeds are coat micromorphology were found to be the most helpful not always appropriate for the identification of South diagnostic characteristics of South American Cucurbita American species, which include many closely related species. taxa, such as C. maxima Duchesne ssp. maxima, C. maxima Duchesne ssp. andreana (Naudin) Filov. and C. moschata Keywords Seed coat Á Taxonomic identification Á (Lam.) Poir. Cucurbita Several papers have been written about the testa tissues of Cucurbita seeds, most of them providing qualitative descriptions (Lott 1973; Stuart and Loy 1983) but only a Introduction few (Singh and Dathan 1972; Teppner 2004) provide data that is useful for differentiating species. The present study Cucurbita seeds, including archaeological specimens, are is the first archaeobotanical analysis of the characteristics typically identified to species from the qualitative charac- of South American Cucurbita seed testae. In this paper, we teristics of their external morphology, particularly colour, discuss our results, describing the microscopical features of margin and funicular attachment (Cutler and Whitaker fresh, dehydrated and experimentally carbonised Cucurbita 1961; Lira Saade 1995). However, these features are not seed coats that are useful for the identification of archae- always appropriate for the identification of dried or charred ological Cucurbita seeds. archaeobotanical remains (Smith 2000). To address this General features of Cucurbita seeds Communicated by M. van der Veen. Cucurbita seeds are flattened and ellipsoid, with a rela- V. Lema (&) Á A. Capparelli tively more or less pronounced marginal bulge. According Departamento cientı´fico de arqueologı´a, Laboratorio de to Singh and Dathan (1972) the seed coat is comprised of Etnobota´nica y Bota´nica Aplicada, Museo de Ciencias Naturales de La Plata, Paseo del Bosque s/n, 1900 La Plata, Argentina four layers of tissue: epidermis, hypodermis, sclerenchyma, e-mail: [email protected] and aerenchyma, which surround an ‘‘inner zone’’. The epidermis consists of radially elongated cells, which have M. L. Pochettino more or less thickened non-lignified walls (Esau 1977). Laboratorio de Etnobota´nica y Bota´nica Aplicada, Museo de Ciencias Naturales de La Plata, These cells vary in their length, with longer ones typically Paseo del Bosque s/n, 1900 La Plata, Argentina occurring around the marginal bulge (Singh and Dathan 123 S278 Veget Hist Archaeobot (2008) 17 (Suppl 1):S277–S286 1972; Teppner 2004). The hypodermis consists of small 1. Transverse Sections (Tr) were obtained from micro- cells with slightly thickened lignified walls where thick- pyle (C1), middle (C2), and chalaza or basal (C3) ening forms a reticule (Esau 1977;Lott1973). The portions of the seed. Successive sections of C1 and C3 sclerenchyma layer gives the seed coat its firmness were labelled as ‘‘a’’ and ‘‘b’’ as they bordered C2. (Hayward 1953) through sclereids with festooned lignified Between 5 and 11 seeds of each taxon were analysed walls (Esau 1977; Singh and Dathan 1972). The aeren- which resulted in the analysis of 103 thin sections from chyma is comprised of cells with arm-like extensions, all the taxa analysed. which are interspersed with enlarged intercellular spaces 2. Tangential Longitudinal sections (TgLg) were taken (Esau 1977; Hather 2000; Singh and Dathan 1972). Lott from the middle zone (C2). The testa was removed (1973) describes the tissue between the sclerenchymatous from the seed. For the purposes of separating the layer and the cotyledons as a spongy parenchyma formed sclerenchyma tissue, but maintaining its integrity, by dead irregular-shaped cells with numerous outgrowths. tangential sections were immersed in a chemical maceration solution (50% hydrogen peroxide 100 volumes and 50% acetic acid) and boiled for Methodology 10–15 min. To achieve a better understanding of sclerenchyma morphology this type of sectioning was Microscopy was used to study 63 seeds of Cucurbita fici- performed on two specimens: a single C. maxima ssp. folia Hubber, C. moschata, C. maxima Duchesne ssp. maxima cv. criollo seed and a single C. maxima ssp. andreana, C. maxima ssp. maxima cv. zapallito; C. maxima andreana ssp. maxima cv. hubbard; C. maxima ssp. maxima cv. in- 3. Radial Longitudinal section (RdLg) The same proce- gles and C. maxima ssp. maxima cv. criollo. Samples of dure described above for the tangential longitudinal cultivars were provided by the Faculty of Agronomy, La section was used to obtain a radial view of the Plata National University (U.N.L.P.) and C. maxima ssp. sclerenchyma cells of a single C. maxima ssp. maxima andreana samples were collected from several localities in cv. criollo seed. Cordoba province, representing three kinds of pollination The Group 2 seeds (n = 7) were dehydrated by air- (free, autogamic and crossed). Voucher specimens were drying and manually fractured in the middle (C2) for deposited in the Laboratory of Ethnobotany and Applied studying with a stereoscopic microscope as well as SEM Botany (Faculty of Natural Sciences, U.N.L.P.) and in a (at magnifications 80, 370 and 7009). personal collection at the Archaeological Scientific At first, we only considered the qualitative features of Department of La Plata Museum of Natural Sciences the seed coat. As the study progressed, we recognised that (U.N.L.P.). there was a need for quantitative analyses because there Samples were divided into three groups for different were so many qualitative similarities among closely related types of analysis. Seeds in Group 1 (n = 35) were hydrated taxa, for example C. maxima subspecies and cultivars. and the testae cut into thin sections that were observed Therefore, measurements were taken of seed coat cells employing light microscopy (LM). Only the testa tissue from images recorded with a Motic Image Plus 2.0 web layers were studied. camera and its measuring software. To check the accuracy The Group 1 seeds were subjected to three types of thin of the Motic Image Plus 2.0 measurements, the results were sectioning (see Fig. 1): compared with those made with a drawing tube attached to a Leica microscope (DM/LM). Measurements were also calculated from SEM images using the programme Image Tool 3.0. From each measurement, we calculated means and range of width, length and height of cells. Between 20 and 50 measurements on each taxon were taken. Data on C. maxima ssp. maxima were produced by combining the four cultivars. The seeds in Group 3 (n = 21) were charred to observe how carbonisation alters seed shape and size due to expansion or shrinkage of the internal tissue. The samples comprised dried seeds of C. maxima ssp. maxima cv. cri- ollo, C. maxima ssp. andreana, C. moschata and C. ficifolia, which were charred in both an aerobic and Fig. 1 Sections of the studied Cucurbita seeds anaerobic atmosphere in an electric muffle oven at 300°C 123 Veget Hist Archaeobot (2008) 17 (Suppl 1):S277–S286 S279 for 120 minutes. We measured seed dimensions (length, C3b) there are shorter elongated cells, which appear like width and height) before and after charring using digital ‘‘horns’’, and do not surround the short cells (Fig. 2c). This calipers. The best preserved specimens were those charred observation has not been reported previously in the litera- in the aerobic atmosphere. They were manually fractured ture although this pattern is described for C. maxima ssp. in their C2 region and their testae were observed under maxima (Fig. 2e). This observation highlights the impor- SEM at magnifications of 80, 370 and 7009. tance of selecting the proper cross section location to make correct identifications. In both C. ficifolia and C. moschata the bases of the long cells are situated in the submarginal Results and discussion bulges, respectively (Fig. 2a, b; Teppner 2004). Our observations of the C2 region of C. maxima ssp. Seed areas useful for taxonomic identification maxima showed that most of the specimens of the four cultivars have a few long cells, which are not much longer Cutting transverse sections from different parts of the seed than the short ones, forming a protuberance (Fig. 2e). (C1, C2 and C3) allowed us to examine the tissue in detail However, some specimens of cultivars criollo and ingle´s and to conclude that C2 has the most homogeneous tissue had long cells surrounding short cells encompassing one- morphology. Therefore, it was decided that further taxo- third (or less) of the marginal bulge. In some specimens of nomic analysis should be made in this region. It was also landraces from the Argentinean northwest, seed coat type 3 decided that both the flattened area of the seed corre- with a narrow submarginal bulge and long cells encom- sponding to the major faces and the marginal bulge should passing the marginal bulge as proposed by Teppner (2004) be examined. was recognized, therefore it is thought that more specimens of this subspecies should be investigated for this feature. Marginal bulge diagnostic characters Contrary to Teppner’s (2004) report that in C.