Version classiqueVersion mobile
OpenEdition Books

AGRUMED: Archaeology and history of citrus fruit in the Mediterranean

 | 
Véronique Zech-Matterne
, 
Girolamo Fiorentino

Quantitative evaluation of modern Citrus seed shape and comparison with archaeological remains discovered in Pompeii and Rome

Anna Maria Grasso, Fabio Mavelli et Girolamo Fiorentino

Texte intégral

  • 1 Corner 1976.

1The taxonomic identification of seeds and fruit generally entails the recognition of macro and microscopic morphological and anatomic characteristics, as well as the measurements of biometric characteristics.1 Indeed, a long tradition of studies in this sector has led to the creation of standardised and functional criteria for many taxa. These usually combine a small number of standard measurements (e.g. length, width and thickness) with qualitative observations that identify the shape (e.g. elongated, rounded); however, in some cases these results are dependent on the perceptions of the researcher.

  • 2 Vaughan 1970.
  • 3 Pagnoux et al. 2013.
  • 4 IBPGR 1988.
  • 5 The latest list of descriptors no longer considers the biometric parameters of the seeds to be deci (...)
  • 6 2013.

2Furthermore, application of the traditional morphological and biometric criteria to seeds of the Citrus genus is not sufficient to recognise individual species,2 due to the wide intraspecific morphological variability.3 The International Board for Plant Genetic Resources4 has identified some descriptors that concisely express the biometric characteristics of the seed (length, width, weight),5 shape (fusiform, clavate, cuneiform, ovoid, deltoid, globose, semi-spheroid, other) and surface texture (smooth, wrinkled, hairy). For the first time, some of these were selected by Pagnoux et al.6 and adopted for the taxonomic attribution of unknown archaeological samples recovered from Pompeii and Rome. As already stated, however, the size and shape of the individual seeds are fairly variable and naked-eye observation is not always sufficient to determine their species. It should also be pointed out that the microscopical surface texture is not always clearly visible on seeds sampled from archaeological contexts, because the taphonomic dynamics can distort this evidence or even render it completely unreadable.

  • 7 Size aspects were included in the analysis because, to the authors’ knowledge, there are no studies (...)

3In order to complement traditional methods, this study proposes to use quantitative techniques to analyse the seed’s shape and to process the data by means of multivariate analysis.7 The aim of this study is to analyse the diversity of Citrus seeds first in a controlled modern collection then in archaeological samples, in order to highlight the similarities and differences between them.

4In the last twenty years (increasing exponentially in the last decade) image analysis techniques have played a crucial role in studying the shape of various organisms, with numerous applications in the botanical and archaeobotanical sector, among others. This research is based on the correlation between shape and the genetic properties that determine it, these differences can then be used to support taxonomic identification.

  • 8 Kuhl, Giardina 1982; Rohlf, Archie 1984.
  • 9 For a summary see Costa et al. 2011.
  • 10 Orrù et al. 2013; Ucchesu et al. 2015.

5Complex morphological analyses use closed-profile methods to extract and statistically describe the shape outline of a digital image. Among these procedures, the most effective is the Elliptic Fourier Analysis (EFA);8 this approximates the closed boundary of an image by summation of its harmonic functions weighted by numerical coefficients. The values of the coefficients constitute a dataset of morphological descriptors that can be used to classify items by means of their shape outline. The classification is then performed by applying multivariate statistical methods. In the case of seeds, fruit and other parts of the plant, this approach has been successfully applied to numerous taxa.9 Recently,10 a criterion was proposed that integrates the data from the outline-based method including: 1. certain morphometric characteristics (e.g. perimeter, area); 2. the ratios between measurements (e.g. convex perimeter and crofton perimeter); 3. shape indices (e.g. roundness factor). By means of stepwise Linear Discriminant Analysis (LDA), the combined information was used to identify the most informative descriptors and to define a discriminant model. The morphometric analysis of the Citrus genus seeds was conducted precisely in accordance with this integrated approach.

1. Materials

  • 11 Citrus species have different degrees of genetic diversity: it is greater for C. reticulata, C. med (...)
  • 12 In agreement with: Rovner, Gyulai 2007; Pagnoux et al. 2015.

6The controlled modern collection was composed of eight species – C. maxima, C. medica, C. reticulata, C. myrtifolia, C. aurantium, C. aurantifolia, C. sinensis, C. limon  provided by the citrus collection CRB INRA-CIRAD Citrus in San Giuliano, Corsica (France).11 For each species, 30 seeds were chosen at random and analysed12 to form a “training set”; no preparatory treatment was necessary.

  • 13 Mineralisation and waterlogging usually leads to the loss of the seeds interior and/or of some exte (...)
  • 14 Fiorentino, Marinò 2008.
  • 15 Pagnoux et al. 2013.
  • 16 Fiorentino et al. 2014.
  • 17 Pagnoux et al. 2013.

7The archaeological samples were composed of two mineralised seeds from Pompeii and nine waterlogged seeds from Rome.13 The seeds from Pompeii were in a good state of conservation; unfortunately, the cellular structure of the epicarp was not always perfectly conserved in the seeds from Rome. The former were recovered following the excavation of a cistern below the Temple of Venus, whose contents were archaeologically dated to the 3rd-2nd centuries BC.14 The characteristics of the shape, the presence of the ventral crest and the structure of the surface cells ruled out the possibility that they were seeds of the Maloideae species, but did indicate that they may belong to C. medica.15 The seeds from Rome came from a votive deposit, sealed below the pavement of the Carcer Tullianum situated in the northern part of the Roman Forum, dated radiometrically to the Augustan Age (27 BC-14 AD).16 These are believed to belong to the limon species.17

2. Methods

8For each seed, the length, width and thickness were measured to create a biometric data set. Principal Component Analysis (PCA) was performed on this set in order to assess the variability of modern seeds and to evaluate if these parameters, which are extensively used in classical analysis, were sufficiently discriminant for the Citrus genus.

  • 18 Kuhl, Giardina 1982.
  • 19 Iwata, Ukai 2002.

9Subsequently, the digital images of all the seeds were processed to calculate the Elliptic Fourier Descriptors (EFDs).18 These data were then used to perform the PCA, in order to evaluate if they were discriminant. The extracted coefficients were used also to reconstruct the average shape of the seeds by species, using Inverse Fourier Transform,19 in order to visualize their morphological variability and to provide a comparison catalogue for specialists.

  • 20 In agreement with Ucchesu et al. 2015.

10This study also proposes an alternative approach that integrates information of shape and dimensions. Therefore, in order to better characterise them, 18 morphometric descriptors were determined: Area, Perimeter, Convex Perimeter, Crofton perimeter, Perimeter ratio Pconv/Pcrof, Feret (Dmax), MinFeret (Dmin), Feret ratio, Max ellipse axis, Minor ellipse axis, Circ. (shape factor), Round, Eq. Circular diameter, Fibre length, Curl degree (Dmax/F), Convexity degree (Pcfrof/P), Solidity, Compactness degree.20 The data obtained for each individual seed based on the 18 morphometric characteristics and the 77 Fourier descriptors formed the training dataset that was then used for statistical analysis.

  • 21 The descriptors selected with this procedure are: Perimeter, a3, d2, Perimeter ratio (Convex perime (...)

11Linear Discriminant Analysis (LDA) was applied to the training dataset in order to determine a discriminant model to classify the archaeological seeds. In fact, LDA minimises the variance within groups and maximises the variance among groups, thereby providing a reliable classification model of the samples in the training set. Moreover, this approach made it possible to identify the best descriptors from among the 95 proposed by using a stepwise method of selection. Twelve descriptors were selected,21 and these were the only ones used in the discriminant model. This model was then evaluated by the leave-one-out cross-validation procedure.

  • 22 In agreement with Pagnoux et al. 2013.

12Once the discriminant model had been validated, it was used to classify the archaeological seeds and assign them to the species constituting the training dataset. A probability of attribution was also estimated obtaining scores considered good if ≥ 0.75 and very good if ≥ 0.90.22

  • 23 Yoshioka et al. 2004.

13In order to understand whether the characteristics of the seed’s shape also reflected precise phylogenetic relationships and, in light of this, be useful in interpreting the results of the LDA applied to the archaeological dataset, PCA and hierarchical clustering23 were conducted only on data related to the twelve selected descriptors in both the modern collection (average values of each species) and the archaeological samples.

3. Results

  • 24 2013.

14The length, width and thickness of each of the 30 seeds from the 8 species in the modern collection confirmed that there was considerable biometric variation, not only between species but also within species. The data (fig. 1) did not always coincide with the data reported in the recent study by Pagnoux et al.,24 perhaps partly due to the greater number of samples examined in this study. Specifically, there was a broad dispersion of length in C. maxima, as well as in C. limon and C. myrtifolia. PCA performed on the three biometric features of the modern dataset identified only two macro groups: one composed of the species maxima, aurantium and sinensis and the other composed of the remainder (fig. 2).

Fig. 1 - Measurements of 30 modern Citrus species seeds.

Fig. 1 - Measurements of 30 modern Citrus species seeds.

Fig. 2 - PCA performed on the 3 biometric features of the modern dataset.

Fig. 2 - PCA performed on the 3 biometric features of the modern dataset.

15A similar result was obtained by performing PCA on the dataset of 77 Elliptical Fourier coefficients (data not shown).

16Furthermore, using the coefficients calculated with the inverse Fourier Transform, it was also possible to reconstruct and graphically present the morphological variability for each species (fig. 3). Depending on the species, between five and ten principal components are needed to account for 95% of the total variance, with the first two to four components accounting for more than 80%. The visualisation of the shape variations, described by each of the principal components, indicated that the first principal component tends to be connected to the length-to-width ratio of the seed, the second component to the position of the seed’s centroid and the third to the shape of the distal and proximal parts. The remaining components correlated with different variations from the first three; these are difficult to interpret but are generally marginal (none of them accounting for more than 5% of variance). In summary we can say that the most homogeneous species are C. limon and C. reticulata and, more generally, it can be observed that variation in the shape and orientation of the apex and hilum are the most informative features.

Fig. 3 - Variation of Citrus seed shape within each species and the effect of each principal component. Each line represents the case that principal components takes ±2σ and mean, respectively.

Fig. 3 - Variation of Citrus seed shape within each species and the effect of each principal component. Each line represents the case that principal components takes ±2σ and mean, respectively.

17Finally, LDA was performed on the 95 descriptors dataset (18 morphometric descriptors and the 77 Fourier descriptors) which allowed us to define a discriminant model based on only 12 descriptors (data not show) selected by the stepwise procedure. This model was found to have correctly classified 81.3% of the original grouped cases in the training dataset (76.3% of cross-validated grouped cases). The most effectively discriminated species were C. aurantifolia, C. maxima and C. myrtifolia. In contrast, the greatest problems were observed with C. limon (which can be erroneously attributed to C. medica and C. reticulata) and C. sinensis (which can be confused with C. aurantium). The graphic representation of the seed’s LDA, from the eight species present in the collection, is shown in fig. 4, which plots the first two discriminant functions.

Fig. 4 - Linear Discriminant Analysis plot (the centroid of each species is represented).

Fig. 4 - Linear Discriminant Analysis plot (the centroid of each species is represented).

18Classification of the archaeological seeds made it possible to attribute the two seeds from Pompeii to C. aurantifolia with a very good probability. In contrast, the seeds from Rome were attributed in three cases to C. aurantium, in one case to C. reticulata with a very good probability and in one case to C. aurantium with a good probability. The remaining seeds were attributed to C. aurantifolia, C. myrtifolia and C. limon; however, this was considered invalid because the probability was below 75%. The PCA and Hierarchical Cluster Analysis performed on the twelve descriptors of the archaeological dataset, and on the average values for each species of the training set, confirmed the attribution of the seeds discovered in Pompeii to C. aurantifolia; in contrast, it showed disparate values for the sample from Rome, with a tendency towards C. reticulata (fig. 5 and 6).

Fig. 5 - Modern species and archaeological seeds from Pompeii and Rome (Principal Component Analysis plot).

Fig. 5 - Modern species and archaeological seeds from Pompeii and Rome (Principal Component Analysis plot).

Fig. 6 - Cluster analysis plot from modern species and archaeological seed data.

Fig. 6 - Cluster analysis plot from modern species and archaeological seed data.

4. Discussion and research perspectives

19Tracing the geographical dissemination pathways and the history of fruiting plants, as with vegetables and more generally all fruit that do not require special processing in order to be consumed (e.g. roasting), is always problematic, because their remains are rarely conserved in archaeological contexts. In the case of the Citrus genus, discoveries are extremely rare, and even when they are made, the quantity of conserved remains is minimal. To this, it may be added the difficulty of taxonomic attribution and recognition that further complicates and hampers the research. It is therefore necessary to develop a protocol of analysis that makes it possible to enhance the information potential of the few archaeological remains discovered.

20The traditional approach (measurement of the dimensions and analysis of the shape) has shown that although there are some characteristics that clearly identify certain taxa, there remain overlaps that only a comprehensive, integrated study can resolve. Indeed, the descriptors selected by LDA on the basis of the training set produced good results in terms of discrimination, which bodes well for continuation of the research. This would entail broadening the control collection and making use of taxa selected on the basis of specific issues to be resolved, such as phylogenetic relationships between species or the distribution of cultivars in particular areas.

  • 25 Nicolosi et al. 2000; Penjor et al. 2013.
  • 26 Wu et al. 2014.
  • 27 Gulsen, Roose 2001.
  • 28 Barkley et al. 2006; Bayer et al. 2009.
  • 29 Curk et al. 2015.
  • 30 Cfr. Pagnoux et al. 2013.
  • 31 2011: 88.
  • 32 2013.
  • 33 Pagnoux et al. 2013.

21Recent studies on the origins of the Citrus species cultivated today seem to agree that there are three ancestor species: C. maxima, C. medica and C. reticulata.25 Regarding the taxa used in the present study, it is argued that C. aurantium, and subsequently also C. sinensis, were hybridised from C. maxima and C. reticulata;26 C. limon was hybridised from C. medica and C. aurantium,27 while C. aurantifolia is believed to be the result of crossing C. micrantha with C. medica28 and, lastly, C. myrtifolia is probably a type of C. aurantium.29 As a result of this study, hypotheses have been formulated regarding the history of the species that still exist today. Obviously, there may also be others that are no longer extant  a particularly valid point in the case of the Citrus genus, due to its modes of reproduction and speciation. The most recent studies propose that in Italy, in antiquity, C. medica and C. limon were both known, and in some cases were perhaps even cultivated.30 In reality there is no unequivocal evidence of their presence, but there are elements that attest to the presence of the Citrus genus, which, partly on the basis of literary and iconographic sources, identify with one or another of the two cited species, depending on the case. The application of these categories might, however, be reductive, as the results of the analysis applied to the archaeological sample presented here seem to indicate. Indeed, the seeds of Pompeii are very similar to modern C. aurantifolia. Van der Veen,31 who found lime seeds and rind fragments at the medieval ports of Quseir al-Qadim (Egypt), has seen them and, after personal observation, supposed that they might be lime. Conversely, Pagnoux et al.,32 who have studied the surface pattern using SEM images, affirm that they have the same characteristics of modern C. medica: the seed-coat exhibits cells chained in longitudinal and parallel lines, arranged in fascicles. Thus, these conflicting determinations may suggest the presence in Pompeii of an ancient cultivar based on a C. medica hybrid rather than C. medica or C. aurantifolia, itself. The results of the analyses of the seeds from Rome seem to indicate the same thing. Indeed, first and foremost, the sample exhibits a certain degree of dispersion (perhaps due to the absence in the modern collection of varieties that might have yielded a less equivocal attribution), with a tendency towards the species C. aurantium/C. reticulata. As already mentioned, these two species played a role in the origin of the modern lemon, and it is therefore possible that the remains discovered in Rome also belong to a cultivar that was derived from these species but was not precisely identifiable with modern C. limon, as Pagnoux et al. proposed.33

22In conclusion, it can be affirmed that the results set out here indicate the possibility that the Citrus species present in antiquity may be different from our modern ones. Certainly, there is a need to expand the modern collection, adding other cultivars and endemic varieties, in order to better characterise every species. However, it is also necessary to construct a database of seeds discovered in archaeological contexts, to make it possible to compare samples with a certain geographical and temporal contiguity, in order to fully understand the circulation of Citrus particularly in the Mediterranean.

Bibliographie

Barkley et al. 2006: N.A. Barkley, M.L. Roose, R.R. Krueger, C.T. Federici, Assessing genetic diversity and population structure in a Citrus germplasm collection utilizing simple sequence repeat markers (SSRs), Theoretical and Applied Genetics, 112, p. 1519-1531.

Barracosa et al. 2007: P. Barracosa, J. Osório, A. Cravador, Evaluation of fruit and seed diversity and characterization of carob (Ceratonia siliqua L.) cultivars in the Algarve region, Scientia Horticulturae, 114, p. 250-257.

Bayer et al. 2009: R.J. Bayer, D.J. Mabberley, C. Morton, C.H. Miller, I.K. Sharma, B.E. Pfeil, S. Rich, R. Hitchcock, S. Sykes, A molecular phylogeny of the orange subfamily (Rutaceae: Aurantioideae) using nine cpDNA sequences, American Journal of Botany, 96, p. 668-685.

Corner 1976: E.J.H. Corner, The seeds of Dicotyledons, Cambridge.

Costa et al. 2011: C. Costa, F. Antonucci, F. Pallottino, J. Aguzzi, D.-W. Sun, P. Menesatti, Shape Analysis of Agricultural Products: A Review of Recent Research Advances and Potential Application to Computer Vision, Food and Bioprocess Technology, 4, p. 673-692.

Curk et al. 2015: F. Curk, G. Ancillo, F. Ollitrault, X. Perrier, J.-P. Jacquemoud-Collet, A. Garcia-Lor, L. Navarro, P. Ollitrault, Nuclear Species-Diagnostic SNP Markers Mined from 454 Amplicon Sequencing Reveal Admixture Genomic Structure of Modern Citrus Varieties, PLoS One, 10, 5: e0125628 <https://doi.org/10.1371/journal.pone.0125628doi:10.1371/journal. pone.0125628>.

Fiorentino, Marinò 2008: G. Fiorentino, G. Marinò, Analisi archeobotaniche preliminari al Tempio di Venere di Pompei, in P.G. Guzzo, M.P. Guidobaldi (eds.), Nuove ricerche archeologiche nell’area vesuviana (scavi 2003-2006), Roma, p. 527-528.

Fiorentino et al. 2014: G. Fiorentino, V. Zech-Matterne, E. Boaretto, A. Celant, S. Coubray, N. Garnier, F. Luro, C. Pagnoux, M.-P. Ruas, AGRUMED: The history of Citrus fruits in the Mediterranean. Introductions, diversifications and uses, Antiquity. Project Gallery, 88, 339 <http://antiquity.ac.uk/projgall/fiorentino339/>.

Freeman 1974: H. Freeman, Computer Processing of Line-Drawing Images, ACM Computing Surveys, 6, 1, p. 57-97.

Gulsen, Roose 2001: O. Gulsen, M.L. Roose, Lemons: diversity and relationships with selected Citrus genotypes as measured with nuclear genome markers, Journal of the American Society for Horticultural Science, 126, 3, p. 309-317.

Hammer et al. 2001 : Ø. Hammer, D.A.T. Harper, P.D. Ryan, PAST: Paleontological Statistics Software Package for Education and Data Analysis, Palaeontologia Electronica , 4, 1, p. 1-9.

IBPGR 1988: International Board for Plant Genetic Resources, Descriptors for Citrus, Rome.

IPGRI 1999: International Plant Genetic Resources Institute, Descriptors for Citrus, Rome.

Iwata, Ukai 2002: Y. Iwata, H. Ukai, SHAPE: A computer program package for quantitative evaluation of biological shapes based on elliptic Fourier descriptors, Journal of Heredity, 93, 5, p. 384-385.

Iwata et al. 1998: H. Iwata, S. Niikura, S. Matsuura, Y. Takano, Y. Ukai, Evaluation of variation of root shape of Japanese radish (Raphanus sativus L.) based on image analysis using elliptic Fourier descriptors, Euphytica, 102, p. 143-149.

Kuhl, Giardina 1982: F.P. Kuhl, C.R. Giardina, Elliptic Fourier features of a closed contour, Computer Graphics and Image Processing, 18, p. 236-258.

Nicolosi et al. 2000: E. Nicolosi, Z.N. Deng, A. Gentile, S. La Malfa, G. Continella, E. Tribulato, Citrus phylogeny and genetic origin of important species as investigated by molecular markers, Theoretical and Applied Genetics, 100, 8, p. 1155-1166.

Ohsawa et al. 1998: R. Ohsawa, T. Tsutsumi, H. Uehara, H. Namai, S. Ninomiya, Quantitative evaluation of common buckwheat (Fagopyrum esculentum Moench) kernel shape by elliptic Fourier descriptor, Euphytica, 101, p. 175-183.

Orrù et al. 2013: M. Orrù, O. Grillo, G. Lovicu, G. Venora, G. Bacchetta, Morphological characterisation of Vitis vinifera L. seeds by image analysis and comparison with archaeological remains, Vegetation History and Archaeobotany, 22, p. 231-242.

Pagnoux et al. 2013: C. Pagnoux, A. Celant, S. Coubray, G. Fiorentino, V. Zech-Matterne, The introduction of Citrus in Italy with reference to the identification problem of seed remains, Vegetation History and Archaeobotany, 22, p. 421-438.

Pagnoux et al. 2015: C. Pagnoux, L. Bouby, S. Ivorra, C. Petit, S.-M. Valamoti, T. Pastor, S. Picq, J.-F. Terral, Inferring the agrobiodiversity of Vitis vinifera L. (grapevine) in ancient Greece by comparative shape analysis of archaeological and modern seeds, Vegetation History and Archaeobotany, 24, 1, p. 75-84.

Penjor et al. 2013: T. Penjor, M. Yamamoto, M. Uehara, M. Ide, N. Matsumoto, R. Matsumoto, Y. Nagano, Phylogenetic Relationships of Citrus and Its Relatives Based on matK Gene Sequences, PLoS One, 8, 4: e62574 <https://doi.org/10.1371/journal.pone.0062574>.

Rohlf, Archie 1984: J.F. Rohlf, A.W. Archie, A comparison of Fourier methods for the description of wing shape in mosquitoes (Diptera: Cuculidae), Systematic Zoology, 33, p. 302-317.

Ros et al. 2014: J. Ros, A. Evin, L. Bouby, M.-P. Ruas, Geometric morphometric analysis of grain shape and the identification of two-rowed barley (Hordeum vulgare subsp. distichum L.) in southern France, Journal of Archaeological Science, 41, p. 568-575.

Rovner, Gyulai 2007: I. Rovner, F. Gyulai, Computer-Assisted Morphometry: A New Method for Assessing and Distinguishing Morphological Variation in Wild and Domestic Seed Populations, Economic Botany, 61, 2, p. 154-172.

Schneider et al. 2012: C.A. Schneider, W.S. Rasband, K.W. Eliceiri, NIH Image to ImageJ: 25 years of image analysis, Nature Methods, 9, p. 671-675.

Smykalova et al. 2011: I. Smykalova, O. Grillo, M. Bjelkova, M. Hybl, G. Venora, Morpho-colorimetric traits of Pisum seeds measured by an image analysis system, Seed Science and Technology, 39, p. 612-626.

Terral et al. 2010: J.-F. Terral, E. Tabard, L. Bouby, S. Ivorra, T. Pastor, I. Figueiral, S. Picq, J.-B. Chevance, C. Jung, L. Fabre, C. Tardy, M. Compan, R. Bacilieri, T. Lacombe, P. This, Evolution and history of grapevine (Vitis vinifera) under domestication: New morphometric perspectives to understand seed domestication syndrome and reveal origins of ancient European cultivars, Annals of Botany, 105, p. 443-455.

Ucchesu et al. 2015: M. Ucchesu, M. Orrù, O. Grillo, G. Venora, A. Usai, P.-F. Serreli, G. Bacchetta, Earliest evidence of a primitive cultivar of Vitis vinifera L. during the Bronze Age in Sardinia (Italy), Vegetation History and Archaeobotany, 24, 5, p. 587-600.

Van der Veen 2011: M. Van der Veen, Consumption, trade and innovation. Exploring the botanical remains from the Roman and Islamic ports at Quseir al-Qadim, Egypt, Frankfurt.

Vaughan 1970: J.G. Vaughan, The structure and utilization of oil seeds, London.

Venora et al. 2009: G. Venora, O. Grillo, C. Ravalli, R. Cremonini, Identification of Italian landraces of bean (Phaseolus vulgaris L.) using an image analysis system, Scientia Horticulturae, 121, 4, p. 410-418.

Wu et al. 2014: G.A. Wu, S. Prochnik, J. Jenkins, J. Salse, U. Hellsten, F. Murat, X. Perrier, M. Ruiz, S. Scalabrin, J. Terol, M.A. Takita, K. Labadie, J. Poulain, A. Couloux, K. Jabbari, F. Cattonaro, C. Del Fabbro, S. Pinosio, A. Zuccolo, J. Chapman, J. Grimwood, F.R. Tadeo, L.H. Estornell, J.V. Munoz-Sanz, V. Ibanez, A. Herrero-Ortega, P. Aleza, J. Perez-Perez, D. Ramon, D. Brunel, F. Luro, C. Chen, W.G. Farmerie, B. Desany, C. Kodira, M. Mohiuddin, T. Harkins, K. Fredrikson, P. Burns, A. Lomsadze, M. Borodovsky, G. Reforgiato, J. Freitas-Astua, F. Quetier, L. Navarro, M. Roose, P. Wincker, J. Schmutz, M. Morgante, M.A. Machado, M. Talon, O. Jaillon, P. Ollitrault, F. Gmitter, D. Rokhsar, Sequencing of diverse mandarin, pummelo and orange genomes reveals complex history of admixture during citrus domestication, Nature Biotechnology, 32, 7, p. 656-662.

Yoshioka et al. 2004: Y. Yoshioka, H. Iwata, R. Ohsawa, S. Ninomiya, Analysis of petal shape variation of Primula sieboldii by elliptic Fourier descriptors and principal component analysis, Annals of Botany, 94, p. 1-8.

Notes

1 Corner 1976.

2 Vaughan 1970.

3 Pagnoux et al. 2013.

4 IBPGR 1988.

5 The latest list of descriptors no longer considers the biometric parameters of the seeds to be decisive indicators: IPGRI 1999.

6 2013.

7 Size aspects were included in the analysis because, to the authors’ knowledge, there are no studies that clarify the role of environmental and genetic factors in the development of Citrus genus seeds (cfr. Rovner, Gyulai 2007; Costa et al. 2011).

8 Kuhl, Giardina 1982; Rohlf, Archie 1984.

9 For a summary see Costa et al. 2011.

10 Orrù et al. 2013; Ucchesu et al. 2015.

11 Citrus species have different degrees of genetic diversity: it is greater for C. reticulata, C. medica and C. maxima and lower for the other species, but for this first study we did not consider this aspect.

12 In agreement with: Rovner, Gyulai 2007; Pagnoux et al. 2015.

13 Mineralisation and waterlogging usually leads to the loss of the seeds interior and/or of some external characteristics, while charring often affects the seeds shape; therefore it is possible to compare these archaeological seeds to the controlled modern collection.

14 Fiorentino, Marinò 2008.

15 Pagnoux et al. 2013.

16 Fiorentino et al. 2014.

17 Pagnoux et al. 2013.

18 Kuhl, Giardina 1982.

19 Iwata, Ukai 2002.

20 In agreement with Ucchesu et al. 2015.

21 The descriptors selected with this procedure are: Perimeter, a3, d2, Perimeter ratio (Convex perimeter/Crofton perimeter), c2, Solidity degree, Shape factor, Feret ratio (Minimum diameter/Maximum diameter), Maximum ellipse axis, Minimum ellipse axis, Area, d7.

22 In agreement with Pagnoux et al. 2013.

23 Yoshioka et al. 2004.

24 2013.

25 Nicolosi et al. 2000; Penjor et al. 2013.

26 Wu et al. 2014.

27 Gulsen, Roose 2001.

28 Barkley et al. 2006; Bayer et al. 2009.

29 Curk et al. 2015.

30 Cfr. Pagnoux et al. 2013.

31 2011: 88.

32 2013.

33 Pagnoux et al. 2013.

Table des illustrations

Titre Fig. 1 - Measurements of 30 modern Citrus species seeds.
URL http://books.openedition.org/pcjb/docannexe/image/2211/img-1.jpg
Fichier image/jpeg, 96k
Titre Fig. 2 - PCA performed on the 3 biometric features of the modern dataset.
URL http://books.openedition.org/pcjb/docannexe/image/2211/img-2.jpg
Fichier image/jpeg, 264k
Titre Fig. 3 - Variation of Citrus seed shape within each species and the effect of each principal component. Each line represents the case that principal components takes ±2σ and mean, respectively.
URL http://books.openedition.org/pcjb/docannexe/image/2211/img-3.jpg
Fichier image/jpeg, 256k
Titre Fig. 4 - Linear Discriminant Analysis plot (the centroid of each species is represented).
URL http://books.openedition.org/pcjb/docannexe/image/2211/img-4.jpg
Fichier image/jpeg, 108k
Titre Fig. 5 - Modern species and archaeological seeds from Pompeii and Rome (Principal Component Analysis plot).
URL http://books.openedition.org/pcjb/docannexe/image/2211/img-5.jpg
Fichier image/jpeg, 72k
Titre Fig. 6 - Cluster analysis plot from modern species and archaeological seed data.
URL http://books.openedition.org/pcjb/docannexe/image/2211/img-6.jpg
Fichier image/jpeg, 118k

Auteurs

Laboratorio di Archeobotanica e Paleoecologia – Università del Salento, Lecce, Italy

Dipartimento di Chimica, Università degli Studi, Bari, Italy

Laboratorio di Archeobotanica e Paleoecologia – Università del Salento, Lecce, Italy

© Publications du Centre Jean Bérard, 2017

Conditions d’utilisation : http://www.openedition.org/6540

Acheter