Study area

The study was carried out in two clonal test with different environments. The first is located on land owned by Agropecuaria Santa Genoveva, in San Francisco, state of Campeche (Trial 1) at 19º33’26.53” N, 90º01’33.96” W, at 82 masl. The soil is of the Litosol type, with a clay texture. Rain precipitation is 1 300 mm per year and the average annual temperature is 23 °C. The second test was established on a private property in the Isla municipality, state of Veracruz (Test 2) at 18º04'50.21" N, 95º32'0.75" W, at an altitude of 56 m. The soil of the place is of the Cambisol type, with a sandy texture, precipitation of 2 000 mm per year and average annual temperature of 25 °C (^{Sánchez-Monsalvo et al., 2019}).

To compare some characters, the same genotypes from the clonal bank of the El Palmar Experimental Field of the Instituto Nacional de Investigaciones Forestales Agrícolas y Pecuarias (INIFAP) in El Palmar, Tezonapa, Veracruz were used, which is located at 18º30'19.29" N, 96º4515.48 "W at an altitude of 180 m. The type of soil is Acrisol with a sandy clay loam texture, annual precipitation of 2 888 mm and mean annual temperature of 24.4 °C (^{Sánchez and Velázquez, 1998}).

Design and establishment of clonal tests

The plant material used to establish the clonal trials was propagated by grafting outstanding genotypes from different provenances: (Bacalar, Quintana Roo; Calakmul, Campeche; Cardel, Cárdenas and Teapa, Tabasco; Comalapa and Palenque, Chiapas; Misantla, Tezonapa, Tierra Blanca, Tinaja, Papantla, and Los Tuxtlas, Veracruz; Lombardo and Tuxtepec, Oaxaca; Yucatán; Zona Maya, Unknown; Non-specific México and Guatemala), determined by means of progeny-provenance tests carried out by INIFAP. The two clonal trials were established in 2012 in a randomized complete block experimental design with 90 clones and six replications; the clonal bank was established in 2013 with 25 replications per clone.

Assessed variables

The morphological descriptors described by ^{Méndez and Sánchez (2012)} were used. Four complete leaves were taken from the middle part of the crown. A sample for each cardinal point of the tree was evaluated and the data was averaged. The growth variables and morphological characters of the stem and crown were measured in the two experiments, but not in the clonal bank, since the trees are periodically pruned to obtain cuttings.

Quantitative characters

The total height (m) was measured with a Haga altimeter and the normal diameter (1.3 m high) of the stem (cm) with a Jackson MS 283D/10M diametric tape. The volume of the stem (dm³) was estimated with the ^{Sánchez and García’s (2009)} equation:

The straightness of the stem (degrees) was evaluated from the deviation of the stem with respect to the vertical line that it should follow, for which the Spirit level application of a mobile phone was used, placed at 1.5 m high from the tree and the observation it was expressed in degrees. The length of the leaf (cm), its width (cm) and the length of the petiole (cm) were measured with a Truper TFC-30ME graduated tape, Petiole length (mm), leaflet length (mm) and secondary leaflet width (mm) were measured with a Digimatic D 6 P Vernier caliper. In addition, the number of pairs of secondary leaflets and the shape index of the secondary leaflet (ratio between length / width) were determined.

Qualitative characters

The shape of the stem at a height of 1.30 m from the ground was categorized into: semicircular (1), rhomboid (2), deltoid (3), ellipsoid (4), ovate (5), pentagonal (6) and circular (7). The type of branching was classified as monopodial (1), sympodial (2), dichotomous (3) and polycotomic (4). The orientation of secondary leaflets was upwards (1), outwards (2) and downwards (3). The arrangement of secondary leaflets was evaluated as alternate (1), opposite (2) and alternate-opposite (3). The secondary leaflet shapes were falcate (1), lanceolate (2), ovate (3), elliptical (4) and oblong (5). The shape of the leaflet base was categorized as oblique (1), rounded (2) and attenuated (3). Finally, the shape of the apex of the secondary leaflet was evaluated as acute (1), muchronous (2), cuspid (3), caudate (4) and awned (5).

Statistical analyses

Statistical analyzes were performed for each test, for the data as a whole from the two experiments and for the clonal bank with the SAS statistical package (^{SAS, 2002}). In order to evaluate the repeatability of morphological characters, that is, that they are homogeneous in size and shape between branches of a clone and different between clones, the following analyzes were carried out: for the values of the seven qualitative descriptors, the frequency distribution, for the 12 quantitative characters the mean, maximum, minimum value and coefficient of variation were obtained. For the 19 descriptors as a whole, the Principal Components analysis (PCA) was performed to determine which ones contribute the greatest weight to the variance. Once the PCA was performed, the descriptors that contributed low values to the components were excluded and the analysis was made again with the seven quantitative descriptors that contributed the most to the components in the clonal tests and four for the clonal bank. With the seven descriptors with the highest weight in the PCA, a clustering analysis was carried out with the average link method (AVERAGE). To avoid multicollinearity problems in the PCA, the volume variable was excluded since it presented a high Pearson correlation with diameter and height; likewise, the length of secondary leaflets was excluded because it showed a high correlation with leaf width.

Qualitative descriptor frequencies

The predominant descriptors in clonal trials 1 and 2 were the circular stem shape, the most desirable shape (42 and 41 %), and the type of sympodic branching (63 and 72 %). The type of desirable branching is monopodic, which registered a low frequency in both evaluation sites (<19 %).

The morphological descriptors with the highest frequency in the three evaluation sites (test 1, 2 and clonal bank) were: the arrangement of alternate secondary leaflets (61.6, 65 and 69 %), the lanceolate form of the secondary leaflet (95, 98 and 94 %), the oblique shape of the base of the secondary leaflet (75, 67 and 70 %) and the mucronate shape of the apex of the secondary leaflet (75, 69 and 69 %). The orientation of secondary leaflets was predominantly outward in trial 1 and in the clonal bank (38 and 56 %), while in experiment 2 the downward orientation predominated (54 %) (Table 2).

Principal Component Analysis

In the Principal Component Analysis based on the correlation matrix of seven quantitative characters in the clonal tests and four in the clonal bank, the first three components explained 77 % of the variance in test 1, 85 % in test 2 , 82 % in the joint analysis of the two clonal trials, and 95 % in the clonal bank (Table 3).

Component one contributed 36.4 % of the total variance in test 1. The characters that contribute significantly were leaf length, secondary leaflet width, leaf width, petiole length; this component is associated with secondary leaves and leaflets. Component two contributed 22.6 % to the variance and the characters with the most weight were the straightness of the stem, the diameter, the length of the leaf and the height of the tree; this component is associated with growth. Component three contributed 18.4 %, and the characters with the greatest weight were the length of the leaf and the width of the secondary leaflets (Table 4).

Component one contributed 45.6 % of the total variance in test 2. The characters with the most weight were the width of the secondary leaflets, the length of the leaf, the width of the leaf and the straightness of the stem. As in site 1, the component was associated with the leaf and secondary leaflets with negative weight in the straightness of the stem. Component two contributed 27.1 % to the variance, in which the straightness of the stem, the length of the leaf and the width of the secondary leaflets were the most determining characteristics; unlike site 1; in this case, diameter and height contributed less. Component three contributed 12.8 % to the variance and the characters with the greatest weight were the length of the leaf and the width of the secondary leaflets, as in test 1 (Table 4).

In the joint analysis of the two clonal tests, component one contributed 42.9 % of the variance, the width of the secondary leaflets, the length of the leaf and the width of the leaf and the straightness of the stem were the variables that most influenced. Component two contributed 25.2 % to the variance, the characters with the greatest weight were the straightness of the stem, the length of the leaf, the diameter, the length of the petiole and the height. Component three contributed 14.3 % to the variance in which, the most determining characters were also straightness, width of secondary leaflets and leaf length (Table 4).

Component one contributed 61.3 % of the total variance in the clonal bank. The characters with the most influence were the length of the leaf, the width of secondary leaflets, and the width of the leaf. Component two contributed 27 % to the variance and the characters with the most weight were the width of the secondary leaflets and the length of the leaf. Component three contributed 7 % to the variance and the width of the leaf and the length of the petiole are the determining characters (Table 4).

The previous results indicate that the contributions of the characters to the main components for each site (trials 1 and 2, their joint analysis and the clonal bank), from the first three components, the characters that coincide in contributing with the greatest weight are the straightness of the stem, the length of the leaf and the width of the secondary leaflets, for which there is stability in the characters of these components between the sites. The positive relationship between stem straightness and diameter also coincides, as well as between the length of the leaf and the width of the secondary leaflets, according to the Biplot diagram (Figure 1).

(a) clonal trial in Santa Genoveva, Campeche, (b) clonal trial in Isla, Veracruz and (c) in a clonal bank in Palmar, Veracruz according to the first two principal components.

Figure 1 Scatter plot of the 90 clones of Cedrela odorata L.  

Other studies have also shown the importance of quantitative leaf variables, such as length, width and area (^{Said et al., 2013};^{Santos et al., 2012}; ^{Toili et al., 2016}). However, the results obtained differ from those recorded by ^{Chimello et al. (2017)}, who described that the variables with the greatest variation were the shape of the leaf, the shape of the leaf apex and the presence or absence of the petiole in Tectona grandis L.

Cluster analysis

The clones that make up each group do not present similar or distinctive morphological characteristics for a particular clone nor are they from the same provenances, but they do have similarities between groups, in terms of growth characteristics and there is a positive relationship between trees with good growth performance and straightness (Table 5).

In test 1, group 11, composed of three clones, showed average values in volume of 170 dm³ and straightness of the stem of 86°; in this group are located two of the 18 clones with greater volume and greater straightness (67 and 95). Group 1 was the second with the highest average performance in volume, composed of 15 clones and was characterized by having clones of average volume of 128 dm³ and straightness of 85°; in this group clones 1, 60, 61, 64, 68, 89 and 108 are present, which are among the 18 with the highest average volume at the site.

Group 12, formed only by clone 98, registered the lowest average volume (70 dm³) and the lowest value of straightness of the stem (78°); this is one of the clones characterized by its lower performance. Other groups with a low average value of volume and straightness were group 8, with an average volume of 82 dm³ and straightness of 76°: in this group clones 57 and 83 meet; the first clone is one of the lowest average performance in volume and clone 6 that had an average volume value of 89 dm³ and a straightness value of 80°; this group has nine clones and eight of them are among the lowest values of the site's volume ranking (39, 73, 80, 86, 97, 103, 105, 116) (Table 6).

In test 2, group 6 showed a greater volume (57 dm³) made up of a single clone (46), and was identified among the 18 clones with the largest volume with a stem straightness value of 84°. The second best performing group was number 8, formed by three clones, in which two of the clones (48 and 63) with the largest volume (average 49 dm³) and a straightness of the stem of 84° were together. Group 12 recorded the lowest volume (27 dm³) and straightness of the minor stem (72°), formed by two clones. Another group with low volume was 7 (also made up of a single clone), with low volume as well (30 dm³), but with a higher stem straightness value (83°) (Table 6).

In the clonal bank, group 7, made up of nine clones, had the longest leaf length (40 cm), but the group with the highest value for the width of the secondary leaflets was group 6 (38 mm). Group 12 was formed by a single clone, with a leaf length of 19.6 cm and width of secondary leaflets of 23 mm (Table 6).

No clustering was observed between clones from the same provenance. The results coincide with the lack of clustering of nine provenances of Tectona grandis L. in southern India, and unrelated to morphological variables in a multivariate analysis (^{Sreekanth et al., 2014}).

Groupings between individuals of the same provenance have been determined in wild and cultivated materials or accessions with specific characteristics (^{Bayramzadeh et al., 2012}; ^{Said et al., 2013}; ^{Baldanzi et al., 2015}; ^{Toili et al., 2016}; ^{Anushka et al., 2017}; ^{Lo Bianco and Mirabella, 2018}). ^{Kjær (1996)} recorded the grouping of T. grandis among provenances according to the geographical origin of the families. Likewise, the correlation between the grouping and the geographical origin of six populations has been confirmed with individuals sampled in natural populations of Adansonia digitata L., whose morphometric differences are correlated with environmental factors (^{Assogbadjo et al., 2006}). In these studies, although such results have been recorded, only one site was analyzed, without considering the effect of the genotype x environment interaction.

Morphological variation is influenced by genetics and the environment that acts specifically on each genotype, which leads to adaptation and, with it, phenotypic plasticity (^{Galván et al., 2018}). The diameter, height and straightness of the stem, as well as the width and length of the leaf are under genetic control, but also by the local environment, so the effect of the site and its interaction with the genotype could explain the lack of clustering between clones or their provenances (^{Hovenden and Vander, 2003}).

Cedrela odorata has high values of genetic differentiation between populations, as well as high morphological variability, due to the fact that its natural environmental distribution is diverse (^{Navarro et al., 2004}). On the other hand, fragmentation and reduction of natural populations due to overexploitation and dysgenic selection suffered by the species may be the reasons for the high differentiation between populations, an issue that prevents genetic flow between populations (^{De la Torre, 2013}). Likewise, the species is capable of developing in environments with different environmental conditions (^{Pennington and Sarukhán, 2005}), which produces differential adaptation of the clones to different environments. This may partially explain the lack of consistency in morphological characters within the clones tested in the present study.

Among the reasons why ramets of the same clone presented differences in their morphological characteristics, it may be that the rootstock affects the development of the ramets, since each of these plants is a different genotype, which can generate a different response in each clone. (^{Schmutz and Lüdders, 1999}). The rootstock exerts influence on the adaptability, vigor and growth in each ramet (^{Hernández et al., 2014}), since an interaction is established between two genetically different individuals, which may even be incompatible (^{Forner, 1984}), which causes heterogeneity between ramets of the same clone (^{Darrouy et al., 2010}). On the other hand, to date there are no studies in which the age-age correlation is estimated, so it is unknown if in any other stage of the evaluated individuals it is possible to identify repeatability in the evaluated characters. There is also the effect of culture (C effect), which causes members of a group of relatives, such as a clone or a family, to present differences between them. The C effects can result from the location of the crook cut in the ortet or the environmental situation at the time when it was grafted (^{Forner et al., 1984}).

For example, the analyzes of the morphological characterization were not useful to obtain variables that may be important in the definition of descriptors of tejocote (Crataegus spp.) In a germplasm bank whose genetic material was obtained from sources of the State of Mexico, Puebla and Chiapas, and were produced by grafting (^{López-Santiago et al., 2008}).

The reduction of variables with the Principal Component Analysis was convenient since the first two components allow understanding 59 and 72 % of the variation in each of the evaluated sites. The quantitative characteristics were more relevant than the qualitative ones for the Principal Component Analysis, as they contributed more weight to the first two components.

Despite the stability in the weight that they provide to the main components, the morphological characters evaluated did not show repeatability between the clones. Due to this lack of repetition of the values of the evaluated characters of the ramets of the same clone, it was not possible to distinguish by means of these morphological descriptors a particular clone, nor the superior clones that are the most interesting for its distinction. In this way, it is not sufficiently advisable to use the characters as descriptors to request the registration of the intellectual property of such clones.

It is worth mentioning that in future studies it will be necessary to carry out tests of distinction, homogeneity and stability (DUS) requested by SNICS (^{SAGARPA, 1996}), only with outstanding clones that are those of interest for the registration and use of their germplasm. It is also necessary to continue exploring other alternatives such as molecular markers (^{SNICS, 2014}).

However, it must be considered that the assessed materials are wild, that they have been selected for their performance in other tests, but they are not materials that have entered an advanced domestication process as occurs in agricultural crops, and therefore, they have not fixed characters that can fully distinguish them. On the other hand, the rootstock must exert influence on the graft, because they are different genotypes and the ramet is the result of the interaction that occurs between both. Currently, the vegetative propagation of C. odorata is carried out by means of grafting, but, to obtain genetically equal clones and improve the evaluation of morphological characters, it will be necessary to carry out rooting of cuttings; this makes it possible to have only one genetic material per clone. Rooting cuttings of C. odorata show good results in small experiments (^{Sampayo et al., 2019}). However, operationally there are no large-scale works, so more research is required in this regard due to the importance of the species. The present study contributes to the formation of bases in the morphological characterization of the species and the registration of outstanding clones of C. odorata before the SNICS, which to date does not exist in the country.