Introduction
Polyembryony results from a biological process known as sporophytic apomixis, which involves the autonomous development of multiple embryos of nucellar origin (in addition to the zygotic or sexual embryo) in the same seed (Nakano et al., 2012; Kishore, 2014). This mechanism is a form of asexual reproduction through seeds resulting in offspring that are genetically identical to the mother plant (Conner et al., 2015; Zhang et al., 2018; Xu et al., 2021); moreover, it is a heritable trait found in many citrus varieties (Kepiro & Roose, 2010; Woo et al., 2019).
Polyembryony is important in citrus breeding because it is exploited for several purposes, such as rejuvenating old clones that have lost their vigor because of continuous vegetative propagation (Mondal et al., 2015), fixing valuable traits, and preserving the hybrid vigor of highly productive genotypes (Hand & Koltunov, 2014; Conner et al., 2015; Sailer et al., 2016; Fiaz et al., 2021). As it is a vegetative propagation generated through seed, recombination and segregation associated with sexual reproduction of hybrid plants are avoided (Wang et al., 2017). Furthermore, this mechanism is important because it represents a low-cost, disease-free, genetically identical rootstock propagation method (Kepiro & Roose, 2007; Shimada et al., 2018), which reduces the possibility of generating segregating zygotic plants, which may lose the quality attributes of that rootstock (Kepiro & Roose, 2007).
Polyembryony is a heritable trait, although the genetic mechanism governing it is not fully determined. García et al. (1999), when studying a progeny population from a cross between C. volkameriana × P. trifoliada, deduced that apomixis in citrus is controlled by six loci. In turn, Hong et al. (2001), based on the results of crossing monoembryonic mandarins (C. clementina) and polyembryonic oranges (C. sinensis), proposed a possible control mechanism of apomixis in the genera Citrus and Poncirus, which involves two complementary dominant genes (called A1 and A2).
Somatic embryogenesis processes in citrus have been reported to depend on one or a few genes (García et al., 1999; Hong et al., 2001; Kepiro & Roose, 2010). Furthermore, most polyembryony is most likely controlled by a dominant gene in the heterozygous state (Kishore, 2014), although modifier genes and minor genes are also present, which influence the proportions of polyembryonic seeds and the level of polyembryony, respectively (Kepiro & Roose, 2010; Simsek et al., 2018). Major and minor genes generate a diversity of genotypes that produce seeds with more than one embryo, a stable trait among commercial citrus varieties (Kepiro & Roose, 2010) and specific to each genotype with respect to a given region (Kashyap et al., 2018).
The percentage of polyembryonic seeds and the average number of embryos per seed can be used to identify monoembryonic genotypes or those with a low percentage of polyembryonic seeds, desirable attributes in breeding programs by conventional hybridization (Pérez-Tornero & Porras, 2008). This is because these genotypes facilitate obtaining hybrid plants; in addition, they can be used as varietal descriptors for the morphological characterization of accessions in germplasm banks (International Plant Genetic Resources Institute [IPGRI], 2000; Woo et al., 2019) and the identification of genotypes with potential for use as rootstock. In this case, genotypes with a high percentage of polyembryonic seeds (with nucellar embryos that generally give rise to uniform plants) are desirable, since, by maintaining their genetic uniformity, they retain their characteristics of tolerance/resistance to biotic and abiotic factors, compatibility with the variety and their level of polyembryony (Arruda et al., 2018).
Polyembryony also has undesirable aspects for citrus breeding, especially in conventional hybridization (Wang et al., 2017; Simsek et al., 2018; Xu et al., 2021). In polyembryonic seeds, embryos of nucellar origin are more vigorous and compete with the zygotic or sexual embryo for both space and nutrients within the seed, and generally eliminate it from early stages of their development (Moore et al., 1993). For this reason, monoembryonic parents, which produce only zygotic seedlings, are preferred by citrus breeders (Singh et al., 2018).
Different strategies or methodologies have been used to estimate polyembryony in citrus. Andrade-Rodríguez et al. (2004), Pérez-Tornero and Porras (2008) and Kashyap et al. (2018) used the direct embryo counting method on seeds obtained from mature fruit, while Shinde et al. (2007) and Kepiro and Roose (2010) estimated polyembryony indirectly by counting seedlings derived from mature polyembryonic seeds germinated in soil or substrates. Bowman et al. (1995) and Guerra et al. (2012) used both methods simultaneously.
Methods using germination of mature seeds (in substrate or soil) have the drawback that small embryos, contained in polyembryonic seeds, lack the energy to germinate and develop a seedling under these conditions (Andrade-Rodríguez et al., 2004); consequently, these ungerminated embryos are not detected and counted, resulting in an inadequate estimation of the number of embryos per seed and percentage of polyembryony. Woo et al. (2019) analyzed mature seeds from different citrus trees to determine whether they were monoembryonic or polyembryonic, but did not determine their level of polyembryony. Other studies have used ovules and microscopy techniques to elucidate the origin of nucellar embryos (Sánchez-Damas et al., 2006).
A good method for estimating the level of seed polyembryony should not underestimate the average number of embryos per seed or the percentage of polyembryonic seeds. To do this, it is necessary to determine the appropriate time for seed development and a methodology to sample and count the embryos that polyembryonic seeds have the capacity to produce. This is to minimize the number of embryos that cannot be detected and, therefore, are not counted.
Regarding the sample size to study polyembryony, Santos et al. (2015) used direct embryo counting in 15 rootstock species, and with the maximum curvature equations method determined that it was advisable to use a minimum of 23 seeds for the purposes of estimating the level of polyembryony and the average number of embryos.
To date, there have been no studies on the aforementioned variables of polyembryony in Mexican lime, nor a reliable method for their estimation. Therefore, the aim of this research was to find an efficient methodology to detect, count and estimate the potential number of embryos per seed, the percentage of polyembryonic seeds and the level of polyembryony in the ‘Colimex’ and ‘Lise’ varieties of Mexican lime, and the ‘Rosenberger’ variety of true lemon.
Materials and methods
The study was carried out between 2019 and 2021 at the facilities of the Tecomán experimental field of the National Institute of Forestry, Agricultural and Livestock Research (INIFAP), located in Colima, Mexico (18° 53’ 7 NL and 103° 50’ 30 WL). The study used four-year-old trees of the ‘Colimex’ and ‘Lise’ varieties of Mexican lime, and four trees of the 'Rosenberger' variety of true lemon, grafted on C. macrophylla Wester, all from the citrus germplasm bank and an experimental lot.
Four strategies for embryo counting were evaluated, which included two stages of seed development and two ways of making observations to estimate the number of embryos per seed and the percentage of polyembryonic seeds: 1) direct counting of embryos in mature seeds (DCEMS), 2) direct counting of embryos in immature seeds (DCEIS), 3) counting of seedlings derived from mature seeds germinated in vitro (CSMSGV) and 4) counting of seedlings derived from mature seeds germinated in substrate (CSMSGS).
Direct counting of embryos in mature seeds
On November 08, 2020 and April 21, 2021, well-developed, open-pollinated fruit with green skin were collected. For the ‘Colimex’ and ‘Lise’ varieties, the fruit had between 90 to 100 days of development from anthesis, and 120 to 130 days in the case of the ‘Rosenberger’ variety trees. All fruit were cut between 5 and 8 mm deep in the equatorial zone, and by turning the halves in opposite directions they were split in two and the seeds extracted. Under the lens of a stereo microscope (BA200, Motic®, China), and with the aid of dissecting forceps and scalpel, a longitudinal cut was made in the seed coats and the embryogenic masses were removed. Embryos were separated and arbitrarily classified into three categories: large (more than 5.0 mm), medium (3.0 to 5.0 mm) and small (less than 3.0 mm). For each seed, the number and size of embryos were recorded individually. A total of 346 seeds of ‘Colimex’, 276 of ‘Lise’ and 168 of ‘Rosenberger’ were analyzed.
Direct counting of embryos in immature seeds
On July 9 and 19, 2019 and April 6, 2021, tender, open-pollinated fruit, approximately 60 days from anthesis and with 1.6 to 2.4 cm in equatorial diameter, were collected for the ‘Colimex’ and ‘Lise’ varieties, and on May 20, 2019 and May 12, 2021, with 70 to 75 days from anthesis and 2.9 to 3.7 cm in equatorial diameter, fruit of the ‘Rosenberger’ variety were collected. The fruit were split in two, as mentioned above, and the immature seeds were extracted. Under the lens of a stereo microscope (BA200, Motic®, China), and with the aid of dissecting forceps and a scalpel, a longitudinal cut was made in the seed coats and the embryos were extracted. The number of embryos per seed was recorded, as well as their morphological development stage, which varied from globular, heart, torpedo to cotyledonary (Andrade-Rodríguez et al., 2004). A total of 273 seeds of ‘Colimex’, 190 of 'Lise' and 229 of ‘Rosenberger’ were analyzed.
Counting of seedlings derived from mature seeds germinated in vitro
On August 27, 2019 and September 23, 2020, fully-developed, open-pollinated fruit with green skin color and 90 to 110 days old from anthesis of the ‘Colimex’ and ‘Lise’ varieties, and 120 to 130 days old in the case of the 'Rosenberger' variety were collected. The fruit were superficially disinfected in the laboratory by immersion for 15 min in a solution of 2.0 % sodium hypochlorite plus 0.5 % Tween 20. Subsequently, under aseptic conditions, they were rinsed three times with sterile water, then split in two (as mentioned above) and the seeds extracted. Under the lens of a stereo microscope (BA200, Motic®, China), and with the aid of dissecting forceps and a scalpel, a longitudinal cut was made in the seed coats and the embryos or embryogenic masses were extracted. Immediately, two seeds were sown per 25 × 150 mm test tube. Murashige and Skoog (1962) culture medium, with 75 g·L-1 sucrose added, pH 5.7, solidified with 8.0 g·L-1 agar and autoclaved at 120 °C and 1.05 kg·cm-2, was used. Cultures were brought to a dark environment at 27 ± 2 °C for the first 12 days and subsequently exposed to natural light (1,000 Lux, measured with a luxmeter 840006, Sper Scientific, Taiwan). After 24 days, the number of seedlings emerged from each seed was recorded. A total of 172 seeds of ‘Colimex’, 172 of ‘Lise’ and 148 of ‘Rosenberger’ were analyzed.
Counting of seedlings derived from mature seeds germinated in substrate
For this methodology, some of the fruit collected in the previous experiment were used. A mixture prepared with silt-textured soil and coconut dust was placed in plastic seed start containers measuring 60 × 30 cm at the base and 40 cm high. Small furrows 0.5 cm deep and 4.0 cm apart were created. The seeds extracted the previous day were placed horizontally 3.0 cm apart inside the furrows and covered with a layer of sand. Sufficient irrigation was applied to moisten the substrate, and a shade structure, constructed with coconut palapas, was placed at a height of 20 cm. Irrigation was applied three times a week. After 35 days, the seeds were removed from the substrate and the total number of emerged seedlings, including malformed seedlings, was counted. A total of 220 seeds of ‘Colimex’, 184 of ‘Lise’ and 224 of ‘Rosenberger’ were analyzed.
Statistical analysis
From the data recorded for each variety and each method, four groups or replicates were made. To calculate the average number of embryos per seed, the total number of embryos detected was divided by the total number of seeds observed or, where appropriate, the number of seedlings emerged from germinated seeds was divided by the total number of seeds sown in vitro or in substrate. To calculate the percentages of polyembryonic seeds, the number of seeds with two or more embryos, or two or more seedlings, was divided by the total number of seeds observed and multiplied by 100. Percentage values were transformed into arc sine values using the equation: y = (arc sine (√(x/100) × 57.3). The frequencies of seeds that presented the different numbers of embryos detected were determined. Statistical analysis was done using a 4 × 3 factorial arrangement with four replications and a minimum of 43 seeds as the experimental unit. Means were compared using Tukey's test (P ≤ 0.05). The Statistix ver. 9 (2008) statistical software package was used.
Results and discussion
Each strategy designed for embryo counting required a different protocol, either by the stage of development of the seeds (mature or immature), by directly counting the embryos, by counting seedlings from germinated seeds or by the two conditions of seed germination (in vitro or in substrate).
Direct counting of embryos in mature seeds
In the three genotypes studied, monoembryonic and polyembryonic seeds were detected in the same fruit. Inside the polyembryonic seeds, masses of compacted and sometimes intertwined embryos were found (Figure 1a), although their separation was relatively simple, which facilitated their counting.
In polyembryonic seeds of the ‘Colimex’ and ‘Lise’ varieties, embryos of different sizes were found in the same seed. Generally, the smallest embryos were observed at the micropylar tip, although small embryos were also detected enclosed between the cotyledons of a larger embryo, or between other larger embryos. Polyembryonic seeds of ‘Lise’ and ‘Colimex’ had one or two large embryos, in proportions of 44.12 and 45.16 %, respectively, with respect to the total number of embryos recorded; of these seeds, 26.17 and 25.53 %, respectively, were of medium size, and 29.70 and 29.30 %, respectively, were of small size. The above indicates that nucellar embryos have asynchronous development within the seeds, which agrees with what was reported by Sanchez-Damas et al. (2006), who observed the same in three citrus rootstocks, and Kashyap et al. (2018) detected it in ‘Khasi’ mandarin (C. reticulata).
The number of embryos was variable even among seeds of the same fruit. Ranges of 1 to 10 embryos per seed were obtained in ‘Lise’, and from 1 to 13 in ‘Colimex’, which is somewhat consistent with what was reported by Kashyap et al. (2018) in 'Khasi' mandarin. In the ‘Rosenberger’ lemon, most seeds had only one embryo, although some seeds with two or even three embryos were detected. In the polyembryonic seeds of this variety, large embryos corresponded to 60.46 %, medium-sized embryos to 11.63 % and small embryos to 27.90 % of the total embryos observed.
Direct counting of embryos in immature seeds
Immature seeds in the three varieties had a soft consistency, a greenish-white color and a slightly translucent seed coat without lignification. Inside, the seeds presented a tissue of variable consistency depending on their stage of development, where the small embryos were found (Figure 1b). In the same fruit, seeds with different stages of maturity could be identified, suggesting that the seeds also have asynchrony in their development.
Embryos at globular, heart, torpedo and cotyledonary stages could be observed in the same seed. This confirms the asynchronous development of embryos, mentioned above in mature seeds, and agrees with what was reported by Kashyap et al. (2018) in ‘Khasi’ mandarin. With this method, from one to six embryos per seed could be detected in ‘Colimex’, from one to seven in ‘Lise’ and from one to four in ‘Rosenberger’.
Counting of seedlings derived from seeds germinated in vitro
In all three varieties, 100 % germination was achieved, as the seeds were favored by the in vitro environment. During germination, the embryogenic masses declumped, the embryos separated and the seedlings developed independently (Figure 1d), which facilitated their counting. Most embryos germinated and developed complete seedlings. Visually, seedling size varied with embryo size. Some very small embryos generated small and, in some cases, aberrant seedlings. The number of seedlings per seed varied from one to ten in ‘Colimex’, from one to eight in ‘Lise’ and from one to three in ‘Rosenberger’.
Counting of seedlings derived from mature seeds germinated in substrate
With this method, germination percentages of 90 and 93 % for ‘Colimex’ and ‘Lise’, respectively, and 75 % for ‘Rosenberger’ were recorded. Most of the seeds that germinated generated complete seedlings, with root, hypocotyl, epicotyl and two embryonic or cotyledonal leaves (Figure 1c). The number of seedlings per seed ranged from one to three in ‘Colimex’ and ‘Lise’, and from one to two in ‘Rosenberger’. These values are very low compared to those obtained with the DCEMS method.
Average number of embryos per seed
Table 1 shows the data on the average number of embryos per seed for the three varieties and the four sampling methods. Statistical analysis detected significant effects (P ≤ 0.01) for sampling methods, varieties and the interaction between both factors. The DCEMS and CSMSGV methods yielded the highest average number of embryos per seed, while CSMSGS had the lowest average. The differences observed between CSMSGV and CSMSGS may be due to the fact that many small embryos were unable to germinate under the conditions provided by the substrate and the environment, while the in vitro culture, with an aseptic medium rich in nutrients, favored the germination of more embryos. Therefore, the CSMSGS method may underestimate the average number of embryos per seed, mainly in Mexican lime varieties; consequently, it is the least recommended for estimating the value of this variable.
Factors | Embryos / seed (average) | Polyembryonic seeds (%) |
---|---|---|
Sampling methods | ||
DCEMS | 2.41 az | 59.09 a |
DCEIS | 1.89 b | 47.58 b |
CSMSGV | 2.33 a | 56.28 a |
CSMSGS | 1.42 c | 40.10 c |
CV (%) | 7.33 | 9.84 |
Varieties | ||
‘Colimex’ | 2.35 b | 68.43 a |
‘Lise’ | 2.53 a | 71.56 a |
‘Rosenberger’ | 1.16 c | 12.31 b |
CV (%) | 7.33 | 9.84 |
DCEMS = direct counting of embryos in mature seeds; DCEIS = direct counting of embryos in immature seeds; CSMSGV = counting of seedlings derived from mature seeds germinated in vitro; CSMSGS = counting of seedlings derived from mature seeds germinated in substrate; CV = coefficient of variation. zMeans with the same letters within each column do not differ statistically (Tukey, P ≤ 0.05).
Another interesting aspect is that with the DCEMS method more embryos per seed were obtained than with DCEIS. This is because with DCEIS the samples were taken from seeds at an early stage of development, where the potential number of embryos in the seeds has not yet been reached and, consequently, only those with a size visible with the stereoscope were detected. Therefore, the DCEMS and CSMSGV methods proved to be the most suitable for estimating the average number of embryos per seed in the varieties studied.
The varieties were also found to be different from each other. ‘Lise’ achieved the statistically highest values for average number of embryos per seed. This is one more of the differences between ‘Lise’ and ‘Colimex’, in addition to others reported by Robles-González et al. (2014) and Robles-González and Manzanilla-Ramírez (2012). On the other hand, ‘Rosenberger’ statistically presented the lowest average.
Interaction of sampling methods with varieties was detected (Figure 2a). Although the Mexican lime varieties behaved similarly in the three methods studied, different statistical averages for the number of embryos per seed were recorded with CSMSGV. In addition, it was observed that the highest average number of embryos per seed in ‘Colimex’ was achieved with DCEMS, followed by CSMSGV, while ‘Lise’ had an inverted relationship between these same methods, although statistically the two varieties were not different. In both varieties, the lowest averages were obtained with the CSMSGS method. The ‘Rosenberger’ variety showed similar behavior with the four sampling methods evaluated.
Percentage of polyembryonic seeds
The statistical analysis detected significant effects (P ≤ 0.01) of the sampling methods, varieties and the interaction between both factors for the percentage of polyembryonic seeds. The DCEMS and CSMSGV methods reached the highest values in this variable and were statistically different from DCEIS and CSMSGS (Table 1). The last method presented the lowest percentages of polyembryonic seeds. As already mentioned, the difference between CSMSGV and CSMSGS may be due to the fact that the very small embryos did not have the ability to germinate in the natural environment provided by the CSMSGS method, while with CSMSGV, the seeds managed to germinate and develop seedlings due to the optimal temperature, moisture and nutrient medium conditions. In the case of the DCEIS method, the lower percentage of polyembryonic seeds, compared to DCEMS and CSMSGV, could be due to the fact that in the immature seeds the process of formation and development of nucellar embryos had not finished.
It is important to note that a method for determining the level of polyembryony should not underestimate the average number of embryos per seed or the percentage of polyembryonic seeds. According to the results, DCEMS and CSMSGV recorded the highest averages for these variables, and, therefore, they can be considered adequate to determine the levels of polyembryony in the varieties studied.
The ‘Lise’ and ‘Colimex’ varieties showed high percentages of polyembryonic seeds and were statistically similar to each other. This suggests that, although morphological and physiological differences have been noted in the trees of these varieties (Robles-González et al.,2014; Robles-González & Manzanilla-Ramírez, 2012), they maintain the same level of polyembryony. On the other hand, the ‘Rosenberger’ variety averaged the lowest percentage of polyembryonic seeds. Likewise, interaction between the factors sampling methods and varieties was detected in this variable (Figure 2b). Mexican lime varieties behaved similarly in three methods, although they were statistically different with CSMSGV. In ‘Lise’, the highest percentage of polyembryonic seeds was achieved with CSMSGV, while in ‘Colimex’ the highest value was obtained with DCEMS. The ‘Rosenberger’ variety had similar behavior in the four sampling methods studied.
Level of polyembryony
To analyze the level of polyembryony, only the data obtained with the DCEMS method were used, since it was the one that best estimated the number of embryos per seed. As can be seen in Table 2, ‘Colimex’ and ‘Lise’, which belong to the species C. aurantifolia, achieved similar and relatively high averages of embryos per seed. On the other hand, ‘Rosenberger’, of the species C. limon, averaged low values and was statistically different (P ≤ 0.01) from the Mexican lime varieties. For the variable percentage of polyembryonic seeds, ‘Colimex’ and ‘Lise’ presented high values and were statistically similar to each other, while ‘Rosenberger’ had the lowest average and was statistically different (P ≤ 0.01) from the Mexican limes. The percentage recorded for ‘Rosenberger’ coincides with what was observed by Pérez-Tornero and Porras (2008) in Italian-type lemons.
Factors | Embryos/seed (average) | Polyembryonic seeds (%) |
---|---|---|
‘Colimex’ | 3.21 az | 80.63 a |
‘Lise’ | 2.90 a | 78.27 a |
‘Rosenberger’ | 1.33 b | 18.37 b |
CV (%) | 7.64 | 11.07 |
CV = coefficient of variation. zMeans with the same letters within each column do not differ statistically (Tukey, P ≤ 0.05).
According to Kishore (2014), and considering the level of polyembryony, species and varieties can be grouped into three categories: 1) slightly polyembryonic (≤ 25 %), 2) moderately polyembryonic (25-50 %) and 3) highly polyembryonic (> 50 %). Therefore, considering the results of this study, ‘Colimex’ and ‘Lise’ are highly polyembryonic (with values of 80.65 and 78.27 %, respectively), while ‘Rosenberger’ is slightly polyembryonic (18.37 %).
Variation in the number of embryos per seed
For this variable, the data obtained with the CDESM method were also used. It was found that the number of embryos was highly variable between ‘Colimex’ and ‘Lise’ seeds. This agrees with Kishore (2014), who points out that the number of nucellar embryos varies between seeds and between species.
Seeds were grouped according to their number of embryos (Table 3), and the following morphotypes were generated as described by Kishore (2014): one (single), two (twins), three (triplets), four (quadruplets), five (quintuplets), six (sextuplets), seven (septuplets), and eight (octuplets) embryos per seed. Sporadically, seeds with 13 embryos were observed in ‘Colimex’ and 10 embryos in ‘Lise’. Seeds with two and three embryos reached the highest frequencies, but as the number of embryos per seed increased the frequencies decreased. Andrade-Rodríguez et al. (2004) found a similar trend in C. volcameriana, whereas Kishore et al. (2012) note that seeds with four and three embryos are the most frequent in apomictic citrus, regardless of the genotypes they studied. In ‘Rosenberger’ lemon, diversity was minimal, as most seeds (81.63 %) had only one embryo, a low proportion of seeds contained two embryos and, sporadically, seeds with three or five embryos were found.
Variety | Frequency of seeds with different numbers of embryos (%) | |||||||
---|---|---|---|---|---|---|---|---|
1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | |
‘Colimex’ | 19.37 bz | 26.89 a | 18.50 a | 12.13 a | 10.11 a | 6.63 a | 2.31 a | 1.44 a |
‘Lise’ | 21.72 b | 25.01 a | 19.60 a | 18.85 a | 7.94 a | 5.43 a | 0.37 a | 0.72 a |
‘Rosenberger’ | 81.63 a | 14.89 b | 3.00b | - | 0.49 b | - | - | - |
CV (%) | 7.88 | 11.98 | 17.11 | 21.77 | 39.08 | 28.29 | 122.13 | 89.12 |
CV = coefficient of variation. zMeans with the same letters within each column do not differ statistically (Tukey, P ≤ 0.05).
Conclusions
The DCEMS and CSMSGV methods allowed detecting and recording the highest averages for number of embryos per seed, as well as the highest percentages of polyembryonic seeds in the Mexican lime varieties ‘Colimex’ and ‘Lise’ and the true lemon variety ‘Rosenberger’.
The ‘Colimex’ and ‘Lise’ varieties are considered highly polyembryonic, presenting 80.65 and 78.27 %, respectively, of seeds with two or more embryos, while the ‘Rosenberger’ variety is slightly polyembryonic with only 18.37 %.