Maize landraces management to avoid transgenic contamination, decreases yield and seed quality

Producing high-quality seeds for agroecological agriculture is needed because the plants will be exposed to a diversity of environments and management practices. Sowing times and population densities can influence the quality of seeds produced by landraces. Further, seeds for agroecological cultivation must not be contaminated by transgenes. In this context, it is necessary to know the agronomic performance of landraces outside of the recommended growing period to avoid gene flow with transgenic cultivars. Thus, the objective of this work was to determine the germination rate, vigor and sanitary quality of seeds of maize genotypes cultivated in an agroecological system at different sowing times and densities. A split-split-plot design was used in the 2016/2017 and 2017/2018 growing seasons. The physiological quality evaluation was based on germination and vigor tests, and the sanitary quality was obtained by visually examining the seeds. The results showed a significant simple effect of genotype, sowing time and density, as well as double and triple interactions among factors for all evaluated traits. The most suitable time for sowing the corn landraces evaluated in an agroecological system in western Santa Catarina is September, when the seeds exhibited the highest percentage of germination and vigor and there was a lower incidence of pathogens. At this time, a density of 55,000 plants·ha–1 had the best result for germination. Late sowing was tested to avoid gene flow from transgenic varieties that usually are sown in September, but the results indicated penalties, such as a decrease in seed yield and quality.


INTRODUCTION
Brazil, in addition to having high biodiversity, is a major food producer and part of this food comes from organic and agroecological systems. Agroecological principles, such as crop diversification, heterogeneous varieties, organic soil management and water conservation, reduce vulnerability to climate change effects (Altieri et al. 2015).
In these agricultural systems, instead of highly homogeneous cultivars, farmers use landraces or local varieties of heterogeneous genotypes (Zeven 1998). Specifically, in the municipalities of Anchieta and Guaraciaba, in western Santa Catarina state, Brazil, it was found there are 2049 small farms with 1513 populations of maize landraces (Costa et al. 2017). The authors note that the region is a microcenter of Zea mays L. Small farmers across the country continuously save, reuse and exchange landrace seeds with neighbors, which favors genetic diversity and in situ on-farm conservation. Harlan (1951) designed microcenters of diversity when a significant diversity is accumulated within a very restricted geographical area.
In this scenario, landraces become important seed sources in all agricultural systems, favoring the adoption of sustainable practices. In addition, landraces can produce more than 7.0 tons per hectare in organic systems (Balbinot Junior et al. 2005;https://doi.org/10.1590/1678

Experimental design
The seed production field was on a small family farm located in Chapecó, Santa Catarina, Brazil (27°03'21"S; 52°40'07"W). The experiment, installed in an area at least 500 m away from another maize crop, was carried out in split-split-plot with three replications. The main plot represented the sowing times, the split-plot represented the genotypes, and the splitsplit-plot represented the sowing density. The experiments were installed during the first half of September (sowing time 1 -ST1), first half of November (sowing time 2 -ST2) and first half of January (sowing time 3 -ST3) in the 2016/2017 and 2017/2018 growing seasons. The sowing densities used were 2.5, 4.0, 5.5 and 7.0 plants·m -1 (thereafter SD1, SD2, SD3 and SD4, respectively). Three landraces (Branco Precoce, Taquara and Língua de Papagaio) cultivated by farmers in the region, and an open pollination variety (SCS 155 Catarina), was used as a control for the growing region.
The ears of ST1, ST2 and ST3 were harvested in the first half of January, mid-March and the first half of June for each growing season, when the seeds reached 20% moisture content. The ears produced in each plot were harvested manually, threshed and the seeds were dried in the shade until they reached approximately 13% moisture content. An average seed sample of 1000 g of each treatment was homogenized and divided to obtain four repetitions of 250 g (Coelho et al. 2010), which were used in assessments of physiological and sanitary quality. The physical and physiological analyses of the seeds were carried out at the Seed Analysis Laboratory of the Agroveterinary Center, Universidade para o Desenvolvimento do Estado de Santa Catarina.

Germination test (GT)
Fifty seeds were used per repetition, arranged on sheets of germitest paper as the substrate. The paper sheets were previously moistened with distilled water in the amount of two and a half times the weight of the dry paper. After the seeds were scattered, the paper sheets were placed in a Magensdorf-type germinator at 25 °C. The first count was performed on day four and the final count was performed on day seven after the test initiated (Brazil 2009).

Accelerated aging (AA)
Accelerated aging was used to determine the vigor (Marcos-Filho 2005). Again, 50 seeds per repetition were placed to age in a plastic gerbox containing 40 mL of distilled water. The seeds were distributed in a single and uniform layer on a stainless-steel screen at the bottom of the gerbox. The boxes were kept in an aging chamber at 45 °C for 72 h. After the aging period, the seeds were placed to germinate; after four days, an evaluation of normal seedlings was carried out as described in the Rules for Seed Analysis (Brazil 2009).

Health test
The seeds of each repetition were examined visually and those that presented typical symptoms or signs of pathogenic agents, such as Fusarium spp., Stenocarpella maydis and Stenocarpella macrospora, were separated and counted (Brazil 2009). The results (incidence of pathogens) were expressed as a percentage of the ratio seeds with pathogens to total evaluated seeds.

Weather data
The monthly average of rainfall, radiation and temperatures from January 01, 2008 through December 31, 2018 were obtained from the Foz do Chapecó Energia S.A. In addition to the same data specifically from September 1, 2016 through May 31, 2018, the average minimum temperatures and average maximum temperatures were obtained from the automated weather station (27°05'24"S; 52°38'05"W), located at Empresa de Pesquisa Agropecuária e Extensão Rural de Santa Catarina, Chapecó, SC, Brazil (http://ciram.epagri.sc.gov.br/) 5 km apart from the experimental. The weather data ( Fig. 1) were used to discuss the obtained results.

Statistical analyses
The germination test and accelerated aging data were transformed using arcsen √x / 100 to meet the theoretical assumptions of the F-test (normality and homogeneity). Then, the data were subjected to a joint analysis of variance and when a significant difference was detected by the F-test, their means were compared by Tukey's test, p < 0.05 of significance. The statistical analyses were performed using the program R (R Core Team 2016).

RESULTS AND DISCUSSION
The analysis of variance indicated that there was a significant triple interaction between sowing times × genotypes × densities for all evaluated variables (data not shown). Since the field experiments were carried out twice in the area, weather data can contribute to explain the obtained results.  The evaluated sanitary attributes in the 2016/2017 growing season indicated that September and November are the sowing times with the lowest incidence of pathogens (Fig. 3a, c, e, g), regardless of genotype and sowing density. Similar results were found for the 2017/2018 growing season (Fig. 3b, d, f, h), for ST1, again with genotypes showing the lowest incidence of pathogens compared to ST2 and ST3. However, the incidence of pathogens on maize plants sown in November varied according to sowing density. The greater the sowing density, the higher the incidence of pathogens in the seeds. The highest incidence of pathogens was when seeds were sown in January, regardless of the crop and sowing density. The results indicate that late sowing favors the development of pathogens, which negatively affects the health quality of the seeds produced. Similar results were found for other maize genotypes, such as a super early hybrid in a monoculture, where a delay in sowing from the beginning to the end of spring increased, synergistically, the incidence of pathogens (Coelho et al. 2019).
Throughout the experiment of the first and second sowing times, the average temperatures were above 20 °C, which is above the minimum temperature necessary for good development of maize plants (Fig. 2). However, for the third sowing time, the average temperature fluctuated, with values below 20 °C at the end of the crop cycle (April to June).
The combined effect of sowing season and genotype on incidence of pathogens on harvested seeds was also variable in the genotypes. When seeds were harvested from plants from the November sowing time of the 2017/2018 growing season, the incidences of pathogens at densities of 4.0 and 5.5 plants·m -1 were 4.5 and 5.5%, respectively, for Língua de Papagaio. This differed from the 15.0 and 12.5% averages for Taquara and 14.5 and 8.50% averages for SCS 155 Catarina, respectively at densities of 4.0 and 5.5 plants·m -1 , the two genotypes with the highest seed incidence of pathogens for these sowing densities. The main pathogens associated with maize seeds responsible for physical damage (discoloration of the seeds, reduced carbohydrate content, proteins and total sugars) are Stenocarpela spp., Fusarium spp., Gibberella zeae, Penicillium spp., and Aspergillus spp. (Mendes et al. 2011;Siqueira et al. 2014).
Previously, in a study carried out with the same genotypes and environmental conditions as the present one, it was verified that the later the sowing is carried out, the greater the percentage of lodged and broken plants (Munarini and Nodari 2021). According to these authors, if both lodged and broken plants are taken together, the percentage of 4.6% in sowing carried out in September increased to 25.36% in sowing in November and to 36.52% in sowing in January. The lodged and broken plants are associated with the occurrence of rot that can be aggravated when the crop is sown late, due to the lower supply of radiation and better agroclimatic conditions for infection of the fungi that cause this problem (Sangoi et al. 2010) Overall, for seed germination, higher germination percentages were observed for the September (91%) and November (87%) sowings, compared to sowing in January (79%), for both analyzed growing seasons (Fig. 4). The smallest variation in germination percentages among genotypes was observed when sowing was carried out in September.
Specifically, the highest percentages of seed germination were observed in the sowing of September and November, compared to the sowing of January, in the two analyzed harvests; however, the genotypes showed different values for the germination percentage across densities and growing seasons (Fig. 4). The Branco Precoce is highlighted, because it showed statistically higher germination rate in September than in January in all evaluated treatment combinations except the sowing at density of 2.5 plants·m -1 of 2016/2017 (Fig. 4a). A similar result was exhibit by the SCS 155 Catarina genotype, which showed statistically superior germination rate in September compared to January in 75% of evaluated treatment combinations. For the genotypes Língua de Papagaio and Taquara, the superiority of the germination rate in seeding in September comparatively to January was proven in 50% of the treatment combinations. Overall, considering the minimum percentage of germination required for the commercialization of seeds in Brazil (85%), 12% of the combination of treatments sown in September, 28% in November and 59% in January did not reach the minimum.
In the present study, a reduction in the percentage of seed germination was observed in seeds produced in second and third sowing times, regardless of the sowing density used. This result may be associated with a longer period of exposure to environmental factors until the seeds reach adequate humidity to harvest (20%). Noal et al. (2014) studied the harvesting time of three maize landraces and found that the ideal harvest humidity for seeds of the genotypes varied from 20 to 30%, since this humidity range had the highest percentages of germination and vigor. In the present study, when sowing was carried out in January, the incidence of diseases was higher compared to the other two sowing times, which reduced the percentage of seed germination. These results indicate that, for the studied genotypes, the ideal sowing time to obtain higher germination percentages is at the beginning of the period indicated for the climate zone for cultivating the species, which is September. In addition, sowing in January is not indicated to obtain seeds with the standard quality required by Brazilian law.      The genotypes tested in the present study differed for percentage of seed germination, demonstrating that there is variability among them (Fig. 4). The performance of the genotypes when sown in September was similar in both growing seasons. The cultivar SCS 155 Catarina exhibited an average germination of 95% in both growing seasons. For the same sowing time (ST1), seed germination was 92% for the Taquara and 90% for Branco Precoce genotypes in the two growing seasons. The Língua de Papagaio genotype had the lowest performance, with an average germination of 85% in both growing seasons. For the second sowing time (ST2), the germination percentages were different among the genotypes in the distinct growing seasons evaluated. While Língua de Papagaio still had the lowest performance in the first growing season, with an average seed germination of 77%, in the second season (2017/2018) its performance was superior to Taquara and SCS 155 Catarina. Variations in genotype performance for each crop were also observed when seeds were sown in January (ST3), reaffirming the existence of diversity among genotypes that are exposed to different environmental factors.
Significant differences among maize landraces for percentage of germination were also verified by Queiroz et al. (2019) as one of the selection criteria in their breeding program. In addition, genetic diversity for physiological seed quality among genotypes was observed by Nerling et al. (2013) when analyzing intervarietal crosses of maize. On the contrary, Martin et al. (2018) evaluated maize seeds of open pollinated varieties and hybrids produced in organic systems and found no difference in germination percentages that were greater than 95%.
In the present study, it was verified that the effect of sowing density varied according to the growing season. Considering the best sowing time (September), in the 2016/2017 growing season, Branco Precoce and Língua de Papagaio had the highest germination percentages at a density of 5.5 plants·m -1 . The genotypes Taquara and SCS 155 Catarina maintained the same germination percentage even at the density of 7 plants·m -1 . In the 2017/2018 growing season, the genotypes Branco Precoce, Língua de Papagaio and Taquara exhibited a reduction in percentage of germination as the sowing seed density increased. However, the SCS 155 Catarina genotype was not affected by density, which is associated with the selection done during the cultivar development. Overall, the results suggest that the maize landraces do not tolerate dense cultivation (or intraspecific competition), especially when sown late in the season. Managing this feature relies on the practices of small farmers, where no high sowing density is used to produce grain, particularly in agroecological or organic polyculture systems.
In the evaluation of seed vigor (Fig. 5), the highest average vigor percentages were obtained from seeds sown in September and November, which were 80 and 68%, respectively, for the 2016/2017 growing season and 74% for both ST1 and ST2 in the following growing season. Sowing in January resulted in seeds with the lowest vigor performance, 66 and 56% in the two analyzed growing seasons, respectively. Percentages of vigor above 80% were verified only for sowing in September and November. Late sowing negatively affects the vigor of the seeds.
Among genotypes, the open pollinated cultivar SCS 155 Catarina and the Taquara landrace showed the highest percentage of seed vigor (87%) when seeds were sown in September at a density of 2.5 plants·m -1 . On the other hand, for the Branco Precoce and Língua de Papagaio landraces, when sowing was carried out in November at densities of 4 and 5.5 plants·m -1 the seeds had a higher percentage of vigor (82 and 89%, respectively). Língua de Papagaio was the most sensitive when sown in January, with an average vigor of 52%.
The effect of sowing density on vigor varied according to the growing season and sowing time. Seed vigor percentage varied from 82% for the density 2.5 plants·m -1 to 70% for the density 7.0 plants·m -1 , when sowing was carried out in September in the 2017/2018 growing season. For the other two sowing times, the behavior in relation to density varied according to the genotype. In general, the higher the sowing density, the lower the seed vigor. The fact that density reduces the penetration of solar radiation into the canopy was noted by Sangoi and Silva (2016) to explain why high maize densities accelerate leaf senescence, reducing the duration of the grain filling period and promoting the production of light grains. In addition, an adverse effect of high sowing density seed production and physiological quality was verified in the present study.
As found in a study with other genotypes, the use of inadequate sowing densities and a delay in seed harvesting decrease the physiological quality of seeds and favor the attack of insects and microorganisms, accelerating the process of seed deterioration (Henning et al. 2011). In addition, sowing at the beginning of the recommended period for a crop in a certain region tends to benefit seed production with the required sanitary standards, due to the low pressure from pathogens (Martin et al. 2007).   The present study brings insight to two main issues. First, it supports the hypothesis that it is possible to produce organic or agroecological seeds that meet the legal requirements for commercial sowing. This result became relevant because high quality seeds from maize landraces can be produced and marketed. Consequently, the entire crop cycle for agroecological or organic systems (Brazil 2003) can be accomplished.
Second, the management practices tested, like sowing delay, do not allow farmers to avoid contamination of their landraces by transgenic hybrids without penalties, in terms of seed quality and quantity. In the region of the present study, which is currently recognized as a microcenter of maize diversity in western Santa Catarina, Brazil (Costa et al. 2017), the spatial isolation, to avoid transgenic contamination of landraces, also cannot be applied because of the proximity among farms, since 17,1% of them (< 10 ha) produce maize landraces and the others nearby cultivate transgenic varieties (IBGE 2017).
Therefore, there is still no agronomic solution for small farmers that cultivate maize landraces in regions where transgenic hybrids are widespread without even minimal contamination. Earlier studies revealed the contamination of landraces after the legal release of transgenic varieties. Maize landraces from that maize diversity microcenter tested positive (19.1% out of 157 samples) for the presence of transgenic proteins (Nerling et al. 2014), which was present in transgenic cultivated varieties in that region. In Spain, maize seed production in an organic area decreased by 75% in Aragon between 2004 and 2007 and by 95% in Catalonia from 2002 to 2008 because of the transgenic contamination (Binimelis et al. 2008).
Genetic contamination is the result of the gene flow that is accomplished by cross-pollination or seed mixture. In addition to commercial deficits, the flow of transgenes and their introgression into the valuable germplasm of important crop plants can cause unexpected consequences, disturb the genetic conservation, and affect public policies and crop management systems (Ellstrand 2003). The present article tested management practices of maize landraces in an agroecological system to avoid cross-pollination. The best one was the same sowing time as used for transgenic varieties. Spatial isolation also could avoid gene flow by cross-pollination, but not in the studied area due to proximity of farms (Costa et al. 2017). However, gene flow through seed mixtures was firstly demonstrated by Quist and Chapela (2001) in México. In that episode, transgenic seeds that contaminated Oaxaca maize landraces came from outside of the country. Fifteen years later, Oaxaca maize landraces continued with the transgenes (Agapito-Tenfen et al. 2017). Additionally, the authors claimed that transgenes extent and frequency depend on the societal characteristics and seed management systems of the local communities. Thus, even the farmers accept change the management practices, seed mixtures could also cause transgenic contamination of their landraces.
Thus, without options for temporal and spatial isolation without penalties, transgenic contamination of cultivating landraces in regions where transgenic varieties were released could happen at a large scale, which compromises the efforts of thousands of farmers that are conserving significant genetic diversity not yet represented in an ex situ germplasm bank.
Two subjects are of great importance for the sequence of studies on the use of landraces. The first refers to the need to test other landraces, because in this study we tested three out of hundreds of them with potential for use. If many tested landraces become available for cultivation, farmers could choose more than one to grow and thus have less risk of losses due to biotic and abiotic stresses. The second subject is related to climate change. Although landraces were tested during two growing cycles, their behavior may not be repeated in years with different climatic conditions from those that occurred during this study.

CONCLUSION
Seeds of maize landraces produced in agroecological systems with higher physiological quality and lower incidence of pathogens, standards required by Brazilian regulation, were obtained for the September (beginning of spring) sowing time at a density of 5.5 plants·m -1 .
When seed sowing is postponed to November (middle of spring), not all landraces or all sowing densities tested provide seeds of high quality and with the required standard due to the adverse effects of the interaction of genotypes × densities × sowing times.
For the first time, evidence was found that postponing planting even further by sowing seeds in January to avoid contamination by maize transgenic hybrids not only the harvested seed acquire the legal standard and the desired quality to be sown next season, but also provoked lesser yield.
Overall, temporal isolation of landraces cultivated in agroecological or organic systems for seed production is not a solution for small farmers to avoid transgenic contamination of maize landrace because of loss of yield and seed quality.
Considering that there were differences between landraces in terms of vigor and seed germination rate in late plantings, it is recommended that farmers test several landraces adapted to their region on a small scale before commercial planting.