Study area

The study was conducted in the Coronango municipality (19° 05' 53.88" to 19° 10' 26.76" N, 98°15' 32.04" to 98° 19' 24.60" W and 2,190 masl), located in the Puebla Valley (^{INEGI, 2020}). The municipality is made up of three auxiliary boards (San Antonio Mihuacán, San Francisco Ocotlán, and San Martín Zoquiapan) and the municipal seat (Santa María Coronango). According to ^{INEGI (2010)} and ^{INEGI (2023)}, the climate is temperate sub-humid with summer rains; the average annual temperature is 14-18 °C and rainfall ranges between 800 and 1000 mm; two types of soil (Arenosol and Phaeozem) predominate in the area. Currently, urban areas are growing on land previously dedicated to agriculture.

Collection and genetic material

The collection of maize landraces was carried out between February and April 2021 in the Coronango municipality. The sampling frame was constituted by the list of peasants in the Coronango municipality registered as corn producers in the Producción para el Bienestar Program spring-summer 2020 cycle, made up of 207 peasants. With the information from this frame, a stratified random sampling technique with Neyman distribution (^{Singh & Mangat, 1996}) was applied, considering boards and head of the municipality as strata. The accuracy was set at 10 % concerning the mean, and the confidence (z_α/2) at 95 %. The sample size obtained was 63 peasants, distributed as follows: Coronango: 4, Mihuacán: 13, Zoquiapan: 14, and Ocotlán: 32. In each site, from the municipality register of peasants, ordered by auxiliary boards and municipal seat, the number of farmers corresponding to the one determined for each stratum was randomly chosen; once the peasant’s names were known, they were contacted to explain them the purpose of the study and maize samples were collected with those who agreed to provide them.

As a result of the collection, 54 samples of maize landraces were obtained (Table 1), which are grown under strictly rainfed conditions, with sowing dates concentrated in the month of May. All of them were included in the evaluation. Three accessions from the CIMMYT germplasm bank, representative of the Chalqueño (CIMMYTMA1398), Cónico (CIMMYTMA10232), and Elotes cónicos (CIMMYTMA1895) races, were used as racial controls. Additionally, six experimental materials of different grain color and earliness were included (three synthetics of white grain: 'Nopalucan Precoz', 'Nopalucan Ultraprecoz' and 'Serdán' and three composite varieties: LHM Blanco, LHM Amarillo, and LHM Azul) and one commercial variety (Niebla^®).

Experimental evaluation

To evaluate the 64 materials, a simple 8×8 lattice was used. The experimental unit consisted of two rows 5 m long and 0.85 m wide, with eleven hills per row, spaced every 50 cm. The useful plot was the total of the experimental unit. Experiments were established in three locations in the Coronango municipality: San Francisco Ocotlán (19° 08' 43.0" N and 98° 16' 20.0" W), San Martín Zoquiapan (19° 08' 10.0" N and 98° 18' 17.0" W) and San Antonio Mihuacán (19° 08' 34.0" N and 98° 18' 29.0" W) (^{INEGI, 2020}); sowings were carried out on May 19, 21 and 24, 2021, respectively.

Experiment management

At the time of sowing, three seeds were deposited per stroke, and 35 days later, thinning was made by two plants per hill. Sowing and crop management were carried out according to the conventional practices of farmers in the region, except for fertilization, for which a dose of 110N-60P-00K was used, applied fractionally: 40N-60P-00K in the first tillage (27 days after sowing) and the remaining nitrogen in the second tillage (40 days after sowing), using diammonium phosphate and urea as sources. Weed control was carried out with tillage and, after the second tillage, with the application of 2,4-D amine and atrazine at a rate of 1 L∙ha^-1.

Variables recorded

The variables days to 50 % tasseling (DTT) and silking (DTS) were quantified considering all the experimental unit, as described by ^{Rocandio-Rodríguez et al. (2014}). Once flowering was concluded, five plants with complete competition were randomly chosen in each plot, in which the vegetative variables described in Table 2 were recorded, as described by ^{Hortelano et al. (2012)}. The tassels of the five plants mentioned were collected to measure the listed attributes, following ^{Rocandio-Rodríguez et al. (2014)}. Finally, at harvest, a sample of five representative ears was taken from each plot, to measure in them and their grains the corresponding characteristics, as explained in the methodology described by ^{Hortelano et al. (2012)}. In addition, several indices were calculated, derived from the variables contained in Table 2.

Statistical analysis

A combined analysis of variance was performed for the 37 variables recorded and then, a Pearson correlation analysis was applied to identify pairs of correlated variables (r ≥ 0.7|), from which, for the subsequent multivariate analysis, the one considered most informative was chosen. With the averages per population for the 27 resulting variables, a cluster analysis was performed, based on a Gower distance matrix, using Ward's Minimum Variance as a grouping method. In the resulting dendrogram, using the pseudo-F criterion, four groups were identified, for which a multivariate analysis of variance was performed, followed by an analysis of variance by groups (considering the materials included as replicates) and a Tukey mean comparison test (P ≤ 0.05). Finally, a principal component analysis was also performed, from which a distribution graph of the materials was generated based on the values of the first three principal components, including the groups identified in the dendrogram. All statistical analyses were performed with the SAS OnDemand for Academics^® program (^{SAS Institute Inc., 2012-2020}).

In a rural area undergoing urbanization, there are still important levels of morphological diversity among cultivated maize.

The grain coloration pattern found in the Coronango municipality reveals that the types reported 50 years ago by ^{Cervantes & Mejía (1984)} for the Puebla Valley are maintained: white, blue, yellow, pinto and red, with the prevalence of white grain populations (70.4 % in residual moisture and 48.1 % in rainfed; 55.6 % at present) and the persistence of varieties with pigmented grain, although with a higher percentage of those of blue color at present (5.5 % in residual moisture and 12.4 % in rainfed in the collection of the years 1970-1971; 25.9 % in the present study). The later result could be attributed to a strategy that production units close to urban areas have generated, consisting of transforming the blue-colored grain into tortillas for sale (^{Lerner & Appendini, 2011}).

Even though the study was carried out in a geographically small space (compared to more extensive studies conducted in the Puebla highlands, such as those of ^{Hortelano et al., 2008}; ^{Hortelano et al., 2012} and ^{Alvarado-Beltrán et al., 2019}), which is also in a process of absorption by the urban environment (^{Guevara, 2017}), important levels of morphological variation were still detected among the landraces. Evidence of the latter is that it was possible to determine the existence of three large groups, which were differentiated from each other, particularly by grain color, earliness, and characteristics related to various plant structures (leaves, tassel, ear, and kernel). The importance of grain color and earliness as grouping attributes of landraces was documented by Muñoz (2013), who explained that within the micro-regions where maize landraces are grown, peasants have structured varietal patterns (understood as systems that combine groups of landraces, environmental strata or environmental levels and the relationships between them), whose components are defined precisely in terms of grain color and earliness. It is common for the components of the varietal pattern in maize to be represented by a set of white grain materials (late cycle) and others with pigmented grains (early cycle), as reported by ^{López et al. (2020}). The relevance of vegetative, tassel, ear, and kernel variables (such as leaf area, number of leaves above the ear, leaf index, width/length and length/width kernel ratios, ear row number and number of primary branches of the tassel, which contributed most to the explanation of the variation found) for group distinction has been demonstrated in previous reports, such as those of ^{López-Romero et al. (2005}), ^{Vidal-Martínez et al. (2018)} and ^{Linares-Holguín et al. (2019)}, among others.

As in previous reports by ^{Hortelano et al. (2008)} and ^{Hortelano et al. (2012)} in micro-regions of the central highlands of Puebla, in this research, the groups of landraces were also differentiated, first by grain color and then by plant characteristics. In agreement with those studies, it was found that plants from pigmented grain materials tended to present lower values in phenological, vegetative, ear, and kernel variables than those from white grain materials. These elements suggest that the diversity structure of rainfed maize landraces in the study area corresponds, in general terms, to that identified in larger areas of the central highlands of Puebla, where maize is grown under residual moisture.

Obtained results show that farmers in the study area have two large groups of maize materials, those with pigmented grain (Group IB) and those with white grain, with the latter subdivided into two groups, one comprised of landraces (the most numerous, Group IIB) and the other (Group IIA), which, due to its characteristics, shows affinity with improved varieties and includes a creolized material. In the case of landraces, the importance of the peasant having the two coloration groups mentioned above is that, as stated by ^{Muñoz (2003)}, ^{Castillo-Nonato (2016)}, and ^{López et al. (2020)}, it enables them to face the variability in environmental conditions, managing the different levels of earliness that are associated with grain color (white, late-cycle and pigmented, early cycle) and, on the other hand, to satisfy various cultural aspects. The presence of creolized materials reflects another strategy followed by farmers to diversify their range of production options, as it results in populations that combine desirable characteristics of improved varieties with those of landraces (^{Bellon & Risopoulos, 2001}).

The maize landraces of the study area show greater morphological similarity with the Chalqueño race.

According to ^{Sierra-Macías et al. (2016)}, the main maize races present in the Puebla Highlands (altitude higher than 2000 m) are Cónico, Elotes Cónicos, Chalqueño, Arrocillo amarillo, and Cacahuacintle, all of them belonging to the Cónico group. Based on the collection made between 1970 and 1971 in the Puebla Valley (the present study area is contained therein), ^{Cervantes & Mejía (1984)} reported Cónico and Chalqueño as the predominant races at that time. Data obtained in the present study suggest that in the explored area, the maize landraces studied have little relationship with the Conico and Elotes Conicos racial controls (which formed a morphologically independent group) and that the greatest affinity (particularly in the case of the white grain and some pigmented materials) was with the Chalqueño racial control; therefore, it is likely that a considerable fraction of the maize currently cultivated in the area corresponds to the aforementioned race. This affinity was also corroborated by comparing the expression levels of various morphological characters recorded in this work with those reported by ^{Wellhausen et al. (1951)} for the Chalqueño and Cónico races; the greatest resemblance was recorded with the former.

The greater association of the maize landraces of Puebla central highlands (particularly those with white grains) with the Chalqueño race and the lesser relationship with the Cónico race is a previously reported situation by ^{Hortelano et al. (2008), Hortelano et al. (2012)} and ^{Alvarado-Beltrán et al. (2019)}, who propose gene flow, recombination between populations, selection imposed by peasants and the greater productive capacity of the Chalqueño-type materials as the reasons to explain such association. In this work, as in those mentioned above, it was found that part of the pigmented maize did not group with the races Cónico or Elotes Cónicos, but neither was it closely related to the Chalqueño race. This may be because they are part of the existing continuum of variation in the region (Cervantes & Mejía, 1984; ^{Hortelano et al., 2008}) or because they are undergoing a process of morphological differentiation that separates them from the representative accessions of such races (^{Hortelano et al., 2012)}. Another factor could be what ^{Herrera et al. (2004)} reported in the sense that Chalqueño is a race that shows different variation degrees and which, among other groups, includes what they called the Chalqueño-Cónico group (with intermediate characteristics between both races) and the Elotes Chalqueños-Chalqueño group, integrated by three subgroups: Chalqueño Cremoso, Chalqueño Palomo and Elotes Chalqueños (with blue grain and floury endosperm). Populations that were not grouped likely belong to one of these groups. Finally, it is possible that, in the case of the creolized materials, they exhibit features that are no longer so typical of races reported as prevalent in the region.

According to ^{Guzzon et al. (2021)}, the reasons that lead to the abandonment of maize landraces (and, therefore, to the loss of diversity) are complex, and include agronomic, ecological, economic, and social factors, and within the latter, they cite increasing urbanization. This led to the assumption that, in the studied area, which is under strong urban growth pressure (^{Hernández, 2019}), the existing diversity would be minimal. The results show that this was not the case, since significant morphological variation was detected, comparable to that existing in the seventies of the last century (except for races, where no samples corresponding to the Cónico race were found). This implies that, despite the pressures they face, the remaining peasants have sought to maintain the cultivated diversity, a situation that is consistent with that reported by ^{Orozco-Ramírez & Astier (2017)} in the Patzcuaro Lake Region and by ^{Lerner & Appendini (2011)} in the Toluca-Atlacomulco Valley. Notwithstanding the above, it is advisable to assess the relevance of undertaking actions aimed at conserving the diversity that still exists.