The study was conducted in two experimental orchards located at Andalusian Institute of Agricultural and Fisheries Research and Training (IFAPA) Centro Las Torres, Alcalá del Río (Seville) in the Guadalquivir River Valley (SW Spain, 37º 30’ 48’’ N; 5º 57’ 46’’ W). Almond plantation was established in January 2016 and was made up of two plots of about 6000 m² each, 500 m apart from each other, managed under organic and conventional production systems respectively, containing almond trees 7 × 6 m spaced. The experimental design was in randomized blocks with four almond cultivars. ‘Guara’, ‘Marcona’, ‘Lauranne’, and ‘Marta’ grafted onto Garnem® rootstock, with eight replications, and four trees per replication. Cultivar ‘Carreró’ was used as a pollinator and was arranged among cv. ‘Marcona’ allowing to have eight trees in each plot. The soil of the experimental plot was a loam soil (43% sand, 26% silt, 31% clay) classified as Entisol group Xerofluvent subgroup Typic. The main physicochemical characteristics of both plot soils are presented in Table 1.

Both plots were drip-irrigated form March to October by using two pipelines with emitters of 2.3 L h^-1 following requirements estimated by García-Tejero et al. (2014, 2020). Meteorological data were collected from a weather station installed at IFAPA Centro Las Torres (Davis Advance Pro2, Davis Instruments, Valencia, Spain): daily mean, maximum and minimum temperature, relative humidity, and accumulated rainfall. The chilling hours model (Chandler, 1942) was used, and accumulated chilling hours throughout the dormant season were summed up (Luedeling, 2012). Fertilization in the organic plot consisted of the annual application of composed beef cattle manure (2 kg m^-2 year^-1) (Table 2) and the use of green cover crops of a mixture legume:cereal (75% Vicia sativa L: 25% Avena sativa L) in 2017, 2019 and 2021, alternate with bean cover (Vicia faba L.) in 2018 and 2020. Cover crops were sowed each year in autumn after the first rains and were mechanically cut and incorporated into the soil in the following March or April with a farm tractor. In the conventional plot, the same fertilization program was followed each year, based on the application of a N-P-K complex fertilizer (15-15-15, 150 kg ha^-1) at flowering. No cover crops were used in the conventional orchard.

Pests and diseases in the organic and conventional plots were treated according to the Regulation (EU) 2018/848 of the European Parliament on organic production and labelling of organic products (OJEU, 2018) and with the regulations for integrated production in Andalusia (BOJA, 2012). The severity of aphid attack was measured in a scale 0-5 according to Arroyo et al. (2013). Maximum severity index of aphid attack was used for comparison among cultivars in both management systems. To avoid weeds, herbicides were applied in the conventional orchard and manual tillage joined to a straw cover around the trees were used in the organic orchard. Almond pruning was similar in both orchards. Almond trees were pruned during the dormant period, following a training vessel.

The flowering period, including the beginning and end dates of flowering, was determined by analyzing two trees per cultivar, repetition, and cultivation management. For this, the BBCH scale was used: beginning of flowering (first open flowers), full flowering (> 50% open flowers), and end of flowering (all petals fallen) (Hack et al., 1992; Sakar et al., 2019). Likewise, the flower density was evaluated for each cultivar, repetition, and treatment, counting the number of flowers of three branches per tree, in the N, ESE, and WSW positions, in a section of one meter from its branching (Loren, 2013). In addition, the kinetics of defoliation was evaluated each 2 weeks from October to December each year, using a scale of 0-5, where 0 is no fallen leaves; 1, up to 20% of fallen leaves; 2, up to 40% of fallen leaves; 3, up to 60% of fallen leaves; 4, up to 80% of fallen leaves and 5, complete defoliation.

Growth and vigor of two trees of each cultivar and repetition in the organic and conventional plots was determined in November of each year, at the end of the growing season, measuring the perimeter of the trunk at 20 cm above the graft. Trunk cross section area (TCSA) was calculated as an estimate of tree growth (Arquero et al., 2005; Arroyo et al., 2013). Content of N and P in the leaves was measured in June of each analyzed year. Fully leaves from five to eight well-exposed non-fruiting spurs were taken for the analysis. Leaves were carefully detached, washed under tap water to remove any attached soil particles, rinsed twice with distilled water, and then dried at 60 °C. Afterwards, dried samples were passed through 40-mesh screens and stored in plastic containers. Leaf analysis was determined by microwave acid digestion and posterior determination by an autoanalyzer (BRAN+LUEBBE, method G-188-97). For P analysis, a sample of grinded leaf was exposed to microwave digestion by using the mixing of nitric and perchloric acid. The element concentration was determined by inductively coupled plasma (ICP). Production efficiency (PE) was estimated as almond in shell production (kg of almond in shell per tree) normalized with the size of the trees (TCSA in cm²). Yield efficiency (YE) was calculated as kernel production (kg kernel per tree) normalized with the size of the trees (TCSA in cm²). Kernel yield was the percentage of seed with respect to the weight of the entire nut fruit (Casanova-Gascón et al., 2019). PE, YE and kernel yield were determined on the total of almond nuts collected in each repetition per cultivar and crop management (4 trees per repetition, and 8 repetitions per cultivar in the two managements).

Statistical analyses were performed using Statistix software (v. 9.0, NH Analytical Software, USA). Data corresponding to flower density, TCSA, N and P leaf content, and production were subjected to analysis of variance (ANOVA) and multiple comparisons of means were analyzed by LSD (least significant difference) at p<0.05. Defoliation in each date analyzed, and maximum severity index of aphid attack were analyzed with the Kruskal-Wallis non-parametric test at p<0.05 level of significance. To quantify the relationship between temperature and defoliation degree, the non-parametric Spearman’s correlation coefficient was used.

Table 3 shows the accumulation of chill units during the cold season and the average temperatures, relative humidities and rainfall during the flowering period registered in the almond orchards. Analyzing the average mean temperature in the period from January 15 to March 15 in the location, an increase of 2.98 ºC was observed from 2018 to 2021 (Fig. 1). The relative humidity did not suffer variations, and the rainfall was very variable throughout the four seasons studied. The chilling hours progressively decreased from 2018 to 2021.

Figure 2 shows the average flowering period of all cultivars studied from the beginning to the end of flowering, in the organic and conventional plots from 2018 to 2021. Results reveal a general early bloomer, as well as a lengthier flowering interval in conventionally managed trees compared to trees under organic management: in 2018 and 2019, the onset of flowering was 2 to 4 days earlier in the conventional plot. However, it was very similar in both plots in 2020 and 2021. The data also show a gradual advance of the onset of flowering from 2018 to 2021, of 12 days in the conventional plot and 13 days in the organic orchard. Regarding flowering interval, trees under conventional management had an average of 19 days of flowering compared to an average of 15 days displayed by trees organically managed. The analysis of flowering by cultivars indicated that ‘Carreró’ and ‘Marcona’ were the earliest in flowering in both, the organic and the conventional plots, although no flowering was produced in cv. ‘Carreró’ in 2018 (Fig. 2). Flowering in ‘Carreró’ started earlier than in ‘Marcona’, fulfilling its function as a pollinator of the latter variety, which is considered not self-fertile and needs cross pollination. In general, flowering came earlier, a few days for each cultivar, in the conventional plot with respect to the organic one, this pattern was repeated in 2018, 2019 and 2021. In 2020, the trend was reversed, and flowering began earlier in the organic plot for all cultivars, except for ‘Carreró’. ‘Lauranne’ was the latest cultivar to start flowering in all the years studied, and in both production systems (Fig. 2). Likewise, the flowering interval increased for all cultivars from 2018 to 2021, probably due to the growth and vigor increase of the trees in this period. There were no significant differences in the flowering interval between both production systems for cvs. ‘Marta’ and ‘Lauranne’ in 2020, and for ‘Lauranne’ in 2021.

The flower density of the different almond cultivars under organic and conventional management from 2018 to 2021 is shown in Table 4. ‘Marcona’ was the cultivar that produced the highest number of flowers in all the years studied, under organic and conventional management. All cultivars increased its flower production in both plots over the four years, significantly differing between conventional and organically managed trees, except for ‘Lauranne’ which showed a similar flower density in both plots in 2021. Cultivars ‘Marta’ and ‘Carreró’ showed the lowest flower production in 2018, 2019 and 2020 in both plots, although ‘Carreró’ had a large flower production in the conventionally managed plot in 2020 and 2021. In general, the mean number of flowers was significantly higher in the conventional plot than in the organic one in all years studied (Table 4).

Seasonal leaf fall was evaluated from October to December for all cultivars. Considering the whole orchards, defoliation began earlier, up to two weeks in some cultivars, and was more intense in the organic orchard than in the conventionally managed plot throughout all the seasons analyzed (Fig. 3). All almond trees from the organically managed orchard showed a significant higher leaf drop and were completely defoliated before those in the conventional orchard. ‘Guara’ and ‘Marcona’ were the cultivars that presented significantly the most premature and intense defoliation, especially in the organic orchard. On the contrary, ‘Marta’ was the most backward cultivar and the one in which an overlap of old non-fallen leaves with new leaves was observed under both managements. ‘Guara’ and ‘Lauranne’ showed the lowest defoliation differences between organic and conventional management. Significant differences in the degree of defoliation between the two managements were only observed in the intermediate evaluation dates (from October 11 to November 23) for all varieties (Fig. 3). The defoliation dynamics of almond trees fluctuated for all years studied. The daily mean temperature in the period from 1st October to 18 December is showed in Fig. S1 [suppl]).

A high negative Spearman correlation was observed between the mean defoliation degree of almond cultivars in each plot and the average temperature of the analyzed period for all varieties in all years studied (Table 5). This correlation was specially observed in warmer periods that occurred in 2019 (October 29 to November 5) and 2020 (November 12 to 19) that corresponded to a slow down of the defoliation in these periods.

The severity of aphid attack was more intense in the organic orchard for all cultivars in all years analyzed. ‘Marcona’ was significantly the most susceptible cultivar to aphid attack in both management systems, not significantly differing from ‘Carreró’. ‘Lauranne’ and ‘Guara’ were the most tolerant, showing scarce incidence in the organic and conventional orchards in all years analyzed (Fig. 4).

Table 6 shows the tree growth and vigor, estimated as TCSA in cm², for each cultivar in the two production systems from 2017 to 2020, as well as the TCSA increase in that period. Data reveal that all cultivars showed significantly higher values of TCSA in conventionally grown almond trees than in organically managed ones in all years. The growth increase in the analyzed period was also higher in the trees grown in the conventional plot for all cultivars. In 2017, there were no significant differences in vigor among cultivars within the organic production system. As the trees developed, differences in growth among cultivars were appreciated: ‘Marta’ was the cultivar that showed the highest growth dynamics, displaying the highest TCSA values and the highest growth increase in both production systems. ‘Lauranne’ was the cultivar that showed the least difference between the two production systems. Likewise, ‘Marcona’ and ‘Carreró’ were the cultivars that presented the lowest growth increase under organic and conventional management, respectively, in all years studied (Table 6).

N and P leaf contents for each cultivar in the two production systems from 2018 to 2021 are shown in Table 7. N concentration ranged from 2.01 to 2.97% and from 2.25 to 3.02% in organically and conventionally managed tress, respectively, not showing significant differences among management systems except for specific cultivars in different years. ‘Lauranne’ was, in general, the cultivar with the highest N leaf content in both management systems. P concentration ranged from 0.08 to 0.17% and from 0.09 to 0.19% for the organic and conventional management, respectively. Significant differences in P content between trees under conventional and organic management were only observed in 2018, with higher values displayed by organically managed almond trees. In the three following years, no significant differences in leaf P concentration were observed between both management systems except for ‘Lauranne’, which displayed a higher leaf P content under conventional management in 2021.

Almond in shell and kernel productions (measured as kg per tree) were significantly higher in the conventional orchard than in the organically managed one, for all cultivars and all years studied, except for ‘Lauranne’ cultivar in 2021 (Table S1 [suppl]). On average, the production of almond in shell from organic plot was 26% of that of conventional plot in 2020 and increased to 57% in 2021. A similar trend showed the kernel production whose value in the organic plot was 28% of that of the conventional plot in 2020, and increased to 55% in 2021.

The difference in size of the trees between both plots have been considered for efficiency calculations. So, Table 8 shows the almond in shell and kernel productions normalized with the size of the trees (TCSA in cm²) to obtain PE and YE, respectively. PE of the organic plot was 41% of that of the conventional one in 2020 and increased to 85% in 2021. In the same way, YE of the organic plot was 32% and reached 83% of that of the conventional plot in 2020 and 2021, respectively. Regarding PE per almond cultivar, differences between both managements were observed for all cultivars in 2020, with significantly higher values for the trees grown under conventional system. In 2021, significant higher PE values were only observed for ‘Marta’ and ‘Marcona’ under conventional management. ‘Lauranne’ showed the best performance in terms of PE under organic and conventional systems in the two seasons analyzed. ‘Guara’ also exhibited high PE, not significantly differing from that of ‘Lauranne’ in 2021. Regarding YE, no significant differences were observed between both managements for cv. ‘Lauranne’ in the two years studied, and significant higher YE values were observed for conventionally managed ‘Guara’, ‘Marcona’ and ‘Marta’ in 2020 but only for ‘Marta’ in 2021.

Kernel yield of organic plot, calculated as the percentage of seed with respect to the weight of the fruit with shell, was 91 and 97% of that of the conventional plot in 2020 and 2021, respectively (Table 8). Regarding kernel yield per almond cultivar, ‘Guara’ was the cultivar that showed the higher kernel yield under organic and conventional management in the two years studied. Differences between both managements were only observed for this cultivar in 2020 which rendered 32% and 40% of kernel yield under organic and conventional systems, respectively. The rest of cultivars showed similar kernel yields under both managements in the two seasons. Marcona was the cultivar that showed the lowest kernel yield under both managements.