Study site

The LVRN is located at 40° south latitude and 63° west longitude, on the southern bank of the Negro river in Argentina. The climate in the region is semi-arid mesothermal, with an average precipitation of 394 mm and a mean annual temperature of 14.2 °C. It is a fertile area, with fine to medium textured soils, and has a network of canals for irrigation water supply.

This study was carried out during the 2019 growing season in a plot with gravity irrigation located at the Chacra Experimental de la Unidad Integrada para la Innovación del Sistema Agroalimentario de la Patagonia Norte (UIISA), in the LVRN, Argentina, where onion was grown the previous year. The soil of the experimental site is clay loam, with a pH of 7.9, electrical conductivity (EC) of 1.2 dS·m^-1, 4.75 % organic matter (OM) and 0.9 % total nitrogen (N).

Experimental design

In a 5 000 m² plot, three blocks of 1 200 m² (12 x 100 m) were set, where three plots of 12 x 25 m were established, one for each fertilization treatment: 1) organic (Co), onion waste and cattle manure compost (8 kg·m^-2) at the time of sowing, 2) mineral (Mi), monoammonium phosphate at the time of sowing (100 kg·ha^-1) and urea when the crop showed the third leaf (50 kg·ha^-1) and 3) mixed (Co+Mi), compost + monoammonium phosphate at the time of sowing (4 kg·m^-2 + 50 kg·ha^-1, respectively) and urea when the crop showed the third leaf (25 kg·ha^-1). Doses were determined according to crop requirements and soil contributions (^{Pellejero et al., 2021}).

On March 19, 2019, Morado INTA garlic was sown, one clove of garlic every 10 cm in double rows on the ridges. Irrigation was carried out at the time of sowing, and from August onwards it was irrigated every 15 days, according to crop demand by the furrow between ridges. Harvest was carried out on December 18, 2019.

Characteristics of compost

Onion processing residues from the Quequen packing company, located in the LVRN, were used. The management of harvested onions involved the removal of roots, upper leaves and outer cataphylls, which was used for composting its high C/N ratio and as a high nitrogen content structuring agent of cattle manure. Mature compost had a pH of 7.8, EC of 2.8 dS·m^-1, OM of 21.6 %, N of 0.9 % extractable phosphorus (P) of 0.23 % (^{Pellejero et al., 2017}).

The variables evaluated in each treatment are described below:

Crop yield. At the time of harvest, developed garlic was collected from one linear meter of ridge. Fresh weight and equatorial diameter were determined for each bulb to classify them by caliber.

Soil samples. 27, 85, 148, 237 and 274 days after sowing (das), three soil samples were taken from each treatment. Each sample consisted of 15 subsamples of the first 5 cm of soil on the crop bed.

Biological indicators: Soil microbial respiration, dehydrogenase activity and esterase activity were determined on the wet soil samples. For the former, 100 g of wet soil, sieved with 2 mm mesh, were incubated with 25 mL of NaOH (0.25 M) for 7 days in the dark at room temperature. Subsequently, 10 mL of hydroxide was titrated with HCl (0.10 M) using phenolphthalein as an indicator. The carbon dioxide produced was quantified by comparison with a soilless blank (^{Alef, 1995}).

Dehydrogenase activity was determined by incubating 3 g of wet soil with 0.5 mL of triphenyltetrazolium (TTC) at 37 °C for 24 h. The hydrolysis product was extracted with three washes with alcohol and brought to 50 mL. The concentration of triphenylformazan (TPF) was determined by absorbance at 480 nm (^{Casida et al., 1964}). Esterase activity was obtained by incubating 1 g of wet soil in 7.5 mL of phosphate buffer with pH of 7.2 and 0.2 mL of fluorescein diacetate (1 µg·mL^-1 de FDA) for 30 min at 30 °C. The reaction was stopped with 7.5 mL of acetone, and the released fluorescein was quantified by absorbance at 490 nm (^{Alef, 1995}).

Chemical indicators: The aerated soils passed through a 2 mm sieve and pH in 1:2.5 solution, EC in saturation paste, sodium adsorption ratio (SAR), N by Kjeldahl, OM by oxidation with potassium dichromate (OM), P by Olsen's method and available potassium (K) extracted with ammonium acetate by flame spectrometry were determined.

Statistical analysis

The data results were submitted to an ANOVA and Fisher’s comparison of means (LSD, P ≤ 0.05). Spearman's correlation coefficient was used to explore correlations between microbiological and physicochemical parameters. All tests were carried out using the statistical program InfoStat 2015 (^{Di Rienzo et al., 2015}).

Garlic crop yield

The total and commercial yield (4 to 6 caliber healthy bulbs) of the three fertilization treatments were similar (Figure 1). The commercial yield (10.99, 12.29 and 10.33 t·ha^-1 for Co, Mi and Co+Mi, respectively) was similar to that observed for the same garlic crop in Mendoza province, Argentina, in the 2017 and 2018 growing seasons (11.36 and 12.38 t·ha^-1, respectively) (^{Fernández, 2019}).

Figure 1 Yield of garlic produced with different fertilizer sources. Co = compost of onion residues and cattle manure; Mi = monoammonium phosphate and urea; Co+Mi = compost + monoammonium phosphate and urea. Each bar corresponds to the average of three samples and lines indicate the standard error. 

Garlic yields using goat or mineral fertilizer were found to be similar in a study carried out in eastern Ethiopia by ^{Tadila and Nigusie (2018)}, although the highest yields were achieved with the combination of both fertilizers at the same doses. ^{Ruiz et al. (2007)} found no differences in onion crop fertilized with different organic sources, such as cattle manure, goat manure or chicken manure, although they did not compare yields with chemical fertilization, but with an unfertilized control that yielded less than the composted treatments. ^{Paterlini et al. (2019)} also observed no differences in lettuce crop yield when testing different doses of chicken litter compost. ^{Carelli and Defendente (2018)} had similar yields when applying organic and mineral fertilization on arugula crop. In this study, obtaining similar yields to chemical fertilization, in the first year of using onion and cattle manure compost in the LVRN, is promising in this first phase of agroecological transition.

Eighty-six percent of the harvested bulbs weight corresponded to medium sizes (4 and 5). Treatment Mi had a higher proportion of large bulbs (caliber 6). In all treatments, small bulbs (caliber lower than 4) represented less than 3% of the total weight of the harvest (Figure 2). In addition to yields, size homogeneity is important for the market, a characteristic that was observed in the compost treatments.

Figure 2 Weight ratio per caliber of garlic bulbs produced with different fertilizer sources. Co = compost of onion residues and cattle manure; Mi = monoammonium phosphate and urea; Co+Mi = compost + monoammonium phosphate and urea. 

A similar distribution to that of the present study was observed in the 2018 season in the entire province of Mendoza, Argentina, with predominance of garlic size 5 (51 %), followed by size 6 (38 %) and size 4 (6 %). Medium sizes accounted for 89 % of morado garlic produced in that province of Argentina (^{Fernández, 2019}).

Soil quality indicators

Soil salinity indicators in Mi treatments showed no changes during the trial. However, in the treatments where compost was applied (Co and Co+Mi), pH was similar to that of Mi at the beginning and increased 9 % during the growing season. Conversely, the EC was higher than that of the soil at the beginning of the cycle and decreased by 53 % during the growing season. SAR values were low in all treatments, with a high dispersion to identify variations during the growing season (Figure 3).

Figure 3 Behavior of soil salinity indicators in garlic crop with different fertilization sources. a) pH in ratio 1:2.5, b) electrical conductivity (EC) in saturation paste and c) sodium absorption ratio (SAR). Co = compost of onion residues and cattle manure; Mi = monoammonium phosphate and urea; Co+Mi = compost + monoammonium phosphate and urea. 

^{Benedicto-Valdés et al. (2019)} noted an 8 % increase in pH when applying cattle manure, starting from a pH of soil and manure similar to each other. On the contrary, when using goat manure or chicken manure, they observe no change in pH. Similarly, but with chicken litter compost, ^{Paterlini et al. (2019)} observed pH increase in one unit for the transplant lettuce crop, which coincides in date with the garlic crop, but has a shorter cycle (55 days). The increase in soil pH could be due to the contribution of cattle manure from the compost used in this study, which would condition its use in areas with alkaline soils such as those of the LVRN. However, this variable had no effect on the quality or yield of the crop.

The EC of soil with cattle manure was higher than that of soil without compost in the trial of ^{Benedicto-Valdés et al. (2019)}, although it is not indicated how long the manure was in the field and whether irrigation was carried out. In the present study, we started from mature compost, so EC was relatively low (2.8 dS·m^-1) and seemed to be associated with the contribution of sodium ions (SAR), which were washed out with successive irrigations. Thus, in the first months after cultivation, EC was the same in all treatments.

At harvest (late spring), the pH of soils with organic fertilizer (Co and Co+Mi) was higher than that of soils with Mi. On the other hand, EC and SAR were similar in the three treatments (Table 1); this was because the multiple gravity irrigations carried out (12 irrigations of 30 mm), the quality of the water from the Negro river used in the LVRN (EC 0.3 mS·cm^-1) and the operation of the drainage system that allows the leaching of soluble salts together with the excess water. Such results are contrary to that observed by ^{Cebadero-Cayetano et al. (2020)}, who obtained a decrease in pH with increasing organic carbon content.

In the first sampling (27 das), the four fertility indicators evaluated (N, OM, P and K) resulted higher in the composted treatments (Figure 4). Total N in soil decreased during the trial, probably due to ammonium washout from cattle manure, which would have caused pH elevation (Figure 4a). OM also decreased during the trial, with a slight increase in the last sampling probably due to the contribution of leaf residues from the garlic crop itself and from weeds (Figure 4b). Soil P content was significantly higher (P ≤ 0.05) in Co treatments, due to the lower mobility of this nutrient in the soil, especially at the alkaline pH of the region (Figure 4c).

Figure 4 Changes in soil fertility indicators in garlic crop with different fertilization sources: a) nitrogen, b) organic matter, c) phosphorus and d) potassium. Co = compost of onion residues and cattle manure; Mi = monoammonium phosphate and urea; Co+Mi = compost + monoammonium phosphate and urea. 

The different chemical forms of N in soil could justify the higher calibers observed in the Mi treatment, although the total yields with the three treatments resulted similar. ^{Rodrigues et al. (2019)} also observed no differences in total N content in pot soils with olive plants after 18 months when applying eight fertilization treatments, which included mineral fertilization, biochar and different organic residues.

The Co treatment significantly (P ≤ 0.05) modified soil fertility indicators when the crop was in the field. In that treatment, higher values of MO, P and K (1.2, 3.5 and 1.5 times, respectively) were observed than in soils with Mi or Co+Mi (Table 2). This condition will favor the successor crop.

^{Rodrigues et al. (2019)} found no differences in P content in soils with different sources of organic residues, but they started from a soil with higher P content and acid pH; in addition, they observed no improvement in OM content with the treatments tested. P availability in soil is associated with neutral pH and high OM contents (^{Andrades & Martínez, 2014}); although ^{Paladino et al. (2019)} had higher organic carbon content and lower EC in urban soils with three years of agroecological production.

The behavior of the biological activities was similar in the three treatments during the crop cycle. In general, the three biological parameters were higher with the Co treatment during the whole cycle; this was due to the contribution of microorganisms present in the onion and cattle manure compost used (Figure 5). The increase in respiration of soil microorganisms in the last sampling could be due to the interaction of this activity with an irrigation event and the mechanical work carried out for weed removal prior to harvest. This effect was not seen in esterase and dehydrogenase activities because, being extracellular enzymes, they could have percolated with irrigation.

Figure 5 Behavior of biological activities of soil garlic crop exposed to different fertilization sources: a) respiration (mg_CO2·kg_ss ^-1 for 7 days), b) esterase activity (µg_FDA·g^-1) and c) dehydrogenase activity (mg_TFF·g^-1). Co = compost of onion residues and cattle manure; Mi = monoammonium phosphate and urea; Co+Mi = compost + monoammonium phosphate and urea. 

Biological parameters at the end of the crop cycle were significantly higher (P ≤ 0.05) with Co, being 37, 26 and 54 % higher than Mi treatment for respiration, esterase activity and dehydrogenase activity, respectively (Table 3). Based on the above, the contribution of organic fertilizer for soil nutrient cycling is promising. ^{Di Ciocco et al. (2014)}, with continuous farming, reported decreases in soil respiration and dehydrogenase activity, justifying the need to contribute microorganisms to the soil.

The contribution of organic matter favors microbial activity, which is why ^{Cebadero-Cayetano et al. (2020)} noted an increase in dehydrogenase activity in plots with a constant supply of litterfall, such as an esparto grass or oak. Similarly, in the present study, the increase in organic matter stimulated the biological activities present in soil nutrient cycling.

Spearman's correlation coefficient shows that soil respiration is significantly (P ≤ 0.05) and positively correlated with EC, SAR and OM; esterase activity, with N and P, and dehydrogenase activity, with pH and SAR. The latter was also negatively correlated with EC (Table 4). ^{Di Ciocco et al. (2014)} had correlation between soil respiration and EC, OM and sodium, in addition to correlating positively with other chemical variables such as P, N and pH, which was not observed in the present study, since the soil showed higher pH and lower content of N and P. These authors analyzed nitrogenase activity, and it correlated positively with other physicochemical variables to those observed with respiration.