A coffee trial was carried out in Santo Antônio do Amparo, MG, Brazil (20˚53'26"S, 44˚52'04"W and average altitude around 1100 m) along three seasons (2016/2017, 2017/2018 and 2018/2019). According to the Köppen international classification, the climate of the region is Cwa, with an average temperature of 19.6˚C and average precipitation of 1648 mm. The experiment was laid out at a Red Latosol (Oxisol) characterized for pH (H₂O), calcium (Ca²⁺), magnesium (Mg²⁺), potassium (K⁺), aluminum (Al³⁺), (H + Al), P, boron (B), iron (Fe), cooper (Cu), manganese (Mn), zinc (Zn) following the methodology described by [45] , and Remaining P (P-rem) following the methodology described by [46] . Ca²⁺, Mg²⁺, and Al³⁺ were extracted using 1.0 mol·L⁻¹ KCl, while Fe, Cu, Zn, Mn, P and K by Mehlich-1, and H + Al by 0.5 mol·L⁻¹ calcium acetate. Ca²⁺, Mg²⁺, Fe, Cu, Zn and Mn were determined by atomic absorption spectrophotometry, K⁺ by flame photometry, P by colorimetry, and Al³⁺ and (H + Al) by volumetry analysis. In the 0 - 0.20 m layer, the soil characteristics were as follows: pH (H₂O), 5.1; organic matter, 3.7 g·dm⁻³; P-rem, 5.95 mg·L⁻¹; P-Mehl, 5.0 mg dm⁻³; K, 160 mg·dm⁻³; Ca, 12 mmol_c·dm⁻³; Mg, 4.0 mmol_c·dm⁻³; Al, 2.0 mmol_c·dm⁻³; H + Al, 54.6 mmol_c·dm⁻³; CEC, 74.7 mmol_c·dm⁻³; base saturation, 26.9%; B, 0.3 mg·dm⁻³; Cu, 4.4 mg·dm⁻³; Fe, 61 mg·dm⁻³; Mn, 6 mg·dm⁻³; Zn, 2.8 mg·dm⁻³; clay, silt, and sand contents of 550, 130, and 325 g·kg⁻¹, respectively. The soil P availability used in the experiment was classified as “Very Low” for coffee orchards [47] . The treatments, applied in the planting furrow, were arranged in an incomplete factorial (2 × 4) + 1, using two P sources (MAP: 11% N, 52% P₂O₅ and Policote coated MAP: 10% N, 47% P₂O₅ and 1.9% Mg), four P rates (40, 80, 120 and 160 g P₂O₅ plant⁻¹) and control (without P application). Policote is a water-soluble polymer additive distributed by Wirstchat [48] . A randomized block design with three replications was used. Each experimental plot had three rows, with eight plants each. The central row was considered in this experiment and two guard rows were discarded. Weed, pest, and disease controls were made in all seasons.

The experiment with Arabica coffee (Coffea arabica L.), cultivar Catuaí Amarelo, was planted in November 2016, in 3.00 × 1.00 m spacing, with one plant per hole. The fertilization conducted at planting was composed of the treatments, 4.0 Mg·ha⁻¹ poultry manure and 4.0 Mg·ha⁻¹ organic compost (44% wet coffee husk, 28% dry coffee husk, 20% coal grinder, and 8% soil). Nitrogen and potassium applications (15 g N + 30 g K₂O plant⁻¹; ammonium nitrate and KCl) were parceled out in three applications (December 2016, January 2017, and February 2017). Plant height was evaluated in the six central plants of each plot in December 2016 and June 2017.

Nitrogen application (30 g N plant⁻¹; ammonium nitrate) was parceled out in three applications (November 2017, December 2017, and January 2018). Plant height and coffee yield were evaluated in the six central plants of each plot in April 2018 and June 2018 (18-month-old coffee plants), respectively. All coffee fruits were hand-harvested to estimate the percentages of immature, ripe, overripe, and dry beans. After sun drying, the fruits were weighed, and their moisture content was adjusted to 12% (w/w). Thus, the final yield was expressed in kg·ha⁻¹ of air-dried fruits.

Nitrogen and potassium applications (60 g N + 60 g K₂O plant⁻¹; ammonium nitrate and KCl) were parceled out in three applications (November 2018, January 2018, and February 2018). Plant height and coffee yield were evaluated, as described above, in November 2018 and June 2019 (30-month-old coffee plants), respectively. Agronomic P efficiency use index [(coffee yield with P—coffee yield without P)/ applied P rate] was calculated [49] with average coffee yields.

All of the statistical procedures were performed with the R Studio software. Normality and homoscedasticity of variances were tested before performing an analysis of variance. Regression analysis was used, when necessary. Qualitative treatment means were compared using the F test (p ≤ 0.05).

P fertilization only increased plant height in April 2018 (Table 1), and it was not different with or without P fertilization in all other time evaluations. There were no differences among P sources in April 2018, but increasing P rates increased plant height (p < 0.05; Figure 1).

ns—statistically insignificant. *—statistically significant (p < 0.05).

P fertilization did not affect coffee bean maturation because the percentages of immature, ripe, overripe, and dry beans were not different among treatments in the 2017/2018 (18-month-old coffee plants) and 2018/2019 (30-month-old coffee plants) seasons. Average observed percentages of immature, ripe, overripe, and dry beans in the 2017/2018 season were 14.58%, 66.21%, 12.71%, and 6.50%, respectively. In the 2018/2019 season, the percentages were 8.01%, 65.29%, 15.69%, and 11.01%, respectively.

P fertilization increased coffee yield only in the 2018/2019 season (p < 0.01; Table 2) when differences among P sources (p < 0.01) and rates (p < 0.01) were observed (Figure 2). Average coffee yield in 2017/2018 season was 156.6 kg·ha⁻¹.

Agronomic P efficiency use (APEU) was different between treatments in the 2018/2019 season. The effects of P sources and rates on the APEU are shown in Table 3.

ns—statistically insignificant. **—statistically significant (p < 0.01).

P fertilization increased plant height up to 73.5 cm, with 89.5 g P₂O₅ plant⁻¹ in April 2018 (17-month-old plants). Higher plants (105 - 109 cm in 24-month-old plants) were reported by [50] on slow-release fertilizer evaluation on the coffee cultivar Topaz. Probably the differences concerning the cultivar and age might explain this.

Plant height reduction is a typical result of P nutrition lack, but this happened only in the Apr/2018 evaluation. Other manuscripts report similar facts. Reference [51] also did not find plant height differences evaluating growth and nutritional disorders of coffee cultivated in nutrient solutions with suppressed phosphorus. Reference [9] did not find plant height differences among P fertilizers in 30-month-old coffee plants, but in 41-month-old coffee plants, P fertilization increased coffee plant height. Probably, with aging, plant height differences between P fertilizer managements could be easily detected.

The first evidence of phosphorus absence is plant growth reduction and dark green color in old leaves in the coffee crop. Neither plant height reduction (except in the April 2018 evaluation), nor dark green color in old leaves were observed without P fertilization in this experiment. Probably visual symptoms will appear with coffee orchard aging.

Estimated coffee yields (2018/2019 season) with different P sources and rates are described in Table 4. The highest coffee yield achieved with MAP application was 566.9 kg·ha⁻¹, which occurred at a rate of 123.9 g P₂O₅ plant⁻¹. The same yield was obtained with Policote-coated MAP at a rate of 30.4 g P₂O₅ plant⁻¹. The P rate used to reach maximum coffee yield with MAP (123.9 g P₂O₅ plant⁻¹) resulted in 37.8% increase in coffee yield with Policote coated MAP (566.9 × 781.4 kg·ha⁻¹). The maximum coffee yield using MAP (566.9 kg·ha⁻¹) was lower than the maximum yield using Policote coated MAP (781.4 kg·ha⁻¹), which used a lower P rate (123.9 × 100.3 g P₂O₅ plant⁻¹, respectively).

Phosphorus fertilization did not affect coffee bean ripening. No significant effects of P treatment on the biochemical composition of coffee beans were reported by [52] . Agricultural management is one of the key factors in determining coffee quality [32] . P fertilization plays an important role in agricultural management. But probably, P fertilization affects coffee quality in other criteria than coffee bean maturation.

Inconclusive responses of the adult coffee plant to P fertilization have been reported in the literature [53] [54] because the coffee plant is considered not very responsive to fertilization with P during the production period [54] . So, P fertilization studies at coffee planting are important to develop healthy and productive orchards.

P fertilization increased coffee yield. Similar results were also reported by [5] [9] [53] [55] [56] . Reference [9] applied P fertilization at coffee planting, while [53] and [55] applied it in a one-year-old coffee orchard and [5] applied it in a 5-years-old coffee orchard.

Reference [9] reported a maximum coffee yield (2243.8 kg·ha⁻¹) with the application of 620.7 g P₂O₅ m⁻¹, in a 30-month-old coffee plant. It is higher than the maximum yield observed in this experiment (809.0 kg·ha⁻¹). This divergence may be explained by cultivar (Acaiá Cerrado × Catuaí Amarelo) and plant density divergences (4081.6 plants ha⁻¹ × 3,333.3 plants ha⁻¹).

The observed yields in the 2017/2018 and 2018/2019 seasons were low because they were the 1^st and 2^nd crop yields, which will increase with age. The results obtained in studies like this highlights the importance of phosphate fertilization at the planting time of coffee orchard.

The recommended P rate for coffee orchards for the used soil in this experiment is 80 g P₂O₅ plant⁻¹ [47] . Estimated coffee yields (2018/2019 season) with 80 g P₂O₅ plant⁻¹ are described in Table 5. P fertilization with Policote-coated MAP increased coffee yield by 46.9% at the recommended P rate.

Increasing crop yields with polymer-coated fertilizers were also reported by [18] [20] [42] [43] [44] [57] [58] [59] and [60] . Higher yield with the same P rate as well same yield with a lower P rate was observed with Policote coated fertilizer. Similar results (fertilizer rate reduction) were also observed by [19] [44] [61] and [62] . This strategy can be used to reduce farm investment, increase agricultural profits, preserve phosphatic rock reserves, and avoid the overuse of phosphate fertilizer.

APEU in the range of 0.13 - 1.28 kg coffee kg P₂O₅^-1 for 30-month-old coffee plants was reported by [9] , while [5] reported APEU in the range of 1.38 - 2.58 kg coffee kg P₂O₅⁻¹ for five-year-old coffee plants.

Results showed that APEU was higher with Policote-coated MAP than with conventional MAP. Higher APEU with Policote-coated MAP explains higher yields obtained with this enhanced efficiency P fertilizer when compared to MAP. The APEU increase by applying Policote-coated MAP was also observed by [19] [39] [42] [44] and [63] , in lettuce, coffee and soybean crops, respectively. Coating phosphate fertilizers with polymer is an innovative option [18] and an emerging technology to improve phosphorus use efficiency. The observed changes in P use efficiency among P fertilizers increased our understanding of enhanced efficiency fertilizers. The obtained results demonstrated that Policote-coated MAP can be used as an enhanced efficiency fertilizer. Results show that Policote-coated fertilizer is a more efficient way to deliver required phosphorus to plants than conventional fertilizer.

Phosphorus fertilization did not affect coffee bean maturation.

Phosphorus fertilization increased coffee yield in 31-month-old plants when differences among P sources and rates were observed. Higher yield with the same P rate as well as the same yield with a lower P rate was observed with Policote coated fertilizer concerning conventional fertilizer.

Policote-coated MAP can be used as an enhanced efficiency fertilizer and is a more efficient way to deliver required phosphorus to plants than MAP.

The target for reducing farm investment, increasing agricultural profits, preserving phosphatic rock reserves, and avoiding the overuse of phosphate fertilizer could be realized through the rational use of enhanced efficiency fertilizers and fertilizer rate use reduction towards 4R’s stewardship.