Article Energy Efficiency of Variable Rate Fertilizer Application in Coffee Production in Brazil 1 2 3 2 Graciele Angnes , Maurício Martello , Gustavo Di Chiacchio Faulin , José Paulo Molin 1, and Thiago Libório Romanelli * Laboratory of Systemic Management and Sustainability, Department of Biosystems Engineering, “Luiz de Queiroz” College of Agriculture, University of Sao Paulo, Piracicaba 13418-900, SP, Brazil; firstname.lastname@example.org Laboratory of Precision Agriculture, Biosystems Engineering Department, Luiz de Queiroz College of Agriculture, University of São Paulo (ESALQ/USP), Piracicaba 13418-900, SP, Brazil; email@example.com (M.M.); firstname.lastname@example.org (J.P.M.) Fatec Shunji Nishumura, Pompeia 17580-000, SP, Brazil; email@example.com * Correspondence: firstname.lastname@example.org; Tel.: +55-19-3447-8523 Abstract: Coffee is a crop of great relevance in socioeconomic terms for Brazilian agribusiness, which is the world's largest producer in cultivated areas. The implementation of precision agricul- ture in the coffee culture has provided countless benefits to its development, which over the years has been cultivated in the same area. However, there is a lack of studies that address the impact of the application of variable-rates inputs in soil on the energy efficiency and sustainability of these systems. This study aimed to analyze how variable-rate fertilization influences energy efficiency in coffee growing. A production area subjected to variable and fixed rates of fertilizer in alternating rows was evaluated. Geo-referenced yield data was collected to assess yield response for fixed and variable rate applications. The energy assessment was combined with the Geographic Information Citation: Angnes, G.; Martello, M.; System (GIS) to determine site-specific energy indicators. To determine the energy flow, only NPK Faulin, G.D.C.; Molin, J.P.; fertilizer applications were considered as inputs and the yield as output. The results obtained indi- Romanelli, T.L. Energy Efficiency of cated that the variable rate fertilizer application has a small difference, indicating greater energy Variable Rate Fertilizer Application efficiency concerning the applied fertilizer and coffee production per crop season. It was observed in Coffee Production in Brazil. −1 −1 in the 06/07 crop, the incorporated energy was 10.7 MJ kg for VR and 10.2 MJ kg for UR and for AgriEngineering 2021, 3, 815–826. −1 −1 the 07/08 crop it was 30.7 MJ kg for VR and 34.9 MJ kg for UR. The energy balance was more https://doi.org/10.3390/ agriengineering3040051 efficient at variable rates, as it provided fertilizer savings without compromising yield. However, the difference between the embodied energy per mass of coffee produced was very small compared Academic Editor: Anna Vatsanidou to the fixed rate. Received: 23 August 2021 Keywords: Coffea Arabica L.; precision agriculture; perennial crop; energy balance Accepted: 17 October 2021 Published: 19 October 2021 Publisher’s Note: MDPI stays neu- 1. Introduction tral with regard to jurisdictional Brazil is the world's largest coffee producer. Coffea arabica L. production occupies a claims in published maps and institu- tional affiliations. total area of 1.5 million hectares, with an estimated production of 47.37 million bags of processed coffee (1 bag = 60 kg), and its largest producer is Minas Gerais state, with 33,46 million bags of processed coffee . Copyright: © 2021 by the authors. Considering the potential yield and profit of coffee, a broad knowledge of high qual- Submitted for possible open access ity coffee production techniques is indispensable for modern coffee growing and new publication under the terms and con- management tools such as precision agriculture should be used . According to Mulla ditions of the Creative Commons At- et al. (2010)  variability of soil attributes will influence the crop management efficiency tribution (CC BY) license (http://crea- and development. In this context, Precision Agriculture (PA) techniques offer manage- tivecommons.org/licenses/by/4.0/). ment systems in which inputs are applied according to the spatial variability of the field AgriEngineering 2021, 3, 815–826. https://doi.org/10.3390/agriengineering3040051 www.mdpi.com/journal/agriengineering AgriEngineering 2021, 3 816 using variable rate technology. PA's main objective is to provide economic and environ- mental benefits . For some crops, such as grains, the study of techniques for monitoring the spatial variability of productivity is already consolidated, mainly due to harvesters having easier access to technological packages, unlike perennial crops such as orange, eucalyptus, and coffee. [5,6]. Some authors have reported the effects of PA on input consumption and productivity indicating positive gains [1,7–10]. On the other hand, other assessments [11,12] also have suggested that different site-specific nutrient strategies were often not economically ben- eficial. Furthermore, little is known about the actual environmental impact of this technol- ogy. Studies that perform economic and environmental analyses to test PA technologies [6,13–15] are rarer than those restricted to harvest evaluation. The determination of energy efficiency is an important tool when one wants to assess the sustainability of production systems . Embodied energy indicates how demanding a production system is in terms of energy. Nutrient balances assess the balance (or not) between the input and output of nutrients into the system and, similarly, inform about the risk of nutrient loss and envi- ronmental contamination due to over-application [17,18]. In energy terms, an agricultural production system can be interpreted as a converter of solar energy into food energy, with the intervention of water, carbon dioxide and semi- manufactured products, such as fuels, fertilizers, pesticides, seeds, among others . Input and output energy flows based on the material flows and production, respec- tively, can be used to assess the environmental sustainability of different management strategies [16,20,21] and nutrients [17,22]. Through PA adoption, data of yield and applied inputs are geographically available, allowing for site-specific economic and environmen- tal indicators to be related to the field variability. For example, studies approaching en- ergy analyses in citrus [6,23] and wheat crops  with site-specific variable fertilizer ap- plication provide maps of energy balance, embodied energy, and energy embodiment. The approach allowed for environmental management and sustainability analysis at the site-specific scale. In coffee production systems, few studies have applied energy assessment, as high- lighted by Lacerda et al. (2014) , Muner et al. (2015)  and Turco et al. (2018)  who applied the methodology to evaluate different coffee production systems. However, none of them evaluated variable rate fertilizer application. This study aimed to analyze how variable rate fertilization influences energy effi- ciency in the coffee growing at one coffee plantation . 2. Materials and Methods 2.1. Coffee Areas and Treatments Relevant data on the experimental design used for the present study are provided in this section and were obtained from a coffee farm located in Patrocínio municipality, Mi- nas Gerais state, Brazil (18°41’31.14” S, 46°49’18.17” W), as it is referred to in the article of Molin et al. 2009 . The studied plot had 4.67 ha, presenting 4-m distance between rows, cultivated with the species Coffea arabica L., Catuaí variety, conducted during two harvest seasons. The soil type was Yellow Oxisol . According to Köppen classification, the climate is classified as Cwa. A field-scale experimental design was used in which variable rate (VR) and uniform rate (UR) treatments were implemented in interspersed bands of two lines across the en- tire area (Figure 1). The design allowed the data to be interpolated separately for each treatment, allowing for comparisons between treatments. The inputs evaluated were N, P, K, whose sources were ammonium nitrate, triple superphosphate, and potassium chlo- ride. AgriEngineering 2021, 3 817 Figure 1. Example of the arrangement of treatments in each area. For the VR, recommendations were made from soil sampling in the 0–0.2 m layer, the samplings were carried out at georeferenced points, and arranged in a grid with equi- distant longitudinal and transverse spacing. For the soil samples in the UR treatment, we tried to respect the methodology used on the farms, where sub-samples are collected in the center of the lines along the area, composing a single sample composed by area . In the VR treatment, each sample consisted of nine sub-samples, spaced parallel to the row of plants by 0.6 m and grouped at an interval of approximately 1.0 m (Figure 1). For leaf sampling, collections were carried out only in the 2007/2008 season on branches located in the middle part of the plant height, the third and fourth pair of leaves, counted from the first pair of leaves with a length greater than 25.0 mm. Fifty-two leaves per sample were removed from thirteen plants close to the sampling point for each treat- ment using the average content in UR treatment. 2006/2007 season a uniform rate of 340.0 −1 kg ha of urea was established based on the expected yield without sampling. The sam- ples were placed in properly identified paper bags and stored in a Styrofoam container with ice to be taken to the laboratory. The samples were sent to certified analysis laboratories to determine exchangeable −3 −3 contents of Phosphorus (P resin, mg dm ) and Potassium (K+, mmolc dm ) in the soil, −1 and for the mineral nitrogen leaf content (N, g kg ), (Table 1). Table 1. Statistical parameters of nitrogen, phosphorus and potassium analysis. Samples (2) (3) (4) Treatment Season Mean Min. Max. SD CV (%) W (1) −1 (leaf N, g.kg ) UR 1 27.4 - - - - - 07/08 VR 8 26.9 26.3 27.6 0.47 1.74 0.942 * −3 (soil P-resin, mg dm ) UR 1 18.0 - - - - - 06/07 VR 20 17.9 12.0 35.0 7.1 39.3 0.791 AgriEngineering 2021, 3 818 UR 1 6.9 - - - - - 07/08 VR 14 6.0 5.0 7.0 0.8 13.1 0.912 * −3 (soil K, mmolc dm ) UR 1 1.50 - - - - - 06/07 VR 20 1.49 1.1 3.5 0.53 35.7 0.642 UR 1 4.50 - - - - - 07/08 VR 19 4.47 3.7 5.0 0.43 9.6 0.778 (1) (2) (3) (4) Number of samples considered after exploratory analysis; Standard Deviation; Coefficient of variation Shapiro- Wilk Statistic; * Data came from a normal distribution at the 5% significance level. For the maps, semivariograms of soil, leaves and yield data were modeled and ad- justed (Table 2) and interpolated using the kriging method. R Studio software was used, with the stats , MASS  and geoR  packages. Table 2. Geostatistical parameters of nitrogen, phosphorus, potassium, and yield in the VR treat- ment. (1) (2) (3) (4) Season Model C0 C1 a (m) SC% Nitrogen (leaf sample) 07/08 Spherical 0.05 0.21 140.0 19 *** Phosphorus (soil sample) 06/07 Spherical 0.53 0.91 109.1 37 ** 07/08 Spherical 0.003 0.001 105.2 75 ** Potassium (soil sample) 06/07 Spherical 0.002 0.01 124.7 17 *** 07/08 Spherical 2680.92 20,777.16 105.2 11 *** Yield 06/07 Exponential 7.9 × 10³ 4.7 × 10³ 69.1 62.4 ** -1 -1 07/08 Exponential 2.1 × 10 1.8 × 10 57.3 53.9 ** (1) (2) (3) (4) nugget effect; structural variance; range in meters; Spatial Dependency (% Nugget). Note: For : ** moderate spatial dependence; *** strong spatial dependence. The results obtained were used to determine the fertilizer application map by nutri- ent house regressions that met the recommended ranges and the expected yield. Fertilizer applications were conducted three separate times for nitrogen, twice for potassium, and with a single application for phosphorus. To establish the fertilizer doses that were ap- −1 plied, the expected yield (kg ha ) and the total length of plant rows (m) per hectare were considered. Furthermore, according to the results of the soil analysis, some adjustments were made to the result of phosphorus and potassium doses that were applied . For fertilizer application, a drag fertilizer with a system by conveyors and gate (vol- umetric) and pneumatic directed distribution to the plant canopy projection, equipped with implementing unit for VR, GNSS receiver (C/A L1 carrier) without correction (Jacto S.A., model not yet commercially available) and pulled by a 31.7 kW tractor. Due to its construction, with two metering mechanisms and independent lateral distribution, it was possible to configure the left side of the machine to apply at a uniform rate and the right side to apply at a variable rate, following the recommendation by the application map previously defined and loaded into the computer of the equipment. The application map provides the equipment with information on the doses to be applied and their geographic locations. For VR recommendations, fertilizer rates were calculated for each 10 × 10 m map pixel based on information from a soil fertility (or leaf nutrition) map and a yield map. The data comprised a period of two agricultural harvest seasons. AgriEngineering 2021, 3 819 The doses for UR to be applied were determined considering expected yield ranges and levels found in the soil analysis report . Linear regressions were established for each nutrient that met the recommended ranges and the expected yield, as presented by Molin et al. (2009) . Recommendations were divided into three times for Nitrogen, twice for Potassium, and a single application for Phosphorus. The fertilizations used Urea (N source), Simple Superphosphate (P2O5) and Potassium Chloride (K2O). 2.2. Material Flow and Energy Analysis The embodied energy was determined based on Romanelli et al. (2010) , where −1 the total input energy for fertilization was divided by yield (MJ kg ). This indicator rep- resents the energy demand by fertilization to produce 1 kg of coffee beans. Input fertiliza- tion energy was determined by multiplying the application rates of nutrients (N, P2O5 and K2O) by the energy indices. These indices represent the amount of energy needed to pro- duce one unit of each input. In this study, only the fertilizers that were used in each treatment were evaluated. The energy index used for each fertilizer was extracted from Pellizzi (1992)  for nitro- −1 −1 −1 gen (N) 74.0 MJ kg phosphorus (P2O5) 12.6 MJ kg and potassium (K2O) 6.7 MJ kg . Output data were obtained by a self-propelled harvester Jacto, model K3 (Máquinas Agrícolas Jacto S.A.), equipped with a volumetric yield monitor, which measures the productivity in liters per hectare of the harvested coffee fruit . To determine the yield during the experiment, samples of the coffee fruits were collected at the discharge outlet of the harvester. These samples were randomly obtained in the area, after they were sun dried for further processing and weighing, to generate a conversion factor, which was −1 applied to the productivity data (green coffee) (L ha ) and made it possible to determine −1 the productivity (kg ha ) of the processed coffee (commodity). For energy output deter- −1 mination, the energy index for coffee was adopted from  who determined 9.72 MJ kg . As noted by Colaço et al. (2020) , the energy incorporated in the mechanical oper- ation was relatively less relevant when compared to the energy incorporated in the ferti- lizer itself. Thus, results depended mainly on the amount of fertilizer applied and the re- sulting yield for each year and treatment. Energy balance (EB, Equation (1)), energy return on investment (EROI, Equation (2)) and energy intensity (EI, Equation (3)) were the indicators of energy efficiency. These in- dicators represent the amount of energy that was made available for each kg of coffee produced. The incorporation of energy (Equations (1) and (2)) refers to the energy re- quired to produce each unit of mass of processed coffee. EB = OE-IE (1) EROI_ = IE/OE (2) EI = IE/Productivity (3) −1 −1 where: EB—energy balance (MJ ha ); IE—input energy (MJ ha ) OE—output energy (MJ −1 ha ); EROI is energy return on investment (non-dimensional), EI is–energy intensity (MJ −1 −1 ha ), Productivity (kg ha ). 2.3. Spatial Variability Analysis Yield maps for the UR and VR treatments and fertilizer application maps (for the VR treatment) were generated for the entire area by kriging spatial interpolation, using in each case only the data from the corresponding ranges. Productivity and application maps were combined in layers and energy flows were determined (10 × 10-m map pixel). Maps were produced using QGIS software (version 2.10; QGIS Development Team, 2019) . AgriEngineering 2021, 3 820 3. Results 3.1. Material Flow By strictly implementing the official fertilizer recommendation and using more de- tailed input information, VR reduced fertilizer application in both seasons mainly for po- tassium and phosphorus, yield varied between seasons, as coffee presents biennially. It is highlighted that the first crop had high productivity and the second one, low. Regarding material flows, variable rate provided a reduction of input consumption of the phosphate and potassium when compared with the conventional fertilization. The VR fertilization resulted in savings of 24.0% and 17.4% on P2O5 and K2O respectively, in 06/07 season, and 17.4% and 20.9% in 07/08 season, for N there was no reduction (Table 3). Table 3. Applied average doses and yield on coffee in seasons for uniform and variable fertilizer application rates. N P2O5 K2O Yield Season Treatment −1 −1 −1 −1 (kg ha ) (kg ha ) (kg ha ) (kg ha ) VR 340 13.7 ± 4.4 427.6 ± 25.6 2753 ± 507 06/07 UR 340 18.0 431.3 2835 ± 383 VR 274 ± 56.4 7.1 ± 4.5 104.3 ± 45.5 702 ± 152 07/08 UR 278 8.6 262.3 655 ± 100 Figure 2 shows the inputs of fertilizers, as it is possible to observe in the 2006/2007 season there was no variation in the application of nitrogen in the variable rate treatment and application of phosphorus and potassium occurred more uniformly. However, in the 2007/2008 season the greatest range occurred for all fertilizers in the variable rate treat- ment. AgriEngineering 2021, 3 821 Figure 2. Applied fertilizer doses in seasons 2006/2007 and 2007/2008. 3.2. Energy Flows −1 −1 The input energy was 28.2 GJ ha for VR and 28.3 GJ ha for UR in season 06/07 and −1 −1 22.2 GJ ha for VR, 22.4 GJ ha for UR in season 07/08. When considering average yield, −1 it was observed in season 06/07 that the difference between treatments was 82 kg ha (2.9%), which resulted in a very similar incorporation of energy between the two treat- −1 −1 ments, 10.7 MJ kg for VR and 10.2 MJ kg for UR. However, for 07/08 harvest season, it −1 −1 was 30.7 MJ kg for VR and 34.9 MJ kg for UR ( Table 4; Table 5). The maximum value energy embodied by nitrogen fertilizer in every kilo of coffee in −1 −1 VR was 14.1 MJ kg and minimum was 6.7 MJ kg in season 06/07. For season 07/08 the −1 −1 maximum value was 53.8 MJ kg and 19.3 MJ kg for minimum value (Table 4). Nitrogen accounted for 85% of the energy incorporated by fertilizers. Implementing the official fertilizer recommendation and using more detailed input information, VR reduced overall fertilizer application in both seasons and the differences between the spatial patterns of each treatment for embodied energy maps by fertilizer are shown in Figure 3. AgriEngineering 2021, 3 822 Table 4. Inputs and outputs used for energy calculation according to area, harvest, and treatment. −1 −1 −1 N (MJ kg ) P2O5 (MJ kg ) K2O (MJ kg ) Season Treatment Max Min mean ± s Max Min mean ± s Max Min mean ± s VR 14.14 6.7 9.09 ± 1.4 0.12 8.12E−08 0.06 ± 0.02 1.84 0.71 1.08 ± 0.24 06 07 UR 14.14 6.7 9.05 ± 1.3 1.84 0.061 0.08 ± 0.012 1.62 0.78 1.03 ± 0.15 VR 53.8 19.3 29.6 ± 5.1 0.5 2.32E−07 0.12 ± 0.08 2.58 0.34 1.0 ± 0.4 07 08 UR 43.8 19.6 32.1 ± 4.4 0.23 0.1 0.16 ± 0.02 3.74 1.67 2.7 ± 0.4 Table 5. Energy summary of NPK fertilizers and yield for each treatment, area, and crop. Difference between treatments and % of the difference between treatments. Fertilizers Δ Δ Season Treatment −1 (MJ kg ) (MJ) (%) VR 10.7 ± 2.2 06/07 −0.5 −4.6 UR 10.2 ± 1.5 VR 30.7 ± 5.3 07/08 4.2 12 UR 34.9 ± 4.8 Figure 3. Maps of embodied fertilizer energy for UR, VR, and the difference between treatments (VR-UR) over two sea- sons. The energy indicators are presented in Table 6. Energy efficiency in this study was determined by measuring the net energy from the fertilization of the arabica coffee. The comparison of the net energy of the fertilization in VR and UR arabica coffee was −0.98 MJ −1 −1 kg coffee in VR treatment and −0.48 MJ kg coffee in season 06/07; however, in season −1 07/08 the yield was low and net energy was the 20.98 MJ kg coffee in VR and 25.18 MJ −1 kg coffee in UR. EROI was higher for UR in 06/07 and for VR in 07/08. The energy incor- porated (EI) was similar for both in 06/07, and slightly lower for VR in 07/08. AgriEngineering 2021, 3 823 Table 6. Energy performance of evaluated treatments. EI EB EROI −1 MJ kg Season Treatment VR −0.98 0.90 10.7 06/07 UR −0.48 0.95 10.2 VR −20.98 0.31 30.7 07/08 UR −25.18 0.28 34.9 4. Discussion Studies that have attempted to measure the economic benefits of variable rate over traditional management typically test the claim that variable rate provides input savings without compromising yield, or that using the same amount of inputs increases yield . Although there is a minute difference in incorporated energy between treatments in sea- son 06/07 (Table 5), our results also point to energy gains when using the variable rate, be- sides observing a reduction in the amount of fertilizer. The biggest difference was in the 07/08 harvest, obtaining an energy reduction value −1 of 4.2 MJ kg of coffee. So, variable rate application presented a reduction in input energy and an increase in the output energy availability, that is, productivity. The variable rate application of fertilizer on coffee was evaluated by Molin et al. (2010)  and they observed a similar trend, with the VR providing 23% savings in phos- phorus and 13% increase in potassium consumption when compared to UR. Regarding productivity, they observed that plots that received fertilizers at VR showed a productiv- ity increase of 34% compared to UR. In our study we observed a productivity increase of 7% only in season 07/08. We observed a reduction for VR in the use of all fertilizers in the 07/08 crop, with a reduction of 21% for K2O and 17% for P2O5 and in the 06/07 crop, the reduction in the use of phosphorus was 24%. When evaluating the energy demand in the orange crop, Colaço et al. (2020)  ob- served that nitrogen was the most demanding input in terms of energy, accounting for 66% and 73% of the total fertilization energy incorporated in treatments VR and UR of orchard 1, and for 56% and 74% in treatments VR and UR of orchard 2. In our study, nitrogen, as the main energy input, represented 89% of the energy in- corporated by fertilizers. The magnitude of N for the energy input was expected, since N has the highest energy index among the evaluated inputs, that is, it requires more energy to be produced than other inputs. As observed through the results in the 06/07 season the energy incorporation between variable and fixed rates showed a small difference, being smaller in the fixed rate since fertilization of nitrogen was the same value in both treat- −1 ments and the yield of the 10 kg ha was larger in UR. In the 07/08 season, the amount of nitrogen applied was different and resulted in a difference of 4.2 MJ kg of incorporated energy between the two treatments. Despite the fact that greater nutrient surpluses cannot necessarily be interpreted as resulting from excessive nutrient application because coffee is perennial, this can, how- ever, indicate that nutrient losses were more likely to occur in the UR treatment than in the VR, as we observed a reduction in excess phosphorus and potassium in both seasons in the VR. VR treatment reduced excess P2O5 in both seasons. Other studies have evaluated the energy sustainability arabica coffee production sys- tems in Brazil. In Muner et al. (2015)  three production systems were evaluated, and the results showed that energy intake related to fertilizer inputs had statistical differences among the three cropping systems. Cultivation systems with good practices showed higher energy consumption, that such systems had higher fertilizer consumption, fol- lowed by conventional systems and organic coffee farms. The reduced contribution re- garding fertilizer inputs of organic coffee farms was justified by its distinctive nutritional AgriEngineering 2021, 3 824 management, which adopted organic fertilization as the base of plant nutrition. These au- thors, highlighted that the highest energy intake of conventional systems (65.9%) was due to an intensive use of fertilizers, which is widely applied in conventional systems. Thus, −1 −1 the embodied energy was 12.4 MJ kg for conventional systems, 4 MJ kg for organic −1 coffee farms, and 14.3 MJ kg for cultivation systems with good practices. In our study −1 the embodied energy was between 10 to 30.2 MJ kg for a year with high productivity and a year with low productivity, respectively. So, another aspect that should be noted is productivity, as coffee presents a biannual production  which is possible to observe in the result in energy embodied ( Table 4; Table 5). This aspect resulted in a 3-fold increase in the embodied energy per kg of coffee in the low production season. The energy intensity (EROI) was higher in the VR treatment in the 07/08 season. This indicates that the VR is more energy-efficient because it incorporates less per kg of coffee produced. Data with the same trend are found by Colaço et al. (2012)  when they eval- uated the VR treatment in wheat and obtained higher EROI values compared to the values obtained with the UR treatment.  estimated Energy Balance and Green House Gas Emission on Smallholder Java Coffee Production of Indonesia, the results showed that the energy value obtained for the arabica coffee transportation and production were 152.36 −1 −1 MJ t and 74.54 MJ t , respectively, while the robusta coffee had an energy value of 158.68 −1 −1 MJ t and 162.74. MJ t , respectively. These authors observed a net energy value positive of 9.1 and 8.8 for arabica and robusta coffee respectively, different from the values ob- served in our study. 5. Conclusions The analysis of an activity through the lens of energy is an important tool for observ- ing production systems that use precision agriculture as a management tool, as it can pro- vide subsidies to monitor the energy potential of crops. The energy indicators pointed to precision agriculture through variable rate as a possible tool capable of improving energy sustainability in a coffee production system. Embodied fertilization energy was reduced −1 by approximately 4.2 MJ kg coffee in a season with a nitrogen rate variable. The VR fer- tilization resulted on savings of 24.0% and 17.4% on P2O5 and K2O respectively, in the 06/07 season, and 17.4% and 20.9% in the 07/08 season. As the difference between the results was small, it is suggested that future studies work with a larger number of crops and areas. In addition, for future studies, it is sug- gested that the energy assessment takes into account the entire production cycle, allowing the total balance of the activity to be observed. Author Contributions: Conceptualization, G.A. and M. M.; methodology, G.A.; M. M. and T.L.R.; software, M.M.; data curation, G.D.C.F. and J.P.M.; writing—original draft preparation, G.A. and M.M.; writing—review and editing, G.A.; M.M.; T.L.R.; G.D.C.F. and J.P.M.; supervision, T.L.R. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Conflicts of Interest: The authors declare no conflict of interest. References 1. Embrapa; Cafés do Brasil. 2021. Valor Bruto da Produção. Available online: https://www.embrapa.br/busca-de-noticias/-/no- ticia/63076571/faturamento-bruto-estimado-dos-cafes-do-brasil-da-especie-arabica-corresponde-a-78-e-de-conilon-a-22-do-to- tal-das-lavouras-cafeeiras (accessed on 11 June 2021). 2. Molin, J.P.; Motomiya, A.V.A.; Frasson, F.R.; Faulin, G.D.C.; Tosta, W. Test procedure for variable rate fertilizer on coffee. Acta Sci. Agron. 2010, 32, 569–575. AgriEngineering 2021, 3 825 3. Mulla, D.J.; Bhatti, A.V.; Kunkel, R. Methods for removing spatial variability from field research trials. Adv. Soil Sci. 1990, 13, 201–213. 4. Lowenberg-DeBoer, J.; Erickson, B. Setting the record straight on precision agriculture adoption. Agron. J. 2019, 111, 1552–1569. 5. Molin, J.P.; Faulin, G.D.C. Spatial and temporal variability of soil electrical conductivity related to soil moisture. Sci. Agric. 2013, 70, 01–05. 6. Colaço, A.F.; Pagliuca, L.G.; Romanelli, T.L.; Molin, J.P. Economic viability, energy and nutrient balances of site-specific fertili- zation for citrus. Biosyst. Eng. 2020, 200, 138–156. 7. Yang, C.; Everitt, J.H.; Bradford, J.M. Comparisons of uniform and variable rate nitrogen and phosphorus fertilizer applications for grain sorghum. Trans. ASAE 2001, 44, 201–209. 8. Wang, D.; Prato, T.; Qiu, Z.; Kitchen, N.R.; Sudduth, K.A. Economic and environmental evaluation of variable rate nitrogen and lime application for claypan soil fields. Precis. Agric. 2003, 4, 35–52. 9. Johnson, R.M.; Richard, E.P. Variable-rate lime application in Louisiana sugarcane production systems. Precis. Agric. 2010, 11, 464–474. 10. Thomason, W.E.; Phillips, S.B.; Davis, P.H.; Warren, J.G.; Alley, M.M.; Reiter, M.S. Variable nitrogen rate determination from plant spectral reflectance in soft red winter wheat. Precis. Agric. 2011, 12, 666–681. 11. Swinton, S.M.; Lowenberg-DeBoer. Evaluating the profitability of site-specific farming. J. Prod. Agric. 1988, 11, 439‒446. 12. Diacono, M.; Rubino, P.; Montemurro, F. Precision nitrogen management of wheat: A review. Agron. Sustain. Dev. 2013, 33, 219‒ 13. Balafoutis, A.T.; Koundouras, S.; Anastasiou, E.; Fountas, S.; Arvanitis, K. Life cycle assessment of two vineyards after the ap- plication of precision viticulture techniques: A case study. Sustainability 2017, 9, 16. 14. Spekken, M.; Molin, J.P.; Romanelli, T.L. Cost of boundary manoeuvres in sugarcane production. Biosyst. Eng. 2015, 129, 112– 15. Colaço, A.F.; Povh, F.P.; Molin, J.P.; Romanelli, T.L. Energy assessment for variable rate nitrogen application. CIGR J. 2012, 14, 85–90. 16. Romanelli, T.L.; Milan, M. Energy performance of a production system of eucalyptus. Rev. Bras. Eng. Agrícola Ambient. 2010, 14, 896–903. 17. Kuosmanen, N. Estimating stocks and flows of nitrogen: Application of dynamic nutrient. Ecol. Econ. 2014, 108, 68–78. 18. Parris, K. Agricultural nutrient balances as agrienvironmental indicators: An OECD perspective. Environ. Pollut. 1998, 102, 219– 19. Ferreira, C.O. Análise energética de sistemas de produção agrícola. Rev. Econ. Energ. 1999, 3. 20. Odum, H.T.; Odum, B. Concepts and methods of ecological engineering. Ecol. Eng. 2003, 20, 339–361. 21. Veiga, J.P.S.; Romanelli, T.L.; Gimenez, L.M.; Brusato, P.; Milan, M. Energy embodiment in Brazilian agriculture: An overview of 23 crops. Sci. Agric. 2015, 72, 471–477. 22. Roberts, D.F.; Kitchen, N.R.; Scharf, P.C.; Sudduth, K.A. Will variable-rate nitrogen fertilization using corn canopy reflectance sensing deliver environmental benefits? Agron. J. 2010, 102, 85. 23. Colaço, A.F.; Molin, J.P.; Romanelli, T.L. Mapping nutrient energy embodiment in variable and fixed rate fertilization in orange. In Sustainable Agriculture through ICT Innovation, Proceedings of the EFITA-WCCA-CIGR Conference, Turin, Italy, 24–27 June 2013. 24. Lacerda Filho, A.F.; Palacin, J.J.F.; Rigueira, J.A.R.; Ribeiro, A.; Melo, E.C. Energy balance in the production of mountain coffee. Renew. Sustain. Energy Rev. 2014, 39, 1208–1213. 25. Muner, L.H.; Masera, O.; Fornazier, M.J.; Souza, C.V.; Loreto, M.D.S. Energetic sustainability of three arabica coffee growing systems used by family farming units in Espírito Santo state. Engenharia Agrícola 2015, 35, 3. 26. Turco, P.H.N.; Esperancini, M.S.T.; de Carvalho Bueno, O.; Oliveira, M.D.M. Balanço energético dos sistemas de produção de café convencional e irrigado. Energ. Agric. 2018, 33, 73–80. 27. Molin, J.P.; Faulin, G.D.C.; Stanislavski, W.M. Yield mapping and variable rate of fertilizers for coffee in Brazil. Acta Hortic. 2009, 1, 261–266, doi:10.17660/ActaHortic.2009.824.30. 28. EMBRAPA SOLOS. Sistema Brasileiro de Classificação de Solos; EMBRAPA Produção de Informação: Brasília, Brazil, 1999; 412 p. 29. The R Development Core Team. R: A Language and Environment for Statistical Computing; R Foundation for Statistical Computing: Vienna, Austria, 2009. 30. Venables, W.N.; Ripley, B.D. Modern Applied Statistics with S, 4th ed.; Springer: New York, NY, USA, 2002; 498 p. 31. Ribeiro Junior, P.J.; Diggle, P.J. geoR: A package for geostatistical analysis. R News 2001, 1, 15–18. 32. Cambardella, C.A.; Moorman, T.B.; Parkin, T.B.; Karlen, D.L.; Novak, J.M.; Turco, R.F.; Konopka, A.E. Field-scale variability of soil properties in central Iowa soils. Soil Sci. Soc. Am. J. Madison 1994, 58, 1501–1511. 33. Romanelli, T.L.; Milan, M. Material flow determination through agricultural machinery management. Sci. Agric. 2010, 67, 4. https://doi.org/10.1590/S0103-90162010000400001. 34. Pellizzi, G. Use of energy and labor in Italian agriculture. J. Agric. Eng. Res. 1992, 52, 111–119. 35. Sartori, S.; Fava, J.F.M.; Domingues, E.L.; Ribeiro Filho, A.C.; Shiraisi, L .E. Mapping the Spatial Variability of Coffee Yield with Mechanical Harvester. In Proceedings of the World Congress of Computers in Agriculture and Natural Resources, Iguacu Falls, Brazil, 13–15 March 2002; pp. 196–205. https://doi.org/10.13031/2013.8330. 36. Franco, G. Tabela de Composição Química dos Alimentos, 9th ed.; Editora Atheneu: São Paulo, Brazil, 1999. 307 p. AgriEngineering 2021, 3 826 37. QGIS v2.10—QGIS Development Team. QGIS Geographic Information System. Open Source Geospatial Foundation Project. 2019. Available online: http://www.qgis.org (accessed on 5 January 2021). 38. Adrian, A.M.; Norwood, S.H.; Mask, P.L. Producers perceptions and attitudes toward precision agriculture technologies. Com- put. Electron. Agric. 2005, 48, 256–271. 39. Camargo, A.P.; Camargo, M.B.P. Definição e esquematização das fases fenológicas do cafeeiro arábica nas condições tropicais do Brasil. Bragantia 2001, 60, 1–6. 40. Harsono, S.S.; Wibowo, R.K.K.; Supriyanto, E. Energy Balance and Green House Gas Emission on Smallholder Java Coffee Production at Slopes Ijen Raung Plateau of Indonesia. J. Ecol. Eng. 2021, 22, 271–283.
AgriEngineering – Multidisciplinary Digital Publishing Institute
Published: Oct 19, 2021
Keywords: Coffea Arabica L.; precision agriculture; perennial crop; energy balance