An interview and some questionnaires were conducted on each farm to determine the volume of application of synthetic nitrogen fertilizers, amendments, organic matter as well as coffee cherry production, concerning the harvest period. CO₂e is equal to the production of N₂O in Global warming potential because of the greenhouse effect, N₂O with respect to CO₂ (GWP). The N₂O_Direct - N was estimated from the sum of two main contributions, nitrogen fertilization and joint contribution of N mineralization, related to the loss of C from soil organic matter, incorporation of organic matter and contribution of N from agricultural residues (coffee leaf litter) (Klein et al., 2006):

Where:

N₂ODirect -N = annual direct emissions of N₂O-N produced from managed soils, kg N₂O-N year^-1 N₂O-Ncontribution N = annual direct emissions of N₂O-N produced by N inputs to managed soils, Kg N₂O-N year^-1 N₂O-NOS = annual direct emissions of N₂O-N from managed organic soils, kg N₂O-N year^-1 N₂O-N_PRP = annual direct releases of N₂O-N from urine and manure inputs to grazing lands, kg N₂O-N year^-1

In Equation 1, the terms N₂O-N_OS + N₂O-N_Nprp do not only take into account direct contributions (N₂O-N_{N Contribution}); therefore, Equation 2 and 3 were used after some mathematics arrangements to adjust to the coffee crop dynamics.

Where:

N₂O-N = annual direct emissions of N₂O-N produced from managed soils, kg N₂O-N/period. FSN = annual amount of N applied to soils in the form of synthetic fertilizer, kg N yearˉ¹/period. FON = annual amount of animal manure, compost, sewage sludge and other N inputs applied to soils by period. FCR = annual amount of N in agricultural waste (aerial and underground), kg N/period. FSOM = annual amount of N in mineral soils being mineralized, related to C loss from soil organic matter as a result of changes in land use or management, kg N/period. CO₂ -C Emission = annual emissions of C by application of limes, kg C period. EF1 = emission factor for N₂O emissions from N inputs, kg (N₂O-N)ˉ¹ (kg N contribution)/period.

In this research, no emissions were found from the application of limes because there were no applications of calcium limestone (CaCO₃) or dolomite (CaMg(CO₃)₂), which lead to CO₂ emissions since they are dissolved and release bicarbonate (₂HCO₃ ^-), which is converted to CO₂ and water (H₂O). The few applications made by the producers were from soluble sources of CaO, which do not contain inorganic carbon; therefore, they are not included in the calculations for the estimation of the CO₂ emissions from applications to the soil, as recommended by Klein et al. (2006).

The nitrogen volume applied in the period (kg of N) was determined from equation 2 FSN, where total nitrogen fertilization contribution was established as follows:

Where:

FSN= Kg CO₂e (per contribution of N)ˉ¹ EF1= 0.01 MN₂O/MN₂= 44/28 is the mass ratio of N₂O to N₂ molecules GWP N₂O = 298. (Eggleston et al., 2006)

The FSOM was estimated from the soil calcination method of loss on ignition, (Zhang and Wang, 2014), which also quantified the FON and FCR in an indirect way. Two soil samples were taken at a depth of 20 cm per farm, 60 in total, at two different times, T1 and T2, in which the calcined organic matter (% MOc) was measured to obtain a single % MOc corresponding to the harvest period. A soil sample was also taken to determine the bulk (dry) density by means of the cylinder method.

For the estimation of C, the soil weight was established as a function of the bulk (dry) density at a depth of 20 cm and the average % MOc. The % C of the soils was calculated based on the IPCC guidelines (Eggleston et al., 2006), which corresponded to 35% organic matter (MO). A C mineralization rate of 1.39 % was estimated as reported by (Cardona and Sadeghian, 2005), for open-air coffees; in addition, a 44/12 kg C to kg CO₂e conversion factor was used.

To quantify the CO₂e fixation rate through the accumulation of biomass for the harvest period (T1 to T2), where T1 was August 2014 and T2 was February 2015, the estimation was based on the age and planting density of the different lots of each farm because the coffee farms usually have several lots of different ages, varieties and agronomic management, which directly influences the biomass storage; for this reason, the change of existence of Carbon (ΔCB) as a function of age and weighted density of the farms was quantified. The measurements were developed at 20 sites randomly in the lot most relevant to the age and weighted density of each farm, for a total of 600 measurements per farm.

To quantify the aerial biomass of coffee at each site, the useful volume of the section was estimated according to Farfan and Rendon (2014) with Equation (4) and multiplied by the density of the coffee wood (0.91 g cm^-3), and then multiplied by the total number of farm trees, which allowed us to estimate the total aerial biomass of the coffee plantation at a given time.

Where:

Vs = Volume of the tree´s section h = Height of section R = Largest radius r = Smallest radius

To estimate the aerial biomass in T2, lower diameters were measured at the base of the stem of each axis (R). The upper diameter (r) was measured in the transition zone from woody to green stem. The height (h) included the section of the woody stem, from the base to the transition zone between the lignified stem and green stem.

The aerial biomass in T1 was projected from the same site or tree in which the biomass of T2 was evaluated. The height was determined by subtracting the T2 height from the total height of tree growth, taking into account the rate of emission of one internode per month, according to recommendations of Ramirez (2014). Therefore, the height difference between the first (upper) internode and the seventh (lower) internode was measured. The lower diameter (R) measurement was taken at a height above the base corresponding to the growth height. The upper diameter (r) was the same as that measured in T2.

The CO₂e fixation was estimated with the ΔCB equation of the existence difference method (Aalde et al., 2006), where the rate of accumulation or growth of the biomass included the sum of the aerial and underground biomass (root) for such period. A coffee biomass/aerial biomass ratio of 27% was estimated (Eggleston et al., 2006).

Where:

ΔCB = annual change in carbon existences of biomass (the sum of aerial and underground biomass) on land remaining in the same category (kg C period). C_T1 = total carbon in biomass for each subcategory of land remaining in the same category at time T₁ (kg C). C_T2 = total carbon in biomass for each subcategory of land remaining in the same category at time T₂ (kg C). C = total biomass carbon for the period T₁ to T₂. A = land area that remains in the same land use category (ha). V = volume of growing venal existence corresponding to the woody volume of the tree and excludes the branches, shoots, foliage and underground components, such as roots), (cm³ ha^-1). i = ecological zone i (i = 1 a n) j = climatic domain j (j = 1 a n) ˉ¹ R = relationship between the aerial biomass and underground biomass, kg d.m. of underground biomass (kg d.m. of aerial biomass). Where R = 0.27 CF = dry matter carbon fraction, kg C (kg d.m.)ˉ¹. Being CF = 0.5 kg C/kg of dm. (Lasco et al., 2006). BCEFS = biomass conversion and expansion factor, for expansion of growing venal existence volume to aerial biomass, tons of aerial biomass growth (cm³ of growing existences volume)^-1.

The quantification of CO₂ accumulated in each fraction of biomass has been calculated through the relationship between the total weight of a CO₂ molecule and the weight of the carbon atom.

In order to estimate the CO₂e fixation per kg of coffee cherry through the annual change equation in carbon existence of biomass, the area of land (A) was modified by the total trees of the farm and the biomass conversion and expansion factor (BCEFS) was replaced by the projection of biomass accumulation from T2 to T1.

The greenhouse gas emission and fixation balance (BefGEIcc), in kg of CO₂e per kg of produced coffee cherry (kg cc), was quantified using Equation 8 according to the guidelines of IPCC (Eggleston et al., 2006).

Four experiment units were established in the locations of Ciudad Bolivar, Giraldo, Retiro-La Ceja and Urrao. Two different parameters were measured for each experiment unit: fixation and emissions.

GHG emission estimates (N₂O-N, FSOM and CO₂-C) were carried out in the four locations on 30 different farms with potential special coffees production. The measurements were taken at two different times: one in the initial period (T1) and another one in the final period (T2).

The GHG fixation estimation was carried out in the same four locations and 30 farms. On each farm, 20 sub-samples were chosen in the most representative lots in relation to the density and age of the crop in order to determinate the annual change in the carbon existence of biomass (ΔCB). The biomass of the coffee was measured the initial period (T1) and the final period (T2).

The different elements that comprised the response variable for both the emission and fixation were analyzed for differences between the locations through an analysis of variance (ANOVA).

The Anova of average GHG emission by application of synthetic nitrogen fertilizers (kg of CO₂e per kg of coffee cherry) (FSN) showed significant differences (P<0.0001) in the mean GHG emissions from nitrogen fertilizer applications in kg CO₂e per kg of coffee cherry between the locations. Table 2 shows the comparison between the locations using the LSD test at a 0.05 significance level. Urrao emitted the most with 0.215 kg of CO₂e per kg of coffee cherry, followed by La Ceja-El Retiro and Giraldo. Ciudad Bolívar and Giraldo emitted less GHG with 0.055 kg and 0.083 kg of CO₂e per kg of coffee cherry, respectively. This was especially influenced by the fact that these locations presented the highest average cherry production in kg per tree, with 1.6 and 1.2, respectively.

Table 2

The average emission by application of nitrogen fertilizers was 0.125 kg of CO₂e per kg of coffee cherry, which accounts for 15.7% of total emissions, as reported by Noponen et al. (2012), in their studies on the quantification of the carbon footprint in conventional coffee in Costa Rica, which was 0.26 to 0.67 kg of CO₂e per kg of coffee cherry, representing 50% of all emissions. Segura and Andrade (2012) reported an emission participation of nitrogen fertilizer applications ranging from 0.033 to 0.117 kg of CO₂e per kg of coffee cherry, for a participation of 68 to 82% of the emissions. These percentages of participation in the emissions differ from the 15.7% quantified in this paper. This is probably due to the fact that these authors did not take into account the losses of C from soil MO and the decomposition of leaf litter.

The Anova of the average GHG emissions by loss of C from MOs, decomposition of leaf litter and input of MO in kg of CO₂e per kg of produced coffee cherry showed significant differences (P = 0.003) in the mean emission by loss of C from MOs, decomposition of leaf litter and input of organic matter in kg of CO₂e per kg of coffee cherry between the locations. Table 3 shows the corresponding average values per location.

Table 3

An emission range from 0.38 to 1.180 was found, with an average emission of 0.674 kg of CO₂e per kg of coffee cherry. Urrao presented statistical differences from all the other locations, duplicating the average emission of CO₂e by loss of C at 1.108 Kg CO₂e per kg of coffee cherry, in comparison with the other locations. This could be mainly because this location had the highest average content of % Moc, with 29.5%, as compared to 17.1% in Ciudad Bolívar, 15% in La Ceja-El Retiro, and 9.5% in Giraldo. Similarly, Urrao presented the lowest averages in the average production of coffee cherry per tree and planting density, with 0.82 kg of coffee cherry and 4.081 trees ha^-1, respectively.

Segura and Andrade (2012), reported 6.95 kg CO₂e per kg of coffee cherry, along with Montilla et al. (2008), who reported 6.23 kg CO₂e per kg of coffee cherry, mainly because of the organic matter inputs. These results are far from the data obtained here since this measurement was taken more directly through the soil calcination method, which is closer to that published by Hergoualc'h et al. (2012).

The Anova of average fixation of GHG by accumulation of biomass in kg of CO₂e per kg of coffee cherry per center (∆GB) did not show significant differences (P=0.117) in the mean fixation produced during the period between the locations. Table 4 presents the corresponding averages per location. However, Urrao and Giraldo fixed the most. The fixation oscillated between 0.682 and 1.459 for an average of 1.068 kg of CO₂e per kg of coffee cherry, which is the same as fixing an average of 8.941 kg of CO₂e ha^-1 year)^-1, which is lower than that reported by Segura and Andrade (2012) of 13.1 kg of CO₂e per kg of green coffee, representing an approximate equivalence of 2.6 kg of CO₂e per kg of coffee cherry, but higher than that reported by Hergoualc'h et al. (2012, 2008).

Table 4

The Anova did not show significant differences (P=0.584) in the mean emission and fixation balances between the locations. Table 5 shows the balance between average fixation and emission of GHG per locations, in kg of CO₂e per kg of coffee cherry. Generally, the production of coffee cherry in areas with a potential for specialty coffee production has a positive balance, fixing between 0.083 and 0.575 kg CO₂e per kg of coffee cherry, for an average of 0.27 kg CO₂e per kg of coffee cherry. Noponen et al. (2012) estimated the balance as a function of the changes in soil C stock, obtaining BefGEIcc values between 0.26 and 0.67 kg CO₂e per kg of coffee cherry, results that are comparable with those found in this research.

Table 5

Hergoualc'h et al. (2012), estimated the emission balance and fixation of GHG in mono-culture of coffee in Costa Rica in 3.83 Mg ha of CO₂e year, also incorporating the traditional emission sources (applications of N fertilizers, limes and organic matter), finding emissions from changes in soil C of 0.27 kg CO₂e per kg of coffee cherry, which are in line with those found in this research when doing conversions (3.2 Mg ha CO₂e per year).

These results indicate that the coffee cherry production process in these locations with the applicable agronomic processes are being carried out in a sustainable and environmentally responsible way in terms of the GHG storage.