Seven hundred fifty kilograms per hectare and per year of a balanced formula, the same drip irrigation for all, the same lot, the same variety, and the same age of plant. The only difference among the seven treatments of a two-cycle trial in Arabica coffee was how that total was distributed among three moments: early flowering, fruit expansion, and maturation. The distribution that concentrated the largest fraction in the fruit expansion and the smallest in maturation, with 250, 350, and 150 kilograms respectively, outperformed the even distribution of 250, 250, and 250 by 8.86 percent in dry grain yield and by 8.47 percent in water use efficiency, averaging over the two years. That almost nine percent did not cost an additional peso in fertilizer. It cost better decision-making regarding timing.
The previous blog of this season dealt with measuring what soil analysis cannot see in the plant, and made it clear that the supply of the land and effective absorption are different numbers. One of the most frequent reasons for that gap is temporal, and it is the focus of this blog. The crop does not consume evenly throughout the year; it concentrates demand in specific phenological windows, and a fertilizer that arrives outside those windows is exposed to losses for weeks before there is a root interested in taking it up.
None of what follows suggests applying more or changing products. The entire exercise consists of taking the amount you were already going to apply and distributing it differently, which probably makes it the adjustment with the highest return per peso invested in all nutrition, and also one of the most demanding in record-keeping, because without knowing in which week each lot is, no fractionation is possible.
Seven hundred fifty kilograms distributed in seven ways
A trial was conducted with adult Arabica coffee, five years old at the start, under drip fertigation, over two complete cycles. The annual total was set at 750 kilograms per hectare of a 20-20-20 formula, and it was always divided into three applications: one at early flowering, one at fruit expansion, and one at maturation. The control treatment distributed the total in three equal parts, while the other six tested unequal combinations while keeping the total identical. Since all management variables and all inputs remained constant, any difference between treatments is attributable to the distribution.
The results are quite clear. The treatments that loaded the largest fraction in fruit expansion achieved the best results in yield and water efficiency, while those that loaded the largest fraction in maturation performed the worst, including declines in net photosynthetic rate and carboxylation efficiency compared to the even distribution. The best treatment surpassed the control in yield and water efficiency, while the worst fell short in both.
There is a second result that is particularly interesting for those selling differentiated coffee. The same winning distribution also achieved the highest contents of total sugars, fat, protein, and chlorogenic acid in the green bean, with increases of 9.48, 3.60, 14.02, and 15.42 percent compared to the even distribution, and the best sensory score of all treatments in medium roast. Two of the distributions, both with a high load at the end of the cycle, fell below the eighty-point cup threshold. Keep that eighty-point threshold in mind, as it will come back at the end when we calculate the cost of making a mistake in timing.
Two neighboring farms can buy the same fertilizer, in the same quantity and at the same price, and end the campaign with almost nine percent difference in yield, with one of them outside the quality range that pays a premium, without either having done anything different except choosing the dates.
The crop does not consume evenly, and the curve demonstrates it
Behind that result is a biology that can be measured, and which has already been measured quite precisely in several crops.
In blueberries, a recent study calculated the requirements for nitrogen, phosphorus, and potassium every ten days across four stages: flowering, fruit set, young fruit, and fruit expansion. The expansion stage concentrated an absorption of between 2.95 and 10.55 times that of the previous stages for nitrogen, between 3.05 and 6.53 times for phosphorus, and between 2.77 and 8.54 times for potassium. These are not subtle differences; they are orders of magnitude within the same cycle, and they describe a crop that consumes little for much of the year and almost everything during a few weeks.
In coffee, the pattern is equivalent although the window is longer. The accumulation of dry matter and nutrients concentrates in the growth and grain filling phases, and monitoring the maturation of the fruit shows that the concentration and accumulation of nutrients continue to change until full maturation, with nitrogen accumulating mainly in the grain and potassium mainly in the pulp. The same body of evidence reports that yield loss can reach 60 percent when nitrogen is not applied during the reproductive stage, which is the most compelling way to say that there are weeks when missing out is very costly.
From this arises the rule that reorganizes the fertilization calendar. The correct question is not how many times a year to fertilize, but in which weeks the crop has active demand and a root capable of taking it up, and that answer is given by the phenology of the lot, not the average of the area or the entire farm. A lot that flowers three weeks later than another needs its application three weeks later, and applying to both on the same day for logistical convenience implies that one of the two will receive the fertilizer outside the window.
There is an additional implication regarding frequency. The study in blueberries found that the availability of nitrogen, phosphorus, and potassium in the soil, although it rose sharply after a high application, fell markedly by day ten. This means that during windows of maximum demand, the problem is not solved with a large dose but with moderate and more frequent doses, and that a single abundant application at the right moment still wastes a good part of itself.
Knowing if you applied in the right week
Do you know today in which phenological week each of your lots is, when the last application was made in each, and how much fraction of the annual total has been applied? In most farms, fertilization is scheduled for the entire property because that is what logistics allow, and the phenology by lot is not recorded anywhere, so fractionation becomes just another calendar and loses exactly what made it valuable. Harvis helps you record the phenological state and application date by lot, see how much fraction of the annual plan each one has applied, compare at the end what you applied against what it yielded, and have the series of several years that allows you to refine the distribution with your own numbers instead of with the table from a trial done elsewhere. Contact us, and we will set it up on the lots you already have defined.
What the harvest takes when it leaves your farm
There is a second calculation that is worth doing alongside the timing one, and that is how much leaves the lot each time a truck departs. Available estimates indicate that one ton of green coffee can carry between 33 to 63 kilograms of nitrogen, 2 to 11 of phosphorus pentoxide, and 47 to 67 of potassium oxide, with variation depending on soil and management. The order of extraction by the harvest places potassium first, followed by nitrogen, and here appears a contrast worth examining closely, because the work on nutritional standards in Hass located the tree's requirement in nitrogen, calcium, and potassium, in that order, different from the order of extraction of the fruit.
Requirement and extraction are different calculations, and neither substitutes the other. The requirement describes what the tree needs to have available to function, including everything it invests in leaves, wood, and roots that remain in the lot. The extraction describes what actually leaves the system and needs to be replenished. A fertilization plan built solely on extraction underestimates what the tree needs during vegetative growth phases; one built solely on requirement loses sight of replenishing what you exported.
The distribution within the fruit also opens a practical decision that many coffee farms already have within reach. Since most of the potassium is concentrated in the pulp and most of the nitrogen in the grain, the pulp that remains in the processing facility is a concentrated flow of potassium that either leaves the lot or returns to it depending on what you decide to do with it. Composting it and returning it does not replace the fertilization plan, but it changes part of the potassium replenishment calculation that you are currently paying for in bulk.
Fractionating is not splitting into three equal parts
The central finding of the trial is that the even distribution, which is what most farms use when they decide to fractionate, was not the best; splitting into three equal parts is an administrative decision disguised as an agronomic decision, it is convenient, easy to explain to the foreman, and does not require knowing in which week the lot is, but it implicitly assumes that the crop consumes equally in all three moments, which is precisely what the absorption curves disprove. In the trial, moving one hundred kilograms from maturation to expansion, without changing the total, produced the complete improvement of almost nine percent.
The error in the opposite direction also matters. The treatments that loaded the largest fraction at the end of the cycle had the worst performance in yield, water efficiency, and cup quality, with two of them falling below the eighty-point threshold we saved above. Applying late is not a softer version of applying on time; it is a decision that costs on two fronts at once, because the nutrient arrives when demand has already decreased and remains in the soil exposed to losses for months.
The operational rule that emerges is simple to state and demanding to execute. Identify in your crop and in your lot the window of maximum demand, which in most fruit trees coincides with fruit expansion, and assign to that window the largest fraction of the annual total. Reduce the fraction of the lower demand phases instead of adding new doses. And within the window of maximum demand, if your system allows it, split that fraction into more frequent applications, because soil availability declines within days.
Where fractionation does not compensate
The first limit is water, and it completely nullifies the exercise when it is not resolved. A fertilizer applied without sufficient moisture in the soil does not dissolve, does not move towards the root, and does not enter, so the recommendation to apply at times of highest demand always comes with the condition of adequate moisture. In rainfed systems, effective windows are defined by rainfall, not phenology, and fractionation must adjust to the intersection of the two. The trial that provides the numbers for this blog was conducted under fertigation with complete irrigation every seven days, and that is a context worth keeping in mind before copying its fractions.
The second limit is logistical and is paid in labor. Each additional entry to the lot costs labor, fuel, and coordination, and in small or difficult-to-navigate farms, that cost can eat up the gains from fractionation. The way to evaluate it is straightforward: multiply the cost of an additional entry by the number of new entries and compare it against the value of the percentage of yield you expect to recover. In systems with installed fertigation, that cost is marginal, and fine fractionation almost always pays; in manual soil application, three or four well-chosen moments are usually the reasonable ceiling.
The third limit is that the date does not fix what the source damages. The same body of literature reports that at least 30 percent of the nitrogen from urea is lost through volatilization when high doses are applied in coffee, and that loss occurs in the first days after application, regardless of how well you have chosen the week. Getting the timing right and losing a third of the nitrogen through volatilization is still inefficient fertilization. The source, incorporation, and stabilization technologies are the subject of the fourth blog of this season, and they are complementary to what we are discussing today, not alternatives.
The fourth point is about transferring the numbers. The fractions from the trial, 250, 350, and 150 out of a total of 750, correspond to a specific crop, plant age, climate, and irrigation system. What transfers is the criterion, loading the window of maximum demand, not the table.
Half a hectare, three dates, and the fertilizer bill
Your own trial fits in one lot and one campaign, and has the advantage of not requiring the purchase of anything additional. Choose a lot and divide it into two parts comparable in soil, age, and variety. Both will receive exactly the same total amount of fertilizer in the year, the same product, and the same formula, and that is what makes the exercise clean. The only difference will be the distribution.
Before applying anything, write down the expected phenology of the lot with approximate dates, marking flowering, fruit set, fruit expansion, and maturation. Walk through the lot every two weeks and note what stage it is actually in, because the written phenology serves for planning and the observed one is what matters. That two-column sheet, date and observed stage, is all the instrument you need.
The control part receives your usual distribution, whatever it may be. The treated part receives the same total shifted towards the expansion window, subtracting from the lower demand phase and without adding anything new. Note each application with date, product, quantity, and responsible party in both parts, and also note the rains or irrigations from the three days before and after each application, as that data will later explain any unusual results.
At the end of the campaign, compare four numbers between the two parts: yield, quality according to the criteria that your buyer pays, the number of entries to the lot required by each distribution, and the cost of those entries. The fertilizer does not enter into the comparison because it was the same, and that is why any difference you find is pure gain or loss from decision-making. Repeat it for a second year before changing the management of the entire farm, because a single campaign can be dominated by the weather.
The fertilizer does not get lost in the bulk, it gets lost in the date
The conversation about fertilization in the field almost always revolves around two questions: which product and how much. The evidence suggests there is a third question, with less commercial visibility and probably more return, which is when. The crop has weeks when almost everything that arrives becomes harvest and months when almost nothing that arrives finds an active root waiting for it, and the difference between getting those windows right or wrong was measured at almost nine percent yield without an additional peso of investment. No supplier is going to sell that, because there is nothing to dispatch.
What you can do starting Monday fits into five decisions. Write down the phenology of your lots with dates and walk through them every two weeks to note the observed stage, because without that record, fractionation becomes just another calendar. Stop distributing the total in equal parts and shift the larger fraction towards the fruit expansion window, subtracting it from the lower demand phase and without increasing the total, as that movement was what produced all the improvement in the trial. If you have fertigation, also split that larger fraction into more frequent applications, since soil availability declines markedly by day ten. Check the soil moisture before each application and postpone if it is not sufficient, because a fertilizer without water does not enter, no matter how correct the date is. And set up the comparison in two parts of the same lot with the same annual total, to have at the end of the campaign your own figure instead of the table from an external trial. The producer who finishes the year knowing what each distribution yielded has a calendar for the next campaign; the one who applied by habit will pay for the same fertilizer for less harvest without knowing where it went.
Do you want to keep track of phenology and applications by lot, and have at the end of the campaign the comparison between what you applied and what each distribution yielded? Contact us via WhatsApp or schedule a demo, and we will show you how Harvis turns your walks and tasks into the series you need to decide the date with data.
References Consulted
- Optimizing drip fertigation at different periods to improve yield, volatile compounds and cup quality of Arabica coffee. Frontiers in Plant Science, 14, 1148616. https://pmc.ncbi.nlm.nih.gov/articles/PMC10272449/
- Response of soil nutrient dynamics and stoichiometric characteristics in blueberry to fertilization rates. (2025). PLOS One. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0335223
- Nutrient dynamics in the berry, bean and husk of six Coffea canephora genotypes throughout fruit maturation. (2025). Scientific Reports. https://www.nature.com/articles/s41598-025-13171-4
- CND and DRIS Methods for Nutritional Diagnosis in 'Hass' Avocado Production. (2025). Horticulturae, MDPI, 11(6), 621. https://www.mdpi.com/2311-7524/11/6/621
- Sadeghian, S. y colaboradores. (2022). Determinants of nitrogen use efficiency in coffee crops, a review. Revista de Ciencias Agrícolas, 39(2). http://www.scielo.org.co/scielo.php?script=sci_arttext&pid=S0120-01352022000200198
- Salamanca-Jimenez, A., Doane, T. A. y Horwath, W. R. (2017). Nitrogen Use Efficiency of Coffee at the Vegetative Stage as Influenced by Fertilizer Application Method. Frontiers in Plant Science, 8, 223. https://www.frontiersin.org/articles/10.3389/fpls.2017.00223/full
- Covre, A. M. y colaboradores. (2016). Nutrient Accumulation in Bean and Fruit from Irrigated and Non-Irrigated Coffea canephora cv. Conilon. Emirates Journal of Food and Agriculture.
- Partelli, F. L., Espindula, M. C., Marré, W. B. y Vieira, H. D. (2014). Dry matter and macronutrient accumulation in fruits of Conilon coffee with different ripening cycles. Revista Brasileira de Ciência do Solo, 38, 214.
- Nutritional Efficiency of Coffea canephora, the Role of Genetic Variability and Nutrient Accumulation. (2025). PMC. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12115185/
- Micro-Sprinkling Fertigation Enhances Grain Yield and Nitrogen Use Efficiency by Reducing N Redundancy and Increasing Root-Water-Nitrogen Spatiotemporal Coordination. (2025). PMC. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12430560/
- Phenophase Transitions and Fertiliser-Mediated Regimes as Determinants of C-N Partitioning and Pedogenic Pathways in Tropical Agriculture. (2026). Agronomy, MDPI, 16(3), 366. https://doi.org/10.3390/agronomy16030366
- Nutrient Requirements and Fertilizer Management Strategies in Plant Cultivation. (2025). PMC. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12525752/
- Influence of Variable Chloride and Sulfur Doses as Part of Potassium Fertilization on Nitrogen Use Efficiency by Coffee. (2023). PMC. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10222592/
- Nutrients Leaching in Response to Long-Term Fertigation and Broadcast Nitrogen in Blueberry Production. PMC. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7696231/
- Oregon State University Extension Service. Nutrient Management for Blueberries in Oregon, EM 8918. https://extension.oregonstate.edu/catalog/pub/em-8918-nutrient-management-blueberries-oregon
- Food and Agriculture Organization of the United Nations. Arabica Coffee Manual, chapter on nutrition and fertilization. https://www.fao.org/4/ae939e/ae939e06.htm