Growth response surface for optimizing fertilization in Received for publication: 7 February, 2017. Accepted for publication: 24 July, 2018 Doi: 10.15446/agron.colomb.v36n2.70274 1 Corporación Colombiana de Investigación Agropecuaria - AGROSAVIA. Centro de Investigación Tibaitatá – Km 14 Vía Mosquera - Bogota, Cundina- marca (Colombia). * Corresponding author: wcardona@agrosavia.co Agronomía Colombiana 36(2), 135-142, 2018 Growth response surface for optimizing fertilization in Andean blackberry (Rubus glaucus Benth.) nurseries Superficie de respuesta en crecimiento para optimizar fertilización en mora de los Andes (Rubus glaucus Benth.) en viveros William Cardona1*, Julio Galindo1, Martha Bolaños1, and María Ramírez1 ABSTRACT RESUMEN Fertilization in soils cultivated with the Andean blackberry has been carried out empirically because there is no complete knowledge on its nutrient requirements. Therefore, the aim of this study was to estimate the effect of variable doses of N, P, K and Ca on the growth of thornless Andean blackberry in nurseries. This research was carried out in the nursery of AGROSAVIA C. I. Tibaitatá (Mosquera, Colombia) using vi- troplantlets sown in peat moss, sand and rice husks (2:1:1). A completely randomized block design with 25 treatments and 15 plants per experiment unit was used. The nutritive solution application frequency was established as every four d for three months with a volume of 44 mL/plant; the concentration was increased each month. Destructive sampling was carried out 30, 60 and 90 d after transplanting, registering plant height, leaf area, root length and volume, number of leaf lets, and leaf, stem and root dry matter. Regression models were used establishing significance (P<0.05 and <0.01) between N, P, K and Ca using SAS 9.3. Doses of 36 N, 43 P2O5, 18 K2O, and 9 CaO g/plant improved the root development in terms of dry matter and length and increased the number of leaf lets, aerial length and leaf dry matter. La fertilización en suelos cultivados con mora se ha realizado empíricamente, puesto que no se tiene pleno conocimiento de sus requerimientos nutricionales. Tomando como base lo anterior, el objetivo de esta investigación fue estimar el efecto de dosis variables de N, P, K y Ca sobre el crecimiento de la mora sin espinas en vivero. La investigación se realizó en los invernaderos de AGROSAVIA C. I. Tibaitatá (Mosquera, Co- lombia), utilizando vitroplántulas sembradas en turba, arena y cascarilla de arroz (2:1:1). Se estableció un diseño en bloques completos al azar con 25 tratamientos y 15 plantas por unidad experimental. La frecuencia de aplicación fue cada cuatro días durante tres meses, con un volumen de solución nutritiva de 44 mL/planta, incrementando la concentración cada mes. Se realizaron muestreos destructivos a los 30, 60 y 90 días después del trasplante, registrando altura de planta, área foliar, longitud y volumen radical, número de foliolos y masa seca en hoja, tallo y raíz. Se trabajaron modelos de regresión determinando significancia (P<0,05 y <0,01) entre variables utilizando el software SAS 9.3. La dosis 36 N – 43 P2O5 – 18 K2O – 9 CaO g/ planta permitió mejor desarrollo radical en masa y longitud, mayor número de foliolos, longitud aérea y masa seca foliar. Key words: v it ropla nt lets, d r y mat ter, opt i mu m dose, application rate. Palabras clave: vitroplántulas, masa seca, dosis óptima, tasa de aplicación. thornless Andean blackberry, which was obtained from the Colombian coffee-growing region, possibly through spontaneous generation (Morales and Villegas, 2012). It has a higher number of productive branches (female) and a lower number of non-productive branches (whips). Additionally, it shows higher tillering than the Andean blackberry with thorns (15 to 20% superior) (Bernal and Díaz, 2006), inferring a consumptive use that is different from that of the genotype with thorns. In this sense, vari- ous researchers have evaluated nutrient consumption in the thornless Andean blackberry, comparing it with the one with thorns. Bolaños-Benavides et al. (2014) carried out applications of 36 g, 43 g, 18 g and 9 g of N, P2O5, K2O Introduction Scientific and technical literature on the Andean black- berry (Rubus glaucus Benth.) yield regarding nutrient re- quirements and extraction levels is still in its initial stages, and there are still big differences in fertilizer applications (sources and amounts). This information is of utmost importance for achieving higher production and yield (Santos et al., 2014). Currently, the Andean blackberry is produced with a mixture of diverse materials and agro- nomic management practices, urgently requiring a dif- ferent type of management. One of these materials is the http://dx.doi.org/10.15446/agron.colomb.v36n2.70274 wcardona@agrosavia.co 136 Agron. Colomb. 36(2) 2018 and CaO per plant, respectively, finding that thornless plants had a higher height with significant differences (P≤0.05) (9.93 cm and 20.83 cm, 30 and 90 d after trans- planting (dat), respectively), as compared to plants with thorns, with height values of 8.11 and 20.55 cm at 30 and 100 d of accumulating nutrients, respectively (Cardona et al., 2016). However, the experience acquired by producers and technicians has led to fertilizing 60 d after planting, and this practice continues every 90 d until one year of establishment is reached (Betancur-Cardona et al., 2014). Adequate and accurate nutrient application for plants starting in the initial growth stages (Marschner, 2011) favors the rooting process and gives seedlings higher vigor (Razaq et al., 2017). Based on this empirical practice and in order to have plants with adequate nutrition when transplanted to uncontrolled conditions, the aim of this study was to estimate optimum doses of N, P, K and Ca in thornless Andean blackberry plants under nursery conditions, developing their genetic potential once they have been established as a commercial crop. Materials and methods Study site location This study was carried out in the Tibaitatá Research Center of the Corporación colombiana de investigación agropecuaria - AGROSAVIA located at 4º41’43.1349’’ N and 74º12’18.7666’’ W, with a mean temperature of 13.1°C, a relative humidity of 80% and an altitude of 2,600 m a.s.l. Plant material Thornless Andean blackberry vitroplantlets were propaga- ted in the micropropagation laboratory of AGROSAVIA. Then, the plant material was placed in transparent trays with inert peat moss, adding 1 g of the fungus Glomus proliferum Dalpé et Declerck strain GB02 (60 to 70 spores g-1) for the hardening stage (30 d). Finally, each plantlet was sown in recipients with 500 g of peat moss: sand: rice husk at a volume: volume ratio of 2:1:1. Experimental design and statistical analyses The treatments were analyzed using a composite central design, defining 25 treatments and applying the following equation: 2k + 2*k + 1 (where k is the number of factors: N, P, K and Ca). This experiment was established in a greenhouse (the cli- matic conditions during the evaluations had a maximum temperature of 21°C, with an average of 19°C and a mini- mum of 17°C; in the case of relative humidity, a maximum humidity of 71%, an average of 57% and a minimum of 35% were registered) in complete randomized blocks with 25 treatments, three (3) blocks and 15 plants per experimental unit. The application frequency was every four d for three months, i.e. 24 applications, administering a volume of 44 mL of nutrient solution per plant. The concentration of the nutrient solution applied increased every month according to plantlet development (Tab. 1). The basis for this nutrient solution was the Hoagland and Amon solution, N, P, K and Ca concentrations according to the fertilization treatment and maintaining the original Mg, S and micronutrient concentrations. Destructive sampling was carried out 30, 60 and 90 d after transplanting (dat). The following vari- ables were registered: plant height (PH) using a ruler with millimeters and measuring from the base of the stem to the growth point, leaf area (LA) using a CI - 202 portable measuring device (CID Bio-Science, USA), and root volume (RV). In this method, the entire root was placed inside the test tube with a known water volume, with the root system volume corresponding to the displaced water. Other vari- ables included the number of leaf lets (NL), leaf dry matter (LDM), stem dry matter (SDM), root dry matter (RDM), and total dry matter or plant dry mater (TDM). In each destructive sampling, the plants were processed as follows: a) petiole and lamina, b) stem, and c) roots. To measure the dry matter, the material was placed in paper bags and placed TABLE 1. Nutrient level specifications for constructing the treatment matrix for 30, 60 and 90 dat according to the Central Composite Design. Code level N P2O5 K2O CaO g/plant Days after transplanting (dat) 30 60 90 Total 30 60 90 Total 30 60 90 Total 30 60 90 Total Lower axial (LA) -1.41 1.6 2.4 3.0 7.0 0.5 0.7 0.9 2.1 2.0 2.4 6.2 10.6 1.6 1.9 3.0 6.5 Low dose (LD) -1 2.3 4.0 5.7 12.0 2.1 3.1 3.8 9.0 4.7 5.7 7.5 17.9 2.0 3.0 4.0 9.0 Average dose (AD) 0 4.9 6.3 12.8 24.0 6.0 9.0 11.0 26.0 9.6 11.4 15.0 36.0 4.2 4.9 6.0 15.1 High dose (HD) 1 4.8 13.4 17.8 36.0 12.0 14.3 16.7 43.0 12.4 17.0 24.6 54.0 6.0 7.0 8.0 21.0 Upper axial (UA) 1.41 7.3 14.2 19.5 41.0 14.0 16.8 19.2 50.0 15.7 21.4 24.4 61.5 6.7 7.8 9.0 23.5 137Cardona, Galindo, Bolaños, and Ramírez: Response surface fertilization for optimizing Andean blackberry (Rubus glaucus Benth.) growth in nurseries in an oven at 70ºC for 24 to 48 h until constant weight; then, an analytical balance (Mettler Toledo Model AB204) was used to measure the dry matter. The statistical analysis included regression models selecting significant terms (P<0.05) of a saturated model with lineal and quadratic effects for doses of N, P, K and Ca with their double interactions, simple effect and interaction over time. The data processing was carried out with SAS 9.3 (SAS Institute, Cary, NC, USA). Results and discussion At 60 d after transplanting, the Andean blackberry plants presented differential values for the recorded allometric variables, showing higher averages for the leaf area, plant height and root length, as compared to 90 d after transplanting (Tab. 2). In contrast, the variables related to accumulation of dry matter in each organ and number of leaf lets presented incremental averages over time. Plant weight accumulation According to the ANOVA (not included), the dry matter accumulation in the leaves, stems, roots and in the entire plant (total) was significantly inf luenced (P<0.05) by the doses of N, K, P and Ca, with some interactions descri- bed below. In all of the plant organs, the biomass varied because the Ca dose effect followed quadratic tendencies, but the response rate was variable over time, indicating that the Ca requirement increased with plant growth, but an excess Ca dose can be detrimental. The P doses had a significant quadratic effect (P<0.05) on RDM accumulation, at a rate that varied according to plant growth; hence, the temporal role in initial root sys- tem growth was found; likewise, K can modify P response through an interaction. These results contrast with those found by Ascencio and Lazo (2001), who argued that P absorption by roots is better explained by total root length and not by root system dry weight, which were taken as a reference point. On the contrary, Bayuelo-Jiménez et al. (2011, 2012) stated that root formation is not related to plant size, but to the efficiency in phosphorus absorption. On the other hand, Benavides-Mendoza (2011) stated that root system dry matter can increase its length. "increasing root dry weight might ref lect an increase in root length" However, when developing biomass in any of the plant organs, the Ca dose effect interacted with the N dose effect consistently over time. This means that the excess Ca dose inhibited or reduced the effect of the N dose. Castaño et al. (2008) evaluated nutritient deficiencies in the Andean blackberry, finding that a lower RDM was registered in the Ca treatment, stimulating cellular division in the meristematic points and stabilizing cells during division. Riveras et al. (2015) mentioned the role that Ca plays in the regulation of nitrogen absorption. A P x Ca effect was observed in the leaves (P<0.05), indicating that the excess Ca also affected the plant response to the P dose in the LDM. The response rate of the N dose was affected by plant growth only in the LDM (P<0.01). Likewise, Yang et al. (2014) stated that absorption in leaves changes over time and has a narrow relationship with the growth rate. More- over, Castaño et al. (2008), in a treatment without nitrogen, found the lowest SDM resulted from the role of this element in growth, tissue formation and dry matter accumulation. TABLE 2. Descriptive statistics of the analyzed variables. Variable 30 dat 60 dat 90 dat Mean Min Max SD Mean Min Max SD Mean Min Max SD LA (cm2) 21.735 0.050 70.120 10.824 98.673 4.080 272.380 77.187 48.980 0.720 279.210 48.872 PH (cm) 9.981 4.000 16.500 2.416 13.007 2.000 29.100 4.345 12.212 2.500 29.500 5.268 RL (cm) 11.071 2.500 24.000 4.054 14.780 1.500 28.000 5.007 10.802 1.000 28.000 5.115 NL (n) 6.824 1.000 17.000 2.060 8.726 3.000 21.000 3.553 13.043 3.000 31.000 5.498 LDM (g) 0.067 0.010 0.226 0.034 0.180 0.005 0.924 0.125 0.265 0.018 1.340 0.252 RDM (g) 0.022 0.003 0.068 0.012 0.054 0.003 0.551 0.047 0.064 0.001 0.325 0.061 SDM (g) 0.022 0.001 0.096 0.013 0.054 0.003 0.272 0.042 0.092 0.005 0.741 0.105 TDM (g) 0.100 0.003 0.324 0.052 0.261 0.001 1.339 0.190 0.383 0.001 2.081 0.393 RV (cm3) 0.420 0.050 5.000 0.431 0.698 0.030 15.000 0.974 0.774 0.050 4.000 0.803 Min: Minimum; Max: Maximum; SD: Standard deviation; LA: Leaf area; PH: Plant height; RL: Root length; NL: Number of leaflets; LDM: leaf dry matter; RDM: root dry matter; SDM: stem dry matter; TDM: total dry matter; RV: Root volume. 138 Agron. Colomb. 36(2) 2018 The R DM was affected by ever y evaluated nutrient; however, according to the analysis of variance, N and K had a stable response over time, indicating that the root absorbs these elements in the amount required by growth. Castaño et al. (2008) found that the highest RDM average was obtained in the treatment without nitrogen, as seen by Cánovas et al. (2016), who stated that a decrease in the nitrogen concentration promotes an increase in root system length during initial growth stages. The SDM responded to the K and Ca doses at a different rate according to the plant growth; however, there was no inter- action between the K and Ca doses, indicating independent functions in this organ. One of the main functions of Ca is the formation of the structural part of the protopectin, keeping cells together as they are located in the medium layer and the primary cell wall (Al-Shemmar et al., 2013). The surface analysis for the RDM response (Tab. 3) showed a positive response to the N dose increase; however, this effect was reduced with the higher doses of Ca. Likewise, there was a quadratic effect towards the maximum RDM, with an increase in the P doses (P2), but it was reduced when high doses of K were applied (Fig. 1). Therefore, doses of Ca and K must be low (CaO ≤ 9 and K2O ≤ 18) in order to favor positive N and P effects. Kim and Li (2016) found that high doses of P favored biomass accumulation in the root system when analyzing the effects of reduced phosphorus on shoot and root growth, partitioning, and phosphorus utilization efficiency in Lantana camara. The most favor- able treatment with these conditions was the one that had the doses of 36 g of N, 43 g of P2O5, 18 g of K2O and 9 g of CaO per plant, which match the highest average observed between the treatments. Another proposed dose is 36 g of N, 50 g of P2O5, 18 g of K2O and 9 g of CaO per plant; this could increase RDM by 12% (0.15 units of weight) accord- ing to the established model. TABLE 3. Regression coefficient estimation and associated error per regression model in terms of RDM, LDM and SDM. Terms RDM LDM SDM Estimation Error Estimation Error Estimation Error Intercept -0.0241 0.0126 -0.9188 0.1684 -0.3721 0.0673 N 0.0050 0.0006 0.0168 0.0036 K 0.0050 0.0022 Ca 0.1295 0.0217 0.0536 0.0087 N x P 0.0001 1.6E-05 N x Ca -0.0002 3.3E-05 -0.0004 0.0002 P x K -2.6E-05 1.0E-05 P x Ca -0.0003 0.0001 K x Ca 0.0001 1.9E-05 P2 3.0E-05 8.0E-06 0.0001 4.1E-05 K2 -0.0001 2.9E-05 Ca2 -0.0041 0.0007 -0.0018 0.0003 r2 0.52 0.54 0.5 All of the coefficients in the table are significantly different from zero (0) according to the t-test at 5% significance. Empty cells indicate terms that were not included in the model because they did not contribute significantly. r2: coefficient of determination; RDM: root dry matter; LDM: leaf dry matter; SDM: stem dry matter. FIGURE 1. The surface analysis for the root dry matter (RDM) response to the P x K interaction. In the case of the LDM (Tab. 3), there was a positive effect, with an increase in the N dose that was reduced with high amounts of Ca (CaO ≥ 15 g/plant) (Fig. 2). Accordingly, the Ca effect had a quadratic tendency, indicating that the LDM followed a curve as a function of the Ca dose, with a maximum response at 15 g of CaO/plant according to the derivative calculations. Moreover, P had a positive effect on 10 20 40 30 R oot dm (g/pant) 0.05 0.06 0.07 0.04 P dose (g/pant) K d os e ( g/p an t) 20 30 50 40 139Cardona, Galindo, Bolaños, and Ramírez: Response surface fertilization for optimizing Andean blackberry (Rubus glaucus Benth.) growth in nurseries the LDM, although it may have been unfavorably affected by the high Ca doses. Finally, the model discarded K, with lower doses chosen because of economic costs (K2O ≤ 18 g/plant). According to these conditions, the dose of 36 g of N, 43 g of P2O5, 18 g of K2O and 9 g of CaO per plant was favorable for the LDM. However, with this model, it is possible to predict that, with a dose of 15 g of CaO per plant (optimum value by derivation), a 4% higher point is reached (0.52 units of weight). For the observed averages, the dose of 41 g of N, 26 g of P2O5, 36 g of K2O and 15 g of CaO per plant also showed a good LDM result. This treat- ment had a Ca level that matched the level recommended by the model, and the effect of reducing the P dose was compensated by the higher N dose; nevertheless, the K dose can be reduced to 18 g/plant according to the model with- out changing the result in the LDM, which would reduce costs for potassium fertilizers. In other words, adjusted proposal is 41 g of N, 26 g of P2O5, 18 g of K2O and 15 g of CaO per plant; however, this result should be validated. Parra-Terraza et al. (2010) found that a higher LDM may be related to higher leaf area values, decreasing the stress caused by Solanum lycopersicum plantlet transplantation, facilitating establishment. For the SDM (Tab. 3), the model had a quadratic tendency in response to the K dose and a maximum weight with 40 g of K2O per plant. Parra-Terraza et al. (2010) assessed the nitrate/ammonium/urea ratio and potassium concentra- tions in S. lycopersicum plantlet production, finding that K concentrations did not present significant differences in leaf, stem and root dry weight. As in the results for the LDM, there was a positive Ca effect that followed a quadratic tendency with a maximum LDM with 15 g of CaO per plant (Fig. 3). Finally, a positive N x P interaction was found, indicating an advantage in increas- ing their doses for higher SDM effects (N>36 g, P2O5>43 g/ plant). The closest treatment for the conditions suggested a dose of 41 g of N, 26 g of P2O5, 36 g of K2O and 15 g of CaO per plant. This treatment had the best response although, when using this model, a maximum SDM with a dose of 43 g of P2O5 per plant is expected, generating an increase of 24% in the SDM (0.25 units of weight); however, this result should be validated. The adjustment of 36 g to 40 g of K2O per plant did not have a strong effect on the SDM according to the model (<1%). Furthermore, the application of N and P was more impor- tant for the leaf and root development, whereas the high application of K and Ca was unfavorable. In this plant development stage, when root system development must be strengthened, the treatments with good levels of N and P must be selected even if they are not the most favorable to increasing stem biomass. Allometric variables The allometric variables were significantly inf luenced (P<0.05) by the N, P, K, Ca doses and their interaction, as described below. On one side, the K and P negative inte- raction had a significant effect on the root volume (RV) (Tab. 4). The effect of Ca was positive and followed the curve with a peak when the Ca dose was 12.5 g per plant. Additionally, the effect of N was positive and lineal, and, according to the model, this treatment achieved high RV values with a dose of 36 g of N, 9 g of P2O5, 54 g of K2O and 9 g of CaO per plant, with an average of 1.24 cm3. However, the Ca dose could be increased to 12.5 g of CaO per plant in order to achieve a 9% higher volume (1.35 cm3). Then, in this proposal, a dose of 9 g of P2O5 per plant is low, but produces a higher response in RV when 54 g of K2O per FIGURE 2. The surface analysis for the leaf dry matter (LDM) response to the N x Ca interaction. FIGURE 3. The surface analysis for the stem dry matter (SDM) response to the combined effect of K and Ca. 15 20 30 25 35 Leaf dm (g/pant) 0.2 0.3 0.4 0.1 N dose (g/pant) Ca d os e (g /p an t) 10 12 16 14 18 20 50 20 40 30 Stem dm (g/pant) 0.16 0.18 0.20 0.14 0.12 K dose (g/pant) Ca d os e (g /p an t) 10 12 16 14 18 20 140 Agron. Colomb. 36(2) 2018 plant is applied. RV results would be better if P dose were increased and K dose were decreased, according to P x K interaction. So, in the data observed, the dose of 24 g of N, 50 g of P2O5, 36 g of K2O and 15 g of CaO per plant had a higher average (1.46 cm3), 30% higher than what was expected by the model (1.01 cm3). Another treatment that can be proposed is 36 g of N, 43 g of P2O5, 18 g of K2O and 12 CaO per plant, which had an expected average of 1.30 cm3 and produced the best RDM development. These results contrast with those found by Koning et al. (2015), who evaluated the effects of nitrogen and phosphate fertilization on the leaf nutrient content, photosynthesis, and growth of Fallopia sachalinensis cv. ‘Igniscum Candy’; they observed a significant increase when the nitrogen concentration was increased in the growth media. Nitrogen had a significant effect (P<0.05) on root system growth (root length and volume) over time; these results agree with the findings of Rivera-Espejel et al. (2014), who evaluated the fertilization of S. lycopersicum with nitrate and ammo- nium, finding significant differences when compared to the absolute control (root that only grew with water). According to Frias-Moreno et al. (2014), high N doses induce a higher development of the stem and leaves in S. lycopersicum, but decrease root system growth. For leaf area (LA) (Tab. 4), the Ca effect was very strong according to the model and showed a response curve with a peak when the CaO dose was 14.6 g per plant (by model derivation). This result is very close to the mean model dose (i.e. 15 g of CaO per plant). In this respect, Vargas-Bolívar et al. (2009) evaluated nutrient deficiencies in S. quitoense in nurseries, finding that the absence of calcium had a significant positive effect on LA as compared to plants irrigated only with water and calcium. Calcium interacts negatively with P; however, the dose effect of P was positive, such as N and K. Therefore, high doses of N, P and K increase LA (≥ 36 g of N, ≥ 43 g of P2O5, and ≥ 54 g of K2O per plant). Xiang et al. (2012) found a decrease in height, leaf area and stem diameter in Glycine max plants cultivated without phosphorus fertilization because this nutrient is essential as a tissue constituent that acts directly on plant height (PH). Because of the Ca x P interaction, a P dose should not be too high, with a Ca dose near the medium level. The results established that the dose of 41 g of N, 26 g of P2O5, 36 g of K2O and 15 g of CaO per plant was most similar to the assessed conditions, obtaining the highest result, 26% higher than what was expected by the model. Luo et al. (2017) stated that a lack of nitrogen inhibits elongation and cellular division. Likewise, White et al. (2015) argued that nitrogen is present in many essential compounds, and because of that a plant grows slowly when there is low availability of this nutrient. These results agree with results found by different authors for plant height (PH) when nitrogen is applied together with phosphorus and potassium (Dhillon et al., 2011; Xiang et al., 2012). For the leaf let number (LN) (Tab. 4), the model indicated the importance of Ca dose management, showing a maxi- mum response level with 11.1 g of CaO per plant according to the derivation model. Moreover, there was a negative Ca interaction with N and P. For the N dose, there was a positive effect, indicating that high doses are recommended for increasing LN (≥ 36 g of N per plant). Nevertheless, including P, the model had a quadratic tendency, where an increase in dose produced a maximum LN response. P doses associated with the maximum were out of the range of the experimental design, so the LN response to 50 g of P2O5 per plant is not known. Finally, K did not inf luence the LN; and therefore, low doses are recommended for decreasing costs (≤ 18 g of K2O per plant). Nonetheless, the treatment with the best response was 36 g of N, 43 g of P2O5, 18 g of K2O and 9 g CaO per plant, but, according to the model, a slightly superior response was obtained with 11 g of CaO (19.3 leaf lets as compared to the 19 leaf lets obtained with 9 g of CaO per plant). Additionally, Preciado-Rangel et al. (2002) mentioned that the number of leaves is related to photosynthesis and, therefore, to a higher production of carbonate skeletons that are used or stored in the stem. On the contrary, Magdaleno- Villar et al. (2006) stated that the number of leaves is not a reliable indicator of plantlet production because it mostly depends on the age of the plant. In addition, the PH (Tab. 4) was defined primarily by the Ca dose, with a quadratic tendency and a peak when the CaO dose was 6 g per plant. This element interacts negatively with N and P. Since the N and P effects were positive in the aerial length, high doses are better (≥36 g of N, and ≥43 g of P2O5 per plant). Li et al. (2016) reported that phosphorus contributes to the growth and elongation of internodes, stimulating cellular division. The effect of K was not significant, so the application of low doses is recommended because of economic costs (≤18 g of K2O per plant); this effect is similar to the one found by Sá et al. (2014), who did not register an effect of potassium applications on the growth of Corymbia citriodora plants, 141Cardona, Galindo, Bolaños, and Ramírez: Response surface fertilization for optimizing Andean blackberry (Rubus glaucus Benth.) growth in nurseries even when they were cultivated in a soil with a low content of this nutrient. Therefore, the most favorable treatment according to these conditions was 36 g of N, 43 g of P2O5, 18 g of K2O and 9 g of CaO per plant; moreover, it had the highest average (20.8 cm). A slight improvement with a dose of 6 g of CaO per plant is expected, but only 2%. Conclusions The dose of 36 g of N (high), 43 g of P2O5 (high), 18 g of K2O (low) and 9 g of CaO (low) per plant allowed good development of the root system under nursery conditions, both in dry matter and in length, as compared to the other evaluated doses. Likewise, it favored a higher number of leaflets, aerial length and leaf dry matter, which are decisive variables in plant establishment under field conditions. Therefore, this dose is adequate for an integrated fertiliza- tion program because it will result in plants with adequate growth, which can later be transplanted to uncontrolled conditions under standard cultivation systems. Furthermore, high doses of K (above the doses proposed) can favor a greater leaf area, but are unfavorable for root system development and other allometric variables. There was a negative interaction between the nitrogen and phosphorus fertilization and the calcium fertilization. The high doses of Ca did not favor the application of elevated levels of N and P, and, under that condition, it was, in general, detrimental for plant growth. These conclusions apply only for the first three months of Andean blackberry plant growth. According to the analysis of the results during the growth period, the plants tended to be saturated quickly, so it would be difficult to extrapolate plant behavior beyond three months. Acknowledgments The authors wish to thank the Ministerio de Agricultura y Desarrollo Rural (MADR) of Colombia and the Corpora- ción colombiana de investigacion agropecuaria – AGRO- SAVIA for financing this study. Literature cited Al-Shemmar, G.N., H.B. Abdurrahman, and G.J. Zedan. 2013. Effect of some agricultural treatments on fruits storage quality of two tomato hybrids cultivated in gypsum soil. J. Genet. Environ. Resour. Conserv. 1(3), 233-246. Ascencio, J. and J.V. Lazo. 2001. Crecimiento y eficiencia de fósforo de algunas leguminosas cultivadas en arena regada con solu- ciones nutritivas con fosfatos inorgánicos de hierro y calcio. Rev. Fac. Agron. 18, 13-32. Bayuelo-Jiménez, J.S., V.A. Pérez-Decelis, M.L. Magdaleno-Armas, M. Gallardo-Valdéz, I. Ochoa, and J.P. Lynch. 2011. Genetic variation for root traits of maize (Zea mays L.) from Purhepe- cha Plateau, under contrasting phosphorus availability. Field Crops Res. 121, 350-362. Doi: 10.1016/j.fcr.2011.01.001 Bayuelo-Jiménez, J.S., I. Ochoa, V.A. Pérez-Decelis, M.L. Magdale- no-Armas, and R. Cárdenas-Navarro. 2012. Eficiencia a fósforo en germoplasma de maíz en la meseta P’urhépecha. Field Crops Res. 121, 350-362. TABLE 4. Regression coefficients and their associated error for regression model terms per plant growth variable. Terms Number of leaflets Leaf area Plant height Root length Root volume Estimation Error Estimation Error Estimation Error Estimation Error Estimation Error Intercept -9.4654 3.5528 -179.1964 29.5621 27.8844 3.3535 -1.4252 0.5423 N 0.3894 0.0752 1.9954 0.2313 -0.5362 0.1657 0.0174 0.0043 P 1.4561 0.4649 0.0178 0.0074 K 0.3789 0.1493 0.015 0.0057 Ca 2.2712 0.455 23.0843 3.8844 1.1567 0.0966 -1.7964 0.4433 0.2197 0.0702 N x Ca -0.0117 0.005 -0.0098 0.004 P x K -0.0004 0.0002 P x Ca -0.0071 0.003 -0.059 0.0307 -0.0106 0.0026 N2 0.008 0.0012 0.0124 0.0034 P2 0.0033 0.0009 0.0043 0.0007 0.0007 0.0004 Ca2 -0.0693 0.0146 -0.793 0.1287 -0.0297 0.0066 0.0526 0.015 -0.0088 0.0024 r2 0.6 0.53 0.96 0.41 0.51 All of the coefficients in the table are significantly different from zero (0) according to the t-test at 5% significance. Empty cells indicate terms that were not included in the model because they did not contribute significantly. r2: Determination coefficient. http://dx.doi.org/10.1016/j.fcr.2011.01.001 142 Agron. Colomb. 36(2) 2018 Benavides-Mendoza, A. 2011. Absorción de iones en la raíz. De- partamento de Horticultura, Universidad Autónoma Agraria Antonio Narro. Buenavista, Mexico. Bernal, J. and C. Díaz. 2006. Materiales locales y mejorados de tomate de árbol, mora y lulo sembrados por los agricultores y cultivares disponibles para su evaluación en Colombia. Cor- poica. C. I. La Selva. Bol. Div. 7, 10-13. Betancur-Cardona, E., E.L. García-Valencia, E. Barrera-Bello, O. Quejada-Rovira, H.D. Rodríguez-Mariaca, and I.C. Arroyave- Tobón. 2014. Manual técnico del cultivo de mora bajo buenas prácticas agrícolas. Gobernación de Antioquia - SENA. An- tioquia, Colombia. Bolaños-Benavides, M.M., W.A. Cardona, W.L. Ramírez, and J.H. Arguelles. 2014. Requerimientos nutricionales (N, P, K y Ca) de Rubus glaucus B., durante crecimiento vegetativo. Memorias XX Congreso Latinoamericano y XVI Congreso Peruano de la Ciencia del Suelo. Cuzco, Peru. Cánovas, F.M, U. Lüttge, and R. Matyssek. 2016. Progress in Botany 78. Springer. Doi: 10.1007/978-3-319-49490-6 Cardona, W.A., O.I. Monsalve-Camacho, J.S. Gutiérrez-Díaz, and M.M. Bolaños-Benavides. 2016. Efecto de N, P, K y Ca sobre crecimiento de mora con tunas en vivero. Memorias XVIII Congreso Colombiano de la Ciencia del Suelo. Villa de Leyva, Colombia. Castaño, C.A., C.S. Morales, and F.H. Obando. 2008. Evaluación de las deficiencias nutricionales en el cultivo de la mora (Rubus glaucus) en condiciones controladas para bosque montano bajo. Agron. 16, 75-88. Dhillon, W.S., P.P.S. Gill, and N.P. Singh. 2011. Effect of nitrogen, phosphorus and potassium fertilization on growth, yield and quality of pomegranate ‘Kandhari’. Acta Hortic. 890, 327-332. Doi: 10.17660/ActaHortic.2011.890.45 Frias-Moreno, N., A. Nuñez-Barrios, R. Perez-Leal, A.C. Gonzalez- Franco, A. Hernandez-Rodriguez, and L. Robles-Hernandez. 2014. Effect of nitrogen deficiency and toxicity in two varieties of tomatoes (Lycopersicum esculentum L.). Agric. Sci. 5, 1361- 1368. Doi: 10.4236/as.2014.514146 Kim, H.J. and X. Li. 2016. Effects of phosphorus on shoot and root growth, partitioning, and phosphorus utilization efficiency in Lantana. HortSci. 51(8), 1001-1009. Koning, L.A., M. Veste, D. Freese, and S. Lebzien. 2015. Effects of nitrogen and phosphate fertilization on leaf nutrient content, photosynthesis, and growth of the novel bioenergy crop Fal- lopia sachalinensis cv. ‘Igniscum Candy’. J. Appl. Bot. Food Qual. 88, 22-28. Doi: 10.5073/JABFQ.2015.088.005 Li, M., X. Shi, C. Guo, and S. Lin. 2016. Phosphorus deficiency inhibits cell division but not growth in the dinof lagellate Amphidinium carterae. Front. Microbiol. 7, 826. Doi: 10.3389/ fmicb.2016.00826 Luo, L., S. Pan, X. Liu, H. Wang, and G. Xu. 2017. Nitrogen de- ficiency inhibits cell division-determined elongation, but not initiation, of rice tiller buds. Israel J. Plant Sci. Doi: 10.1080/07929978.2016.1275367 Magdaleno-Villar J.J., A. Peña-Lomeli, R. Castro-Brindis, A.M. Castillo-González, A. Galvis-Spinola, F. Ramírez-Pérez, and P.A. Becerra-López. 2006. Efecto de tres sustratos y dos colores de plástico en el desarrollo de plántulas de tomate de cáscara (Physalis ixocarpa Brot.). Rev. Chapingo Serie Hortic. 12(2), 153-158. Doi: 10.5154/r.rchsh.2005.11.054 Marschner, P. 2011. Mineral nutrition of higher plants. Academic Press, London. Morales, C.S. and B. Villegas. 2012. Mora (Rubus glaucus B.). pp. 728-754. In: Fischer, G. (ed.). Manual para el cultivo de frutales en el trópico. Produmedios, Bogota. Parra-Terraza, S., E. Salas-Núñez, M. Villarreal-Romero, S. Hernán- dez-Verdugo, and P. Sánchez-Peña. 2010. Relaciones nitrato/ amonio/úrea y concentraciones de potasio en la producción de plántulas de tomate. Rev. Chapingo Serie Hortic. 16(1), 37-47. Preciado-Rangel, P., G.A. Baca-Castillo, J.L. Tirado-Torres, J. Ko- hashi-Shibata, L. Tijerina-Chávez, L., and A. Martínez-Garza. 2002. Nitrógeno y potasio en la producción de plántulas de melón. Terra 20, 267-276. Razaq, M., P. Zhang, H. Shen, and Salahuddin. 2017. Inf luence of Nitrogen and phosphorous on the growth and root morphol- ogy of Acer mono. PLoS One 12(2), e0171321. Doi: 10.1371/ journal.pone.0171321 Rivera-Espejel, E.A., M. Sandoval-Villa, M. Rodríguez-Mendoza, C. Trejo-López, and R. Gasga-Peña. 2014. Fertilización de tomate con nitrato y amonio en raíces separadas en hidro- ponía. Rev. Chapingo Serie Hortic. 20(1), 57-70. Doi: 10.5154/r. rchsh.2012.12.069 Riveras, E., J.M. Alvarez, E.A. Vidal, C. Oses, A. Vega, and R.A. Gutierrez. 2015. The calcium ion is a second messenger in the nitrate signaling pathway of Arabidopsis. Plant Physiol. 169: 1397-1404. De Sá, A.F.L., S.V. Valeri, M.C.P. da Cruz, J.C. Barbosa, G.M. Rezende, and M.P. Teixeira. 2014. Effects of potassium application and soil moisture on the growth of Corymbia citriodora plants. Cerne 20(4), 645-651. Doi: 10.1590/01047760201420041422 Santos, E.M., Í.H.L. Cavalcante, G.B. Silva Júnior, F.G. Albano, F.N. Lima, A.M. Sousa, and L.F. Cavalcante. 2014. Estado nutricional do mamoeiro Formosa (cv. Caliman 01) em função de adubação com NK e espaçamento de plantio. Comunicata Scientiae 5, 29-240. Vargas-Bolívar, M.I., L.A. Calderón-Medellín, and M.M. Pérez-Tru- jillo. 2009. Efecto de las deficiencias de algunos nutrientes en plantas de lulo (Solanum quitoense var. quitoense) en etapa de vivero. Rev. Fac. Cienc. Básicas 5(1), 64-81. White, A.C., A. Rogers, M. Rees, and C.P. Osborne. 2015. How can we make plants grow faster? A source-sink perspective on growth rate. J. Exp. Biol. 67(1), 31-45. Doi: 10.1093/jxb/erv447 Xiang, D.B., T.W. Yong, W.Y. Yang, Y. Wan, W.Z. Gong, L. Cui, and T. Lei. 2012. Effect of phosphorus and potassium nutrition on growth and yield of soybean in relay strip intercropping system. Sci. Res. Essays 7(3), 342-351. Yang, B.M., L.X. Yao, G.L. Li, Z.H. He, and C.M. Zhou. 2014. Dynamic changes of nutrition in litchi foliar and effects of potassium-nitrogen fertilization ratio. J. Soil Sci. Plant Nutr. 15(1), 98-110. Doi: 10.4067/S0718-95162015005000009 http://dx.doi.org/10.1007/978-3-319-49490-6 http://dx.doi.org/10.17660/ActaHortic.2011.890.45 http://dx.doi.org/10.4236/as.2014.514146 http://dx.doi.org/10.5073/JABFQ.2015.088.005 http://dx.doi.org/10.3389/fmicb.2016.00826 http://dx.doi.org/10.3389/fmicb.2016.00826 http://dx.doi.org/10.1080/07929978.2016.1275367 http://dx.doi.org/10.5154/r.rchsh.2005.11.054 http://dx.doi.org/10.1371/journal.pone.0171321 http://dx.doi.org/10.1371/journal.pone.0171321 http://dx.doi.org/10.5154/r.rchsh.2012.12.069 http://dx.doi.org/10.5154/r.rchsh.2012.12.069 http://dx.doi.org/10.1590/01047760201420041422 http://dx.doi.org/10.1093/jxb/erv447 http://dx.doi.org/10.4067/S0718-95162015005000009