This study was carried out at the Department of Plant Science Research farm in the University of Buea (Figure 1). It is located between latitudes 4˚28′30ʺ N and 3˚54′26ʺN and longitudes 8˚57′10ʺ and 9˚30′49ʺE [12], and at an elevation of about 550 m above sea level. The area has a mean annual temperature 28˚C and a humidity of 86% [12]. It has a humid tropical climate with an annual rainfall of 2800 mm [13][14]. The annual sunshine is about 900 - 1200 hours per annum, [14]. The upper elevations of the town tend to be cool and cloudy, while the lower elevations tend to be much warmer and less humid. Cameroon has five agro-ecological zones, with the South West Region located within the Humid Forest Zone with monomodal rainfall [15]. It is characterized by two seasons [16], a short dry season (about 3 months), and a wet season (about 9 months).

Figure 1. Location of the University of Buea, Cameroon.

A composite soil sample was collected from the experimental plots randomly using a soil auger at a depth of 0 - 10 cm, 10 - 20 cm and 20 - 30 cm, The soils were bulked according to each soil depth and homogenous samples were collected and labelled. The soil samples were air-dried and later sieved with a 2 mm sieve and stored in ziploc bags for routine chemical analysis. The samples were sent to the Plant and Soil laboratory of the University of Dschang for chemical analysis according to the method described by [17]. The soil samples were analyzed for nitrogen, phosphorus, potassium, magnesium, calcium, soil pH, organic carbon, exchange acidity and Cation Exchange Capacity (CEC).

A total area of 208 m² of land was cleared, raked and divided into 5 blocks with each block having 4 plots and each block separated from the other by a distance of 2 m. This gave a total of 20 plots with each plot measuring 3 × 3 m (9 m²) with 1 m spacing between plots. The experimental design was a complete randomized block design (CRBD), five treatments and four replicates. The experiment was conducted across three successive harvest periods, spanning from October 29, 2020, to May 2, 2021. The specific harvest periods were:

1)First Harvest: October 29, 2020, to December 29, 2020 (60 days).

2)Second Harvest: December 30, 2020, to March 2, 2021 (60 days).

3)Third Harvest: March 3, 2021, to May 2, 2021 (60 days).

The stated fertilizer rates (e.g., 150, 200, 300, 400 kg/ha) represent the total annual amount (Table 1). This total was split equally across three application cycles: at the beginning of the growing season and immediately following the first and second harvests. Therefore, the quantity applied per individual application was one-third of the stated rate (e.g., 50, 66.7, 100, and 133.3 kg/ha per application, respectively).

Nursery and transplanting of seedlings ofBrachiariaruziziensis

Seeds of Brachiariaruziziensis were obtained from the Institute of Agronomic Research for Development (IRAD) Bagante, Cameroon and the nursery medium were sand and topsoil in a ratio of 1:2. Seeds of Brachiariaruziziensis were manually sown by broadcasting.

Seedlings of similar heights (average height of 10 cm with leaf numbers of 4 - 5 were selected from the nursery and transplanted into the various plots. Transplanting was carried out three weeks after seed germination. The seedlings were transplanted at 50 cm between rows and 25 cm within rows giving a plant density of 40 plants per plot and 80,000 plants/ha (modified from [18][19].

Table 1. Treatment allocation.

Note: The plant population was maintained at a uniform density of 40 plants per plot (equivalent to 80 000 plants per hectare) across all treatments. The quantity of fertilizer per plant used was calculated accordingly.

To assess growth, five plants were selected at random from the middle of each plot and tagged. Plant heights were measured from base to the crown using a graduated measuring rod. The diameters of the plants were measured at the height 5 cm of the base of the stem using a digital vernier caliper. The number of tillers of the tagged plants was counted and recorded. The total number of tillers per tuft for each measurement period was defined as the sum of total number of tillers at the previous measurement and the number of tillers formed after the previous measurement. The numbers of leaves per tagged plant were enumerated, and these measurements were conducted weekly over a span of eight weeks.

The first harvesting was carried out 60 days after transplanting [18] and the plants were harvested at 10 cm above the ground level to facilitate regrowth. Fresh weights were determined using an electronic balance, followed by oven drying at 105˚C until a constant mass was achieved to obtain the dry weight. Three harvest cycles were conducted with a 60-day regrowth interval between each harvest period [12].

Fifty leaves were randomly sampled from each plot, giving a total of 200 leaves per treatment. This implied a total of 1000 leaves for the five treatments. The length and width of each leaf were measured in centimeters using a ruler to calculate the

relative leaf area with the formula

was used where b and L represent the width and length of the leaf respectively. The actual leaf areas of these samples were determined using a leaf area meter (Orsenigo 121TM35). Subsequently, the actual leaf area was graphed against the relative leaf area using Excel software, and a regression line was fitted to depict the relationship between actual and relative leaf areas (Equation 2).

Y = MX + C (2)

where M = gradient, X = Relative leaf area, C = constant.

The model generated from the regression analysis between actual leaf area and relative leaf area revealed that for every unit increase in the predictor variable (relative leaf area), the response variable (actual leaf area) increases by a factor of 1.403.

The regression equation is

This non-destructive modelling approach is crucial for efficient field assessment of forage growth parameters [20]-[22].

At each of the three harvest periods, 500g of leaves was harvested from each treatment oven dried at 65˚C for 48 hours, milled and sieved with a 2 mm sieve. The routine chemical analysis procedure was repeated for the milled plant materials for the nutrients (N, P, K, Mg, and Ca), ash content, crude protein and crude fibre content.

Data collected on agronomic parameters and dry matter of forage samples was subjected to Analysis of Variance (ANOVA) procedure based on the model designed for a complete randomized block design (CRBD). The means were separated using the Dunnett method. All the analysis was done using MINITAB version 17 Statistical Software (2010).

The initial chemical analyses of the experimental soil (Table 2) provided insightful context for interpreting the plant growth responses. The soil was characterized by low levels of key nutrients, particularly nitrogen (0.20%), potassium (0.28 Cmol/kg), and organic carbon (1.11%), while phosphorus was moderately available (19.57 mg/kg). The slightly acidic pH (pH-water 6.53) was generally favorable for nutrient availability. This baseline fertility profile, indicating a soil deficient in the primary nutrients supplied by the fertilizer, established the potential for a strong growth response to NPK 20:10:10 application.

Table 2. The chemical parameters of the soil prior to treatment application.

As anticipated from the soil analysis, significant enhancements in all growth parameters were observed in plants treated with fertilizers compared to the control group (Table 3, P = 0.05). Grouping information, indicated by superscript letters, was determined using the Dunnett Method at a 95% confidence level. Among the treatments, the application of 400 kg/ha of NPK resulted in the highest fresh weight (85.1 tons/ha), whereas the control group exhibited the lowest fresh weight (38.2 t/ha). Similarly, the treatment with 400 kg/ha also demonstrated the highest dry weight (34.7 tons/ha) and this was not significantly different from plants treated with 300 kg/ha NPK, while the control group recorded the lowest dry weight (15.3 t/ha) (Table 3).

Table 3. Effects of fertilizer levels on growth parameters of Brachiariaruziziensis.

The values presented in the table represent the means of three planting regimes, accompanied by their respective standard errors. Grouping information, indicated by the letter “a” was determined using the Dunnett method at a 95% confidence level. Within each column, means not sharing the letter “a” are significantly different from the control treatment.

The regression analysis results revealed that 82.3% of variation in the response variable (actual leaf area) is explained by variations in the predictor variable (relative leaf area) as indicated by the coefficient of determination (Adj. R²) of 0.823 in Table 4. The values for RMSE, MAE, AIC and BIC show how well the model fits the data, the lower these values the better the model.

The analysis of variance results further confirmed the existence of a significant (P < 0.01) relationship between the predictor (relative leaf area) and the response variable (actual leaf area) at the 5% level of significance. This implies that changes in actual leaf area significantly depend on changes in relative leaf area.

Table 4.Model evaluation metrics.

A scatter plot and regression line was used to graphically depict the relationship between actual leaf area and relative leaf area (Figure 2). Results revealed a positive slope indicating a positive relationship between actual leaf area and relative leaf area.

Figure 2.Scatter plot and regression line showing the relationship between the relative leaf area and the actual leaf area.

The results of the effects of fertilizer on nutrient concentrations are presented in Table 5. For all nutrients studied, plant supplied with fertilizer had a significant effect (P = 0.05) except ash content on plant nutrient concentrations.

For nitrogen concentration, plants treated with 300 kg/ha fertilizer recorded the highest percentage of nitrogen was 2.2% and was not significantly different from the amount recorded in the 400 kg/ha B (2.1%) while the control exhibited the lowest N concentration (0.7%). Similar trends were observed for phosphorus and potassium, with the 300 kg/ha fertilizer treatment showing the highest concentrations of phosphorus (2964 mg/kg) and potassium (2510 mg/kg), whereas the control had the lowest concentrations for all nutrients analysed except ash. In the case of calcium and magnesium, plants treated with 400 kg/ha of NPK recorded the highest concentrations—1280 mg/kg and 117 mg/kg, respectively—while the control had the lowest concentrations, at 77.6 mg/kg for calcium and 17.0 mg/kg for magnesium. Plants treated with 300 kg/ha NPK fertilizer had the highest crude protein (CP) (14.9%), crude fibre (13.7%) while plants in the control plots had 12.8% and 5.4 % respectively (Table 5).

Table 5. Nutrient concentration of Brachiariaruziziensis under different treatments.

Table values represent the means of three cutting regimes, presented alongside their corresponding standard errors. Superscript letters (A, B, C, etc.) across rows indicate statistical grouping; means sharing the same letter are not significantly different. Higher numerical values reflect more favorable outcomes except for ash, F- and P-values apply to comparisons across row.