The study plots are situated in the Jujube Preservation Warehouse in Zhangqiu District, Shandong Province, China (36°25′-37°09′ N, 117°10′-117°35′ E). The research regions experience a moderate monsoon climate. The yearly average temperature is 12.8°C, with a maximum monthly temperature in July of 27.2°C and a minimum in January of-3.2°C. There is an average of 600.8 mm of precipitation, 2647.6 h of sunshine, and 192 days without frost throughout the entire year. The growing seasons of C. bungei forests last 130–150 days from May to September.

In March 2017, we established a pure forest with a row spacing of 3 × 4 m using a 2-year-old C. bungei clone (“9–1”). The total planting area was 0.8 ha with a total of 18 plots (384 m² for each plot), and 45 trees (5 rows × 9 columns) were planted in each plot. Since the planting of the trees, no tending had been done. After planting, C. bungei trees had an average height of 4.2 m and diameter at breast height (DBH) of 4.0 cm. Immediately following afforestation, the chemical composition of the soil was assessed, including its pH (7.67), organic matter content (19.64 g/kg), total nitrogen (0.91 g/kg), total phosphorus (0.53 g/kg), total potassium (16.70 g/kg), alkaline hydrolyzed nitrogen (81.88 mg/kg), available phosphorus (32.10 mg/kg), and available potassium (176.82 mg/kg). The majority of the soil in the area is brown loam.

Beginning in May 2018, we selected 9 plots at random and fertilized them as part of a split-plot experiment. Existing fertilization techniques included hole fertilization (HF) and water and fertilizer integration (WF), with no fertilization acting as a control. The HF approach was applied based on the previously fertilized forests of C. bungei (unpublished). The WF method was alluded to as being derived from the fertilization of fruits and crops in the agricultural ecosystem. Three plots were randomly selected for each fertilization treatment, and the distance between plots was around 6–8 m.

Early in May 2018, fertilization studies were carried out, with WF and HF applying the same quantity of fertilizers each year. In the first year, N (24 g/tree), P₂O₅ (8 g/tree), and K₂O (16 g/tree) were applied as fertilizers. Subsequently, the annual amount of fertilizers increased by 20% compared to the previous year. Every year in May, HF applied all of the fertilizers at once. We excavated a hole, with a size of 20 cm in diameter and 30 cm in depth in the south and north of each tree. Then the fertilizers were equally divided into 2 parts and placed into the holes. No water was added after covering the soil. WF divided its annual fertilizers into 12 portions that were distributed equally. Starting on May 1st, we fertilized once every 10 days. The fertilizers were correctly delivered close to the roots after being dissolved in 1000 L of water using an intelligent drip irrigation system (HN-BXE, Huinong Automation Corporation, China). Potassium sulfate was employed as the potassium fertilizer and urea as the nitrogen fertilizer for HF and WF. Calcium superphosphate was utilized as a phosphate fertilizer for HF, whereas ammonium dihydrogen phosphate was used for WF. All Fertilizers were purchased on the Taobao website. The temperature and precipitation data from the nearby meteorological station for the years 2017 through 2022 were displayed in Supplementary Figure S1.

The standard tree represents the average level of trees in each sample plot, including their average height and DBH. Two standard trees were felled in each plot, and a total of 18 trees and 516 branches were collected in November 2021. In short, 6 trees were felled in each of the three treatments. The DBH and crown width (CW) were measured before being cut down. The tree height (TH) and the height of the crown base (HCB) were determined after felling. All branches (except dead branches) were dissected from the crown base to the top. The branch height (BH) (2.30–9.13 m), diameter (BD) (6.21–94.05 mm), angle (BA) (6.3–92.8°), length (BL) (0.45–5.10 m), orientation (BO) (0–355°), number of branches (NB) (20–39 branches), and density of the branches (DB) (3–6 branches·m⁻¹) were measured.

The DBH was the diameter of the trunk at 1.3 m. The CW was the mean of the east–west and north–south projections of the canopy. The TH was the vertical distance from the bottom to the top of the tree. The HCB was the vertical distance from the bottom of the trunk to the crown base. The BH was the vertical distance from the base of the branch to the bottom of the trunk. The BD was the diameter at the base of the branch. The BA was the insertion angle between the branch and trunk pith. The BL was the distance between the base and the tip of the branch. The DB was the ratio of branch number to crown length. In addition, the trees were divided into the individual organ, including stems, branches, leaves, and roots. All of the roots were dug up with an excavator, and any soil that was left on top of the roots was then removed with a brush. The fresh weight of each organ was determined. Then, a small portion of each organ was collected, taken back to the laboratory at low temperature, and dried for water content determination.

The biomass was the difference between the fresh weight and water content of the organ. The total biomass of the tree was the sum of trunks, branches, leaves, and roots. The differences in branch attributes or organ biomass in different fertilization regimes were compared using a one-way analysis of variance (ANOVA) and Tukey’s honestly significant difference test (Tukey’s HSD). Since the data for the branch attributes of the investigated trees had a hierarchical structure, linear mixed-effects models (Equations 1 and 2) were used to assess the significance of variations in treatment for branch attributes at the tree and branch level.

Where y was observed value; μ was the overall mean; ϑ_F was the effect of fertilization regimes; φ_p, φ_pt, and φ_ptbr were the random effects for plot (p), tree (t), and branch (br), respectively.

The restricted maximum likelihood estimation (REML) was employed in the mixed-model analysis, and Tukey’s HSD was utilized to conduct multiple range tests between treatments. Generalized linear mixed models (GLMM) were utilized for count variables (e.g., branch number) in the model-building process, and linear mixed models (LMM) were used for continuous variables (e.g., BD and BL). The following approaches were used to model the branch attributes. The multiple linear regression models were initially run for dependent variables with all pertinent independent variables to select potential factors. The prospective factors with statistical significance, a reasonable ecological explanation, and a variance inflation factor (VIF) that was not greater than three were then used to construct the mixed effects models. All of the above candidate and random components (such as plot and tree effects) were then evaluated using likelihood ratio tests (LRT) when a mixed effects model was constructed using the REML. Factors at a significance level of 0.05 were incorporated into models. Generalized linear models (GLM) or linear models (LM) were replaced in the absence of any discernible random factors. Lastly, the models were updated using the previously selected fixed and random components, and the best-fit models were picked from a range of variance functions primarily based on model simplicity and the Akaike Information Criterion (AIC).

Model performance was evaluated using the root mean squared error (RMSE), which was determined for each model on the original data scale. In the case of GLM, the predictive power R² was calculated as the squared correlation between observations and predictions. The marginal R² (R_m², only the fixed effects) and conditional R² (R_c², including the fixed and random effects) were measured for LMM. The associations between the dependent and all independent variables were expressed using simulation plots built on the fitting models. All statistical analyses and model building were performed using SPSS 25.0 for Windows (IBM-SPSS Inc., Chicago, IL, United States) (Table 1).

The fertilizing regimes had minimal impact on the HCB and CD but markedly increased the TH, DBH, and CR (Table 2). The DBH grew by 7.8 and 19.5% and the TH increased by 1.9 and 8.2% for HF and WF in comparison to CK, respectively. HF and WF produced greater stems, branches, leaves, roots, and total biomass than CK (Table 3). WF dramatically enhanced the total biomass of each organ compared to other fertilizing methods. WF increased the stem biomass by 45.9%, branch biomass by 59.8%, leaf biomass by 63.8%, root biomass by 37.7%, and total biomass by 48.3% compared to CK, respectively. The rank of organ biomass in three fertilization regimes was: stem > root > branch > leaf. The fertilization techniques did not, however, have a significant impact on the allocation proportion of organ biomass.

Amongst the three fertilization regimes, there was a noticeable difference in the total number of branches within the crown (Table 4). Most branches were in WF, and the fewest were in CK. Branch density did not considerably vary in response to fertilization regimes. The amount and percentage of branches at the four orientations for each fertilization treatment did not differ noticeably (Table 5). Thus, the four orientations of the branches within the crown were distributed roughly equally.

Fertilization regimes had no discernible effect on the diameter, length, and angle of the C. bungei branches (Table 4). WF dramatically increased the biggest branch diameter compared to CK and HF. The height of the largest branch in WF was dramatically lowered than CK and HF.

Our results showed that fertilization regimes (i.e., WF and HF) significantly increased the tree height, DBH, and organ biomass in the C. bungei forests compared to no fertilization. The fertility of the soil and availability of nutrients needed for plant growth are improved through soil fertilization (Tumushime et al., 2019; Yang and Karban, 2019; Li et al., 2021). In this study, the WF forests had greater tree growth and organ biomass than the HF forests did. Tree absorption and utilization for liquid fertilizer were superior to solid fertilizer, producing more biomass in the WF forests (Ma et al., 2019; Guan et al., 2022). The forest serves as a massive biomass carbon reservoir while also conserving water. Thus, WF forests may be recommended in terms of exhibiting greater water conductivity and carbon storage capabilities. Moreover, the DBH-TB model stated that it was possible to compute the total biomass of trees fairly quickly and efficiently, saving both time and labor, by precisely determining the DBH value.

The development of crown branches is influenced by a variety of factors, including site characteristics, stand density, management techniques, and tree growth potential. However, due to the limitations – a young stand age and a small number of sample trees – the models given are not suitable for widespread application. The impact of other aspects must also be taken into account while creating a large-scale model. A future study based on more extensive data from various geographic regions is required to create models that might be broadly utilized.

Tree branches develop inconsistently at every age stage. The competition between branches for water, light, and nutrients intensifies as trees get older, which leads to natural pruning and inevitable disruption of the branch indexes. Furthermore, the size of knots as well as the diameter, angle, and length of branches are significantly influenced by stand density. In the future, research on the development of large-diameter timber without knots in C. bungei plantations could focus on the importance of reasonable afforestation density and appropriate tending measures.