Statistical study of the causes of acacia tree deterioration in the Ha’il region

Study area

This research was conducted in the Ha’il region of Saudi Arabia, an arid zone characterized by low rainfall, high temperatures, and limited water availability. The region is home to acacia trees that play a critical ecological and socio-economic role. The study area was selected due to the environmental challenges and the observed decline in acacia populations, making it a suitable location for investigating the factors contributing to their deterioration.

According to Alshammari, Busais & Ibrahim (2017), the Ha’il Region is situated in the northern central part of Saudi Arabia. The area of the region is 118,232 square kilometers. Ha’il is situated in the center of the region. It is the only significant city in the area. Located 600 km from Riyadh, 450 km from Madinah, and 650 km from Tabuk, it is easily accessible from other regional centers. Ha’il has a temperate climate due to its elevation of 915 m above sea level. The study was conducted across 31 locations in Ha’il Province (Fig. 1) based on the mapping provided by Alshammari, Busais & Ibrahim (2017).

According to Climate Data (2025), over the past 20 to 50 years, Ha’il—like much of the Arabian Peninsula—has experienced significant climate changes, particularly in rising temperatures. Although detailed long-term records for Ha’il are limited, broader regional data indicate a consistent warming trend consistent with global climate change. For example, the average national temperature in Saudi Arabia from 1961 to 1990 was approximately 24.7 °C, and temperatures have steadily increased in the decades since. Today, Ha’il records an average temperature of approximately 22.7 °C (72.8 °F).

Rainfall levels in Ha’il have remained low and stable over the years, in keeping with its desert climate. The region receives an average of only 81 mm (3.2 inches) of rain annually, a figure that has seen little change over the decades.

Summer, which runs from June to September, is extremely hot, with temperatures often exceeding 40 °C (104 °F). In contrast, winters are characterized by mild temperatures, with nighttime temperatures sometimes dropping below 10 °C (50 °F). For those planning a visit, April, May, and October offer more comfortable conditions and are considered the best times to explore the region.

Furthermore, we improved the study area’s environmental profile by providing a more thorough analysis of the soil’s physical and chemical properties, which are crucial for understanding Acacia species’ ecological dynamics. The soils in the Ha’il region are mostly sandy to loamy, with minimal organic matter, limited moisture retention, and high alkalinity (pH > 7.5). Alkaline environments, high calcium carbonate levels, and moderate cation exchange capacity (CEC) have a substantial impact on nutrient availability and root development in Acacia trees. Including these soil factors gives a more rigorous ecological context for examining acacia degradation patterns in the area. Including these soil parameters provides a more robust ecological context for analyzing acacia degradation patterns in the region (Farrag, 2012; Salman, 2015).

These climate trends highlight the growing importance of environmental awareness and adaptation strategies in arid regions like Ha’il, especially considering ongoing global climate change.

Data collection

Environmental factors

Data regarding rainfall, temperature, and soil conditions were collected through meteorological records and supplemented by information gathered via questionnaires distributed to residents, landowners, and environmental experts. Participants were asked to share their perceptions of how these environmental variables impact the health of acacia trees in the region.

Findings from Alanazi et al. (2022) provided additional context for interpreting environmental impacts. Their study identified acacia trees in dry environments, such as plateaus, that face heightened vulnerability to drought and pest infestations. Prolonged droughts weaken trees, reducing their ability to fend off pests like the xylophagous beetle Steraspis speciosa. This beetle’s larvae burrow into the bark, causing severe damage to the xylem and phloem, which disrupts water and nutrient transport and leads to tree mortality. The insights from this research supported questionnaire responses indicating that limited water availability exacerbates tree decline in the Ha’il region.

Anthropogenic factors

The role of human activities, including grazing, urbanization, and firewood collection, was examined through both satellite imagery and questionnaire responses. Respondents highlighted overgrazing as a significant issue, particularly in unprotected areas where young acacia trees are unable to regenerate. Additionally, urban expansion and firewood harvesting were reported as key contributors to habitat degradation.

The observations align with the findings of Alanazi et al. (2022), who noted that areas exposed to anthropogenic pressures exhibit higher rates of acacia decline. The combination of human activities and environmental stressors creates a compounding effect, further weakening tree health and increasing vulnerability to pest infestations.

Pest dynamics and their interaction with other factors

Responses from questionnaires indicated a growing awareness of pest activity as a factor in tree decline. Participants noted the presence of beetle infestations in certain areas, corroborating the findings of Alanazi et al. (2022), who documented that Steraspis speciosa preferentially targets older trees in drier habitats. The study also highlighted that pest infestations are more severe in regions with limited soil moisture, suggesting an interplay between environmental stress and pest dynamics.

By integrating questionnaire data with insights from the literature, this study provides a comprehensive understanding of the environmental and anthropogenic factors driving acacia tree decline in the Ha’il region.

Data analysis

Ethical considerations

The study followed ethical guidelines to ensure the integrity of the research process. After obtaining approval from the Research Ethics Committee, the questionnaire was distributed to the target group. Research Ethics Committee (RCE) at the University of Ha’il reviewed and approved this study on 03/02/2025, with research number H-2025-587. Verbal and written consent were obtained from the participants before data collection.

Limitations

The study faced some limitations, including reliance on self-reported data, which may be subject to personal biases or inaccuracies. Additionally, the absence of direct field measurements restricted the ability to validate participants’ perceptions. Future research could integrate opinion-based data with physical assessments for a more robust analysis.

The results of the general data analysis

Figure 2 presents the distribution of responses based on age categories. The highest engagement was recorded among individuals aged 45 and above (44%), followed by the 34–44 age group (32%). In contrast, respondents aged 23–33 accounted for 18%, while the youngest group, 18–22 years old, contributed only 4%. These results suggest that older participants were more inclined to complete the questionnaire, possibly due to greater interest or availability, whereas younger respondents may require more targeted outreach.

Figure 3A illustrates how survey participation varied across different educational backgrounds. Respondents with a university degree made up the largest proportion (47%), followed by those with a postgraduate qualification (28%). In comparison, diploma holders accounted for 12%, while individuals with only a high school qualification represented 11%. The higher response rates among university-educated participants suggest a potential correlation between academic background and engagement in the study.

Figure 3B compares survey participation based on respondents’ places of residence. Most responses came from Hail city (61%), while governorates in Hail contributed 20%. Participation from villages within Hail was considerably lower (3%), whereas 14% of respondents reported living in other locations. The data suggests that urban dwellers were more likely to engage with the survey, whereas those in rural areas may have encountered accessibility challenges or lower awareness of the study.

In Fig. 3C, gender-based participation in the survey shows a clear disparity, with male respondents making up 83% of the sample, while female participation was significantly lower at 17%. This imbalance could be attributed to differences in interest, accessibility, or social factors influencing engagement. To ensure a more representative gender distribution in future studies, additional efforts may be needed to encourage female participation.

Figure 3D presents the most frequently suggested strategies for protecting acacia trees. The most widely supported measure was enforcing strict regulations against illegal tree cutting (24%), followed by both organized grazing (17%) and increasing tree-planting efforts (17%). Additionally, expanding protected areas (11%) and enhancing irrigation methods (9%) were also proposed. These findings reflect a strong awareness of environmental conservation and highlight key areas for policy development and ecological management.

The results of the regression analysis

Table 2 summarizes the descriptive statistics for the variables included in the study. The dependent variable, VAR18, has a mean of 53.3205 with a standard deviation of 23.25940, and the coefficient of variation (CV) is 43.62187151% across 234 observations. Independent variables, such as VAR1 through VAR17, show varying means and standard deviations, reflecting their unique distributions and roles within the model.

Table 3 presents the Pearson correlation coefficients among the variables. VAR18 demonstrates moderate positive correlations with several independent variables, including VAR1 (r = 0.287, p < 0.00) and VAR4 (r = 0.325, p < 0.00), indicating potential predictive relationships.

Table 4 provides a regression model summary, regression coefficient, and Pearson correlation. The adjusted R² value is 0.175, indicating that approximately 17.5% of the variance in the dependent variable is explained by the independent variables. The model’s overall F-statistic (F = 2.695, p < 0.000) demonstrates statistical significance. To evaluate potential multicollinearity among predictor variables, a correlation matrix was created that included all independent and dependent variables. The regression model predicts a dependent variable based on 17 independent variables (VAR1 to VAR17) with a constant term.

$$PredictedValue=9.54+3.72\mathrm{V}\mathrm{A}\mathrm{R}1+\dots +2.20\mathrm{V}\mathrm{A}\mathrm{R}17+\mathrm{ϵ}$$

Variables with large positive coefficients suggest a strong positive relationship with the dependent variable:

• VAR1: 3.72

• VAR4: 3.36

• VAR9: 2.32

• VAR17: 2.20

• VAR5: 1.70

• VAR13: 0.87

• VAR6, VAR7, VAR16, VAR15: Moderate positive values (around 0.6 to 0.8)

Variables with negative coefficients suggest an inverse relationship:

• VAR8: −2.80(strong negative influence)

• VAR14: −2.22

• VAR10: −0.72

• VAR3, VAR12, VAR11: Slightly negative impact.

Multicollinearity diagnostics based on Pearson correlation coefficients revealed notable intercorrelations among several predictors. Specifically, strong positive correlations were observed among VAR11, VAR12, VAR13, and VAR14 (with VAR13 and VAR14 reaching r = 0.777), as well as among VAR15, VAR16, and VAR17. These correlations may indicate redundancy among predictors, potentially impacting the model’s stability and interpretability.

Table 5 outlines the analysis of variance, confirming the model’s statistical significance (p < 0.000). The regression sums of squares (22,061,157 indicates the variance explained by the model, while the residual sum of squares (103,992.804) reflects unexplained variance.

Table 6 lists the unstandardized and standardized coefficients for each predictor. Unstandardized coefficients (B) represent the expected change in the dependent variable due to a one-unit increase in the predictor variable, provided all other variables remain constant. These values are given in the original measurement units and provide useful information about real-world relationships within the data. Standardized coefficients, also known as Beta weights, quantify the effect of an independent variable on the dependent variable in standard deviations. A one-standard-deviation rise in the predictor results in a Beta-sized change in the outcome variable. These coefficients enable direct comparison of the relative strength of predictors within a single model (Field, 2018). Two predictors emerged as statistically significant at the 0.05 level:

• VAR1 (Drought Severity): B = 3.721, SE = 1.624, β = 0.245, 95% CI [0.524–6.918], p = 0.023

• VAR4 (Urban Expansion): B = 3.365, SE = 1.655, β = 0.217, 95% CI [0.071–6.60], p = 0.045

Other variables, such as VAR8 (Soil Erosion) and VAR14 (Air Pollution), showed negative coefficients but were not statistically significant at the conventional level (p = 0.097 and p = 0.292, respectively), though their effect sizes suggest potential relevance.

Figures 4 and 5 display diagnostic plots to assess model assumptions:

• Figure 4: A histogram of standardized residuals confirms normality (Mean = 0, Std. Dev. = 0.963).

• Figure 5: The normal P-P plot shows points closely aligned with the diagonal line, supporting the assumption of normally distributed residuals.

Figure 6 illustrates the relationship between standardized predicted and actual values of the dependent variable (VAR18). The scattered distribution indicates no apparent pattern, supporting the assumption of homoscedasticity.

Table 7 provides results for two normality tests applied to the unstandardized residuals of a dataset with 234 observations. Residual diagnostics supported key model assumptions. Both the Kolmogorov-Smirnov (p = 0.200) and Shapiro-Wilk tests (p = 0.378) indicated no significant deviation from normality. Histogram and P–P plots confirmed that the distribution of residuals was approximately normal. Additionally, the residuals were homoscedastic, as shown by the evenly scattered standardized residuals around zero, without systematic patterns.

The results of predictive modeling

This study applies the CHAID method to develop a classification tree model. The analysis examines the relationship between a dependent variable (VAR18) and a set of independent variables (VAR1–VAR17). The CHAID algorithm partitions the dataset into statistically significant subgroups based on chi-square tests, allowing for the identification of patterns in the data.

The classification tree was constructed using the following parameters:

Parameter

Specification

Dependent variable

VAR18

Independent variables

VAR1–VAR17

Validation method

None (entire dataset used)

Maximum tree depth

3

Minimum cases per parent node:

100

Minimum cases per child node:

50

CHAID algorithm parameters

• Splitting criterion: Chi-square test

• Significance level for splitting: 0.05 (adjusted using Bonferroni correction)

• Merging criterion: 0.05 significance level

• Handling of missing data: Nominal missing values were treated as missing cases

The CHAID decision tree consists of five nodes, including three terminal nodes, with two primary independent variables (VAR13 and VAR1) identified as the most significant predictors.

In Fig. 7, the root node (Node 0) represents the entire dataset with an average value of 53.321. The first significant split is based on VAR13 (p < 0.001, F = 22.327), creating two subgroups:

• Node 1 (mean = 47.062)

• Node 2 (mean = 60.877)

A second split occurs in Node 2, based on VAR00001 (p = 0.002, F = 15.784), forming two additional terminal nodes:

• Node 3 (mean = 70.120)

• Node 4 (mean = 52.625)

The model’s overall risk estimate is 456.846, with a standard error of 42.388, indicating the extent of prediction error. The gain summary (Table 8) provides insights into the effectiveness of each terminal node. Node 3 exhibits the highest mean outcome (70.120), suggesting that individuals in this category have the most distinct response to the independent variables.

The CHAID classification tree effectively identifies VAR13 and VAR1 as the primary variables influencing VAR18. The model segments the data into distinct groups with varying mean values, offering valuable insights into the dataset’s structure. Future studies may improve the model’s accuracy through cross-validation techniques and the inclusion of additional predictor variables.

Conclusion

This research demonstrates that the deterioration of Acacia trees in the Ha’il region is driven by a combination of environmental stressors and human-related activities. Key ecological pressures include prolonged drought, increasing temperatures, and soil degradation, while human influences such as unregulated urban development and excessive grazing intensify the decline. Regression results showed that drought severity and urban expansion significantly impact tree health, explaining 17.5% of the variation. Furthermore, the CHAID analysis pinpointed urban growth and overgrazing as the most influential factors, highlighting specific areas that are most at risk.

A notable strength of this study lies in its innovative methodology, merging community-based surveys, satellite and meteorological data, and advanced statistical modeling (e.g., regression and CHAID decision trees). This integrated approach offers a richer and more accurate picture of the challenges facing Acacia populations.

Given the scale and severity of the issue, immediate and well-informed conservation actions are essential. Recommendations include enforcing stricter land use regulations, enhancing irrigation practices, expanding reforestation and protected zones, and raising public awareness about the ecological importance of Acacia trees. The study provides a scientifically grounded roadmap for stakeholders to protect and sustainably manage these vital desert ecosystems.

Recommendations

• 1.

  Strengthening conservation policies: Implement and enforce regulations to control overgrazing and prevent illegal logging.

• 2.

  Expanding reforestation efforts: Establish protected areas and promote large-scale acacia replanting initiatives.

• 3.

  Promoting sustainable land management: Encourage environmentally responsible agricultural and urban planning practices to minimize ecological degradation.

• 4.

  Enhancing water resource management: Develop and implement improved irrigation techniques to support tree health, particularly in vulnerable areas.

• 5.

  Raising public awareness: Conduct educational campaigns to inform local communities about the ecological and economic importance of acacia trees and promote sustainable land-use practices.

• 6.

  Advancing research and monitoring: Integrate remote sensing technology with ecological field assessments to track tree health, assess conservation effectiveness, and refine predictive models.

By implementing these measures, policymakers, conservationists, and local stakeholders can work collaboratively to mitigate the decline of acacia trees and ensure their long-term sustainability in arid and semi-arid ecosystems.