Climate Change Impacts on Agricultural Infrastructure and Resources: Insights from Communal Land Farming Systems

Abstract

Climate change significantly impacts agricultural infrastructure, particularly in communal land farming systems, where socio-economic vulnerabilities intersect with environmental stressors. This study examined the effects of extreme weather events (floods, droughts, strong winds, frost, and hail) on various agricultural infrastructures—including bridges, arable land, soil erosion control structures, dipping tanks, roads, and fences—using an ordered probit model. A survey was conducted using structured questionnaires between August and September 2023, collecting data from communal farmers (n = 60) in oKhahlamba Municipality, Bergville. Key results from respondents showed that roads (87%), bridges (85%), and both arable land and erosion structures were reported as highly affected by extreme weather events, especially flooding and frost. Gender, the type of farmer, access to climate information, and exposure to extreme weather significantly influenced perceived impact severity. The ordered probit regression model results reveal that drought (p = 0.05), floods (p = 0.1), strong winds (p = 0.05), and frost (p = 0.1) significantly influence the perceived impacts on infrastructure. Extreme weather events, including flooding (p = 0.012) and frost (p = 0.018), are critical drivers of infrastructure damage, particularly for smallholder farmers. Cumulative impacts—such as repeated infrastructure failure, access disruptions, and increased repair burdens—compound over time, further weakening resilience. The results underscore the urgent need for investments in flood-resilient roads and bridges, improved erosion control systems, and livestock water infrastructure. Support should also include gender-sensitive adaptation strategies, education on climate risk, and dedicated financial mechanisms for smallholder farmers. These findings contribute to global policy discourses on climate adaptation, aligning with SDGs 2 (Zero Hunger), 9 (Industry, Innovation, and Infrastructure), and 13 (Climate Action), and offer actionable insights for building infrastructure resilience in vulnerable rural contexts.

1. Introduction

Climate change has emerged as one of the most pressing global challenges, significantly affecting agricultural systems worldwide [1,2,3]. While much of the research has focused on the direct impacts of climate change on crop productivity and food security [4], water availability [5], and pest outbreaks [6]. However, less attention has been given to the effects of climate change on agricultural infrastructure, especially within communal farming systems regions like in KwaZulu-Natal, South Africa. Infrastructure components such as roads, bridges, soil erosion control structures, storage facilities, and fences are critical for ensuring the seamless functioning of agricultural operations [7]. These structures support the timely delivery of inputs, access to markets, and overall productivity [7]. However, the increasing frequency and intensity of extreme weather events, such as floods, droughts, and heavy rains, are undermining the functionality of these infrastructures, particularly in resource-constrained communal farming systems [8].

Recent studies have demonstrated the quantifiable effects of climate-induced events on rural infrastructure. For instance, Ref. [9] reported that over 60% of rural roads in Kenya were severely damaged by recurrent flooding, disrupting agricultural transport networks. Similarly, Ref. [10] highlighted a 45% increase in road maintenance costs due to climate-related degradation in rural Brazil. In South Africa, Ref. [11] documented infrastructure losses exceeding USD 4 million following a single flash flood incident affecting farms in KwaZulu-Natal province. Furthermore, the 2022 South African National Food and Nutrition Security Survey (NFNSS) undertaken by the Department of Agriculture, Land Reform, and Rural Development seeks to redress gaps in agricultural productivity, particularly in rural districts where issues of land access and infrastructure are impediments to market access and sustainable food security. Hence, the work by the task team is underway in the development of the new food and nutrition security plan 2025–2029 that seeks to transform the local food system, commissioned by the UN FAO Food Systems. This research case study contributes to a futuristic outlook of the current existing infrastructural and resource gaps and the reality of climate change in supporting food security initiatives for South Africa and within the province of KwaZulu-Natal. These empirical studies provide critical benchmarks for understanding the financial and operational vulnerabilities of rural infrastructure under climate stress.

Communal farming systems are particularly vulnerable to climate-induced disruptions due to their reliance on shared infrastructure and limited capacity to implement mitigation and adaptation strategies [12]. Infrastructure degradation caused by climate change can result in cascading challenges, including delayed land preparation, planting, and harvesting; spoilage of crops; and significant income and livelihood losses [13]. For example, heavy rains often render rural roads impassable, damage bridges, resulting in the isolation of farmers from markets and critical agricultural services [14]. Similarly, soil erosion caused by extreme weather events compromises the integrity of erosion control structures, reducing land quality and exacerbating environmental degradation [15].

This study is novel in that it explicitly investigates how multiple types of infrastructure are simultaneously affected by different extreme weather events within a communal farming context, which is an area underexplored in the current literature. Existing research often isolates infrastructure types or neglects how interconnected climate vulnerabilities impact both infrastructure and farming systems holistically. This study bridges that gap by integrating multivariate analysis of both socio-demographic and environmental variables across six infrastructure categories.

Furthermore, the study tests the following hypotheses: (i) Extreme weather events (floods, drought, strong winds, frost, and hail) have a significant effect on perceived infrastructure degradation among communal farmers. (ii) Socio-demographic factors (such as gender, type of farmer, education, and access to climate information) significantly influence perceptions of climate-induced infrastructure damage.

The study focuses on the oKhahlamba Local Municipality in the KwaZulu-Natal province, South Africa, a region characterized by mountainous terrain, high rainfall variability, and a population heavily reliant on smallholder and subsistence farming.

The primary objective of this study is to evaluate the extent to which climate change affects agricultural infrastructure in communal land farming systems and to explore the downstream effects on farming practices and livelihoods. Specifically, it investigates how different types of infrastructure are affected by distinct climate stressors and what socio-demographic characteristics influence the perceptions of these effects. The study further provides evidence-based insights for policymakers on practical interventions such as the prioritization of high-risk infrastructure (e.g., rural roads and erosion structures), implementation of early warning systems, and community-based maintenance programs. By doing so, this research contributes to the global discourse on agricultural resilience and climate adaptation, aligning with United Nations Sustainable Development Goals (SDGs)particularly SDG 2 (Zero Hunger), SDG 9 (Industry, Innovation, and Infrastructure), and SDG 13 (Climate Action). The study offers transferable insights that can be applied in similarly vulnerable rural regions worldwide, especially those with communal land governance systems.

2. Materials and Methods

2.1. Study Area

The study was conducted in oKhahlamba Local Municipality, situated within the uThukela District of KwaZulu-Natal Province, South Africa. One local municipality was selected: oKhahlamba Local Municipality (Figure 1), which, located in the western region, encompasses a vast and diverse area that includes the town of Bergville and the Drakensberg (oKhahlamba) Mountains [16]. Geographically, it lies at 28°43′52″ S 29°20′30″ E, with an area spanning approximately 3466 km². The municipality is part of the oKhahlamba–Drakensberg World Heritage Site, renowned for its breathtaking landscapes and ecological significance. This region is characterized by a temperate climate with an average annual rainfall of approximately 800–1200 mm, predominantly during the summer months [16]. The Drakensberg (oKhahlamba) foothills are strategic areas for food security in the province [17].

Agriculture plays a pivotal role in the livelihoods of oKhahlamba’s rural population, who rely heavily on communal farming systems [18]. The main agricultural activities include the cultivation of maize, dry beans, and various vegetables, alongside livestock farming [18]. The municipality’s infrastructure, such as rural roads, soil erosion control structures, bridges, and irrigation systems, underpins these agricultural practices, facilitating market access and supporting subsistence and smallholder farmers. With a population of approximately 137,724 people, predominantly living in rural areas, the region represents a hub of traditional, subsistence, and commercial agricultural practices [18].

In recent years, climate variability has emerged as a critical challenge for the municipality, manifesting through extreme weather events such as heavy rainfall, droughts, and unseasonal frosts. These events have exacerbated soil erosion, disrupted farming activities, and severely damaged essential agricultural infrastructure [16,18]. Furthermore, the mountainous terrain and fragile soils of the Drakensberg exacerbate the region’s vulnerability to climate-induced environmental changes.

The local municipality is an important contributor to the region’s ecosystem services, such as water provision, biodiversity conservation, and carbon sequestration. However, the municipalities face challenges related to soil erosion, deforestation, and increasing infrastructure vulnerability due to climate change [16,18]. These challenges are compounded by limited financial and technical resources available to communal farmers, making the municipality a critical case study for understanding the impact of climate change on agricultural infrastructure in resource constrained contexts. The municipality’s reliance on agriculture for economic and subsistence purposes highlights the importance of this study, which aims to examine how climate change affects agricultural infrastructure and the subsequent implications for farming activities and livelihoods. By focusing on this region, the study provides insights into the vulnerabilities and resilience strategies of communal farming systems in the face of climate change.

Figure 1. The location of oKhahlamba Local Municipality (Source: [19]).

Figure 1. The location of oKhahlamba Local Municipality (Source: [19]).

The Drakensberg Climate Potential (Figure 2) provides further spatial insight into the environmental heterogeneity of the oKhahlamba Local Municipality. This climate potential reveals that the region encompasses areas of variable climate suitability for agriculture, ranging from zones with very good to poor climate potential. These differences are shaped by altitude, rainfall patterns, and ecological sensitivity, especially within the oKhahlamba Drakensberg Park and surrounding foothills. The presence of high-value biodiversity areas, major water bodies, and limited community access roads highlights both the ecological significance and infrastructural challenges faced by communal farmers. This variability in climate potential has direct implications for agricultural productivity and resilience, particularly as climate change intensifies. This further underscores the critical need for climate-informed agricultural planning in rural systems, where infrastructure such as roads, bridges, and arable and grazing lands are increasingly vulnerable. As such, this spatial context enriches the study’s analysis by linking climate variability to the functional state of rural infrastructure, which is foundational to sustaining smallholder agriculture in this mountainous region.

2.2. Questionnaire Design

The study employed a structured questionnaire to collect data on the perceived impacts of climate change on agricultural infrastructure and associated farming practices. The instrument was developed based on a review of climate change vulnerability literature and expert consultation. Questions were designed to assess socio-demographic characteristics, climate exposure, access to information, and the extent of infrastructure impacts from extreme weather events. Responses for infrastructure impacts were captured using a three-point ordinal scale (1 = less affected, 2 = moderately affected, 3 = highly affected).

The study fully represented the sample size recognized to contribute to rural food security initiatives using smallholder farming enterprises. There were 70 registered farmers within the capacity development program for over 10 years. The study included 60 farmers who participated, considering that the other 10 were excluded as they formed part of the pilot survey.

To ensure reliability and validity, the questionnaire was pre-tested with a sample of 10 respondents drawn from the study area. Pre-test results were used to refine ambiguous questions, improve phrasing, and confirm the cultural and contextual relevance of the questions. The survey included built-in ethical protocols aligned with the Declaration of Helsinki, including informed consent, confidentiality assurances, and voluntary participation.

2.3. Data Collection and Analysis

2.3.1. Data Collection

Ethical clearance was granted by the Mangosuthu University of Technology Ethics Committee (NSci/03/2021). The survey was conducted between August and September 2023 using face-to-face interviews conducted by trained field researchers fluent in both English and isiZulu languages, ensuring linguistic inclusivity. The study employed purposive sampling to select the 60 active communal farmers who met the following criteria:

I.

Active involvement in farming within communal land systems.

II.

Dependence on infrastructure such as roads, bridges, fences, and soil erosion structures.

III.

Direct experiences with climate change impacts on farming activities.

The purposive sampling approach was chosen to ensure that respondents possessed firsthand experience of climate impacts on agricultural infrastructure. This method was deemed more appropriate than random sampling given the study’s focus on perceptions and experiences specific to vulnerable farming communities. However, we acknowledge its limitations, including potential selection bias and lack of statistical generalizability.

2.3.2. Analytical Framework and Empirical Model Specification

This study employed separate ordered probit regression models using Stata Version 18.5 to analyze the factors influencing farmers’ perceptions of climate change impacts on rural agricultural infrastructure. The ordered probit model is well suited for this analysis because the dependent variables are measured on a three-level ordinal scale, capturing increasing levels of perceived damage from climate change: (1) Less affected, (2) Moderately affected, and (3) Highly affected.

The six dependent variables used in the analysis represent different components of agricultural infrastructure, specifically: (i) Roads, (ii) Bridges, (ii) Dipping tanks, (iv) Fences, (v) Arable land, and (vi) Soil erosion control structures. Each of these variables reflects the farmer’s perceived level of impact of climate change on that specific infrastructure type. Given the ordinal nature of each outcome, the ordered probit model was chosen over binary models to capture more nuanced gradations in perception. Moreover, it was preferred over the multinomial logit model because the response categories have a natural ranking, even though the distances between them are not known.

In this model framework, there exists an unobserved (latent) continuous variable ${\mathbf{Y}}_{\mathit{i}}^{*}$ for each infrastructure type that reflects the severity of perceived climate impact. This latent variable is modeled as a function of observable characteristics ${\mathit{X}}_{\mathit{i}}$ and an error term ${\mathsf{\epsilon }}_{\mathit{i}}$, such that (Equation (1)):

$${\mathbf{Y}}_{\mathit{i}}^{*}={\mathit{\beta }}^{\prime }{\mathit{X}}_{\mathit{i}}+{\mathsf{\epsilon }}_{\mathit{i}}$$

(1)

where:

• ${\mathbf{Y}}_{\mathit{i}}^{*}$ is the latent and continuous measure of climate-induced damage to infrastructure;
• ${\mathit{X}}_{\mathit{i}}$ is a vector of explanatory variables (e.g., gender, age, education, type of farmer, climate exposure);
• $\mathit{\beta }$ is a vector of parameters to be estimated;
• ${\mathsf{\epsilon }}_{\mathit{i}}$ is the normally distributed error term.

However, we do not observe ${\mathbf{Y}}_{\mathit{i}}^{*}$ directly. Instead, the observed ordinal outcome ${\mathbf{Y}}_{\mathit{i}}^{*}$ is defined as below on Equation (2):

$${\mathit{Y}}_{\mathit{i}\mathit{j}}=\left\{\begin{array}{ccc}\mathbf{1}\hfill & \hfill \mathit{i}\mathit{f} {\mathit{Y}}_{\mathit{i}\mathit{j}}^{*}\le {\mathit{\mu }}_{\mathbf{1}}\hfill & \hfill (\mathit{L}\mathit{e}\mathit{s}\mathit{s} \mathit{a}\mathit{f}\mathit{f}\mathit{e}\mathit{c}\mathit{t}\mathit{e}\mathit{d})\\ \mathbf{2}\hfill & \hfill \mathit{i}\mathit{f} {\mathit{\mu }}_{\mathbf{1}}<{\mathit{Y}}_{\mathit{i}\mathit{j}}^{*}\le {\mathit{\mu }}_{\mathbf{2}}\hfill & \hfill (\mathit{M}\mathit{o}\mathit{d}\mathit{e}\mathit{r}\mathit{a}\mathit{t}\mathit{e}\mathit{l}\mathit{y} \mathit{a}\mathit{f}\mathit{f}\mathit{e}\mathit{c}\mathit{t}\mathit{e}\mathit{d})\\ \mathbf{3}& \hfill \mathit{i}\mathit{f} {\mathit{Y}}_{\mathit{i}\mathit{j}}^{*}>{\mathit{\mu }}_{\mathbf{2}}\hfill & \hfill (\mathit{H}\mathit{i}\mathit{g}\mathit{h}\mathit{l}\mathit{y} \mathit{a}\mathit{f}\mathit{f}\mathit{e}\mathit{c}\mathit{t}\mathit{e}\mathit{d})\end{array}\right.$$

(2)

where the threshold ${\mathit{\mu }}_{\mathbf{1}}$ and ${\mathit{\mu }}_{\mathbf{2}}$ are re estimated alongside the regression coefficients. The cumulative probabilities for each response category are derived using the standard normal cumulative distribution function $\Phi$ as shown in Equation (3):

$$\begin{gathered} \mathit{P}\left(\mathit{Y}\mathit{i}=\mathbf{1}\right)=\mathit{\Phi }\left(\mathit{\mu }\mathbf{1}-{\mathit{\beta }}^{\prime }\mathit{X}\mathit{i}\right) \\ \mathit{P}\left(\mathit{Y}\mathit{i}=\mathbf{2}\right)=\mathit{\Phi }\left(\mathit{\mu }\mathbf{2}-{\mathit{\beta }}^{\prime }\mathit{X}\mathit{i}\right)-\mathit{\Phi }\left(\mathit{\mu }\mathbf{1}-{\mathit{\beta }}^{\prime }\mathit{X}\mathit{i}\right) \\ \mathit{P}(\mathit{Y}\mathit{i}=\mathbf{3})=\mathbf{1}-\mathit{\Phi }(\mathit{\mu }\mathbf{2}-{\mathit{\beta }}^{\prime }\mathit{X}\mathit{i}) \end{gathered}$$

(3)

The explanatory variables ${\mathit{X}}_{\mathit{i}}$ include:

• Socio-demographic characteristics: gender, age, education level, and type of farmer;
• Contextual factors: duration of observing climate change, access to climate information, source of information, distance to farm, and use of indigenous knowledge;
• Perceived impact of extreme weather events: drought, flooding, frost, hail, and strong winds.

Each dependent variable was modeled independently to allow for infrastructure-specific interpretation and to better capture variations in perceived vulnerability. While multivariate ordered probit models would allow for estimation of correlated residuals, such models require more restrictive assumptions, larger sample sizes, and more complex estimation procedures (e.g., the generalized structural equation model or Bayesian approaches) [20]. Given the modest sample size (n = 60) and the focus on individual infrastructure responses, the use of separate ordered probit models was methodologically appropriate and offered clearer interpretability.

This modelling approach has been validated in similar agricultural and environmental perception studies, such as [21] on climate adaptation in Ethiopia, and [22] on smallholder resilience in South Africa.

2.3.3. Model Comparison and Statistical Justification

To evaluate the best-fitting approach, we compared the ordered probit model against a standard linear regression model (OLS) using statistical performance metrics—the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC)—as shown in

Table 1. Model comparison.

Model Type	AIC	BIC	Log-Likelihood	Degrees of Freedom
Ordered probit	80.14	113.65	−24.07	16
OLS Regression	113.20	144.62	−41.60	15

. These criteria penalize model complexity while rewarding model fit, with lower values indicating better model performance.

As shown above, the OP model dramatically outperforms the OLS model, with significantly lower AIC and BIC values. This supports the theoretical argument that OLS is not suitable for ordinal dependent variables, especially when the response categories (e.g., low, moderate, high impact) are non-continuous and non-equidistant. The OLS model had weak explanatory power (Adjusted R² = 0.0426; p = 0.3174), failing to capture meaningful variation in the data.

By contrast, the ordered probit model (ordered probit), not only respects the ordinal structure of the dependent variables but also improves fit and interpretability. Furthermore, the ordered probit adds value by modeling multiple correlated outcomes jointly, accounting for unobserved systemic effects (e.g., infrastructure systems damaged together during a single flood).

This empirical and theoretical justification reinforces the selection of ordered probit as the most appropriate model for the research objectives.

2.3.4. Independent and Dependent Variables

Table 2. Description and measurement of dependent variables.

Dependent Variables	Categories (Ordinal Scale)	Measurement Description	Justification
Impact on Bridges	1 = Less affected, 2 = Moderately affected, 3 = Highly affected	Farmer perception of flood/wind/frost-induced damage on bridges used to access farms or markets.	Bridges are critical for input delivery and produce movement.
Impact on Roads	1 = Less affected, 2 = Moderately affected, 3 = Highly affected	Perceived extent of road surface degradation or access issues due to climate events (e.g., flooding, drought).	Roads are essential for logistics and mobility.
Impact on Arable Land	1 = Less affected, 2 = Moderately affected, 3 = Highly affected	Soil quality, erosion, or reduced usability due to flooding, frost, or drought, as reported by the farmer.	Core agricultural resource—damage reduces food production [23].
Impact on Erosion Structures	1 = Less affected, 2 = Moderately affected, 3 = Highly affected	Perceived damage to soil bunds, contours, or vegetation barriers due to extreme rainfall, hail, or wind.	Controls land degradation, especially on sloped fields [24].
Impact on Dipping Tanks	1 = Less affected, 2 = Moderately affected, 3 = Highly affected	Reported physical damage or functionality loss in livestock dipping tanks due to storm or flood events.	Key infrastructure for livestock health and disease control.
Impact on Fences	1 = Less affected, 2 = Moderately affected, 3 = Highly affected	Reported breakage or degradation of fencing due to wind, hail, or erosion effects.	Fences protect crops and livestockdamage can result in losses.

outlines the six ordinal categorical dependent variables used in the analysis, representing perceived climate-induced damage to agricultural infrastructure. Each variable was measured on a three-point ordinal scale, with severity levels based on farmers’ experience of damage from extreme weather events. The table includes definitions, measurement procedures, and justification for inclusion.

Table 3. Justification for the inclusion of hypothesized independent variables.

Variable	Type	Expected Influence	Justification	Reference
Gender	Categorical (0 = Female, 1 = Male)	(+/−)	Gender influences farming roles and exposure to infrastructure. Women may report different impacts than men due to differentiated access or usage patterns.	[25]
Age Category	Categorical (1 = 18–29 Years, 2 = 30–44 Years, 3 = 45–59 Years, 4 = 60–70 Years 5 = 71 or older)	(+/−)	Older farmers may have more experience observing climate effects but lower adaptability to infrastructure changes.	[26]
Education Level	Categorical (1 = No Education, 2 = Primary school, 3 = High School, 4 = TVET College 5 = University)	(+)	Higher education levels may enhance understanding of climate risks and reporting accuracy regarding infrastructure vulnerability.	[27]
Farmer Type	Categorical (1 = Subsistence Farmer, 2 = Smallholder Farmer, 3 = Commercial farmer)	(+)	Full-time or commercial farmers may perceive greater infrastructure damage due to increased reliance on these systems.	[28]
Duration of Climate Observation	Ordinal (1 = Past 20 Years, 2 = Past 10 Years, 3 = Past 5 Years, 4 = Last Years)	(+)	Longer exposure to climate change increases the likelihood of observing infrastructure degradation.	[10]
Access to Climate Information	Categorical (1 = Yes, 2 = No, 3 = Not sure)	(+)	Access to climate information may improve awareness and reporting of climate-related infrastructure damage.	[29]
Source of Climate Information	Categorical (1 = Radio/TV, 2 = Newspaper, 3 = Extension Officers.)	(+/−)	The type of information source may influence how farmers perceive and act on climate infrastructure risk.	[29]
Farming Distance (to Plot)	Categorical (1 = less than 5 km 2 = More than 5 km)	(+/−)	Distance may influence exposure and dependence on roads and bridges.	[29]
Indigenous Knowledge Use	Categorical (1 = Detect weather, 2 = Restore soil & plant health, 3 = Treat livestock diseases, 4 = Purify water)	(+/−)	Use of traditional knowledge may affect perception and preparedness for infrastructure impacts.	[30]
Level of Impact—Flooding	Ordinal (1 = Low, 2 = High, 3 = Extreme)	(+)	Floods are major climate stressors, causing damage to roads, bridges, and land.	[31]
Level of Impact—Drought	Ordinal (1 = Low, 2 = High, 3 = Extreme)	(+)	Droughts weaken roads and soil structures, especially unpaved ones.	[10]
Level of Impact—Frost	Ordinal (1 = Low, 2 = High, 3 = Extreme)	(+)	Frost can damage both crops and associated infrastructure like tanks and pipes.	[32]
Level of Impact— Hail	Ordinal (1 = Low, 2 = High, 3 = Extreme)	(+)	Hail may physically damage erosion control structures and fencing.	[33]
Level of Impact—Strong Winds	Ordinal (1 = Low, 2 = High, 3 = Extreme)	(+)	Wind may destroy fencing and structures, and damage topsoil, affecting erosion systems.	[9]

presents the independent variables included in the multivariate ordered probit model, detailing their type, hypothesized directional influence on perceived infrastructure impact, justification based on theory and empirical literature, and supporting references. These variables capture socio-demographic, economic, and climate exposure dimensions relevant to communal farmers’ climate vulnerability.

3. Results

3.1. Descriptive Analysis

The results in

Table 4. Descriptive analysis of independent variables.

Variables	Measurements	Percent (%)	Frequency (n)
Gender	1 = Male	72	43
	2 = Female	28	17
Age	1 = 18–29 years	3	2
	2 = 30–44 years	17	10
	3 = 45–59 years	36	22
	4 = 60–70 years	42	25
	5 = 71 or older	2	1
Education	1 = No education	10	6
	2 = Primary school	30	18
	3 = High school	55	33
	4 = TVET college	2	1
	5 = University	3	2
Type of Farmer	1 = Subsistence	18	11
	2 = Smallholder	78	47
	3 = Commercial	4	2
Duration Observing Climate Change	1 = Past 20 years	48	29
	2 = Past 10 years	33	20
	3 = Past 5 years	12	7
	4 = Last year	7	4
Access to Climate Information	1 = Yes	93	56
	2 = No	5	3
	3 = Not sure	2	1
Source of Climate Information	1 = Radio/TV	93	56
	2 = Newspaper	7	4
	3 = Extension officers	0	0
Farming Distance	1 = <5 km	92	55
	2 = >5 km	8	5
Use of Indigenous Knowledge	1 = Detect weather	3	2
	2 = Restore soil & plant health	87	52
	3 = Treat livestock diseases	8	5
	4 = Purify water	2	1

reveal significant demographic, educational, and farming characteristics among respondents, shedding light on their exposure to climate change and its impacts. The sample is predominantly male (72%), consistent with studies showing male dominance in farming, particularly in rural areas where men often engage in more commercial-oriented agriculture while women focus on household subsistence [34,35]. Furthermore, the farming population is aging, with 78% of respondents aged 45–70 years. This aligns with global trends, particularly in developing regions, where youth migration to urban areas leaves older generations managing farms, contributing to concerns about the sustainability of rural agriculture [36].

Table 4 also shows that education levels highlight a significant barrier to agricultural innovation, with 10% of respondents having no education and 85% limited to primary or high school education. Limited access to higher education (only 3% have university qualifications) reflects challenges in rural areas, where educational opportunities are constrained. This is critical, as studies indicate that higher education correlates with the adoption of advanced farming technologies and climate-smart practices, which are vital for building resilience against climate change [37]. The dominance of smallholder farmers (78%) in the sample is typical of rural agricultural systems in developing countries, where small-scale farming is the backbone of food security but often faces resource constraints that exacerbate vulnerabilities to climate change [37].

Regarding climate change, the majority of respondents (93%) have access to information, primarily through radio and television, while agricultural extension services are notably absent. This reflects a reliance on mass media for disseminating agricultural information in rural areas, but it also highlights the limitations of agricultural extension systems, which are often underfunded and understaffed making it difficult to meet the extension-to-farmer ratio [38]. Moreover, 48% of respondents have observed the impacts of climate change over the past 20 years, suggesting increasing awareness of its long-term effects. Proximity to farms, with 92% located within 5 km, further underscores the smallholder context, where accessibility to land and resources plays a critical role in farmers’ adaptation strategies [39].

The impacts of extreme weather events are stark, with flooding identified as the most severe challenge82% of respondents reported extreme impacts from flooding, highlighting its destructive potential. This finding is consistent with global evidence showing that flooding is one of the most devastating climate-related events, particularly for smallholder farmers with limited adaptive infrastructure [40]. Similarly, drought is a significant concern, with 67% experiencing extreme impacts. This is in line with studies showing that rain-fed agriculture, common among smallholder farmers, is particularly vulnerable to prolonged dry spells, reducing crop yields and water availability [41]. Hail and frost also pose considerable risks, with high and extreme impacts reported by over 70% of respondents for both events, corroborating research on their potential to cause sudden and widespread crop damage. Strong winds were another major concern, with 91% of respondents reporting severe impacts, reflecting their capacity to disrupt agricultural activities and compound existing vulnerabilities.

3.2. Level of Impact of Extreme Weather Events on Agricultural Resources and Infrastructure

The results from the survey on the impact of extreme weather events like flooding and heavy rainfall on agricultural resources and infrastructure indicate a significant vulnerability of key farming assets to extreme weather events. The results show that bridges are among the most severely affected agricultural infrastructure components in the study area, with 85% of respondents indicating that bridges were highly affected by climate change-related impacts. In contrast, only 10% reported that bridges were moderately affected, and a mere 5% considered them to be less affected. The study has revealed that the study area is dominated by gravel roads with low-lying bridges (Figure 3), and in most instances without bridges, rendering the area to be highly vulnerable to flooding during heavy rainfall (Figure 4a,b). This situation disrupts access to essential farming services, collapsing food systems. Arable land was reported to be heavily affected, with 81% of respondents indicating that it was highly affected by flooding. This suggests that flooding severely damages crops, erodes soil, and reduces the productivity of farmland, making it highly susceptible to the adverse effects of extreme weather. A smaller portion of respondents (12%) reported moderate effects, and only 7% indicated that their arable land was less affected, highlighting the widespread damage flooding causes to farmland.

Access roads to agricultural fieldsessential for transporting inputs, moving harvested produce, and facilitating market accessemerged as the most severely affected type of infrastructure in the study area.

Table 5. Level of impact of extreme weather events.

Affected Agricultural Resources and Infrastructure	Less Affected	Moderately Affected	Highly Affected
Percent (%)
Bridges	5	10	85
Arable land	7	12	81
Dipping tanks	13	32	55
Fences	22	15	63
Roads	5	8	87
Soil erosion control structures	2	17	81

shows that a striking 87% of respondents reported that roads were highly affected, primarily due to flooding and related extreme weather events. In contrast, only 8% indicated that roads were moderately affected, and a mere 5% reported them as less affected. This highlights the major disruption caused by flooded roads, which impede farmers’ ability to access markets, transport goods, and receive emergency services, further compounding the economic and operational challenges farmers face during floods.

In terms of dipping tanks, which are crucial for managing livestock health, 55% of respondents reported that they were highly affected by flooding, while 32% indicated moderate effects. This points to the significant disruption flooding causes in maintaining livestock health, as damaged dipping tanks can hinder efforts to control livestock parasites, such as ticks and associated diseases. Only 13% reported that dipping tanks were less affected, emphasizing the vulnerability of these facilities during flooding events. Fences, another critical component for livestock management and protecting crops, were highly affected by floods in 63% of cases, with 15% reporting moderate effects. Fences play a key role in keeping livestock contained, restrict theft and safeguarding crops from damage, and their destruction during flooding can lead to further agricultural losses. The extensive impact on fences suggests that flooding significantly disrupts farm operations and poses a threat to the security of agricultural land.

Soil erosion control structures, which are essential for preventing soil degradation, reducing runoff of water, and maintaining land productivity, were highly affected in 81% of cases. Only 2% reported less affected structures, and 17% reported moderate effects. Flooding compromises the effectiveness of these structures, leading to soil erosion and long-term degradation of land. This result underscores the vulnerability of erosion control infrastructure during extreme weather events, which exacerbates the challenge of sustaining soil health and agricultural productivity.

3.3. Multicollinearity Test of Variables

The multicollinearity among the independent variables was assessed by using the Variance Inflation Factor (VIF), which is acceptable if the values are below 10. The findings in

Table 6. Multicollinearity test of variables.

Variable	VIF	1/VIF
Level of impact by frost	2.45	0.407762
Level of impact by hail	2.04	0.489306
Age	1.79	0.559771
Level of impact by drought	1.56	0.641908
Gender	1.50	0.666705
Level of impact by strong winds	1.49	0.672510
Access to CC infomation	1.46	0.685325
Use of indigenous knowledge	1.44	0.693894
Type of farmer	1.34	0.748868
Farming distance km	1.33	0.750033
Level of impact by flooding	1.29	0.775382
Duration of observing CC	1.26	0.795859
Source of CC information	1.23	0.815966
Education	1.21	0.824897
Mean VIF	1.53

indicate no multicollinearity issues since all the VIF values were below this threshold [42].

3.4. Econometric Analysis: Multivariate Ordered Probit Regression Analysis

The results presented in

Table 7. Multivariate ordered probit regression results showing the determinants of climate change impact on agricultural infrastructure.

	Agricultural Infrastructure
Variables	Bridges	Arable Land	Dipping Tanks	Fences	Roads	Soil Erosion Control Structures
Independent Variables	Coefficient (Standard Errors), p-value *,**,***
Gender	1.392 (1.074) n.s	−0.346 (0.687) n.s	−1.040 (0.477) **	0.772 (0.531) n.s	3.224 (1.488) **	2.499 (1.021) **
Age	−0.171 (0.369) n.s	0.518 (0.373) n.s	−0.162 (0.261) n.s	−0.006 (0.264) n.s	−0.374 (0.429) n.s	0.056 (0.438) n.s
Education	−0.367(0.294) n.s	−0.093 (0.302) n.s	−0.075 (0.249) n.s	−0.393 (0.228) *	−0.352 (0.331) n.s	−0.136 (0.296) n.s
Type of farmer	1.381 (0.776) *	0.183 (0.662) n.s	1.698 (0.499) ***	0.426 (0.486) n.s	2.239 (0.937) **	1.546 (0.736) **
Duration of observing CC	−0.182 (0.288) n.s	−0.171 (0.292) n.s	0.206 (0.224) n.s	0.255 (0.226), 0.258	−0.146 (0.391) n.s	−1.066 (0.394) ***
Access to CC information	4.916 (677) n.s	5.434 (577.774) n.s	1.729 (1.002) *	5.209 (345.225) n.s	6.996 (560.984) n.s	4.371 (509.127) n.s
Source of CC information	4.606 (1272) n.s	−1.421 (0.998) n.s	0.953 (0.818) n.s	−0.576 (0.824) n.s	6.580 (769.737) n.s	−1.548 (0.879) *
Farming distance km	5.86511(883) n.s	6.187 (606.253) n.s	1.309 (0.947) n.s	0.858 (0.836) n.s	0.912 (1.228) n.s	−0.769 (1.010) n.s
Use of indigenous knowledge	0.302 (0.719) n.s	2.383 (1.281) *	0.394 (0.608) n.s	1.015 (0.571) *	−0.150 (0.929) n.s	0.780 (0.766) n.s
Level of impact by frost	−0.341 (0.595) n.s	1.279 (0.563) **	1.016 (0.430) **	−0.463 (0.413) n.s	0.473 (0.653) n.s	1.235 (0.663) *
Level of impact by strong winds	0.742 (0.470) n.s	0.844 (0.439) *	−0.194 (0.394) n.s	0.620 (0.381) n.s	0.475 (0.656) n.s	0.801 (0.500) n.s
Level of impact by flooding	0.923 (0.471) **	1.052 (0.581) *	−2.249 (0.894) **	0.591 (0.437) n.s	−0.809 (0.893) n.s	0.779 (0.664) n.s
Level of impact by drought	1.392 (1.073) n.s	−0.471 (0.359) n.s	0.130 (0.304) n.s	−0.441 (0.326) n.s	1.046 (0.504) **	−0.145 (0.382) n.s
Level of impact by hail	−0.171 (0.369) n.s	−0.872 (0.542) n.s	−0.223 (0.366) n.s	0.534 (0.382) n.s	−1.014 (0.622) n.s	−1.364 (0.631) **
LR Chi2 (14)	17.40	23.45	40.67	23.23	22.41	24.49
Prob > Chi2	0.2357	0.0533	0.0002	0.0566	0.0707	0.0399
Pseudo R2	0.2797	0.3275	0.3525	0.2139	0.3883	0.3835
Log likelihood	−22.393558	−24.070507	−37.359034	−42.696815	−17.650134	−19.689231
AIC	76.78712	80.14101	106.7181	117.3936	67.30027	71.37846
BIC	110.2966	113.6505	140.2276	150.9031	100.8098	104.888
No. of observations	60	60	60	60	60	60

Coefficients are presented with standard errors in parentheses. Statistical significance is denoted as follows: * p < 0.10; ** p < 0.05; *** p < 0.01. n.s = not significant.

highlight the significant impact of various socio-demographic and environmental factors on agricultural infrastructure under the effects of climate change. These findings underscore the differentiated vulnerabilities faced by farmers based on their gender, farming type, exposure to climate information, and experience with extreme weather events such as frost, flooding, and drought.

3.4.1. Roads, Bridges, and Arable Land

Table 7 shows that gender plays a significant role in how farmers perceive the impacts of climate change on agricultural infrastructure. Gender is positively associated with the perceived impact of climate change on roads (coef. = 3.224, p = 0.030) and soil erosion control structures (coef. = 2.499, p = 0.014), suggesting that male farmers were more likely to report high levels of infrastructure damage. This could be due to their greater involvement in land preparation, transport, and infrastructure use. These results align with [43], who found that gendered roles affect the perception of climate risks, with men being more involved in infrastructure management. Male farmers were more exposed to the physical effects of infrastructure failures.

Flooding is positively associated with the perceived impact of climate change on bridges (coef. = 0.923, p = 0.050) and arable land (coef. = 1.052, p = 0.070), corroborating previous findings by Diakakis et al. (2020) [11], who reported frequent bridge and field damage during extreme rainfall events. Frost was significantly associated with damage to arable land (coef. = 1.279, p = 0.023), as low temperatures affect soil quality and crop viability.

The type of farmer was significant for both bridges (coef. = 1.381, p = 0.075) and roads (coef. = 2.239, p = 0.017), indicating that full-time and commercial farmers are more likely to perceive severe infrastructure impacts. These farmers rely more heavily on infrastructure for production and market access, making them more sensitive to its degradation.

3.4.2. Dipping Tanks and Fences

Table 7 further shows that gender was negatively associated with the perceived impact on dipping tanks (coef. = −1.040, p = 0.029), implying that female farmers perceived lower levels of impact, possibly due to less frequent use of livestock infrastructure. In contrast, education was marginally significant for fences (coef. = −0.393, p = 0.084), suggesting that more educated farmers perceived fewer impacts, potentially due to better management or maintenance practices.

The type of farmer also played a critical role in dipping tank vulnerability (coef. = 1.698, p < 0.001), with smallholder and commercial farmers reporting higher perceived impacts. Flooding showed a strong negative association with dipping tanks (coef. = −2.249, p = 0.012), likely reflecting physical inundation and damage to these ground-level structures during rainstorms. Access to climate information was significant (coef. = 1.729, p = 0.084), suggesting that betterinformed farmers are more aware of potential infrastructure damage, aligning with [29], who emphasized the role of information in enhancing risk perception.

3.4.3. Soil Erosion Control Structures

For erosion structures, gender again had a significant positive effect (coef. = 2.499, p = 0.014), and the type of farmer remained significant (coef. = 1.546, p = 0.036), reiterating the theme that those more directly engaged in farm operations perceive higher vulnerability. Frost (coef. = 1.235, p = 0.062) and hail (coef. = −1.364, p = 0.030) were also statistically significant. These results indicate that climatic extremes such as sudden temperature drops and hail events can cause physical erosion or blockages in water diversion structures. These findings align with [44], who highlighted weather-induced degradation of conservation infrastructure in rural landscapes. The source of climate information had a marginally significant negative association (coef. = −1.548, p = 0.078), implying that conventional channels like radio/TV may not be effective in communicating technical risks about erosion structures.

4. Discussion

This study investigated the perceived impact of climate change on rural agricultural infrastructure in communal farming systems using a multivariate ordered probit model. The analysis identified statistically significant socio-demographic and climatic predictors for infrastructure degradation, including gender, education, farmer type, access to climate change information, and extreme weather events such as flooding, frost, and drought.

4.1. Socio-Demographic Determinants of Perceived Infrastructure Impact

Gender emerged as a significant predictor across several infrastructure types. Specifically, being male was associated with significantly greater perceived impacts on roads (coef. = 3.224, p = 0.030) and soil erosion control structures (coef. = 2.499, p = 0.014, 95), while it was negatively associated with perceived damage to dipping tanks (coef. = −1.040, p = 0.029). This finding suggests that male farmers are more exposed to or aware of infrastructure deterioration due to their higher involvement in transport, land preparation, and erosion management activities. Conversely, female farmers may engage less with dipping tanks or rely on different animal husbandry practices. These findings support the gendered vulnerability framework proposed by [45], which asserts that gender roles shape access to and perception of environmental resources and infrastructure.

While age was not statistically significant in the multivariate model (p > 0.10), its role warrants interpretive attention. Older farmers may perceive higher impacts due to accumulated experience, historical knowledge and indigenous knowledge of climate variability. Their deeper familiarity with local infrastructure also likely makes them more sensitive to gradual degradation. Moreover, older farmers often have reduced access to modern technologies or alternative coping mechanisms, making them more reliant on existing, often fragile, communal infrastructure.

Education was marginally significant for fence infrastructure (coef. = −0.393, p = 0.084), indicating that farmers with higher education levels reported lower levels of impact. This could be attributed to a better understanding of climate adaptation practices or more diversified livelihoods that reduce infrastructure dependency. This aligns with research by [46], who found that education enhances adaptive capacity through improved risk assessment and resource use.

The type of farmer was a statistically significant predictor for multiple infrastructures. For example, commercial and smallholder farmers reported significantly higher perceived impacts on dipping tanks (coef. = 1.698, p < 0.001), roads (coef. = 2.239, p = 0.017), soil erosion structures (coef. = 1.546, p = 0.036), and bridges (coef. = 1.381, p = 0.075). These groups are more reliant on consistent and functional infrastructure for accessing markets, transporting inputs and outputs, and managing livestock. This finding aligns with [47], who emphasized that commercial farmers’ reliance on external infrastructure increases their vulnerability to climate-induced disruptions.

4.2. Role of Information Access and Indigenous Knowledge

Access to climate change information was positively associated with higher perceived impacts on dipping tanks (coef. = 1.729, p = 0.084). This suggests that better-informed farmers may recognize threats more clearly or are more observant of physical degradation. Although extension services were not directly measured, this result indirectly supports the idea that timely and relevant climate information can raise awareness, consistent with [29]. However, the source of climate information (e.g., radio/TV) was negatively associated with the perceived impact on erosion structures (coef. = −1.548, p = 0.078), possibly due to generic messaging that fails to highlight technical infrastructure risks.

The use of indigenous knowledge was significant for arable land (coef. = 2.383, p = 0.063) and fences (coef. = 1.015, p = 0.076). While indigenous practices are beneficial for adapting to climate variability (e.g., through crop diversification or weather forecasting), they may not fully compensate for physical infrastructure degradation, particularly in managing erosion or protecting livestock areas. This partial limitation of indigenous knowledge echoes findings from [48].

4.3. Impact of Extreme Weather Events on Infrastructure

The results confirm that extreme weather events are key drivers of infrastructure vulnerability. Flooding was significantly associated with damage to bridges (coef. = 0.923, p = 0.050) and dipping tanks (coef. = −2.249, p = 0.012). Bridges are highly exposed to high-flow water events and are structurally vulnerable to washout, while dipping tanks—often located in low-lying areas—are prone to flooding-related erosion and waterlogging.

Frost had significant effects on arable land (coef. = 1.279, p = 0.023), dipping tanks (coef. = 1.016, p = 0.018), and erosion control structures (coef. = 1.235, p = 0.062), reflecting how cold stress damages soil and weakens constructed barriers. These results align with [49], who documented frost as a contributor to the collapse of small-scale conservation infrastructure.

Drought significantly affected roads (coef. = 1.046, p = 0.038), likely due to surface cracking, soil shrinkage, and weakening of unpaved roadbeds. This matches findings by de [10], who noted increased maintenance costs for rural roads during drought conditions. Additionally, strong winds showed marginal significance for arable land (coef. = 0.844, p = 0.054), indicating that wind-related soil erosion contributes to perceptions of land degradation. It is important to note that certain types of infrastructure like roads and bridges are more vulnerable to extreme weather events due to continuous neglect and infrequent or even no maintenance by relevant authorities in communal areas compared to commercial farms.

The damage to the infrastructure in these communal areas delays agricultural activities such as accessing inputs in towns, land preparations, planting, etc. Such delays lead to severe crop yield reductions and livestock mortality, with negative impacts on the local livelihoods, resulting in household food insecurity.

The limitation of this study is that it focused on the Northern Drakensberg areas only, which was due to financial constraints. The study could not cover the Southern Drakensberg areas, which is also affected by climate variability in the form of high rainfall intensity and drought. Furthermore, the study did not capture other resources such as machinery, equipment, and agricultural inputs as practical constraint, as there was limited time to fully explore the damage during the peak of events. The study was conducted a few weeks later, due to road damage and furthermore, this restricted immediate access for field visits to conduct on-time assessments.

5. Conclusions and Policy Implications

This study examined how climate change is perceived to affect critical agricultural infrastructure within communal land farming systems in oKhahlamba Local Municipality, South Africa. Using separate ordered probit regression models, the analysis identified that climate-induced events (particularly flooding, frost, and drought) pose escalating risks to infrastructure such as roads, bridges, dipping tanks, and soil erosion control structures. These risks are further influenced by socio-demographic characteristics, farming commodity and experience, and access to climate information.

Emerging climate hazards, especially extreme rainfall events and seasonal frost, are expected to increase in intensity and frequency over the coming years. These phenomena are most likely to disrupt already fragile infrastructure, impede market access, exacerbate erosion, and impair livestock management. Importantly, the study found that commercial and smallholder farmers, who rely heavily on infrastructure for production and trade, are more likely to perceive infrastructure as vulnerable, highlighting their central role in resilience planning.

The findings also reveal that farmers with access to climate information or indigenous knowledge report more nuanced perceptions of risk, emphasizing the importance of both formal and informal knowledge systems in adaptive decision-making. However, disparities in access and interpretation suggest the need for more localized and inclusive communication strategies.

From a policy standpoint, the results call for targeted climate adaptation policies that integrate infrastructure resilience into broader agricultural planning. This includes not only physical investments in durable infrastructure, but also capacity-building for farmers, gender-sensitive support systems, and coordinated institutional responses to manage and anticipate future climate shocks.

Recommendations

To mitigate these impacts, several targeted recommendations are proposed. First, investments in climate-adaptive and climate-resilient infrastructure are essential to improve the durability of roads, bridges, dipping tanks, and soil erosion control systems against extreme weather events. Technologies such as flood-resistant structures and erosion control mechanisms should be prioritized. In addition, investments in good -quality roads and high-level bridge construction are critical, as this infrastructure improves connectivity and facilitates access to essential services and markets, thus contributing to sustainable food systems, improved rural livelihoods, and sustainable food security.

Second, gender-sensitive adaptation strategies should be integrated into climate policies, recognizing the unique challenges faced by male and female farmers and promoting equitable access to resources.

Third, tailored support for smallholder and subsistence farmers, including access to funding, technical assistance, and agricultural insurance, is critical to enhancing their resilience to climate impacts.

Fourth, promoting capacity building and education on climate change adaptation through extension services and access to reliable climate information will empower farmers to adopt sustainable practices.

Fifth, there is urgent need for government to implement its Policies and Strategies including National Climate Change Adaptation Strategy; Climate Smart Strategic Framework; National Spatial Development Framework; and Presidential Climate Change Commission.

Lastly, integrating indigenous knowledge with modern solutions can provide holistic and context-specific approaches to managing agricultural infrastructure under changing climatic conditions.

Considering the aforementioned proposed study recommendations, the authors further propose an in-depth interrogation of the key production systems used in communal land. The authors propose immediate responsiveness fostered by local government in exploring the varied degrees of severity in both crop and livestock production systems. Exploratory interventions should prioritize establishing local facilities that support the agricultural resilience of resources, such as establishing community seed banks that can preserve genetic diversity of crops, including indigenous seeds, as a form of insurance against crop losses. In addition, investment in innovative and indigenous rural storage facilities must be locally enacted to reduce losses and spoilage. Moreover, the accessibility of mobile veterinary services is fundamental to mitigate against climate change’s impact on livestock health and disease management. As the trends of climate change continue, monitoring and evaluation from impact assessments managed by local government should be more agile in rural settings and more futuristically inclined rather than event-reactive. Future research must explore an adaptive capacity analysis bearing in mind the interrelated challenges reported, which have a two-pronged impact on mixed-crop farmers using communal land farming systems. Addressing these challenges requires a collaborative effort among policymakers, agricultural stakeholders, and rural communities to ensure sustainable agricultural systems and the long-term resilience of rural livelihoods. The limitation of the study is that the survey was not conducted in other sub-wards around oKhahlamba Local Municipality due to the limitation in financial resources. Future research is necessary that will focus on the impact of climate change on local food systems based on the climate variability effects on agricultural infrastructure attested to in this study.