The Ngwezi River Basin is a topographic depression located in Southern Province and is bounded on the south by the Zambezi River as far west by western province boundary. The basin was formed in response to down-warping of the interior of the Southern Africa terrain (Haddon, 2005). Figure 1 below shows the study area which is Ngwezi River Basin. It is one of the major sub catchments of the Zambezi River and covers an area of about 4,338 km². It has been selected for this particular study as it is one of the driest regions in Zambia characterized as arid region climatically. It is a vulnerable region which is known for having past instances of major heat waves such as in 2010, it recorded 46.8 °C and 50 °C in 2016 causing high amount of fatalities. For this study, the Ngwezi River Basin was divided into three segments based on key hydrological reference points as shown in Figure 1. The upstream section extends from the earth dam to the northern part of the basin. The middle reach covers the stretch between the earth dam and the hydrometric station, while the downstream section extends from the hydrometric station to the southern end of the basin.

This study adopted a sequential explanatory mixed-methods design to investigate community-based water management practices in the Ngwezi River Basin, Southern Zambia. The design integrated quantitative household surveys with qualitative key informant interviews (KIIs), focus group discussions (FGDs), and participant observation. The quantitative phase provided measurable patterns of household water practices and perceptions, while the qualitative phase explored the socio-cultural, institutional, and governance dimensions shaping these practices. Four core areas of inquiry—rainwater harvesting, soil management for water retention, social dynamics in water access, and crop diversification—were selected based on their relevance to community adaptation and their alignment with IWRM principles (Global Water Partnership (GWP), 2000). Data integration occurred during the interpretation stage, where qualitative insights contextualized quantitative findings to produce a holistic understanding of adaptive water governance.

The Ngwezi River Basin has an estimated population of 87,000 (CSO, 2022). Using Yamane (1967) formula for finite populations, a target sample of 398 households was calculated to achieve a 95% confidence level with a 5% margin of error. To enhance clarity and representativeness, the study adopted a mixed sampling design combining stratified random and purposive sampling approaches. In the first stage, stratified random sampling was applied to ensure proportional representation across the basin’s three geographic zones—upstream, middle reach, and downstream—each considered a distinct stratum based on hydrological and socio-economic characteristics. From community-prepared household lists, 130 households were randomly selected within each stratum using a random number generator to avoid selection bias.

In the second stage, purposive sampling was employed within each stratum to ensure the inclusion of vulnerable groups such as female-headed and child-headed households. Local leaders and community registers were used to identify these households, and their representation in the sample reflected their actual proportion in the population (e.g., if 12% of households were female-headed, approximately 12% of the sample in that stratum came from this group). These households were proportionally integrated to reflect actual demographic representation while minimizing bias in quantitative estimates.

After accounting for non-responses and incomplete questionnaires, a total of 390 valid household surveys were retained, representing a 92% response rate. Replacement sampling within strata was undertaken to minimize non-response bias and maintain proportional balance across zones. Although the purposive inclusion of vulnerable households slightly reduces the sample’s full statistical generalizability, the approach strengthened the study’s analytical depth by systematically capturing the experiences of marginalized groups while preserving the integrity of randomization across the wider population.

The quantitative phase included structured face-to-face household surveys, which were administered to household heads or, where applicable, the primary household water decision-makers. In cases where the household head was unavailable or not primarily responsible for daily water collection and management, the individual identified by the household as the primary decision-maker—typically a spouse or adult member responsible for sourcing and using water—was selected to ensure data accuracy and contextual relevance. Only one standardized questionnaire was used as the main survey instrument, and no other survey tools were administered. The questionnaire was divided into four clearly defined thematic sections designed to capture household-level adaptation and water management practices reflected in the study results. These themes included:

All questions in these sections were primarily closed-ended to enable quantitative comparison across the three basin zones, with a few open-ended prompts added to capture contextual explanations of household coping and adaptation behaviors. The single questionnaire was adapted from validated frameworks relevant to community water and agricultural management, including IWRM household assessment framework (Basu et al., 2021) and the Rural Water Governance Toolkit (Bisung and Elliott, 2014). Specifically, items on adaptive practices and institutional participation were drawn from the IWRM framework, while questions on governance, local innovations, and coping strategies were derived from the Rural Water Governance Toolkit. Each component was localized through consultation with community leaders and pilot testing to ensure contextual relevance.

The instrument was pre-tested with 30 households outside the main study area to refine question clarity, language, and response options. Content validity was established through expert review by three specialists in water governance and climate adaptation, who assessed whether each questionnaire item adequately represented the intended constructs—rainwater harvesting, soil management, social water practices, and crop diversification. Their feedback led to adjustments in terminology, sequencing, and item phrasing to ensure comprehensive coverage of all thematic areas and contextual appropriateness.

Translation into Tonga followed forward–backward translation procedures to maintain conceptual and linguistic equivalence. The qualitative phase included twelve key informant interviews (KIIs) conducted with purposively selected stakeholders: three government officials from the water sector, two local government representatives, two agricultural extension officers, two traditional leaders, and three community water committee members. A single semi-structured interview guide was used to explore institutional perspectives on water governance, integration of local practices within IWRM, and community adaptation strategies.

Additionally, nine focus group discussions (FGDs) were conducted—three per basin zone (upstream, middle, and downstream)—with participants segregated by gender to capture gender-differentiated experiences of water access and adaptation. This meant that one male FGD, one female FGD, and one mixed-gender FGD were held in each zone. Each group comprised 8–12 participants of varied ages, livelihoods, and degrees of water dependency. Participants were purposively selected to include child-headed households, elderly community members, and youths, representing diverse perspectives on water scarcity and coping responses. Discussions explored lived experiences of water shortages, traditional and collective water management, and community-based adaptation practices.

Quantitative data were entered and cleaned using SPSS version 29. Finally, the principal researcher conducted non-participant observation for two weeks in each basin zone, documenting household water collection routines, agricultural water use, and communal water-sharing activities. Detailed field notes and reflexive journals were maintained to triangulate quantitative and qualitative findings and minimize observer bias.

Quantitative data were entered, cleaned, and analyzed in statistical software. Descriptive statistics (frequencies, percentages) summarized household practices. Percentages were calculated relative to the total number of household survey respondents (n = 390) across the whole basin. Associations between geographic zones (upstream, middle, downstream) and adaptive strategies were tested using Chi-square (χ²) tests of independence. Missing values (<5% of cases) were treated with multiple imputation to preserve sample integrity. Qualitative data from KIIs, FGDs, and observations were transcribed verbatim, coded inductively, and analyzed through thematic analysis. Two independent coders conducted first-cycle coding, resolving discrepancies through consensus. Emergent themes highlighted adaptation strategies, governance challenges, and community perceptions. Credibility was enhanced through, peer debriefing, and intercoder checks. Mixed-methods integration was achieved during interpretation through side-by-side comparison of quantitative and qualitative results, enabling triangulation and a more nuanced understanding of local adaptive water management practices.

Communities across the Ngwezi River Basin employ a wide range of rainwater harvesting techniques, shaped by hydrological conditions and local knowledge. The most common strategies include rooftop rainwater harvesting, farm ponds, percolation pits, contour trenches, bunds and embankments, sand dams, and zai pits (Table 1). Distinct regional patterns were evident. Upstream households relied more on rooftop rainwater harvesting (n = 52; 40%) and contour trenches (n = 39; 30%), reflecting the need to capture rapid runoff on steep terrain. Downstream communities made greater use of farm ponds (n = 65; 50%) and sand dams (n = 52; 40%), taking advantage of seasonal flooding. Middle-reach communities adopted a combination of bunds (n = 59; 45%), farm ponds (n = 46; 35%), and percolation pits (n = 39; 30%) to balance runoff management and groundwater recharge. As one upstream farmer noted, “The rain does not stay long here—we must catch it quickly from the roof or by digging trenches.”

Several respondents emphasized the absence of formal technical support for these initiatives. One middle-reach farmer explained, “We build ponds ourselves because no one from the government comes to show us how; when NGOs came, they only stayed for a short time.” Similarly, a downstream community leader noted, “WARMA officers came once to register our boreholes, but they never helped us with these sand dams.” These accounts suggest that while rainwater harvesting aligns with IWRM principles of participation, it remains largely a community-driven effort with minimal institutional backing.

Water governance across the basin is also shaped by social practices, with three prominent forms identified: illegal water collection from private farms, traditional rainmaking ceremonies, and cross-community sharing of water resources (Table 2). Illegal water collection was reported most frequently downstream (n = 59; 45%), where water scarcity and competition are most intense during dry seasons. Rainmaking ceremonies were more prevalent upstream (n = 78; 60%), where traditional beliefs continue to influence water management. Cooperative water sharing was practiced across all regions but was highest downstream (n = 72; 55%) and in the middle reach (n = 65; 50%), reflecting local agreements to ensure equitable access. An elder upstream explained, “We still gather to call on the ancestors for rain—it keeps our traditions alive.”

Respondents also highlighted tensions arising from unequal access to formal water infrastructure. A downstream resident lamented, “The farms with private boreholes do not allow us to fetch water; when we try, they say it is illegal, yet we are all suffering from the same drought.” Another participant added, “The government talks about water permits, but no one in our village has ever received one—we just use what we have.” These sentiments underscore the coexistence of customary and formal systems, often resulting in overlapping and sometimes conflicting governance arrangements.

Crop diversification emerged as a key adaptation strategy to manage rainfall variability and maintain food security. Households reported adopting drought-resistant crops, crop rotation with water-efficient species, perennial crops, and intercropping (Table 3). Upstream farmers showed the highest adoption of drought-resistant crops (n = 78; 60%) and intercropping (n = 85; 65%), while middle-reach farmers emphasized intercropping (n = 65; 50%) and perennial crops (n = 46; 35%). Downstream households, facing more severe water stress, prioritized intercropping (n = 52; 40%) to maximize yields under constrained conditions. As one upstream farmer explained, “When rains are poor, sorghum and millet still give us something to eat.” However, participants reported limited collaboration between agricultural officers and local farmers. A middle-reach farmer noted, “We hear about new crops from the radio, not from the extension officer—they rarely visit.” Another respondent commented, “We learn from each other through field days organized by women’s groups, not the government.” These verbatims reveal that adaptive knowledge circulates primarily through social networks rather than formal institutional channels, reflecting weak cross-sectoral coordination in the basin.

Soil conservation and moisture-retention practices were widely reported across the basin, with contour plowing, terracing, mulching, and cover cropping being the most common (Table 4). Contour plowing (n = 78; 60%) and terracing (n = 72; 55%) were most prevalent upstream, consistent with efforts to stabilize steep slopes. Middle-reach households frequently practiced cover cropping (n = 65; 50%) and mulching (n = 52; 40%) to maintain soil fertility and conserve water. Adoption was lower downstream, though mulching (n = 39; 30%) showed signs of increasing uptake. An upstream farmer highlighted the importance of soil conservation, noting, “Without terraces, the rains would wash everything away.” In focus group discussions, farmers expressed frustration that government and NGO soil management projects were often top-down and unsustained. One participant stated, “They came with contour markers but never returned to see if we used them.” Another added, “Chiefs help us organize land care days, but the ministry does not support these efforts.” These perspectives suggest that while communities actively conserve soil, institutional follow-through and integration remain limited.

Chi-square tests were used to assess variation in the adoption of water management and agricultural practices across regions as shown in Table 5 below. Significant regional differences were found in several key practices, including rooftop rainwater harvesting (p = 0.003), farm ponds (p = 0.001), bunds and embankments (p = 0.04), sand dams (p = 0.005), and illegal water collection (p = 0.009). These results reflect how local hydrological, social, and economic factors shape water management behaviors across the basin. Focus group narratives further confirm that these regional variations correspond to differences in institutional presence. For example, one upstream respondent remarked, “We sometimes see government officers because we are near the town, but people downstream say they have never seen them.” This spatial disparity helps explain why community reliance on informal systems intensifies toward the downstream areas, where formal institutional engagement is weakest.

Rainwater harvesting in the Ngwezi River Basin is not merely a pragmatic reaction to ecological constraints but reflects what Boelens et al. (2019) describe as “situated rationality” within hydrosocial territories—landscapes where flows of water intersect with social relations and political authority. The results demonstrate statistically significant spatial variation in adoption: rooftop harvesting (p = 0.003), farm ponds (p = 0.001), sand dams (p = 0.005), and bunds/embankments (p = 0.04). These patterns confirm that adaptation is not randomly distributed but embedded in both physical geography and institutional context.

In upstream areas, where steep slopes accelerate runoff and formal water services are minimal, 40% of households have turned to rooftop harvesting and contour trenching. These practices create micro-scale infrastructure that harnesses vertical hydrological flows while compensating for state neglect. Downstream households, by contrast, rely heavily on farm ponds (50%) and sand dams. Their strategies draw on the seasonal flooding of lowlands and customary land tenure arrangements that allow communal experimentation with floodplain ecology. Such practices echo findings from arid regions of Kenya and India, where decentralized and low-cost harvesting approaches have consistently outperformed centralized infrastructure in enhancing resilience and accessibility (Pachpute et al., 2009; Biazin et al., 2012).

Institutional dynamics are equally influential. The coexistence of community-led rainwater structures and limited formal intervention reveals a parallel governance system where informal ingenuity substitutes for bureaucratic absence. While WARMA and local councils promote basin planning, their engagement in small-scale rainwater harvesting remains peripheral, leaving most community efforts unregistered and technically unsupported.

In zones where state presence is limited, communities practice institutional bricolage—blending modern rainwater tanks with ancestral trenching and bunding techniques. This confirms that adaptation in practice rarely follows top-down policy prescriptions but emerges from locally assembled arrangements that mix formal interventions with inherited ecological knowledge (Parida et al., 2025). Such hybrid practices represent latent sites of institutional interaction, where local actors reinterpret policy objectives within existing social networks. The Ngwezi case thus illustrates that rainwater harvesting is not a stopgap survival strategy but an evolving form of hydrosocial innovation with broader lessons for other semi-arid regions.

When viewed through the IWRM framework, these decentralized harvesting initiatives reflect the principle of participatory water management and local-level decision-making. However, the weak coordination between community initiatives and WARMA’s formal basin plans exposes a policy–practice gap that undermines the integration of indigenous strategies into official IWRM mechanisms. Bridging this gap would operationalize IWRM’s principle of subsidiarity by enabling community-led systems to complement formal infrastructure through local knowledge and accountability.

Water governance in the basin is shaped as much by cultural institutions as by formal policy. Statistically regional difference of the practices such as rainmaking ceremonies (p = 0.001) and illegal abstraction (p = 0.009) highlight that water management is simultaneously spiritual, social, and political. Rainmaking ceremonies, reported by 60% of upstream respondents, serve not only as spiritual invocations but also as community rituals that reinforce collective identity and hydrosocial obligations. As an elder explained, “We still gather to call on the ancestors for rain. It keeps our traditions alive.” These findings resonate with Zvingowanisei and Chirongoma (2024) and Mverecha and Midzi (2025), who document similar roles of spiritual ecologies in Zimbabwe, Ethiopia, and Mali. Yet within mainstream IWRM frameworks, such practices are often marginalized as “non-scientific,” illustrating a persistent blind spot in global water governance discourse.

Regional difference in water sharing (p = 0.002) and unauthorized abstraction (45%) practices further underscores governance asymmetries and infrastructural inequities. These practices should not be dismissed as illegality or moral failure; rather, they reflect conditions of exclusion where formal allocation systems clash with customary norms (Mirnezami et al., 2024; Nhapi, 2009). The results point to a broader politics of legitimacy: where communities perceive state systems as inaccessible or unjust, they turn to alternative arrangements that sustain water access but sit outside official frameworks. This reflects a dual authority structure, in which formal rules coexist and often compete with traditional governance, producing a complex field of overlapping jurisdiction.

This duality—spiritual governance upstream and informal redistribution downstream—demonstrates that hydrosocial resilience in the basin is inseparable from contested authority and institutional hybridity. From an IWRM standpoint, these findings expose the governance disconnect between statutory institutions and customary systems. Formal mechanisms under WARMA emphasize water permits, compliance, and catchment management boards, yet lack culturally responsive channels for including traditional leaders and ritual practices in decision-making. Bridging these institutional silos by recognizing traditional ecological knowledge as part of basin governance could foster co-management and reduce enforcement conflicts. Integrating these social and spiritual governance forms into sub-catchment councils could strengthen IWRM’s social acceptability and promote equity in water allocation.

Crop diversification emerges as a crucial adaptation pathway, though its significance varies spatially. The prevalence of the practice of planting drought-resistant crops was higher in the upstream region; however, this difference was not statistically significant (p = 0.64) and further the prevalence of perennial cropping was significantly higher in middle-reach areas. (p < 0.001). Here, mixed agroforestry systems—particularly cassava cultivation—buffer households against seasonal fluctuations in both food and water security.

Upstream households reported the highest adoption of intercropping (65%) and drought-tolerant cereals like sorghum (60%), strategies that blend ancestral seed knowledge (e.g., the local maize variety Gankanta) with NGO-supported extension programs. One farmer explained: “When rains are poor, sorghum still give us something to eat.” These patterns mirror regional findings in South Africa (Omotoso and Omotayo, 2025) and the Eastern Peninsula (Navarrete et al., 2024), where agrobiodiversity is increasingly recognized as ecological insurance against climate shocks.

Crucially, diversification here is not simply about substituting crops but about experimenting, sharing knowledge, and redistributing risk through farmer-to-farmer networks, informal seed banks, and women’s groups. This confirms that resilience is socially constructed: it depends less on the availability of technologies and more on the circulation of knowledge and support within communities. Ignoring these endogenous pathways risks undermining the very capacities that enable adaptation in volatile environments.

Institutionally, the limited coordination between agricultural extension officers, local farmer groups, and WARMA illustrates that cross-sectoral integration under IWRM remains weak. Current programs prioritize irrigation and commercial productivity over indigenous diversification systems, despite the latter embodying IWRM’s principles of efficiency and equity. These findings also highlight that current formal agricultural and water governance programs in Zambia—while aligned with IWRM’s multi-sectoral vision—tend to prioritize irrigation and commercial agriculture over smallholder adaptive diversification. By contrast, local diversification aligns more closely with IWRM’s sustainability and equity dimensions, suggesting that policy realignment toward community-supported, small-scale adaptive farming could enhance both water productivity and ecosystem resilience.

Soil management strategies reveal how farmers address the twin challenges of land degradation and water scarcity. Statistically significant adoption of contour plowing (p < 0.001) and widespread use of terracing upstream (55%) illustrate how households respond to erosion threats by investing labor into land stabilization. As one farmer observed: “Without terraces, the rains would wash everything away in some areas.”

Middle and downstream farmers relied more heavily on cover cropping and mulching, though these patterns were statistically non-significant. This differentiation reflects localized risk profiles: upstream, erosion is the main threat, while downstream, nutrient loss and evaporation dominate. The findings align with East African studies (Wawire et al., 2021; Mutungi et al., 2025), which show that land-based measures are often more practical than large-scale water infrastructure, particularly where capital access is limited.

However, the broader governance challenge persists. In Zambia, as elsewhere, land and water governance remain institutionally fragmented—split across ministries, donor projects, and customary tenure regimes. While IWRM is promoted as a unifying framework, in practice it often functions as a rhetorical ideal with little operational coherence. The Ngwezi evidence shows that soil and water conservation initiatives are rarely implemented jointly, reflecting weak institutional integration between the Ministry of Agriculture, WARMA, and traditional authorities. The Ngwezi evidence suggests that moving toward a landscape-based governance model—which recognizes hydrosocial interdependencies and supports local stewardship—may be more effective than continuing to treat land and water as administratively separate domains.

This fragmentation reflects the weak institutional link between the Ministry of Water Development and WARMA on one hand, and the Ministry of Agriculture and local traditional authorities on the other. Building functional coordination platforms at catchment level could translate IWRM principles into practice, enabling chiefs, farmer groups, and state officers to co-manage soil-water resources under shared accountability. Embedding soil-water conservation initiatives within IWRM catchment strategies—while acknowledging the custodial authority of chiefs—would address a key structural limitation in Zambia’s current governance landscape.

Across rainwater harvesting, ritual water governance, crop diversification, and soil management, a clear pattern emerges: communities in the Ngwezi River Basin are not passive beneficiaries of climate adaptation programs but active architects of resilience. Their strategies combine ecological intelligence, institutional improvisation, and social cooperation to craft workable solutions under persistent water stress. These practices collectively illustrate the adaptive capacity of rural societies operating at the margins of formal governance.

The Ngwezi case resonates strongly with global experiences of community-led adaptation in water-scarce environments. In the Andes, for example, local communities maintain centuries-old canal irrigation and waru-waru raised fields that regulate soil moisture and mitigate frost (Boelens, 2022). In the Sahel, farmers employ zai pits and contour bunds to harvest water and rehabilitate degraded soils, thereby improving crop yields under variable rainfall (Zougmoré et al., 2014). Similarly, in Rajasthan, India, traditional johad tanks and stepwells have been revived as decentralized storage systems that buffer communities against prolonged droughts (Saxena and Hughes, 2023). These parallels demonstrate a shared global logic: when ecological knowledge is embedded in social institutions, resilience becomes locally produced rather than externally prescribed. The Ngwezi Basin’s innovations—sand dams, communal water-sharing, and hybrid crop diversification—thus mirror a broader transregional pattern where cultural embeddedness and community agency, rather than centralized engineering, anchor adaptive water governance (Nyika and Dinka, 2023).

By foregrounding indigenous and hybridized strategies, this study underscores the limitations of standardized climate adaptation frameworks that overlook local agency. Instead, it advocates for context-specific, culturally attuned, and socially grounded approaches that integrate customary norms, gendered roles, and knowledge co-production. In doing so, it contributes new empirical evidence from Southern Africa—a region underrepresented in global water governance scholarship—and enriches comparative understandings of hydrosocial resilience across arid and semi-arid landscapes.

It is also important to note that the study employed a purposive sampling strategy, focusing on representative upstream, middle-reach, and downstream communities within the Ngwezi Basin. The results therefore capture in-depth contextual dynamics rather than patterns that are statistically generalizable to all of Zambia.

Crucially, by situating these findings within the IWRM framework, the study reveals that indigenous water management systems already embody key IWRM principles—participation, equity, sustainability, and inclusivity—often more effectively than formal institutional mechanisms. The persistent disconnection between WARMA, agricultural agencies, and traditional authorities highlights a structural governance gap that constrains the realization of integrated management. Bridging this divide requires institutional hybridization: establishing participatory catchment forums, co-management agreements, and joint monitoring mechanisms that recognize both customary authority and statutory mandates.

At the policy level, this synthesis offers an actionable pathway for localizing IWRM in Zambia and comparable semi-arid regions. Integrating indigenous practices into formal basin planning would enhance legitimacy, foster equity in water allocation, and promote adaptive learning across governance scales. Ultimately, reconciling indigenous systems with formal frameworks transforms IWRM from a technocratic model into a living, socially embedded process—capable of delivering equitable and climate-resilient water security for vulnerable communities.

This study has a few limitations. Data were collected once, so seasonal shifts in water scarcity and practices may not be fully captured. The sample was well distributed (n = 390) but included some purposively selected vulnerable households, which slightly limits generalizability. Because many practices (e.g., illegal abstraction) were self-reported, responses may contain social desirability bias. The quantitative analysis relied mainly on chi-square tests, showing association rather than causation. Finally, institutional findings reflect mostly community and local-leader perspectives, not a full audit of WARMA or district records. Despite these limitations, the findings provide robust and contextually grounded insights into how communities in the Ngwezi River Basin adapt to water scarcity and engage with local governance systems.