Heterogeneity in integration of greenhouse gas mitigating interventions among rural and Peri-urban smallholder dairy farms in Nakuru County, Kenya

Abstract

Dairy production is a priority livestock system for reduction of greenhouse gas (GHG) emissions and is a target of a number of GHG mitigating interventions (GHG-MI) recommendations. However, uncertainty exist regarding how smallholders are integrating these GHG-MI, which include improving feeding, breeding, health and welfare, and manure management. This study tested hypothesis that integration of GHG-MI is heterogeneous in smallholder dairy farms. A sample of 291 farms were obtained in a cross-sectional survey in rural and peri-urban wards around Nakuru city, Kenya. Frequency data on practicing a GHG-MI were subjected to Principal Component Analysis followed by Cluster Analysis to derive distinct clusters, resulting in three clusters: cluster one and two each comprising 36.8 % and cluster three 26.4 % of the farms. These were respectively labelled low (LM), average (AM) and high (HM) mitigating farms based on the number of GHG-MI frequently practiced by atleast half of the farms. The majority of the HM farms frequently practiced most of the GHG-MI and LM farms practiced the least. Out of 11 feeding practices examined, atleast 6 in 10 HM farms frequently practiced four whereas atleast 5 in 10 of AM or LM farms frequently practiced two. In examining selection for six traits, five were frequently selected by at least 8 in 10 of HM farms whereas atleast 5 in 10 of AM or LM farms only frequently selected two traits. Of seven good animal health and welfare practices examined, all were frequently practiced by atleast 6 in 10 of HM farms. Among AM and LM farms, atleast 5 in 10 practiced six good animal health and welfare. While dry-lot manure storage was only a frequent practice on AM farms, manure spreading on land was a more frequent practice across all clusters. These findings reveal that smallholder dairy farms are responsive to mitigating GHG emissions, however heterogeneity in their integration of GHG-MI warrants research to compare HM with the AM and LM farms. Feed additives and effective manure management should be entry points for extension service promoting climate-smart livestock interventions, because only a few farms used them, though are effective GHG-MI in dairy systems.

1 Introduction

The dairy sector faces the critical challenge of balancing a growing demand for dairy products with the urgent need to reduce its environmental footprint, particularly greenhouse gas (GHG) emissions. This is because the livestock sector accounts for two-thirds of the agricultural GHG emissions, estimated at 17% of the total global GHG emissions. In the dairy sector, methane accounts for up to 96% of the sector’s total GHG output (Food and Agriculture Organization of the United Nations, 2017).

Smallholder dairy farming in Sub-Saharan Africa has an estimated 9 tCO₂e per ton of milk, which is one of the highest GHG emission intensities globally (Hakuzimana et al., 2021). This places smallholder dairy systems as a target for urgent GHG mitigation interventions. In Kenya, the dairy subsector contributes approximately 20% of agricultural GHG emissions from 4.3 million dairy cows, underscoring the sector’s significance for climate mitigation (Onyango et al., 2024). Dairy cattle are ruminants, with their enteric fermentation accounting for half of the sector’s total emissions (Ntinyari and Gweyi-Onyango, 2021). Without mitigation interventions, dairy emissions in Kenya are projected to increase as annual milk consumption grows from 110 kg to 220 kg per person by 2030 (Vernooij et al., 2024). For this reason, dairy production is a priority in the national climate goal commitments (Adenaeuer et al., 2021; Huber et al., 2023).

Kenya’s nationally determined contributions (NDCs) target a 7% annual reduction in GHG emissions by 2030 (Chelang’a et al., 2025c; Wilkes et al., 2020). The dairy sector can contribute to the NDCs by integrating several GHG-mitigating interventions (GHG-MIs) recommended for dairy farms. These GHG-MIs can be grouped into enteric methane-reducing feed additives, improved feeding, selective breeding, enhanced animal health and welfare, and low-emission manure management (Beauchemin et al., 2022). In dairy farms, GHG-MI integration is bundled, but the extent of bundling can be variable (Caulfield et al., 2023; Chelang’a et al., 2025c). This has implications for the role of smallholder dairy farms in achieving GHG reduction in their farm management. Variability in bundling of GHG-MI implies that the integration can be associated with heterogeneity (Adenaeuer et al., 2024; Korir et al., 2023).

Use of feeds and feed additives has the potential to reduce methane by up to 11% on average. For instance, Rotz et al. (2025) observed that some feed additives reduce methane emission by up to 10%, though with large variation across farms, from 9.5 to 23%, depending on how the feed additive was integrated with manure handling. Feeding sweet potato vine silage (SPVS) can reduce milk emission intensities by up to 11% (Merbold et al., 2023). Other works reported that reducing forage-to-concentrate ratios, lowering dietary protein, and improving feed efficiency can reduce methane emissions by up to 10% (Rotz et al., 2025), but baseline feeding practices can have some influence. Improved feeding can be cost-effective, generating negative abatement costs due to concurrent increases in milk yield and farm revenue (Khatri-Chhetri et al., 2020).

Beyond improved feeding and feed additives, manure management interventions are options for achieving greater GHG emission reduction. For example, anaerobic digesters can achieve 7 to 27% reduction of a farm’s carbon footprint, and the greatest benefits are when a high proportion of manure is handled in liquid or slurry form (Rotz et al., 2025). Other manure-handling interventions, including sealed storage covers with flaring, slurry acidification, and solids separation, could have similar effects, but their attainable emission reductions need to be quantified (Kavanagh et al., 2021; Li et al., 2024; Kalson et al., 2025).

Change in animal management can reduce GHG emissions, though magnitude may be modest. For instance, reducing herd replacement rates, improving reproductive fertility, and lowering age at first calving have been associated with a reduction in emission intensity per unit of milk produced (Merbold et al., 2023). Changes in herd genetic composition are also a GHG-MI; for example, crossbreeding toward high-yielding potential can substantially reduce enteric emissions, both absolute and intensity (Park et al., 2025). Evidence from Kenya shows that some GHG-MI can close milk yield gaps by 39 to 49% across agro-ecological zones, while reducing GHG emission intensities (Arenas-Calle et al., 2024).

Integration of the GHG-MIs varies between production systems. In intensive systems, animal-related emissions account for up to 85% of total emissions. In these intensive systems, improved feeding and targeted manure and enteric mitigation are important GHG-MIs (Dida et al., 2025; Mottet et al., 2017).

The wide variations in the extent of GHG reduction across different GHG-MIs point to variable responsiveness to climate-smart livestock practices. Although integration of GHG-MI in smallholder dairy farms can achieve climate change mitigation goals, empirical evidence is lacking on how farmers bundle these various interventions on their dairy farms. Most studies focus on biophysical emission quantification or single-intervention assessments, with limited attention to underlying heterogeneity in bundling GHG-MI recommended for adoption (Korir et al., 2023; Caulfield et al., 2023); yet, it is critical for designing targeted, effective interventions (Mwaura et al., 2024). This study aims to close this knowledge gap by examining the extent of heterogeneity in the integration of GHG-MI among smallholder dairy farms in rural and peri-urban wards around Nakuru city, Kenya.

2 Materials and methods

2.1 Study design

The study was an analytical observational design. It was implemented through a cross-sectional random farm survey to collect primary data needed to test the research hypothesis that integration of GHG-MI is heterogeneous in smallholder dairy farms. The data needed included both quantitative and qualitative data from a random sample farm, to align with the conditions for using an analytical observational design (Kothari, 2004).

2.2 Study area

The study was conducted in Nakuru County (Figure 1), one of the 47 counties in Kenya. The county has a land area of approximately 7,500 km² located between Longitudes 35.41 ° East or 35 ° 24′ 36″ East and 36.6 ° East or 36 °36′ 0″ East and Latitude 0.23° North or 0 ° 13′ 48” North and 1.16 ° South or 1° 9`36” South. The altitude ranges from 1,400 to 2,970 meters above sea level (m.a.s.l), with average annual precipitation exceeding 1,400 mm in the highlands and approximately 500 mm in the semi-arid lands (County Government of Nakuru, 2018, 2023).

Figure 1

The county is home to Nakuru, the fourth-largest city in Kenya, with a population of 2.35 million people, of which 51.3% live in urban centers (County Government of Nakuru, 2018; KNBS, 2019). This county is Kenya’s third-leading dairy producer, with urbanization as a significant driver of intensification and market orientation in dairy farming. As urbanization expands, farmlands are diminishing, resulting in high stocking densities of livestock, which farmers have to maintain through heavy purchases of external feeds and other inputs (Braamhaar et al., 2025). The county government has responded to this development trend by prioritizing improved productivity, incomes, and environmental sustainability in dairy farming through the promotion of a range of GHG-MI. This study targeted rural and peri-urban wards around Nakuru city, selected because the resulting interactions between urbanization and dairy production practices have implications for developing low-emission dairy production systems.

2.3 Sampling size and the sampling process

The target population was smallholder dairy farms in rural and peri-urban wards around Nakuru city. The minimum number of dairy farms needed in the survey was estimated to enable distinguishing between rural and peri-urban wards, with a difference of 20% in herd size (TLU) at a 95% confidence interval for an 83.7% coefficient of variation observed (TLU 8.6 ± 7.2) observed in Nakuru County (Braamhaar et al., 2025). The minimum number of dairy farms to be surveyed was estimated to be 136 in each of these areas (rural and peri-urban), by applying Poate and Daplyn (1993)‘s equation:

Where = 1.96 for a 95% confidence interval, was the 83.7% coefficient of variation (0.837), and was the level of difference (0.20).

Four peri-urban wards were selected from Nakuru town West and Nakuru town East electoral constituencies, while four rural wards were selected from constituencies that neighbor the two town constituencies. The selection of the wards aligned with the administrative boundaries defined in the Nakuru County Integrated Development Plan 2023–2027 (County Government of Nakuru, 2023). The total 272 dairy farms were spread proportionately to the four wards in rural and peri-urban areas.

A dairy farm was the sampling unit, which was randomly selected along a transect line drawn between a pair of landmarks. These were permanent features, such as a dairy cooperative facility, trading center, school, church, or big hotel. The field enumerator drew a transect line between a pair of landmarks, then sampled every fifth farm on the right and on the left for interview and observation.

2.4 Collecting data on GHG emission mitigating practices

Enumerators were guided to the farms by local livestock extension officers to administer a structured questionnaire. The interest was on the frequency of practicing GHG-MI. A list of GHG-MI was compiled from literature sources. Preference was given to GHG-MI reported as being effective in dairy systems from simulation modeling studies. From the compiled list, GHG-MIs were grouped into feeding, feed additives, breeding, animal health, animal welfare, and manure management. In the farm survey, the frequency of practicing a specific GHG-MI in each dairy farm was measured on a 0 to 5-point Likert scale: 5 = very frequently, 4 = frequently, 3 = occasionally, 2 = rarely, 1 = very rarely, and 0 = never used.

2.5 Analysis

The data was entered in Excel, cleaned, verified, then exported to SPSS for statistical analysis, using Principal Component Analysis (PCA) and Cluster Analysis (CA) to determine the extent of heterogeneity in GHG-MI integration among smallholder dairy farms. Before PCA, data standardization was implicitly achieved through the use of a correlation matrix in SPSS, which is equivalent to performing z-score transformation on the variables. This approach is appropriate for this dataset since the GHG-MI practices were measured on a uniform 0–5 Likert scale across all variables. Using the correlation matrix ensured that each variable contributed equally to the extracted components, preventing variables with higher variance from dominating the analysis and producing unbiased and comparable factor loadings across all GHG-MI practices.

Subsequently, a run of PCA with Varimax rotation was performed to reduce dimensionality by transforming correlated variables into a smaller number of uncorrelated Principal Components (PCs), each representing an independent dimension. Components with eigenvalues greater than one (based on Kaiser–Guttmann criterion) that explained the highest variance and factor loadings greater than 0.5 were used for subsequent Cluster Analysis to generate relatively heterogeneous clusters of farms. This hierarchical clustering pre-grouped the farms based on similarity, followed by K-means clustering to refine the cluster membership and maximize internal similarity. Distinct cluster typologies were derived using the resulting PC1s from each set of GHG-MI in hierarchical and K-means clustering.

Each cluster represented a distinct typology and was the basis for identifying the cluster heterogeneity in GHG-MI integration among farms. Cross-tabulation analysis was performed to examine associations between GHG mitigation intervention (GHG-MI) practices and farm cluster membership. The Pearson Chi-Square (χ²) with p < 0.05 significance was used to test for significant associations between practice adoption and farm clusters.

3 Results

3.1 Improved feeding and use of feed additives

With PCA fitted for the improved feeding and use of feed additives, the Kaiser–Meyer–Olkin measure (0.718) indicated sampling adequacy and data sufficiently suitable for principal component analysis. Bartlett’s test of sphericity (Chi-square = 878.061, df = 91, p = 0.000) showed that the correlation matrix was not an identity matrix, indicating meaningful component structure within the data. Five principal components were extracted explaining 62.6% of the total variance, with principal component 1 accounting for 22.83% of the variance (Table 1).

In Table 2, rotated factor loadings are associated with the specific feeding and feed additive practices on each principal component. All feeding practices loaded positively on associated PCs, indicating mutually reinforcing feeding practices. Feeding practices loading on PC1 indicated that farms improved feed quality (0.827) by feeding balanced rations (0.840) and sweet potato vine silage (0.559). Feeding practice loading on PC2 indicated that fermented feeds (0.787) are fed together with additives, such as nitrates (0.661) and molasses (0.689). Factor loadings on PC3 suggest that concentrate feeding (0.802) was practiced when feeding high-quality forage (0.793) and early harvested grass (0.699). Factor loadings on PC4 indicated that feeding high-quality hay (0.743) was supplemented with oil cakes (0.582), while the factor loading on PC5 represents a unique feeding practice, vaccinating against rumen microbes.

3.1.1 Improved herd and breeding management

The Kaiser–Meyer–Olkin measure (0.741) for the improved herd and breeding practices indicated sampling adequacy and suitability for Principal Component Analysis (Table 3). The Bartlett’s test of sphericity (Chi-square = 557.272, df = 21, p = 0.000) provided strong confirmation that the correlation matrix differed substantially from an identity matrix and that meaningful component structure existed within the data. Two principal components with eigenvalues greater than one were extracted, collectively explaining 59.1% of the total variance, of which 41.2% of the variance was explained by Principal Component 1.

Rotated factor loadings on the two PCs are presented in Table 4. Practices loading on Principal Component 1 suggested that farmers culling for low performance (0.744) do select for high growth rate (0.804), high milk yield (0.728), and longevity (0.797). This pattern is indicative of breeding for high-performance efficiency. Practices loading on PC2 suggested that selection for high feed use efficiency (0.779) is associated with first insemination before 24 months of age (0.792).

3.1.2 Improved animal health and welfare

For animal health and welfare practices fitted to PCA, the Kaiser–Meyer–Olkin measure (0.630) indicated sampling adequacy for principal component analysis (Table 5). Bartlett’s test of sphericity (Chi-square = 259.780, df = 21, p = 0.000) indicated a non-identity correlation matrix, justifying component extraction. Three principal components with eigenvalues greater than one collectively explained 64.1% of the total variance, of which 31.3% was from Principal Component 1.

In Table 6, all rotated factor loadings are positive across the extracted three principal components. Practices loading on PC1 suggested that provision of veterinary services (0.731) is reinforced by ensuring water access (0.706) and regular deworming (0.596), which aligns with the integration of good animal health and welfare. Practices loading on PC2 indicated that animal welfare practices of providing animals with shade (0.816) and good ventilation (0.757), while practices loading on PC3 indicated that integrating animal health (routine vaccination, 0.754) with animal welfare (provision of bedding, 0.760).

3.1.3 Improved manure management

The improved manure management practices fitted to PCA (Table 7) had a Kaiser–Meyer–Olkin measure (0.545), indicating that sampling adequacy was relatively modest for principal component analysis. This suggests that the partial correlations among some variables were comparatively large, likely reflecting the diversity and heterogeneity of manure management practices among the sample farms. Bartlett’s test of sphericity (Chi-square = 466.708, df = 36, p = 0.000) indicated a non-identity correlation matrix and meaningful component structure within the data. Three principal components were extracted with eigenvalues greater than one; collectively, they explained 58.5% of the total variance, of which 21.1% is from PC1.

The rotated factor loadings on PC1 presented in Table 8 indicated that good and poor manure practices were found together on a farm. Dry lot manure storage (0.849) and composting manure storage (0.562) are good manure management practices and were practiced together with poor management of uncovered manure storage (0.769). Practices loading on Principal Component 2 suggest that farms burning manure (0.794) or having a solid lagoon (0.790) were less likely to spread manure on land (−0.623). Practices loading on PC3 suggested a pattern that liquid or slurry systems (0.860) were associated with biogas use (0.857).

3.1.4 Typologies of heterogeneity in GHG-mitigating interventions

The extracted PC1 from each GHG-MI, improved feeding and use of feed additives, breeding, herd, health and welfare, and manure management, proceeded to Cluster Analysis to derive distinct clusters of GHG-MI farms. Cluster Analysis returned three clusters: cluster one and two each comprising 36.8% (n = 107) and cluster three 26.4% (n = 77) of farms. To enable easier interpretation, the frequency of practicing a specific GHG-MI was collapsed into two: frequent practice (5 = very frequently, 4 = frequently) and rare practice (3 = occasionally, 2 = rarely, 1 = very rarely, and 0 = never used). Based on the number of GHG-MIs frequently practiced by at least half of the farms, the clusters were labeled low mitigating (cluster three), average mitigating (cluster one), and high mitigating (cluster two). Frequent practice of most of the GHG-MI was observed among high-mitigating farms, followed by the average-mitigating farms and low-mitigating farms. For each GHG-MI cluster, results are presented on the frequency of farms practicing specific GHG-MI.

First, Table 9 presents a description of the sample farms by the three GHG-MI clusters. Overall, women were only 3.8% more than men, but not different between the clusters (p = 0.720). Education level was higher among average- and low-mitigating farms, with those having attained at least secondary level education more (16.8 to 21.9%) than in the high-mitigating farms (p = 0.011). Although cluster difference was not observed in dairy intensification (p = 0.398), high- and average-mitigating farms had more (7 to 13%) intensified dairy farms relative to low-mitigating farms.

Table 10 shows the percentage distribution of farms within the three clusters by the frequency of integration of improved feeding practices. A total of 11 feeding practices were examined, of which four were frequently practiced by at least 6 in 10 (63 to 73%) of high-mitigating farms, while only two practices were frequent with at least 5 in 10 average- and low-mitigating farms. Feeding high-quality forage (51%) supplemented with concentrates (64%) was frequent among high-mitigating farms, while harvesting grass early (55.5%) to feed high-quality forage (62.3%) was frequent among low-mitigating farms. Regardless of the mitigation cluster label, only a few farms frequently fed fermented feeds and nitrate additives (<8%), oil cakes, sweet potato vines and molasses (11 to 19%), and feeding quality hay (14 to 31%).

Table 11 summarizes cluster farms by frequency of improved herd and breeding practices on the farm. Six practices were examined, of which five related to trait selection were frequently practiced by at least 8 in 10 (84 to 99%) of high mitigating farms compared to only two traits by at least 5 in 10 of the average- and low-mitigating farms. These two practices were feed use efficiency (53 to 54%) and milk yield (53 to 67%). While at least 9 in 10 (91.6%) high-mitigating farms frequently culled for low performance, inseminating heifers by 24 months of age was a rare practice (16 to 29%) across the three clusters.

Table 12 is the results of sample farms within clusters by frequency of practicing good animal health and welfare management. Seven practices were examined, of which all were frequently practiced by at least 6 in 10 (64 to 97%) of high-mitigating farms. Six of the practices were frequent among at least 5 in 10 (51 to 100%) of average mitigating farms, and four practices were frequent among at least 6 in 10 (55 to 92%) of low-mitigating farms.

In Table 13, the distribution of farms within clusters by frequency of practicing manure management is presented. Eight practices were examined, of which spreading manure on land was the most frequently practiced, but fewer among average-mitigating farms (54.2%) than (p = 0.0010) among low-mitigating farms (77.9%) or high-mitigating farms (95.3%). The next most frequent practice was uncovered manure storage, observed more frequently among average mitigating farms (79.4%) than (p = 0.001) among low-mitigating farms (63.6%). Dry lot manure storage, a good practice for mitigating GHG emissions, was only a frequent practice among average mitigating farms. Other good manure management practices for mitigating emissions were rare (<10%), including biogas, composting, and solid lagoon. Manure composting was only frequent in average mitigating farms (24.3%).

4 Discussion

This study aimed to close the knowledge gap on understanding heterogeneity in the integration of GHG-MI among smallholder dairy farms. A representative sample of rural and peri-urban farms from wards around Nakuru city, Kenya, was used. Heterogeneity was ascertained, with three farm typologies of GHG-MI integration derived, which point to varying responsiveness to climate-smart livestock practices. This observed heterogeneity implies that farmers perceive and value a portfolio of recommended GHG-MI (improved feeding, herd and breeding management, animal health and welfare, and manure management) differently. This could stem from high adoption costs, inadequate extension support, or unaffordable climate finance credit (Akzar et al., 2023). Farmers faced with these barriers are more likely to place higher value on immediate tangible benefits than on emission reduction potential, which they cannot quantify (Vernooij et al., 2024). Some workers have found this to be a reason that also slows down adoption of GHG-MI among smallholder farmers (Maindi et al., 2020; Mburu et al., 2024). For policy, heterogeneity in integration of GHG-MI implies that supporting the transition to a low-emission dairy sector needs a shift from generic recommendations to targeted interventions that account for producer diversity (Adenaeuer et al., 2024; Korir et al., 2023; Vogel and Beber, 2022; Chelang’a et al., 2025c).

The observed heterogeneity aligns with literature that has observed that variation in technology adoption and management is typical in smallholder systems (Arndt et al., 2022; Huber et al., 2023). The heterogeneity reflects social and spatial diversity in response to climate-smart interventions (Musafiri et al., 2020). In Brazil, Vogel and Beber (2022) identified four farm groups with carbon footprints ranging from 1.75 to 3.27 kg CO₂ eq per kg of fat- and protein-corrected milk. Chen et al. (2026) showed that the potential of herd management strategies for GHG mitigation depended on both the strategy chosen and the initial productivity of the herd.

Furthermore, integration of a portfolio of GHG-MI aligns with livestock sustainability frameworks that advocate for bundled interventions to achieve environmental, economic, and societal benefits (Díaz de Otálora et al., 2024; Balcha et al., 2023). For instance, Oduor et al. (2025) observed only positive correlations among climate-smart agriculture practices in smallholder dairy systems, indicating synergistic complementarities rather than substitutability. Farmers tend to bundle mutually reinforcing interventions to improve overall farm performance. Chen et al. (2026) demonstrated that combining health, feeding, and reproductive management strategies produced mutually reinforcing effects on both emission reduction and farm gross margin, without detectable trade-off between mitigation and profitability. These complementarities reinforce the case for promoting bundles of GHG-MI rather than isolated interventions and have implications for extension services and development agencies, which need to consider promoting bundles of GHG-MI practices rather than isolated practices.

In this study, the majority of the farms (63%) only frequently integrated a few GHG-MI, indicating a huge opportunity to tap unrealized gains in promoting low-emission dairy systems. Targeted support may be relevant to these low- and average-mitigating farms to unlock emission reduction gains while sustaining productivity and livelihoods.

Importantly, GHG-MI integration did not increase with intensification or education levels, suggesting that factors other than farm and farmer attributes may be at play (Arndt et al., 2022; Huber et al., 2023). Differences in social, institutional, and economic factors likely do explain why farmers with similar intensification and education levels differ markedly in GHG-MI bundling intensity. Cooperative membership, social capital, entrepreneurial orientation, extension services, access to early warning systems, and herd size have been reported to significantly increase adoption of climate-smart practices (Chelang’a et al., 2025c; Oduor et al., 2025; Mamo and Yayeh, 2023). Collectively, these findings suggest that improving access to inputs, credit, extension, and farmer networks can influence the scaling of climate-smart practices more than education or intensification alone. Furthermore, the findings have implications for climate change mitigation policy promoting low-emission dairy need not be tied to intensification.

Farms with high mitigation had integrated more of feeding and breeding practices, which can be associated with increased productivity that subsequently reduces emission intensities. Improved feed digestibility and nutritional planning have been associated with lower emission intensity in dairy systems. Arndt et al. (2022) and Ayemele et al. (2025) attained a 33% decrease in enteric methane emissions when feeding mineral-based feed additives, 58% methane reduction with up to 4% feed additive dosage, and a 39% decrease with total mixed ration systems than partial mixed ration systems. However, at present, the frequent use of feed additives (such as nitrates, molasses, and fermented feeds) was consistently low. This could be pointing to a lack of appropriate technical knowledge and high market prices associated with the use of these GHG-MI (Branca and Perelli, 2020).

In the studies by Chelang’a et al. (2025b) and Mamo and Yayeh (2023), farmers ranked feed formulation as a highly preferred climate-smart dairy strategy, but uptake of feed supplements remained low due to high market prices. Adoption of GHG-MI has been significantly associated with awareness of productivity benefits, especially higher milk yields and lower feed costs, suggesting that extension programs should emphasize economic gains alongside climate benefits (Chelang’a et al., 2025a). Similarly, Vogel and Beber (2022) reported that sustainable intensification programs in Brazil prioritize feeding and herd management improvements before more technically and financially demanding interventions. This is relevant to the present sample of dairy farms.

The breeding practices that were frequently integrated were selection for feed use efficiency, milk yield, growth rate, longevity, and culling for low performance. This selection for a number of production and functional traits suggests that dairy farms implement a selection index for improved performance, which is important in reducing emission intensity. With high-yielding potential, the same volume of milk can be produced with fewer cows, consequently contributing to lowering emissions, both absolute and intensity. Graham et al. (2024) has estimated that breeding practices can reduce emission intensity by up to 27.4%, though variation is large (3.9 to 27.4%) depending on the intervention package applied.

Although improved dairy breeding practices are expected to support increasing intensification (Brito et al., 2021), this was not true in the present study. Instead, dairy intensification showed no pattern with integration of GHG-MI. Achieving early age at first insemination and calving has implications for replacement rate in the herd. In the present study, early age at insemination was a rare practice. This points to a challenge in herd fertility management, including limited breeding services, poor heat detection, and inadequate extension support (O’Shea et al., 2021; Korir et al., 2023). In the central highlands of Kenya, Oduor et al. (2025) observed high overall adoption of improved breeding practices, but performance in early insemination and fertility management remained weak. However, some authors have observed modest adoption of artificial insemination in central Kenya highlands due to access and service constraints (Chelang’a et al., 2025b). Farmers express strong preference for high-yielding sexed semen, but actual adoption remains marginal.

Good animal health and welfare are supportive of lowering emission intensity through achieved high performance and reduced morbidity and mortality (Food and Agriculture Organization of the United Nations, 2019). Veterinary interventions targeted to endemic diseases have been associated with 7 to 24% reduction in herd-level GHG intensity through low replacement needs in the herd, which results in lower absolute emissions (Statham et al., 2020). Animal health management was the most preferred climate-smart dairy strategy, with deworming and tick control ranked highest due to their visible effects on milk yield (Chelang’a et al., 2025b). In contrast, vaccination and curative measures were less preferred because their production benefits were less apparent. Strategic vaccination is beneficial if informed by access to early warning systems and health practices (Oduor et al., 2025).

Provision of shade, clean drinking water, and temperature regulation enhance good animal welfare status. In this study, these were frequent practices regardless of intensity of mitigation. Mottet et al. (2017) observed that practicing good animal welfare can achieve up to 10% reduction in absolute GHG emissions.

However, provision of bedding was a rare practice, probably because farmers fail to consider this as a good animal welfare practice that can enhance productivity (Khoza et al., 2020). Rare provision of bedding may reflect awareness and implementation gaps. While Mamo and Yayeh (2023) identified low awareness as a key barrier to adopting climate-smart livestock practices, Chelang’a et al. (2025a) observed that farmers’ confidence in implementing practices significantly increased GHG reduction uptake. These findings suggest a need for extension to raise awareness that provision of bedding is an important animal welfare practice. It has an indirect association with improved health and performance, and hence potential to reduce emission intensity.

Good manure management, such as biogas and slurry systems, was rare across all clusters, except dry-lot manure storage, which was only frequent among average mitigating farms. Frequent manure management was spreading manure on land and uncovered manure. This reveals low integration of more effective manure management practices for GHG mitigation. The situation could be related to several barriers, including structural, institutional, and investment capacity of farmers. Jesse et al. (2020) identified high capital investment requirements, limited technical maintenance capacity, and weak policy support as barriers constraining the integration of biogas technologies in smallholder systems. These barriers are widely documented across East Africa. Oduor et al. (2025) observed that only 2% of dairy farms were using biogas, which concurs with observations by Mamo and Yayeh (2023) in Ethiopia, which they explained is due to limited awareness and finance.

Dairy intensification creates more opportunities for effective manure collection and hence potential for emission mitigation (Food and Agriculture Organization of the United Nations, 2019; Wilkes et al., 2020; Oduor et al., 2025). In zero-grazing, keeping cows in confinement and floored housing eases the collection of urine and dung, allowing waste to be directed into bio-digesters or separation systems. With this manure management, it is possible to cut down methane emissions by 35% (Dida et al., 2025). In the present study, less than 3 in 10 farms had zero-grazed dairy units, implying missing out on a huge opportunity to reduce emissions through dairy intensification. Although zero-grazing presents good opportunity for improving manure management, it increases labor demands, feed costs, and gendered burdens.

4.1 Policy and practical implications for sustainable smallholder dairy systems

The observed heterogeneity in the integration of GHG-MI suggests that low-emission dairy policies should account for differences among smallholder dairy farm typologies. Uniform approaches may be less effective than targeted strategies that reflect farmers’ existing levels of mitigation practice adoption.

Although fewer (36%) farms demonstrated being highly responsive to low-emission practices, they provide evidence that smallholder dairy farms can be relied upon to support implementation of the nationally determined commitments in the Paris Agreement for GHG emission reduction. These farms have integrated more feeding practices and selection for production efficiency. However, the GHG-MI they have integrated shows a need to include feed additives and good manure-handling practices, such as biogas and slurry, to unlock the manure management gains.

The large proportion (64%) of farms with average-to-low response to low-emission practices presents huge opportunity to bolster response to low-emission practices in smallholder dairy systems. Low integration of GHG-MI limits their potential to contribute to emission reduction goals. It is necessary for extension service to identify these farms for targeted capacity building that demonstrates the synergies between feeding, breeding, and health practices. Use of a living lab set up is an option to instigate co-innovation capacity on GHG-MI, so that these farmers actively participate in emission reduction goals. In the living lab, they would co-learn and co-innovate around integrating feed additives and effective manure management as entry points to low-emission dairy systems.

With few smallholder farms (36%) actively embracing GHG-MI, policy direction needs to consider how to enable access to climate finance for smallholder dairy farmers effectively. Climate finance needs to be packaged as bundled GHG-MI that go to reduce investment and knowledge barriers. Living labs, together with farmer cooperatives and demonstration-based learning platforms, may provide effective channels for disseminating these bundled approaches. These approaches also encourage consideration of farmer preferences, needs, and priorities in the co-design of GHG-MI with high potential to deliver multiple, immediate, and visible productivity benefits. Promising candidates for GHG-MI include deworming, feed formulation, and water access, biogas and vaccination, and dry lot manure handling.

4.2 Limitations and recommendations for future research

Because of poor practice in recording performance on smallholder farms, the bulk of the data were self-reported and recalled. This introduces reporting biases, despite our efforts to complement with on-farm observational assessments and independent validation of GHG-MI on the farm. The study examined the integration of GHG-MI recommended in the literature for achieving low-emission dairy production, but did not aim further to validate reported estimates of greenhouse gas emissions reduction. Rather, the aim was to examine the extent to which the recommended GHG-MI are being integrated to lead to GHG emission reduction.

The cross-sectional nature of the study also limits causal inference regarding the factors associated with GHG-MI integration. Longitudinal studies would provide greater insight into adoption dynamics over time. Further research is also needed to evaluate the economic feasibility of different GHG-MI bundles across farm typologies, including assessments of costs, benefits, and returns to investment. Given the potential influence of gender on technology adoption and farm decision-making, future studies could examine gender-differentiated patterns of GHG-MI uptake. Finally, the role of farmer living labs, cooperatives, service providers, and other value-chain actors in facilitating the scaling of GHG-MI for low-emission dairy systems warrants further investigation.

References

• 1

  AdenaeuerL.BreenJ.HaydenA. (2021). Insights in overcoming the non-adoption of voluntary agricultural GHG mitigation measures in Ireland. Econ. Agro Aliment.22, 1–26. doi: 10.3280/ECAG3-2020OA11042

  • CrossRef
  • Google Scholar

• 2

  AdenaeuerL.SidhoumA. A.SipiläinenT.KlingV.NiemiJ. K. (2024). Finnish farmers' adoption of feed additives for GHG mitigation in dairy farming: barriers and perspectives. EuroChoices24, 63–71. doi: 10.1111/1746-692X.12472,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 3

  AkzarR.UmbergerW.PeraltaA. (2023). Understanding heterogeneity in technology adoption among Indonesian smallholder dairy farmers. Agribusiness39, 347–370. doi: 10.1002/agr.21782

  • CrossRef
  • Google Scholar

• 4

  Arenas-CalleL.JenningsS.ChallinorA. (2024). Agro-technology for climate-smart agriculture and resilience to climate extremes in sub-Saharan Africa. Environmental Research: Food Syst., 1:21001. doi: 10.1088/2976-601X/ad50d9

  • CrossRef
  • Google Scholar

• 5

  ArndtC.HristovA. N.PriceW. J.McClellandS. C.PelaezA. M.CuevaS. F.et al. (2022). Full adoption of the most effective strategies to mitigate methane emissions by ruminants can help meet the 1.5 °C target by 2030 but not 2050. Proc. Natl. Acad. Sci. U. S. A.119:e2111294119. doi: 10.1073/pnas.2111294119

  • CrossRef
  • Google Scholar

• 6

  AyemeleA. G.Ngueuyim NonoJ. K.FotsidieH. G.NgabaM. J. Y.TilahunM.MaL.et al. (2025). A meta-analysis of feeding strategies for dairy cattle performance and greenhouse gas mitigation across the world. Resour. Environ. Sustain.22:100271. doi: 10.1016/j.resenv.2025.100271

  • CrossRef
  • Google Scholar

• 7

  BalchaE.MenghistuH. T.ZenebeA.TeferiT.HadushB. (2023). Climate-smart agricultural practices: a case of dairy cooperative farmers in Agula and Maychew, northern Ethiopia. Carbon Manag.14:1880. doi: 10.1080/17583004.2023.2271880

  • CrossRef
  • Google Scholar

• 8

  BeaucheminK. A.UngerfeldE. M.AbdallaA. L.AlvarezC.ArndtC. (2022). Current enteric methane mitigation options. J. Dairy Sci.105, 9297–9326. doi: 10.3168/jds.2022-22508,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 9

  BraamhaarD. J. M.Van der LeeJ.BebeB. O.OostingS. J. (2025). From rural to urban: exploring livestock farming practices in urbanizing landscapes. Agric. Syst.225:104297. doi: 10.1016/j.agsy.2025.104297

  • CrossRef
  • Google Scholar

• 10

  BrancaG.PerelliC. (2020). 'Clearing the air': common drivers of climate-smart smallholder food production in eastern and southern Africa. J. Clean. Prod.270:121900. doi: 10.1016/j.jclepro.2020.121900

  • CrossRef
  • Google Scholar

• 11

  BritoL. F.BédèreN.DouhardF.Rojas de OliveiraH.ArnalM.PeñagaricanoF.et al. (2021). Review: genetic selection of high-yielding dairy cattle toward sustainable farming systems in a rapidly changing world. Animal15:100292. doi: 10.1016/j.animal.2021.100292

  • CrossRef
  • Google Scholar

• 12

  CaulfieldM. E.GrahamM.GibbonsJ.McNicolL.WilliamsP.ChadwickD.et al. (2023). Marginal abatement cost curves of climate-smart agricultural practices to mitigate greenhouse gas emissions from smallholder dairy farms in Kenya. Res. Square2023:9239. doi: 10.21203/rs.3.rs-3179239/v1

  • CrossRef
  • Google Scholar

• 13

  Chelang’aN. C.MathengeM.OtienoD. O.SassiM. (2025a). The determinants of greenhouse gas reduction levels among smallholder farmers: insights from the adoption of climate-smart dairy strategies in Central Kenya. Front. Clim.7:1593584. doi: 10.3389/fclim.2025.1593584

  • CrossRef
  • Google Scholar

• 14

  Chelang’aN. C.MathengeM.OtienoD. O.SassiM. (2025b). Smallholder farmers' preferences for climate-smart dairy strategies in Central Kenya. NJAS97:2533955. doi: 10.1080/27685241.2025.2533955

  • CrossRef
  • Google Scholar

• 15

  Chelang’aN. C.MathengeM.OtienoD. O.SassiM. (2025c). Promoting sustainable dairy production amid climate change: adoption of climate-smart dairy strategies and welfare effects on farmers in Central Kenya. Front. Sustain. Food Syst.9:610. doi: 10.3389/fsufs.2025.1598610

  • CrossRef
  • Google Scholar

• 16

  ChenL.ThorupV. M.ØstergaardS. (2026). Herd management strategies for greenhouse gas emission mitigation in dairy production: a simulation study. J. Dairy Sci.109, 5334–5346. doi: 10.3168/jds.2025-26699,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 17

  County Government of Nakuru. (2018). County Integrated Development Plan (CIDP) 2018–2022. Available online at: https://maarifa.cog.go.ke/sites/default/files/2022-08/CIDP%20-%20Nakuru%20-%202018%20-%202022.pdf

  • Google Scholar

• 18

  County Government of Nakuru. (2023). Nakuru City County Integrated Development Plan 2023–2027. Available online at: https://nakuru.go.ke/wp-content/uploads/2024/05/Nakuru-CIDP-2023-2027.pdf.

  • Google Scholar

• 19

  Díaz de OtáloraX.del PradoA.DragoniF.BalaineL.PardoG.WiniwarterW.et al. (2024). Modelling the effect of context-specific greenhouse gas and nitrogen emission mitigation options in key European dairy farming systems. Agron. Sustain. Dev.44:940. doi: 10.1007/s13593-023-00940-6

  • CrossRef
  • Google Scholar

• 20

  DidaM. F.GarciaS. C.GonzalezL. A. (2025). Greenhouse gas emissions of confinement and pasture-based dairy farms: implications for mitigation. J. Dairy Sci.108, 11026–11040. doi: 10.3168/jds.2025-26566,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 21

  Food and Agriculture Organization of the United Nations. (2017). Global Livestock Environmental Assessment Model (GLEAM 2.0) — Assessment of Greenhouse gas Emissions and Mitigation Options. FAO. Available online at: https://www.fao.org/gleam/en/.

  • Google Scholar

• 22

  Food and Agriculture Organization of the United Nations. (2019). Animal Health and Greenhouse gas Emissions Intensity. FAO Animal Production and Health Working Paper No. 19. Available online at: https://www.fao.org/animal-health/en.

  • Google Scholar

• 23

  GrahamM. W.ÖzkanŞ.ArndtC.González-QuinteroR.KorirD.MerboldL.et al. (2024). Toward compatibility with national dairy production and climate goals through locally appropriate mitigation interventions in Kenya. Agricultural Systems, 214:104098. doi: 10.1016/j.agsy.2024.104098

  • CrossRef
  • Google Scholar

• 24

  HakuzimanaJ.HabimanaJ. D.MunyamboneraJ. P. D. (2021). The carbon footprint of smallholder dairy farming in sub-Saharan Africa: a review. Int. J. Prog. Sci. Technol.25, 478–493. doi: 10.52155/ijpsat.v25.1.2737

  • CrossRef
  • Google Scholar

• 25

  HuberR.TarruellaM.SchäferD.FingerR. (2023). Marginal climate change abatement costs in Swiss dairy production considering farm heterogeneity and interaction effects. Agric. Syst.207:103639. doi: 10.1016/j.agsy.2023.103639

  • CrossRef
  • Google Scholar

• 26

  JesseO. O.DanielO.ShemO. W.AsaahN. (2020). Constraints limiting the improvement of manure management as climate smart technology for smallholder dairy farmers. Afr. J. Agric. Res.16, 1155–1116. doi: 10.5897/AJAR2020.15018

  • CrossRef
  • Google Scholar

• 27

  KalsonA.AhmadQ. N.VarshaKhaiperM.MamtaK.PareekJ.et al (2025). Effect of Manuring and Biofertilization on NPK Content, Uptake and Post-Harvest Soil Nutrient Availability in Mung Bean (Vigna radiata). Asian Journal of Soil Science and Plant Nutrition.11, 35–43. doi: 10.9734/ajsspn/2025/v11i2508

  • CrossRef
  • Google Scholar

• 28

  KNBS. (2019). Kenya Population and Housing Census Volume IV: Distribution of Population by Socio-Economic Characteristics. Kenya National Bureau of Statistics. Available online at: https://www.knbs.or.ke/wp-content/uploads/2023/09/2019-Kenya-population-and-Housing-Census-Volume-4-Distribution-of-Population-by-Socio-Economic-Characteristics.pdf

  • Google Scholar

• 29

  KavanaghI.FentonO.HealyM. G.BurchillW.LaniganG. J.KrolD. J. (2021). Mitigating ammonia and greenhouse gas emissions from stored cattle slurry using agricultural waste, commercially available products and a chemical acidifier. Journal of Cleaner Production, 320:126251. doi: 10.1016/j.jclepro.2021.126251

  • CrossRef
  • Google Scholar

• 30

  Khatri-ChhetriA.WilkesA.OdhongC.Mitigation Options and Finance for transition to low-Emissions Dairy in Kenya CCAFS Working Paper no. 329. Wageningen, the Netherlands, CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS) (2020). Available online at: https://hdl.handle.net/10568/110568.

  • Google Scholar

• 31

  KhozaS.De BeerL. T.Van NiekerkD.NemakondeL. (2020). A gender-differentiated analysis of climate-smart agriculture adoption by smallholder farmers: application of the extended technology acceptance model. Gend. Technol. Dev.25, 1–21. doi: 10.1080/09718524.2020.1830338,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 32

  KorirL.ManningL.MooreH. L.LindahlJ. F.GemechuG. (2023). Adoption of dairy technologies in smallholder dairy farms in Ethiopia. Front. Sustain. Food Syst.7:1070349. doi: 10.3389/fsufs.2023.1070349

  • CrossRef
  • Google Scholar

• 33

  KothariC. R. (2004). Research Methodology: Methods and Techniques. 2nd Edn New Delhi: New Age International Publishers.

  • Google Scholar

• 34

  LiJ.JiangR.TangX. (2024). Assessing psychological factors on farmers’ intention to apply organic manure: An application of extended theory of planned behavior. Environment Development and Sustainability, 26, 2467–2491. doi: 10.1007/s10668-022-02829-y

  • CrossRef
  • Google Scholar

• 35

  MaindiC.OsugaI.GichehaM. (2020). Advancing climate smart agriculture: adoption potential of multiple on-farm dairy production strategies among farmers in Murang’a County, Kenya. Livestock Research for Rural Development.32:63. Available online at: https://www.researchgate.net/publication/340377132_Advancing_climate_smart_agriculture_adoption_potential_of_multiple_on-farm_dairy_production_strategies_among_farmers_in_Murang’a_County_Kenya

  • Google Scholar

• 36

  MamoB.YayehD. (2023). Smallholder farmer’s adoption of climate smart livestock production: practices, status and determinants in Hidebu Abote Woreda, Central Ethiopia. FARA Res. Rep.7, 545–562. doi: 10.59101/frr072343

  • CrossRef
  • Google Scholar

• 37

  MburuM. N.MburuJ.NyikalR.MugeraA.NdambiA. (2024). Assessment of socio-economic determinants and impacts of climate-smart feeding practices in the Kenyan dairy sector. Mitigation and Adaptation Strategies for Global Change, 29, 1–25. doi: 10.1007/s11027-024-10131-7

  • CrossRef
  • Google Scholar

• 38

  MerboldL.GrahamM. W.ArndtC.González-QuinteroR.KorirD.LeitnerS.et al. (2023). Multiple mitigation strategies can lead to GHG emissions reduction in Kenyan dairy systems. Presented at the 3rd one-day symposium of the Animal Task Force and the EEAP Commission on Livestock Farming Systems, Lyon, 28 August 2023. Nairobi, Kenya: ILRI. Available online at: https://hdl.handle.net/10568/135928

  • Google Scholar

• 39

  MottetA.De HaanC.FalcucciA.TempioG.OpioC.GerberP. (2017). Livestock: on our plates or eating at our table? A new analysis of the feed/food debate. Glob. Food Secur.14, 1–8. doi: 10.1016/j.gfs.2017.01.001

  • CrossRef
  • Google Scholar

• 40

  MusafiriC. M.MachariaJ. M.Ng'etichO. K.KiboiM. N.OkeyoJ.ShisanyaC. A.et al. (2020). Farming systems' typologies analysis to inform agricultural greenhouse gas emissions potential from smallholder rain-fed farms in Kenya. Sci. Afr.8:e00458. doi: 10.1016/j.sciaf.2020.e00458,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 41

  MwauraF. M.NgigiM. W.ObareG. (2024). Do smallholders have a role to play in atmospheric greenhouse gas removal? Insights from western Kenya. Sci. Afr.24:e02206. doi: 10.1016/j.sciaf.2024.e02206,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 42

  NtinyariW.Gweyi-OnyangoJ. P. (2021). “Greenhouse gases emissions in agricultural systems and climate change effects in sub-Saharan Africa,” in African Handbook of Climate change Adaptation, eds. Leal FilhoW.OgugeN.AyalD.AdelekeL.Da SilvaI. (Berlin: Springer), 1–25.

  • Google Scholar

• 43

  OduorF. O.OdhiamboM. O.NyangwesoP. M. (2025). Toxic or synergistic relationship: the interdependence and determinants of dairy climate smart agriculture practices in the north rift and western parts of Kenya. Clim. Smart Agric.2:100084. doi: 10.1016/j.csag.2025.100084

  • CrossRef
  • Google Scholar

• 44

  O’SheaR.BougardM.BreenJ.StylesD. (2021). Pathways toward inclusive low-emission dairy development in Tanzania: producer heterogeneity and implications for intervention design. Agric. Syst.190:103073. doi: 10.1016/j.agsy.2021.103073

  • CrossRef
  • Google Scholar

• 45

  OnyangoJ.VellemaS.CraneT.AtelaJ. (2024). Searching for synergies for low emission development, the case for dairy subsector in Kenya. Regional Environmental Change, 24:114. doi: 10.1007/s10113-024-02275-8

  • CrossRef
  • Google Scholar

• 46

  ParkH.ParkS.FufaA. J. (2025). A study on GHG emission mitigation scenarios for dairy farm in Ethiopia. J. Anim. Environ. Sci.27, 1–16. doi: 10.11109/JAES.2025.27.1.001

  • CrossRef
  • Google Scholar

• 47

  PoateD.DaplynP. (1993). Data for Agrarian Development. Cambridge: Cambridge University Press.

  • Google Scholar

• 48

  RotzC. A.BriggsK.HristovA. N.LeytemA.YoungE.HagevoortR.et al. (2025). Strategies for mitigating greenhouse gas emissions from US dairy farms toward a net zero goal. J. Dairy Sci.108, 9755–9777. doi: 10.3168/jds.2025-26392,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 49

  StathamJ.ScottH.StathamS.ActonJ.WilliamsA.SandarsD. (2020). Dairy cattle Health and Greenhouse gas Emissions pilot Study: Chile, Kenya and the UK. Global Research Alliance on Agricultural Greenhouse Gases, Global Dairy Platform. Available online at: https://globalresearchalliance.org/wp-content/uploads/2020/10/Dairy-Cattle-Health-and-GHG-Emissions-Pilot-Study-Report.pdf.

  • Google Scholar

• 50

  VernooijV.SchoneveldG.VellemaS.CraneT.KihoroE. (2024). From fix to fitting: connecting low-emission development with multilevel smallholder dairy practices in Kenya. NJAS96:965. doi: 10.1080/27685241.2024.2334965,

  • Pubmed Abstract
  • CrossRef
  • Google Scholar

• 51

  VogelE.BeberC. L. (2022). Carbon footprint and mitigation strategies among heterogeneous dairy farms in Paraná, Brazil. J. Clean. Prod.349:131404. doi: 10.1016/j.jclepro.2022.131404

  • CrossRef
  • Google Scholar

• 52

  WilkesA.WassieS.Odhong'C.FravalS.Van DijkS. (2020). Variation in the carbon footprint of milk production on smallholder dairy farms in central Central Kenya. J. Clean. Prod.265:121780. doi: 10.1016/j.jclepro.2020.121780

  • CrossRef
  • Google Scholar