![Farm-specific yield increases for maize. Left: bar plots showing increase in harvest yield at each farm plot with the Ombeyi cluster. Right: same but for the Wawidhi cluster. Dotted lines denote the average yield increase for each cluster (Ombeyi: 60.4±15.7 [%]; Wawidhi: 80±25.4 [%]).](https://www.researchgate.net/profile/Fatima-Haque/publication/385887254/figure/fig1/AS:11431281290818475@1731814671714/Farm-specific-yield-increases-for-maize-Left-bar-plots-showing-increase-in-harvest_Q320.jpg)
Content uploaded by Fatima Nasrin Haque
Author content
All content in this area was uploaded by Fatima Nasrin Haque on Nov 17, 2024
Content may be subject to copyright.
www.fluxcarbon.earth contact@fluxcarbon.earth
N o v e m b e r 2 0 2 4
Yield Increases for smallholder
farmers in Sub Saharan Africa via
Enhanced Rock Weathering
Preliminary results from a smallholder pilot
study in Kisumu County, Kenya
Yield Increases for Smallholder farmers in Sub
Saharan Africa via Enhanced Rock Weathering:
Preliminary results from a smallholder field trial in
Kisumu County, Kenya
November 2024
Fatima Haque, Vincent J. Clementi, Benjamin Möller, Laura Bastianini, Cavince Odhiambo, Susan
Sagina, Herine Ondolo, Collins Kechir, and Sam Davies
Abstract
Enhanced Rock Weathering (ERW) has emerged as a promising geochemical carbon dioxide removal
(CDR) technology with the potential to improve both soil health and crop yields, particularly in
smallholder farming systems. This study, conducted in a collaboration between Flux and the United
Nations Convention to Combat Desertification (UNCCD), investigates the agronomic and social
co-benefits of ERW deployments in Kisumu County, Kenya. Targeting women smallholder farmers, the
research focuses on maize cultivation in nutrient-poor soils where traditional inputs like lime and
fertilisers are often inaccessible due to cost constraints.
The study involved 56 farmers, each with divided plots—one control and one treated with volcanic rock
powder sourced from a local quarry. Our findings demonstrate significant agronomic benefits, with an
average yield increase of 71.17% ± 15.5% and an aggregate yield increase of 47.47% ± 5.73% in maize
yield on treatment plots compared with the control plots. Whilst follow-on soil samples are yet to be
taken, it is assumed that the yield increases are due to the rock powder enhancing soil nutrient
availability, improving pH balance, and supporting plant growth. These results highlight the potential of
ERW to not only sequester carbon but also positively impact agricultural productivity, offering economic
benefits to smallholder farmers in sub-Saharan Africa.
Annexes
A. List of farmers removed from study
B. Harvest and yield data
C. Complete soil analysis
D. Example Soil Analysis certificate
E. Maps of deployment
F. Accompanying Images
1
Foreword
At the time of publication, this study represents the first and only deployment of Enhanced Rock
Weathering with smallholder farmers in Africa in a real world setting (that is to say not a controlled plot
or single farm experiment). This project aimed to fit entirely within smallholder farmers normal farming
practices in the region, and to understand not only the agronomic implications of ERW for soil health,
but also the operational challenges and requirements for this environment. This comes with both
challenges, which have been highlighted in this report, and opportunities for learning. If ERW is to
become a realistic pathway to boosting yields and soil health for smallholder farmers in Africa, we must
rapidly deploy in the actual context, and at scale, to understand how it works.
In self-publishing this report as a whitepaper, it is our intention to speed up this learning cycle, generate
interest in the topic from researchers and interested parties, and generate further research projects - as
soon as possible. It is our ambition to publish as a peer-reviewed paper when final soil analysis is
complete in 2025, but that getting out in its current form now will be useful to the community.
Before publishing this whitepaper we invited a group of reviewers from various backgrounds within the
CDR and ERW community to offer feedback and comments to improve the rigour of the paper and
make it more useful for the reader. We are incredibly grateful for the detailed and thoughtful comments
from these reviewers that have helped improve this work.
All further comments and feedback are most definitely welcome and appreciated and should be
directed to contact@fluxcarbon.earth
Background and Motivation
Enhanced rock weathering (ERW) is a geochemical carbon dioxide removal (CDR) approach that has
been identified as a requisite negative emission technology to limit warming to <2 °C above
pre-industrial levels1. The weathering reaction that facilitates permanent CDR in agricultural settings can
also deliver agronomic co-benefits for soil health, plant nutrition, and crop yield, which makes ERW a
promising pathway to address current and future food insecurities2,3. This may be particularly true for
smallholder farmers, who account for an estimated 83% of the 570+ million farmers globally4, and the
surrounding communities.
Smallholder farms (<2 ha) are the dominant farm size in much of the world (particularly in South(east)
Asia, Latin America, and sub-Saharan Africa5) and play a central role in global food production6.
However, much of our understanding of the agronomic co-benefits from ERW originate from studies
that have taken place in or replicate the soils, climates, and farming practices that broadly characterise
agricultural systems in Europe, North America, and other developed nations7-10. This limits our ability to
extrapolate these findings since temperate climates, socioeconomic mechanisms, and agricultural
practices throughout much of the “Global North” often differ greatly from those elsewhere, particularly
in tropical regions, which host a majority of smallholder farms4,5.
The tropics are uniquely positioned to scale ERW technology whilst delivering co-benefits throughout
the developing world11. Experiments in Latin America and Southeast Asia have demonstrated the CDR
potential of tropical ERW and that the dissolution of silicate rock powders can increase plant resilience,
nutrient supply, and crop yield12,13. Notably, there is a lack of research on the efficacy of ERW with
smallholder farming practices, as well as studies on ERW in sub-Saharan Africa.
2
Sub-Saharan Africa could be a hotspot for ERW. The warm and humid climate provides ideal conditions
for ERW reaction kinetics, and there are enough basaltic deposits to sequester ~1 billion tonnes of
carbon dioxide14. Despite the abundance of family and smallholder farming (33+ million), there exists
widespread food insecurity in part because cultivation takes place in acidic and nutrient-poor soils15,16.
Therefore, partnerships between ERW suppliers and smallholder farmers in sub-Saharan Africa can not
only help ERW achieve gigatonne scale for CDR, but also provide the agricultural material to alleviate
food stress in one of the most populated regions on Earth. Achieving this, however, requires a robust
effort to remove the logistical and financial barriers to accessing ERW technology.
Collaboration between Flux and UNCCD
Flux is an enhanced rock weathering company based in Nairobi, Kenya. Flux is working to unlock the
potential of ERW across Africa, aiming to sequester hundreds of millions of tonnes of CO2whilst
simultaneously improving soils and yields for millions of farmers across the continent. As part of its
mission, Flux is running several experiments and trials to answer pressing research questions, and is
partnering with academic and philanthropic institutions interested in ERW in Africa.
In March 2024, Flux partnered with the United Nations Convention to Combat Desertification (UNCCD)
on a pilot study to assess the agronomic and social co-benefits of ERW deployments for women
smallholder farmers in Kisumu County, Kenya. Smallholder farms in this region are often <1 ha and used
to grow maize, sugarcane, and other crops for subsistence and income. However, liming agents and
fertilisers are not widely available or used due to cost limitations.
In this study, Flux provided the feedstock material that was required and farmers were encouraged to
use their standard farming practices to ensure that this study reflected real-world conditions.The
feedstock was selected based on local availability to minimise transport emissions, mineral composition
with a high theoretical CO₂ sequestration potential, and its composition of calcium and magnesium to
enhance soil health. Here, we report on preliminary results from this pilot, which highlight clear
agronomic co-benefits associated with spreading rock powders on maize crops across two cluster
populations.
Study Design
Geographic, Geologic, and Agricultural Setting
This study focuses on smallholder farm clusters in two wards - Ombeyi and Wawhidi - in Kisumu County,
which is located in Western Kenya (Figure 1). Positioned along the equator, the region experiences
warm daytime temperatures (~28°C on average) and substantial precipitation (~1900 mm) year round,
with most rainfall occurring during the two wet seasons from March-May (~230 mm/month) and
October-December (~210 mm/month). The tropical climate conditions support robust agricultural
activity at the commercial and smallholder scales, with Kisumu County serving as a major producer of
sugarcane, cassava, and rice in the region.
In addition to the ideal climatic conditions, Western Kenya is located in the central sector of the East
African Rift System. Geologically, the region is characterised by extensional tectonics, commencing in
the Oligocene, and associated, silica-undersaturated mafic volcanism.17 This volcanic activity has
produced rock types that appear particularly well suited for ERW, as they contain high amounts of
3
fast-weathering Ca-, Mg-, and Na-silicates. Among these are nephelinites, which occur in multiple
volcanic centres throughout Western Kenya.18
Figure 1: Study area. A) Map of Kenya in eastern Africa. Red shows the location of Kisumu County in
western Kenya. B) Kisumu County outlined by the red dotted line. C, D) Clusters with control plots
shown in yellow and treatment plots in pink.
Project Timeline
Event
Timeline
Funding Approved
Nov 2023
Sensitisation Meetings
Dec 2023
Farmer Onboarding & Mapping
Jan 2024
Baseline Soil Sampling
Jan 2024
Rock Spreading
Mid-Feb to mid-Mar 2024
Harvest & Yield Analysis
Jul to Aug 2024
Soil Analysis Certificates Issued to Farmers
Sep 2024
White Paper Publication
Nov 2024
4
2nd Soil Sampling (yet to be done)
Jan 2025
Future soil sampling / yield analysis
TBD
Onboarding of Farmers
In a community based pilot such as this, thorough sensitization, informed consent, and community
engagement were key components of the successful implementation of this study (Figure 2). Flux first
conducted a county-level stakeholder engagement initiative with key personnel from Kisumu County,
including the county commissioner and the ministers for Women and Gender, Environment, Agriculture,
and Business. Flux then engaged local community leaders to identify which community clusters could
meet the project requirements (i.e., size and number of farms, climatic conditions, crop types, and
willingness to participate). Once these clusters were identified, field agents held community meetings
to explain the project and delivered information in three languages: Swahili, Luo, and English. Farmers
who wished to participate were then onboarded.
Figure 2: Onboarding process.
A total of 56 smallholder farms used to cultivate maize in
the Wawidhi and Ombeyi communities of Kisumu County
were onboarded for the study. During the community
meetings, over 100 farmers volunteered to be part of the
trial. The 56 farmers selected from this initial group were
based largely on their field size, to ensure there was
adequate land for both treatment and controls not smaller
than 0.05ha.
Each farmer's plot was divided roughly in half into two
sections: a control plot and a treatment plot (referred to as
the “Application” plot). These two plots were subdivisions
from within one field that had similar usage in the past.
To ensure accurate mapping, the Fields Area Mapper app was used to collect area data for each farm;
maps of the two clusters are shown in Figure 2. Farmers were assigned unique identification numbers
based on their cluster and plot location. All farms in the Wawidhi cluster were prefixed with “KW”
(Kisumu - Wawidhi), while those in the Ombeyi cluster began with “KO” (Kisumu - Ombeyi). This was
followed by a sequential number and a letter to indicate the plot type: "C" for Control or "A" for
Application. For instance, a plot ID “KW21A” refers to Wawidhi farm number 21 with an application
treatment.
There was no cost to farmers during this deployment - which replicates the model that Flux sees for
scaling ERW with smallholders on the continent. Where hand spreading was used, farmers were
5
compensated at an appropriate daily rate. For mechanical spreading, local contractors were used.
Farmers also received free soil analysis and advice.
Rock Spreading
On the treatment plots, a 0-4 mm rock
powder was applied at a rate of 20
tonnes per hectare (Figure 3). This was
done using a combination of mechanical
spreading and hand spreading. The
mechanical spreading was carried out by
a local contractor using a 1.5 tonne
rear-mounted twin-disc lime spreader.
Hand spreading was carried out using
teams of 3-5 farmers with buckets of rock
powder that were collected from a
central location.
Rock powder was applied between the
1st and 2nd ploughing, and incorporated
into normal farming practices with no alteration from Flux.
Fitting in with regular farming practices
One of the key aims of this pilot was to implement ERW without disrupting what farmers normally do -
this will be crucial if ERW is to scale. Farmers in the two clusters from this pilot followed roughly the
same farming practices per harvest cycle of their maize crop, which is outlined in the diagram below.
Figure 3: Schematic showing normal farming practices in the region for this harvest cycle.
Feedstock
The feedstock used in the field trials was sourced as a by-product from a quarry located in Homa Bay
County in western Kenya, approximately 110 km from the field sites. The material was sourced at a
particle size of 0-4 mm. This particle size range was chosen for logistical feasibility as it was readily
available at the quarry site and able to be procured in time for the trial. The volcanic rock used in this
6
trial is a nephelinite, consisting primarily of clinopyroxene and nepheline (Table 1). It is rich in calcium
and magnesium, resulting in a high theoretical gross sequestration potential of 0.45 tCO2/trock (Table 2).
The silicate rock powder was analysed for heavy metal content using both ICP-MS and ICP-AES for
comparison with regulatory standards for soil and fertiliser. To date, Kenya has not established national
standards for potentially toxic elements in agricultural soils or inorganic soil improvers. We therefore
deferred to maximum permissible soil limits set by the World Health Organization (WHO) and Food and
Agriculture Organization (FAO), and to the European Union (EU) for inorganic fertilisers. Data from this
exercise are displayed in Table 3.
Heavy metal concentrations for the feedstock used in this study are less than the WHO/FAO
permissible limits for soils with the exception of Co, Cr, Cu, and Ni. However, metal content is below the
maximum limits for As, Cd, Cu, Ni, Pb, Zn as established by the EU for inorganic fertilisers. We estimate
based on the Cr and Ni content of this feedstock that several hundred applications at a rate of 20
tonnes per hectare would be required to attain soil concentrations that exceed the WHO/FAO
permissible limits, which suggests that this feedstock is safe for use in ERW field trials.
Table 1: Mineral composition of the feedstock as determined via XRD.
Mineral
Mineral formula
Composition %
Pyroxene (augite/hedenbergite)
(Ca,Na)(Mg,Fe,Al,Ti)(Si,Al)2O6
49.70
Nepheline
(Na,K)AlSiO4
25.10
Magnetite
Fe3O4
8.40
Feldspar (orthoclase/sanidine)
(K,Na)(Si,Al)4O8
2.70
Sodalite
Na8Al6Si6O24Cl2
2.70
Mica (biotite/muscovite)
K(Mg,Fe)3[AlSi3O10](OH,F)2
2.20
Natrolite
Na2Al2Si3O10•2(H2O)
2.10
Hydroxylapatite
Ca5(PO4)3(OH,F,Cl)
2
Vermiculite/Clay
(Mg,Fe,Al)3(Al,Si)4O10(OH)2•4(H2O)
2
Analcime
NaAl(SiO6)•(H2O)
1
Other minor phases
N/A
2.1
7
Table 2: Major element composition of the feedstock material as determined via XRF. *Note: The
Na2O value was derived as an average of two additional analyses. These were required as the
original XRF analysis used sodium peroxide as a flux for the preparation of the fusion beads. Sodium
was determined via atomic absorption spectrometry.
Oxide
Composition %
SiO2
37.2
Al2O3
8.5
Fe2O3
17.03
CaO
14.98
Na2O
3.96*
K2O
1.72
MgO
7.80
TiO2
5.75
MnO
0.24
P2O5
0.71
Total
98.16
Table 3: Trace element concentrations in feedstock determined by ICP-MS or ICP-AES (denoted by
asterisk *).
Element
EU limits for fertiliser19 [mg/kg]
Feedstock content (this study) [mg/kg]
As
40
4
Cd
1.5
0.12
Co
—
55.4
Cr
—
114*
Cu
300
234*
Mn
—
1856*
Ni
100
68.8
Pb
120
16.7
Se
—
<1
Zn
800
124*
Farming Practices
The study maintained the diversity of farming practices to reflect real-world conditions, recognizing that
each farmer uses different methods, seeds, fertilisers, manure, and faces varying climatic conditions
8
and soil types. In the end, fertiliser was not used by farmers on any plot during this study owing to the
increased price and lack of soil testing available to identify the most beneficial products. Use of manure
sourced from local livestock was reported by 50% of the farmers but was not quantified or tracked
during this study, however, the same practices were applied to both control and treatment plots by all
farmers. No farms used any form of irrigation due to the high rainfall in the area.
The subtle variations in farming practices guided the decision to include both a control and a treatment
plot for each farmer. Although additional repetitions per plot were considered, this was ultimately not
pursued due to operational constraints.
Soil Sample Collection
Soil sampling was conducted by one
Flux associate with assistance from
three stakeholders in the community.
These 4 team members (all women)
were trained in soil sampling and
collection at KALRO (Kenya
Agriculture and Livestock Research
Organisation) in Kibos, and became
agents for Flux in their respective
communities. This greatly helped with
community engagement and provided
a necessary skill that can be shared
for future sampling in the region.
Image: Farmers from within the
communities undertake soil sample collection training at KALRO laboratory.
Health and Safety
Throughout the trial, personal health and safety were paramount. To mitigate risks associated with
heavy machinery, we implemented strict safety protocols. Only trained operators, equipped with
high-visibility gear and protective equipment, were allowed to operate the tractor and spreader.
Regular maintenance ensured the machinery's safe operation.
Dust inhalation was a potential concern. We minimise exposure by restricting access to deployment
areas during windy conditions, providing dust masks to all personnel, and conducting regular safety
briefings. Despite no reported incidents, we plan to enhance future deployments with higher-rated
masks, extended safety training, and community first-aid lessons.
Methodology
Soil Characterization
Baseline soil samples were collected before planting, with one mixed sample taken per control and per
treatment plot. Samples were obtained using a systematic sampling method, with 10-15 cores which
were composited, taken at 5-metre intervals in a ‘W’ pattern across the plots to ensure a random and
representative sample. The samples were taken to a depth of 30 cm. The cores were not GPS logged
9
in this study but should be for future trials and as best practice for ERW MRV. This process resulted in a
total of 112 soil samples (one per control and one per treatment plot per farmer) being sent to the
KALRO for determination of texture classification, soil pH, organic carbon and nitrogen content, and
major and trace elemental composition following standard analytical protocols.
Unfortunately, some results were never returned by the lab and a duplicate of the soil samples was not
retained for these samples. The missing results are randomly distributed and can be attributed to
human error at the lab. Future studies should hold duplicate samples in an archive to mitigate the risk
of this error.
The soils of the Ombeyi cluster are, according to available soil maps20, haplic, partially sodic and
calcaro-cambic xerosols with arenosols, and minor calcic, sodic cambisols. In the Wawidhi cluster, pellic
and vertic, partially sodic vertisols and gleysols dominate. The low exchangeable sodium content (Table
4) in the topsoil of the fields included in the study indicates that these particular plots are not affected
by sodification.
A summary of the soil analyses for the Wawidhi and Ombeyi clusters, presented in Table 4, outlines the
baseline properties evaluated prior to treatment. The majority of farms in both the Wawidhi and Ombeyi
clusters are characterised by a clay soil texture. In the Wawidhi cluster, 90% of the farms have a clay
soil texture, 7% are classified as silt clay, and 3% as silt clay loam. In the Ombeyi cluster, 90.5% of the
farms are clay, with the rest classified as silt clay. The exchangeable sodium percentages of ca. 0.51 %
and 0.46 % respectively indicate that the particular
Changes in soil characteristics due to ERW will be evaluated via post-application sampling, to be
conducted during January 2025. The results will be published in an addendum to this report.
10
Table 4: Characterization of soil. Error represents 1 SE.
Parameter
Method
Wawidhi Cluster
Ombeyi Cluster
Soil Texture
Sand (%)
Silt (%)
Clay (%)
Hydrometer method
Clay
20.7 ± 1.5
28.3 ± 1.8
51.0 ± 1.1
Clay
37.0 ± 1.4
13.9 ± 1.0
49.1 ± 2.1
Soil pH
pH meter on 1:1
water:soil slurry
6.51 ± 0.07
6.81 ± 0.09
Total Nitrogen (%)
Kjeldahl method
0.15 ± 0.1
0.25 ± 0.01
Total Organic Carbon (%)
Walkley-Black method
1.53 ± 0.06
2.55 ± 0.12
Phosphorus (ppm)
Mehlich I extraction
28.20 ± 0.71
31.97 ± 7.23
Copper (ppm)
0.1M HCl extraction
1.24 ± 0.1
1.05 ± 0.14
Iron (ppm)
98.92 ± 4.06
37.03 ± 4.65
Zinc (ppm)
5.14 ± 0.34
3.60 ± 0.37
Yield Analysis
To ensure representative and comparable yield measurements between plots, a 10 m x 20 m ‘Harvest
Box’ was measured and roped off in each plot (Figure 4). The Harvest Boxes were positioned to give
maximum coverage of the greatest variation of yield within a particular plot, and located away from the
edges of fields where footpaths and livestock can disrupt growing. All of the maize plants within this
box were then harvested, and the cobs bagged and weighed separately from the rest of the field. The
total weight of the cobs from within the Harvest Box was then recorded as the ‘Harvest Weight’. This
yield can therefore be expressed as kg/0.02 ha.
Figure 4: Diagram of harvest analysis collection protocol.
Maize Nutritional Analysis
After the cobs were weighed, the farmers air-dried them as per normal practice for 3-5 days. They then
removed the kernels from the cobs. A 1 kg sample of these kernels were sub-sampled from the
application and control kernels per farmer and sent to KALRO for nutritional analysis. These results are
11
Potassium (meq/100g)
1N Ammonium
Acetate at pH 7,
Ethanol, 1N KCl at pH
7
1.36 ± 0.06
1.05 ± 0.07
Calcium (meq/100g)
13.91 ± 0.83
35.09 ± 3.94
Magnesium (meq/100g)
3.41 ± 0.04
3.75 ± 0.06
Manganese (meq/100g)
0.60 ± 0.03
0.30 ± 0.03
Sodium (meq/100g)
0.51 ± 0.04
0.46 ± 0.04
Cation Exchange
Capacity (meq/100g)
32.2 ± 1.12
50.34 ± 2.84
still pending and will be updated here to explore whether any difference in nutritional qualities
between control and application maize is observed.
Image: An overhead photo of a harvest box being implemented in the field.
Results
Loss of farms from study
Although 56 farmers were initially onboarded, several plots were lost from the study at various stages,
which are detailed below. As such, the results here were collected from 31 smallholder farms from the
Wawidhi and Ombeyi clusters.
Table 5: Number and reason for losses of participants throughout the trial. Detailed explanations can be
found at Annex A
Registered
interest in Trial
Initially
Onboarded
Rock Spread
Harvest Data
Collected
Data Utilised in
Study
Participants
100
56
46
34
31
Losses
—
44
10
12
3
Reasons for
losses
Farm size
too small
Spouses not
engaged
Requested
payment
7 due to
spouses
3 Planted
before
spreading
10 destroyed by
floods
1 harvested before
analysis
1 farm eaten by
donkeys
3 contaminated
yield results
12
Land ownership and participation constraints
Seven farmers withdrew their farms from the study due to pressure from their spouses. The motivation
of the study was to understand the impact with women smallholder farmers, so this was the subsection
that was engaged. However, this left some of the male spouses of the farmers feeling left out and
under-informed, and ultimately led to their withdrawal from the study. Future pilots should aim to
engage and educate all stakeholders within a community to reduce the likelihood of conflict or ill
feeling from other parties.
Flooding
This area of Kisumu County received unseasonably high and long rainfall, which resulted in flooding in
some areas. Ten farms had either their entire crops or one side of the plot destroyed by flooding.
Although some crops were subsequently replanted and small yields were documented, the data would
not have been applicable for this study. It would be an interesting further study to explore the effect of
flooding on the weathering efficiency of silicate rocks, and if it is subject to run-off or being
re-deposited.
Other challenges
Other challenges came from coordination with farmers and fitting in with practices. For example some
farmers planted before we were able to spread rock, and some harvested before our teams arrived to
conduct the harvest box analysis - thus losing any yield data.
Preliminary Observations
Preliminary observations were made 7 weeks after planting, but were not accurately quantified on all
plots. Aerial photographs taken at this point showed visible differences on some plots, but were not
taken for every farm.
13
Image: Overhead view of a farm in the Wawidhi cluster showing visible differences in maize
production between an application plot (left) and control plot (right) under normal farming practices
after seven weeks.
Harvest Yield Analysis
After roughly four months of growing, farmers began to harvest their maize crops as they became
ready. The harvest came earlier by two weeks in the Wawidhi cluster compared to Ombeyi. Statistically
significant (p<<0.05) increases in crop yield for the Ombeyi cluster (60.4±15.7%) and the Wawidhi cluster
(80±25.4%) are attributed to the application of silicate feedstock (Figure 7; Figure 8). To report yield
changes comprehensively, we present both the Average Yield Changes and the Aggregate Yield
Change. The Average Yield Changes, calculated by averaging the yield change across all fields, shows
a yield change of 71.17 ± 15.5%. This measure reflects the variability in yield across individual fields,
offering insights into field-specific performance and highlighting any variability or potential outliers in
the data. In contrast, the Aggregate Yield Change of 47.47 ± 5.73%, which is derived by averaging the
control and treatment data separately and calculating the yield difference between them, provides a
broader assessment of the treatment’s overall effectiveness across the entire study area. Possible
modes of action are considered in the discussion section below.
Figure 5: Farm-specific yield increases for maize. Left: bar plots showing increase in harvest yield at
each farm plot with the Ombeyi cluster. Right: same but for the Wawidhi cluster. Dotted lines denote the
average yield increase for each cluster (Ombeyi: 60.4±15.7 [%]; Wawidhi: 80±25.4 [%]).
The average yield increases are strongly influenced by the very large increases observed on a small
subset of the plots, specifically on those with a low control yield. The least productive quartile saw an
average yield increase of 169.8 %, whereas the most productive quartile, with control yields well above
the national average, saw an average increase of 28.9 %.
14
Table 6: Summary of average yield increases by control yield. It is apparent that low-productivity plots
saw the highest relative yield impact of rock powder application. For comparison, the average corn yield
in Kenya was 1900 kg/ha in 2023/2024.21 Complete yield results are available at Annex B.
Control yield [kg/ha]
n
Average yield increase [%]
< 1600
10
143.5
1600 - 2600
11
44.2
> 2600
10
28.5
Socioeconomic Impact of Yield Increases
The average yield per hectare on control plots was 2085 kg. The average on treated plots was 3075
kg. At the August 2024 price of 43KES (Kenyan Shillings) per kg of maize, this difference of 989 kg per
hectare is equivalent to $326. For context, a farmhand worker in this region could expect to earn $60
to $80 monthly. This increase in yield also enhances food security by providing more maize for both
consumption and sale, reducing dependency on external sources. Moreover, higher income and
productivity can stimulate community economic growth and foster long-term agricultural sustainability.
Discussion and Further Study
The yield data presented in this first report provides strong indication that ERW deployments have the
potential to significantly increase maize yields in smallholder farming systems of Western Kenya. The
encouraging results justify further research, and operational learnings will improve the setup of future
trials.
While the results from the post-application soil sampling will be required to understand the exact effect
that the enhanced weathering had on soil fertility parameters, the known composition of the rock
powder along with literature data allows us to identify possible modes of action, including:
1. pH increase due to silicate weathering and resulting improvement in nutrient availability9
2. Release of macro and micro nutrients (P, K, Ca, Mg, Zn, Cu, Fe) from the silicate minerals during
weathering9,22
3. Improved P availability as a result of silicic acid formation23
4. Increased pest and disease resistance due to higher plant-available silicon24
15
5. Improved physical soil properties (water infiltration, water holding capacity)13
6. Increased soil CEC due to high content of zeolite group minerals in the rock powder25
It is intriguing that the substantial yield increases shown in this study were achieved on soils with a
relatively high starting pH (6.49 and 6.78 for Wawidhi and Ombeyi cluster respectively), a high clay
content and CEC, and reasonably high available P and K, and that neither of these parameters directly
correlates with the magnitude of yield change. It therefore seems unlikely that the agronomic effect is
(exclusively) due to pH regulation or macro nutrient release, two processes commonly demonstrated in
ERW trials9,21.
In this study, ca. 50% of farmers reported using locally sourced livestock manure; however, the quantity
and frequency of application were not tracked, which may impact nutrient availability and crop yield
results. Recognizing the potential influence of manure on soil nutrient levels and yield outcomes, future
trials should include systematic quantification of manure use to improve the accuracy of nutrient
contribution assessments. Although we did not collect detailed manure data in this study, each plot’s
control and treatment design accounts for variability introduced by individual farmer practices. This
approach strengthens the reliability of treatment effects, despite the variation in unquantified manure
use across the farms.
Building upon the insights and challenges identified in this pilot study, future research will focus on
enhancing the study design, addressing limitations, and expanding the scope to better understand
weathering kinetics and the agronomic and socioeconomic impacts of ERW in smallholder farming
systems in Sub-Saharan Africa.
Future trials will evaluate a broader range of application scenarios. This will include soils developed
over felsic basement rocks of the rift shoulders, where soil pH and nutrient status are generally lower,
indicating more favourable weathering conditions and a potentially higher agronomic impact of ERW. To
account for additional variables beyond soil and rock powder characteristics, a more detailed
documentation of farming practices, including the type and amount of organic fertiliser applied, will be
conducted. Pest and disease incidence will be monitored to account for this (potentially
underestimated) parameter.
To better capture intra-plot heterogeneity in future trials, we will consider increasing the number of soil
and yield replicates per plot. For soil sampling, this will involve creating composite samples from
different zones within each plot, while for yield measurements, multiple smaller harvest boxes will be
established. This approach will allow for a more detailed assessment of variability within plots,
enhancing the statistical robustness of the data. The field trials will be supplemented by mesocosm
experiments in a controlled environment, which will allow us to investigate the impact of individual
variables in isolation.
Also, we recognize that the absence of historical yield data for control plots limits our ability to fully rule
out selection bias, as it is unclear whether yield differences may partially stem from pre-existing
conditions. Future trials should incorporate a baseline yield assessment to better account for natural
productivity differences, which would provide a clearer measure of ERW's direct impact on crop yields
In addition to the investigation of changes in soil fertility indicators and crop yield, future studies will
also include measurements targeted at the quantification of CDR. Both established and novel
measurement approaches will be tested for their applicability in tropical smallholder farming systems,
aiming to unlock their vast potential for ERW.
16
Conclusions
The preliminary findings from this trial indicate that ERW can lead to significant agronomic benefits for
smallholder farmers in Kisumu County, Kenya. The average yield increase of 71.17% ± 15.5% and the
aggregate yield increase of 47.47% ± 5.73% collectively demonstrate that the application of silicate rock
powder significantly enhances soil health and boosts crop productivity. This remarkable improvement
may be partly due to the initially low baseline yield on these marginal lands, where soil conditions were
suboptimal for high productivity. The addition of silicate minerals likely enriched the soil with essential
nutrients like calcium and magnesium, while also raising soil pH, which can improve nutrient availability.
These changes not only support healthier crop growth but also contribute to increased food security
and economic benefits for local communities, creating a more resilient agricultural system. Our results
highlight ERW's potential as a sustainable farming practice that simultaneously addresses climate
change through carbon sequestration and supports the livelihoods of smallholder farmers in
Sub-Saharan Africa. While promising, further research is essential to solidify these findings, address
fundamental challenges, and optimise the application of ERW at scale.
We encourage interested parties from both academia and industry to reach out and discuss with us the
findings presented here, as well as the possibility of collaboration in future research.
References
1. Masson-Delmotte et al. (2021). Climate change 2021: The physical science basis. Cambridge
University Press. 2391 pp.
2. Beerling. (2017). Enhanced rock weathering: biological climate change mitigation with co-benefits for
food security? Biology Letters 13, 20170149.
3. Manning and Theodoro. (2020). Enabling food security through use of local rocks and minerals. The
Extractive Industries and Society 7, 480-487.
4. Lowder et al. (2016). The number, size, and distribution of farms, smallholder farms, and family farms
worldwide. World Development 87, 16-29.
5. Samberg et al. (2016). Subnational distribution of average farm size and smallholder contributions to
global food production. Environmental Research Letters 11, 124010.
6. Herrero et al. (2017). Farming and the geography of nutrient production for human use: A
transdisciplinary analysis. Lancet Planet Health 1, e33-42.
7. Haque et al. (2019). Co-benefits of wollastonite weathering in agriculture: CO2sequestration and
promoted plant growth. ACS Omega 4, 1425-1433.
8. Haque et al. (2020). CO2sequestration by wollastonite-amended agricultural soils: An Ontario field
study. International Journal of Greenhouse Gas Control 97, 103017.
9. Skov et al. (2024). Initial agronomic benefits of enhanced weathering using basalt: A study of spring
oat in a temperate climate. PLoS ONE 19(3), e0295031.
10. Vienne et al. (2022). Enhanced weathering using basalt rock powder: Carbon sequestration,
co-benefits, and risks in a mesocosm study with Solanum tuberosum.Frontiers in Climate 4, 869456.
17
11. Edwards et al. (2017). Climate change mitigation: Potential benefits and pitfalls of enhanced rock
weathering in tropical agriculture. Biology Letters 13, 20160715.
12. Larkin et al. (2022). Quantification of CO2removal in a large-scale enhanced weathering field trial on
an oil palm plantation in Sabah, Malaysia. Frontiers in Climate 4, 959229.
13. Swoboda et al. (2022). Remineralizing soils? The agricultural usage of silicate rock powders: A
review. Science of the Total Environment 807, 150976.
14. Boudinot et al. (2023). Enhanced rock weathering in the Global South: Exploring potential for
enhanced agricultural productivity and carbon dioxide drawdown. IGSD-PxD. 42 pp.
15. Wiggins and Keats. (2013). Leaping & Learning: Linking smallholders to markets. Agriculture for
Impact, Imperial College London. 120 pp.
16. Giller et al. (2021). Small farms and development in sub-Saharan Africa: Farming for income or for
lack of better options? Food Security 13, 1431-1454.
17. Furman (2007). Geochemistry of East African Rift basalts: an overview. Journal of African Earth
Sciences, 48(2-3), 147-160.
18. Le Bas (1987). Nephelinites and carbonatites. Geological Society, London, Special Publications,
30(1), 53-83.
19. EU Regulation 1009/2019. Annex 1, Part 2, PFC 3(B): Inorganic soil improver. 129 pp.
20. RCMRD (2023). Kenya Soils Map (https://opendata.rcmrd.org/maps/rcmrd::kenya-soils-map/about)
21. USDA (2024): Country Summary Kenya (https://ipad.fas.usda.gov/countrysummary/?id=KE)
22. Beerling et al. (2024) Enhanced weathering in the US Corn Belt delivers carbon removal with
agronomic benefits. Proceedings of the National Academy of Sciences, 121(9), e2319436121.
23. Scherwietes et al. (2024) Local sediment amendment can potentially increase barley yield and
reduce the need for phosphorus fertilizer on acidic soils in Kenya. Frontiers in Environmental Science,
12, 1458360.
24. Weerahewa and Somapala (2016) Role of silicon on enhancing disease resistance in tropical fruits
and vegetables: a review. OUSL Journal 11:135–162
25. Szatanik-Kloc et al. (2021). Effect of low zeolite doses on plants and soil physicochemical properties.
Materials, 14(10), 2617.
18
Annex A - List of farms removed from study
Farmer ID
Reason for Removal from Study
1
KO5
Rock not spread - Spouse declined
2
KO13
Rock not spread - Spouse declined
3
KW17
Rock not spread - Spouse declined
4
KO26
Rock not spread - Spouse declined
5
KW23
Rock not spread - Spouse declined
6
KW25
Rock not spread - Spouse declined
7
KW28
Rock not spread - Spouse declined
8
KO16
Rock not spread – Planted before Flux team arrived to spread rock
9
KO29
Rock not spread -Planted before Flux team arrived to spread rock
10
KW21
Did not plant on the control side
11
KO12
Affected by flooding – Rains overwhelmed the control and application areas.
12
KO14
Affected by flooding - Rains overwhelmed the control and application areas.
13
KO25
Affected by flooding - Rains overwhelmed the control and application areas
14
KW3
Affected by flooding - Rains overwhelmed the control and application areas.
15
KW7
Affected by flooding - Rains overwhelmed the control and application areas
16
KW9
Affected by flooding - Rains overwhelmed the control and application areas
17
KW12
Affected by flooding - Rains overwhelmed the control and application areas
18
KW19
Affected by flooding - Rains overwhelmed the control and application areas
19
KW22
Affected by flooding - Rains overwhelmed the control and application areas
20
KW29
Affected by flooding -Rains overwhelmed the control and application areas
21
KO18
Control side destroyed by flooding
22
KO21
Harvested before Flux team arrived to do analysis
23
KW21
Harvest Errors
24
KO21
Harvest Errors
25
KO18
Harvest Errors
19
Annex B - Complete harvest and yield change data
Harvest Weight from inside
harvest box (10mx20m) (kg)
#
Field ID
Control
Treatment
Yield Change
1
KO1
106.38
144.95
36.26%
2
KO2
24.3
78.7
223.87%
3
KO3
32.79
46.32
41.26%
4
KO4
36.85
37.45
1.63%
5
KO6
56.81
76.84
35.26%
6
KO9
54.69
72.85
33.21%
7
KO10
36.85
52.4
42.20%
8
KO15
30.1
49.95
65.95%
9
KO17
32.63
43.86
34.42%
10
KO19
28.85
45.35
57.19%
11
KO20
69.34
77.98
12.46%
12
KO22
39.83
63.59
59.65%
13
KO23
40.09
59.64
48.77%
14
KO24
12.36
31.39
153.96%
15
KW1
42.4
72.3
70.52%
16
KW2
32.65
58.4
78.87%
17
KW5
86.22
139.6
61.91%
18
KW6
15.75
26.6
68.89%
19
KW8
72
89
23.61%
20
KW10
23.6
59.5
152.12%
21
KW11
12.15
40
229.22%
22
KW13
51.38
74.15
44.32%
23
KW15
31.55
35
10.94%
24
KW33
39.84
52.98
32.98%
25
KW16
66.18
72.43
9.44%
26
KW18*
72
89
23.61%
27
KW20
56.3
62.3
10.66%
28
KW34
54.9
75.9
38.25%
29
KW26
24.63
36.72
49.09%
30
KW31
9.2
48.2
423.91%
31
KW32*
46.1
60.8
31.89%
Average Yield
Change ± SE
71.17 ± 15.50%
Average Yield
43.18
63.68
Aggregate
Yield Change ±
SE
47.47 ± 5.73%
20
Annex C - Complete Soil Analysis Data
Soil analysis data for all plots can be found in the supporting documents on ResearchGate.
21
Annex D - Example Soil Analysis Certificate given to farmers
22
Annex E - Deployment Maps - Wawidhi Cluster
23
Annex E - Deployment Maps - Ombeyi Cluster
24
Annex F - Accompanying Images
Image: Overhead of one of the plots showing rock spread on the right hand side. During this phase, the
rock had got wet from rainfall and was harder to spread, which resulted in more passes with the tractor.
Image: Drone photo showing one of the fields within the deployment setting. Access with the tractor
was fairly reasonable despite the small sizes of the plots.
25
Image: One of our community field agents giving a briefing to farmers in Wawidhi Cluster before
spreading operations began. Using community members as agents helped reduce frictions for the
project and is highly recommended for future studies.
Image: Women farmers from the Wawidhi Cluster see the rock powder for the first time.
26
Image: The ‘Backhoe’ loader filling up the lime spreader with silicate rock powder before an
application. The weight of rock held per scoop of the backhoe was calculated to allow for accurate
filling of the spreader before heading to a field site.
Image: Farmers in Wawidhi celebrate receiving their soil sample analysis in Sept 2024
27
