PhD Graduands

The high uptake level of cloud computing services has correlated with an increase in providers of cloud services, who are offering comparable solutions marketed at different prices and at distinctive Quality of Service (QoS) levels. This portends a decision challenge to users of those cloud services, who have to make a selection or a comparison between the available providers of cloud services in so far as performance of their cloud solutions are concerned. Even though there exists computational models for developing QoS measuring tools, they are not vendor agnostic therefore hampering cross vendor functionality comparison.

The decision challenge difficulty is further made worse in view of the blueprint of the current cloud QoS monitoring framework, where the results are stored in the cloud provider’s platform after monitoring for the user to query later. This raises trust issues as far as the results are concerned during service level agreement evaluation.

Further, the rise in the number of providers of cloud services, claiming to offer better services than their competitors, has brought impetus on the need for users of cloud services to authenticate the QoS measured by the different tools of cloud solution providers.

Noting the unavailability of vendor neutral cloud QoS measuring tools, cloud clients have not option but to depend on the tools offered by the cloud providers who also double as their cloud service providers, due to vendor specificity of the tools. In a situation where the user has multiple providers for the same services, with an aim of increasing redundancy, it becomes difficult to liken the QoS results obtained from the various providers since the tools used in measuring cannot be ported across platforms.

To abate the decision challenge and enable cross cloud performance comparison, various research have been done culminating in probable solutions, like the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS), Heterogeneous Similarity Metrics (HSM), Event Based Multi Cloud Service Applications Framework, Multiple-Cloud Monitoring platform, MUSA framework and the PRECENSE framework.

Whereas there is existence of research trying to address the cross cloud performance comparison, the shortcoming is that they rely on the use of existing vendor specific tools, customized for the specific cloud providers infrastructure which are then spread across different cloud providers, while in some instances the use of customized agents installed in various cloud providers’ platform.

The other outstanding shortcoming of the solutions meant to address the service choice overload problem is that they rely on synthetically generated data as opposed to relying on the actual data generated during the time the platform is in use, as well as relying on historical data of past clients.

This research addressed the existing gap by developing a cloud QoS monitoring framework from which a vendor agonistic cloud QoS monitoring model can be developed. The focus was on Software as a Service (SaaS) cloud computing solutions. In development of the framework, the research focused on the location of the QoS monitoring tool, the intention of monitoring, and the mode of access to the cloud services.

It was discovered through this research that to actualize development of a vendor neutral model, the location of the cloud QoS monitoring tools could be shifted from the cloud provider’s platform to the client devices, and use of internet browsers as the mode of access to the services.

This culminated in development of a vendor neutral SaaS cloud QoS monitoring tool, prototyped as a bowser extension, which was implemented using JavaScript, Node.js and SQLite database. To confirm the applicability of the developed vendor neutral tool, it was implemented on chrome browser. The monitoring tool developed from the vendor agonistic model was then used to monitor the QoS of five international SaaS cloud companies, which are, Salesforce, Google, Hubspot, Shopify and Microsoft.

The QoS parameters monitored by the vendor neutral tool were service stability, service response time of the service and service availability, which are the main quantitative parameters for cloud QoS as far as performance is concerned. For validation purposes, results from the vendor agnostic tool and the cloud service providers’ integrated tools were compared using a case study approach. The vendor agonistic tool proved more handy compared to the cloud providers’ integrated tools with regards to the number of QoS parameters it could monitor, as well as the fact that it can be used for performance comparison across different cloud platforms.
The vendor agonistic cloud QoS monitoring solution has capabilities to measure three main QoS metrics, namely, availability of the service, service stability and response time of the service, unlike the cloud providers’ solutions, which have capability to only measure one quantitative metric. Hubspot, Gsuite and Salesforce have a service availability measuring capability, while Shopify has a service response time and service availability capability only.

Due to its vendor neutrality, the tool can also be handy to cloud service users in so far as performance comparison across clouds is concerned for providers offering similar services, prior to the user making a choice on which cloud provider to procure for long term services.

The tool was subjected to Google docs and Microsoft 365 cloud services for comparison performance between 6th October 2020 to 27th October 2020, under the same computing platform and Internet conditions. From the comparison, the average service response time for Google was 4.47 seconds while for Microsoft was 6.04 seconds. Both platforms had an availability of 100% since at no time during experimentation did any of the platform report a platform failure leading to outage of services.

Whereas the availability is 100%, the fluctuations in the service response time were higher for Microsoft at 5.966 seconds than for Google at 2.003 seconds, meaning the Google platform was more stable than the Microsoft platform. In terms of evaluating the trustworthiness of the results reported by the different cloud computing platforms, this research developed a quantitative trust model for evaluation trust of the various cloud providers. This involved a comparison of the results from the vendor neutral model and the results from cloud provider’s embedded monitoring tools, using the common metric monitored by the two solutions, service availability, and a new parameter, confidence interval, introduced by this research.

From the trust evaluation done between 6th October 2020 to 27th October 2020, it was noted that the two compared cloud providers, Google and Microsoft, were both trustable since the results they reported were within the confidence interval of those reported by the vendor neutral model. In terms of overall performance it was found that Google performed better than Microsoft. Where a decision is to be made on whose services to procure, the user can factor in the decision making process this performance measures. Future works from this research could be extended to monitor infrastructure as a service based as well as platform as a service based cloud solutions. Additional works could also focus on other common aspects used by all cloud providers at the client side, for example the operating system, where the monitoring capability could be installed as a utility on the operating system.

Fisheries including tunas play a significant role in contributing to the national economy. The
management and development of tunas in Kenya has been challenged by the paucity of information
and knowledge on this fishery. The main objective of this study was to interrogate the scientific
robustness of the Kenya national fisheries policy and legal framework on tuna management in the
country. The study sourced catch data from primary and secondary sources. Fish samples from
artisanal catches were collected monthly from August 2015 to December 2016 from five landing
sites at the Kenyan coast; Amu (Lamu), Shella (Malindi), Mnarani (Kilifi), Watamu and Old Town
(Mombasa). Historical catch data from 24 foreign flagged longline tuna fishing vessels licensed to
harvest fish in Kenya Exclusive Economic Zone (EEZ) for the period February 2012 to March
2017 was obtained from Kenya Fisheries Service (KeFS). These data sources were complemented
by literature reviews, interviews and field visits. The data and information collected was analyzed
and used to assess the temporal and spatial variation in fish catch rates, species composition and
distribution, size distribution, growth parameters and mortality rates for some artisanal fishery
species. A total of 2686 individuals of tunas weighing 31,672 Kg representing 15 species, 13
genera and three families (Scombridae, Istiophoridae and Xiphiidae) from artisanal fishery were
sampled. Results of this study revealed that Thunnus albacares, Xiphias gladius and
Scomberomorus spp significantly contributed to the coastal fishery in Kenya accounting for 40%,
27.7% and 8%, respectively, of the total catch of tunas sampled. Fish catch rates varied with sites
with Amu recording the highest CPUE of 23.7 KgFisher-1Trip-1

, closely followed by Mombasa

with 19.16 KgFisher-1Trip-1

. Watamu, Shella (Malindi) and Mnarani recorded 11.1 KgFisher-1Trip-
1

, 7.5 KgFisher-1Trip-1

and 6.2 KgFisher-1Trip-1

, respectively. Fish catch rates varied monthly with

low catch rates reported from December 2015 to April 2016. The highest CPUE of 21.6 KgFisher-
1Trip-1 was recorded in the month of August 2015 while the lowest, 3.0 KgFisher-1Trip-1

, was
observed in December of the same year. Most of the tunas were landed during the South East
Monsoon (SEM) between May to October. Results of the 24 longline tuna vessels flagged to six
countries (China, Taiwan, Oman, Seychelles, Spain and Mauritius) fishing in the Kenya Exclusive
Economic Zone (EEZ) indicate that a total of 1833 individuals of tunas weighing 1,519,398 Kg
were harvest based on the declared catch. Thunnus albacares, Katsuwonus pelamis and Thunnus
obesus dominated the catch accounting for 43%, 29% and 17% of the total catch, respectively.
Analysis of Similarities (ANOSIM) of catch composition indicated significant differences been

vessels flagged to different states. Vessels flagged to Spain and Seychelles contributed 48% and
27% of the total catch, respectively. Vessels falged to Mauritius recorded the highest CPUE of
42,000 KgVessel-1Trip-1

followed by Spain with 5,392.6 KgVessel-1Trip-1

. Results of the analysis
for the size distribution, life history strategies, mortality and exploitation rates indicated that over
90 % of Thunnus albacares and Xiphias gladius sampled in artisanal fishery were juveniles.
Exploitation rates were also above the optimal indicating that the two fisheries are experiencing
growth overfishing. A review of the Kenya fisheries policy and legislative framework reveals that
it is coherent with the regional and international law with focus to scientific aspects. However, the
challenge is with effective implementation of tuna fishery management and development measures
informed by science, in particular biological aspects namely the Total Allowable Catch (TAC),
Total Allowable Effort (TAE) and precautionary approach. Information generated by this study
will certainly broaden the scientific knowledge and understanding about tunas in Kenya and inform
policy for their effective conservation and management.

Insider threat to information security is increasingly becoming a challenge to information security managers. One of the biggest challenges is not a lack of strong and robust policies, but that of ensuring full or highest rate of compliance with the policies. This is more compounded by the threats posed by insiders who have unfettered access to information systems assets. It is no surprise then that despite heavy investments in ensuring information security infrastructure, institutions still face the highest rates of information security breaches. Numerous studies have been conducted to provide insights and models on information security mitigations. However, very few studies have considered the policy compliance culture phenomenon. Among those who have considered the mixed methodology approach, none of the scholarly studies have considered grounded theory methods. The overall objective was to establish the relationship existing between organizational culture and information security compliance culture. As part of the Specific objective, the study intended to; 1) explore the relationship that exists between organizational culture and the actual information security compliance culture in universities in Kenya, 2) explain the relationship that exists between organizational culture and the actual information security compliance culture in universities in Kenya through theory generation, 3) and validate the theoretical model that predicts information security compliance culture.

The study employed an exploratory sequential mixed-method research design. This followed the QUAL-Quan principles. The population of this study was the Universities in Kenya. The study was divided into two phases namely, the model development phase and the model validation phase. The model development phase was designed to achieve two objectives namely: exploring the factors that impact information security compliance culture and explaining the relationships between the emerging factors and information security compliance culture through theory generation. The model validation phase was designed to test and validate the emergent theory through a semi-structured questionnaire. The model development phase adopted a grounded theory methodology while the model validation phase adopted the survey questionnaire approach.

The resulting theory was analysed and discussed in terms of model development and model validation. In the model development phase, several themes emerged which upon consolidation, were grouped into 4 main thematic groupings namely, demographic-oriented themes, organizational-oriented themes, individual-oriented themes, and information security compliance culture-oriented themes. The organizational oriented themes were further sub-grouped into the organizational level factors and moderating factors. The same was also done for individual-oriented themes to generate the individual-level factors and the factors moderating the individual-level factors. The study thereafter generated a theoretical model that explained a relationship between organizational initiatives, independent behavioral trends, management support, individual demographic interventions, and external organizational interventions towards information security compliance culture (ISCC). The model validation phase produced findings that supported the emergent theoretical model by having factor loadings that significantly supported the model among other parameters that were tested.

The study makes a main theoretical model contribution which is highlighted based on the model developed in phase one and the validated theoretical model. The model is adaptable to future researchers interested in covering information security compliance studies. The other contribution that this study makes is the methodological contribution which is also discussed in line with the efficiency of the procedures this study efficiently adopted. Further, the application of mixed methods as adopted in this study will provide insights to future information systems researchers to consider when deciding on how to conduct behavioral related studies. In terms of practice, the emergent theoretical model will be beneficial to practitioners in formulating checklists geared towards strengthening information security compliance regimes within their policy directions. This study is important because it provides a theoretical direction and methodological directions for future exploration of information security-related studies.

Keywords: Insider Threats, Information Security, Compliance Culture, Mixed Methods, Grounded Theory

Plastics enter the ocean inform of either large debris or microplastics that are a product of breakdown from the large debris or are primarily microplastics suchs as beads used in beauty products. Plastic pollution impacts in oceans are considerable world over. Microplastics (MPs) are tiny plastic particles measuring between 0.1μm and 5000 μm, make an important part of plastic pollution and form a pathway to the aquatic food web including humans. And although this is a global problem, there are limited studies along the Kenyan Coast in the Western Indian Ocean (WIO). This study looked at MPs in the surface waters, sediments and fish within three sites (Tudor, Port Reitz and Mida Creek) that were sampled in January/February and September 2018. Water samples were collected by towing neuston nets of 500 μm (large) and 250 μm (medium) mesh sizes and sieving 50 litres of seawater through a 20 μm net (small) size. Sediment samples were collected from the intertidal zone using a 3.6 cm diameter corer up to 10 cm deep. Fish were obtained from fishermen on site and local landing beaches. Samples were digested in 10 % Potassium Hydroxide and, microplastics extracted by Thompson’s improved density separation protocol using super saturated (0.8 μm) Sodium Chloride (NaCl) solution (1.2g cm-3). Total concentrations of MPs in both water and sediments, was highest in Tudor Creek followed by Port-Reitz and finally Mida. In the water column smallest MPs (20-250μm) recorded the highest concentration while in the sediments, MPs of the large size (500-4999 μm) were the most abundant. In fish, MPs concentration was highest in demersal fish followed by pelagic fish and in both types, omnivores recorded highest MPs concentrations followed by carnivores and finally grazers. Polyethylene (PE) polymers were the most abundant (63.9 %), followed by polypropylene (PP) (27%), while 9.1 % were unknown. Based on the results, the marine ecosystems along the Kenya coast are polluted with MPs.

Much attention is given to the study of generalized distributions because, adding more parameters gives flexibility in the shape, location, scale and tail(s) of a distribution. Existing literature shows that the generator approach is one of the desired methods of
constructing generalized distributions. Also, the mixture method is used because it gives flexibility to distributions.

In this work we determine various methods of construction and moments of the logistic distribution and its generalizations. The methods considered are; the difference of two standard Gumbel random variables, Burr differential equation, transformations and mixtures. The moment generating function has also been obtained. We also show the application of the cdf of the logistic distribution in determining the probability of default in logistic regression using data from a money lending company in Kenya - Mobipesa Ltd.

The generalized logistic distributions have been constructed using various transformations, the Burr differential equation, mixtures of Gumbel, beta 1 and beta 2 distributions. A new distribution ’extended standard logistic’ has been introduced using beta distributions and their generalizations. GLIV, extended GLIV and Exponential generalized Beta II distributions have been obtained whence with their special cases. A clear pattern of construction of the generalized logistic distributions is established. We further determine the discrete and continuous mixtures of minimum and maximum order statistic distributions from the standard logistic and exponentiated logistic distributions. The mixing distributions used are zero truncated Poisson, binomial, negative binomial, geometric and the logarithmic series distribution. The minimum and maximum order statistics distributions have been constructed alongside their hazard and survival functions

The other major contribution of this work is the introduction of the Logistic Inverse Gaussian distribution distribution. This distribution has been constructed from the properties if the modified Bessel functions of the third kind. Its log-likelihood function and moments have been studied. Continuous mixtures of the logistic with location parameter and shape parameter have been studied. The mixing distributions used include; logistic, exponential, gamma I, gamma II, inverse gamma, half logistic and reciprocal inverse Gaussian. The mixed distributions have been expressed in terms of the modified Bessel functions of the third kind. The rth moments, hazard and survival
functions have been determined. Since we have quite a number of generalized logistic distributions and their special cases, it is not possible to consider all their mixtures using all the mixing distributions proposed by Nadarajah and Kotz (2004), whatever is not done has be left for further research.

Abstract Kinetoplastids are human pathogens with devastating economic and health effects, which include Leishmania and Trypanosoma species from flagellated protozoans. With the developed technology platform that allows the generation of high atomic level resolution of pathogen ribosome’s crystal structures, we demonstrate that rRNA is a target of choice for the development of next-generation drugs. In addition, using several novel and transformative technologies we have developed, we demonstrate that the modular nature of rRNA facilitates the development of in vitro assays, structure determination, molecular-modeling, and compound screening studies for drug design. We employ computational homology and de novo modeling to reveal an atomic-level view of Leishmania and Trypanosoma ribosome and use the information of the rRNA structure and movement to design anti-infective-like compounds that target biologically functional ribosome RNA motifs in a predictable manner. This was performed by screening the pathogen box and the natural product databases where we got the best 40 compounds that bind well to the predicted motifs. Further analysis was conducted and mode of action of how the binding happens explained at the conclusion. Therefore, developing additional measures to control these “neglected tropical diseases” becomes increasingly clear, and we believe that the opportunities for developing drugs, diagnostics, vaccines, and other tools necessary to expand the knowledge base to combat these diseases have never been better.

Keywords: Neglected tropical diseases; homology and de novo modeling; rRNA motif; In vitro; Screening

Many actuaries worldwide use Systematic Mortality Risk (SMR) to value actuarial prod-
ucts such as annuities and assurances sold to policyholders. Data availability plays an

essential role in ascertaining the SMR models’ accuracy, and it varies from one country to
another. Incorrect stochastic modeling of SMR models due to paucity of data has been a

problem for many Sub-Saharan African countries such as Kenya, thus prompting modifi-
cations of the classical SMR models used in those countries with limited data availability.

This study aims at modelling SMR stochastically under the collateral data environment

such as Sub-Saharan African countries like Kenya and then apply it in the current actuar-
ial valuations. This thesis has formulated novel stochastic mortality risk models under the

collateral data setup. Kenya population data is preferably integrated into the commonly

applied stochastic mortality risk models under a 3-factor unitary framework of age-time-
cohort. After testing SMR models on the Kenyan data to assess their behaviours, we

incorporate the Bühlmann Credibility Approach with random coefficients in modeling.
The randomness of the classical SMR models is modeled as NIG distribution instead of
Normal distribution due to data paucity in Kenya (use of collateral data environment). The
Deep Neural Network (DNN) technique solves data paucity during the SMR model fitting
and forecasting. The forecasting performances of the SMR models are done under DNN

and, compared with those from conventional models, show powerful empirical illustra-
tions in their precision levels. Numerical results show that SMR models become more

accurate under collateral data after incorporating the BCA with NIG assumptions. The
Actuarial valuation of annuities and assurances using the new SMR offers much more

accurate valuations when compared to those under classical models. The study’s find-
ings should help regulators such as IRA and RBA make policy documents that protect all

stakeholders in Kenya’s insurance, social protection firms, and pension sectors. For areas
for further research, one can use the BCA approach for Sub-Saharan African countries
with similar demographic characteristics and Hierarchical BCA in SMR modeling.

Infectious diseases are a public health concern, as they cause serious illnesses in humans. Among infectious diseases, malaria, leishmaniasis, and bacterial diseases account for about a million fatalities per year. Globally, two hundred million cases of malaria were reported in 2019, with over four hundred thousand fatalities. Africa is the epicentre of malaria, accounting for at least 94 percent of all reported cases. Similarly, the World Health Organization reports 30,000 new infections of visceral leishmaniasis yearly, with 20,000 deaths. Incidences of cancer and coinfection with bacteria have been reported in malaria and leishmaniasis patients. Cancer accounted for about ten million fatalities and 19.3 million new cases globally. To reduce the burden of these diseases, chemotherapies are widely practiced. However, drug –resistance has been a persistent problem. In addition, the most effective drugs are costly and/or toxic, necessitating the continuous search for novel drugs. Plant derived heterocyclic compounds, including flavonoids have been
reported to show activities against pathogenic infections and cancer. Thus, the antimalarial, antileishmanial, anticancer, and antibacterial properties of some members of the Leguminosae and Moraceae families which are known to be rich sources of flavonoids and other phenolics were studied. Hence, phytochemical analysis of the dichloromethane/methanol (1:1) extracts of Mundulea sericea and Tephrosia uniflora (both Leguminosae) and Strebulus usambarensis (Moraceae) yielded thirty-one compounds. Phytochemical analysis of S. usambarensis stems and roots resulted in the identification of three novel naphtho-benzofuran derivatives, named usambarin A (110), B (111), and C (112). Eight new naphthalene derivatives named usambarin D (113), E (114), F(115), G (116), H (117), I (118), and L (125), phenyl-1-benzoxepin derivative (120), two flavans (119 and 126), and four known compounds. Similarly, the analysis of the stems and roots of M. sericea yielded ten known compounds; three flavanonols, two flavanols, an isoflavone, one rotenoid, two pterocarpans, and the sterol stigmasterol. Phytochemical investigation of the stems of T. uniflora also yielded one new -hydroxydihydrochalcone (134) and three known compounds (an isoflavone, -hydroxydihydrochalcone and a rotenoid). NMR, X-ray crystallography, UV spectroscopy, electronic circular dichroism and mass spectrometry were used to determine their structures.

The crude extract of M. sericea roots exhibited antiplasmodial effect against chloroquine-resistant (W2) (IC50 of 0.6 μg/mL) and chloroquine-sensitive (3D7) (IC50 1.8 μg/mL) strains of Plasmodium falciparum. Among the major compounds from this plant, lupinifolinol (129) (IC50 of 2.0 μM against the W2 strain, and 6.6 μM against the 3D7 strain), and mundulinol (64) (IC50 of 5.9 μM
against the W2 strain and 2.4 μM against the 3D7 strain) were active. The antileishmanial activity of selected compounds was tested against L. donovani strains, both antimony-sensitive (MHOM/IN/83/AG83) and antimony-resistant (MHOM/IN/89/GE1), of which sericetin (130) was active against antimony-sensitive (IC50 5.0 M) and antimony-resistant (IC50 38.0 μM) strains. Dehydrolupinifolinol (128) was also active against the antimony-sensitive strain (IC50 9.0 μM). The isolated compounds from S. usambarensis were tested against E. coli and B. subtilis.

Usambarin D (113) had moderate antibacterial activity against B. subtilis (MIC = 9.0 M), however, the other compounds examined were inactive (MIC >100 M). All the tested compounds were inactive against E. coli. Some of the identified compounds from S. usambarensis and M. sericea were investigated for their cytotoxicity against; lung (A549), breast (MCF-7), immortal human hepatocytes (LO2), liver (HepG2), and human lung/bronchus cells (epithelial virus-transformed, BEAS-2B) cancer cell lines. Usambarins A (EC50 of 65 μM) and B (EC50 of 92 μM) were weakly cytotoxic against the breast cancer cell line, while compounds 114, 115, 119, 120, and 122 were not cytotoxic to MCF-7 with EC50 > 200 μM. The current study has revealed that M. sericea, S. usambarensis, and T.uniflora possess a spectrum of metabolites with unique structural features with potential in the development of antimalarial, anticancer, antibacterial and antileishmanial agents.

Climate variability changes ultimately impacts on agriculture and subsequently food
productivity and security. In Kenya, milk production is predominantly smallholder and
dependent on rain fed agriculture. To ensure that dairy farmers are empowered to effectively
prepare, adapt and mitigate the effect of extreme climate changes, this study aimed at
investigating the effect of climate change on milk production in smallholder farms; a case of
Nandi county, Kenya. Primary data sources of data included observed and climate model
outputs (precipitation, maximum and minimum temperature), fodder availability (Normalized
Difference Vegetation Index –NDVI and Soil Moisture), milk production (milk marketed).
Secondary data was sourced through structured questionnaires, focus group discussion, and
key informant interviews. The study used concurrent triangulation research design to allow
mixed-methods research methodologies. Trend analysis and spatial plots were used to analyse
spatiotemporal variability of past and future climate (2021-2050) was based on RCP45 and
RCP85. Relationship between climate and milk production were based on correlation and multi
regression analysis. Graphs and pie charts were also used to present the results.
Past and projected precipitation showed bimodal patterns with high spatial and temporal
variability with remarkable differences between baseline and projected precipitation under
RCP45 (-19.5% to 11.0%) and RCP85 (-9.5% and 26.3%) scenarios. Past and projected
maximum and minimum temperatures showed increasing trends. Monthly NDVI and soil
moisture values were higher in April and November while seasonal values were high/low in
JJA/DJF indicating high/low fodder availability. Milk production showed positive change from
2007 to 2016 with highest/lowest values in April/December. Computed percentage change in
seasonal milk production showed increases of up to 186% (MAM), 183% (JJA), 202% (SON),
and 214% (DJF) whereas annual milk production showed increases of up to 204%. The
lowest/highest correlation coefficients were found in precipitation/minimum temperature at lag
0, 1 and 2 while the selected models based on different predictors based on climate and fodder
availability showing positive relationship with milk production.
Over 79% of household involved in milk production in Nandi County are male headed.
Although drought was found to be the leading climate hazard affecting their grazing practices,
other factors such as rainfall variability, rainfall unpredictability and extreme temperatures also
affected grazing practices. The survey results indicated that observed changes in milk
production, the amount of water available for the animal, body condition of the animal, heat

detected and growth of calves and heifers were negative in almost all the wards in the County
implying that climate change had negatively impacted on dairy productivity. The most
important source of animal feed were natural pastures mainly from farmers own farms (86.9%),
crop residue (62.6%), planted fodder such as Nappier grass (39.4%), communal land (19.2%)
and others purchased their fodder (16.2%) with most farmers depended on natural pasture from
their farms (76.5%).
Majority of farmers planted fodder in less than 0.5 acres of land for Napier (79.7%), Sorghum
(54.3%), Rhodes grass (57.3%), Kikuyu Grass (49.4%), Lucerne (71.9%) and fodder Tree
(82.1%) and conserved/preserved crop residue (88.2%), hay (39.9%) and silage (35.4%).
Communal lands were noted to be overgrazed and very little fodder was available with the
grass growth not beyond one foot. Methods used to address negative experiences of climate
change included use conserved hay/silage (44.2%), buying of commercial feeds (40.9%), use
crop residue (74.6%), moving of animals to other farms (8.8%) and selling of animals (17.4%).
Other measures adopted by households to help them avert negative climate change included
use of new fodder types/varieties, new planning methods, intercropping different fodder,
conservation and preservation practices. Smallholder farms had also adopted climate smart
agricultural technologies such as compost making (18.6%), use of biogas (2.5%), water
conservation (56.6%), disease control (95.4%), planting of fodder trees (30.1%), reducing the
number of animals (36.6%) and breeding using AI (63.4%).
The study findings indicate that dairy productivity is highly sensitive to climate. Moreover,
fodder availability which is also vulnerable to changes in climate significantly influences milk
production. Given the high spatial and temporal variability in these environmental factors, it is
expected that the projected change will significantly challenge future dairy productivity
especially in Nandi County of Kenya. The study recommends the need to improve on
monitoring of weather and climate by increasing observation stations and development weather
and climate products targeting milk production. There is also need to develop climate smart
fodder varieties/production methods and adoption of the use of climate smart fodder
varieties/production methods. Moreover, policy makers need not only to promote the use of
climate smart fodder varieties/production methods but also mainstreaming climate change
information into development planning, budgeting and implementation at national and county
levels.

Nuclear forensics (NF) is a systematic and scientific methodology designed to identify, categorize, and characterize seized nuclear and radiological materials (NRM). The aim of NF is to reveal the geographical origin, process/production history, age and intended use of the NRM to prevent future diversions and thefts, thereby strengthening the national security of a country. The limitation in NF at the moment is the absence of suitable methodologies to directly, rapidly and non-invasively analyze limited size of NRM under concealed conditions. Laser based spectroscopy and spectral imaging techniques (laser induced breakdown spectroscopy (LIBS) and Laser Raman microspectrometry (LRM)) combined with machine learning techniques (ML) possess the power to conduct direct, rapid NF analysis of limited size NRM with accuracy and precision. The uranium lines at 386.592 nm, 385.957 nm and 385.464 nm were identified as nuclear forensic signatures of uranium in uranium trioxide bound in cellulose and uranium ore surrogates (uranium mineral ores and high background soil samples). The detection limit for uranium in cellulose was determined at 76 ppm. Uranium lines were
divided into resonant and weak lines, depending on the signal-to-background ratio. Resonant and weak uranium lines were utilized to develop multivariate calibration models in artificial neural network (back-propagation algorithm). The calibration models employing weak U-lines and resonant U-lines predicted uranium content in the certified reference material (CRM) (RGU1(400 ppm)) with relative error of prediction (REP) at 4.32% and 9.75% respectively. The model using weak U lines predicted the uranium concentration in the mineral ores (uranium) obtained from various parts of Kenya utilizing weak uranium between (103 - 837) ppm. The models were further validated with RGUMix (101 ppm) in addition to RGU-1 (400ppm). The calibration model using weak U-lines predicted the uranium concentration in RGUMix (101 ppm) and RGU-1 (400 ppm) at REP = 2.97% and 2.14% respectively, while using resonant U-lines predicted at REP = 69.07% and 4.22% respectively. The poor sensitivity of the resonant lines to changes in the low concentration uranium may account for the high REP of the model using resonant U-lines. On successful validation of the calibration model utilizing
weak U-lines, the uranium concentration in the uranium mineral ores was predicted ranging from (112 - 1000) ppm. LIBS spectra of HBRA soils of Kenya combined with principal component analysis revealed patterns that related to their origin. The uranium mineral ores collected from various parts of Kenya were successfully grouped into their mineral mines (origin) applying PCA to selective spectral regions. NF signatures associated with uranium molecules in uranyl nitrate, uranyl sulphate, uranyl chloride and uranium trioxide samples were identified at 865 cm-1, 868 cm-1, 861 cm-1, and 848 cm-1respectively using LRM (laser λ= 532 nm, 785 nm). Spectral imaging using these signatures on sample spiked with trace uranium (150ppm), HBRA soil samples and uranium mineral ore samples demonstrated the distribution of uranium molecules. Thus, ML techniques, in combination with laser-based techniques and
spectral imaging techniques, have the potential to not only perform rapid, direct, minimally intrusive qualitative and quantitative analysis of trace uranium (typical of NF), but also aid in the attribution of uranium ore surrogates to their origin and distribution of uranium molecule through the different layers of these samples.