To evaluate the spatiotemporal dynamics of land use land cover changes (LUCCC) in Kapkatet wetland, imageries for the years 1986 (Landsat 4-5 TM), 2000 (Landsat 7 ETM+), and 2019 (SENTINEL) were obtained from USGS/ESA databases. All the acquired images were referenced to the Kenyan Coordinate System using UTM projections and WGS 84 datum on Arc GIS 10.2 software. Considerations made while obtaining these images were on seasons over which they were captured. Only those of dry seasons across the three years were utilized in this study. This is because during dry seasons, there is minimal cloud cover hence higher precision, and also various anthropogenic activities are rampant particularly on the wetland during such periods [25].

Pre-processing and processing activities of Landsat and Sentinel images were done using Arc GIS 10.2. This involved the process of geometric image correction for clarity over the identified periods (1986, 2000 & 2019). For digital image processing, false colour composites were created using bands 2, 3, and 4 for each of the images. The images were then geo-referenced in the UTM projection WGS84-reference ellipsoid. GIS data which consisted of Global Positioning System (GPS) ground-truthing data were entered into ArcGis 10.2 software. Supervised classification of the satellite imagery was used to produce land use wetland cover classes. The maximum likelihood classification technique was performed using the spectral bands 2, 3, and 4 in each of the satellite images. Feature extraction, selection of training data, and selection of suitable classification approaches were carried out to generate each of the wetland cover classes of the study area. Accuracy assessment was achieved by comparing samples of pixels from the classification results and the ground-truthing data collected using a GPS machine from the field.

Overall accuracy was calculated by dividing the number of pixels for a particular class with the overall number of reference pixels for that particular class. To derive the information on the wetland cover and wetland change over these three periods, four distinct classes (Table 1) were only chosen from the wetland regions. They were selected based on observable features identified in the photographs captured during the field survey and ground-truthing exercise. Post classification comparison method was used for change detection whereby images of different dates were first classified and labelled individually and the classified

images were then compared and changed areas extracted. An error matrix method was used to check for classification accuracy. All the other computations were done using ArcGIS 10.4 software. Variations in terms of wetland cover classes were deduced from the findings with varying percentages of classes over the past three decades. Computation for wetland area change over the periods (1986-2019) was done using the following method;

% Δ = Wetland area Δ ( 2019 − 1986 ) Original Wetland area ( 1986 ) × 100

where; %∆ is Percentage Change.

Household Survey: Both structured and semi-structured questionnaires were used to obtain information on the drivers of wetland encroachment. The purposive sampling technique was adopted to identify households located within a radius of ≤1 km from the wetland and those whose parcels of land extend to this wetland. Because of time and resources, respondents (n = 100) consisting of an adult in each of the identified households, willing and capable of providing the required information, were allowed to fill in the questionnaires with the help of two trained field assistants. According to the gender disparities, 69 were male and 31 were female drawn from the two counties. In cases where an adult was not available or not willing to be interviewed, the next household was chosen. The data from the household survey were entered and coded using SPSS software and thereafter, quantitative data subjected to descriptive statistics to obtain frequencies and percentages.

Focused Group Discussions (FGDs): This study utilised four FGDs, three in Bomet County (Chesilyot, Togomin and Cheptangulgei Villages) and one in Kericho County (Rungut village). The number of FGDs was higher in Bomet County, because a larger portion of the wetland is located in this county. The villages selected represented the entire region of the study area. A total of 40 residents participated in the four FGDs consisting of youths, local leaders and outstanding men and women in the community. The researcher liaised with both the village elder and the assistant chiefs in finding suitable places for holding these discussions. Moderators were chosen to guide the discussions.

The coverage of reeds/papyrus in 1986 (2,362,500 m²) was higher than the one for the year 2019 (1,349,400 m²), showing a loss over time. The coverage of this

LULC has been on a decreasing trend for the last three decades, decreasing by 5% between 1986 and 2000 and further by 11% from 2000 to 2019. The population of the Kapkatet region just like any other part of the country has been the major driver which contributed to the change. Demand for more food by this population has necessitated the expansion of farming lands, thus contributing to the degradation of wetland vegetations. [26] also reported a similar trend in their study where there was a loss of papyrus in Dunga (50%), Koguta (47%), and Kusa (34%) between the years 1964 and 2000. An elderly participant during an FGD in Chesilyot village reported having been observing a declining trend of birds scavenging on this wetland over the past three decades. This is an indirect parameter that could be linked to the decline in reeds/papyrus cover. Planting non-native vegetations like eucalyptus trees could also be attributed to the decline in papyrus/reeds cover. [27] acknowledged that eucalyptus species planted on riparian zones have invasive characteristics that alter with the floristic diversity, stand structure and the diversity of wetland vegetations.

Similar to reeds/papyrus, tree coverage initially recorded in 1986 (431,100 m²) was determined to be higher than the one recorded for 2019 (275,500 m²), an indication of tree cover loss. From 1986 to 2000, tree cover marginally increased by 1% before eventually decreasing by 3% between 2000 and 2019. The most predominant exotic vegetations along the periphery of this wetland are the eucalyptus trees. Owing to its fast growth and market demand forces, eucalyptus farming in this area is becoming a lucrative business. The external market demand from mushrooming industries such as the Kenya Tea Development Authority (KTDA) owned factories namely, Kapkatet, Litein, Chelal and Tebesonik; private tea factories like Korara and Bureti tea factories mainly rely on the locals for the supply of firewood [28]. This explains why there was a decrease in tree cover between the years 2000 and 2019. Additionally, in recent years the government through its devolved function of the environment has reinstated and enforced the 2009 regulations on the use and management of wetlands, riverbanks, lakeshores and seashores [23].

The coverage of open grounds however has been increasing over the years. The coverage in 1986 (354,600 m²) was found to be lower than the coverage for 2019 (889,300 m²), hence an increased open ground. Generally, open ground recorded a 3% increased coverage between 1986 and 2000 and a further 21% increased coverage from 2000 and 2019. The expansion of open grounds was noted to be as a result of crop cultivation and burning of wetland vegetations especially during dry seasons as part of the pre-fallowing process engaged by some of the residents whose lands extend to this wetland. Zones adjacent to the wetland were also found out to be dominated by exotic trees, tea bushes or horticultural farms, a similar feature reported by [22] on this wetland. This is an indication that natural vegetations which initially dominated the watershed areas were destroyed and replaced by exotic species including crops.

The extent of coverage of the papyrus and reeds experienced a percentage change from −5% between the years 1986 and 2000 to a percentage change of −11% in the periods between 2000 and 2019. This change is attributable to the increased percentage change of open grounds from 3% (1986-2000) to 21% (2000-2019) which accounted for the decline in the papyrus and reeds. This indicates that this wetland has over time been encroached hence affecting the spatial extent of the vegetation cover in the wetland. Anthropogenic activities carried out in the area have also greatly contributed to the dwindling of this wetland. An elderly man during FGD in Chesilyot Village highlighted that it was a norm in the community to undertake cultural festivals for example, initiation ceremonies which were conducted annually around December in specific localities within this wetland. Also, he further mentioned that initiates used to construct their huts using reeds and twigs sourced from the wetland. There was a need therefore to preserve this ecosystem which comprehends what was observed through satellite imageries where reeds covered a higher percentage (68%) as compared with open grounds and the disturbed reeds that covered 10% and 9% respectively.

Study results indicated that 49% (n = 100) of the respondents perceived land conversion as an impetus of wetland encroachment in the study area. Field visits undertaken during dry periods were a true manifestation of how rapidly this wetland was being converted for economic use. Wetland vegetations especially papyrus reeds were being burnt by different landowners in favour of the growth of non-perennial crops, livestock grazing fields, the establishment of woodlots,

fish ponds and harvesting of soils for brick-making. These findings were similar to what was reported in the Nyando Wetland where such economic activities eventually contributed to the loss of its biodiversity, altered fish breeding grounds and bird habitats [29]. Horticultural farming in some sections of the wetland was also noticeable with the majority of the FGD participants claiming the availability of water and fertile soils in such areas during dry periods as motivation for farming. This is in line with the study conducted by [30] who also found out that the major reasons for wetlands degradation in Tanzania were a result of population pressure that led to the demand for more agricultural land. Studies conducted in Tempisque Basin by Daniels & Cumming [31] also showed that demand for more land contributed to the conversions of the wetland in areas where they were most accessible and unprotected.

On the drivers of encroachment, 27% (n = 100) of the respondents indicated the introduction of invasive species as a factor of encroachment. Most of those who had lived in the area for more than 20 years (76%, n = 100), stated that vegetation on this wetland indeed had changed over the past decades. Furthermore, during FGD discussions, one of the participants who happened to be elderly noted also that papyrus reeds which initially were dense, are now in patches with grass, exotic trees, and shrubs taking over the entire wetland region. This information buttressed the findings from satellite imageries that showed that wetland size and vegetations have declined in the past three decades. This finding is also consistent with that of [32] who noted that the introduction of invasive species was one of the drivers of wetland change. A similar problem was also experienced in the riparian zones of the Western Cape of South Africa where eucalyptus among other invasive species led to the degradation of the wetland [27] [33]. The ability of eucalyptus trees to alter the vegetation structure and wetland hydrology has also contributed to the reduction of food resources for the avifauna in the area. The allelopathic nature attributed to eucalyptus comes with new ecological niches which appear to be benefiting alien species over the native ones. Invasion of wetland grasses was also noticeable in some parts of the wetland due to the increased use of NPK fertilisers by tea farmers. This corroborates with the findings by [22] [34].

Eutrophication especially in Sub-Saharan Africa poses a threat to surface waters including wetland ecosystems. According to the household respondents, 10% (n = 100) attributed this factor as among the drivers of wetland encroachment in the area, citing that indeed some of the residents have established tea plantations along the periphery of this wetland. The majority of the farmers in the area engage in tea farming [23] which requires a constant supply of fertilizers containing supplements of nitrogen, phosphorus, and potassium (NPK). Considering the terrain and drainage pattern in the area, often much of the runoffs from these farms laden with traces of NPK find their way to the wetland, changing its vegetation regime [22] found out the pH levels on this wetland were low, a factor which he attributed to high levels of chemicals especially nitrogen and phosphorus emanating from the adjacent tea farms. Dissolved wastes from livestock droppings while grazing on the wetland vegetations also contributed significantly to eutrophication in this area [22] [35]. The findings are also consistent with those of [36] who indicated that Las Tablas de Daimiel wetland has been impaired by eutrophication, salinisation and pollution as a result of discharging municipal and agricultural wastes toward the wetland. Siltation due to marram road was observed in sections close to a footbridge heading to Kapletundo Ward which had converted a section of the wetland into a field. It was further observed that solid wastes carried by water runoffs from Kapkatet Town were being deposited on the site which is attributed to the continued decline in wetland species. Car washing activities were also highlighted by one of the FGD participants at Chesilyot as a major contributor to water pollution. Laboratory analysis of water from Daraja Sita, a section on this wetland, showed that it had high conductivity rates than all the other sampled areas [22].