Water Hyacinth's Extent and Its Implication on Water Quality in Lake Victoria, Uganda

Water hyacinth (Eichhornia crassipes) degrades and obstructs the integrity of freshwater ecosystems. However, little attention has been paid to monitoring water hyacinth's spatial extent, its determinants, and its effects on water quality in Lake Victoria, Uganda. The specific objectives of this paper are to (i) assess the spatial extent and distribution of water hyacinth; (ii) examine the determinants of water hyacinth distribution, and (iii) assess its impact on water quality. High-resolution satellite images (2016–2019) were obtained and used to monitor the spatial extent of the water hyacinth, a household survey was conducted to examine the determinants of the water hyacinth's extent and patterns while water samples were drawn and analysed for physicochemical properties. Results show that the coverage and distribution of water hyacinth varied over space and time. Water hyacinth coverage primarily increased with a decrease in water surface area. The perceived factors that triggered the water hyacinth spread included the morphology of the Bay, effluent discharge, strong winds, speed of water current, water-level changes, ferry navigation, and construction activities at the shore. Water parameters significantly impacted by hyacinth were pH, TP, BOD, COD, DO, turbidity, and transparency. This study recommends the strict development and implementation of integrated weed control measures, catchment management plans, and point and nonpoint pollution source control.


Introduction
Water is an irreplaceable and indispensable natural resource, vital for life on earth, economic development, and human well-being [1,2]. Although 71% of the earth's surface is water, not all the water is accessible and suitable for all uses. Useable water is meagerly available in fresh water streams and inland lake systems. However, even in these conditions, water quality is continuously deteriorating, thus raising sustainability concerns [3]. Te water quality deterioration due to pollution is currently the principal challenge to water resource management [4][5][6]. Water is polluted if it cannot serve a particular purpose resulting from processes that alter its physical, chemical, and biological constituents as it moves through the various spheres of the hydrological cycle [7].
Physical, chemical, and biological constituents defne water quality and its suitability for various uses [8]. Tese components are afected by several factors, including storm runof, nitrifcation from decayed matter, water hyacinth, toxic and hazardous substances, oils, grease, litter, rubbish, and land use such as industrialization, farming, mining, and forestry activities, which signifcantly contribute to water quality degradation [9][10][11]. Land uses either increase the concentration of nutrients or suspended materials (as is the case with agricultural land use) or increase the supply of heavy metals and toxic substances in water (like is the case with industrial activities) [12].
Among the biological sources of water quality deterioration is the eutrophication from aquatic plants [3] such as the water hyacinth (Eichhornia crassipes), a free-foating perennial monocotyledonous plant belonging to the family Pontederiaceae. Tis hydrophyte possesses the potential to alter water nutrient cycles and impact aquatic life [13,14]. Te water hyacinth degrades and damages freshwater systems, compromising water quality and threatening the quality of life [15]. Aquatic weeds represent one of the growing challenges for biosecurity and water resource management worldwide [16]. Te costs associated with the management of this aggressive waterweed are enormous. In Africa, the damages are estimated at an annual cost of $100 million [17].
Risks related to aquatic weeds have been on the rise due to climate change efects and increased nutrient enrichment, as well as other organic and inorganic pollutants from various anthropogenic activities [18]. Despite threats posed by these weeds and their relative increase in spatial coverage, there have been minimal monitoring and management eforts. Besides, their spatial distribution and confguration remain poorly quantifed and less understood particularly in less developed economies [19,20]. Timely detection and up-to-date information regarding water hyacinth distribution are crucial in understanding its spatial confguration and propagation rates [21]. Monitoring and mapping the spatial confguration of water hyacinths are necessary to provide essential information for proper mitigation and control and ensure the continued provision of goods and services by the water bodies under such threats [19]. With the recent developments in remote sensing science and geographical information technologies, it is possible to undertake such resource assessment and monitoring tasks with ease. Tese technologies enhance our ability to acquire spatial data and study and map landscape features such as vegetation for timely inventory and assessment of such resources [22]. Satellite data can capture the spatial and temporal distribution of aquatic macrophytes in a timely and cost-efective approach [23,24].
Previous studies in the Lake Victoria basin have focused on urban eutrophication and its spurring conditions, the socioeconomic impact of water quality deterioration, and the impact of efuent discharge on water quality [25][26][27][28][29]. However, little attention has been paid to monitoring the spatial extent of water hyacinths, their determinants, and their efects on water quality in Lake Victoria. Moreover, studies detecting the spatial distribution and confguration of water hyacinth involving the use of Geographical Information Systems (GIS) and relatively high-resolution remote sensing data such as Sentinel-2 imagery are scanty in the region. With the help of remote sensing and GIS tools, essential information for proper mitigation and control of the waterweed can be acquired and thus reduce contamination levels of the water in the lakes. Terefore, the specifc objectives of this paper are to (i) assess the spatial extent and distribution of water hyacinth; (ii) examine the determinants of water hyacinth distribution, and (iii) assess its impact on water quality in Lake Victoria, Uganda. It is therefore imperative to map the distribution and assess the efect of this alien aquatic plant species on the quality of water in the lake, so that appropriate control and management measures are implemented to keep contamination at unproblematic levels.

Description of the Study Area.
Tis study was conducted in the Murchison Bay which is part of Lake Victoria. It covers parts of Kampala city and Mukono and Wakiso districts in Central Uganda. Te Bay stretches between latitudes 0°13′5″ N-0°18′67″ N and longitudes 32°36′59″ E-32″40′27″ E, forming an extension of Lake Victoria ( Figure 1). Lake Victoria is located in the south east of Kampala city, lying between latitudes 0°10′00″ N-0°30′00″ N and longitudes 32°35′00″E−32°50′00″ E with an average elevation of 1,224 meters above sea level. Temperatures around the Bay range from 25 to 32°C while winds are around 6.9 km/h north [30,31].
Te Murchison Bay covers an area of about 62 km 2 but with a catchment area of approximately 282 km 2 . Te depth of the Murchison Bay in 2004 was 7 meters, but by 2008, it had dropped by 1½ meters [26,31]. Te Bay is further split into inner and outer sections as their characteristics difer tremendously. Te inner Murchison Bay is a semienclosed small water body with an area of 25 km 2 and a length of 5.6 km of the main lake section. Tis section is relatively shallow with an average depth of 3.2 m but deep towards the main lake area with a convoluted shoreline and narrow at the exit to the outer Murchison Bay. Tese facilitate the mixing of water between the inner and the outer Bays [25]. Te inner Murchison Bay forms the main abstraction point for portable water supplied to the expansive population around Kampala city.
Te major channels/wetlands that drain into the Murchison Bay include Nakivubo, which drains Kitante and Lugogo channels with inlets into the inner Murchison Bay; Kansanga wetland, which stretches into the Ggaba shoreline; Kinawataka, which drains industrial centres of Nakawa and Kyambogo; and Namanve wetland [28].

Spatial Extent and Distribution of Water Hyacinth.
High-resolution satellite images covering the Murchison Bay were acquired from Sentinel-2 archives manned by the United States Geological Survey (USGS) (https://glovis. usgs.gov/web-link). Te images were for the period between 2016 and 2019 with Sentinel-2 MSI tiles covering the study area. A single image was downloaded for each year and this had to be of the dry period (between January and March), during which there is less cloud cover to mask ground features. Images selected were those with less than 5% cloud cover as image analysis targeted the visible bands (RGB and IR). Sentinel-2 images were preferred to Landsat data due to the high spatial resolution of the former (Sentinel with bands in 20 * 20 meters) compared to the latter (Landsat with 30 * 30 meters). Since the launch of its frst satellite in 2013, Sentinel data have become more and more applied in landscape mapping, thus serving as an alternative to coarse resolution Landsat series data [22]. Te images were atmospherically corrected using the Dark Object Subtraction (DOSI) model under the semiautomated classifcation (SCP) embedded in Quantum GIS (QGIS) 3.12 software.
To determine the pattern and distribution of water hyacinth in the Murchison Bay, the preprocessed Sentinel-2A images were further processed using the maximum likelihood supervised classifcation algorithm in QGIS. Te model distinguishes pixel properties for diferent land uses and cover (for which water hyacinth was part) based upon input training data of pixels representing the predefned land use/cover classes (Table 1). Based on these data, the algorithm groups the remaining pixels on an image into the created classes. Te maximum likelihood classifcation model was selected for the satellite imagery classifcation in this study because of its high precision in land use and cover classifcation as reported in previous studies, e.g., [19,22]. Moreover, Sentinel-2 data had never been applied in water hyacinth studies in the Murchison Bay.
In addition, feld data collection was conducted to record the location of the water hyacinth using GPS (primary data) during November-December 2019 and January-February 2020. Tese were randomly generated sampling points across the Murchison Bay, following water hyacinth-infested areas. Tese points were used in a training data set for mapping the extent and pattern of water hyacinth.
Te postprocessing of the classifed Sentinel-2 images involved the computation of areal statistics for the cover classes for the images corresponding to the study period (2016 to 2019). Using discriminate analysis, the various changes in coverage of the water hyacinth vis-à-vis other covers in the Murchison Bay were determined, which indicated the pattern and distribution of the water hyacinth in the Bay over the study period. Te results are presented in tables and graphs. Te QGIS semiautomatic classifcation plug-in allows for the extraction of several classifcation accuracy statistics such as overall accuracies, user's accuracy, producer's accuracy, and kappa efciency (Semi-Automatic Classifcation Plugin Documentation, release 5.3.2.1. 2017).

Perceived Determinants of Water Hyacinth Distribution.
Te study adopted a cross-sectional research design to establish determinants of water hyacinth extent and pattern in the Murchison Bay as perceived by the residents. Te design followed a quantitative approach to gathering data from respondents using structured questionnaires. Te targeted respondents' categories included ofcials from the Fisheries Department and National Water and Sewerage Corporation (NWSC) and traders and fshermen stationed at Port Bell, Ggaba, and Mulungu landing sites. A sample of 201 respondents from the abovementioned categories was drawn following purposive and stratifed sampling techniques. First, the respondents' categories were defned on the criterion that they are involved in water resource management and are directly afected by water hyacinths and on the fact that they are more knowledgeable about the problematic waterweed (water hyacinth) in their areas of jurisdiction.  Secondly, a stratifed sampling technique was employed to select respondents from three landing sites around the Murchison Bay. Sixteen respondents (16) were selected from the Fisheries Departments at Ggaba, Mulungu, and Port Bell landing sites, respectively. One hundred and twenty (120) respondents were randomly selected from the three landing sites and 13 respondents from the National Water and Sewerage Corporation at Ggaba.
After determining the target respondents, semistructured questionnaire copies were hand delivered to collect participants' perceptions of the physical and human factors responsible for water hyacinth distribution in the Murchison Bay over the years. Te main section of the questionnaire required respondents to rank the factors that they thought to infuence the water hyacinth pattern and distribution in the Murchison Bay. Up to 15 factors were presented for ranking on a scale of 1-4 to show the extent to which a factor determines water hyacinth extent and distribution in the Bay (where 1 indicates the least level and 4 indicates the highest level of determination). Data obtained were computer coded in the Statistical Packages for Social Scientists (SPSS) computing program, version 23.0. Te data were then analysed using both descriptive and inferential statistical techniques. Mean and standard deviation statistics were generated from the rated responses, and the results are presented in tables. Pearson's chi-square (X 2 ) test was performed to establish whether the rated factors were related to water hyacinth spread and distribution across the Bay. Te relationships were tested at alpha level 0.05. Te analysis yielded Pearson's chi-square, likelihood, and p value statistics, which are also presented in tables.

Efects of Water Hyacinth on Physicochemical Water
Quality Properties. For water physicochemical property analysis, water samples were collected from stationary foating water hyacinth areas and in water hyacinth-free environments (open lake). Ten pairs of sampling locations were determined, corresponding to the two environments. Te sampling points had to be located at an average distance of 500 meters from one another. From each sampling environment, three samples were drawn in relation to water depth (i.e., near the water surface, middle, and at the bottom) ( Figure 2). Water samples were then collected using a 1000 ml water sample collector and subsamples were poured into 500 ml plastic water sample bottles (Figure 3), which were stored in boxes before transportation to NWSC laboratories at Ggaba and Lubigi, for the analysis of specifc water quality parameters of interest in this study. Te bottles were washed with nitric acid to remove any form of contaminants and to ensure that the physical properties of the water samples were maintained. Te parameters of interest included pH, water temperature, total phosphate (TP), dissolved oxygen (DO), biochemical oxygen demand (BOD), electrical conductivity (EC), chemical oxygen demand (COD), turbidity, and transparency. Tese were selected specifcally because they are key indicators of overall water quality and thus impact human health, water production, and ecosystem health [3,12,32].
Te sampling locations were accessed using a motorized boat, and at each sampling point, coordinates were recorded using Garmin Global Positioning Systems, whereas sampling in the open-water environment was done randomly and sampling in the water hyacinth environment was done purposively and systematically (following 500 m mean distance interval). Two diferent sampling occasions were conducted. Te frst sampling activity was conducted between September and December 2019. Tis period represented samples for the wet season of the study area climate zone. Te second sampling activity was conducted between January and February 2020. Tis period represented sampling for the dry season. Tus, a combination of data from two diferent seasons accounted for any variations in water quality brought about by seasons (November/December 2019 and January/February 2020).
DO, temperature, and transparency were tested and recorded in the feld, while turbidity, pH, TP, EC, BOD, and COD were tested in the laboratory using set standard procedures [33,34] (Figure 4). Temperature and DO were measured using a dissolved oxygen meter. Te device was Table 1: Te land cover/use types' classifcation system used for the Murchison Bay area.

Land cover/use class Description
Built-up/settlements Land consisting of residential areas, commercial buildings, and slums and associated infrastructure such as roads Burnt/bare earth Areas with burnt vegetation and/or exposed earth as a result of vegetation removal Lake Areas covered by lake water in the Bay Forest Areas under naturally existing and/or planted tree cover Water hyacinth Areas covered by the water weed and host wetland vegetation within the water body  immersed in the collected water sample, and the results for both parameters were displayed on the device's digital screen. Te temperature was recorded in degrees whilst DO was recorded in mg/L. Transparency on the other hand was measured using a Secchi disk, where the device was dipped into the water at every sampling point and the depth at which the disc was no longer visible was recorded in meters [35]. While in the laboratory, pH and EC were measured by the electrometry method using a pH/EC multimeter (Hach Sension + MM374). Tis device has two probes, one for measuring pH and the second for measuring conductivity. 100 ml of the sample was poured into a 100 ml beaker and the probes were lowered into the sample before starting the machine. Te sample was stirred using a magnetic stirrer until a stable reading was obtained and displayed on the equipment display screen. Te device displays both the pH and EC (mS/cm) values, which were recorded. pH has no units while EC was measured in mS/cm.
Te turbidity of the water samples was determined using a turbid meter (Hach TL 2300). Te sample was uniformly mixed and poured into a 40 ml cell up to the mark and then inserted into the machine to read of turbidity values in nephelometric turbidity units (NTU) displayed on the device's screen. COD determines the amount of oxygen required for the oxidation of organic matter using a strong chemical oxidant such as potassium dichromate under refux conditions [36]. Tis test is widely used to determine the same types of pollution as the BOD expressed in milligrams per litre (mg/L). COD was determined by the oxidation of organic matter using acid dichromate solution, followed by spectrophotometric determination. Te digestion tube and caps were washed with 4 ml H 2 SO 4 to prevent contamination. Two ml of the sample was poured into the digestion tube, followed by adding 2.0 ml of potassium dichromate digestion solution. Te abovementioned process allowed an acid layer to be formed under the sample digestion layer. Cap tubes were swirled several times to mix completely, without inverting the tubes. Te solution was placed in a preheated oven of 150°C for 2 hrs. Tis was followed by reading the concentration of the sample with the help of a spectrophotometer DR 6000.
BOD measures the amount of oxygen consumed through the biochemical degradation of organic carbon, inorganic materials, and nitrogenous compounds present in waste water over a specifed incubation period usually 5 or 7 days. It was determined by the preparation of dilution water by transferring a desired volume of water into a bottle and then saturating the water sample with DO by aerating with organic-free fltered air, adding 1 ml of each phosphate bufer, MgSO 4 , CaCl 2 , and FeCl 3 solutions/l of saturated water, mixing thoroughly before starting to use, while the preparation of DO was conducted by adding the specifc volume of the sample to the individual BOD bottles of known volume [37], flling the bottles up to the brim with sufcient dilution water, reading DO1 using the dissolved oxygen meter (Hach), then taking the initial reading, and tightly sealing the bottle leaving no air bubbles and Te Scientifc World Journal incubating for 5 days at 20°C. After the 5-day incubation, residual DO was determined in the samples.
To determine total phosphates (in mg/L), organically combined phosphorus and all phosphates were converted to orthophosphate. To release the phosphorus as orthophosphate from organic matter, a wet oxidation technique was applied. Tis was based on wet oxidation with potassium per sulphate. Te same procedure for orthophosphate determination was followed. Te procedure involved the following: taking 25 ml diluted or whole samples, acidifying with 1 ml H 2 SO 4 , 0.04 M, adding 5 ml digestion reagent, mixing thoroughly and preparing blank (25 ml distilled water) and phosphate standard by taking 25 ml of known standard concentration, and treating both the blank and phosphate standards in the same way as the sample.
Physicochemical property data obtained using both feld and laboratory methods were largely numeric and thus analysis involved the use of parametric statistical techniques. Tese data were organised in a Microsoft spread sheet and then imported into the R statistical computing environment. Using this program, frst, exploratory and descriptive statistics were computed including maximum, minimum, 1 st quartile, median, 3 rd quartile, mean, variance, and standard deviation for each of the physicochemical water quality properties. Tese were computed for the two data sets representing water hyacinth and nonwater hyacinth environments, and the analysis was meant to summarize the data and give a snapshot of the emerging diferences and similarities in the water quality parameters from the two sampling environments. In the second phase, data on the water quality parameters were subjected to two-way analysis of variance (ANOVA). Tat is, type III sums of squares [38] were computed on each of the water quality variables' data in relation to the sampling environment and water depth.

Spatial Extent and Distribution of Water Hyacinth.
Results from the satellite imagery classifcation indicate that, in 2016, water (42%) and built-up areas (24%) were the most predominant land use/cover types, followed by forest vegetation (15%) ( Table 2 Table 3) reveal that water hyacinth extent and distribution on the lake were highly infuenced by the sheltered morphology of the Bay, efuent discharge, strong winds, the speed of water currents, change in lake water level, construction activities at the shore, and ferry navigation (with average rating between 2.4 and 4) according to the respondents' opinions. However, water temperature, humidity, biotic colonization, hyacinth species, herbaria, water depth, fsh hatcheries, and fshing gear received rating scores below 2.0 which on a scale of 1 to 4 is below average in terms of their importance in infuencing water hyacinth distribution in the Murchison Bay. Te results imply that the majority of the respondents believe that much of the water hyacinth proliferation is due to man's infuence through sewage efuent discharge, construction works, and ferry navigation in the Bay.
Te Scientifc World Journal construction at the shore, fshing boats and nets, herbaria, and ferry navigation determined water hyacinth extent and distribution positively and signifcantly (p < 0.05) varied spatially across the Murchison Bay. Tis implies that these factors signifcantly infuenced water hyacinth spread and distribution, but in selected sections of the Bay, and thus not universally considered signifcant determinants. On the other hand, the perception that temperature, sheltered Bay, speed of water current, hyacinth species, water depth, and botanic gardens were important determinants was not signifcantly related to location diference in the Bay (p > 0.05). Tis result implies that water hyacinth spread and distribution in the Bay were equally perceived to be signifcantly determined by water temperature, sheltered morphology of the Bay, speed of the water, hyacinth species, water depth, and proximity to botanic gardens. Terefore, these factors are signifcant drivers of water hyacinth spread and distribution irrespective of the location in the Bay.

Efects of Water Hyacinth on Physicochemical Water
Quality Properties. Analysis of water physicochemical properties revealed higher pH in the open lake environments with values ranging between 7.3 and 10.8 as compared to that in water hyacinth environments (Table 4). Te water pH difered signifcantly between sampling sites and lake depth ( Table 5) which means that water hyacinth and water depth signifcantly (p < 0.05) afected water pH in the Murchison Bay. However, the interactive efect of these two variables was not signifcant, implying that the two variables afected water quality independently. Electrical conductivity average values in both water hyacinth-infested areas and open lake sites difered slightly (Table 4). However, the results indicated that water depth signifcantly afected the electrical conductivity of water. Te ANOVA results for the interaction of the two factors (environment and lake depth) revealed no statistically signifcant efect of these variables on water EC (p > 0.05) ( Table 5). Te results signify that water EC in the Murchison Bay was signifcantly altered by water depth rather than by water hyacinth infestation.
Te results also showed that water temperature in water hyacinth environments was slightly higher (27°C) than that in open lake water (26°C) on average (Table 4). However, ANOVA results indicated that although temperature varied between the water hyacinth sites, open lake, and lake depth, the diferences were not statistically signifcant (p > 0.05) ( Table 5). In addition, none of the interactions between the three factors had a statistically signifcant efect on water temperature. Descriptive statistics further revealed diferences in the DO in the three sampling environment categories. However, water hyacinth-infested sites registered lower DO compared to an open lake environment (7 mg/L vs. 9 mg/L). Te variations in the DO were signifcant for independent measurements related to lake depth and sampling environment but not statistically signifcant for the combined variables. Tis implies that water hyacinth deprived infested environments' water of DO.
Higher turbidity was also reported in water hyacinthinfested areas as compared to open lake sites. Te efect of lake depth on the other hand was not signifcant on turbidity (p > 0.05). Additionally, the interactive efect of the sampling environments and lake depth on turbidity was also insignifcant. Te results relate to the fact that the water hyacinth negatively contributed to water turbidity in the Murchison Bay. Similarly, in terms of transparency, water hyacinth-infested areas were less transparent as compared to open lake sites. In addition, the efect of lake depth, as well as the interactive efect of the two factors (environment and depth), was not statistically signifcant (p > 0.05) ( Table 5). Tis means that the water hyacinth increased the concentration of suspended materials in water which lowered transparency in the Murchison Bay.
Te efect of the water sampling environment on total phosphates was statistically signifcant (p < 0.05). However, the efect of water depth and the interactive efect of the three factors on TP were not signifcant. Further, the efects of sampling environment and depth on the concentration of TP were independent of each other. Tis result suggests that water hyacinth signifcantly afected the concentration of    (Table 4) and the efect of the two variables was statistically signifcant, but the efect of lake depth was not signifcant (p > 0.05) accounting for the variations in COD. Te interactive efect of the three factors was also not signifcant. Tis means that water hyacinth can account for an increase in BOD and COD in the Murchison Bay.

Spatial Extent and Distribution of Water Hyacinth.
Te extent and pattern of distribution of water hyacinth varied largely over space and time. Tis study reveals that the increase in water hyacinth extent over the Murchison Bay is related to the fndings in a study in [39] while assessing changes in water hyacinth coverage over water bodies in the northern Bangalore using Indian Remote Sensing Satellite LISS-II and III images of the years 1988-2001. Teir study indicated that the area under water hyacinth increased in the recent years which consequently reduced the area under open water. Te major areas of contention in the current study fndings with those of [39] are due to the fact that water hyacinth coverage changes alternated between increases and decreases over the years. Te present study established that the water hyacinth was mainly concentrated in the northern parts of the Murchison Bay. Tis revelation is also echoed in [40] who reported that the water hyacinth attained a maximum lake-wide extent of approximately 17,374 ha by 1998 on the northern shores of Lake Victoria to which the Murchison Bay belongs. Tis points towards the area of intervention in terms of control of the waterweeds. Te results also indicate that water hyacinth coverage largely increased with a decrease in water surface area, which means that water hyacinth reduces the exposed water surface for other environmental processes such as atmospheric water transfer.

Perceived Determinants of Water Hyacinth Distribution.
Tis study established that the sheltered morphology of the Bay, efuent discharge (sewage), strong winds, the speed of the water currents, water-level changes, construction activities at the shore, and ferry navigation strongly determined the water hyacinth pattern and distribution in the Murchison Bay as perceived by the respondents although with variations in the level of infuence. Tis revelation is directly implied in the report in [21] that currents constitute the dispersion of water hyacinth propagule and stolon which makes the weed get distributed and colonize new areas within a short time. Te speed of water currents is thus an abiotic factor for the colonization of new areas with considerable importance for the potential propagation of the infestation in a given territory. However, temperature and humidity insignifcantly infuenced water hyacinth extent and distribution in the Murchison Bay. Te overall temperatures and humidity over the Bay are however generally high (above 18°C and 70%, respectively) on average. Te current study also indicated that the infuence of water depth on water hyacinth extent and distribution is minimal as perceived by the respondents in the Murchison Bay catchment. However, previous studies [13] have shown that both the depth of the water and changes in lake water levels are important for the growth and expansion of water hyacinths. Te reports suggest that the plants have more roots in deep waters than in shallow waters, while the leaf area and the summer growth of the plant are greater in shallow waters [41]. Tis implies that whereas people in the Murchison Bay thought that water depth plays an insignifcant role in water hyacinth distribution and extent, the factor is crucial as even the results from mapping showed more concentration of the water hyacinth on the shores of the lake where lake depth signifcantly reduces.

Efects of Water Hyacinth on Physicochemical Water
Quality Properties. Tis study further assessed the efect of water hyacinth on water quality, and from the analysis, water hyacinth altered the aquatic environment just as was reported in [14] leading to increased water temperature, concentrations of turbidity, COD, and BOD. Temperature is a major determinant of many chemical reactions that take place in water [32]. Warmer temperatures (>25°C) hold less dissolved oxygen which is key for the survival of aquatic organisms but also increase the solubility and consequent toxicity of compounds such as zinc and lead [1]. Average water temperatures of 25-27°C recorded in this study are within the permissible limits of the World Health Organization (30°C) as well as those reported in similar studies [42]. Although diferent, the temperature variation in the two environments was not signifcant. Te higher temperature values under the water hyacinth could be attributed to the heat generated from the breakdown of organic matter, below the hyacinth [13]. In addition, the location of the Bay in the equatorial belt could be the reason for the small and insignifcant variation in temperature reported. Te water hyacinth was noted to lower dissolved oxygen. Dense water hyacinth mats not only interfere with free oxygen transfer between the water and the atmosphere but also limit the mixing of the water by wind, leading to lower levels of dissolved oxygen [14]. While high DO levels add taste to water, it also has a highly corrosive efect on water pipes when at extreme values [42]. Biological processes related to plant decomposition can lead to a reduction in the concentrations of DO [43], which could explain the lower DO levels in the water hyacinth environment. Concentrations between 5 and 10 mg·L −1 are ideal for the proper functioning of aquatic systems, which gives validity to the ranges reported in this study. Dissolved oxygen is dependent on water temperature and the biological demand of the lake system [3]. Temperature afects the ability of water to dissolve oxygen to solubility at diferent temperatures, in that, a lower temperature improves the dissolution of oxygen compared to a higher temperature. In the current study, higher temperatures were recorded under water hyacinthinfested areas which are accounted for by the fact that decomposing water hyacinth releases heat that warms the water and this reduces the dissolved oxygen [44].
Water hyacinth also increased the acidity of water in the Murchison Bay. Te lower pH values in areas infested by the water hyacinth could be attributed to the accumulation of carbon dioxide below the weed [20]. pH extremes negatively impact water quality as well as determine coagulants used in water purifcation and treatment [45]. pH values of more than 7 have been reported to change the taste of water and make its treatment more costly [46] while lower pH values (less than 5) are associated with the corrosion of metals due to the higher levels of acidity [47]. Tis damages metallic water distribution infrastructure such as pipes in addition to contamination of the water being distributed. Reference [48] reports that water pH determines the solubility (the amount that can dissolve in water) biological availability (the amount that can be utilised by aquatic life for chemical constituents such as nutrients (phosphorus, nitrogen, and carbons) and heavy metals (lead, copper, and cadmium). Te pH values reported in this study, although slightly diferent, were in optimal ranges within the permissible ranges by WHO standards between 6.5 and 8.5 [8].
Water hyacinth also increased turbidity levels of the lake water in the study area. Te biological and chemical reactions in water hyacinth-infested areas increase constituents within the water, which increases turbidity levels [49]. Te increased materials not only increase the costs of disinfection but also increase the risks of inhabitation by pathogenic organisms, which all complicate the process of water production as more chemicals are required for disinfection and coagulation [50]. Suspended materials also lead to the clogging of flters leading to increased run-time [51]. Suspended materials further attract metals such as lead, mercury, and chromium, as well as other organic pollutants, which deteriorate the quality of water. Te materials also lower transparency as they limit the amount of light transmissible through the water. Tis is consistent with the fndings in [15] that water clarity can be greatly modifed by the waterweed as well as the reduced concentration of nutrients such as nitrogen and phosphorus. Decreased transparency has also been associated with a breakout in algae bloom because of increased photosynthesis rates, hence requiring more chlorine during peroxidation. Te mean turbidity value obtained for this study is very high when compared with the WHOrecommended value of 5.00 NTU, and this can lead to an increased demand for chlorine during disinfection [51].
Findings revealed low electric conductivity of water under the water hyacinth-covered lake areas. Te low EC values (142.3 mS/cm) are indicative of relatively good water quality and an indicator of low total dissolved solids (TDS) as has been reported in related studies like [42,52]. Te signifcance of electrical conductivity is in its proportion of saltiness, which enormously infuences the water taste and, in this manner, signifcantly afects the convenience of water. Besides, it is associated with high levels of corrosiveness, which is also detrimental to the metallic water production infrastructure. Te WHO standards provide for an EC value of not more than 400 mS/cm [53].
Although the BOD and COD levels in the water hyacinth environment were slightly lower, the variation in open water was insignifcant. Increasing values of COD are often caused by increased organic matter in the water [54]. Similarly, water with a high COD has a high chlorine demand and requires high chlorine dozes to fully disinfect [27]. Tis is also because high COD values are associated with increased organic pollution in the water [29]. Te more organic material there is in the water, the higher the BOD used by the microbes will be [1]. Te fact that BOD was higher in underwater hyacinth environments is proof that the water hyacinth increases the need for oxygen in water [46]. Water with high BOD requires increased use of coagulants to achieve efective clarifcation, thus an important indicator of overall water quality. While unpolluted water has BOD values of less than 5 mg/L, values up to 32.85 mg/L as reported in this study show some level of pollution in the water requiring a higher level of treatment.
While some studies have demonstrated the potential use of the water hyacinth and other aquatic plants in reducing suspended solids, dissolved solids, electrical conductivity, hardness, biochemical oxygen demand, chemical oxygen demand, dissolved oxygen, nitrogen, phosphorus, heavy metals, and other contaminants [41,54,55], the fndings from the present study indicate otherwise. In [13], water hyacinth is reported to have signifcantly increased water conductivity and total dissolved solids. However, these contradictions could be explained by the fact that the water hyacinth performs the purifcation functions only under controlled/managed conditions but for the alien infestations in the Murchison Bay in Lake Victoria and other freshwater lake systems in the region, it is possible that water hyacinth instead uncontrollably creates conditions for deterioration of water quality. Terefore, there is a need to manage the water hyacinth for water purifcation under controlled conditions and also convert it to other uses such as farm mulch and biogas generation as suggested in [54,56].

Conclusion
Te extent and distribution of the water hyacinth in the Murchison Bay vary over space and time but are more concentrated on the northern shores. Te results revealed shifts and diferences in area coverage of water hyacinth in the Bay over the four years (2016-2019). Water hyacinth signifcantly afects water quality, in some cases, outside the WHO maximum-minimum permissible limits. Results from this study indicate that parameters such as DO, turbidity, pH, BO, and total phosphates are not within the permissible range of the WHO 2020 guidelines. Te efect of sampling depth was only signifcant on pH, EC, BOD, and DO whilst the interactive efect of environment and depth was insignifcant for all water quality parameters. Te determining factors of water hyacinth extent and distribution pattern largely vary over space. Te water hyacinth determinants include strong winds, herbaria, fshing gear, construction activities at the shore, water-level changes, fsh hatcheries, ferry navigation, humidity, and biotic colonization.

Data Availability
Te data used to support the fndings of this study are available from the corresponding author upon request.

Conflicts of Interest
Te authors declare that there are no conficts of interest regarding the publication of this article.