The study area is located in the Rift Valley Lakes of Ethiopia and encompasses two lakes, namely Koka and Dembele, which are the largest freshwater lakes in the central part of the country (Figure 1). Lake Koka is located between 8°29’00” N latitude and 39°10’00” E longitude, at 1590 m.a.s.l. elevation. The estimated total area of the lake is 255km², with a length of 20 km and a width of 15 km. The lake has a shoreline length of 19.5–20.5 km and has a maximum and minimum depth of 14 m and 9 m, respectively. Mojo and Awash rivers are the two major rivers draining to the lake. One big town, Modjo, is located within the Koka Catchment. Though Koka Lake was originally developed for hydropower generation, it is currently used for fish production, irrigation, and flood control in downstream areas. The predominant land use/cover in Koka Catchments is agriculture (both irrigated and rain-fed) and rural settlements.

Lake Dembel lies between 8°00’00” N latitude and 38°50’00” E longitude, at 1636 m.a.s.l., and has an estimated surface area of 485 km², with a width of 20 km and a length of 31 km. The lake has a maximum depth of 8.95 m and a mean depth of 2.5 m and is considered the shallowest lake in the Rift Valley. The lake’s surface area fluctuates between 435 and 485 km² depending on the water inflow. The two major rivers that flow into Lake Dembel are the Ketar and Meki Rivers. Lake Dembel is a freshwater lake that supports different uses, including water supply, irrigation, fishing, and recreation. The lake is being degraded primarily because of various water and land-use activities in its catchment. There is ongoing agricultural transformation. Both small-scale farmers and large horticultural companies use huge amounts of pesticides and fertilizers that have led to nutrient loading, a conducive environment for the proliferation of macrophytes like water hyacinth, which is currently invading the lake at an alarming rate. The expansion of water hyacinth has been threatening the existence of the water body and surrounding iodiversity.

Water flows through the riverbanks, and runoff contributes to water hyacinth seed transport into the lakes (Churko et al., 2024). Similar studies indicated that the feet of animals and beaks of birds contribute to the spread of water hyacinths because they feed in sites of water hyacinth infestation and carry the seed over considerable distances by their bodies. Common aquatic macrophytes around and in Lake Dembel include Cyperus alopecuroides Rottb. (aquatic sedge), Phragmites karka (Retz.) Trin., and Paspalidium geminatum (Forssk.).

This study was implemented in three districts, Lume, Dugda, and Adami Tulu Jido Kombolcha, in the East Shewa Zones of the Oromia Regional State in Ethiopia. Zones and districts were purposively selected due to the threats of water hyacinths on Dembel and Koka lakes, which are found in these locations. The lakes are among the foremost water bodies, providing several socioeconomic benefits to the local people and the surrounding areas, including the capital city (Addis Ababa). Kebeles¹ were randomly selected based on their common boundaries with lakes and the level of water hyacinth invasion. In this case, a list of Kebeles with high water hyacinths (where the whole shoreline of the lakes is covered by the weed) and no water hyacinth infestation was identified through discussion with experts working on water hyacinth management in the Lume, Dugda, and Adami Tulu Jido Kombolcha districts. In addition, the sites were verified by field visits to the study area. Six Kebeles were selected from Lume, six from Dugda, and four from Adami Tulu Jido Kombolcha. Based on the proximity of individual households within the kebele, the respondents were grouped as adjacent (who conducted agricultural activities up to 300 meters from the lakes) and not adjacent to water hyacinth-invaded lakes (who conducted agricultural activities 300 meters away from the lakes). The total sample size of households was determined by Equation 1.

where = n₀ is the sample size, N is the total number of households, and e is the level of precision (in this case, 0.05).

Based on the data obtained from the districts’ Environmental Protection and Agricultural Development offices, the total number of households of selected kebeles was 8,654. Accordingly, the sample size is 382, considering the probability that some of the respondents might not be available for an interview or refuse to be interviewed. 5% of the sample (18) households will be added to make up the total sample of 400.

The proportionate sampling technique is employed to determine the sample size from each kebele. As shown in sample households adjacent and not adjacent to invaded lakes were systematically selected by interval and lottery method. After following the first sample drawn, every k_th element in the list was selected until the proportion number reached from all sample Kebele. In the Lume district, 102 samples were selected; 41 are adjacent, and 61 are non-adjacent to the lake. Similarly, from the total 119 samples selected from Dugda, 78 are adjacent, and 41 are not adjacent. Similarly, 125 samples selected from Adamitulu Jidu Kombolcha, 61 are adjacent, and 64 are not adjacent to the invaded lake.

The type of data required to address this research is a fully quantitative cross-section of data collected from the samples adjacent and not adjacent to lakes using a structured questionnaire. Primary quantitative data were collected from sample households using a structured questionnaire. The questionnaire is managed and collected using the Kobo Toolbox.² One supervisor and four data collectors were recruited, and training was given. The data is collected from sample respondents in a house-to-house, face-to-face interview for three weeks from May 12, 2024, to June 15, 2024. All data gathered through household interviews were supported by related annual and periodic reports and published articles to triangulate the information. In addition, key informant interviews were held at the federal, regional, and district levels by engaging experts and people with experience in water hyacinth management. Furthermore, the study conducted focus group discussions in which 6–8 focus-grouped discussants were engaged, considering educational background, experience in water hyacinth management, status in the community, and other community social values. Therefore, data for this research was obtained from households, focus group discussions (FGD), and key informant interviews (KII).

Descriptive and econometric models are employed to address the impact of water hyacinth. Descriptive analysis was used to compare demographic and socioeconomic variables for individuals grouped into adjacent and not adjacent to the nearby Koka and Dembel lakes. It is also compared based on before and after the expansion of the weed in two periods. The data are described by mean, frequency, and percent. Inferential statistics such as t-tests and chi-squares were applied to validate the result. The impact was assessed using propensity score matching and difference in difference between the two sample groups. These methods were preferred among others to see their difference in livelihood due to the water hyacinth invasion. The counterfactual is constructed by modeling the relationship between dependent and independent livelihood variables and grouping variables in a regression analysis.

The following Table 1 summarizes the group, dependent, and explanatory variables. Group variable identifies sample households based on their proximity to the Lakes. It is hypothesized to affect the outcome (dependent) variables. The dependent variable includes income from different sources and livestock production. Explanatory variables are also expected to influence the dependent variable in both groups of respondents.

Demographic and socioeconomic characteristics of sample households living in the study area are compared based on their proximity to water hyacinth-invaded lakes. From the total 400 samples, approximately 87% of the respondents were willing to answer questions. Table 2 shows that a greater number of the sample respondents’ farmland is adjacent to the lake (52%), male-dominated (78%), and married (81%). Similarly, a large number of the samples have their own land (96%) and are involved in on-farm activities (78%), and their livelihood depends on the lakes (84%). Nevertheless, except for household heads involved in off-farm activities, all other characteristics do not significantly vary based on whether their farmland is proximate to the lakes.

Of the total respondents (348 HH), only 11% (HH = 38) are involved in off-farm income generation activities. Approximately 41.3% of them (16 HH) were involved in salaried employment, carpentry, local alcohol preparation, trading manufactured goods, and remittance (Figure 7A). The remaining 58% (HH = 22) were engaged in other off-farm activities such as selling roasted fish, daily labor, and sand mining (Figure 7B).

Quantifying the impact of water hyacinth is critical in reducing the negative consequences on income and food security in developing countries. From the result in section 3, there is a significant difference in the households’ income adjacent and not adjacent to the lakes. The water hyacinth invasion declined the total and on-farm incomes of households adjacent to the lakes. Although water hyacinth infestation hampers the on-farm income by reducing crop yield (Honlah et al., 2019), the impact varies according to the intensity of the weed (Villamagna and Murphy, 2010).

Controlling weed competition for nutrients and light (Horvath et al., 2023) increases the yield of wheat (Minhas et al., 2023), maize (Ali et al., 2011), and soybeans, among others (Sepat et al., 2017), which is expected to increase the on-farm return. On the other hand, crop yield reduces due to failure to control water hyacinth (Damtie et al., 2022) and blockage of the irrigation canals (Churko et al., 2023a,b; Maulidyna et al., 2021). Similarly, land covered by the weed completely reduces farmers’ return (Tewabe et al., 2017), and even smallholders incur additional costs of production (Churko et al., 2023a,b; Getahun and Kefale, 2023). Invasive weeds are also a habitat for other invasive species, adversely affecting agricultural production and productivity (Pratt et al., 2017). In this regard, controlling the expansion of water hyacinths would be crucial for improving households’ livelihoods in the lakes.

In developing countries, smallholder farmers are engaged in off-farm activities to diversify their household income (Adeoye et al., 2019; Liu et al., 2019). Off-farm income, comparable to the on-farm, has contributed to food security and poverty reduction for smallholders (Babatunde and Qaim, 2010). Considering its role, households’ off-farm income in the study area is directly or indirectly associated with the Lakes. Some incomes linked to Lakes are selling fried fish, daily labor in vegetable production, and sand mining. Other incomes unrelated to the lakes are salary, trade, local beverage sales, remittance, and carpentry work. Nevertheless, the off-farm income declined for households adjacent to the invaded lakes compared to the other group of respondents.

Water hyacinth has its own advantage in generating income rather than solely concluding it as a dangerous invasive weed. The weed has an opportunity to enhance household income by using it as an input for making handicrafts (Harun et al., 2021), biogas (Wang, 2021), and compost among the benefits (Muhsin et al., 2023; Nega et al., 2022). Hence, measures to control water hyacinth should consider its diverse economic and environmental benefits.

Livestock plays a vital role in the livelihoods of smallholders, particularly in food security and poverty reduction. Similar to income from on-farm crop production, it is the main source of income for the majority of smallholders (Ashley et al., 2018), the mainstay of livelihood (Chaminuka et al., 2014), and contributes to the economy of a developing country (Herrero et al., 2013).

In relation to water hyacinth, this study did not find any significant impact on livestock production between invaded and non-invaded areas. There are some assumptions about the livelihood of the study area. First, the farming system: households in the area are more dependent on vegetable and crop production. Second, feed sources: if the livestock feed is other than grazing land, the weed may not have a significant impact. Finally, the size of land and herd: it may happen if households do not have enough grazing land and a significant number of livestock sizes. In sum, it is against some findings (Chapungu et al., 2018; Ogunlade, 2002). For instance, it was found that the weed influenced livestock production by invading the grassland. It impacted feed supply, and Damtie et al. (2022) showed its cause of livestock death.

Similar to livestock, fisheries also play a vital role in livelihood (Temesgen et al., 2019; Rommens et al., 2003) and food security (Funge-Smith and Bennett, 2019; Lauria et al., 2018). Our finding shows that water hyacinth has no significant impact on fish income in the invaded areas. The harvested fish species in the study area vary due to the presence of water hyacinths, which may or may not affect the total amount of fish caught (Segbefia et al., 2019). Some studies revealed that water hyacinth creates a conducive environment for certain fish species, such as killifish (Hill et al., 2021), which could explain why the weed’s effect on fish income was not significant in this context. In contrast, some reports show that water hyacinth has led to increased production costs and market supply shortages (Villamagna and Murphy, 2010), reduced fish harvests (Enyew et al., 2020; Tewabe et al., 2017), and disrupted fish business operations (Ezama, 2019; Ogunlade, 2002). These factors have contributed to declines in income in other regions, highlighting the potential variability in water hyacinth’s impact on fisheries.