Land use/land cover dynamics, trade-offs and implications on tropical inland shallow lakes’ ecosystems’ management: Case of Lake Malombe, Malawi

ABSTRACT Lake Malombe supports various ecosystem services (ESs). However, it is increasingly experiencing human-induced pressure. This study used geospatial, household surveys, focus group discussion, key informant interviews, consultative meetings, and field observation on assessing the Land use/Land cover dynamics (LULCD), its trade-offs, and implications on ESs. The findings demonstrated a decrease in forest land (52,932 ha to 78,983 ha) at the expense of cultivated (52,932 ha to 78,983 ha) and settlements (7054 ha to 17,595 ha). Changes in ecological indicators such as fishery, river flow, soil erosion, turbidity, biodiversity, invasion of alien species, scenic beauty, extinction of some species, frequent flooding, cultural value, and carbon sequestration were significantly (p-<0.05) linked to some LULCD classes. The study findings are significant to policymakers, ecosystem managers, the local population, and various stakeholders in understanding conflicting interests and policy priorities to balance the lake ecological restoration and human welfare.


Introduction
Tropical inland shallow lakes' catchments support diverse ESs such as regulatory, purification, supporting, provisioning, culture, and aesthetic services (Guo et al., 2019;Aneseyee et al., 2020;Makwinja et al., 2021a). Regulatory services are crucial in maintaining a world where people can live and control the adverse effects of floods, disasters, and diseases (Mengist et al., 2020). Terrestrial, marine, and freshwater ecosystems sink anthropogenic carbon emissions, with gross sequestration of 5.6 gigatons of carbon per year (IPBES, 2018). An estimated 4 billion people rely primarily on ecosystem provisioning services (EPSs) as a source of natural medicines (Brockerhoff et al., 2013). More than 2 billion people rely on wood fuel to meet their primary energy needs (Van Der Kroon, et al., 2013). Rural and urban populations in the least developed countries depend on EPSs for their sustenance (Plisnier et al., 2018). In Africa, meeting local food production demands is highly linked to the vast expansion of agricultural lands at the expense of natural forests and grasslands (Ngaruiya, et al., 2017;Roser & Ritchie, 2019;Shehab et al., 2021;Watson et al., 2019). With the rapid population explosion and the ongoing socio-economic activities, landscape alterations in Africa have been expected. Catchment degradation due to the vast expansion of settlements, and agricultural production instigated by high rate (2.3%) of human population growth, progressively ground cover removal, socio-economic development, and climate change has been linked to ESs erosion, particularly in tropical inland shallow lakes (Aneseyee et al., 2020;Davivongs et al., 2012;Ewunetu et al., 2021;Gondwe et al., 2019;Nkwanda et al., 2021;NSO, 2018;Seutloali & Beckedahl, 2015). Many researchers such as Lambin and Geist (2006), Sharma et al. (2011), and Makwinja et al. (2014) have also acknowledged that LULCDs affect ES functioning and human livelihoods through a reduction in water supply, reservoir storage capacity, agricultural productivity, and world-ecology. Vallet et al. (2018) also documented that a high rate of land conversion to satisfy the demand for food production has increased pressure on ESs. About 55% of tropical forest in Africa has been lost to agriculture, and this trend is typified in the SADC region's inland shallow lakes'characterized by rapid population growth (Kehoe et al., 2017).
In Malawi, landscape transformation has become the emergent trend, particularly in inland tropical shallow lakes driven by rapid population growth, climate change, and environmental degradation Makwinja et al., 2019;Njaya, 2009). Malawi is ranked as one of the countries in the SADC region experiencing forest loss, with estimates show that about 30,000-40,000 hectares of land are lost annually due to increased agricultural activities and excessive wood and charcoal biomass consumption (Ngwira & Watanabe, 2019;Nkwanda et al., 2021). Over 90% of the population predominately depends on farming for their livelihood (Chingala et al., 2017;Kaland-Joshua et al., 2011). The shallow lake catchments are continuously invaded and transformed into agricultural land as the farming activities intensify, even the areas designated as buffer zones with ecological consequences such as habitat alteration, increased landscape degradation, severe soil erosion, heavy loss of ecological functions, and drastic decrease in the lake ecosystem productivity (Jamu et al., 2003;Kafumbata et al., 2014;Makwinja et al., 2021a;Ratner et al., 2012). Climate-related disasters such as prolonged drought and seasonal floods further stress the lake ecosystem, triggering landscape degradation (Likoya, 2019;Makwinja & M'balaka, 2017;Robledo et al., 2012). With a least Human Development Index of 0.4726, Malawi is experiencing unprecedented freshwater ecosystem loss (Pullanikkatil et al., 2018), and future projection indicates the worst (Cacho et al., 2020;GoM, 2016;Lorena, 2018). Currently, acute food shortages have become chronic among the rural population, which is worrisome as far as the local population's sustenance is concerned (Ellis et al., 2003;FISH, 2015;Froese et al., 2016;MacPherson et al., 2012;Makwinja et al., 2021c;Mueller & Geist, 2016;Njaya, 2007). Lake Malombe provides the best example, with many studies pointing towards increased local population exposure to vulnerability and catastrophic decline in the lake ecosystem productivity (Dulanya et al., 2014;FISH, 2015;Jamu et al., 2011;Kapute, 2018;Makwinja et al., 2021a). The concept of landscape management, associated trade-offs, and implications under the changing catchment, mainly in the inland shallow lakes such as Lake Chilwa, Lake Malombe, Lake Chiuta, has not been adequately discussed (Mendham et al., 2012). Much of the studies in these lakes have focused on the fishery, ecology, rural livelihoods, stock assessment, co-management, and governance (Dulanya et al., 2014;Hara, 2006;Hara & Njaya, 2016;Makwinja et al., 2021a;Tweddle et al., 2015).
There is generally insufficient data regarding the extent to which the landscape of these lakes has been transformed, their associated trade-offs, and their ESs implications. Therefore, the overall objective of this study is to assess the intensification of landscape transformation in the tropical freshwater shallow lakes' catchment, the potential trade-offs, and the implications citing Lake Malombe in Southern Malawi as a case study representing scenarios in other inland shallow lakes in Malawi and Africa. Specifically, the study was conducted to answer the following research questions: (i) To what extent has the Lake Malombe landscape been transformed over the past years? (i) What are the potential trade-offs and opportunity costs associated with such landscape transformation? (iii), What are the potential lake ESs implications linked to such transformation?
It is connected to the Lake Malawi South East Arm through the 19 km stretch of the Shire River basin-an outlet of Lake Malawi (FISH, 2015). Lake Malombe lies entirely within the Great African Rift Valley complex and is characterized by a series of major and minor faults. It has a high productive catchment dominated by calcimorphic soils, which occur along the rift valley floor. Typical calcimorphic soils include mopanosols, dark grey, sandy-clay soils with low permeability, and alluvial soils, grey to brown (Government of Malawi, 2014). The eastern part of the catchment has a low risk of soil erosion as it is partially protected by the forest reserve and Liwonde National Park. The western part is deforested and is prone to soil erosion. Lake Malombe is fed by water from Lake Malawi via a stretch of the Upper Shire River and shares the same aquatic ecology with Lake Malawi, including a high level of fish biodiversity, genetic plasticity, and endemism (FISH, 2015). Limnologically, Lake Malombe is a shallow, turbid, and nutrient-rich lake with shelving vegetated shores compared to Lake Malawi (Makwinja et al., 2021d). The lake is highly productive because of the water column, which mixes freely, recycling the bottom nutrients. Although the current trend indicates otherwise, the lake once produced 17% (15,00 tons) of Malawi's total fish biomass landings in the 1980s (Njaya, 2007).

Data collection process
Primary data were collected using a combination of participatory and formal procedures. The structured household questionnaire was used to capture various information relevant to the research objective. The household survey questionnaire consisted of three sections: the first section captured direct and indirect ESs. The second section captured the general information about access and use of lake ESs, and the third section asked for detailed information about each ESs' changes, impact, vulnerability, and coping strategies. The household survey adopted the cluster sampling technique. The three traditional authorities (Chowe, Mponda, Chimwala) were delineated into clusters (the group village headmen). The 30 × 12 clusters model was used as a guiding principle (Makwinja et al., 2021c). A total of twelve clusters was chosen based on their proximity to the lake. In addition, six university graduates coordinated face-to-face interviews at the respondents' houses or the fish landing sites. The required sample size was calculated using a simple population proportion formula to assume that the study area's population had unknown proportions and was assumed to be heterogeneous (i.e. 50/50 split).
Where n r = sample size and z = value from the standard normal distribution reflecting the confidence level (z = 1.96 for 95% level of confidence) of unknown population proportion (p). The p = 0.5 value was used (assuming maximum heterogeneity), Ɛ is a margin of error. Office, District Fisheries Office, and the Liwonde National Park Office. Following the discussion from the diverse stakeholders, important lake ESs were ranked based on (i) ability to meet the local population's basic needs and (iii) the likelihood of obtaining enough quality data on the ESs for computation. The Focus group discussion (FGDs) explored a greater depth of ESs dynamics using a community as a unit of analysis. When conducting FGDs, questions were presented as guidelines for discussion using four participatory research tools: resource mapping, institutional analysis, cause-effect analysis, and well-being analysis. The FGDs were conducted along with household surveys and indepth key informant interviews. To facilitate openness, women and men formed separate groups. Each group had an average of 10 people. The FGDs involved fishers, crew members, farmers, fish processors, natural resource governance leaders, traders, and traditional leaders in all villages. All groups consisted of residents and migrants of all gender groups, including women, men, and youths.
The FGD was complemented with direct field observation. The rationale behind direct field observation was observing livelihood strategies, socioeconomic activities, governance issues, culture, and customs. Direct field observation provided rich data set through an interpretive approach to people's socio-economic reality and subjective meanings by eliciting and observing what ES is essential to them. The researcher gathered the evidence during the field observation while interpreting the local communities' interpretations of the ESs. Direct observation in the Lake Malombe catchment was done from September to November 2019. Lake Malombe West is the central hub of fishing activities with a high level of migration and mobility. The upper catchment about 19 km away from Lake Malombe is Mangochi Town. The local communities first misinterpreted the researcher as a trader, and many households came to present their various products derived from the lake ecosystem, such as fish, woven craft, fruits, and birds. After some weeks, the researcher could interact with the communities, and the field observation began. The researcher inquired about the diverse ESs derived from the lake, trade-offs, social networks, governance issues, culture, and customs. This was achieved by walking every morning and evening to various fish landing sites, markets, upland areas, and the lake catchment to observe various economic activities in the study area. The researcher also engaged 30 key informants residing in the lake catchment. The in-depth key informant interviews were conducted after seeking consent from the respondents. The key informants were selected based on their experience of ecosystem dynamics and the duration of residing in the catchment. Preferably those who stayed for more than 50 years were selected. A snowball sampling technique was used in choosing informants, where each informant was used to identify one or two other possible informants through networks. The number of key informants increased after adding each informant until the sample size grew with each subsequent interview and eventually became saturated where no significant new information was gathered. Interviews were chosen for their relevance to the conceptual questions rather than their representativeness.

Satellite images pre-processing and land use classification
The Landsat TM images (covering Landsat 5, 7, and 8) of LULC of 1989LULC of , 1999LULC of , 2009, and 2019 were produced from spectral Landsat imagery with a spatial resolution of 30 m retrieved from the United States Geological Survey Website (USGS, http://glovis.usgs.gov/) using Worldwide Reference System Path 29 and Row 32. Seasonal errors were reduced by selecting the images with similar calendar dates. The four bands (band 2, 3, 4, and 5) combinations were considered for image classification. A minimum of 10-30 band cases per class was used to train and validate pixels from TM imagery. Approximately 200 representative training pixels for all classes were selected based on stratified random sampling for 1989, 1999, 2009, and 2019 images. A separate dataset of approximately 180 points and a ground truth classified image were collected to assess generated prediction maps.

Landsat image interpretation
The atmospheric correction was not required for image classification of the same calendar date because it is the same as subtracting a constant from all pixels in a spectral band. The Landsat TM imagery was obtained at Level 1 T and was already geometrically corrected and orthorectified. The Landsat TM imagery was imported into the ENVil 5.4 software. The image georeferencing accuracy was checked with a reference map of the study area accessed from the Malawi Department of Survey. All the input images had the same map and projection data with the same number of layers. After layer stacking, sub-setting was performed to get a portion of a large image file into one smaller file to reduce the image file size, focus the region of interest, and speed up the processing time. The exact number of bands (Band 2, 3,4, and 5) were used for classification in all cases to minimize the biases caused by different band combinations. The training and validation data during optimization were generated using polygon vectors designed for each land-cover type. Supervised training-based region of interest mapping was used for selecting land cover types, and five classes (Table 1) were determined from 1989,1999,2009, and 2019 Landsat imagery.
Arc GIS software version 10.7.1 was used to compute LULC change-traditional matrix using overly procedure to quantify the area converted from a particular LULC class to another LULC category during the study period. The annual rate of change was determined as shown in the equation below Where r is the annual rate of change for each class, S 1 and S 2 are areas of each LULC class at a time t 1 andt 2 respectively.

Normalized difference vegetation index
The Normalized Difference Vegetation Index (NDVI) was used to assess the presence of green vegetation and was computed as follows: NDVI values ranged from −1 to 1 means the higher the NDVI value, the higher the fraction of green vegetation present in the area. Landsat band 4 (0.8-9.94 mm) measured the reflectance in the NIR region, and Band 3 (0,63-0.69 mm) measured the reflectance in the Red region. However, for Landsat 8, the NIR and Red regions had different wavelength ranges. Therefore, they were computed using bands 4 and 5 for Red and NIR, respectively, which resulted in values ranging from 0-200 and fit within an 8-bit structure. The green color showed the presence of vegetation, and other colors show the absence of green vegetation. These attributes helped classify the images.

Maximum likelihood (ML) classification
This study used the maximum likelihood classification algorithm for image classification because it considers the class signature's variances and covariances when assigning each cell to one of the classes represented in the signature file. The algorithm used by the ML tool is based on Baye's theorem of decision making, where cells in each class sample in the multidimensional space are normally distributed. In this study, a class was characterized by mean vector and covariance matrix. The statistical probability was computed for each class to determine the membership of the cells to the class. Each cell was classified to the class to which it has the highest probability of being a member. The algorithm for computing the likelihood D of unknown measurement vector X belongs to one of the known classes, was based on the following Bayesian equation: Where D = weighted distance (likelihood), c =a particular class, X =the vector measurement of the candidate pixel, M c =the mean vector of the sample of class c, a =percent probability that candidate pixel is a member of class c (defaults to 1.0 or is entered from prior knowledge), cov c = the covariance matrix of the pixels in the sample of classc, cov c j j= determinant of cov c (matrix algebra), ln =natural logarithm function,T=transposition function (matrix algebra)

Accuracy assessment
In order to evaluate the performance of the classifiers, the accuracy assessment was carried out using a validation dataset, assuring distribution in a rational pattern so that a specific number of observations were assigned to each category on the classified image. The Kappa accuracy was computed as given below: This is an area dominated by herbaceous rather than woody plant species. The marshes are typically found at the edges of the lakes and streams, forming a transition between aquatic and terrestrial ecosystems. Grasses, reeds, and rushes mostly dominate them.

Water bodies
Water bodies include permanent lakes, rivers, streams, seasonal or permanent wetland, intermittent pools, perennial marshy and human-made dams Where r is the number of rows in the matrix, X ii is the number of observations in row i and column i (the diagonal elements),x xþ i and x xþ are marginal totals of row r and column i respectively, and N is the number of observations. The overall accuracy measured the proportion of the assessed area classified correctly. The user's accuracy measured the proportion of pixels classified as belonging to a class that genuinely classified as belonging to that class. The user's and producer's accuracy measurements are related to commission and omission errors.

Statistical analysis
Qualitative data were decoded, translated into English, and analyzed using content analysis for related themes. The analysis involved coding in generating initial themes among the codes and reviewing and naming the themes. The identification of related themes was based on the historical pattern within the data. On the other hand, Geospatial LULC Landsat satellite data were analyzed using GIS and remote sensing approaches. ArcGIS version 10.7.1 and ENVil 5.4 software were used. The mixed-effects linear regression model was used to explain the extent to which LULC classes influence ESs in the study area. Quantitative data were analyzed using descriptive and inferential statistics. STATA version 15 was used in the data analysis.

The accuracy assessment and LULC classes
The overall producer's accuracy assessment for 2019 was 89.36 with 85.01 for the forest, 89.93 for marshes, 90.05 for cultivated land, 93.93 for shrubs/grass/bare land, and 82.15 for waterbody (Table 2).  (Table 3). Like other lake ecosystems in Malawi, Africa, and the globe, most of the local population living in the catchment depends on farming activities (Schuyt, 2005). Figure 2a and b show that about 75.5% of the respondents own a farm for agricultural activities, and cereal crops such as maize are ranked the highest.      Figure 3 show that the land-use class for cultivation has increased significantly. The cultivation includes upland areas, flood plain, and wetlands, predominately cultivated for rainfed and irrigated agriculture ( The study further shows that land under cultivation increased from 52,932 ha to 78,983 ha from 1989 to 2019. In Ethiopia, Hailu et al. (2020) also reported a tremendous increase in cultivated land from 1973 to 2019.
The increase in cultivated land was done at the expense of forest land, which can be attributed to the spike of small-scale farming as the lake fishery and waterbody declining (Figure 4a and b) (Hailu et al., 2020;Powlson et al., 2011;Yaro, 2013). Evidence in Figure 4a shows that the total annual fish catch slightly fluctuated positively from 1976 to 1984 and registered a peak within 1988 to 1990s and then declined sharply from 1990 to 2000 with the lowest recorded from 2000 to 2004 and later slightly fluctuated positively from 2005 to 2016. Figure 4a further shows that fishing effort declined as the fisherfolks temporarily or permanently abandoned fishing during low catches and increased pressure on the landscape due to increased demand for cultivated land at the expense of forest. Many studies in Malawi and Sub-Saharan Africa also reported increased cultivated land at the expense of forest land, biodiversity, fishery, and waterbody Wasige et al., 2013;du Toit et al., 2018;Munthali et al., 2020). In Malawi, Palamuleni et al. (2011) reported an 18% increase in agricultural land in the Upper Shire river catchment from 1989 to 2002. Jamu et al. (2003), on the other hand, reported increasing deforestation in the Likangala River of Lake Chilwa catchment, Southern Malawi. In Zimbabwe, particularly the Buzi subcatchment, Chemura et al. (2020) also reported a similar trend. In the Ndembera watershed, Tanzania, Hyandye et al. (2018) reported increasing agricultural land and evergreen forests by nearly 10% and 7%, respectively, from 2013 to 2020. Bare land/shrubs, grassland has also been transformed into other land-use types. Figure 3 shows that the lake catchment and lowland areas are characterized by dense cultivation, densely populated, and treeless dominated by shrub and grassland. The area under study was found that the bare/shrub/grassland increased from 1989 to 2019, which aligns with the findings of Daniel (2008), Tolessa et al. (2017), and Siraj et al. (2018). The increase in bare/shrub/grassland is attributed to the high rate of forest reserve conversion into bare land (plate 1a) as the demand for charcoal production and agricultural activities increases. Plate 1b shows that wetland is converted into a cultivated while forest into bareland /shrubs/ grassland.
Charcoal production is one of the income generation activities for the Lake Malombe local population. This study found that in 2019, the hilly part of Lake Malombe catchment remained bare, with increased gullies developing each rainy season as trees are continuously harvested. As a result, the waterbody size decreased from 33,300 ha in 1989 to 30,483 in 2019 (Table 3). Arnhold et al. (2014) explained that mountainous farmland cultivation at the expense of forest land triggers severe soil erosion. The rate of soil loss in Lake Malombe catchment though not yet estimated, could be more than 29 t/ ha/year estimated at the national level (Mzuza et al., 2019). One of the most outstanding phenomena of this present findings is that human activities are strongly linked to the current status of the lake ES dynamics which agree to the recent work by Dulanya et al. (2014), who evidenced that Lake Malombe has been experiencing accelerated eutrophication in the recent past 100 years. Magadza (2010) also had similar findings in Lake Kariba and Zambezi River Valley. The main reason for decreased waterbody in Figure 4b is the prolonged drought, siltation, and rainfall decline as temperatures in the catchment increase (Likoya, 2019;Ngongondo et al., 2020;Power, 2010).

Trade-offs between ecosystem services
Lake Malombe provides multiple ecosystem services to both local and global communities. These diverse ESs generates potential trade-offs (Farley, 2012;De Groot et al., 2010;Kremen, 2005;Langner et al., 2017;Martinez-Harms et al., 2015). Figure 3 shows that the local population is continuously modifying the Lake Malombe landscape for unsustainable settlements. Such enormous socio-economic development has put more pressure on the lake landscape, making it more vulnerable to extreme climate events and natural hazards such as floods, drought, and long-term effects of ES degradation. An attempt was made to bring some significant ESs provided by Lake Malombe to the local and global communities' attention to raise awareness. This was achieved by introducing a set of ecological indicators that do not directly measure ESs but demonstrate the lake's ecological status in response to landscape dynamics. Biodiversity loss was selected as an indicator because it is associated with greater resource use efficiency within the lake ecosystem and forms a significant component of its socio-ecological system (Geist, 2011). Loss of biodiversity can compromise the livelihood of the local population, future conditions of the lake, species composition, and food-web diversity (Schallenberg et al., 2013). Loss of biodiversity is further linked to cultural value and carbon sequestration changes, and these too were selected as indicators. Lake fishery was selected as a significant ESs indicator because it represents harvestable food and is a significant resource for the Lake Malombe population. For example, Lake Malombe is best known for the mass depletion of fish biomass in Africa . This mass depletion exposed the local population to risks and vulnerability, forcing them to devise unsustainable strategies to deal with the fisheries resource scarcity (Makwinja et al., 2021c), negatively impacting the lake ecosystem functions (Makwinja et al., 2021e). The loss of macrophytes from Lake Malombe 1960s caused a sudden regime shift from clear-water to turbid states (Dulanya et al., 2014). Turbidity is linked to human pressure due to increased demand for cultivated land and settlements as the fishery collapses-hence in this study, it was selected as an indicator. Invasion of alien species is a critical anthropogenic activity affecting the ESs, with freshwater considered ecosystems the most impacted by species invasions. In this study, the invasion of alien species was selected because it is linked to other forms of environmental degradation instigated by human activities. Extinction of some species is linked to change in habitats and over-exploitation instigated by increased human population (Pelletier & Coltman, 2018) and hence was selected as an ecological indicator. The indicators mentioned above were introduced into the mixed-effect regression model. Table 4 shows that land degradation in the Lake Malombe catchment is highly linked to socioeconomic development. The model demonstrates a significant (p < 0.05) trade-off between the built-up areas and fishery, turbidity, biodiversity loss, and extinction of some species. Increased built-up land use areas in the lake catchment are done at the expense of biodiversity, some EPS such as fishery, supporting and regulatory services such as water purification and carbon sequestration). Aquatic plants such as Pennisetum purpureum schumach, Phragmites mauritianu, Typha latifolia, and Cyperus papyrus play a significant role in climate regulation through carbon sequestration (Kayranli et al., 2010;Were et al., 2021), water quality purification through the removal of heavy metals (Bernardini et al., 2016;Ceschin et al., 2021;Parzych et al., 2016), fecal pathogens removal (Kipasika et al., 2016), nutrients regulation, habitant provision (Bornette & Puijalon, 2011) and flood regulation (Rooney et al., 2013;Scheffer et al., 2003;Wang et al., 2019). Aquatic plants such as Typha domingensis, Terminalia sericea, Azadirachta indica found in

02, Prob > Chi squared =0.000, ns indicates not significant while ** and * indicate significance at 0.01 and 0.05 probability level of Confidence
They culturally provide services to local communities and cure diseases such as bilharzia, pneumonia, diarrhea, antiseptic wounds, and Malaria (Pullanikkatil et al., 2018). However, these dominant aquatic plants are increasingly exploited and used for constructing temporary shelters, local tea rooms, and woven curios. The findings agree with Zohary and Ostrovsky (2011), who noted that the weakening of keystone species, loss of biodiversity, and increased internal nutrient loading are linked to increased socio-economic activities such as land use transformation for settlements. Hou et al. (2020) noted that major socio-economic events such as the vast expansion of built-up areas strongly impacted flood risk mitigation capacity and water quality in Yangtze Plain freshwater lakes. Similar trade-offs are reported by Asadolahi et al. (2018) in Iran, Lake Chilwa basin, in Malawi (Mvula & Haller, 2009;Pullanikkatil et al., 2013), and Lake Chad (Zieba et al., 2017). Lake Malombe fishery, for a very long period, supported the livelihoods of thousands of local populations from different parts of the catchment (Pinnegar et al., 2016). However, the fishery has faced a series of combined threats, including over-exploitation, pollution, and invasive species (Walker, 2012). Jul-Larsen et al. (2003) reported a severe decline of Copadichromis spp (Viginalis kajose) from 937 metric tons in 1991 to 412 tons/year in 2001. Dulanya et al. (2014) also noted that Lake Malombe had experienced a catastrophic decline in fish stocks. Researchers such as Makwinja et al. (2014) and Xu et al. (2020) pointed out that the decline in fish production is highly linked to the fishery's overdependence. Similar findings have been reported in inland lakes and wetlands in Malawi and tropical regions (Kafumbata et al., 2014;Kosamu et al., 2017;Njiru et al., 2010). The mixed-effect regression results in Table 5 demonstrated a significant (p <0.05) negative relationship between the fishery and cultivated land-class, suggesting that the collapse of the fishery in Figure 4a opened new alternative livelihood opportunities beyond fishing (FISH, 2015). Farming in the Lake Malombe flood plain is one of the most prominent activities that has witnessed a substantial expansion over the past years ( Figure 3). As the water body area shrinks (Figure 3), the flood plain areas have become attractive for farming throughout the year (Aneseyee et al., 2020;Vehrs & Heller, 2017). The agricultural yield derived from cultivating these emerging landscapes is high, and this has been a pull factor for immigration into the catchment leading to rapid population explosion and decreased landholding capacity. However, EPSs such as cropping are done at the expense of cultural and aesthetic values, biodiversity, some valuable endemic animal and plant species, the capacity of the lake catchment to regulate flood, river flow, erosion, and climate. Several researchers, such as Keesstra et al. (2016), have evidenced these trade-offs, who concluded that the rate of species extinction, turbidity, soil erosion, and flooding increased with increasing cultivated land. Kay et al. (2009), Kanyika-Mbewe et al. (2020, and Nkwanda et al. (2021) also reported similar trade-offs. Table 3 shows that the increase in agricultural activities in the Lake Malombe catchment is done at the expense of forest land and natural wetland vegetation. A similar observation has been made by several authors  such as Allan, 2004), Schuyt (2005), Palamuleni et al. (2011), andKafumbata et al. (2014) in Malawi, Dibaba et al. (2020) in Ethiopia, Uwimana et al. (2018) in Southern Rwanda, Sibanda and Ahmed (2021) in Zimbabwe, and Thonfeld et al. (2020) in Tanzania. Figure 3 shows that forest land has been cleared out from 1989 to 2019, leading to loss of terrestrial and aquatic biodiversity, diminishing supporting, regulatory, cultural, and aesthetic services. Table 6 shows that forest land-use class had a significant (p < 0.05) positive regression coefficient with river flow, biodiversity, scenic beauty, and carbon sequestration. Conversely, soil erosion, turbidity, extinction of some species, and frequent flooding had a significant (p < 0.05) negative regression coefficient.

Log likelihood =−65.278, prob >chi square =0.000, Wald Chi square =416.06, ns indicates not significant while ** and * indicate significance at 0.01 and 0.05 probability level of Confidence
During FGDs, it was pointed out that the rate at which terrestrial and aquatic fauna are displaced has increased as forest land decreases (Government of Malawi, 2014). Díaz et al. (2009) also emphasized the link between climate regulation through carbon sequestration and the stability or increase in vegetation over a long time. The increased farming is done at the expense of forest land, shrubs, mangroves, marshes, and vegetation. Loss of forest land has a severe negative ESs impact. The findings agree with the 2018 IPBES report, which indicates that agricultural expansion alongside rapid population growth and increased food consumption has come at the expense of forests, wetlands, and grasslands, compromise climate regulation, supporting, cultural and aesthetic ecosystem service values. In Malawi, Njaya et al. (2011), Nagoli and also observed in the Lake Chilwa Basin.
On the other hand, Jogo and Hassan (2010) suggested that diversifying livelihoods from agriculture while improving economic well-being can enhance the conservation of the tropical freshwater shallow lakes ESs. Climate regulation is essential for both global and local communities. However, it is not a key priority among the local population and cannot be traded off with EPSs such as cropping, fuelwood, and charcoal production. The integrity of the forest ecosystem in Lake Malombe catchment can be achieved through the introduction of several schemes, including compensation of local forest managers for ESs provided to a broader community-an initiative which is done in Costa Rica also known as payment ecosystem services (Chadzon, 2008;Morse et al., 2009). The benefits of this initiative will also boost other ESs such as water regulation, pollination, habitat provisioning, and biodiversity conservation. Turpie et al. (2008) also highlighted that concentration effort should be on ESs such as water supply, carbon sequestration as umbrella services to achieve a range of conservation goals. Sachedina and Nelson (2012) also supported this approach and acknowledged that it could solve global environmental challenges.
The increase cultivated at the expense of forest coupled with climate change impact is done at the expense of water bodies. Figure 3 shows that the water bodies in the catchment have been shrinking over the years due to severe land degradation. During the key informant interviews, the following anecdote was expressed. "Some important rivers such as Mikongo, Lusalumwe, Namingundi, Nangapoche, Msuka, Lutende, Unga, Lugola, Litisa, Lilole, Liwesa, Nanyumbu, Msinje, Mpilipili, Luchimwa, Lipinda, Mchoklola, Namasawi, Luwelele, Ngapani, Namangandwe, Masongola, Mandimba, Nyenyezi, Lingamasa, Masanje, Sangadzi, Nansenga, Mpale, Plate 2: Msinja river dried up (a), a sand bar extending to the shoreline as Lake Malombe shrinks (b). Note: Rodgers Makwinja captured photos (a and b) during the field survey conducted in October 2019 in Lake Malombe Mtamankhokwe, Namingundi, Koche, Nakundu, Thema and Kabudira drain into Lake Malombe, Lake Malawi and Shire river have recently dried up". Said the BVC chairperson at Khande Beach, October 2019. Plate 2a and b confirmed that the rivers have dried out, and lake volume has considerably declined.
The findings agree with Hecky et al. (2003), Kosamu et al. (2012), , Pullanikkatil et al. (2018), and Ngongondo et al. (2020) in Lake Malawi, Elephant Marsh wetland, and Lake Chilwa. Furthermore, Table 7 shows that the regression coefficient for the fishery, turbidity, frequent flooding, and biodiversity displayed a positive regression coefficient and was significant at the 0.05 level of confidence, suggesting that a decrease in water bodies could result in a decrease in fishery and biodiversity (Gownaris et al., 2016;Ng'onga et al., 2019;Wang et al., 2019).
However, turbidity is also reduced and increases with increased water bodies due to erosion (Wantzen & Mol, 2013;Xu et al., 2020). Frequent flooding is also reduced due to prolonged drought in the study area. However, under extreme rainfall, the intensity of flooding events increases (Government of Malawi, 2014). The findings align with Bond et al. (2008), who acknowledged the loss of habitats due to the shrinking of water bodies under the severe drought. Nkwanda et al. (2021) also noted the strong relationship between water quality degradation and aerial coverage of the water bodies in the Lilongwe River catchment. The scenario depicted in Lake Malombe also applies to other African tropical inland freshwater shallow lakes (Elka & Laekemariam, 2020). For example, in Malawi,  reported that Lake Chilwa in Southern Malawi experienced severe water recession due to prolonged dry spells and frequent drought leading to the disruption of ESs. In Ethiopia, Lake Hiromasa disappeared from 1989 to 2005, resulting in the loss of aquatic biodiversity (Yilma, 2010). In Kenya, Lake Naivasha, an official 'Ramsar Site'-30,000 ha, turned into a shallow mud pool during the 2009 drought resulting in a decline in the aquatic ecosystem (Ogola et al., 2012). Once the most significant freshwater Lake, Lake Chad shrunk dramatically in the last 40 years (Pham-Duc et al., 2020).

Conclusion
Lake Malombe ESs support both local and global communities. Changes in the landscape have benefited a few individuals and cause chronic constraints to the majority sharing the common pool resources. This study demonstrates how the landscape dynamics changes instigated by human and climatic-induced factors shape Lake Malombe socioecological system. The study further demonstrates that the local population will continue to depend on EPSs for their sustenance, and increasing demand will shrink other essential ESs making the local population more vulnerable and trapped in the vicious cycle of poverty. The study findings are of practical importance to policymakers, the local population, and various stakeholders in understanding competing interests and policy priorities to balance ecosystem management and human welfare. Engaging diverse stakeholders is required to reduce the risks of competing interests and sectoral policies conflict. The discussion in this study further highlights future research gaps. There is a need to understand how the local population embraces an effective governance system to achieve sustainable integrated freshwater ecosystem management.

PUBLIC INTEREST STATEMENT
The urgent need to conserve various inland tropical freshwater shallow lakes is one of the global challenges ecologists and policymakers face. However, a management policy can cause unprecedented consequences if done without ample evidence. Therefore, an informed decision is necessary to reduce unexpected trade-offs between human demand and the need to safeguard multiple freshwater ecosystem services. The Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES) initiated the need to link scientific evidence and management policies. Against this background, the researcher tried to assess Lake Malombe LULCDs, trade-offs, and potential implications on the ESs to provide appropriate policy recommendations in formulating and implementing Freshwater Ecosystem Management approach to fisheries in Malawian and global freshwater lakes. Prof Emmauel Kaunda is the Vice-Chancellor of Lilongwe University of Agriculture and Natural Resources, and Director of African Center of Execellence in Aquaculture and Fisheries (AquaFish). He received his PhD from Rhodes University where he got the esteemed "Best PhD. Seminar Award". In 2013, he received 3rd Prize of the IMPRESSA by RUFORUM as the best scientist in Africa. He is a co-founder of African Fisheries Experts Network and also a coordinator of the African Union Aquaculture Working Group and Coordinator of the NEPAD Fish Node of the Southern Africa Network on Biosciences (SANBio).
Dr Tena Alamirew is academic researcher with 25 years experience-lecturing numerous courses and supervising undergraduate and graduate students. He has been involved in the design and effective implementation of several national and international collaborative research programs that support graduate students. He demonstrated his leadership attributes while serving as the academic research Vice of Haramaya University. He is a founding direcror of Ethiopian Institute of Water Resources and serve as the Deputy Director of Water and Land Resource Center and heads the research division.

Declaring conflict of interest
The authors of this paper declare that there is no conflict of interest.

Source funds
The African Centre of Excellence for Water Management (ACEWM) provided financial support for this study under the World Bank's African Centres of Excellence (ACE II) Project, Grant number GSR/9316/11.

Data availability
All data generated during this study are available upon request from the corresponding author.