Perceptions and knowledge of ecosystem services in urban river systems, Eastern Cape, South Africa

The ecosystem services concept has been studied in rural contexts but not as much in urban landscapes, particularly aquatic ecosystems in Africa. This means we have little knowledge of how urban populations perceive ecosystem services (ES) supplied by rivers in urban centres. To �ll this gap, this paper assesses local people’s perceptions and knowledge of ecosystem services supplied by rivers in urban centres using the Swartkops River in the Nelson Mandela Bay metropolitan area in the Eastern Cape as a case study. The survey results indicated variation in the extent people felt the river provided a variety of services. There was almost universal agreement that the river never provides freshwater which likely re�ects high levels of pollution in the river. An interesting aspect of the model results were differences in results related to the importance of conceptual and experiential knowledge of ES. For provisioning services, like recreation, experiential knowledge of the river was more important than conceptual knowledge, whereas for intangible ES such a regulating services, conceptual knowledge was important than experiential knowledge.


Introduction
It is projected that by 2050, two-thirds of the world's population will be living in cities (United Nations 2018).The phenomenon of rapid urban population increase is very prevalent in the developing world, especially in Africa.Since 1990, the number of African cities increased from 3,300 to 7,600, and urban centres have added 500 million people (OECD/UN ECA/AFDB (2022).In South Africa for instance, over 70% of the population will live in cities by 2030 (COGTA, 2016).These rates of change place anthropogenic stresses on natural ecosystems in and around cities. Multiple reasons have been offered to explain the rapid urbanisation in Africa, including people migrating from rural to urban centres in search of social services and economic opportunities as well as proximity to administrative centres that are often situated in cities (Njoh 2003;Güneralp et al. 2017;Heinrigs 2020).As a result of the growing urbanisation and urban population, natural systems such as rivers and wetlands within cities have become seriously degraded (Uluocha and Okeke 2004; Walsh et al. 2005;Collier et al. 2015;Xu et al. 2019).Recent studies on the so-called urban river syndrome have shown that the hydrology, ecology, biodiversity, water quality and geomorphology of urban rivers have been severely altered due to pollution, land cover change, poorly designed and over-burden sanitation services, run-off from roads and rail networks, industrial and domestic waste discharges (Capps et al. 2016;Jackson et al. 2016;Odume 2017).Systemic governance failure and urban poverty have also been identi ed as critical contributors to river ecological degradation in cities (Odume et al. 2022).As in many other parts of the world, the implications are that rivers in many cities in Africa have become seriously degraded, compromising their ecological function and the services they provide to society (Capps et al. 2016;Chen et al. 2022).In the Nelson Mandela Bay Metro (NMBM) in the Eastern Cape of South Africa for example, rivers and streams within its urban landscape appeared to be seriously impacted (Odume 2017).This then raises an important but less researched question on how urbanisation re-con gures relations between rivers and people, and how local urban populations perceive ecosystem services supplied by urban rivers in cities as well as the factors that may mediate such perceptions.
The ecosystem services concept has largely been studied in rural contexts but not much in urban landscapes, particularly aquatic ecosystems in Africa (Haase et al. 2012;Keeler et al. 2019).For this reason, the knowledge of how urban communities experience, perceive and value the services urban rivers provide remain sparse.It is thus important to assess local people's perceptions and knowledge of ecosystem services in cities for several reasons.First, such an assessment has the potential to surface the ecosystem services that are most important to people ( Given the criticality of understanding knowledge of local people and how they perceive ecosystem services, research efforts have recently begun to seriously focus on this subject.For example, Rodriguez et al. ( 2006) conducted a study on local people's knowledge of ecosystem services supplied by Opuntia spp. in Peru.The study found that people had more knowledge of, and assigned more value to, ecosystem services that have direct use value such as non-timber forest products, than they did to others.Lewan and Söderqvist (2002) conducted a study in Southern Sweden in which they investigated people's knowledge of ecosystem services supplied by a river system.Among their study samples were ordinary citizens, politicians, government o cials and researchers.The study found that many respondents, including researchers and technically trained people, show little understanding of the ecosystem services concept.In South Africa, Murata et al. (2019) also found that local people had knowledge of provisional ecosystem services such as drinking water and construction timber, but the same cannot be said of regulatory and supporting services.The results from these studies highlight the urgent need to engage with local communities to accelerate the uptake of the concept of ecosystem services in society as a whole as a way of surfacing and strengthening the people-nature relations.In this paper, we seek to distil people-river relations in the Nelson Mandela Bay Metro (NMBM) by assessing local people's perceptions and knowledge of ecosystem services supplied by rivers in urban centres.We also assess factors that may mediate local people perceptions of ecosystem services.The paper lls important gaps in the literature.First, it assesses perception of ecosystem services of river ecosystems in an urban environment.Prior review of ecosystem services has noted that there are few studies of aquatic ecosystems more generally (Hossu et al. 2019), particularly in an urban context (Haase et al. 2012;Keeler et al. 2019).Second, the study focus on South Africa implies contribution from Africa.Prior review of related studies has noted that most scholarship came primarily from the United States, China, and Europe (Cockerill 2016;Grizetti et al. 2016;Targetti et al. 2021;Zhang et al. 2021).This means we have little knowledge of how urban populations perceive ecosystem services supplied by rivers in urban centres within an African context.

Materials and Methods
The river is located in the Nelson Mandela Bay Metropolitan Municipality (NMBM) in the Eastern Cape Province of South Africa (Fig. 1).It forms two main tributaries, Elands and KwaZungu Rivers, and it discharges into the Indian Ocean in Algoa Bay near the city of Port Elizabeth (Odume 2017).The Swartkops estuary is considered ecologically important as it has the third largest inter-tidal salt marsh of South African estuaries and ranked 4th most important in the country in terms of biodiversity and conservation (Odume et  Many communities within the Swartkops River catchment are impoverished, and this impoverishment is spatially distributed across the catchment area.It has been reported that in the residential areas of Zwide to Uittenhage more than 60% of households depend on less than USD 300 per month as opposed to the city of Port Elizabeth area with higher household income (Nelson Mandela Bay Municipality 2022).This differential standard of living and income leaves a greater percentage of the Swartkops population depending on the ecosystem services provided by the river for their livelihood.For example, harvesting of traditional medicines, shing, bait collecting, angling and spiritual activities are among the bene ts derived from the river catchment.

Survey Design and Data Collection
Questionnaires were used to assess perceptions of Ecosystem Services (ESs) of the Swartkops River.Survey data were collected from residents who were 18 years and older living within the river catchment.A team of ten trained eld staff collected 181 surveys between May 2021 and June 2021.Ethical approval (Ethics Review No. 2019-0808-990) for the research project and data collection was obtained.The survey was conducted in English and orally translated into the local language isiXhosa, when necessary.
The survey contained a set of questions that asked about four different types of ecosystem services: provisioning (food, medicine, freshwater), regulating ( ood reduction, regulation of local climate, waste transportation and puri cation), cultural (recreation, spiritual worship, cleansing and puri cation, aesthetic value) and supporting services (maintenance of species variety).Respondents were asked to indicate if the river always, sometimes or never provides these services.The survey also contained questions to collect information about respondents' knowledge of ecosystem services (ES) and their length of residence near the river.Lastly, a group of questions collected information about respondents' demographic and socio-economic characteristics.

Methods
In addition to descriptive statistics and basic tabulations of the data, people's likelihood of indicating "always", "sometimes" or "never" for a particular ecosystem service was analyzed using multinomial logit models.These models are particularly useful for categorical data and in this study, provide an understanding of the likelihood that a respondent chooses a particular answer based on a set of characteristics.The general speci cation of the models estimated in this paper is as follows (Medwid and Mack, 2021): Pr (y = J) = where j = 1 corresponds to always and j = 2 corresponds to sometimes.The choice J corresponds to never and is the reference category in the analysis that follows.The variable z is a vector of respondent characteristics that serve as independent variables in the models.This variable where β 0 is the intercept for each model, β i (i = 1, 2,…,n) is a vector of slope coe cients for each of the independent variables in a vector X i the contents of which are speci ed below in Table 1.

Length of Residence
A categorical variable that takes on a value of 1 if the respondent has lived near the river less than 5 years, 2 if the respondent has lived near the river 5-10 years, 3 if the respondent has lived near the river for more than 10 years Models were estimated in Stata13.1 using the mlogit command and robust standard errors.The coe cients of the models are relative risk ratios which can be interpreted similar to an odds ratio.Relative risk ratios greater than one (1) indicate that a respondent is more likely to answer in a particular way, relative to a base category.For example, a relative risk ratio of 1.30 indicates a respondent is 30% more likely to respond in a particular way compared to respondents in the never category.A relative risk ratio less than one (1) indicates a respondent is less likely to answer in a particular way, relative to respondents in the base category of never.

Results
Table 2 below presents the pro le of the respondents in the survey dataset.From a demographic perspective, the majority of respondents are male.In terms of age, survey participants are fairly evenly split between the three age categories.The same cannot be said for socioeconomic characteristics, most respondents (65%) make between R2000 and R10 000 a month and 71% live in formal settlements with municipal services.Over 80% of respondents have lived near the river for over ve years.This variable is important because it is related to experiential knowledge of the river and its characteristics.Greater than 5 years 54% Aside from living near the river, another factor that may impact knowledge about the services provided by the river is an awareness of the concept of ecosystem services.Our sample was evenly split in terms of prior knowledge; 50% of residents had prior knowledge of ecosystem services and 50% did not (Fig. 2).
Figure 3 displays information about the extent respondents indicated the river provides particular types of ES.Respondents were less likely to indicate the river "always" provides ES.The regulation of local climate is an exception to this trend; 33% of respondents indicated that the river always provides this service.People were more likely to indicate that the river "sometimes" provides a variety of services including food (54.7%), medicine (48%), recreation (48%), water transportation and puri cation services (46%).People were also more likely to indicate the river never provides a particular ESs.A clear example of this trend is the provision of freshwater.Almost 80% of respondents indicated the river never provides this service.Respondents were also likely to say the river never provides water transportation and puri cation services (47%), as well as recreation (40%), and aesthetic value (pleasing to look at) (39%).In fact, respondents were largely split between the river sometimes providing particular services and never providing particular services.Three examples of this split opinion about services are the extent to which the river is pleasing to look at, provides spiritual services, and provides water transportation and puri cation.
Table 3 provides a summary of model results for our two main variables of interest, prior knowledge of ecosystem services and length of residence.Appendices A-H provide the full model results.Models were not estimated for freshwater provision because nearly all respondents indicated the river never provides this service.Thus, there was not enough variability in the data to estimate multinomial logit models.The purpose for presenting the results in this way is to test the extent that knowledge of ecosystem services is important to understanding how surveyed respondents viewed the range of ES offered by the river.Here, ecosystem services are an indicator of conceptual knowledge of ES while the length of residence variable is an indicator of experiential knowledge about river services.The results are organized by type of ecosystem service: provisioning, regulating, cultural, and supporting services.
A review of the chi-square test statistic in appendices A-D indicates that all of the models, save the one for food are statistically signi cant at the 1% level.The model for food services is statistically signi cant at the 5% level.The pseudo R-squared values range from a low of 0.1287 for food services to a high of 0.2594 for water transportation and puri cation services.This means approximately 13% of the variation in food services provision is explained by the model while about 26% of the variation in water transportation and puri cation services is explained by the model.
When both prior knowledge of ecosystem services and length of residence are included in the models, the chi-square statistics get larger and the statistical signi cance improves.All models, save the one for ood regulation are statistically signi cant at the 1% level.The model for ood regulation is signi cant at the 5% level.The pseudo R-squared values also increase and range in value from 0.181 for spiritual services to a high of 0.3003 for water transportation and puri cation services.
Aside from overall model explanatory power, there are also variations in the linkages between the main independent variables of interest and each ecosystem service.For the provision of food services, practical knowledge informed by people's length of residence near the river was more important than conceptual knowledge of ecosystem service (i.e.prior knowledge of the concept of ecosystem services).For surveyed respondents that indicated the river sometimes provides food services, prior knowledge of ES becomes insigni cant when length of residence is included in the model.Further, only residents that have lived near the river between ve and ten years were more likely to indicate that the river sometimes provides food services.
For the provision of medicine, experiential knowledge is also more important than conceptual knowledge of ESs.Residents who lived near the river more than ten years were less likely to indicate that the river always provides medicine.
For regulating services, conceptual knowledge of ES is relatively more important than experiential knowledge of ES.In the models for ood regulation and the regulation of local climate, prior knowledge of ES was statistically signi cant, even after including length of residence in the model.In the models for ood regulation, respondents with prior knowledge of ES were more likely to indicate the river always and sometimes provides ES.In the models for regulation of local climate, both prior knowledge of ES and length of residence were important in explaining the responses of people who responded that the river sometimes regulates the local climate.
The model results for supporting services are similar to those for regulating services; prior knowledge of ES is more important than experiential knowledge of the river.For both species variety and water transportation and puri cation services, people were more likely to respond that the river always and sometimes provides these services if they had prior knowledge of ES.
The results for cultural ES are more varied.For recreation services, length of residence rather than ES was more important to understanding the always responses.People living near the river for more than ve years were less likely to indicate that the river always provides recreation services.Conversely, prior knowledge of ES was more important to understanding the sometimes responses; people with prior knowledge of ES were more likely to indicate that the river sometimes provides recreational services.
For spiritual services, both conceptual and experiential knowledge of ES were important to explaining the always responses.Prior knowledge of ES made respondents more likely to indicate that the river always provides spiritual services.However, people living near the river for more than ten years were less likely to indicate that the river always provides spiritual services.Only length of residence was important in explaining the sometimes responses for spiritual services.People living near the river between ve and ten years were more likely to indicate that the river sometimes provides this service.
For aesthetic services, knowledge of ecosystem services was more important in explaining the always responses; those with prior knowledge of ES were more likely to say the river always provides this service.Length of residence however was more important to explaining the sometimes responses.People living near the river between ve and ten years were more likely to indicate the river sometimes provides this service.
While the overall model results discussed above indicate that all of the variables are useful in explaining the variation in survey responses, some variables are individually signi cant in explaining respondents' survey choices.Individual factors are important to assess because they may mediate local people's perceptions of ecosystem services.Table 4 provides a summary of results for the other control variables in the models that contain both prior knowledge of ecosystem services and length of residence.
Gender was particularly important for the food and aesthetics models; men were more likely than women to indicate that the river always provides this service.They were also more likely to indicate that the river sometimes provides spiritual services.
In terms of age, the results are mixed across age groups and different types of ecosystem services.For example, respondents between the ages of 35-44 were more likely to state that the river always and sometimes provides species variety.They were also more likely to state that the river sometimes provides recreational services.Older respondents however were less likely to indicate the river sometimes helps to regulate the local climate and provides spiritual services.They were more likely to indicate however that the river sometimes provides species variety.
Education was not individually signi cant in most of the models, save the model for water transportation and puri cation services.Here, residents with a high school degree were less likely to indicate the river always provides this service.Income was individually signi cant in many of the models where people indicated the river sometimes provides a particular service.In these instances, upper-middle income survey respondents with monthly incomes between R5,000-R10,000 were more likely to say that the river sometimes provides recreational services and species variety.The highest income survey respondents with monthly income greater than R10,000 were more likely to say that the river sometimes regulates local climate and sometimes provides species variety.They were less likely to indicate that the river provides aesthetic services however.
Lastly, type of housing was individually signi cant in several of the models.Residents in formal housing with municipal services were less likely to indicate the river always provides species variety.Residents living in both informal and formal housing with municipal services were also less likely to indicate the river always provides water transportation and puri cation services.The same was true for climate regulation and species variety; residents in housing with municipal services were less likely to say the river sometimes regulates the local climate and sometimes provides species variety.Residents of informal housing with municipal services were less likely to say the river provides aesthetic services.

Discussion
The goal of this paper was to assess local people's perceptions and knowledge of ecosystem services supplied by rivers in urban centres.We also assessed factors that may mediate local people's perceptions of ecosystem services.In this paper, we assessed how education, length of residence, housing, income, age, and prior knowledge of ecosystem services concept may mediate people's perception of the ecosystem services supplied by rivers in urban centres.
Survey results indicated variation in the extent people felt the river provided a variety of services.For many services, people were more likely to indicate the river sometimes provided a service instead of always.One service however for which there was almost universal agreement was the provision of freshwater; most respondents indicated the river never provided this service.This is likely because of high levels of pollution of the river.The Swartkops River is regarded as one of the most polluted river systems in the Eastern Cape of South Africa (Tshithukhe et al. 2021).High levels of E. coli, pharmaceutical compounds, metals and other chemical pollutants have previously been reported (Odume et al. 2012; Farounbi and Ngqwala 2020; Vumazonke et al. 2020;Ohoro et al. 2021).It is possible that residents' familiarity with, and experiential knowledge of the river water quality may have shaped the perception that the river never provided freshwater.Another possibility would be the level of water service delivery within the catchment.Over 100% of households had access to piped municipal water, minimising their reliance on the river for drinking water (Nelson Mandela Bay Metro 2022), which is a contrast in rural-based studies that have indicated that people often perceived rivers as providing freshwater (Smith and Moore 2011).Our results highlight the importance of knowledge of the local environment in shaping the perception of people regarding ecosystem services (Moutouama et al. 2019) as well as socio-economic status and access to municipal services.
An interesting aspect of model results were difference in results related to the importance of conceptual and experiential knowledge of ES.To capture these types of knowledge, prior knowledge of ES was included as a measure of conceptual knowledge and length of residence was included as a measure of experiential knowledge.For provisioning services such as food and cultural service like recreation, experiential knowledge of the river was more important than conceptual knowledge, whereas for intangible services such a regulating, conceptual knowledge was more important than experiential knowledge.Therefore, seems that experiential knowledge plays a key mediating role for ecosystem services that have direct use value such as provisioning of food and cultural services such as recreation.Similar observations have been made by Rodriquez et al., (2006), Lewan and Söderqvist (2002) and Murata et al. (2019) who reported that people tend to have more knowledge of and assigned value to ecosystem services that have direct use value.
Rivers provide important regulatory and maintaining services such as ood control, local climate regulation, erosion prevention and maintaining population and habitats (Kaiser et al. 2020).For the most part, these regulatory services do not have direct use value.The model results indicate that conceptual knowledge of ES were comparatively more important than experiential knowledge for the regulatory and maintaining services.People who had prior knowledge of the concept of ecosystem services tend to indicate that the river sometimes or always provide these regulatory services.What this implies is that awareness raising and educating local people about the concept of ecosystem services can contribute to a better appreciation of the diversity of ecosystem services urban rivers provide, beyond those with direct use values Zedda (2023).It needs to be noted that level of formal education does not necessarily translate to being aware of or having knowledge of the concept of ecosystem services, as education does not seem to play any signi cant role in our model.For example, in a study that involves diverse participants including politicians, government o cials and researchers, Lewan and Söderqvist (2002) found that the participants showed little understanding of ecosystem services concept.This implies that awareness raising and education should be designed speci cally towards messaging the concept of ecosystem services among local people.
Model results for cultural services suggested mixed importance of both conceptual and practical knowledge of ES.For the most part, cultural services such as recreation can be perceived as having direct use value, hence the importance of experiential knowledge, but services such as sense of identity may not, hence the relevance of conceptual knowledge.In the present study, there was a split in responses between sometimes and never for cultural services such as recreation and aesthetic.The research participants were of the view that the river does not always provide these services.As already indicated, the Swartkops River is seriously polluted although the level of pollution is seasonally mediated (Farounbi and Ngqwala 2020; Vumazonke et al. 2020).The awareness of the river pollution among the resident may have contributed to the responses, as residents may be aware of the risk posed in using polluted water for recreational purposes.
Although our study shed interesting light on river-people relationality in an urban context, it is important to note some limitations of the present study which is based on a case study of one urban river at one point in time.Study results obtained in the future may differ from the ones presented in the present study.A different sample may also have different views about the river.These limitations suggest some avenues for future research.One avenue is the extension of the current study framework to other urban rivers.A second extension is to examine the linkages between urban river pollution and the perception of ecosystem services.
An understanding of the ecosystem services may impact how people interact with urban rivers, but little is known about perceptions of urban river services on the African continent, and in South Africa in particular.To ll this knowledge gap, this paper analysed residents' perceptions of urban river services in the Swartkops River in the Nelson Bay Municipality in the Eastern Cape.The results of this examination suggest a linkage between river pollution and people's perceptions of a variety of ecosystem services.This overarching result suggests that one step in solving the problem of urban river degradation is nding innovative ways to strengthen people-river relations so that they value river systems more highly and take proactive measures to protect these important ecosystems.Prior knowledge of ecosystem services Figure 3 Perceptions of respondent regarding whether the river always, sometimes or never provides a particular ecosystem service

Declarations
Martín-López et al. 2012; Hartel et al. 2014; Hossu et al. 2019).Second, it can reveal the ecosystem services that local people are most willing to conserve because of their perceived importance and value (Murata et al. 2019).Third, it can reveal important synergies and trade-offs that can occur when certain ecosystem services are consumed or conserved (Martin-Lopez et al. 2012; Hartel et al. 2014; Hossu et al. 2019).Fourth, the perception and knowledge of ecosystem services by urban communities can form the basis for mobilising individual, collective and relational agencies for positive actions given the urgent need to reverse the trajectory of ecological degradation of urban river systems (Everard and Moggridge 2012; Chen 2017; Odume et al. 2022) Fifth, such an assessment can inform the design and implementation of ecological interventions that are in line with local values (Harrison et al. 2010).

Figure 1 Location
Figure 1

Figure 2
Figure 2 (Odume 2017;Adams et al. 2019)is threatened by several sources of water pollution including agricultural runoff, the discharge of raw and poorly treated wastewater e uent sewage from informal settlements and municipal treatment works respectively, removal of riverbed materials and alien invasive plants(Odume 2017;Adams et al. 2019).Population growth and informal settlements within the catchment were implicated as important contributors to observed ecological degradation in the catchment(Odume et al. 2023).

Table 1
A binary variable that takes on a value of 1 if the respondent does have prior knowledge of ES and 0 if no