Developing Macroinvertebrate Biotic Indices in Nigerian Urban-Agricultural River Catchments: Is the Continuous Scoring System More Effective than Discrete Scoring System?

Abstract

The evaluation of the ecological consequences of anthropogenic stressors is a critical challenge in the management of the environment. Multimetric indices (MMIs) are one of the biomonitoring tools that have been widely explored to assess the ecological health of riverine systems globally, as MMIs have proven to be extremely effective, owing to their ability to incorporate data and information from both structural and functional assemblages of organisms and the entire ecosystem. Currently, there are very few MMIs developed in Nigeria to assess the ecological health of riverine systems, and none of the MMIs was developed for river stations draining urban and agricultural catchments. In order to close this gap, we developed and validated a macroinvertebrate-based MMI for assessing the ecological health of river systems in the Niger Delta area of Nigeria draining urban and agriculture catchments. Furthermore, we also compared the effectiveness of both continuous scoring and discrete systems for the development of MMI. Physico-chemical variables and macroinvertebrates were collected from 17 well-marked out stations that spread throughout 11 different river systems. The stations were classified into three categories based on the degree of impact: least-impacted stations (LIS), moderately impacted stations (MIS), and heavily impacted stations (HIS). Sixty-seven (67) candidate macroinvertebrate metrics were potentially tested, and only five metrics were deemed significant and ultimately retained for integration into the final Niger Delta urban–agriculture MMI. The following five metrics were chosen to remain in use for the MMI development: Chironomidae/Diptera abundance, %Odonata, Margalef index, Oligochaete richness and logarithmic-transformed relative abundance of sprawler. Notable performance rates of 83.3% for the least-impacted stations and 75% for the moderately impacted stations were found during the index’s validation using a different dataset. However, for the stations that were most affected (i.e., the HIS), a 22.2% performance rate was noted. The Niger Delta urban–agriculture MMI was adjudged to be suitable as a biomonitoring tool for riverine systems subjected to similar combined stressors of urban and agricultural pollution.

1. Introduction

Human activities have greatly affected rivers globally, leading to disruptions in the movement of energy and matter, posing a threat to the survival of various organisms [1,2,3,4,5,6,7,8]. Multiple studies have indicated that human activities at the catchment level can be seen in the physical disruption of local habitats of river stations, leading to water pollution that negatively impacts aquatic communities [9,10,11]. Due to their inherent functions, freshwater ecosystems not only rank among the most exploited systems globally but also directly bear the impacts of human activities within their catchments [12]. Among the world’s freshwater ecosystems, Afrotropical freshwater ecosystems are among the most threatened ecosystems globally despite their unique biodiversity and high endemism [13,14]. Rural-to-urban migration across countries within sub-Saharan Africa is leading to a rapid increase in urban growth, resulting in intensified human activities in the catchment areas of forested riverine systems [1,15,16,17]. Concerted efforts have been made to use biological indicators to better understand and evaluate the health of freshwater ecosystems. Macrophytes, algae, protozoa, macroinvertebrates, and fish are examples of biological indicators that have been explored [18]. This all-encompassing approach of using biological indicators together with physico-chemical variables enables a thorough assessment of the ecological status, providing a holistic perspective on the dynamics and overall health of freshwater ecosystems.

In Nigeria, basic assessments of stream biota are sometimes used to supplement physico-chemical analysis for the management and monitoring of water quality [19]. Even while a physico-chemical analysis is widely used, it has limits, especially when it comes to providing detailed information about the state of degradation of water bodies. To track actual changes in freshwater environmental conditions, physico-chemical and biological parameters have been explored as early-warning indicators [20]. Because of their responsiveness to both man-made and natural pressures, biological indicators, in particular, provide valuable insights. There are numerous biomonitoring techniques and instruments available for tracking how pollution affects riverine ecosystems. These strategies include multivariate approaches, biotic indices, functional feeding groups, and multimetric indices [17,21]. Multimetric indices (MMIs), in particular, have proven to be extremely effective, owing to their ability to incorporate data and information from both structural and functional assemblages of organisms and the entire ecosystem in which they inhabit [21]. The MMIs have been developed and applied to a wide range of biological components, such as phytoplankton [22], macroinvertebrates [17,23,24,25], and fish [26]. There have been a few prior studies in the Afrotropical region, such as those conducted in South Africa [27,28], East Africa [29,30], and West Africa [17,24,31]. While these studies set the pace for the development of MMIs in a continent like Africa that has been understudied, the emphasis on macroinvertebrate-based multimetric indices at the regional level is distinctive. The incorporation of regional factors ensures a nuanced understanding of the unique characteristics and challenges within Nigeria, contributing valuable insights to the broader context of biomonitoring using MMIs in sub-Saharan Africa.

Multimetric indices (MMIs) are a class of biotic indices that include measures of abundance, composition, richness, and variety, in addition to taxonomic, trait, and functional metrics. The resilience and effectiveness of this all-encompassing method have been observed to outperform that of conventional biomonitoring indicators [32,33]. In developing MMIs, two scoring methods have been widely explored, namely the continuous and discrete scoring system [17,34]. The continuous scoring system involves the scoring of metrics using numerical values (e.g., 0–1, 0–10, etc.), while the discrete scoring system involves the use of a range of numerical values that have been predetermined (e.g., 1,2,3,4,5, etc.; [17,34]). The statistical behaviour of continuous scores is less subjective as there are allowances for fractional values (e.g., 0.5, 1.5, etc.), unlike the discrete scoring system in which arbitrary ranges of scores are used with no allowance for fractional values. Further, for a continuous scoring system, some metric scores may increase with the level of disturbance in particular stations, and in such cases, the metric scores are rescaled by subtracting the said scores from a potential maximum score [35]. This is not applicable to the discrete scoring system, as there is no allowance for rescaling scores of metrics should there be variation in their projected level of disturbance in stations before the scoring proper [35]. Finally, due to the standardized nature of the continuous scoring system, biological condition classes (e.g., poor, fair, and good) are easily interpretable by river managers, unlike the discrete scoring system that may need an expert to interpret the biological condition classes if there is any variation in the level of disturbances from the initial projected disturbances [36]. Several organisms have been used in the past to develop MMIs such as birds, fish, macroinvertebrates, plankton, and macrophytes employing either continuous or discrete scoring systems [8,17].

Of the varieties of organisms employed to generate MMIs, macroinvertebrates are particularly noteworthy due to their global focus. Their importance as primary consumers in the food chain and web of aquatic ecosystems, as well as their sampling simplicity and ubiquity, are the reasons for this attention [31,37,38,39]. In addition, aquatic ecosystems depend broadly on macroinvertebrates, which include a diversity of aquatic insects, mollusks, and crustaceans. As secondary producers, they have a significant impact on the flow of energy up and down the food chain, which affects the general well-being and dynamics of aquatic habitats [31]. Aquatic macroinvertebrates such as snails, crustaceans, and insect larvae respond differentially to a variety of environmental conditions. They are sessile and sometimes not very mobile, and they live close to the bottom sediments and the water column, making them vulnerable to various stressors disrupting their natural habitat, such as agriculture, urbanization, and deforestation [40]. Further, the outright removal of forests for agricultural and urban development purposes within the catchments of the freshwater ecosystem, most especially in the Nigerian’s Niger Delta region, calls for serious attention.

The relevance of the Niger Delta region for global biodiversity conservation efforts has been highlighted by its identification on a global scale as a hotspot for biodiversity [41]. Before now, the bulk of riverine systems in Nigeria’s Niger Delta pass through wooded catchments that are home to mangrove swamps, with white and red mangroves predominating in particular [24,42,43,44,45]. That has changed now due to urban and agricultural activities around some of the catchments of the rivers within the region. To this end, it is pertinent to assess the current health states of riverine systems in the region to ascertain the level of degradation the systems have undergone. The majority of the studies conducted to date on evaluating the health of rivers and streams in this region, despite its ecological significance, have concentrated on the composition, diversity, and abundance of the organisms [19,42,43]. Furthermore, the application of multimetric indices is consistent with the current understanding of ecological health as a complex construct influenced by a range of interrelated factors. This is in contrast to the single biotic index methodology [21,44], which only considers data obtained from individual organisms. Hence, in this study, we developed a macroinvertebrate-based MMI for selected riverine ecosystems draining both urban and agricultural catchments in the Niger Delta. We also explored the effectiveness of using both continuous and discrete scoring systems in the development of MMI. Finally, we validated the developed MMI with separate datasets to test the applicability of the developed MMI.

In recent times, there have been serious debates on the kind of scoring system used in awarding ecological classes (e.g., fair, poor, moderate, and good) for MMI development. For instance, Ruaro et al., [34] frowned at the use of a discrete scoring system, stressing that the discrete scoring system does not take into consideration the continuation of scores awarded to metrics during MMI development in a bid to award the appropriate ecological classes to metrics incorporated into MMIs. Also, Edegbene [46] stressed the preference of continuous scoring over the discrete scoring systems in a study where MMI was developed for selected forested riverine ecosystems in Nigeria; hence, our exploration of the use of both continuous and discrete scoring systems to ascertain the one that is more effective for the development of MMIs. This novel contribution to the science of applied aquatic ecology is crucial due to the limited research on the development of biomonitoring tools using biotic indices in Africa, most especially in Nigeria where the current study was conducted. The study significantly enhances the scarce literature on the development of macroinvertebrate-based MMIs for monitoring riverine systems in Nigeria.

2. Materials and Methods

2.1. Study Area

We conducted this study in 11 basins that drain urban–agricultural catchments in Edo and Delta States within the Niger Delta region of Nigeria. The study area is characterized by wet and dry seasons, particular of a tropical climate. The dry season spans between October and March, and wet season spans between April and September [43]. The mean yearly temperature of the study area is 28 °C, with 2000–3500 mm average yearly rainfall and 85% yearly humidity [47].

2.2. Station Selection/Sampling Design

We explored 17 stations distributed in 11 river basins: Edor (2 stations), Eriora (1), Ethiope (2), Anwai (1), Obosh (2), Ogba (2), Ossiomo (2), Orogodo (1), Owan (1), Umu (1), and Umaluku (2). We selected the sampling stations randomly within third- and sixth-order rivers (Figure 1). Further, we ensured selected stations were markedly different in terms of anthropogenic stressors (agricultural activities and urbanisation). The GPS coordinates of the 17 sampled stations are presented in

Table 1. GPS coordinates of the sampled rivers stations in the Niger Delta region of Nigeria.

Rivers	Station Codes	Latitude	Longitude
Owan	R1	7.10205452755	5.84794547245
Ossiomo	R2	6.32945277800	5.76458055600
Ossiomo	R3	6.33000000000	5.72100000000
Ogba	R4	6.30872876199	5.58750000000
Ogba	R5	6.29583333333	5.57916666667
Eriora	R6	5.50200000000	6.18500000000
Orogodo	R7	6.27054166700	6.19773333300
Obosh	R8	6.22200000000	6.62000000000
Obosh	R9	6.21300000000	6.62900000000
Anwai	R10	6.24246388900	6.70266388900
Edor	R11	5.61900000000	6.14400000000
Edor	R12	5.53500000000	6.06000000000
Umaluku	R13	5.51200000000	5.99600000000
Umaluku	R14	5.50400000000	5.97100000000
Umu	R15	5.99300000000	6.30700000000
Ethiope	R16	5.85318333300	6.14970277800
Ethiope	R17	5.80300000000	6.09400000000

. We sampled macroinvertebrates for a period of five years (2008–2012) for two seasons (wet and dry) in a bid to put in perspective seasonal context to develop a standardized multimetric index (MMI). The wet-season samplings were conducted between April and September, while the dry-season samplings were conducted between October and March in every sampling circle. Please note, sampling was not done all year round per every sampling circle.

2.3. Physico-Chemical Variables Sampling

Surface-water physico-chemical samples were collected and analysed for each study station on every sampling expedition. We analyzed nine physico-chemical variables in the current study, including water temperature, water depth, stream flow (flow velocity), electrical conductivity (EC), pH, dissolved oxygen (DO), 5-day biochemical oxygen demand (BOD₅), and nutrients (nitrate and phosphate). Details on both the in situ readings and laboratory analyses of all physico-chemical variables are contained in our previous publication [47].

2.4. Macroinvertebrates Sampling

Sampling of macroinvertebrate was conducted following SASS5 protocol (South Africa Scoring System Version 5 protocol) standard procedure [48]. A kick net of dimension 30 × 30 cm² and 500 µm were used in collecting macroinvertebrates from stone, silt, mud, sand, and vegetation biotopes on every sampling expedition per sampling station. Four-minute sampling per biotope per station was conducted on every sampling expedition. Collected macroinvertebrates were placed in sample bottles containing 70% ethanol before transporting them to the laboratory for further analysis. Identification of macroinvertebrates was done using taxonomic guides by the following authors [49,50,51]. Macroinvertebrates were identified at family level due to scarcity of local identification guides in Nigeria.

2.5. Station Classification/Categorisation Using Physico-Chemical Variables

We classified the 17 samples into three potential impact categories including least-impacted stations (LIS), moderately impacted stations (MIS), and heavily impacted stations (HIS). Principal Component Analysis (PCA) was used to correlate physico-chemical variables with the 17 sample stations in the present study area. Of the 17 stations categorised in the study, two stations were LIS, which correspond to 11.76% of the entire sampled stations, seven (MIS) of which correspond to 41.17% of the entire sampled stations, and eight (HIS) of which correspond to 47.06% of the entire sampled stations (See Table 2 of Edegbene et al. [47].) Details on the station classification/categorisation and the corresponding analyses can be found in Edegbene et al. [47]. The three impact categories were used for further analysis all through.

2.6. Macroinvertebrate-Based Metrics

At first, we considered 67 candidate metrics for the MMI development, and they were selected mainly from earlier studies on the responses of macroinvertebrates to anthropogenic stressors in the tropics ([3,17,24,52] Appendix A 

Table A1. Selected candidate metrics used for the development of macroinvertebrate-based MMI in the urban–agricultural polluted stations in the Niger Delta region of Nigeria. Metric measures: A = Abundance, B = Composition, C = Richness, D = Diversity, E = Traits, Note: X = shows metric that failed in a particular test; √ = shows metric that was retained for integration into the final MMI.

Metrics	Definition of Metrics	Metric Measures	Metric Selection Tests				Retained Metrics for Final MMI
			Sensitivity	Seasonality	Redundancy	Signal/Noise
EPT Abun	Absolute abundance of Ephemeroptera Plecoptera and Trichoptera	A	X
Eph Abun	Absolute abundance of Ephemeroptera	A	X
Tri Abun	Absolute abundance of Trichoptera	A	X
ETOC Abun	Absolute abundance of Ephemeroptera Trichoptera Odonata and Coleoptera	A	X
Chi Abun	Absolute abundance of Chironomidae	A			X
Chi+Oli Abun	Absolute abundance of Chironomidae+Oligochaeta	A		X
Oli Abun	Absolute abundance of Oligochaeta	A		X
Dip Abun	Absolute abundance of Diptera	A	X
Mol+Dip Abun	Absolute abundance of Mollusca+Diptera	A	X
Dec Abun	Absolute abundance of Decapoda	A	X
Mol Abun	Absolute abundance of Mollusca	A	X
Mol+Dec Abun	Absolute abundance of Mollusca+Decapoda	A	X
Col Abun	Absolute abundance of Coleoptera	A	X
Odo Abun	Absolute abundance of Odonata	A	X
Hem Abun	Absolute abundance of Hemiptera	A	X
Col+Hem Abun	Absolute abundance of Coleoptera+Hemiptera	A	X
EPT/Chi Abun	Absolute abundance of Ephemeroptera Plecoptera and Trichoptera/Chironomidae	A	X
ETOC/Chi Abun	Absolute abundance of Ephemeroptera Trichoptera Odonata and Coleoptera/Chironomidae	A	X
ETOC/Dip Abun	Absolute abundance of Ephemeroptera Trichoptera Odonata and Coleoptera/Diptera	A	X
Chi/Dip Abun	Absolute abundance of Chironomidae/Diptera absolute	A					√
%EPT	Ephemeroptera, Plecoptera and Trichoptera relative abundance	B	X
%Eph	Ephemeroptera relative abundance	B	X
%ETOC	Ephemeroptera, Trichoptera, Odonata and Coleoptera relative abundance	B	X
%Tri	Trichoptera relative abundance	B	X
%Chi	Chironomidae relative abundance	B		X	X
%Chi+Oli	Chironomidae+Oligochaeta relative abundance	B		X
%Oli	Oligochaeta relative abundance	B		X
%Dip	Diptera relative abundance	B	X
%Dec	Decapoda relative abundance	B	X
%Mol	Mollusca relative abundance	B	X
%Mol+Dec	Mollusca+Decapoda relative abundance	B	X
%Odo	Odonata relative abundance	B					√
%Hem	Hemiptera relative abundance	B	X
%Col	Coleoptera relative abundance	B	X
%Col+Hem	Coleoptera+Hemiptera relative abundance	B	X
%Mol+Dip	Mollusca+Diptera relative abundance	B	X
EPT Rich	Ephemeroptera, Plecoptera and Trichoptera richness	C	X
Eph Rich	Ephemeroptera richness	C	X
Tri Rich	Trichoptera richness	C	X
Dip Rich	Diptera richness	C				X
ETOC Rich	Ephemeroptera, Trichoptera, Odonata and Coleoptera richness	C	X
Chi Rich	Chironomidae richness	C			X
Chi+Oli Rich	Chironomidae+Oligochaeta richness	C			X
Mol Rich	Mollusca richness	C	X
Col+Hem Rich	Coleoptera+Hemiptera richness	C	X
Col Rich	Coleoptera richness	C	X
Hem Rich	Hemiptera richness	C	X
Odo Rich	Odonata richness	C	X
Oli Rich	Oligochaeta richness	C					√
Dec Rich	Decapoda richness	C	X
Eve Ind	Evenness index	D			X
Sha Div	Shannon-Weiner index diversity	D		X
Mar Ind	Margalef index	D					√
Sim Div	Simpson diversity	D				X
Log Aer	Logarithm of relative abundance of aerial: spiracle	E	X
Log SoE	Logarithm of relative abundance of soft and exposed	E	X
Log CaT	Logarithm of relative abundance of cased/tubed	E	X
Log Opa	Logarithm of relative abundance of preference for opaque water	E	X
Log FrL	Logarithm of relative abundance of free-living	E	X
Log Spr	Logarithm of relative abundance of sprawler	E					√
Log Swi	Logarithm of relative abundance of swimmer	E	X
Log DeF	Logarithm of relative abundance of detritus (FPOM)	E	X
Log DeC	Logarithm of relative abundance of detritus (CPOM)	E	X
Log VeS	Logarithm of relative abundance of very small (<5mm)	E	X
Log Sma	Logarithm of relative abundance of small, >5–10mm	E	X
Log Lav	Logarithm of relative abundance of larva	E	X
Log Shr	Logarithm of relative abundance of shredder	E		X

). The selected metrics were in the structural and functional assemblage measures of macroinvertebrates, including composition, abundance, richness, diversity, and trait measures. Details on the definition and calculation of metric measures are contained in our previous publications [17,24].

2.7. Selection of Metrics and Index Development

A five-step method was used for the selection of metrics as derived from the following author methods [24,53,54,55,56]. Series of tests were used in screening the 67-candidate metrics selected for the current study. The tests are as follows: (a) sensitivity, (b) seasonality, (c) redundancy, (d) signal-/noise-repeatability, (e) metric scoring/integration, and (f) metric validation. The MMI was developed using macroinvertebrate datasets collected for a period of 3 years between 2008 and 2010, while MMI applicability and effectiveness were assessed using macroinvertebrate datasets collected for a period of 2 years between 2011 and 2012.

2.7.1. Sensitivity

Sensitivity of metrics was tested by comparing the performance of the said metric in the LIS, MIS, and HIS [24] by visualising the metric discrimination potentials on box plots. Two criteria were considered in this case. Firstly, a non-overlap of LIS metric interquartile ranges (IQRs) with that of MIS and HIS was considered sensitive [24]. Secondly, in the event that there was overlap of the metric IQRs but the median of the metric is without the IQRs, we consider the metric to be sensitive [24]. We further confirmed the sensitivity of each metric by performing a non-parametric Mann–Whitney (U) test to ascertain their significance level. The U test was considered appropriate for the test for sensitivity confirmation, as the normality test conducted on the selected candidate metrics using Kolmogorov–Smirnov test returned the datasets as not normally distributed. Based on the U test, metrics with p-values greater than 0.05 were eliminated [57].

2.7.2. Seasonality

Seasonality test was conducted by visualising on box plots the discrimination of each metric between dry and wet seasons [58]. The confirmation of metric seasonal stability was confirmed with Kruskal–Wallis test, and metric with a p-value that is less than 0.05 was adjudged as not seasonally stable [24]. Hence, they were eliminated for further analysis. To avoid confusion on the effects of agricultural and urban pollution on the seasonal stability of selected metrics, only candidate metrics on the LIS samples were used for the test for seasonality [3].

2.7.3. Redundancy

Metric redundancy was evaluated using Spearman’s correlation coefficient (r) analysis. Redundancy of metric was defined as metric that have r value that is greater than or equal to 0.78 (r ≥ 0.78). Redundant metrics were removed from further analysis, except they are ecologically significant [55].

2.7.4. Signal/Noise (Repeatability)

Signal/noise (repeatability) test was performed by the calculation of the variance of metrics in all the samples from all stations for the signal, while for noise, the variance of the samples from the LIS was calculated. Metrics with S/N value of ≥2 indicated high variability across the different stations. Hence, it does not distinguish well among stations, and we retained such metrics. On the other hand, metrics that were very noisy with S/N value of <2 were rejected [55,58].

2.7.5. Metric Scoring/Integration: Continuous Scoring System versus Discrete Scoring System

For continuous scoring system, metric scoring was done by integrating metrics with different range of values into the final MMI. To achieve this, we used a score range of 0–1 to standardise each metric by using the 5th and 95th percentiles of the values of LIS [46,56]. A score of 0 or 1 for each metric was awarded following two-step procedures. For metrics that negatively responded predictably to increasing urban–agricultural pollution, we used 95th percentile of the LIS as the value of the scoring ceiling and 5th percentile of the HIS as the value of the scoring floor. We standardised each metric as follows: metric value = (metric value of LIS minus scoring floor)/(scoring ceiling minus scoring floor). For metrics that positively responded predictably to increasing urban–agricultural pollution, we used 5th percentile of the LIS as the value of the scoring ceiling and 95th percentile of the HIS as the value of the scoring floor. Metrics were standardised as follows: metric value = (scoring floor minus metric value of HIS)/(scoring floor minus scoring ceiling). The final multimetric index was arrived at by adding each metric score. Overall, three ecological classes were arrived at, namely: good, moderate, and bad water quality.

We standardized metrics integrated into MMI for the discrete scoring system using maximum value, upper quartile (75%), mid-quartile (50%), lower quartile (25%) and minimum value the metrics [58,59]. We assigned a score of 1 to metrics that were predicted to decrease with increasing urban-agricultural pollution, if the value is lower than the minimum value of LIS, a score of 3 was awarded to a metric, if the value is between the minimum value and less than 25% of the LIS, while score of 5 was awarded to a metric if the value of LIS is greater than or equal to lower quartile (25%). Conversely, metrics that were predicted to increase with increasing urban–agricultural pollution were assigned a score of 1, if the value is above the maximum value of the LIS. A score of 3 was assigned, if the value is above 75% or below the upper quartile (75%) of the LIS. A score of 5 was assigned, if the value is below the upper quartile (75%) of the LIS. Also, for the discrete scoring system, three ecological classes were allotted, including good, moderate, and bad water quality.

2.7.6. Metric Validation

The applicability and efficacy of the final MMI for both the continuous scoring and discrete scoring systems were tested. The final multimetric index scores were computed for each station sampled per sampling expedition between 2011 and 2012 using 17 stations that fall within the LIS and the HIS. The index applicability effectiveness was assessed by comparing the number of stations that fall within good, moderate, and bad water ecological class water quality.

2.8. Relationship between Physico-Chemical Variables and Integrated Metrics

To ascertain the relationship between physico-chemical variables and the integrated metrics, we first conducted a linearity and unimodality test using DCA (detrended correspondence analysis). The test returned the datasets to be linear, as the DCA showed a gradient length <3 [60]. Hence, we used RDA (redundancy analysis) while correlating physico-chemical variables with integrated metrics. Highly multi-colinear variables (r ≥ 0.80) were dropped on the RDA ordination analysis. Additionally, we tested for level of significance of the first two axes of the RDA using global significance (Monte Carlo test @ 999 permutations; [60,61]).

2.9. Data Analyses

The vegan package version 2.5.4 within the R-programing language was used to compute PCA, DCA, RDA, and Monte–Carlo tests [62,63]. We used Version 13.4.14 of Statistica to construct box plots and compute Kruskal–Wallis test. Mann–Whitney (U), Kolmogorov–Smirnov, and Spearman (r) correlation coefficient were conducted on PAST (Palaeontological Statistical Software) version v4.03 [64], while variance and percentile distribution were conducted with Microsoft Excel 2016 version.

3. Results

3.1. Metric Sensitivity and Seasonality Tests

Eighteen (18) metrics out of the 67-candidate metrics initially selected were sensitive as they discriminated satisfactorily between LIS, and MIS/HIS, as shown from the constructed box plots and the non-parametric Mann–Whitney (U) test conducted (Appendix A Table A1). Figure 2 shows the visualisation of five out of the 18 metrics that were sensitive.

Of the 18 metrics that were sensitive, 11 metrics were seasonally stable (Appendix A Table A1). Figure 3 shows the visualisation and corresponding Kruskal–Wallis p value of five of the 11 metrics that were seasonally stable.

3.2. Metric Redundancy and Signal/Noise (Repeatability) Tests

Of the 11 metrics that were seasonally stable, seven proved to be non-redundant (

Table 2. Redundancy analysis showing metric multi-colinearity (r ≥ 0.78, p < 0.05) in this study. Note: values in boldface were significant at p < 0.05.

	Chironomidae Abundance	Chironomidae/Diptera Abundance	%Odonata	Diptera Richness	Chironomidae Rich	Chironomidae+Oligochaete Richness	Oligochaete Richness	Shanon Diversity	Margalef Index	Evenness Index	Logarithm Relative Abundance of Sprawler
Chironomidae Abundance	0.00	1.1 × 10−14	0.005	5.1 × 10−10	4.6 × 10−10	1.2 × 10−9	0.0180	8.6 × 10−8	0.00056	0.20	0.82
Chironomidae/Diptera Abundance	0.77	0.00	0.060	0.0023	0.00050	3.6 × 10−5	0.0056	0.00028	0.030	0.051	0.053
%Odonata	0.34	0.21	0.00	0.07	0.20	0.25	0.27	1.2 × 10−6	0.00049	0.098	0.031
Diptera Richness	0.68	0.35	0.22	0.00	1.3 × 10−30	2.2 × 10−19	0.090	3.3 × 10−9	1.7 × 10−10	0.020	0.42
Chironomidae Rich	0.64	0.42	0.16	0.93	0.00	3.0 × 10−18	0.54	3.4 × 10−8	3.5 × 10−8	0.062	0.23
Chironomidae+Oligochaete Richness Rich	0.63	0.48	0.13	0.84	0.83	0.00	3.3 × 10−7	2.0 × 10−10	1.6 × 10−10	0.21	0.31
Oligochaete Richness	0.29	0.32	0.13	0.21	0.077	0.57	0.00	0.00054	0.00035	0.63	0.099
Shanon Diversity	0.50	0.47	0.50	0.53	0.53	0.60	0.40	6.3 × 10−32	1.8 × 10−13	0.012	0.022
Margalef Index	−0.20	0.22	0.20	−0.29	−0.23	−0.16	0.058	0.025	0.00	0.044	0.00083
Evenness Index	0.40	0.28	0.42	0.68	0.60	0.68	0.40	0.8	−0.23	0.00	0.10
Logarithm relative abundance of Sprawler	0.04	0.24	0.28	0.10	0.14	0.13	0.20	0.13	0.22	0.40	0.00

). The signal/noise test conducted showed two out of the seven non-redundant metrics to be noisy. The metric includes Diptera richness and Simpson diversity. Hence, they were removed from integration into the final MMI (Appendix A Table A1). The remaining five metrics and their corresponding signal/noise values are: Chironomidae/Diptera abundance (2.56), % Odonata (3.98), Oligochaeta richness (2.90), Margalef index (2.78), and the logarithmic-transformed relative abundance of sprawler (3.97), and they were integrated into the final MMI (Appendix A Table A1).

3.3. Scoring/Integration of Metric and Index Development

3.3.1. Continuous Scoring System

As with metric scoring, the 95th and 5th percentiles of the LIS and HIS were used as either the scoring floor or scoring ceiling, depending on the predictable response of each metric to urban–agricultural pollution. The 95th and 5th percentiles of the retained metric at the LIS and HIS are presented in

Table 3. Metric values 95th and 5th percentiles of the five retained metric components integrated into the final MMI in this study-continuous scoring system.

Metrics Retained	LIS Percentiles		HIS Percentiles
	95th	5th	95th	5th
Chironomidae/Diptera Abundance	0.50	0.35	0.46	0.00
%Odonata	20.13	15.22	18.06	0.68
Oligochaete Richness	2.00	1.30	0.05	0.00
Margalef Index	8.98	7.70	5.41	3.47
Logarithm relative abundance of Sprawler	1.64	1.36	0.97	0.19

. The final MMI was calculated by adding the scores of each of the five retained metrics after the signal/noise test. The standardization metrics were done using the 95th and 5th percentiles of the LIS and HIS (

Table 4. Final MMI formula for the retained metric components in this study-continuous scoring system. Note: MVa (Chi/Dip Abun at HIS metric value), MVb (%Odo at LIS metric value), MVc (Oli Rich at HIS metric value), MVd (Mar Ind at LIS metric value), and MVe (Log Spr at HIS metric value).

Retained Metrics	Predicted Response to Urban–Agricultural Pollution	Metric Formula	Simplified Metric Formula
Chironomidae/Diptera Abundance (MVa)	Increase	(0.46-MVa)/(0.46–0.35) → (0.46-MVa)/0.11	(0.46-MVa)/0.11)
%Odonata (MVb)	Decrease	(MVb-0.68)/(20.13–0.68) → (MVb-0.68)/19.45	(MVb-0.68)/19.45
Oligochaete Richness (MVc)	Increase	(0.05-MVc)/(0.05–1.30) → (0.05-MVc)/−1.25	(0.05-MVc)/−1.25
Margalef Index (MVd)	Decrease	(MVd-3.47)/(8.98–3.47) → (MVd-3.47)/5.51	(MVd-3.47)/5.51
Logarithm relative abundance of Sprawler (MVe)	Increase	(0.97-MVe)/(0.97–1.36) →	(0.97-MVe)/−0.39

). We awarded a score of 0 to any metric value that is 0 or less, and a score of 1 to any metric value that is >0. Finally, we assigned three ecological classes based on the final index scores: good (4.0–5.0), moderate (2.0–3.0), and bad water quality (0.0–1.0).

3.3.2. Discrete Scoring System

The metric values of the LIS were used as thresholds for computing the scores for each metric (

Table 5. Selected metrics statistics used for the development of the MMI in this study-discrete scoring system.

Retained Metrics	Statistical Calculations					Score
	Max. Value	75%	50%	25%	Min. Value	1	3	5
Chironomidae/Diptera Abundance	0.50	0.48	0.38	0.35	0.34	>0.50	≥0.48–<0.50	<0.48
%Odonata	20.40	18.02	16.22	16.00	15.00	<15.00	15.00–<16.007	≥15.94
Oligochaete Richness	2.00	2.00	2.00	2.00	1.00	>2.00	2.00	<2.00
Margalef Index	9.00	8.80	8.26	7.95	7.64	<7.64	7.64–<7.95	≥7.95
Log Spr	1.67	1.57	1.42	1.37	1.36	>1.67	≥1.57–<1.67	<1.57

). We arrived at the final MMI by summing up the scores of the five metrics that scaled through all the tests, and 5–25 was used as the range for the index value since five metrics were retained for integration into the final MMI (5 × 5 = 25). Based on this value range, three ecological classes were assigned for the MMI based on the final index scores: good water quality (19–25), moderate water quality (12–18), and bad water quality (5–11).

3.4. Metric Validation

3.4.1. Continuous Versus Discrete Scoring Systems

We used 17 stations that fall within LIS and HIS using the dataset we collected within 2011 and 2012 in a separate river system flowing through urban–agricultural catchments within the Niger Delta region. For the continuous scoring system, the validation of the MMI results revealed that 0.00% of the selected stations sampled for metric validation had good water quality, 94.12% of the selected stations sampled had moderate water quality, and 5.88% of the selected stations sampled had bad water quality. Thus, the result of the MMI validation for the continuous scoring system showed that the final MMI is more applicable to moderately polluted river systems draining urban–agricultural catchment in the Niger Delta region of Nigeria. Hence, the continuous scoring system cannot be said to be totally more effective in determining the ecological status of river-draining urban–agricultural catchments in the Niger Delta region of Nigeria across all the ecological classes identified in this study.

On the other hand, for the discrete scoring system, the validation of the MMI results revealed that 1.89% of the selected stations sampled for metric validation had good water quality, 75.47% of the selected stations sampled for metric validation had moderate water quality, and 22.64% of the selected stations sampled had bad water quality. Thus, the result of the metric validation for the discrete scoring system showed that the final multimetric index is also more effective in moderately polluted river systems draining urban–agricultural catchment in the Niger Delta region of Nigeria. Therefore, the discrete scoring system cannot also be said to be more effective in determining the ecological status of rivers draining urban–agricultural catchments in the Niger Delta region of Nigeria.

Overall, in the present study, there is no distinct difference between the effectiveness of the continuous and discrete scoring systems.

3.4.2. Relationship between Physico-Chemical Variables and the Metrics Integrated into the MMI in This Study

The first axis of the RDA explained 80.5% of variations and the second axis explained 19.45% of variations with corresponding Eigen values of 31.087 and 0.064, respectively. The global statistical significance test (Monte Carlo) test revealed that the first two axes of the RDA showed no significant difference (p > 0.05). A positive correlation was noted between pH and %Odonata and logarithmic relative abundance of the sprawlers, and they correlate negatively with LIS at the first axis and MIS at the second (Figure 4). Water temperature, nitrate, and BOD₅ correlate strongly positively with Oligochaete richness and Chironomidae/Diptera abundance at HIS at the first axis (Figure 4).

4. Discussion

The selected rivers stations in the Niger Delta region of Nigeria were classified into three impact classes, which include LIS, MIS, and HIS based on physico-chemical characteristics. This classification aimed to assess the diverse levels of impact on water quality related to predominant land-use types, including urban and agricultural land uses. Changes in water quality in the study area are mainly influenced by forestry, agriculture, and urbanization. Stations designated as MIS and HIS showed significantly higher nutrient contents in both urban and urban–agriculture catchments.

A thorough selection process resulted in the retention of only five measures for incorporation into the final MMI–urban–agriculture from a pool of 67 potential macroinvertebrate metrics tested. Following a thorough investigation, 18 of the 67 potential metrics demonstrated effective discriminatory abilities, clearly distinguishing between the LIS, the MIS, and the HIS, confirming their sensitivity. Notably, metrics such as Oligochaete, Chironomidae+Oligochaete, and Diptera/Chironomidae that have been widely reported to be pollution-tolerant were identified as critical contributors to the efficacy of abundance, composition, and richness measurements in determining impact levels [17,24,33]. Chironomidae and Oligochaete composition and abundance have been consistently reported to dwell more in river stations subjected to increased urban and agricultural influences [33]. These specific family/order of macroinvertebrates, recognized for their tolerance to a variety of contaminants, have increased in quantity or prevalence, acting as dependable indicators of environmental degradation in such environments. As revealed in previous studies e.g., [19,33,65], the links between Oligochaete and Chironomidae and nutrient enrichment have been a focal point in freshwater biomonitoring. Chironomidae genera, such as Chironomus, can trap oxygen from the air using blood tissues (haemoglobin) and, thus, resist biological enrichment and contamination in water [65,66]. Recent findings by Macedo et al. [67] highlighted the distribution and diversity of Chironomidae in the headwaters of streams within the catchments of a Neotropical savanna affected by hydroelectric power plants, implying that catchment disturbances negatively impact stream functionality, resulting in the distribution and diversity of pollution-tolerant taxa such as Oligochaete and Chironomidae. These organisms thrive in nutrient-rich habitats influenced by anthropogenic activities such as agricultural runoff and urban pollution. The importance of researching the links between Chironomidae and Oligochaetes and nutrient levels stems from their potential use as bioindicators [66].

This strategy of selecting final measurements based on their sensitivity to water-quality deterioration is consistent with comparable tactics used in studies such as the one conducted by Mereta et al. [3]. By carefully evaluating the biological preferences and responses of various macroinvertebrate taxa, the chosen metrics become effective instruments for the assessment of the deteriorating impact of water quality on riverine ecosystems. This method improves the precision and reliability of multimetric indices, increasing their utility in biomonitoring and ecological assessments. Numerous research has looked into the ecological implications of these macroinvertebrate taxa in connection to nutrient levels [33,44]. The susceptibility of Chironomidae, a family of aquatic insects, and Oligochaete, an order of segmented worms, to nutrient-rich conditions is well known. Their abundance and distribution in aquatic habitats frequently reflect variations in nutrient concentrations, particularly when organic matter and fertilizer inputs are increased.

According to this study, agricultural activities and urbanization have a negative impact on rivers in the Niger Delta. Previous research has shown that industrial, municipal, and agricultural discharges can alter river and stream water quality [68,69]. This suggests that agricultural and urban pollution are influencing the composition and function of macroinvertebrates. This influence is visible in the current study, where taxa that are resistant to pollution, such as Chironomidae and Oligochaete, predominate.

4.1. Metric Validation: Applicability and Effectiveness of Continuous Versus Discrete Scoring Systems

The signal/noise test and subsequent removal of noisy metrics showed a methodical approach to improving the MMIs. Because of the noise in this setting, these measurements may exhibit considerable fluctuation or uncertainty, making them less credible indicators for ecological assessment. The validation process for the developed MMI–urban–agriculture, using different datasets, revealed excellent performance in the moderate-water-quality ecological class. The continuous scoring system achieved a 94.12% success rate, while the discrete scoring system achieved 75.47% for the moderate-water-quality ecological class. However, performance in the good- and poor-water-quality ecological classes was less impressive. This pattern highlights the prevalence of moderate pollution levels, indicating that the final multimetric index is particularly effective at assessing and categorizing river systems in urban–agricultural catchments with moderate pollution. The relatively high number of stations classified as having poor water quality raises concerns about the overall suitability of the discrete scoring system in defining the ecological status of river systems in urban–agricultural catchments in the current study area.

The findings support the urban river syndrome, which is characterized by increased nutrient levels, suspended particles, and changes in channel structure and stability in urban rivers [70,71,72,73]. Recognizing these effects highlights the impact of human activities on water quality in the rivers of the Niger Delta’s urban, urban–forestry, and urban–agriculture catchments. Hence, there is a need for the development of comprehensive tools in this region, like water-chemistry analysis, environmental integrity evaluation, and biological information [74,75,76]. For instance, Stoddard et al. [54] found that combining natural environmental variations with human-induced pressures introduced additional variability in multimetric indices and could significantly bias ecological assessments. Other research has demonstrated that natural landscape elements influence macroinvertebrate communities in Neotropical Savanna streams [77], and that these natural influences can have a more pronounced effect on macroinvertebrate responses than human-induced pressures. Nonetheless, for the creation of an MMI, testing the stability of selected metrics with natural seasonal variations is essential [57]. Some studies on seasonal variability supported the hypothesis that this stage is significant [78].

There have been a series of debates that continuous scoring is more responsive, applicable, and less subjective than the discrete scoring system [4,34,44,79], but this study has shown that both scoring systems have almost similar levels of applicability and effectiveness in determining the ecological status of river-draining urban–agriculture catchments. For instance, Blocksom [79] had opined that although discrete scoring system had an effect on the final multimetric index variability, it could alter the ecological categorisation in a fairly minor rate. This supports the fact that the issues of applicability and effectiveness of the developed multimetric indices are not with the metric scoring system adopted by researchers but with the criteria adopted for reference station selection, size of stations marked for a study, sample size, and metric selection criteria in building multimetric indices [34,74]. Conversely, Stoddard et al. [54] stated that the continuous scoring system should be maintained in order to get metrics that are best performing to be integrated into multimetric indices. Also, consistency and reproducibility are required when deciding on the metric scoring approach to be adopted in developing multimetric indices [52]. Overall, we suggest that the use of both continuous and discrete scoring systems in developing future multimetric indices to confirm their level of applicability and effectiveness in determining the ecological status of riverine systems draining urban–agriculture catchments and other land-use types around riverine-system catchments across the globe.

4.2. Correlating Physico-Chemical Variables with Integrated Metrics

The RDA findings provide a thorough examination of the correlation between physico-chemical parameters and macroinvertebrate metrics in urban–agricultural river systems. However, at the HIS on Axis 1, BOD, nitrate, and water temperature exhibited strong positive correlations with Oligochaete richness and Chironomidae/Diptera abundance, indicating that these metrics are influenced by pollution indicators associated with higher-impact categories when a station is influenced by a single dominating stressor classification based on abiotic factors, which is a reliable strategy. This was validated by the connection of %Chironomidae+Oligochaeta with water temperature, conductivity, BOD and nitrate, and in our study’s RDA. The inclusion of these measurements is consistent with earlier research that has included similar indications in multimetric indices [3].

Because of their susceptibility to pollution, the metrics linked to abundance and composition are popular candidates for inclusion in multimetric indices. Several research in the field supports this recognition [25,33,44]. Further, because of their sensitivity to disturbances, particularly pollution, abundance and composition metrics are significant indicators in biomonitoring studies. Findings of similar studies have emphasized the discriminating potential of several diversity indices [23]. Diversity measures are important in biomonitoring studies because they provide information on the structure and variability of the communities of macroinvertebrates in response to environmental changes.

This study used different measures of diversity like Simpson diversity, Evenness, Shannon–Wiener, and Margalef’s indices, successfully distinguishing between various conditions. Specifically, the Evenness index showed particular sensitivity, indicating its usefulness in distinguishing between different types of impacts. This finding is congruent with previous studies e.g., [17,23,24], which found that several diversity indices have significant discriminatory potential. Edegbene et al. [17,24] used the Shannon diversity and Margalef indices for the development of MMI for a Nigerian river, demonstrating the utility of diversity indices as biomonitoring tools for understanding river system ecological health. Furthermore, Aura et al. [30] incorporated Shannon diversity and Margalef indices into multimetric indices, highlighting the versatility and application of diversity measures in evaluating water-quality status. The sensitivity of these indicators to various human-induced stressors is widely accepted [80,81]. They are frequently used in the assessment of wetlands [82]. The inclusion of Chironomidae and Oligochaeta in the construction of the MMI emphasizes their complementary roles in increasing the index’s sensitivity. The MMI becomes more robust in its ability to reflect and respond to environmental stressors by integrating these two groups, boosting its effectiveness as a bioassessment tool. Numerous studies undertaken in various climatic zones have used these metrics as important components in the construction of MMI [58,83,84]. While many studies have shown that the abundance of Chironomidae increases with habitat degradation [85], several authors [86] recommend identifying them at the genus or species level before using them as water-quality indicators. Notably, Chironomidae/Diptera and Oligochaeta have been found as bioindicators in freshwater systems that respond positively to rising anthropogenic activity. This was validated by the connection of %Chironomidae+Oligochaeta with water temperature, conductivity, BOD, and nutrient levels (nitrate) in our study’s Redundancy Analysis (RDA). The inclusion of these measurements is consistent with earlier research that has included similar indications in MMIs [3,17]. Diptera and Oligochaete presence have been noted to exhibit a negative correlation with a previously established MMI, as reported by Chowdhury et al. [86].

Although some tropical studies have shown that metric measures related to functional assemblage (e.g., feeding types or habits) respond differently to increased anthropogenic influences [87], many authors have recommended using trophic measures for biomonitoring, especially in multimetric systems [5,57,88]. The impact of urbanization and agricultural activities on the quality of river water and biological communities have become major public concern [10,89]. This concern is warranted due to the documented negative effects on river biological balance in such areas. This is emphasized in this study, particularly with increasing rural–urban mobility and continuous agricultural activities in the Niger Delta catchments under investigation.

5. Conclusions

The developed MMI for rivers draining urban–agriculture catchments in the Niger Delta region of Nigeria proved to be robust and effective for detecting various human-induced impacts, including changes in land use and physical and chemical degradation. For LIS (least-impacted stations), the metric performed exceptionally well, showing an 83.3% correspondence with the physico-chemically based station classification. There was a 75% correlation between the index values and the physico-chemically based classification for MIS (moderately impacted stations). However, only a 22.2% correspondence was found for HIS (heavily impacted stations), showing reduced precision in such circumstances.

The MMI is a reliable monitoring tool specifically tailored to the study area, with potential applications in establishing biomonitoring networks in other highland regions across the country. Due to its adaptability to various anthropogenic stressors, the index is a valuable asset for monitoring the health of riverine ecosystems in the specified regions. Its ability to capture the effects of land-use changes, as well as physical and chemical alterations, enhances its usefulness for comprehensive environmental monitoring. Therefore, it is a crucial tool for environmental managers and government authorities engaged in the regular monitoring of rivers and streams affected by urban and agricultural pollution. The current study contributes greatly to our knowledge of the ecology of riverine ecosystems in the Niger Delta, particularly those subjected to urban stressors and agricultural practices. However, the current study is not all-encompassing with regards to detecting the levels of disturbances around riverine systems across different landscapes, as we only used specific land-use types. Hence, we recommend that a more robust MMI should be developed that will include all existing land-use types in the studied region of Nigeria.