Exploring Macroinvertebrates Ecological Preferences and Trait-Based Indicators of Suspended Fine Sediment Effects in the Tsitsa River and Its Tributaries, Eastern Cape, South Africa

: The taxonomy-based response pattern of macroinvertebrates to sediment stress is well established, with tolerant taxa increasing in impacted conditions, while sensitive taxa decrease along a deteriorating water quality gradient. However, the distribution patterns of traits in response to environmental stress gradient, including suspended sediments, remain unclear, particularly in Africa, where trait-based studies are under-explored. We examined the distribution patterns of macroinvertebrate traits along a suspended sediment stress gradient and identiﬁed tolerant and sensitive traits for suspended sediment stress. We sampled macroinvertebrates and environmental variables seasonally in winter, spring, summer and autumn of 2016 to 2018 in eight selected sites in the Tsitsa River and its tributaries. We selected 12 traits and ecological preferences, resolved them into 47 trait attributes, and analysed them using the RLQ and fourth-corner analyses. Our results revealed that macroinvertebrate traits and ecological preferences were differentially inﬂuenced by ﬁne suspended sediments in the Tsitsa River and its tributaries. Traits such as a preference for CPOM, collector-ﬁltering, and a high sensitivity to oxygen depletion, were deemed sensitive to suspended sediments stress, exhibiting positive associations with the control sites, and negatively associated with any of the environmental parameters (sediment grain sizes, turbidity, TSS and EC). Tolerant indicator traits included a high tolerance of oxygen depletion, skating and a preference for FPOM. The fourth-corner analysis results indicated that suspended ﬁne sediment grain sizes, (including coarse sand, ﬁne silt and clay) were the most important variables inﬂuencing macroinvertebrate trait distribution patterns during the dry season, while gravel, mud and medium sand were more important during the wet season. Overall, our study provided critical insights towards trait-based responses of macroinvertebrates communities to suspended sediment stress, key information that could stimulate the development of macroinvertebrate trait-based biomonitoring tools for the assessment of suspended sediment stress in the Afrotropical region.


Introduction
Sedimentation is among the most common freshwater ecosystem stressors impacting on macroinvertebrate communities [1][2][3]. Fine sediments' impacts on aquatic biota are wide-ranging and can be profound because their effects can be complex and are mediated by a range of factors including exposure duration, particle size distribution, sediments load, sources, geomorphological setting, and the vulnerability of resident biota [4][5][6]. Sediment effects on macroinvertebrates can be direct, e.g., clogging of fragile and exposed gills and filter-feeding structures, burial, abrasion; or indirect effects, e.g., alteration of the physical and chemical condition of streams such as reduction of dissolved oxygen and increases in turbidity [1,[7][8][9].
Macroinvertebrates assemblages have been reported to change in response to humaninduced stressors in river and stream ecosystems. Globally, the taxonomy-based ap-identification as not all traits are constrained by taxonomy, for example, body shape, body size, and many ecological preferences of macroinvertebrates at family level [35]. Thus, in the present study, we examined the effects of suspended fine sediments on macroinvertebrate communities by means of multivariate RLQ and fourth corner analysis. Eight sites that represent an increasing fine sediments loads were selected for this study. The objectives of this paper, therefore are to (i) explore the distribution of macroinvertebrate ecological preferences and biological traits in relation to elevated suspended fine sediments and (ii) identify trait-based indicators potentially useful for monitoring sedimentation effects.

Study Area and Selected Sampling Sites
This study was conducted in the Tsitsa River catchment in five streams, which include the Tsitsa, Qurana, Pot, Little Pot Rivers, and the Millstream (Figure 1). The Tsitsa River catchment is situated in the Eastern Cape Province of South Africa, with a catchment area of about 4924 km 2 . The Tsitsa River and its tributaries form part of the broader Mzimvubu River catchment. The Tsitsa rises in the Drakensberg 15 km to the south-east of Rhodes, a small town close to Maclear and about 80 km west of Mount Frere, and flows eastwards. The Tsitsa catchment is subject to widespread gully erosion, having more than 10,000 gullies [36]. Gullies, together with the duplex and dispersive soils, are the contributing factors of fine sediments influx into the river system. The Tsitsa River serves as an ecological asset in providing basic ecosystem services such as water and aquatic resources that are beneficial to rural and peri-urban communities on the catchment. Additionally, the river is one of the few remaining rivers in South Africa in their near-natural state. The river is a source of water supply for subsistence agricultural activities such as livestock farming, and rural settlements that rely on it for drinking and cooking.
The study was conducted seasonally at eight selected sampling sites over a period of two years, beginning in late winter (August 2016), spring (October 2016), summer (December 2016), autumn (March 2017), winter (July 2017), spring (September 2017), summer (December 2017) and ending in autumn (March 2018). The selected sites included two sites in the Tsitsa River (Sites 1 and 2), one site in the Qurana River (Site 3), two sites in the Millstream (Sites 4 and 5), two sites in the Pot River (Sites 6 and 7) and one site in the Little Pot River, i.e., Site 8 ( Figure 1). The sites were selected to indicate a gradient of sediment impact based on turbidity, total dissolved solids (TSS) and land use practices (privately owned, well-maintained catchment versus communally owned, poorly maintained landscape). The site classification was based on the extent of erosion, land-use practices and a previous study [5] Other factors that were considered when selecting the sites were the availability of macroinvertebrate biotopes: stones, vegetation and gravel, sand and mud. Sites 1, 2 and 3) were considered as the highly sedimented sites. The Sites in the Millstream (i.e., Site 4 and Site 5) were considered as moderately sedimented. Sites 6, 7 (i.e., Pot upstream and Pot downstream) and Site 6 (i.e., Little Pot) were situated in the well-maintained area of the catchment and were collectively referred to as the control sites (CLS).

Sampling of Selected Water Quality Variables and Fine Sediment Grain Sizes
Physico-chemical variables were measured seasonally at all the sampling sites. For each sampling event, the selected physico-chemical variables measured on-site include dissolved oxygen (DO), electrical conductivity (EC), temperature and pH using a multiparameter probe; model H198, Hanna instrument. Turbidity was measured on site using the portable turbidity Orbeco-Helliage 966 Metre. Water samples were collected and transported to the laboratory for analysis of total suspended solids (TSS). TSS was measured through the filtration and oven-dried method according to the protocol described by [37].
Water samples for sediments grain size characterisation were collected using the disturbance technique [38]. To sample suspended sediments, an open-ended, cylindrical polyethylene bucket (height 75 cm; diameter 48.5 cm) was carefully inserted into the water column. The water column within the cylindrically shaped container was then agitated using a wooden pole of about 15 cm long. The agitation of the water column was done so that the stream bed is not disturbed, in order not to collect settled fine sediments. While the water was still vigorously in motion, suspended fine sediments samples were collected and then through filtered 2000 µm pore size sieve into 250 mL acid washed sampling bottles. Filtration remove particles larger than 2000 µm such as debris. Filtered samples were then transported to the laboratory and refrigerated until analysed. Suspended fine sediments grain sizes were characterised using the Mastersizer 3000 laser diffraction particle size analyser as fully described in [5]. The fine sediment grain size output from the Mastersizer 3000 was further analysed for grain size classes using the GRADISTAT version 8.0 [38]. Suspended fine sediments were separated into gravel, mud, sand, silt and clay. Sand was composed of very coarse sand (>1000-2000 µm), coarse sand (<2000-1000 µm), medium sand (<1000-500 µm), fine sand (<500-250 µm) and very fine sand (<250-125 µm), silt in µm was composed of very coarse silt (<125-63 µm ), coarse silt (<63-31 µm ), medium silt (<31-16 µm), fine silt (<16-8 µm), very fine silt (<8-4 µm) and clay (<4 µm).

Macroinvertebrates Sampling
Concurrent with physico-chemical sampling, macroinvertebrates were collected using a 30 × 30 cm, 1000 µm mesh net in accordance with the South African Scoring System version 5 (SASS5) protocol [15]. The SASS5 protocol is a standardised kick sampling technique that requires the collection of macroinvertebrates from three distinct biotopes: stones (in-current and out-of-current), sediments (gravel, sand and mud; GSM), and aquatic and marginal vegetation. The samples were pooled and analysed as composite samples per site, per sampling event. Three replicate samples were collected per site per sampling event over the study period. Samples were preserved in jars containing 75% ethanol and transported to the laboratory for sorting and identification [39]. In the laboratory, the samples were identified to family levels according to the identification guide by [40].

Selected Macroinvertebrates Traits and Ecological Preferences
A total of 12 traits and ecological preferences were selected for the study and further resolved into 47 attributes ( Table 1). The selection of trait and ecological preferences was informed by the literature and mechanistic link between fine sediment modes of stress and the particular traits. Overall, the selection of traits and ecological preferences were informed by (i) mechanistic relationships between the trait and fine sediment modes of stress, (ii) availability of trait and ecological preference data, and (iii) ease of measurement and observation. For example, respiration is selected for analysis because fine sediments have been hypothesised to clog respiratory trait attributes such as gills [11]. Feeding was also selected because, like respiration, organisms which feed by filtering particulate organic matter have been shown to have their feeding apparatus clogged by elevated fine sediments. Accumulation of fine sediments has also been shown to alter food quality, cover stable surfaces on which many macroinvertebrates feed [38]. Concerning velocity preference, for example, Odume et al [6] and Jones et al [7] [ argued that velocity mediates fine sediment impact on macroinvertebrates because, at a higher velocity, the frictional force between organisms' body surfaces and the moving fine sediments is likely to be aggravated causing increased abrasion of soft and exposed body surfaces. Information on trait and ecological preferences was retrieved from the newly compiled trait database for South African macroinvertebrates at family level [24,41] and supplemented by other sources [5,42,43]

Data Analysis Exploring the Distribution Pattern of Ecological Preferences and Traits and Identifying Trait-Based Indicators of Elevated Suspended Fine Sediments
A two-way multivariate analysis of variance (MANOVA) was employed as a statistical test method to test for differences in suspended sediment grain sizes and physico-chemical variables between the sites and seasons. MANOVA simultaneously used multiple independent variables to compare the sites in terms of the sediment particle size distribution and the physico-chemical variables. While using MANOVA, the basic assumptions of normality and homogeneity of variance need to be examined using the Shapiro-Wilk test and the Levene's test, respectively. If assumptions are not met, then data were transformed logarithmically, but normalized if assumptions were still not met. One-way ANOVA followed by a Tukey's post hoc test was undertaken to indicate where the significant differences lay.
The macroinvertebrate data were consisted in two matrices, a taxon-site matrix (L) and a trait-taxon matrix (Q), with the trait data fuzzy-coded and abundance data log (x + 1) transformed. Fuzzy coding (Supplementary Data, Table S1) was used to describe the association of each taxon to each trait attribute. Affinity scores ranging from 0 to 5 were used, considering potential functional variation between species within a family and also in between life stages within a taxon [33]. Each taxon per trait attribute was assigned a score of 0 indicating no affinity, 5 high affinity to a trait attribute [33]. The fuzzy coding was particularly useful because working at the family level, it allows for potential plasticity, variability, and functional diversity that exist within a given family. The approach has been widely used in other studies [42,44,45].
RLQ analysis for each season was used to relate fine sediment grain sizes, TSS, and turbidity, DO, EC (R), macroinvertebrates taxa (L), and the traits and ecological preferences (Q) along a gradient of increasing sedimentation. RLQ was developed by Doledec [5] for three-dimensional data analysis: environmental data, taxa and trait data. In RLQ ordination, the first ordination (correspondence analysis, CA) is performed on the taxa data set L-table, second ordination (principal component analysis, PCA) on the environmental data sets, in this case, the sediment grain sizes, turbidity, EC, TSS and DO, R-table, which links the taxa data set to the physico-chemical variable data set by using the sample scores result of the CA as row weights. A third ordination was performed to links the L data set to the trait data set by using the taxon score results of the CA as row weights. A final ordination (combined RLQ) analysis was performed that simultaneously conducts ordination on the three separate ordinations (CA, PCA, and HS) by searching for a linear combination of traits-taxon scores in the traits-taxon scores in Q-HS ordination and physico-chemical variables sample scores in R-PCA ordination, by maximising the covariance between Q and R through L ordination. The significance of the RLQ analysis was tested using the Monte Carlo permutation test with 999 permutations at alpha = 0.05. The RLQ ordination thus allows the spatial visualisation of the distribution of traits and ecological preferences in relation to the sediment grain sizes and physico-chemical variables. RLQ were undertaken using the ADE-4 statistical package for R version 3.4.1 in R-programming environment [46].
To identify potential trait-based indicators of suspended fine sediment stress, ecological preferences and trait attribute associated with sites in the Tsitsa River and Qurana River were designated as tolerant traits, whereas trait attributes associated with the control sites were designated as potential fine sediment sensitive traits. The Fourth-corner analysis [46] was then further conducted to confirm designated trait-based indicators of suspended fine sediment effects. The Fourth-corner analysis is a multivariate permutational test that searches for significant association between traits and environmental variables. In this study, it was used to test the association between suspended sediment grain sizes, turbidity, TSS, and DO with the selected traits/ecological preferences. A trait was confirmed to be a fine sediment tolerant if it was positively associated with highly impacted sites and positively correlated with at least one environmental variable (TSS, EC, turbidity) or any of the suspended grain sizes associated with the highly sedimented sites. A trait was confirmed as a sensitive indicator trait if the correlation was significantly negative with TSS, turbidity, or any of the suspended grain sizes associated with the highly sedimented sites [47].

Physico-Chemical Variables and Grain Size Distribution
The means, standard deviations, and ranges of the basic physico-chemical variables including dissolved oxygen (DO), total suspended solids (TSS), Electrical conductivity (EC) and turbidity, recorded in the Tsitsa River and its tributaries for the current study are presented in Table 2 and only Turbidity and TSS were statically significantly different between sites. Sites (i.e., TSU, TSD, and QHR) were mainly dominated by very fine sand, very fine silt, medium silt, very coarse silt, and clay (Table 2). Whereas, the two sites situated in the Millstream (i.e., MLU and MLD) were primarily dominated by clay, constituting more than 60% of the volumetric grain sizes at the two sites. Grain sizes within the control sites were evenly distributed, with the dominant grain size (i.e., coarse silt) constituting only 15% of the overall difference in (Table 2). Overall differences in grain sizes were statistically significant (p < 0.05) across the sites, but not across the two seasons as shown in (Table 3). The interactions between the sites and seasons in terms of the suspended sediment grain sizes were statistically not significant (p > 0.05). The oneway ANOVA results indicated that very fine sand, very coarse silt, medium silt, and fine silt were significantly higher at TSU, TSD, QHR compared with the MLD, MLU, and CLS. The rest of the grain sizes did not differ statistically between sites. Table 2. Means, ± standard deviations, and ranges (in parentheses) for the volumetric distribution of suspended fine sediment grain sizes and physico-chemical variables across sites in the Tsitsa River and its tributaries. Different superscript letters for very fine sand, very coarse silt, medium silt, fine silt, and total suspended solids across sites that indicate significant differences (p < 0.05) revealed by Tukey HSD post-hoc test. The same superscript letter between sites per variable indicates no significant differences (p > 0.05). The sediment grain sizes are reported a volumetric fraction of 1.

Suspended Sediment
Grain

Spatial Distribution of Traits and Ecological Preferences
The results of the RLQ analysis during the dry season revealed that the first two axes explained 99.1% cumulative variance of the dataset. The first axis accounted for 81.94% and the second axis 17.16% of the total variance. The ordination plot revealed that medium silt, coarse sand, and medium sand were clustered together and were positively associated with QHR, MLU, MLD, and CLS. During the dry season, except on few occasions, the control site (Site 6-8) and the moderately sedimented sites, i.e., 4 and 5 (MLU and MLD) were closely clustered together and were mainly influenced by increasing DO and clay. The traits that were mainly associated with the control sites were a preference for climbing, aerial respiration, temporary attachment and skating. These traits were associated with taxa such as Dyticidae, Oligonuridae, Baetidae, Syphidae and Muscidae (Figure 2). The highly sedimented Sites 1 (TSU) and 2 (TSD) and Site 3, Qurana River (QHR) were clustered together. Ceratopogonidae, Baetidae, Caenidae, Leptophlabidae and Helodidae taxa were found to be associated with these sites. Traits associated with these taxa include a preference for very fast-flowing waters, predation and scraping. These sites clustering was mainly influenced by increasing fine sand, very fine silts and TSS (Figure 2). For the wet season, the first two axes of the RLQ explained 92.87% cumulative variance. The first axis accounted for 81.87% variance, and the second axis 11% total variance. The ordination plot revealed that MLU, QHR), TSD, and TSU) were clustered together and showed a positive association with fine sand, coarse silt, fine silt, turbidity, and TSS. Collector-filterers, shredders, CPOM, and a preference for slow-flowing waters were the traits attributes associated with these sites (Figure 2). These traits were associated with taxa such as Oligonuridae, Syphidae, Dyticidae and Muscidae.

Identifying Trait-Based Indicators of Fine Sediment Stress
The fourth-corner analysis was used to test the significance of individual trait-environment association, and to further explore the significant association between the individual trait attribute/ecological preference and suspended sediments grain sizes, turbidity, EC, DO and TSS, (Figure 3). During the dry season, only coarse sand, fine silt and clay correlated significantly with macroinvertebrate traits and ecological preferences (Figure 3). A positive correlation was detected between fine silt, and large (>20 to 40 mm) and very large (>40 mm) body sizes, possession of lungs and spherical body shape, whereas clay was negatively correlated with crawling, a high tolerance of DO depletion and all attributes of velocity preference (m/s), except slow-flowing (0.1-3 m/s). Coarse sand indicated significant positive correlations with gills, crawling, CPOM, scraping, a preference for stone biotope, aerial, and all attributes of sensitivity/tolerance to dissolved oxygen depletion and velocity preference, except very fast-flowing (>0.6 m/s) ( Table 4).  Table 1, for traits and ecological preferences, physico-chemical variables and suspended grain sizes in Table 2   −0.034 * −0.052 0.034 * During the wet season, a total of 36 traits attributes and ecological preferences were significantly correlated with suspended fine sediments grain sizes. Out of the 36 traits/ ecological preferences that were significantly correlated, 18 traits attributes and ecological preferences such as small body size, medium body size, gills, crawling, swimming, a preference for FPOM, CPOM as food sources, collector-filtering, scraping, a preference for the stone biotope, high sensitivity to DO depletion, a moderate sensitivity to DO depletion were negatively significantly correlated with gravel and mud. Conversely, 18 traits attributes and ecological preferences were significantly positively correlated with coarse sand and medium sand. These traits include small, medium body sizes, gills, crawling, swimming, streamlined body shape, cylindrical body, a preference for FPOM, CPOM, predation, permanently attached, and aquatic respiration. Of the traits that negatively correlated with coarse sand and fine sand, CPOM, collector-filterer, high sensitivity to DO depletion were associated with the control sites. These traits were therefore deemed sensitive traits to elevated suspended fine sediments (Table 4). Of the traits that showed a positive correlation with medium sand and very fine sand, a high tolerance to DO depletion, skating, and a preference for FPOM were associated with the highly sedimented sites. These traits were thus deemed tolerant traits (Table 4).

Discussion
In this study, we explored the influence of elevated sediments on the distribution patterns of macroinvertebrate traits and ecological preferences in selected streams in the Tsitsa River Catchment, Eastern Cape Province of South Africa. Our results showed that elevated suspended sediments differentially affected macroinvertebrate traits and ecological preferences, consistent with other freshwater studies, e.g., [8,[47][48][49], marine [29] and wetlands ecosystems [29] who have demonstrated significant effects of fine sediments on macroinvertebrate traits. Traits such as collector-filterer, CPOM, and a high sensitivity to DO depletion were positively associated with the control sites; they were identified as potentially sensitive traits of suspended fine sediments. These traits and preferences exhibited significant positive correlations with any of TSS, EC, turbidity and suspended sediments grain sizes (Figure 3). Conversely, a high tolerance to DO depletion, skating and a preference for FPOM were associated with the highly sedimented sites and were deemed tolerant indicator traits of suspended fine sediments.
Macroinvertebrate feeding activities and preferences for food are commonly reported to be affected by elevated fine sediment loads as they are linked to nutritional quality or impaired access to food resources [8,50]. In the present study, CPOM was associated with less sedimented sites, indicating that elevated sediments may have affected macroinvertebrate that feeds on detritus, CPOM, through the reduction in food quality. When suspended sediments are deposited on stream bottom or substrates, including CPOM; the fine sediments can cover the food items, thereby reducing the quality/palatability and access to food for macroinvertebrates shredders, feeding on CPOM [50]. For example, [51] who investigated the effects of fine sedimentation on CPOM availability and shredder abundance in Alpine streams in the Pellice River, Italy found elevated fine sediments to significantly reduced the amount of (CPOM), affecting the abundance of invertebrate shredders. Similarly, [8] found invertebrate shredders to be the only feeding groups that demonstrated reduced abundance with increased fine sediments in two lowland UK streams. Further, the palatability of CPOM for shredders depend on the initial actions of microbes on leaf litter [52]. However, suspended sediments can inhibit the rate of microbial breakdown of litter, through reduced light energy and temperature [52], thereby affecting the nutritional quality of CPOM for macroinvertebrate shredders. With regards to collector-filters, it is likely that suspended sediment in water column negatively affected filterers through clogging of respiratory and feeding apparatus, and thus their predominant association with the less sedimented sites. The sensitivity of CPOM and collector-filterers are consistent with other studies [52,53]. Rabeni et al. [50] studied changes in functional feeding groups associated with sedimentation and found a greater proportion of gatherers and a lower proportion of filterers and scrapers in four Missouri USA streams. Buendia et al. [54] who assessed the ecological effects of fine sediment in Isábena River, NE Iberian Peninsula Spain also found filterers to be the most sensitive macroinvertebrate feeding groups.
DO plays a critical role in the distribution of macroinvertebrates, particularly with regards to those species that have high sensitivities to DO depletion. Taxa sensitive to DO depletion were associated with the control sites, whereas those tolerant of DO depletion were associated with the sediment-impacted sites, providing support for the importance of DO in structuring stream communities [55]. Sediment delivery from catchment areas that are rich in organic materials is likely to stimulate microbial activities that can cause DO depletion. Moreover, increased sediment loads can also impact on the vertical distribution of oxygen, thereby influencing the depth to which organisms may burrow [7]. Thus, taxa (e.g., the EPTs) that have high sensitivity to DO, can be affected severely, and tolerant taxa of Chironomidae and Oligochaeta are favoured, and thus, the distribution pattern of sensitivity/tolerance to dissolved oxygen depletion observed in this study. Further, infiltration and deposition of fine sediment into the riverbed has been reported to modify macroinvertebrate community structure and functioning [53]. Taxa with low dissolved oxygen requirements frequently dominate substrates characterised by a high proportion of fine sediment (see [50], with absence of taxa vulnerable to fine sediment through damage of gills [47,56].
Macroinvertebrate filter-feeding structures are usually prone to clogging, particularly when levels of suspended sediments are elevated [7]. Collector-filterers that filter suspended FPOM from water column have been considered the most intolerant macroinvertebrate functional feeding group, [56,57]. Although fine sediments can lead to a reduction in macroinvertebrates that feed on FPOM [50,53], they can also serve as an important source of organic food particles [58]. In the present study, macroinvertebrate preferring FPOM proved tolerant of suspended sediments; thus, it is likely that suspended sediments increased the food availability for macroinvertebrates feeding on FPOM, thereby increasing their occurrence in the highly sedimented sites [58]. However, traits are unlikely to respond to environmental stress in isolation, but rather a combination of traits will determine the response of an individual species to a stressor [59]. Since collector-filters that feed on FPOM were earlier identified as sensitive indicators of suspended sediments in this study, it is possible that the observed response of FPOM was mediated by the interactive and correlative effects of other traits [59,60]. The mediating effects of traits on other traits to stressors have been observed in other studies, e.g., [61][62][63]. Nevertheless, previous studies have found macroinvertebrates feeding on FPOM to increase along suspended fine sediment gradient [64,65].
Concerning the skaters, their tolerance to suspended sediments was expected as they are active surface swimmers that live mainly on the water surface and thus can escape the effects of suspended sediments in the water [24]. Most actively swimming taxa are able to escape from danger and seek refuge [51]. Thus, the significant association of skaters with the highly sedimented sites suggests that skaters can actively move out of highly sedimented areas and return when conditions are normalised, and thus their tolerance in this study. The tolerance of actively mobile taxa such skaters observed in the Tsitsa River have been demonstrated in other studies investigating the effects of fine sediments on macroinvertebrate traits [21,28]. For example, Buendia et al. [54] who investigated the ecological effects of sediments on macroinvertebrates structure and function in the Isábena River, Spain observed active swimmers to be the most tolerant locomotion trait, increasing with increased sedimentation. The study attributed the positive correlation of swimmers with fine sediment to the ability of these actively swimming animals to move out of the most impaired areas. Overall, the results were in line with previous studies of sedimentation effects in sediment-impacted river catchments, with macroinvertebrates responding differentially to sediment stress. Thus, the identification of indicator traits of fine sediments is an important step towards the development of trait-based tool specific for monitoring sediment effects for Afrotropical river systems.

Conclusions
The results of this study showed that macroinvertebrates traits seem to be affected differentially by suspended fine sediments. Traits such as collector-filters, aquatic respiration, predators, a preference for very fast-flowing waters, a preference for low DO, skating and a preference for FPOM were associated with the highly sedimented sites. These sites were mainly influenced by increasing turbidity, TSS and sediments grain sizes such as fine sand and very fine silt. Traits such as a preference for low DO, skating and a preference for FPOM were regarded as tolerant traits, which can potentially increase resilience on taxa possessing them. On the other hand, traits such as CPOM, collector-filterers a high sensitivity to DO were deemed sensitive traits, and taxa possessing these traits are likely to be vulnerable to elevated suspended sediments. Overall, the present study provided information that could facilitate the development of trait-based biomonitoring tools for assessing the effects of elevated suspended sediments on riverine ecosystems, particularly in Africa, where trait-based studies remain scarce.