Trichoptera, Odonata, and Diptera as Efficient Surrogate Taxa for Monitoring Macroinvertebrate Assemblages in Urban Amazonian Streams

2.1 Study Area

Manaus has experienced rapid urban expansion since the 1970s, driven by industrial development, increasing the population from ~300,000 (1970) to 1.8 million (2010), with ~99% concentrated in urban areas. This led to uncontrolled land occupation along riverbanks, deforestation, siltation, and domestic sewage discharge (Bentes 2005; Couceiro et al. 2007; Martins et al. 2017).

We sampled 21 streams in the municipality of Manaus, Amazonas, Brazil (Figure 1) during both the dry season (November/December 2004) and the rainy season (May 2005), representing intra-annual variation. In addition, we collected data in September 2010 (dry season) to assess interannual variation in relation to the 2004/2005 data. All streams were sampled in each of the three seasons, with each stream being sampled only once per season. Streams were selected along a gradient of impact, with 13 streams affected by urbanization and/or deforestation and eight in forest remnants.

The climate of the Manaus region is classified as humid equatorial (Af according to the Köppen-Geiger classification), characterized by two distinct seasons: a rainy season from December to May and a dry season from June to November. The average annual temperature is 26.7°C, with monthly humidity ranging from 79% to 88%. The average annual rainfall is about 2300 mm.

2.2 Invertebrate Sampling

In each stream, we established a 60 m longitudinal transect. Along this transect, we collected samples at evenly spaced 15 m intervals (0, 15, 30, 45, and 60 m), using a “D” shaped aquatic entomological net (570 cm², 1 mm² mesh). Sampling was conducted in two microhabitats (sediment/edge litter and sediment/edge litter in the central stream bed), resulting in five samples per stream. Samples were preserved in plastic bags containing 90% ethanol and transported to the laboratory for processing. Upon arrival, samples were washed under running water through a metal sieve (125 µm mesh) and subsequently identified to the lowest taxonomic level possible using relevant literature (Hamada et al. 2019).

2.3 Abiotic Variables

We assessed stream depth and width using a ruler with millimeter markings. Water velocity was estimated by timing the passage of a plastic float over one meter. Electrical conductivity was measured with a portable meter (WTW, model LF90), and pH was recorded using a portable potentiometer (WTW, model pH90). Temperature and dissolved oxygen near the stream bottom were measured with a portable oximeter (YSI, model 55). Deforestation rates for each stream were determined using a 2003 Landsat satellite image, which was classified to distinguish forested and deforested areas within 100 m buffers around each stream's GPS-acquired geographic coordinates (Garmin).

2.4 Data Analysis

The congruence of intra- and interannual abiotic variables, aquatic invertebrates, and the relationship between biotic and abiotic variables was assessed using Procrustes analysis. This analysis compares two configurations obtained by ordination (Jackson 1995; Gower 2001). We used Procrustes analysis to examine the congruence between two datasets: (1) the abundance of invertebrate assemblages at the lowest possible taxonomic level, and (2) the abundance and occurrence (presence/absence data) of taxa at the family and genus levels. Specifically, we focused on Diptera (family level only), Odonata, and Trichoptera. For Procrustes analysis, we used the axis scores obtained from Principal Coordinates Analysis (PcoA; Legendre and Legendre 2012) with log-transformed (log (x + 1)) data to mitigate the influence of outliers. We applied Bray-Curtis dissimilarity for taxa abundance, Sørensen dissimilarity for taxa occurrence, and Euclidean dissimilarity for abiotic variables to quantify dissimilarities. The goodness of fit between matrices was assessed using the Procrustean permutation test (PROTEST). In this test, lower statistic values indicate higher concordance between groups (Peres-Neto and Jackson 2001). To determine the significance of these relationships in PROTEST, we performed 10,000 permutations. We performed all analyses in R (R Development Core Team 2024) using the vegan package (Oksanen et al. 2016).

3.1 Abiotic Data

High congruence of abiotic data was observed within intra-annual seasons (81%; Figure 2). In contrast, the congruence between the abiotic data of 2010 and the dry (65%) and rainy (62%) seasons of 2004/2005 was moderate (Figure 2). These results indicate that the interannual variation in abiotic data in Amazonian urban streams exceeds their intra-annual variability.

3.2 Biotic Data

In 2004/2005, we collected 13,910 specimens distributed across 77 taxa; in 2010, we collected 15,277 specimens distributed across 82 taxa (see Supporting Material). Diptera, Trichoptera, and Odonata accounted for 77% of the abundance and 61% of the observed richness in 2004/2005. By 2010, these three orders accounted for 53% of the abundance and 65% of the observed richness.

We found high intra-annual congruence in total fauna abundance (82%) and invertebrate occurrence (84%). The congruence of invertebrates between the 2010 data set and different seasons in the dry and rainy seasons of 2004/2005 was 77% and 76%, respectively, regardless of whether abundance or occurrence data were considered (Figure 2). In contrast to the abiotic data, these results suggest that there is relatively low variation in the invertebrate populations within and between the years in the urban streams studied.

3.3 Congruence: Invertebrates X Abiotic

The invertebrate occurrence and abundance (all taxa) datasets showed moderate correlation with the abiotic data across all three collection seasons (56%–70%; Figure 3). Most subsets showed similar levels of congruence with the abiotic data compared to the invertebrate data. The highest level of congruence with the abiotic data was observed in the abundance of the Odonata genera in 2010 (78%), while the lowest congruence was found in the abundance of the Diptera family in 2010 (42%; Figure 3). The macroinvertebrate abundance subset showed the least variation in congruence among the three collection seasons (2%). In contrast, Odonata genera (abundance = 29%; occurrence = 27%) showed the greatest variation in congruence between abiotic data and fauna across the three collection seasons. In general, concordance values between biotic and abiotic data were similar when considering abundance and occurrence data. These results suggest that overall, invertebrate family and occurrence data have a similar relationship with abiotic data compared to lower taxonomic levels. Additionally, Trichoptera, Odonata, and Diptera may have stronger associations with abiotic data than the full invertebrate assemblages from Amazonian urban streams.

3.4 Congruence: Taxonomic Levels

In all seasons, the congruence between low-level invertebrates (genera) and families was greater than 93% for both abundance and occurrence data (Figure 4). Abundance and occurrence data for Trichoptera, Odonata, and Diptera genera and families were highly correlated (71%–96%) with the low-level invertebrate assemblage. The congruence between invertebrate abundance and occurrence data was greater than 94%. Occurrence data resulted in up to 7% loss of information compared to abundance data (Figure 4). Congruence was generally higher than 60% across all comparisons between years and seasons (Figure 5).

4.1 Abiotic Data: Intra- and Interannual Variation

We found high intra-annual congruence in abiotic variables in urban streams of Manaus. Seasonal abiotic variations include fluctuating water temperatures during the hot and rainy season and increased total organic carbon during the dry and cold season (Nava et al. 2015). Our study is consistent with previous findings in the region, which did not find significant differences in abiotic variables over the course of a year. Continuous sewage discharge into the streams studied may contribute to these small, transient changes in abiotic conditions during increased precipitation events (Santos et al. 2006).

We found moderate interannual congruence in abiotic variables in the urban streams of Manaus. Abiotic variables such as nitrogen, phosphorus, and oxygen show year-to-year variability influenced by precipitation patterns and land use changes (Atique and An 2020). Therefore, the observed low congruence between sampled years in urban streams of Manaus was expected and is consistent with previous studies in the region reporting increased environmental degradation due to deforestation and domestic sewage from a growing human population (Martins et al. 2017; Ferreira et al. 2021; Couto et al. 2024). These authors observed changes over time, including decreased oxygen levels and increased values of total nitrogen, total phosphorus, deforestation rates, total impervious area, water temperature, pH, and electrical conductivity.

4.2 Biotic Data: Intra- and Interannual Variation

Despite high variability in environmental variables between study years, invertebrate assemblages showed high intra- and interannual congruence. In general, these assemblages are affected by environmental changes (Brasil et al. 2020). However, previous studies in urban streams of Manaus did not record differences in invertebrate richness, abundance, or composition over time (Martins et al. 2017; Couceiro et al. 2021). Therefore, the observed high congruence may be related to the dominance of Chironomidae, Oligochaeta, and Psychodidae across collection years (median = 82% ± 26%) and within the same year (median = 80% ± 27%) in the urban streams of Manaus. The low intra- and interannual variation in invertebrate assemblages reinforces their utility as bioindicators due to their relative stability over time and low variability due to climatic factors (Holt and Miller 2011).

4.3 Biotic Data: Surrogates

We found that family and occurrence data can be a good proxy for genus-identified data, with minimal information loss of less than 10% across all collection seasons. Similar results have been reported for Amazonian stream invertebrates along an agricultural gradient (Brito et al. 2018; Martins et al. 2022) and Pantanal streams (Valente-Neto et al. 2018). This approach allows for efficient environmental assessment studies even in laboratories that lack expertise in some or all groups of aquatic invertebrates, thereby extending their applicability to a wider range of personnel. This inclusivity is crucial for conservation efforts in aquatic environments far from major research centers (Giehl et al. 2014; Brito et al. 2018). In addition, using occurrence data can reduce the time required for environmental assessments, as invertebrates often have high abundances (Martins et al. 2021). For instance, our study recorded an average of 1,253 individuals per stream in 2004/2005.

We found that Diptera, Odonata, and Trichoptera can be good proxies for low-level invertebrate identification data, with minimal information loss of less than 16%. Diptera, particularly the family Chironomidae, dominated all sampled streams due to their remarkable adaptability to different aquatic environments. Certain species within the genus Chironomus are highly tolerant to environmental impacts, while genera such as Stenochironomus are sensitive indicators of environmental conditions (Trivinho-Strixino 2011). Previous studies in the Amazon region have identified Odonata and Trichoptera as reliable indicators of aquatic invertebrate assemblages (Brito et al. 2018; Martins et al. 2022). Odonata exhibit extensive dispersal capabilities, and their two suborders, Anisoptera and Zygoptera, show different responses to environmental changes (Corbet 1999; Oliveira-Junior and Juen 2019). Trichoptera are typically associated with minimally disturbed environments and strongly correlate with riparian vegetation (de Faria et al. 2017). Although composite metrics incorporating multiple orders often better reflect disturbance gradients by integrating diverse environmental responses (Stanford 2012; Brito et al. 2018), focusing on a single order can be practical when specialists are not available for all groups, thereby increasing efficiency, reducing costs, and saving time in biomonitoring efforts (Valente-Neto et al. 2025).