Urbanization decreased the multitrophic sh richness and ecosystem functioning in neotropical streams

Urbanization is regarded a major global threat to biodiversity and ecosystem functioning. Streams are among the most severely affected ecosystems due to worldwide urbanization. An increase in urbanization causes water quality deterioration and loss of habitat heterogeneity in streams. However, it is unclear how water quality deterioration and loss of habitat heterogeneity due to urbanization affect multitrophic diversity and performance of ecosystem functioning. We conducted 2,400 samplings in six streams across Uruguay to investigate how increases in urbanization (area and percentage of urbanization) affect the richness of three trophic groups of shes and the standing stock biomass of the streams. We investigated the direct and indirect effects, mediated by water quality deterioration and habitat heterogeneity of the urbanization on carnivore, omnivore, and detritivore sh richness and standing stock biomass of streams. The increase in urbanization (area and percentagem) in the streams signicantly decreased the richness of carnivores, omnivores, and detritivores shes. The increase in urbanization also strongly decreased habitat heterogeneity and increased water quality deterioration, which indirectly decreased the carnivore, omnivore, and detritivore sh richness. Urbanization also had strong negative effects on the standing stock biomass of the streams. Our study illustrates that urbanization promotes water quality deterioration and loss of habitat heterogeneity in streams, which indirectly causes loss of multitrophic sh richness and biomass production of streams.


Introduction
The human population is rapidly expanding and approximately 8 billion people currently inhabit the planet earth, which is projected to rise to 10 billion in the next few decades (United Nations 2018). Humanity is experiencing a shift to urban living; consequently, it is expected that until 2050, aproximately about 68 % of the global human population will live in urban areas (Grimm et al. 2008;United Nations 2018). Urbanization is one of the major global human-induced threats to biodiversity (McKinne 2006;Batáry et al. 2018), driving drastic impacts on the populations of birds (Batáry et al. 2018), microbes (Flies et al. 2020), plants (Aronson et al. 2014), arthropods (Van Nuland and Whitlow 2014), and shes (Steffy and Kilham 2006). Urbanization also impairs ecosystem functioning, such as biomass production, and pollination (Peng et al. 2020;Rivkinet al. 2020).
Streams are one of the most severely impacted ecosystems by urbanization due to (Paul and Meyer 2001). Increasing urbanization has a negative impact on stream quality in terms of ow regime and channel morphology alteration (Sung and Li 2010). Urbanization induces water quality deterioration by increasing nutrient levels and decreasing dissolved oxygen availability in the streams (Walsh et al. 2005). Excessive increases in nutrient levels, such as those of phosphorus and nitrogen, accelerate eutrophication, leading to plant overgrowth and harmful algal blooms, which decrease the water's oxygen levels and cause sh death (Dodds 2006;Wurtsbaugh et al. 2019). These impacts may be accompanied by other changes such as increasing stream ow and suspended solids (Paul and Meyer 2001;Walsh et al. 2005). In addition, urbanization is associated with homogeneous streams (Bernhardt and Palmer 2007) because urban streams have low plant diversity, which is vital to increasesing habitat heterogeneity (Ferreiro et al. 2011;Ferreiro et al. 2013), and its substrates are dominated by ne particles (such as sand and silt), which provide few refugia and habitat availability (Kukuła and Bylak 2020). Urban streams are also shallower and have a narrower width than natural streams due to erosion, which is elevated in its channel (Walsh et al. 2005).
So far, we know relatively little about how the loss of water quality and habitat heterogeneity due to urbanization affects different trophic groups of sh, mainly in hyperdiverse neotropical streams where studies are scarce. However, we could expect that whether urbanization decreases habitat heterogeneity and water quality, this should have a negative impact on sh multitrophic richness. This is because increasing habitat heterogeneity increases the richness of multiple sh trophic groups (Zeni and Casatti 2014), which occur through a variety of pathways: (i) increasing niche dimensionality (i.e., the number of available ecological niches) (Stein and Kreft 2015), stabilizing the food chain, allowing different trophic groups to co-exist (Haddad et al. 2011);(ii) increasing the number of shelters and refugia, favoring lower trophic groups (McCann et al. 2005); and (iii) increasing primary productivity and predator abundance (Haddad et al. 2011), increasing the amount of energy transfer and size of the food-web (i.e., allowing more trophic groups in the system).
Likewise, streams with good water quality often have a higher sh multitrophic richness than polluted streams (de Carvalho et al. 2020), especially in highest trophic groups (e.g., carnivores), which are more sensitive to environmental degradation (Estes et al. 2011). For instance, in eutrophic urban streams, harmful algae, heavy metals, and pesticides in sediments are ingested by lowest trophic groups, resulting in the accumulation of a diversity of pollutants in their bodies, which can be passed along the food web until these reach the apex predators (Grupta 2018). In apex carnivorous, the concentration of some pollutants is ampli ed by a million times or more (Schartup et al. 2019).
Finally, if urbanization decreases sh multitrophic richness, this could impair ecosystem functioning (e.g., decreasing sh standing stock biomass) because a high richness of multiple trophic groups is fundamental to maintaining ecosystem functioning (Soliveres et al. 2016). Theoretical and experimental studies have illustrated that high multitrophic richness sustains higher biomass production through two biological mechanisms-complementarity and selection (Loreau and Hector, 2001;Maureaud et al. 2019). The complementarity mechanism occurs when more trophic levels lead to higher biomass production through positive interactions, and the selection mechanism occurs when a dominant trophic level leads to higher biomass production (Loreau and Hector 2001). Moreover, urbanization could cause greater negative effects on biomass production if high trophic sh levels are lost, as they are the ones that most contribute to biomass production (Maureaud et al. 2019).
In this study, we assessed the multitrophic richness of sh (represented by three different trophic groups: carnivores, omnivores, and detritivores) at 12 sites of six neotropical streams during the four seasons of the year. These sites are subject to different degrees of urbanization pressure upstream of the sampling sites (represented by the area size [km²] and percentage of the drainage basin of each sampling point that is occupied by urban areas). We measure the habitat heterogeneity, water quality, and standing stock biomass in each stream. We aimed to investigate how urbanization directly and indirectly (through habitat heterogeneity and water quality) affect the richness of three trophic groups of sh and its consequence on standing stock biomass of the streams. We predicted that (i) as urbanization pressure (de ned by area and percentage) in streams increase, there is a decrease in the richness of the three trophic groups of sh. This occurs because (ii) urbanization decreases habitat heterogeneity and increases water quality deterioration. Finally, (iii) the loss of trophic groups of sh due urbanization should have negative effects on the standing stock biomass of the streams.

Study Site
Uruguay is located in the biogeographical region of Pampas, and the Koeppen climate classi cation of the basin corresponds to a "Cfa" humid subtropical climate, having humid summers (mean temperature above 22°C) and mild to cool winters (mean temperature above 0°C) (Kottek et al. 2006). The south of Uruguay is the area of the country with the most deteriorated watercourses (Goyenola et al. 2015;Benejam et al. 2016), particularly, the Colorado stream basin (166 km 2 ). The principal land use in this bain is characterized by intensive agriculture (i.e., winery, deciduous fruits, and vegetables) along with mixed with important urban and industrial areas. This basin has a clear impact when compared to other basins in the country; in this sense, effects in sh communities have already been detected (Benejam et al. 2016;Vidal et al. 2018). However, the speci c effect of the urbanization in streams on this agricultural basin has not been studied.
This basin is composed of six streams, from south to north, and is named "Cañada del Juncal" (CJ), "Cañada del Dragón" (CD), "Arroyo Las Piedras" (ALP), "Cañada de las Conchillas" (CCH), "Arroyo Colorado" (AC), and "Cañada del Colorado" (CC). The studied streams suffer from different urbanization pressures, such as cities and industries (Benejam et al. 2016;Vidal et al. 2018). Moreover, the urbanization impacts on these streams may be in uenced by the area and/or percentage of the stream basin occupied by urban regions. Here, we estimated the area (de ned by Km²) and percentage (%) of urbanization on each stream; speci cally, area of urbanization is de ned as area of the stream basin that is occupied by urban regions, and percentage of urbanization is the percentage of the stream basin that is occupied by urban regions. Importantly, the percentage of urbanization takes into account the size of the stream basin, whereas the area does not. We selected 12 sampling sites subject to different areas and percentage of urbanization in their drainage basins ( Fig. 1). In addition, these streams present differents limnological variables, which are directly related to urban pressure. For instance, streams with larger area and percentage of urbanization have higher values of nutrients (total phosphorus and nitrate), total suspended solids, organic matter, and conductiviy and lower values of dissolved oxygen than less urbanized streams (Table S1).

Fish sampling
Samplings were carried out at 12 sites in one annual cycle (2004)(2005) during the four climatic seasons. At each site, all sampling of water quality, habitat chacteristics and shes were performed in a reach of 50 m. Fish were sampled using electro shing (Type FEG 1000) by applying 50 electric pulses in each stream site. The sampling effort involved 50 samples realized in each climate season (fall, spring, winter, and summer) at each of the 12 sites (totaling 2,400-point sampling). Collected sh were euthanized with an overdose of 2-phenoxyethanol solution (1 mL.l − 1 ) (Teixeira de Mello 2020), and then xed in formaldehyde (10 %). In the laboratory, all sh samples were identi ed taxonomically (Table S2), counted, and weighted (0.01 g). Fish species richness, trophic position, and biomass (g.m − 2 ) in each site was determined from our own feeding trials and from iterature (Teixeira de Mello et al. 2012;Teixeira de Mello et al. 2016). We found three different trophic groups of sh (omnivores, carnivores, and detritivores) in the studied sites (Table S2).

Water quality
In the same 2,400 electrical points, we also measured six environmental variables, which represent indicators of water quality: dissolved oxygen (DO, mg.L), suspended organic matter (SOM, ug.L), total suspended solids (TSS, µg.L ), total phosphorus (PT, µg.L), and nitrate (µg.L) contents and conductivity (Table S1). We investigated the distribution of each variable and carried out a principal component analysis (PCA) with the variables previously standardized and centered. The rst PC axis represented the majority of variation present in the original six variables (66.6 %), and this axis was negatively correlated with DO (r = -0.38) and positively correlated with the SOM (r = 0.42), TSS (r = 0.41), PT (r = 0.43), nitrate (r = 0.42), and conductivity (r = 0.36; Table  S3). The rst PC-axis could represent a clear proxy of water quality deterioration because as nutrients, conductivity, and suspended material increase, oxygen availability decreases. The PC1 axis was termed water quality deterioration.

Habitat heterogeneity
We also measured environmental information for the analysis of habitat heterogeneity. To address heterogeneity in streams as comprehensively as possible, we adopted the classi cation system of Stein and Kreft (2015), which divides the facets of heterogeneity into ve subject areas, of which three were applied in our study: vegetation, microscale topography, and soil diversity.

Quantifying habitat heterogeneity
There are several different measures to represent each of the three facets (Stein and Kreft 2015). We selected measures that better capture the major heterogeneity of the whole streams, which are relevant for sh trophic groups. For instance, for the vegetation facet, we used the percentage of plant presence. Plant presence increases habitat heterogeneity by providing habitats that allow species coexistence and increase sh biodiversity (Moi et al. 2020). For the soil facet, we measured the diversity of the substrate (sediment) types. We classi ed the substrate type based on the grain size of the substrates found, which were clay, sand, boulder, and rocks. We then calculated the Shannon index of these substrates using the 'vegan' package (Oksanen et al., 2013) in the R software (R Core Team 2018). The Shannon index has been used to assess the diversity of substrate types in heterogeneity studies (Stein and Kreft 2015). The diversity of substrate types is related to more pristine environments and high trophic sh diversity (Peressin et al. 2020). For the microscale topography facet, we used the coe cient of variation (CV) of the depth of the streams. The depth is often reduced in urban streams, and this occurs due to erosion, which decreases habitat heterogeneity (Walsh et al. 2005).
We then summarized these three facets of heterogeneity into a multivariate heterogeneity by carrying out a PCA with these three measured previously standardized (i.e., percentage of plant presence, diversity of substrate types, and CV of depth). The rst PC axis accounted for the majority of the variation in data (71 %), and was positively correlated with the diversity of substrate types (r = 0.59), percentage of plant presence (r = 0.59), and CV of depth (r = 0.53; Table S4). The PC1 axis was named multivariate habitat heterogeneity.

Data analysis
We evaluated the relationship of the area and percentage of urbanization with the richness of the three trophic groups of sh (carnivores, omnivores, and detritivores) using generalized additive mixed-effect models (GAMMs). We used GAMMs because the scatterplots of the area and percentage of urbanization versus response variables showed non-linear patterns. The normality and homogeneity of variance of the residuals were analyzed using histograms and by plotting the residuals versus tted values. The GAMMs was tted using the gamm4 package in the R software (Wood and Scheipl 2017), considering seasons of the year in each of the 12 sites in the six streams as a random effect to account for temporal pseudo-replication in the data.
We also applied structural equation modelings (piecewiseSEM; Lefcheck 2016) to test how urbanization (area and percentage) affects the richness of multiple trophic groups mediated by effects on multivariate habitat heterogeneity and water quality deterioration and how this impacts the standing stock biomass of the streams. We also tted piecewiseSEM models with all the individual heterogeneity facets (percentage of plant presence, diversity of substrate types, and CV of depth) rather than multivariate habitat heterogeneity to test how urbanization affects the three trophic groups of sh through effects on each facet and their consequences to standing stock biomass. We checked the multicollinearity in each component model by calculating the variance in ation factor (VIF) for each predictor. A VIF > 3 evidence possible collinearity, but was not present in our data. The piecewiseSEM models were created using linear mixed-effect models (LME; Pinheiro et al. 2013) considering seasons of the year in each site as a random effect to account for temporal pseudo-replication. The biomass production was log-transformed to achieve normality in the residuals. The standardized coe cient was shown for each path, and the indirect effects were estimated by coe cient multiplication. The path's signi cance was obtained by maximum likelihood, and model t was evaluated using Shipley's test of dseparation through Fisher's C statistic where P > 0.05 indicates a adequate model.

Results
During the four climate seasons at the 12 sites, we caught 7,910 sh individuals, and 28 species belonging to carnivores, omnivores, and detritivores trophic groups. We found that the richness of the three trophic groups of sh responded signi cantly to the increase in the area and percentage of urbanization on streams (Table 1). Speci cally, the richness of carnivores (Fig. 2a, b), omnivores (Fig. 2c, d) and detritivores (Fig. 2e, f) signi cantly decreased as the area and percentage of urbanization increased on streams (P < 0.05, Table 1). In addition, the richness of carnivores sh had a steeper reduction as urbanization pressure increased on streams (R² = 0.325 for area and R² = 0.302 for percentage of urbanization; Table 1). The positive (+) or negative estimate effects of the area and percentage of urbanization on response variables is shown next to R² adj . Edf = estimated degrees of freedom.
The piecewiseSEM revealed strong indirect effects of urbanization on the richness of the three trophic sh groups mediated by effects on habitat heterogeneity and water quality deterioration, and such relationships explained 62 % and 65 % of the variation in the standing stock biomass of the streams (Fig. 3). The urbanization area indirectly decreased the carnivore sh richness by increasing water quality deterioration (r = -149; Fig. 3a). Conversely, the multivariate habitat heterogeneity indirectly increased the carnivore sh richness because of decreased water quality deterioration (r = 0.180). Urbanization pressure (area and percentage) also decreased the omnivore sh richness by decreasing the multivariate habitat heterogeneity (area: r = -0.387; percentage: r = -0.346; Fig. 3). Also, there was a positive relationship between the omnivore and carnivore sh richness (β = 0.232; Fig. 3). We also found that the urbanization area and percentage directly decreased standing stock biomass, whereas carnivore sh richness increased the standing stock biomass of the streams (Fig. 3). The water quality deterioration indirectly decreased the sh standing stock biomass by decreasing the carnivore sh richness (r = -100).
Finally, we found similar results when all heterogeneity facets were used instead of multivariate habitat heterogeneity (Fig. S1). The increasing urbanization (area and percentage) decreased all three heterogeneity facets (percentage of plant presence, soil diversity, and CV of depth) and increased water quality deterioration ( Fig. S1). Urbanization (area and percentage) also indirectly decreased the omnivore and detritivore sh richness by decreasing the percentage of plant presence and substrate diversity, respectively (Fig. S1).
Furthermore, the percentage of plant presence decreased the water quality deterioration; thus, indirectly increasing carnivore and detritivore sh richness. Similarly, the percentage of plant presence increased the omnivore sh richness, and soil diversity increased the detritivore sh richness (Fig. S1).

Discussion
The results reveals that the increase of the urbanization in streams is highly positively correlated to a decrease in the richness of carnivore, omnivore, and detritivore shes, and loss of standing stock biomass. The results suggest that the reduction in the richness of these three trophic sh groups may be explained by the fact that the increased urbanization decreased the water quality (by increasing eutrophication and decreasing oxygen availability) and habitat heterogeneity (by decreasing plant presence, soil diversity, and depth) of the streams. This have negative indirect effects on richness of the three trophic groups and ultimately on standing stock biomass of the streams. These ndings agree with other aquatic and terrestrial studies (Walsh et al. 2005;Merckx et al. 2018;Melliger et al. 2018), thus con rming the general expectation that urbanization cause the loss of trophic diversity and impairs ecosystem functioning due to changes in the quality and heterogeneity of the environments.
The reduction of the habitat heterogeneity with the increase of urbanization in streams may cause negative effects on the richness of multiple trophic groups of sh through severel processes such as (i) by decreasing the length of environmental gradients and the number of habitat types, making it di cult for some trophic groups to remain in the system (Stein and Kreft 2015;Ortega et al. 2018); and (ii) by impairing trophic coexistence through decreasing the number of shelters and refugia for the lowest trophic groups, allowing the highest trophic groups to exclude it. Indeed, we found that habitat heterogeneity has a positive effect on the richness of small omnivore shes, which indirectly contributes to increases the richness of carnivore shes. This suggests that in less urbanized streams, habitat heterogeneity appears to mediate trophic coexistence and promote multitrophic richness of shes (Burdon et al. 2019, Penone et al. 2019. By contrast, as urbanization pressure in the streams increase, habitat heterogeneity decrease, which likely impairs the coexistence between trophic sh groups. The results also showed a high deterioration of water quality in highly urbanized streams, which strongly decreased the carnivore sh richness. This likely re ects the fact that the highest trophic groups such as carnivores, are more sensitive to environmental deterioration than the lowest trophic groups (Estes et al. 2011).
In the studied streams, the abundance of omnivores shes such as Cnesterodon decemmaculatus increased with urbanization, whereas overall community richness, especially the of the carnivore shes, decreased (Benejam et al. 2016). This occurs because omnivore shes are more tolerant to water quality deterioration, for instance eutrophication and oxygen depletion (Benejam et al. 2016;Vidal et al. 2019). Moreover, as carnivore sh richness decreases, tolerant omnivores are favored due to a reduction in predation pressure. Importantly, the ndings show that habitat heterogeneity increases water quality, which occurs mainly due to plant presence. Plants stabilize aquatic ecosystems by removing nutrients from the water column and controlling sediment resuspension (Levi et al. 2015). This increases water clarity, thus making aquatic ecosystems healthier (Moi et al. 2020;Silva et al. 2021), consequently increasing the richness of multiple trophic groups of shes, especially of the apex predators (Moi et al. 2020).
We found that all individual heterogeneity facets decreased with increasing urbanization in the streams. In particular, the steeper reduction of the plant presence in more urbanized streams was most likely driven by the high surface runoff (water ow) of these streams, which prevents submerged plants from settling (Paul and Meyer 2001;Pickett et al. 2011). In addition, urbanization increases nutrient levels and suspended organic matter content in the water (Grimm et al. 2008;Pickett et al. 2011), favoring the dominance of small oating plants or lamentous algae (O'Hare et al. 2018;Ardón et al. 2020). The dominance of these two primary producers in turn cause dark and anoxic conditions under water, which provides few opportunities for the submerged plant and animal life (Janse and Van Puijenbroek 1998). Conversely, in less urbanized streams, we observed oligo/or mesotrophic conditions, which favors a high plant diversity (especially of submerged type) and multitrophic richness of shes. We found that plant presence directly increased the richness of small omnivore sh and indirectly increased the richness of carnivorous sh. This occurs because plants play a structural role in offering refuge to different consumer groups, typically increasing richness of multiples trophic groups of sh in streams (Argentina et al. 2010). These ndings provide a strong support for the prominent role of plant presence in shaping interactions between trophic groups in neotropical streams as well as in terrestrial ecosystems (Scherber et al. 2010). Additionally, our ndings are somewhat unique by showing that urbanization causes loss of plants in the streams, which indirectly may have negative impacts on the trophic web of these ecosystems.
The low substrate diversity and depth of urbanized streams are based on the idea that urbanization increases the sediment supply and bankfull discharge into the stream channel, leading to erosion (Paul and Meyer 2001).
The erosion decreases the channel depth as its width increases (Walsh et al. 2005). Moreover, urbanization also changes the substrate texture, which is dominated by ne particles (silt and sand), whereas gravel and rocks decrease in urbanized streams (Papangelakis et al. 2019). The low substrate diversity and depth have negative effects on richness of multiple trophic groups of sh because streams become quite shallow and have homogeneous ne substrates (Walsh et al. 2005; Jesús-Crespo and Ramírez 2011). In turn, homogeneous substrates provide few refugia for the lowest trophic groups, allowing them to be easily consumed by the predators (Kukuła and Bylak 2020). In contrast, larger substrate particles exhibit greater stability, accumulate more basal resources, and provide effective refuge to the smallest trophic groups (Kukuła and Bylak 2020), and as our results showed also increases the richness of detritivorous shes. In addition, with decreasing substrate diversity and depth of streams, the diversity of benthic macroinvertebrates also decreases drastically (Paul andMeye 2001, Burdon et al. 2020), which may indirectly have negative effects on omnivorous shes, since macroinvertebrates are a vital food items for omnivorous shes.
The SEM revealed that urbanization has a strong direct negative effect on sh standing stock biomass in the streams. Moreover, the negative effects of urbanization on habitat heterogeneity and water quality were propagated across richness of the three trophic groups of sh, which indirectly decreased the standing stock biomass of the streams. These results point to the fact that urbanization, directly and indirectly, may impair the functioning of streams by decreasing biomass production, which is also a important ecosystem service (La Notte et al. 2017). Our results indicate that this occurred because of increased urbanization in the streams, the richness carnivorous, omnivorous, and detritivorous shes decreased, and particularly of carnivores, which was the trophic group that most positively affected the standing stock biomass of the streams. Carnivorous shes are key components in aquatic ecosystems and may affect biomass production through direct and indirect pathways by structuring the food web and controlling interactions between the lowest trophic groups and consequently their biomass (Rock et al. 2016). Moreover, most carnivorous shes in our study are generalists (i.e., they are able to feed on benthos and pelagic sh), which is similar to those found in the Europe (Maureaud et al. 2019). This suggests that carnivorous generalist shes may strongly in uence standing stock biomass of streams. Thus, increasing urbanization might lead to the loss of these apex predators, indirectly impairing the ecosystem functioning at a global scale. Our results add to the growing body of evidence that trophic downgrading on planet earth, mainly by massive declines in apex predators (here, originated by urbanization) have strong consequences on the functioning of ecosystems (Estes et al. 2011, Ripple et al. 2014).
Both area and percentagem of urbanization had negative effects on the richness of carnivorous, omnivorous and detritivorous shes and stock biomass of the streams, suggesting a consistent negative impact of urbanization on biodiversity and the functioning of these environments. mportantly, the higher values of area and percentagem of urbanization in the studied streams were 10 km² and 40 %, respectively, indicating a moderated urbanization pressure in these streams (Walsh et al. 2005), which demonstrate that these streams are sensitive to urbanization. Despite this, cities have advanced over the natural streams in the studied region, causing its deterioration over recent years (Alvareda et al. 2020). Our results indicate that, as urbanization pressure on natural streams increases, this should cause further trophic deterioration in these environments, with drastic consequences on its functioning. It is also important to highlight that the studied streams in Uruguay are highly diverse systems (Teixeira-de Mello et al. 2012), and its deterioration due to urbanization will cause great biodiversity loss, which also may implies a loss of biomass production (i.e., ecosystem service).

Conclusion
The results of the present study demonstrate that urbanization directly decreases the richness of carnivorous, omnivorous and detritivorous shes and cause loss of standing stock biomass of neotropical streams. The negative effects of urbanization in the richness of trophic groups of sh occurs particularly due to loss of habitat heterogeneity and water quality in more urbanized streams. The obtained data also show that carnivorous shes sh are the trophic group most impacted by urbanization. Despite this, this trophic group has a key role in increasing the standing stock biomass of the streams. Our study has great ecological relevance as urban areas currently cover approximately 2% of the Earth's land surface (Grimm et al. 2008), and according to the United Nations (2018), urbanization continues to rapidly expand, with about 68 % of the global human population living in urban areas by 2050. We highlights that more studies are necessary to predict the consequences of urbanization on natural ecosystems, and special focus should be given to highly diverse ecosystems (such as streams studied here) as these ecosystems are facing an accelerated urbanization process.

Declarations
Author contribution statement DAM and FTM conceived the manuscript idea. DAM performed the statistical analyses. DAM and FTM wrote and revised the manuscript.