Water quality shapes the community structure of zoobenthos in rivers but only has a minor effect on the fatty acid composition of zoobenthos and salmonids

Many river ecosystems in the boreal zone have faced remarkable changes due to intensive human activities, including land‐use changes in the catchments and channelization. Recently, restoration efforts have increased, hoping to restore a more natural hydromorphology. We studied the community structure of benthic macroinvertebrates (zoobenthos) in seven rivers in Eastern Finland, using samples that covered pre‐ and post‐restoration periods, to examine how environmental metrics relate to the zoobenthos community structure, and the fatty acid (FA) composition and content of Ephemeroptera, Plecoptera, and Trichoptera nymphs and larvae. We also analyzed the FA composition and content of land‐locked salmon (Salmo salar m. sebago) in three of the rivers and brown trout (Salmo trutta) in two of the rivers. Zoobenthos communities differed significantly among most of the rivers: 24% of the differences between the zoobenthos communities were driven by water quality parameters related to the loading of terrestrial organic matter (water color, pH, and iron concentration). Temporal changes in zoobenthos communities could not be fully attributed to restorations. The FA composition of zoobenthos was mostly explained by phylogenetic origin (47%). However, especially mayfly Heptagenia sulphurea (Ephemeroptera: Heptageniidae) had variable FA composition and content among the rivers suggesting an environmental quality indicator role for this species. FA composition and content of salmonids were mostly size‐dependent (24%), but river identity also influenced their FA composition (22%). Our results indicate that water quality affects the availability of essential FAs for consumers by altering the zoobenthos community structure and their FA composition and content.


| INTRODUCTION
Benthic macroinvertebrates (zoobenthos) are important food for many riverine fishes including salmonids.Many taxa, especially species from Ephemeroptera (mayflies), Plecoptera (stoneflies), and Trichoptera (caddisflies) (EPT-taxa), are sensitive to environmental quality, and are routinely used to evaluate the ecological state of European rivers according to EU water framework directive (WFD, 2000/60/EY).Zoobenthos groups vary systematically in their fatty acid (FA) composition (Makhutova et al., 2016), and feeding guilds exploit different resources (Cummins, 1973) with specific FA composition (Makhutova et al., 2011).In aquatic ecosystems, phytoplankton and benthic algae are the primary synthesizers of important omega-3 (ω3) polyunsaturated FA (PUFAs), of which especially the long-chain eicosapentaenoic acid (EPA,20:5ω3) and docosahexaenoic acid (DHA,22:6ω3) are physiologically important for aquatic consumers.Consumers mainly depend on algal production of these ω3 PUFAs because they cannot synthesize physiologically essential FAs (EFAs) arachidonic acid (ARA, 20:4ω6), EPA, and DHA de novo (Cook & McMaster, 2002).Since the content of EPA and DHA vary among the algal taxa and water quality changes can influence algal community composition, water quality changes can also affect the availability of these essential compounds higher up in the food web (Peltomaa et al., 2017;Taipale et al., 2016).Correspondingly, Guo et al. (2018) found that the FA composition of epilithic algae affects the FA composition of zoobenthos in riverine systems.Since fish need ω3 PUFAs to maintain normal physiological functions (Ahlgren et al., 2009;Glencross, 2009;Tocher, 2010), it is important to examine how environmental qualities affect the availability of these EFAs in zoobenthos and in higher trophic level consumers, such as fish.
Most rivers and streams in Finland have been heavily modified by human activities: channelization and dredging for timber floating (Jutila, 1992) and drainage of peatlands.These modifications have led to the loss of habitats for salmonids and many aquatic invertebrates.
Furthermore, the majority of Finnish forests are utilized for industrial forestry with significant impacts on the water quality, especially the concentration of dissolved organic carbon (DOC) and amount of suspended particulate organic material in the aquatic ecosystems have increased in recent decades (Aaltonen et al., 2021;Albrecht et al., 2023;Laudon et al., 2009;Lepistö et al., 2021;Nieminen et al., 2015;Rääpysjärvi et al., 2016).The so-called "browning" phenomenon, indicative of an increase in DOC and iron concentrations due to multiple mechanisms affects many waterbodies in the northern hemisphere (de Wit et al., 2016;Lepistö et al., 2021;Monteith et al., 2007).Most of the rivers and streams in North Karelia have a high humic (DOC) content (Rouvinen, 2010), which may have a major influence on the community structure of zoobenthos (Kesti et al., 2021;Robbins et al., 2020) and thus sets the main criteria for river types in the EU WFD classification system.
To overcome the biodiversity declines caused by the hydromorphological modifications of rivers, significant hydromorphological restoration efforts have been made during the past decades.While the restorations aim to improve overall biodiversity, they have often been driven by the need to increase recruitment habitats for salmonids, mostly brown trout (Salmo trutta) (Marttila et al., 2019).Overall, the results of restoration efforts have been monitored poorly and the documented results have been variable.While most of the studies have found positive responses of zoobenthos communities to restoration (Albertson et al., 2011;Kil & Bae, 2012;Louhi et al., 2016;Muotka et al., 2002;Pilotto et al., 2018;Suurkuukka et al., 2014;Verdonschot et al., 2016), some have reported little or no changes in the community composition (Smith et al., 2020;Tetu et al., 2016).It is also commonly reported that zoobenthos communities are heavily obscured by river restoration procedures and show lower biomass and diversity acutely after restorations (dos Reis Oliveira et al., 2019;Louhi et al., 2011;Molina-Moctezuma et al., 2021).
In this study, we focus on small to medium-sized boreal rivers in Eastern Finland and their native zoobenthos and fish fauna.Some of the rivers support resident populations of endangered brown trout and some are stocked with hatchery-reared juveniles of critically endangered landlocked Atlantic salmon (Salmo salar m. sebago).Virtually, all the waterways in this area have been modified by humans, and substantial restoration efforts have been conducted on several occasions during the past 50 years.In the years 2011-2015, several rivers were modified to create new breeding habitats for the endangered salmonids (Rouvinen, 2010).Zoobenthos reported as favorable food sources for salmonids include larvae of Diptera and caddisflies, and nymphs of stoneflies, mayflies, and dragonflies (Khrennikov et al., 2007;Regerand et al., 2002;Shustov et al., 2012).
Hydromorphological conditions in the river, eutrophication, and high humic content can affect the quality of zoobenthos assemblages via changes in the community structure, as well as via changes in the FA composition and content within taxa (Kesti et al., 2021;Strandberg et al., 2023;Taipale et al., 2016).Changes in the communities of these prey species or their FA composition and content could thus influence the overall availability of EFAs to salmonids (Strandberg et al., 2023).Specifically, we aimed to examine (1) which river characteristics drive zoobenthos community structure, (2) how past river restorations might affect zoobenthos community structure, and (3) if the FA composition and content, especially physiologically essential FAs ARA, EPA, and DHA of zoobenthos and salmonids differ among the rivers.

| Study rivers
We sampled zoobenthos from seven rivers with past restoration activities (Figure 1, Table 1, Table S1).We focused on rivers in North Karelia in Eastern Finland to limit the impact of regional differences.The selected rivers were restored at different periods: Kalliojoki, Koitajoki, Kuusoja, and Venejoki at the end of the 1990s, Ala-Koitajoki and Naarajoki during the year 2010, and Hanhijoki and Ulkkajoki in 2015 (Rouvinen, 2010).These rivers were restored mainly by increasing channel complexity by adding varying-sized stones and gravel suitable for spawning large salmonids, but two sites in River Ala-Koitajoki (Hiiskoski and Räväkkäkoski) were also restored by translocation of small stones with attached aquatic mosses (Hynninen & Vehanen, 2022;Rouvinen, 2010).We retrieved physical and chemical data of six of the sampled rivers from the Finnish Environment Institute (SYKE) Hertta-database (www.syke. fi/avoindata): Ala-Koitajoki, Hanhijoki, Kalliojoki, Kuusoja, Ulkkajoki, and Venejoki.The data had been collected as a part of monitoring for EU WFD.There was no physical-chemical data for Naarajoki.
The selected rivers have a relatively high water color (ranging 110-190 mg L À1 Pt, Table 1), which is typical for rivers in the study region (Rouvinen, 2010).

| Zoobenthos community analyses
Supplementary zoobenthos community data, as collected according to EU WFD standards (i.e., four pooled kick-net samples collected from shallow rapids per river), were retrieved from the Hertta-database in December 2020 for four of the rivers: Ala-Koitajoki, Hanhijoki, Kalliojoki, and Kuusoja.For the rest of the rivers (Naarajoki, Ulkkajoki, and Venejoki), we relied on the self-collected primary data (Figure 1, Table S1).The zoobenthos community data were sorted according to the years from restoration for each river (before [À] or after [+] restoration), with five-year intervals.The total time scale for the zoobenthos community data related to the time (years) from restoration was: In River Ala-Koitajoki À0 to 4 to +5 to 9, in River Hanhijoki À0 to 4 to +5 to 9, in River Kalliojoki +10 to 14 to +20 to 24, in River Koitajoki +10 to 14 to +20 to 24, in River Kuusoja +5 to 9 to +20 to 24, in River Ulkkajoki +0 to 4, and in River Venejoki +20 to 24, respectively.We calculated the relative abundance (percentage, %) of each taxon from the total number of individuals of all taxa present in the samples from the community.

| Zoobenthos sampling and preparation for FA analysis
Zoobenthos field sampling and preparation procedures were conducted following Kesti et al. (2021).Zoobenthos were collected with a standardized kick-net sampling method (SFS-EN 27828).Samples from River Ala-Koitajoki were collected in August 2018, and the rest of the samples were collected in August-October 2019, respectively (Figure 1, Table S1).Zoobenthos samples were filtered through a 0.5 mm mesh-sized sieve, transported to the laboratory, and first sorted according to class/family and stored in Eppendorf tubes at À80 C until further identification.For further identification, samples  were briefly thawed and rinsed with MilliQ water.We identified EPTtaxa to species level, when possible.After identification, the samples were stored at À80 C until further analyses.
We selected members of the EPT-taxa for the FA analyses (Table S2).Prior to the analysis of FAs, the samples were freeze-dried using Christ ALPHA 1-4 Ldplus (Martin Christ Gefriertrocknungsanlagen GmbH, Osterode, Germany) and further pulverized with mortar and pestle.
2.4 | Salmonid sampling and preparation for FA analysis Salmonids (S. salar and S. trutta) were electrofished from four rivers as a part of fish monitoring in collaboration with Natural Resources Institute Finland (LUKE) except the River Kuusoja salmonids that were selfcollected (Figure 1, Table S1).Endangered and protected salmonids were caught using the electrofishing method (SFS-EN 14011) under the licenses from the Regional Centre for Economic Development, Transport, and the Environment (ELY-centre) (POSELY/1015/5716-2016, POSELY/1738/5716-2017). Salmonid body mass (to 0.1 g) and length (to 1.0 mm) were measured in the field (Table S3).All captured salmonids were assumedly juveniles, but the fish were not dissected for maturity analysis.In the laboratory, we dissected part of the dorsal muscle tissue from the fish.The samples were stored at À80 C until further analyses.The muscle tissue was freeze-dried and pulverized with mortar and pestle.

| FA analysis
We analyzed the FA composition and content from the members of the EPT-taxa (Table S2).FA extraction was done with 2:1 chloroform:methanol (by volume) extraction following Folch et al. (1957).We used gas chromatography (GC) and mass spectrome- FA extraction from the salmonid samples was done in the same way, using 5-10 mg of pulverized tissue.For salmonid GC and MS, we used the same instrument as mentioned before, but we used split injection (20:1) with an oven program: The starting temperature was 150 C, after which raised from 150 to 180 C at 1 C min À1 , then to 210 C at 2 C min À1 , then to 230 C. The final temperature was held for 2 min.The total running time for samples was 49.00 min.
We used saturated FA 23:0 (Nu-Check prep., Elysian, MN, USA) as an internal standard.GLC-538 (Nu-Chek prep) was used for the calibration.Mass spectra and GLC-538 were used for the identification of FAs.

| Statistical methods
We used permutational multivariate analysis of variance (PERMANOVA) to investigate the effects of the river, sampling site within the river, and time from restoration on the zoobenthos community structure.We utilized a nested design in the PERMANOVA analyses, where sites were nested in rivers.PERMANOVA was also used to analyze the taxon-, river-, and site-specific differences in the FA composition of zoobenthos.Finally, PERMANOVA was used to analyze the taxon-, river-, and site-specific differences in the FA composition of salmonids.For zoobenthos, we ran an unrestricted permutation of raw data using type III sum of squares.For salmonids, total body length was used as a covariate in the PERMANOVA analyses, so we ran a permutation of residuals under a reduced model using type I sum of squares.
We used similarity percentage (SIMPER) analysis to examine, which taxa were driving the differences in the zoobenthos community structure among the rivers.Additionally, SIMPER was used to identify which FAs were driving the differences in the zoobenthos and salmonid FA composition among the rivers, sampling sites, and taxa.
We used a non-metric multidimensional scaling (NMDS) ordination to visualize the differences in the FA composition of zoobenthos among the taxa and rivers.NMDS ordinations and PERMANOVA were based on Euclidean distance.We used stress values to describe how well the ordinations described the data: Stress values <0.2 were considered acceptable, whereas stress values >0.2 were considered random ordinations (Clarke, 1993).The proportion of zoobenthos taxa in a river community and the proportion of individual FAs from the total FAs of zoobenthos and salmonids were arcsine square root transformed before the analyses.
Additionally, we used distance-based linear modeling (DistLM) to evaluate how much of the variation in the zoobenthos community structure could be explained by different river characteristics.We terrestrial organic matter (tOM) loading (pH, water color, and iron concentration), and bottom material (e.g., percentage of detritus, rocks, and water mosses) as listed in Table 1.
We used Kruskall-Wallis H, one-way analysis of variance The taxa that contributed to the differences in the zoobenthos community structure among the rivers mainly belonged to the EPTtaxa, along with some other taxa (Table S5).In river-specific analysis, the taxa that contributed the most to the aforementioned differences were: In River Ala-Koitajoki: caddisfly Hydropsyche pellucidula

| Variables explaining zoobenthos community structure
The factors explaining most of the variation in the zoobenthos community structure among the rivers were factors related to tOM loading (24% of the variation, Pseudo-F 7,55 = 3.42, p < 0.001) and bottom material (22% of the variation, Pseudo-F 7,55 = 3.9197, p < 0.001) (DistLM).
We found significant temporal changes in the zoobenthos community structure within two rivers in relation to the time from restoration (PERMANOVA, Pseudo-F 6,56 = 2.84, p < 0.001).PERMANOVA pair-wise tests revealed that the differences were statistically significant in Rivers Koitajoki and Kuusoja.In River Koitajoki, there was a statistically significant difference between the communities +10 to 14 and +15 to 19 years after restoration (t = 3.16, p < 0.05).The differences were mostly related to the proportions of chironomids (Chironomidae) and caddisfly Hydropsyche in the communities: Chironomidae were more common in the +10 to 14 years than in the +15 to 19 years after restoration communities, whereas Hydropsyche were less common in the +10 to 14 years than in the +15 to 19 years after restoration communities (SIMPER, Contribution: Chironomidae 14.43%, Hydropsyche 6.0%) (Table 2).
In River Kuusoja, there was a statistically significant difference between the communities +5 to 9 years and +20 to 24 years after restoration, (t = 1.53, p < 0.05), and the communities +10 to 14 and +20 to 24 years after restoration (t = 1.36, p < 0.05) (PERMANOVA, pair-wise tests).The differences were mostly related to the proportion of Ephemeroptera and Plecoptera: mayfly H. lauta was more common T A B L E 2 Proportions of different zoobenthos taxa and their contributions to the dissimilarities of zoobenthos community structure between communities differing according to time from restoration (SIMPER).in the +5 to 9 years than in the +20 to 24 years after restoration communities, and stonefly Leuctra hippopus (Plecoptera: Leuctridae) was more common in the +10 to 14 years than in the +20 to 24 years after restoration communities (SIMPER, Contribution: H. lauta 4.4%, L. hippopus 4.6%) (Table 2).In other rivers, no statistically significant differences in zoobenthos community structure were observed in relation to the time from restoration.
Within Ephemeroptera, we found statistically significant differences in Heptagenia sulphurea FA composition and content among the rivers.Significant differences were found between Rivers Ala-Koitajoki and Venejoki, Rivers Ala-Koitajoki and Naarajoki, and Rivers Ala-Koitajoki and Kalliojoki (Table S6).Most of the differences in EFA percentage and content were related to the percentage and content of ARA.There was also a statistically significant difference in the percentage of EPA between Rivers Ala-Koitajoki and Naarajoki and ω3/ ω6 ratio between Rivers Ala-Koitajoki and Kalliojoki (Tables 3 and 5).
In Trichoptera, we found statistically significant differences in the FA composition of Polycentropus flavomaculatus between Rivers Kalliojoki and Venejoki.We also found a significant difference in the FA composition between the two sites in River Ala-Koitajoki (Hiiskoski and Räväkkäkoski) for Hydropsyche angustipennis (PERMANOVA pairwise test, t = 1.69, p < 0.05) (Table S6).The percentage and content of EFAs in Trichoptera species also differed among the rivers.In P. flavomaculatus, there was a significant difference in ARA content between Rivers Kalliojoki and Venejoki.In Lepidostoma hirtum, we found a significant difference in the ω3/ω6 ratio between Rivers Ala-Koitajoki and Naarajoki (Tables 3 and 5).
No statistically significant differences were detected in Plecoptera species with respect to FA composition and content among the rivers.There was a significant difference in the salmon FA composition among three of the study sites (Ala-Koitajoki, Naarajoki site in Lieksanjoki and Ruunaa site in Lieksanjoki) (PERMANOVA, Pseudo-F 3,84 = 16.683,p < 0.001) (Figure 3).The essential FA DHA contributed to the differences in the salmon FA composition among these rivers (Table S7).Additionally, we found statistically significant differences in the FA composition between the two sites from River Ala-Koitajoki (Räväkkäkoski and Hiiskoski) (PERMANOVA, pair-wise test, t = 1.6204, p < 0.05).The FAs mostly responsible for the differences were DHA (16.1%), 12:0 (9.7%) and alpha-linolenic acid (ALA) (9.7%) (SIMPER, Table S7).

| FA composition and content of salmonids
Salmon FA content differed statistically significantly among the rivers.There was a significant difference in the total FA content between Rivers Ala-Koitajoki and Naarajoki (Figure 3).In essential FA content, most of the differences were found between Rivers Ala-Koitajoki and Naarajoki, but we also found a significant difference in the ARA content and ω3/ω6 ratio between Rivers Ala-Koitajoki and Ruunaa (Tables 4 and 5).
In brown trout, statistically significant differences in FA composition were detected between Rivers Ala-Koitajoki and Kuusoja (PERMANOVA, pair-wise test, t = 5.43, p < 0.001).The FAs mostly responsible for the differences were DHA (22.5%), 16:0 (12.2%) and ALA (10.4%) (SIMPER, Table S7).Brown trout EFA composition T A B L E 3 Mean percentage (%) and content (c, μg mg DW-1) of essential fatty acids (EFA), total fatty acid (FA) content (c, μg mg DW-1), and ω3/ω6 ratio with standard deviations (±) of zoobenthos in rivers.and content also differed statistically significantly between these rivers.There was also a significant difference in the total FA and EPA content, as well as the ω3/ω6 ratio and DHA percentage between these rivers (Tables 4 and 5).The total FA content, EFA composition and content, and ω3/ω6 ratio of salmonids in the study rivers are presented (Table 4).

| DISCUSSION
Expectedly, zoobenthos communities differed among the study rivers with variation being explained mainly by water quality parameters related to water color and DOC concentration.There was intraspecific variation in the FA composition and content of certain EPT-taxa, irrespective of the measured environmental characteristics of their home river.Our results indicate that the availability of physiologically EFAs to salmonids is primarily regulated by the zoobenthos community structure, but also by the intraspecific variability of EFAs in zoobenthos.Importantly, we found differences in the FA composition and content of salmon and brown trout irrespective of the body size of the fish, indicating potential dietary differences between these species.

| Zoobenthos community structure
Zoobenthos community structure is affected by local (river-specific), regional (species pool), and large-scale (e.g., climate, longitude, and latitude) factors (Sandin, 2009).Properties of the catchment area also play a role in determining zoobenthos community structure (Hämäläinen et al., 2007).The studied rivers were situated relatively close to each other, so the factors affecting the zoobenthos community structure were likely local rather than climatic.The zoobenthos communities were generally unique in each river, and the differences were mostly driven by differences in EPT-taxa abundances.Species of these taxa have different tolerances for multiple environmental conditions (Ficsor & Csabai, 2021;Smith et al., 2007), and they are, therefore, used as the key indicators in EU WFD-based ecological status classification.
Water quality parameters related to tOM loading mainly explained the zoobenthos community structure among the studied rivers.These rivers are dark in water color (ranging 110-190 mg L À1 Pt), which is typical for rivers in the study region (Rouvinen, 2010).Increased concentration of DOC can reduce the abundance and biodiversity of zoobenthos communities (Arzel et al., 2020;Brüsecke et al., 2022) but some taxa, such as mayfly T A B L E 4 Mean percentage (%) and content (c, μg mg DW À1 ) of essential fatty acids (EFAs), total fatty acid (FA) content (c, μg mg DW À1 ), and ω3/ω6 ratio with standard deviations (±) of salmonids in rivers.Note: Values with statistically significant differences ( p < 0.05) between each EFA, total FA content, and ω3/ω6 ratio between rivers have been highlighted and noted with letters.

Species
Baetis and Chironomidae might even benefit from it (Bellamy et al., 2019;Robbins et al., 2020).The concentration of DOC (especially humic matter) also decreases the pH (Oliver et al., 1983), which has been found to drive the taxonomical diversity of zoobenthos communities, together with nutrient availability (Baker et al., 2022;Heino et al., 2003).Similarly to our observation on river benthos, terrestrial DOC also contributes to the differences in the zoobenthos community structure among boreal lakes (Kesti et al., 2021;Strandberg et al., 2023).
In addition to tOM loading, characteristics of riverbed substratum (bottom material) explained zoobenthos community structure among the studied rivers.Diverse riverbed substrata support zoobenthos diversity (Huttunen et al., 2022); particularly many hydropsychids prefer specific substrata (Ficsor & Csabai, 2021).Aquatic macrophytes offer suitable microhabitats for several zoobenthos taxa, and macrophyte coverage can greatly influence the zoobenthos community structure (Huttunen et al., 2017).They also provide attachment sites for net-spinning zoobenthos taxa (Richardson & Clifford, 1986).Particularly water mosses are important hiding places for many key invertebrates and are thus central in river restoration (Korsu, 2004;Muotka & Laasonen, 2002).In the site Hiiskoski in River Ala-Koitajoki, water mosses were transplanted in 2018 as a part of the restoration.However, it can take several years for the moss-planting to influence the zoobenthos community structure (Hynninen & Vehanen, 2022).Therefore, the moss-transplantations were likely too recent to produce a clear impact on zoobenthos community structure in site Hiiskoski.
Only two rivers (Koitajoki and Kuusoja) showed changes in the zoobenthos community structure related to the time from restoration.
In Koitajoki, the changes in the community structure were mostly related to the increased proportion of the caddisfly Hydropsyche and the decreased proportion of Chironomidae.In River Kuusoja, the changes included several other taxa, but the changes were not clearly indicative of improving diversity.Due to the lack of proper reference sites for the past restorations, these results should be treated with caution as they could reflect overall land use and climatic changes or other temporal changes in these rivers.The interpretation is further hampered by the lack of pre-restoration data for the rivers.
In general, river restoration includes the addition of gravel and boulders to the riverbed (Louhi et al., 2011;Luhta et al., 2012), which creates new bottom substrata and alters the flow regime of the river.
These procedures usually decrease water velocity (Marttila et al., 2016), which could affect the zoobenthos community structure.The addition of gravel and boulders could benefit different EPT-taxa by providing them with new microhabitats.Improvement of water quality, especially the rising pH, has also been reported as beneficial for hydropsychids (Ficsor & Csabai, 2021).Sasaki et al. (2005) found that an increased concentration of nutrients and organic matter might also be beneficial for hydropsychids.Elevated levels of phosphorous and nitrogen, however, are harmful to less tolerant zoobenthos (Smith et al., 2007).

| FA composition and content of zoobenthos
Despite the strong control of phylogeny, we found significant riverand even site-specific differences in the FA composition and content of certain EPT-taxa.
We found biomarker FAs indicative of diatoms, bacteria, and tOM within the Ephemeroptera taxa.Especially mayfly H. sulphurea FA composition and content differed significantly among the study rivers.Heptagenia are grazers, feeding on periphyton (Merritt & Cummins, 1984).The differences in their FA composition among the rivers indicate that the periphyton composition and/or quality could differ among the study sites.Yet, it should be noted that many zoobenthos taxa, despite their associated feeding guilds, can be considered opportunists (Tierno de Figueroa et al., 2019).The FA profiles of scrapers may strongly correlate with the physical variables of the watershed.Specifically, PUFAs are positively associated with canopy cover, whereas saturated FAs (SAFAs) are negatively correlated with increased canopy cover (Wang et al., 2022).Also, we found in a previous study (Kesti et al., 2021) that shore type affected the FA composition of Heptagenia.These findings indicate that Heptagenia are generalists.
We found biomarker FAs indicative of diatoms, bacteria, and tOM also in Trichoptera taxa.Cashman et al. (2016) found high autochthonous signatures from the FA composition of Trichoptera, but some evidence suggests that site characteristics might also affect the FA composition of Hydropsyche (Cashman et al., 2016).Even though many hydropsychids are classified as filter-feeders (Cummins, 1973), there is evidence that they also practice selective feeding and omnivory (Basaguren et al., 2002;Ficsor & Csabai, 2021;Hellmann et al., 2013;Torres-Ruiz & Wehr, 2020).Previous study has documented high levels of 12:0 and 18:1ω9 and low levels of 16:1ω7 and EPA in Hydropsychidae, which may indicate a high contribution of allochthonous detritus in their diet (Descroix et al., 2010).Our results indicate both autochthonous and allochthonous resource utilization in the documented Trichoptera taxa.
There was one significant exception among the hydropsychids in their FA composition.The FA composition of Hydropsyche newae did not differ between Rivers Ala-Koitajoki and Naarajoki, despite the difference in the latitude between the two rivers.Unfortunately, we could not acquire physico-chemical data from River Naarajoki, so we cannot say if the physico-chemical characteristics were similar between these two rivers.This is worth discussing since it has been reported that the members of Hydropsychidae have very limited abilities to synthesize and modify their dietary FAs (Torres-Ruiz et al., 2010).This would indicate that the food sources were similar between these two rivers.Thus, it was even more interesting to find differences in the FA composition of H. angustipennis between the two sites in River Ala-Koitajoki (Hiiskoski and Räväkkäkoski).These differences in the FA composition could indicate differences in their diet between the two sites.Hiiskoski, right below the upstream Lake Koitere, has been restored and aquatic mosses have been transplanted onto the site (Hynninen & Vehanen, 2022).In Hydropsyche, ARA has been reported to originate from aquatic mosses (Torres-Ruiz & Wehr, 2020).Recent studies have also indicated that aquatic mosses can be an important food source for certain invertebrates (Kalachova et al., 2011;Labed-Veydert et al., 2021) and might influence the FA composition of their consumers.In the case of Hiiskoski, the differences may also be explained by the proximity of the lake and outlet effect, affecting the hydropsychids in Hiiskoski more than the hydropsychids in the downstream Räväkkäkoski.
There was a significant difference in the ω3/ω6 ratio in caddisfly L. hirtum between Rivers Ala-Koitajoki and Naarajoki.The ω3/ω6 ratio was exceptionally high in River Ala-Koitajoki which indicates strong utilization of autochthonous resources (Guo et al., 2016;Torres-Ruiz et al., 2007).L. hirtum is classified as a shredder (Azevedo-Pereira et al., 2006), but it has been shown to display detritivory, utilizing fine and coarse detritus (Basaguren et al., 2002).Unfortunately, we did not have data on the river characteristics of extensively restored River Naarajoki, but River Ala-Koitajoki had only little (0-5%) fine and coarse detritus, which could contribute to the differences in L. hirtum ω3/ω6 ratio between the rivers.
Polycentropus are carnivorous, net-spinning caddisflies (Philipson, 2010).We found statistically significant differences in P. flavomaculatus FA composition between Rivers Kalliojoki and Venejoki, which did not differ in their zoobenthos community structure.
Therefore, we assume that the food sources utilized by their prey items differ between these two rivers.However, as we have no information on the FA composition of their prey items, the differences in the P. flavomaculatus FA composition between these rivers cannot be attributed to a known source in this study.As mentioned above, Hydropsyche have very little ability to modify their dietary FAs (Torres-Ruiz et al., 2010), but to the best of our knowledge, there are no studies regarding Polycentropus on this matter.
We did not find any statistically significant differences in the FA composition and content of Plecoptera taxa among the rivers.Most of the Plecoptera species in this study were classified as shredders (L.hippopus, N. flexuosa, and P. intricata) (Cummins, 1973), except D. bicaudata, which is a predatory stonefly (Huhta et al., 1999).This would indicate that the food sources utilized by different Plecoptera taxa were similar among the study rivers.

| FA composition and content of salmonids
The FA composition of salmonids was affected more by the river identity than phylogeny.Juvenile salmon and brown trout inhabit slightly different microhabitats with salmon preferring higher water velocity than trout, which could have been predicted to cause diet-related differences in their FA composition.However, total body length had a significant impact on the FA composition suggesting significant ontogenetic niche shifts or alternatively could be caused by selective FA retention/metabolization.In the Great Lakes, length, together with condition factor/muscle lipid content, were found to be significant predictors for muscle FA composition in different fish species, including salmonids (Arnillas et al., 2023).Contrary to our results, Naesje et al. (2006) found no differences in the total lipid content between juvenile salmonid cohorts, when differences in fish body mass were considered.As juvenile salmonids grow, the content of monounsaturated FAs (MUFAs) decreases and the content of EPA and DHA increases (Murzina et al., 2016;Nemova et al., 2015).Consistently, the proportion of DHA in brown trout from River Kuusoja was higher than elsewhere.
Despite the greater influence of length on the FA composition of salmonids, we found small differences in the FA composition and content among the rivers, sites, and species.Based on the results, we can predict that the differences in the FA composition and content of salmonids among the rivers can be attributed to the differences in both the community structure and FA composition and content of zoobenthos among the rivers.Food quality can influence the growth rate of salmonids (Berge et al., 2009) so changes in the quality of zoobenthos could potentially impact salmonid populations in ecologically relevant ways.Also, salmonids with higher total FA content show better winter survival rates and improved swimming performance (Litz et al., 2017).Hence, differences in the community structure and/or the FA composition and content of zoobenthos might influence salmonid recruitment and parr survival.
Unfortunately, the salmonid samples from the Ruunaa site in River Lieksanjoki were collected from locations with no zoobenthos sampling.Thus, we cannot assess how the community composition and/or the FA composition and content of zoobenthos in Ruunaa affect salmonids in this area.Ruunaa is situated upstream from Naarajoki site in the same River Lieksanjoki system so their water quality and zoobenthos community compositions could largely resemble each other.Supporting this, the differences in the EFA percentage and content in salmonids were very small between Naarajoki and Ruunaa.
The amount of DHA was very low in the studied zoobenthos taxa.
Most freshwater insect larvae completely lack or have very low levels of DHA (Guo et al., 2016).Salmonids, however, can elongate and desaturate long-chain PUFAs from their shorter-chain analogs (Murzina et al., 2016;Nemova et al., 2015;Tocher, 2003).Despite the low levels of DHA in the zoobenthos taxa in this study, juvenile salmonids could satisfy their nutritional demand by consuming zoobenthos taxa abundant in its shorter chain analogs ALA and EPA (Vesterinen et al., 2021).Changes in the ALA and EPA composition and content could, thus, affect the DHA composition and content of salmonids, which could explain some of the observed differences in the FA composition and content of salmonids between the rivers.
Also, environmental changes can affect the food items available for salmonids.High water temperature and low current speed are favorable for the mass reproduction of terrestrial insects, in which shortchain PUFAs are more characteristic (Nemova et al., 2015).Browning of waters may also affect the foraging behavior and negatively impact the growth rate of visually foraging fish (van Dorst et al., 2020), so increased DOC might negatively affect salmonids.Future research questions should be pointed toward examining, how rivers with clear differences in water quality (especially in DOC concentration) affect the FA composition of salmonids.

| CONCLUSIONS
Despite the strong control of phylogeny on the FA composition of zoobenthos, differences in the FA composition and content of certain EPT-taxa were identified among the rivers and between the sites.
Especially mayfly H. sulphurea appeared to respond to local conditions suggesting that Heptagenia are generalists whose FA characteristics could be used as biomarkers of environmental quality.Using the publicly available environmental data, we observed temporal changes in the zoobenthos communities in Rivers Koitajoki and Kuusoja.The changes, whether resulting from river restorations or other environmental trends, could cascade to changes in the FA content and concentration of salmonids in these rivers.This opens a whole new research field: whether the quality of zoobenthos has an ecological effect on the recruitment of salmonids beyond the quantity of available prey.Due to the lack of diet data for the fish, we cannot link the environmental variables with FAs in salmonids, but future experimental studies could resolve this potentially important question in conservation physiology.

ACKNOWLEDGMENTS
The corresponding author would like to thank Jenny & Antti Wihuri Foundation and The Finnish Cultural Foundation for their grants, which supported the writing of this article.This study was also supported by Academy of Finland grant (#310450) to Paula Kankaala.We thank LUKE, especially Jorma Piironen and Matti Janhunen, for providing us with the fish samples and for their assistance and guidance on the selection of study rivers.We would also like to thank UEF, especially Jaakko Haverinen and Pyry Pihlasvaara for their help in fish sampling.Finally, we would like to thank Jussi Vesamäki for his help in the zoobenthos sampling and preparation and Jiri Vihavainen for his help in the laboratory work.
Abbreviation: N/A, no data available.
try (MS) to quantify and identify our FAs.FA extraction, GC, and MS for zoobenthos were done followingKesti et al. (2021).We used an Agilent 6890N GC (Agilent Technologies, Wilmington, DE, USA) equipped with a mass selective spectrometer (Agilent 5973N).The column was an Agilent DB-23 (0.25 mm Â 60 m Â 0.25 μm).The gas chromatography temperature program was as follows: The starting temperature was 50 C for 1 min, after which raised from 50 to 150 C at 15 C min À1 , then to 170 C at 0.5 C min À1 , then to 230 C at 2 C min À1 .The total running time for samples was 77.67 min.
used a stepwise selection of environmental factors with adjusted R 2 as model selection criteria.Several of the environmental predictors were strongly correlated, thus these predictors were assigned to specific indicators, based on their collinearity and environmental attributes.The indicators were morphometry (river length and size of the catchment area), oxygen (summertime concentration and saturation percentage), nutrients (concentration of phosphorous and nitrogen), Body size had a great influence on the FA composition of salmonids, with total body length explaining 24% of the differences in the FA composition (PERMANOVA, Pseudo-F 1,86 = 44.73,p < 0.001).F I G U R E 2 Non-metric multidimensional scale (NMDS) ordination of the fatty acid (FA) composition of zoobenthos.Species are presented with different markers.Each river is represented by its own color.The ordination was based on Euclidean distance.FAs that correlate strongly (Pearson r > 0.6) with either of the axes are visualized with vectors.The 2D stress value for the ordination was 0.15.[Color figure can be viewed at wileyonlinelibrary.com]Nevertheless, we found a statistically significant difference in the salmonid FA composition among the rivers (Pseudo-F 3,84 = 7.09, p < 0.001) (Figure3), with river identity explaining 22% of the variation.Taxon alone explained only 4% of the variation between the two salmonid species (Pseudo-F 1,86 = 2.88, p < 0.05).
Taxon and river in combination explained less of the differences in zoobenthos FA composition than taxon alone.This supports the F I G U R E 3 Non-metric multidimensional scale (NMDS) ordination of the fatty acid (FA) composition of salmonids.Species are presented with different markers.Each river is represented by its own color.The ordination was based on Euclidean distance.FAs that correlate strongly (Pearson r > 0.7) with either of the axes are visualized with vectors.The 2D stress value for the ordination was 0.09.[Color figure can be viewed at wileyonlinelibrary.com]

1
Zoobenthos community data and sampling sites and salmonid sampling sites.Detailed coordinates and sampling sites are presented in Table S1.Background orthophotos are an open aerial photo dataset (2023) provided by the National Land Survey of Finland.[Color figure can be viewed at wileyonlinelibrary.com]Physical and chemical factors of the study rivers.
ANOVA, and Bonferroni post hoc analyses were conducted using IBM SPSS Statistics 27 (IBM Corp., Armonk, New York, USA).
T A B L E 5 Statistically significant (p < 0.05) test results for different FAs (individual EFAs, ω3/ω6 ratio, total FA) for individual zoobenthos and salmonid species.
Note: Mean values and standard deviations (±) for FAs have been given (% = percentage of total FAs, c = FA content [μg/mg DW À1 ]).Statistical tests with degrees of freedom (df) and statistical significance ( p) are also shown.