Linking Riparian Forest to the Functional Diversity of Ephemeroptera, Plecoptera, and Trichoptera in First-Order Tropical Streams

Abstract

Riparian forests are important for maintaining aquatic biodiversity, yet they face increasing pressure from logging activities. This study assessed the functional diversity of Ephemeroptera, Plecoptera, and Trichoptera (EPT) in 30 Amazonian first-order streams across three riparian forests: pristine, selectively logged, and conventionally logged. We evaluated four habitat attributes linked to ecosystem functioning (canopy cover, water temperature, sediment organic matter, and small woody debris) and calculated two indices of functional diversity: richness and divergence. Functional diversity was highest in pristine streams, intermediate in selectively logged streams, and lowest in conventionally logged streams. Functional richness and divergence declined significantly in conventionally logged forests, indicating a loss of ecological traits and potential reductions in ecosystem functions. We also observed that canopy cover, sediment organic matter, and woody debris were positively associated with EPT functional diversity, while water temperature had a negative association. These findings highlight that conventional logging leads to the functional homogenization of aquatic insect assemblages, compromising key ecological processes. Selective logging that maintains riparian buffers may preserve functional diversity, even though these differences may be influenced by site-specific environmental conditions. Our results underscore the importance of conserving riparian integrity to sustain the resilience and functioning of tropical stream ecosystems in logged landscapes.

1. Introduction

Riparian forests serve as transition zones between terrestrial and aquatic ecosystems [1]. They play an important role in regulating the input of allochthonous resources such as leaves, branches, fruits, and small terrestrial invertebrates [2]. However, landscape alterations in these areas can destabilize food web structures, reduce primary production rates, and decrease habitat availability and food resources [3,4,5]. These forests have been significantly converted into pastures and agro-industrial plantations [6,7] and have been subject to timber extraction [8]. Approximately one-quarter of the world’s tropical forests are managed for logging, and this practice in riparian zones can have significant effects on adjacent aquatic ecosystems [9]. It can alter water temperature, impacting aquatic metabolism, and reduce the input of organic matter [10].

Timber harvesting generally follows two main approaches. Conventional logging removes vegetation without sustainable management, drastically reducing canopy cover and ecosystem services [8]. In contrast, selective logging maintains protective buffers along streams and, when properly managed, helps to preserve ecosystem services, conservation value, and mitigate biodiversity loss [11]. In tropical regions, conventional logging has resulted in extensive forest loss in recent decades, which is a significant concern because the degradation of these areas threatens critical ecological interactions [12,13]. On the other hand, selective logging may help conserve high-value forest patches essential for maintaining the fish food resources and aquatic insects [3,8].

Despite the ecological importance of aquatic insects, such as Ephemeroptera, Plecoptera, and Trichoptera (EPT), there is limited information on how logging activities affect their functional attributes linked to feeding habitat or respiration. EPT taxa are ecologically significant and valuable indicators of environmental conditions [14]. Understanding how riparian disturbances influence EPT assemblage is crucial, as changes in their functional traits reflect anthropogenic impacts [15]. EPT taxa exhibit diverse life-history strategies and play key roles in aquatic ecosystems, such as shaping organic matter dynamics, nutrient cycling, and predation [16]. These functions support ecological balance and directly influence the health of aquatic ecosystems [17]. Therefore, it is important to assess how logging practices impact the functional roles of EPT in logged areas. Analyzing functional diversity involves looking beyond species counts to examine the distinct roles that different taxa play within ecosystems [18,19]. This approach reveals trophic complexity and indicates ecosystem resilience and recovery capacity following disturbances [20,21].

This study investigates the functional diversity of EPT in first-order tropical streams across three riparian canopy covers: pristine forest, selective logging, and conventional logging. We assessed habitat attributes linked to ecosystem functioning [22], including water temperature (°C), sediment organic matter (g), fine woody debris (g), and riparian canopy cover (%). We hypothesize that EPT functional diversity would differ between streams in pristine and conventionally logged forests due to decreases in riparian canopy cover, sediment organic matter, fine woody debris, and increases in water temperature. Conventional logging creates fragmented landscapes with fewer large trees, reducing the allochthonous resources necessary to maintain aquatic habitats [8]. In contrast, we expect functional diversity to be maintained in selectively logged forests, since selective tree removal tends to preserve a relatively riparian canopy, ensuring continued input of food resources and providing shade that helps regulate stream temperature.

2. Materials and Methods

2.1. Study Area

This study was carried out in the middle Capim River basin, in southeastern Pará state, in the eastern Brazilian Amazon (Figure 1). This area encompasses 148,091 hectares that have undergone an extensive land conversion from forest to mechanized farms, silviculture (Eucalyptus sp.), and areas earmarked for logging. However, 98,000 hectares are covered with primary rainforests, dominated by Sapotaceae, Leguminosae, Lecythidaceae, and Bignoniaceae species. In the study area, the climate is tropical hot and humid, corresponding to Köppen’s Am type, with rainfall of approximately 2000 mm/year, relative air humidity of 80%, and a well-defined seasonality: a dry season from July to November and a rainy season between December and June [23].

We sampled 30 first-order streams selected based on the use of the riparian vegetation. The first area was pristine forest (10 streams), conserved in the Rio Capim Forest Management Unit, which belongs to Cikel Brasil Verde Madeiras LTDA. In this area, we observed continuity between the riparian vegetation and the adjacent forest, and the absence of human impacts. These streams have an average length of 1087 m, an average depth of 1.08 m, and an average canopy cover of 97%. The streams have a substrate composed of fine sediment, mostly fine and medium sand. The second area was the selectively logged forest (10 streams), managed in the Rio Capim Forest Management Unit, which belongs to Cikel Brasil Verde Madeiras LTDA. In this area, commercially valuable trees are marked and selectively removed to minimize disturbance to adjacent habitats. In addition, logging cycles occur after 30 years of forest regeneration, and the riparian vegetation along streams is relatively preserved. These streams have an average length of 1082 m, an average depth of 1.06 m, and an average canopy cover of 73%. The streams have a substrate composed of fine sediment, mostly fine and medium sand. The third area was the conventionally logged forest (10 streams), outside the Rio Capim Forest Management Unit. In this area, timber is removed without forest or riparian protection. These streams have an average length of 1069 m, an average depth of 1.06 m, and an average canopy cover of 22%. The streams have a substrate composed of fine sediment, mostly fine and medium sand. Cikel Brasil Verde Madeiras LTDA maintains the pristine forest exclusively for conservation and the selectively logged area for sustainable forest management, in compliance with national regulations. Moreover, in both selectively and conventionally logged forests, native trees are maintained and not replaced by exotic species. Detailed habitat attributes are provided in

Table A1. Environmental characteristics (stream length, width, and depth; riparian canopy cover; and water temperature) of the 30 first-order tropical streams used for EPT sampling within the three different riparian forests (pristine, selective logging, and conventional logging) in the eastern Amazon. Riparian canopy cover (%) was calculated using the approach proposed by [24]. All variables are expressed as mean ± standard deviation.

Riparian Forest	Length (m)	Width (m)	Depth (m)	Canopy Cover (%)	Water Temperature (°C)
Pristine forest
Stream 1	1095	4.0 ± 1.2	0.9 ± 1.05	96 ± 1.03	21.2 ± 1.41
Stream 2	1071	4.3 ± 1.1	1.1 ± 1.04	96 ± 1.03	21.3 ± 1.31
Stream 3	1126	5.1 ± 1.1	1.3 ± 1.05	95 ± 1.02	21.2 ± 1.15
Stream 4	1065	4.5 ± 1.2	1.1 ± 1.02	97 ± 1.01	21.1 ± 1.22
Stream 5	1192	5.8 ± 1.2	1.2 ± 1.03	97 ± 1.01	21.2 ± 1.23
Stream 6	1064	5.2 ± 1.2	0.8 ± 1.03	97 ± 1.02	21.5 ± 1.21
Stream 7	1073	4.5 ± 1.3	0.9 ± 1.02	98 ± 1.02	21.3 ± 1.25
Stream 8	1051	4.7 ± 1.1	1.1 ± 1.02	98 ± 1.04	21.3 ± 1.32
Stream 9	1072	4.3 ± 1.1	1.2 ± 1.04	96 ± 1.01	21.1 ± 1.35
Stream 10	1057	4.3 ± 1.2	1.2 ± 1.02	98 ± 1.03	21.1 ± 1.42
Selective logging
Stream 1	1026	4.7 ± 1.2	1.1 ± 1.01	72 ± 1.01	21.9 ± 1.23
Stream 2	1085	4.5 ± 1.2	1.3 ± 1.03	74 ± 1.01	22.1 ± 1.25
Stream 3	1063	4.3 ± 1.1	1.3 ± 1.01	75 ± 1.03	21.8 ± 1.19
Stream 4	1055	5.9 ± 1.3	0.8 ± 1.05	74 ± 1.02	22.1 ± 1.25
Stream 5	1078	5.5 ± 1.1	0.8 ± 1.02	73 ± 1.02	22.5 ± 1.21
Stream 6	1117	4.6 ± 1.1	0.9 ± 1.01	73 ± 1.01	22.6 ± 1.28
Stream 7	1092	5.1 ± 1.4	1.1 ± 1.02	74 ± 1.03	22.3 ± 1.22
Stream 8	1093	4.2 ± 1.3	0.9 ± 1.03	74 ± 1.03	21.9 ± 1.25
Stream 9	1125	4.7 ± 1.3	1.3 ± 1.03	73 ± 1.01	22.4 ± 1.21
Stream 10	1095	4.6 ± 1.2	1.1 ± 1.02	74 ± 1.01	22.5 ± 1.22
Conventional logging
Stream 1	1034	5.4 ± 1.1	1.2 ± 1.02	21 ± 1.02	25.5 ± 1.81
Stream 2	1059	5.1 ± 1.1	1.1 ± 1.01	24 ± 1.02	25.4 ± 1.65
Stream 3	1047	4.6 ± 1.2	1.2 ± 1.01	22 ± 1.01	26.1 ± 1.41
Stream 4	1116	5.9 ± 1.2	1.2 ± 1.02	22 ± 1.01	25.4 ± 1.33
Stream 5	1082	5.2 ± 1.2	0.8 ± 1.04	23 ± 1.01	25.3 ± 1.45
Stream 6	1049	4.7 ± 1.3	0.8 ± 1.03	19 ± 1.02	26.1 ± 1.21
Stream 7	1112	4.6 ± 1.3	0.9 ± 1.02	21 ± 1.02	25.2 ± 1.37
Stream 8	1075	4.5 ± 1.2	1.1 ± 1.01	24 ± 1.03	26.1 ± 1.29
Stream 9	1069	4.5 ± 1.1	1.1 ± 1.01	23 ± 1.02	25.7 ± 1.31
Stream 10	1047	4.6 ± 1.1	1.2 ± 1.04	23 ± 1.02	24.9 ± 1.28

.

2.2. Field Sampling Design

Sampling for assessing EPT assemblage and habitat attributes was conducted at all sites between September to October 2019, when water levels were sufficiently low to allow access and robust quantitative sampling in the streams. We used a minimum geographical distance of 3 km between each stream to preserve the independent sampling unit and select streams from different sub-basins. This design was conducted to avoid the potential impacts of stream connectivity [3]. In the middle course of each stream, we established one 150 m reach. Sampling was carried out in the upstream direction, where habitat attributes were sampled initially to avoid potential bias due to water column disturbance.

2.3. Measurement of Attributes Linked to Habitat Functioning

We measured four habitat attributes—water temperature (°C), sediment organic matter (g), small wood debris (g), and riparian canopy (%)—to test our hypothesis that EPT functional diversity would differ between streams with pristine and logged forests due to a loss in habitat functioning. Along each reach, we randomly took 10 water temperature measurements to estimate the average temperature of each stream. The measurements were taken around 9 a.m. using a digital probe at the bottom of the water column to avoid differences caused by daily variation. We randomly collected 10 sediment samples along each reach using a 0.1 m² hand net with 63 µm mesh to quantify the sediment organic matter. Twenty grams of the dry sediment (40 °C for 48 h) was ignited in a muffle furnace at 550 °C for 4 h to estimate the sediment organic matter for each stream (0.001 g precision). In addition, we randomly collected 10 samples of leaf litter using a 0.1 m² quadrat, which were used to estimate the average small wood debris for each stream (0.001 g precision). We measured the average riparian canopy for each stream using the approach proposed by [24]. We randomly took 10 hemispherical photographs along each stream using a smartphone equipped with a fisheye lens. The photographs were taken with the smartphone held horizontally and later analyzed using the GLAMA software 3.0 (Gap Light Analysis Mobile Application) [24].

2.4. Sampling of EPT Assemblage

For a standard sampling of the EPT specimens and to cover all microhabitats, we used a reinforced rectangular (30 by 15 cm) hand net with 300 µm mesh. We sampled each reach for 30 min, manually disturbing the substrate to a depth of approximately 10 cm. At each stream, we washed the sample collected in the field to remove coarse particles, and the remaining material was immediately placed in individual plastic buckets and preserved in 70% ethanol. Each sample was again washed in the laboratory through a 300 µm mesh. The EPT specimens were counted and identified at the genus level using identification keys for Amazon aquatic insects [25].

2.5. Functional Trait Selection

We selected three functional traits linked to niche dimensions and resource acquisition (

Table 1. Presence (1) and absence (0) and functional traits of EPT genera in the three different forests in the eastern Amazon. Information about functional traits is sourced from [26,27,28].

Order	Genus	Presence and Absence			Functional Traits
		Pristine Forest	Selective Logging	Conventional Logging	Main Food Type a	Feeding Habit b	Respiration c
Ephemeroptera	Baetodes	1	1	0	Fd	CG	GT
	Apobaetis	1	1	0	FdMi	CGSC	GT
	Farrodes	1	1	0	FdCd	SH	G
	Campylocia	1	0	0	FdCd	SH	G
Plecoptera	Anacroneuria	1	0	0	Ma	PR	G
Trichoptera	Marilia	1	1	1	FdMi	CGSC	GT
	Leptonema	1	1	1	FdCdMa	CFSHPR	GT
	Nectopsyche	1	1	1	FdCd	CGSH	GT
	Cernotina	1	1	0	Ma	PR	T
	Chimarra	1	1	0	Fd	CF	T
	Hydroptila	1	1	0	Mi	PI	T
	Helicopsyche	1	0	0	Fd	CG	T

^a Main food: fine detritus (Fd); coarse detritus (Cd); microphytes (Mi); macroinvertebrates (Ma). ^b Feeding habit: collector-gatherer (CG); shredder (SH); scraper (SC); predator (PR); collector-filterer (CF); piercer (PI). ^c Respiration: gill (G); tegument (T).

) associated with ecosystem functioning [26]. These traits were defined to include measurable attributes that may reflect individual insect behavior and association with habitat. These attributes include the main food types (macroinvertebrates, microphytes, and coarse and fine detritus), feeding habits (collector-gatherer, collector-filterer, shredder, scraper, predator, and piercer), and respiration (tegument and gill). Trait values are sourced from [26,27,28].

2.6. Functional Diversity Indices

We combined the presence–absence data of EPT assemblage with functional trait information to calculate two functional indices: functional richness and functional divergence [29]. Functional richness represents the total niche space filled by all species in a community, where the niche space is measured as the convex hull volume [19]. Functional divergence measures how species are distributed within the convex hull volume of traits and is independent of functional richness [19,29]. When using presence–absence data, this index ranges from zero (species present similar attributes) to one (several species with unique attributes are present, indicating high resource partitioning). All functional indices were calculated using the FD package 1.0–12.3 [30] in the GNU R 4.4.1 [31].

2.7. Statistical Analysis

We used a mixed-effects model based on the negative binomial distribution to test whether EPT functional diversity would differ between streams with pristine and logged forests. This model was used to minimize the effects of possible non-independence of the sampling units [32]. In addition, mixed-effects models provide a flexible and powerful tool for modeling within-group correlation and minimizing any effects of overdispersion [32]. In this approach, each stream was considered a sampling unit. We used the riparian cover (pristine, conventional, and selective logging) as the fixed effect. The random effect was included with the factor sampling unit (30 streams), taking into account the random intercept. Where significant, we conducted a post hoc test for the mixed-effects model for pairwise comparisons using medians. Assumptions of homogeneity of variances, normality, and uniformity of residual distributions were checked using residual plots after modeling.

We used a mixed-effects random forest regression in the randomForest package 4.7 [33] in GNU R 4.4.1 [31] to identify patterns in EPT functional diversity associated with the attributes linked to habitat functioning. This model was used to minimize the effects of possible non-independence of the sampling units [34]. We used the riparian cover (pristine, conventional, and selective logging) as the fixed effect. The random effect was included with the factor sampling unit (30 streams), taking into account the random intercept. We overtrained the model by selecting a subsample from the whole dataset by bootstrapping. Afterwards, we ran the training model progressively [35], using the following settings: mtry function = 1 to 4; ntree function = 200, 300, 400, 500, 600, 800, 1000, and 10,000. Observations not included in the bootstrap subsample were defined as out-of-bag (oob) samples and were used to create an oob estimation of the generalized error in the model [35]. The optimum number of mtry and ntree to be used in the final model was selected from the model with the smallest generalized error. We chose the final random forest model with the highest variance explained (pseudo-R² value) for the oob samples. The relationship between EPT functional diversity and the attributes linked to habitat functioning was described using partial dependence plots.

3. Results

We recorded 12 EPT genera (Table 1). In the pristine forest, we found all EPT genera (four Ephemeroptera, one Plecoptera, and seven Trichoptera), with different respiration strategies, a broader spectrum of feeding habits (generalists and specialists), and feeding on a greater diversity of food types (Table 1). In the selectively logged forest, we found nine EPT genera (three Ephemeroptera and six Trichoptera). In these streams, we observed a slight loss of feeding habits but a still-maintained good diversity of food types and respiration strategies (Table 1). However, in the conventionally logged forest, we observed only three EPT genera (three Trichoptera) and a considerable loss of functional traits (Table 1).

The loss of functional traits between the pristine and logged forests was corroborated by the significant decrease of functional richness (F_2,27 = 0.92; p < 0.001; R² = 0.84;

Table A2. Calculated values of functional richness and functional divergence of EPT assemblage within the three different riparian forests (pristine, selective logging, and conventional logging) in the eastern Amazon.

Riparian Forest	Functional Richness	Functional Divergence
Pristine forest
Stream 1	0.43	0.83
Stream 2	0.46	0.85
Stream 3	0.43	0.84
Stream 4	0.45	0.85
Stream 5	0.34	0.80
Stream 6	0.46	0.85
Stream 7	0.46	0.85
Stream 8	0.43	0.84
Stream 9	0.46	0.85
Stream 10	0.46	0.83
Selective logging
Stream 1	0.29	0.82
Stream 2	0.37	0.85
Stream 3	0.08	0.76
Stream 4	0.40	0.84
Stream 5	0.40	0.85
Stream 6	0.26	0.81
Stream 7	0.09	0.70
Stream 8	0.29	0.73
Stream 9	0.17	0.82
Stream 10	0.12	0.70
Conventional logging
Stream 1	0.009	0.67
Stream 2	0.03	0.62
Stream 3	0.009	0.67
Stream 4	0.009	0.67
Stream 5	0.009	0.67
Stream 6	0.03	0.62
Stream 7	0.03	0.62
Stream 8	0.009	0.63
Stream 9	0.008	0.63
Stream 10	0.009	0.63

) and functional divergence (F_2,27 = 0.07; p < 0.01; R² = 0.45; Table A2) (Figure 2). In terms of functional richness, this difference varied greatly throughout the three forests (Figure 2a). Although the pristine and selectively logged forests share EPT genera (Table 1), we observed that the selectively logged forest lost functional traits associated with collector-gatherers, shredders, and predators. However, this difference was negligible for functional divergence (low R² value) and observed only between pristine and conventionally logged forests (Figure 2b).

The final random forest model showed that EPT functional diversity was strongly associated with the attributes linked to habitat functioning, explaining 78.3% of the total variance. We observed a positive association between functional richness and divergence with the riparian canopy, small wood debris, and sediment organic matter (Figure 3). In contrast, we observed a negative association with water temperature (Figure 3). This model was obtained with mtry = 3 and ntree = 600, since performance was not substantially enhanced after these parameters. In addition, the oob generalized error was 0.007. The high variance explained and low oob generalized error indicated the high predictive power of the model.

4. Discussion

Our results show that conventional logging in riparian forests is associated with a significant decrease in the functional diversity of EPT, which supports our initial hypothesis. This decline in functional traits may be linked to a substantial loss of ecosystem functioning due to the disappearance of more diverse EPT groups. A similar pattern has been observed in areas deforested for oil palm plantations in Southeast Asia [7,9]. Our findings underscore the importance of preserving riparian integrity to maintain biodiversity and ecological function in aquatic ecosystems [36]. In our study, pristine riparian forests exhibited a wide variety of respiratory strategies and feeding habits, including both specialist and generalist EPT genera. This functional diversity may be associated with greater microhabitat complexity and distinct ecological niches [10,37]. For example, the structural heterogeneity provided by woody debris and leaf litter facilitates niche partitioning and allows the coexistence of taxa with differing ecological requirements [38,39]. In addition, woody debris may serve as both a refuge from flow and a substrate for biofilm development, while leaf accumulations support detritivore communities through gradual organic matter processing [22]. This habitat diversification is particularly important for EPT assemblages, as different life stages often utilize distinct microhabitats: In general, Plecoptera require clean interstitial spaces in coarse substrates and Trichoptera depend on specific particle sizes for case-building [25]. Furthermore, the vertical stratification of organic matter (from suspended fine particulates to buried coarse debris) creates a resource gradient that supports diverse functional feeding groups, enhancing ecosystem processes such as organic matter breakdown and nutrient cycling [17].

In contrast, we observed a simplified EPT assemblage in conventionally logged forests, with only three Trichoptera genera and a significant loss of functional traits. This functional homogenization was also reported by [14] in streams in the Amazon associated with mining, indicating a consistent pattern in tropical aquatic ecosystems. The absence of shredder and collector genera highlights a significant depletion of food resources, such as coarse and fine particulate organic matter [1,40]. Our results confirm this correlation, showing decreased riparian cover, woody debris, and sediment organic matter in conventionally logged streams are linked to lower EPT functional diversity. Decreased woody debris inputs limit habitat complexity, reducing niche availability for shredders that rely on stable substrates [27,41]. In addition, lower sediment organic matter disrupts detrital processing pathways, impairing collector taxa dependent on fine particulate organic matter [2]. These combined stressors lead to biotic homogenization, favoring generalist species with narrow ecological tolerances. Furthermore, we found a negative effect of increased water temperature on EPT functional diversity. Similar findings have been found in temperate and tropical streams, where rising water temperatures due to forest cover loss have led to declines in sensitive EPT genera [10,14,15,40,41]. Increased water temperatures may significantly alter aquatic metabolism, leading to heightened microbial respiration rates and organic matter decomposition, raising the demand for dissolved oxygen. This may result in hypoxia risks and water quality degradation that impair stream ecosystem services [4]. On the other hand, reducing riparian cover may increase light availability in streams, promoting algal growth that supports scraper functional groups of Ephemeroptera and Trichoptera [1]. However, the streams analyzed have turbid water and their substrate is composed of fine sediment, which may limit algal development.

Selective logging, on the other hand, retained relatively higher functional diversity, particularly among Ephemeroptera and Trichoptera groups. Although some functional traits were lost in these forests, the decline was not statistically significant, suggesting that reduced-impact logging may help preserve aquatic ecosystem functions. Similar results were reported by [8] in Eastern Amazonian streams, where selective logging maintained greater Odonata diversity than conventional logging. However, our findings showed a substantial decline in Plecoptera genera, which are highly sensitive to habitat disturbance. Although pristine and selectively logged forests shared some Trichoptera taxa, the loss of functional groups of Plecoptera may disrupt important ecosystem processes, reducing resource partitioning [36].

In conclusion, the loss of tropical riparian forests is linked to significant declines in EPT functional diversity. Thus, our findings emphasize the need for management practices that protect riparian ecosystem processes. While selective logging offers potential for conserving EPT functional diversity, it still results in losses of ecologically important groups such as Plecoptera. In this sense, vegetated buffers that comply with regulatory standards are important for maintaining functionally diverse macroinvertebrate assemblages [9]. Additionally, active restoration strategies focusing on native tree species that produce coarse litter could effectively recover ecological functions [37]. Our study focused on a single macroinvertebrate group (EPT) susceptible to environmental changes, which limits the generalization of our findings to other aquatic taxa. Additionally, the spatial scale of our research was restricted to headwater streams, and broader-scale patterns may differ. Moreover, our study may be limited by the confounding of treatment and location, as each forest-use type corresponds to a distinct site. Consequently, some of the observed differences in functional diversity may be influenced by forest management practices and site-specific environmental conditions. Thus, future research should evaluate functional traits across a wide range of macroinvertebrate taxa and larger spatial scales to better understand aquatic ecosystem responses to various riparian forest management practices.