Study area
Zarin Gol River is one of the Gorgan-Rud tributaries with its geographical location of longitude 54° 57′ and latitude 35° 52′. The maximum volumetric discharge of river water is 150×106 m3/year with the sandy gravel-bed river (Gholizadeh et al. 2017). The sampling sites were located in the indigenous forest area of the basin (Fig. 1).
The forest cover of the area consisted of mostly oak, European hornbeam (Parrotia persica), maple, Lindens (Zelkova carpinifolia), and English yew (Taxus baccata L.) with an average canopy cover of 52.7%, stretched over 81.7% of the total area. Rangelands cover most of the upper elevations of the basin and around some of the villages including 10.1% of the basin (Nasr Nasrabadi 1998).
The downstream areas of the river, which are affected by human activities, were excluded from this study. Therefore, only the impacts of natural disturbance were included as significant factors in shaping benthic macroinvertebrate communities in the study area.
The catchment area and its maximum elevation were about 342.82 km2 and 2800 m, respectively. Heavy rainfall caused the flood events that occurred in this stream in the summer (August 2014 and 2017) because the permeability of soil is low while the soil slope is high (Fig. 2). The historical hydrographs analyzed in this study were from 1987 to 2017. The first survey before the higher water levels (the floods), conducted in June 2014 and 2017, was regarded as a reference sample. Figure 3 presents two of the floods, illustrating the relevant variables derived from the flood hydrograph. Due to high discharge, the water level rose significantly on August 8, 2014 (average rainfall of 27.5 mm in less than 24 h), August 19 and 20, 2017 (average precipitation of 22.5 mm) (Fig. 3), and 9 months later (June 2018); the data were collected to examine the recovery of communities.
Macroinvertebrate collection
Macroinvertebrates were collected by a Surber sampler (900 cm2 area and 250-μm mesh size) with 3 replications (diagonal transect across a riffle for 10 m and all sample collected in different parts of the same riffle) from 7 stations (3 replications × 7 stations × 5 months = 105 samples) along the Zarin Gol River, Iran, 1 month before (June 2014 and 2017) and after the flood (September 2014 and September 2017) and 9 months after the flood (June 2018). Samples were preserved in formaldehyde (final concentration 4%) until further analysis and poured the contents of the sieve into the tray to separate the organisms from the underlying particles under the stereoscope. Macroinvertebrates were identified in the level of family or genus based on valid keys (Pennak 1953; Needham 1976; Pescador et al. 2004).
Biological measurement criteria
To evaluate the effects of environmental factors on aquatic life in the running waters, data from macroinvertebrate communities were collected utilizing a multimetric index of biodiversity. The multimetric index of biotic integrity is known as the Biological Assessment Profile (BAP) score, which is used for the biological assessment of water quality. BAP includes four components such as species richness, Ephemeroptera-Plecoptera-Trichoptera (EPT) richness, Hilsenhoff Biotic Index (HBI) (Hilsenhoff 1987), and percent model affinity (PMA) (Novak and Bode 1992). The result of each criterion is placed on a standard 10-criterion scale, and then, the average of these values is computed. A four-tiered scale of water quality (non-impacted: BAP>7.5, slightly impacted: BAP 5–7.5, moderately impacted: BAP 2.5–5 or severely impacted: BAP<2.5) has been established for the obtained BAP Score (NYSDEC 2012). The results of sampling in the middle or extreme classes show significant effects on aquatic life, which are included in the list of disturbed waters. Samples with a low organism density usually need a higher percentage of samples for processing, while samples with high organism density generally need a small percentage of samples for processing to reach the same subset of 100 organisms.
Statistical analysis
Two analyses were applied to illustrate the total effects of floods on macroinvertebrates assemblages, to indicate how specific taxonomic groups were affected, and also to represent how the rate of different indices and the composition of the assemblages are recovered after the floods. In the first step, analysis of variance (ANOVA) and multiple-comparisons procedures (Tukey test) were utilized to determine if the BAP scores, component metrics, and density changed significantly (α = 0.05) among the five sampling events. Therefore, this test determines whether the impact of flooding was significant and how long each metric takes to be recovered.
Second, multivariate methods with PRIMER-E Version 6 software were applied to measure temporal changes in community composition. A square-root transformation was used for macroinvertebrate and Bray-Curtis distance (Primer-E Ltd) was applied to analyze the community similarities between samples. A one-way analysis of similarities (ANOSIM) test was applied to examine the null hypothesis indicating that there are no differences between the periods of surveys (Clarke and Gorely 2006). Permutation distribution of the R-statistic in ANOSIM was used to measure differences in community composition at survey periods (Clarke and Gorely 2006). There are no differences in community composition with values close to 0 (R <0.25) whereas there is a significant difference in the community composition with the values close to 1 (R> 0.75) (Clarke and Gorely 2006). Also, a nonmetric multidimensional scaling ordination (nMDS) was used to illustrate the similarity of macroinvertebrate communities between the survey periods (before, after the flood, and recovery time) and to demonstrate information on the Bray-Curtis distance matrix. The evaluation of categorized composition was performed by nMDS at all stations before and after the flood and during the recovery period. Considering the distance, and the location of the samples in the species ordination space relative to the floods, some inferences were drawn about the resistance, resilience, and recovery of macroinvertebrate communities. Similarity percentage (SIMPER) analysis was conducted to identify taxa that contributed greatly to sample dissimilarity between sets detected in the nMDS (Clarke and Gorely 2006). SIMPER conducted paired comparisons of all samples; therefore, these comparisons could determine which species generate the most difference between the surveys (Clarke and Gorely 2006). Using the groups of studies that illustrated the most changes in community composition, sampling stations were classified into three classes of flood effects. These classes were demonstrated as before the flood (June 2014 and 2017), after the flood (September 2014 and September 2017), and recovery time (June 2018). To calculate the mean of differences between the three classes, SIMPER was applied based on the occurrence of individual species.