Ecological uniqueness of fish communities from streams in modified landscapes of Eastern Amazonia

Ecological Indicators 111 (2020) 106039

Contents lists available at ScienceDirect

Ecological Indicators journal homepage: www.elsevier.com/locate/ecolind

Ecological uniqueness of fish communities from streams in modified landscapes of Eastern Amazonia

T

⁎

Híngara Leãoa, , Tadeu Siqueirab, Naiara Raiol Torresc,d, Luciano Fogaça de Assis Montagd a

Instituto Federal do Piauí, Rodovia PI 247, KM 7, Portal dos Cerrados, CEP: 64860-000 Uruçuí, Piauí, Brazil Departamento de Ecologia, Instituto de Biociências, Universidade Estadual Paulista (UNESP), Avenida 24A, 1515, Jardim Vila Bela, CEP: 13506-900 Rio Claro, São Paulo, Brazil c Programa de Pós-Graduação em Ecologia Aquática e Pesca, Universidade Federal do Pará, Rua Augusto Corrêa, 01, Guamá, CEP: 66075-110 Belém, Pará, Brazil d Laboratório de Ecologia e Conservação, Instituto de Ciências Biológicas, Universidade Federal do Pará, Rua Augusto Corrêa, 01, Guamá, CEP: 66075-110 Belém, Pará, Brazil b

A R T I C LE I N FO

A B S T R A C T

Keywords: Beta diversity Ichthyofauna Local contribution to beta diversity (LCBD) Species contribution to beta diversity (SCBD) Land use

Ecological uniqueness is one aspect of beta (β) diversity that shows the relative contribution of sites (local contribution to beta diversity – LCBD) and taxa (species contribution to beta diversity – SCBD) in the formation of a unique environment in terms of species composition, and may be directly related to habitat quality. Our objective was to evaluate the uniqueness of fish communities from streams in modified areas of the Amazon, and to investigate the main environmental predictors at local and landscape scale. We sampled 58 streams in the Capim river basin (Pará, Brazil) located within areas of preserved forest, reduced-impact logging, conventional logging and pasture. We found greater β diversity and higher LCBD in the pasture areas, making this land use the largest contributor to β diversity in the study area. At local scale, this high contribution was primarily influenced by environmental heterogeneity, thalweg depth, percentage of land use and cover (all positively), and volume of large woody debris in the riverbed (negatively). This indicates that the β diversity of fish is highly affected by streams with greater thalweg depth, possibly due to the reduced amount of large woody structures entering streams within pasture areas. These streams also showed greater environmental heterogeneity due to the large variation in disturbance levels of this area, which renders their sites suitable for potential occupation by different species, making them high contributors (high LCBD) and also leading them to present high β diversity. On the other hand, areas with a higher percentage of forest at landscape scale (including preserved forest, reducedimpact logging and conventional logging) were the main contributors, while pasture areas had a higher percentage of exposed soil. We did not find any association between SCBD values and the habit of the species, as the taxa that contributed most to β diversity can be classified as having reduced niches (specialists) as well as broader niches (generalists). We conclude that modified areas may contribute substantially to β diversity because they have a distinct species combination, however different patterns can be observed at local and landscape scales.

1. Introduction Understanding the causes behind declines in biodiversity has been one of the most important undertakings of contemporary ecology (Gurevitch and Padilla, 2004; Wood et al., 2013; Dirzo et al., 2014). Consequently, ecologists have compiled information on the distribution of species at various locations to analyze and interpret the variation in taxonomic composition between sites with differing disturbances (Solar et al., 2015; Phillips et al., 2017). This variation in species composition, called beta diversity (β), has been widely assessed in ecological studies

⁎

with different approaches (Mac Nally et al., 2004; Gutiérrez-Cánovas et al., 2013; Knop, 2016), and can provide information on biodiversity loss and biological invasions (Karp et al., 2012; Marini et al., 2013; Socolar et al., 2016). Recently, Legendre and De Cáceres (2013) suggested the partition of β diversity into two components: local contribution to beta diversity (LCBD) and species contribution to beta diversity (SCBD). The LCBD consists of the relative contribution of each sampling site to the β diversity patterns of a region, revealing sites that have a unique species assemblage, i.e. possessing high richness and high conservation value,

Corresponding author. E-mail address: [email protected] (H. Leão).

https://doi.org/10.1016/j.ecolind.2019.106039 Received 3 July 2019; Received in revised form 20 December 2019; Accepted 23 December 2019 1470-160X/ © 2019 Elsevier Ltd. All rights reserved.

Ecological Indicators 111 (2020) 106039

H. Leão, et al.

being the most highly modified area. Firstly, we sought to assess the β diversity of fish within different areas of land use. Secondly, we identified the uses and taxa that most contribute to β diversity, and finally, we investigated which factors are most important in explaining the observed patterns of uniqueness using local and landscape scale predictors. Our first hypothesis is that β diversity should decrease correspondingly with the change in gradient, therefore streams located within areas of conventional logging and pasture should present the lowest values. In addition, the limited effect of reduced-impact logging on habitat structure and ecosystem processes (Miller et al., 2011; Calvão et al., 2016; Prudente et al., 2017) should show a β diversity similar to areas of preserved forest. As for the contribution of streams, our hypothesis is that those which drain forest and/or pasture areas have a greater contribution to β diversity, as they represent the extremes of the modification gradient. This is based on the idea that streams within forest areas should present higher richness because they contain both rare and common species, whilst streams within pasture areas should contain an exceptionally low number of species, and possibly present “newcomers” or exotic species due to the new environment created. For the contribution of species, our hypothesis is that those with specialised habits contribute more significantly to β diversity, since they are possibly present in only a few localities and are likely abundant in these areas.

or low species richness, indicating degraded sites that require ecological restoration. This uniqueness may be related to environmental and spatial factors, disturbance and scale of the study (Silva and Hernández, 2014; Lopes et al., 2014; Sor et al., 2018). The SCBD shows the relative importance of each taxon in affecting β diversity. This index can be associated with several intrinsic characteristics of the species, such as niche position and range, degree of occupation, abundance and biological traits (Heino and Grönroos, 2017; Silva et al., 2018), which can be determined by their levels of tolerance along an environmental gradient. Ecological uniquness may be directly related to habitat quality, since preserved and disturbed sites are strong contributors to β diversity patterns (Legendre, 2014). However, the contribution of modified sites presents a worrying scenario. The rapid advance of anthropogenic activities and consequent decline of pristine forests has led to a reduction in habitat and microhabitat combinations exploited by species with different ecological needs (Schneider and Winemiller, 2008). These changes to the environment may favor species that have broad niche breadth (generalists) over those with reduced niches (specialists), modifying the species composition of these sites and resulting in communities dominated by tolerant taxa and with few or none sensitive ones (Batáry et al., 2007; Allard et al., 2016). This dynamic process can make a particular area more unique due to the low species richness in the community. For example, the conversion of forests into open areas reduces habitat availability and causes the elimination of rare or sensitive species, whilst favoring the maintenance of those more adapted to the new environment (Silva et al., 2014). Thus, some sites present high ecological uniqueness due to the low β diversity promoted by the loss of species. Some factors are important in determining β diversity and therefore can influence patterns of uniqueness. Among them, environmental heterogeneity is one of the key predictors (Ceschin et al., 2018), because it entails greater variability in environmental conditions and resources. By this means, heterogenous environments allow the colonization and coexistence of species with different ecological requirements (Hutchinson, 1957) and provide new opportunities for a greater number of species (Clarke et al., 2008), contributing positively to the high β diversity of a region (Astorga et al., 2014; Zorzal-Almeida et al., 2017). Another important factor in determining β diversity is the scale at which the different ecological processes act. For example, local scale variables can have a stronger influence on community assemblage than those at landscape scale, and vice versa (Zorzal-Almeida et al., 2017, Montag et al., 2018), affecting β diversity, and consequently, uniqueness (Heino et al., 2017). Although the assessment of ecological uniqueness is relatively new, many works have been developed in recent years using this approach, especially in freshwater environments (Simões et al., 2015; Vilmi et al., 2017; Tolonen et al., 2018). However, studies are still incipient in river ecosystems, which are among the most threatened by human activities (Dudgeon et al., 2006; Castello et al., 2013), particularly through landuse practices that involve resource extraction and services (Allan, 2004). In addition, this approach remains poorly explored with fish (e.g. Bourassa et al., 2017; Arantes et al., 2017), and our work contributes to an even smaller number of studies evaluating factors that contribute to the β diversity of ichthyofauna in modified streams and their responses to environmental gradients (Bojsen and Barriga, 2002; Göthe et al., 2015; Edge et al., 2017). Our objective in this study was to assess the ecological uniqueness of fish communities from streams in modified areas of the Amazon. The sampled streams are located along a gradient of intense resource extraction dominated by different land uses, including reduced-impact logging (i.e., pre- and post-management planning), conventional logging (i.e., unmanaged) and pasture surrounded by forest fragments. We assumed the existence of an anthropogenic modification gradient, where forest areas were considered the most preserved, followed by reduced-impact logging, conventional logging and pasture, the latter

2. Material and methods 2.1. Study area Our study was carried out in the Capim river basin, situated in the northeastern part of the state of Pará, Eastern Amazonia (Fig. 1). The region has a sub-montane dense ombrophilous vegetation type (Veloso et al., 1991) and is described as Tropical Wet “Af” according to the Köppen Climate Classification System (Peel et al., 2007). The annual average temperature in the region is 27.2 °C, with annual average rainfall of approximately 1800 mm (Watrin and Da Rocha, 1992). The region is formed by sedimentary plateaus of low altitude, in addition to floodplains that accompany waterways, which are more emphasised along the lower Capim river (Monteiro et al., 2009). The region stands out due to the intense process of land occupation, with a long history of land-use modification (Almeida and Uhl, 1998). Among the primary economic activities are timber extraction (Tritsch et al., 2016) and cattle farming (Barona et al., 2010). Timber extraction is one of the oldest local activities in the region and has long been conducted without any planning management or assessment of forest resilience. Currently, it is carried out primarily by companies who utilize techniques which aim to mitigate impacts on the environment (Putz et al., 2012). Furthermore, the expansion of pasture areas for cattle farming has become a major economic force within the region, contributing to a significant increase in deforestation rates (Barona et al., 2010). As a result, the primary vegetation index continues to decrease and deforestation can be characterized by a mosaic of activities in the region (Leal et al., 2016). 2.2. Sample design Collections were carried out between August and October, coinciding with the region’s dry season, from 2012 to 2015. The dry season was chosen as this allowed for greater efficiency in the collection of fish, as well as to avoid seasonal variation (Jaramillo-Villa and Caramaschi, 2008; Prudente et al., 2017). We sampled 58 streams distributed as follows: 13 in preserved forest (2012–2013), 21 in reduced-impact logging areas (2012), 10 in conventional logging areas (2014) and 14 in pasture areas (2015). For the measurement of local scale environmental variables and collection of organisms, a section of 150 m was delimited in each stream, subdivided into ten segments (15 m), and separated by 11 transverse transects labeled ‘A’ to ‘K’ in a 2

Ecological Indicators 111 (2020) 106039

H. Leão, et al.

Fig. 1. Location of 58 sampled streams in areas with different land uses practices in the Capim river basin, Eastern Amazonia.

Association, 2013), fixed in 10% formalin solution and transferred to 70% alcohol after 48 h. Specimens were identified to the lowest possible taxonomic level using specialized literature and expert assistance. Collection was carried out under license number 4681-1 granted by the Sistema de Autorização e Informação em Biodiversidade – SISBIO. Specimens were deposited in the Ichthyology Collection of the Museu Paraense Emílio Goeldi (MPEG) in Belém (PA). The collection of biological material was conducted in accordance with the Comissão de Ética de Uso de Animais – CEUA No. 8293020418 (ID 000954), as well as Conselho Nacional de Controle de Experimentação Animal (CONCEA) guidelines.

upstream direction (Fig. S1 in Supplementary Material). 2.3. Fish sampling The collection of fish was carried out by two collectors using 55 cm diameter hand nets with 2 mm mesh between opposing nodes. For further comparison of the streams we established a sampling effort of 18 min per segment, totaling three hours of collection. This method of sampling is considered efficient in evaluating fish assemblages in loworder streams (Uieda and Castro, 1999) and has been widely used to assess patterns of ichthyofauna in Amazonian streams (e.g. Prudente et al., 2017; Ferreira et al., 2018). Collected fish were euthanized with a lethal dose of anesthetic (Eugenol; American Veterinary Medical 3

Ecological Indicators 111 (2020) 106039

H. Leão, et al.

targets in the images, and converts the pixel values to reflectance. This processing was performed in PCI Geomatics V10.1 software (Hill, 2007) using the ATCOR Ground Reflectance module. After atmospheric correction, the PCI Geomatics OrtoEngine module was used to create a mosaic of the images. The REIS mosaic was subjected to object oriented classification using Ecognition 9 software (Definiens, 2009). Validation of the classification was carried out with images of the TerraClass Project provided by the National Institute of Space Research (Almeida et al., 2016). We subsequently calculated the Kappa index, which reflects the quality of the classification, according to Landis and Koch (1977) and Piroli (2010). Land use and cover was classified into: (i) forest, comprising areas occupied by dense ombrophilous forest at different stages of development; (ii) fallow, resulting from natural processes of succession after total or partial suppression of the primary vegetation by anthropogenic actions or natural causes; (iii) agriculture, areas with mechanized or family farming, use of chemical products, fertilizers and the presence of herbaceous plants, in addition to a homogenization observed by satellite images; (iv) pasture, with areas occupied by intensive and/or extensive livestock farming; and (v) exposed soil, with areas devoid of vegetation and severe degradation, including urban areas with networks of dirt roads and highways. The dimensions of each class of land use and cover were quantified in km2 throughout the microcatchments and subsequently converted into a percentage (%), which constituted our predictive landscape scale variables (Table 1).

Table 1 Descriptive statistics of potential predictor variables. Variables in bold are those selected by Akaike’s information criterion (AIC). Landscape variables were used in a Principal Component Analysis (PCA) and the first axis (PCA1) used as predictor variable in the linear models. LWD = large woody debris; SD = standard deviation; Min = minimum; Max = maximum. Variables

Mean

SD

Min

Max

Local Thalweg depth (cm) Bank angle (degrees) Heterogeneity of water flow Substrate embedded in fine sediments (%) Leaf litter (%) Natural shelter Number of LWD in the riverbed Volume of LWD in the riverbed Exposed soil Canopy cover Mid-layer cover Ground-layer cover

27.67 25.72 0.13 56.33 31.71 104.16 21.07 2.89 2.58 49.37 54.84 40.26

12.93 11.17 0.07 14.95 16.36 36.83 21.01 3.72 5.36 22.42 19.83 16.64

9.36 8.14 0.00 19.89 0.00 22.50 0.00 0.00 0.00 0.23 3.86 16.14

58.75 56.14 0.31 85.82 61.54 204.32 150.67 25.21 27.84 91.59 109.20 93.52

Landscape Forest (%) Fallow (%) Agriculture (%) Pasture (%) Exposed soil (%) Environmental heterogeneity

79.28 3.96 1.74 8.70 4.75 2.79

30.95 8.30 4.22 14.95 9.13 1.26

9.19 0.00 0.00 0.00 0.00 0.91

100.00 39.91 20.49 60.10 36.08 7.12

2.4. Local scale environmental variables

2.6. Data analysis

Local scale environmental variables were measured in accordance with US Environmental Protection Agency (EMAP/US-EPA) stream assessment guidelines (Peck et al., 2006), adapted for tropical streams by Callisto et al. (2014). This guideline provides a broad set of raw data that allows the calculation of several variables (Kaufmann et al., 1999), among which we selected those potentially capable of predicting impacts in tropical streams based on several studies carried out in the Amazon region (e.g. Calvão et al., 2016; Juen et al., 2016; Prudente et al., 2017; Ferreira et al., 2018). These variables represent different habitat characteristics of streams, such as channel morphology (thalweg depth and bank angle), hydraulic (heterogeneity of water flow), substrate (embedded in fine sediment and leaf litter), fish shelter (natural shelter), woody debris (number and volume of large woody debris in the riverbed) and riparian vegetation (exposed soil, canopy, mid-layer and ground-layer cover) (Table 1). A brief description of the collection and calculation of these variables can be found in Supplementary Material.

2.6.1. Calculation of LCBD and SCBD A schematic diagram showing statistical analyses can be found in Supplementary Material (Fig. S2). β diversity was calculated in accordance with Legendre and De Cáceres (2013). Initially, the abundance and presence/absence data of the species were subjected to a Hellinger transformation, an adequate index for analyzing β diversity. In this case, our goal was to assess whether there are differences between quantitative and qualitative data. Next, we calculated the total β diversity for each land use and cover class. The generated index ranges from 0 to 1, where 1 indicates maximum dissimilarity and 0 indicates maximum similarity between sites. To test for differences in β diversity between land uses we used a One-way Analysis of Varience (ANOVA) followed by Tukey’s test (p < 0.05) for multiple comparisons, with the assumptions of normality and homogeneity of the variances being fulfilled. In addition, we calculated the contribution of streams and species to total β diversity, both for abundance data (hereafter LCBDab and SCBDab) and for presence/absence data (hereafter LCBDp/a and SCBDp/ a). High values of LCBD indicate sites with high ecological uniqueness, that is, they present quite different species composition from the others and, therefore, contribute more to β diversity (Legendre and De Cáceres, 2013). Streams with above-average LCBD values were considered the largest contributors (Mimouni et al., 2015). SCBD represents the relative importance of each taxon in influencing patterns of β diversity, signaling those species that present high variation between sites in the study area or that are abundant in the few places where they occur (Legendre and De Cáceres, 2013). Thus, species present in all communities have zero value in their contribution. Taxa with higher than average SCBD values were also considered to be major contributors to β diversity (Sor et al., 2018). In both cases (LCBD and SCBD), the sum of the indices should be equal to 1, since they represent a relative contribution. To test for differences in stream contribution (LCBDab and LCBDp/a) between the land uses we used a non-parametric Kruskal-Wallis test (p < 0.05), as the data did not fulfill the assumptions of a parametric test. We then made multiple comparisons using an associated posteriori test (Dunn, 1964). Considering that LCBD may be correlated to the species richness and abundance of a community (Heino and Grönroos,

2.5. Classification of land use and cover (landscape scale variables) Characterization of land use and cover was performed in the 58 microcatchments using several geoprocessing programs. Through the ArcGis 10.1 program (ESRI, 2014) we delineated the microcatchments upstream of the sampling section. The catchment was extracted from Shuttle Radar Topograph Mission (SRTM) Digital Elevation Model data with spatial resolution of 30 m, obtained free of charge from the United States Geological Survey (https://earthexplorer.usgs.gov/). The classes of land use and cover were identified through Digital Image Processing of the RapidEye Earth Imaging System (REIS) optical sensor. REIS images are for commercial use, however the data were acquired for research purposes free of charge from the Ministry of Environment at http://geocatalogo.mma.gov.br/. Selected images corresponded with time of ichthyofauna sampling and characterization of local scale variables. The REIS images were acquired, orthorectified and projected on the WGS 84 geodetic DATUM. The images were subject to atmospheric correction, a process that attenuates the effects of the atmosphere on the spectral response of the 4

Ecological Indicators 111 (2020) 106039

H. Leão, et al.

2.6.6. Generalized linear models (GLM) In order to identify the predictor variables which primarily explain stream uniqueness (LCBDab and LCBDp/a), we used Generalized Linear Models (GLMs) with Gaussian distribution (Zuur et al., 2009). In constructing the models we included the following predictor variables: environmental heterogeneity (using distance to the centroid), standardized local and landscape scale environmental variables (PCA1). In this complete model, we used a stepwise procedure based on the Akaike Information Criterion (AIC) in order to obtain the model of best fit (Akaike, 1981). The selected model was the one with the lowest AIC value. No correction method was required in the models as no overdispersion of the data was detected.

2017), we performed a Spearman correlation in order to verify any relationship (for LCBDab and LCBDp/a). Thus, it was possible to verify if the land uses that contributed the most to β diversity (higher values of LCBD) were more or less rich and abundant. We also correlated SCBD (SCBDab and SCBDp/a) with species abundance and the number of sites occupied by them, since these characteristics have also been shown to be correlated with species contribution (Heino and Grönroos, 2017). 2.6.2. Niche range of species In order to classify species as habitat specialists or generalists, we used the outlying mean index (OMI) (Dolédec et al., 2000). This analysis calculates the niche breadth (or tolerance) of each species through the relationship between their abundance and the environmental variables measured. High tolerance values indicate taxa which occur in a wide range of environmental conditions (generalists), while low values indicate taxa distribution over limited environmental conditions (specialists). We performed a simple regression (p < 0.05) to assess whether the contribution of species to beta diversity (SCBD) is influenced by their niche range (species tolerance).

2.6.7. R statistical packages All analyzes were performed in the R program (R Core Team, 2016). The β diversity and LCBD and SCBD indices were computed using the function “beta.div” available in the adespatial package (Dray et al., 2017). The niche breadth of species was calculated using the function “niche” of the ade4 package (Dray and Dufour, 2018). The distances used as measures of heterogeneity and eigenvectors (PCNM) were obtained through the “betadisper” and “pcnm” functions, respectively, both available in the vegan package (Oksanen et al., 2011). The selection of eigenvectors was performed through the function “forward.sel” from the packfor package (Dray et al., 2016). The VIF analysis was performed using the function “vif” in the car package (Fox et al., 2015). The ANOVA was performed with the function “aov” and the GLM was adjusted using the function “glm”, both from the stats package. Finally, to select the best AIC-based model we used the function “stepAIC” in the MASS package (Ripley et al., 2015).

2.6.3. Spatial autocorrelation analysis We examined the potential influence of spatial autocorrelation on LCBD values for subsequent linear model construction. Using the geographic coordinates of the sampling sites, we constructed a fluvial distance matrix between all pairs of streams to obtain spatial filters through the Principal Coordinates of Neighboring Matrices method (PCNM; Dray et al., 2006). The fluvial distance provides a better representation of spatial patterns generated by the dispersion of fish along a dendritic network, differing from the patterns observed in linear distances for these organisms (Landeiro et al., 2011). The distances between streams were calculated using the ArcGIS Network Analysis extension (ESRI, 2014). The selection of eigenvectors used to test autocorrelation was done through the forward-selection method, which selected four eigenvectors (Table S1). As we observed spatial autocorrelation for LCBD, we extracted the residual data for the construction of the models (Borcard et al., 2018).

3. Results The total β diversity was 0.456 for abundance data and 0.484 for presence/absence data, with significant difference between the land use and cover classes (ANOVA: βab: F(3.54) = 3.599, p = 0.019; βp/a: F(3.54) = 3.182, p = 0.031). The pasture area presented the highest β diversity (βab = 0.400; βp/a = 0.467), while forest (βab = 0.268; βp/ a = 0.348), reduced-impact logging (βab = 0.291; βp/a = 0.395), and conventional logging (βab = 0.231; βp/a = 0.354) presented lower values and a similar pattern of β diversity (Fig. 2a and b). Based on the abundance data, the peer-to-peer comparison showed that only pasture and conventional logging were significantly different (p = 0.022). These land uses also presented higher and lower environmental heterogeneity (Fig. 3a and b), respectively. When considering species occurrence, β diversity was different only between forest and pasture areas (p = 0.037). Finally, we did not find any difference between forest and reduced-impact logging, for both abundance data (p = 0.966) and presence/absence data (p = 0.635) (Fig. 2a and b) (for other comparisons, see Table S3). These two treatments also presented a similar pattern of environmental heterogeneity (Fig. 3a and b). LCBD values ranged from 0.007 to 0.050 for abundance data, with 18 streams contributing above average (0.017), and between 0.010 and 0.042 for presence/absence data, with 22 streams presenting a larger contribution (Table S4). We found a difference in uniqueness between land use and cover (Kruskal-Wallis; LCBDab: p = 0.004; LCBDp/a: p = 0.009), and the highest values were also observed for pasture areas, making this land use the largest contributor to β diversity (Fig. 2c and d). The peer-to-peer comparison showed that pasture was different to all other land uses for abundance data (forest: p = 0.001, reducedimpact logging: p = 0.002, conventional logging: p = 0.040), and presence/absence (forest: p = 0.016, reduced-impact logging: p = 0.001, conventional logging: p = 0.025) (for other comparisons, see Table S4). Forest streams contributed little to the β ß diversity of ichthyofauna (low LCBD values), presenting a similar pattern of contribution to reduced-impact logging and conventional logging areas (Fig. 2c and d).

2.6.4. Environmental heterogeneity The local scale environmental variables were standardized and used in a Multivariate Dispersion Analysis (PERMDISP; Anderson, 2006) to evaluate the heterogeneity of the streams, one of our predictive variables. The mean distances of each site to the centroid of its group (land use and cover classes) were used as a measure of environmental heterogeneity. Land uses with high heterogeneity were represented by streams with the largest distances. Additionally, we created an ordination plot through a Principal Coordinate Analysis (PCoA) using a Euclidean distance matrix to visualize the heterogeneity of land use in two-dimensional space. 2.6.5. Multicollinearity analysis We used the Variance Influence Factor (VIF) to verify multicollinearity between all the predictive variable candidates, being: land use and cover (forest, reduced-impact logging, conventional logging and pasture), environmental heterogeneity (using distance to the centroid), local and landscape scale environmental variables. Those that presented a VIF ≥ 10 were considered strongly correlated (Curto and Pinto, 2010). In this step, we excluded land use and cover (VIF = 89.15) and applied a Principal Component Analysis (PCA; Legendre and Legendre, 2012) with the landscape scale variables, among which a strong correlation was identified. We used the first axis (PCA1) as a predictor variable, which was selected as it presented an eigenvalue greater than that expected through the broken-stick criterion (Jackson, 1993), explaining 68% of the variation in the data (Table S2). 5

Ecological Indicators 111 (2020) 106039

H. Leão, et al.

Fig. 2. Difference in β diversity (2a and 2b) and local contribution to beta diversity (LCBD) (2c and 2d) among the land use classes based on abundance and presence/absence data of species. FOR = forest; RIL = reduced-impact logging; CVL = conventional logging; PAS = pasture. Different letters indicate significant differences among the land use classes (p < 0.05).

LCBDab did not present any relationship with species richness (r = −0.081; p = 0.548) and showed a weak negative correlation with fish abundance (r = −0.261; p = 0.048). This indicates that in general, streams that present high uniqueness in their composition can also have low abundance, a condition observed primarily in the pasture areas. On the other hand, LCBDp/a presented a weak negative relationship with species richness (r = −0.352; p = 0.006), but did not

show any relationship with abundance (r = 0.032; p = 0.810). In this case, streams that present high ecological uniqueness have low species richness, but in this instance, sites are included in the different categories of land use and cover. Regarding analysis of the community, 21,123 individuals belonging to 80 species were captured (Table S5). Maximum SCBD values were found for abundance data, and 22 species presented values higher than

Fig. 3. Principal Coordinate Analysis (PCoA) showing environmental heterogeneity among the land use classes. FOR = forest; RIL = reduced-impact logging; CVL = conventional logging; PAS = pasture. 6

Ecological Indicators 111 (2020) 106039

H. Leão, et al.

4. Discussion

the mean (0.012). Hyphessobrycon heterorhabdus (Ulrey, 1894) (0.145), Iguanodectes rachovii Regan, 1912 (0.102) and Apistogramma gr. regani (0.079) were the three largest contributors to β diversity. For presence/ absence data, 32 species presented higher than average SCBD values (0.012), with Brachyhypopomus sp. 2 (0.033), Iguanodectes rachovii (0.033) and Microcharacidium weitzmani Buckup, 1993 (0.032) possessing the highest values (Table S6). The niche range did not explain the contribution of species to abundance data (SCBDab: R2 = 0.146, F(1.78) = 1.505, p = 0.224), and although we found a significant relationship with presence/absence data, the model explanation was very low (SCBDp/a: R2 = 0,019, F(1,78) = 13,38, p < 0,001). Thus, both specialist and generalist species contributed to β diversity, refuting our hypothesis that the greatest contribution would be from specialist species. The above-mentioned species occurred within all land uses, with the exception of Brachyhypopomus sp. 2, which was the only species absent in pasture areas and present in all other land uses. Both SCBDab (r = 0.866; p = < 0.001), and SCBDp/a (r = 0.913; p = < 0.001) showed a high correlation with the number of sites occupied by the species. Rare species with low occurrence and abundance, such as Hoplerythrinus unitaeniatus (Spix & Agassiz, 1829), Acanthodoras cataphractus (Linnaeus, 1758), Megalechis picta (Müller & Troschel, 1849) and Tetranematichthys wallacei Vari & Ferraris, 2006 presented extremely low contributions to β diversity (≤0.0001). All predictive environmental variables potentially capable of influencing the contribution of streams (LCBD) are described in Table 1. The model of best fit obtained through the AIC for LCBDab included seven predictor variables (AIC = −445.3), among which four were significant in explaining the contribution of streams (R2-adjusted = 0.412) (Table 2). LCBDab was positively related to the environmental heterogeneity of land uses, depth of talvegue and PCA1, and negatively related to the volume of woody debris in the riverbed. The first PCA axis used as a predictive variable was positively correlated with the percentage of primary vegetation, responsible for most of the data variation (68%), and presented a negative relationship with the percentage of pasture and exposed soil. The forest, reduced-impact logging and conventional logging areas together comprise the highest percentage of forest and the sum of their contributions exceeds the contribution of pasture alone at this scale. For LCBDp/a the model with the lowest AIC includes only two variables (AIC = −481.3), both important in explaining the contribution of the streams, however the explanation of the model was very low (R2-adjusted = 0.136), with environmental heterogeneity presenting a positive relationship, and the volume of large woody debris in the riverbed a negative relationship with the response variable (Table 2).

The results do not support our hypothesis that land uses with a higher degree of disturbance would have a negative effect on β diversity. The effects of anthropogenic disturbances on community assemblage can be quite variable and depend on a number of factors, such as the initial ecological conditions of the site, the magnitude, type and uniformity of modification to the environment, as well as the individual sensitivity of each taxa to the disturbance (Hawkins et al., 2015). In this case, we believe that these factors may have contributed to the high β diversity of ichthyofauna in more modified areas. Although streams within pasture areas are all modified, they can be characterized by variation in the level of environmental modification, where the effects of disturbance in this land use are not uniform and, consequently, increase the environmental and biological heterogeneity among the sites. Streams may undergo a variety of land use effects while flowing through the landscape and such effects cannot necessarily be directly connected to any single land use type (Tóth et al., 2019). Therefore, we would like to emphasise that β diversity can increase, decrease or remain unchanged in the face of anthropogenic impacts, depending on the processes involved in the making of similar (biotic homogenization) or dissimilar (biotic heterogenization) species composition between different sites (Socolar et al., 2016). For example, modified streams (such as those located in pasture areas) may exhibit different levels of degradation or may have different levels of riparian and within-stream habitat structure. This can introduce unexpected variability among stream communities, which means that high beta diversity may not always indicate a reference environmental condition and low beta diversity may not necessarily indicate a degraded environmental condition (Dala-Corte et al., 2019). Although we did not observe expected patterns for streams within forest (low β) and pasture (high β) areas, we did find low β diversity in the conventional logging areas as expected for more modified areas. Most studies commonly find a negative relationship between β diversity and anthropogenic impacts for several taxonomic groups (Bojsen and Barriga, 2002; Ekroos et al., 2010; Solar et al., 2015; Knop, 2016). However, in some cases β diversity may increase in response to such disturbance (Flohre et al., 2011; Göthe et al., 2015; Hawkins et al., 2015; Fugère et al., 2016). This occurs because the changes can cause convergence in community composition (low β) by increasing the niche selection of disturbance tolerant species, but can also cause divergence (high β) through the increase of filters capable of selecting different species along environmental gradients (Myers et al., 2015). Areas of forest and reduced-impact logging did not show any difference in β diversity, supporting our hypothesis of similarity between both. Many studies have demonstrated that the application of managed timber harvesting techniques, such as reduced-impact logging, considerably reduces residual damage in forests where these practices are carried out (Miller et al., 2011; Edwards et al., 2014), causing less damage to biodiversity in relation to other land-use activities (Allard et al., 2016; Calvão et al., 2016). We also did not observe any differences in environmental heterogeneity between these areas and those of forest, possibly because they are similar in their structural habitat descriptors, thus not permitting any detection in changes to the β diversity of fish in managed areas. High stream contribution to the β diversity of ichthyofauna was also observed in pasture areas, while other land use and cover classes presenting a similar pattern of contribution. This indicates that the groups of streams in pasture areas are the most different in relation to the set of streams studied. Although streams in forest areas can also significantly contribute as predicted in our hypothesis, this was not observed in our results. Streams within areas of preserved forest share many similar characteristics, which possibly favor a similar set of species as well. On the other hand, streams within pasture areas are surrounded by forest fragments with distinct characteristics and densities varying in their levels of disturbance. It is important to note that this heterogeneity is

Table 2 Generalized Linear Model (GLM) results for LCBDab and LCBDp/a. PCA1 represents the first axis of Principal Component Analysis (PCA) performed with landscape variables. LWD = large woody debris. Variables

Estimate

SE

t

p

LCBDab (Intercept) Environmental heterogeneity Thalweg depth (cm) Bank angle (degrees) Ground-layer cover Number of LWD in the riverbed Volume of LWD in the riverbed PCA1

0.010 0.003 0.003 −0.001 0.001 −0.001 −0.002 0.002

0.002 0.001 0.001 0.001 0.001 0.001 0.001 0.000

4.921 3.733 3.361 −1.915 1.424 −1.717 −2.051 4.667

0.000 0.000 0.002 0.061 0.161 0.092 0.046 0.000

LCBDp/a (Intercept) Environmental heterogeneity Volume of LWD in the riverbed

0.014 0.001 −0.002

0.001 0.000 0.001

10.713 2.789 −2.907

0.000 0.007 0.005

7

Ecological Indicators 111 (2020) 106039

H. Leão, et al.

tolerances can generate different patterns of spatial distribution (Leibold et al., 2004), as these environments provide a greater niche breadth for the species (Tews et al., 2004; Astorga et al., 2014), including in areas subject to land-use practices (Weibull et al., 2000). Thus, although pasture areas are recognized as damaging to the environment (Fearnside and Barbosa, 1998; Casatti et al., 2006), the high variation in environmental characteristics of these areas in our data likely reflects the different stages of the deforestation process, promoting increased heterogeneity in this area. In addition, although streams within pasture areas present similar levels of deforestation, they may have different histories of land use alteration (e.g. time of first logging), and this timeline may be an important determinant of the diversity found today (Brejão et al., 2018), and may even be responsible for higher β diversity in the region. We also found a negative relationship between abundance of fish and LCBDab, as well as between species richness and LCBDp/a. This indicates that streams which presented a high uniqueness in their composition had a low abundance and richness of fish. In general, the streams located in pasture areas were the most unique and had the lowest abundance of fish, and those with the highest LCBD values showed relatively low richness. This suggests that streams within pasture areas may harbor a set of unique species, even if these species are tolerant or the sites present low richness, thus contributing strongly to β diversity of the region (Heino et al., 2017). Other studies also found a negative relationship between LCBD and the species abundance and richness of different taxonomic groups (Da Silva and Hernández, 2014; Heino and Grönroos, 2017; Landeiro et al., 2018). As emphasized by these authors, ecological uniqueness can not be considered synonymous with high species richness. We did not find any association between values of SCBD and the habit of the species, as the taxa that contributed most to β diversity can be classified as having reduced niches (specialists) as well as broader niches (generalists). However, we observed that some species which contributed most to SCBDab were the ones which contributed least to SCBDp/a and vice versa. For example, Hyphessobrycon heterorhabdus was the biggest contributor to SCBDab, but when considering only the occurrence of this species its contribution was below the general average. Similarly, Brachyhypopomus sp. 2 showed good contribution to SCBDp/a, but reduced contribution when considering species abundance. Among other examples, Pyrrhulina aff. brevis, Nannostomus nitidus Weitzman, 1978 and Synbranchus marmoratus Bloch, 1795 also presented a similar pattern. This confirms the effect of abundance on the degree to which species contribute to β-diversity (Legendre and De Cáceres, 2013). Hence, species that present high total abundances in the data should also show high variation in abundance between sites, and thus exhibit high contribution (Heino and Grönroos, 2017). Considering that higher values of SCBD indicate heterogeneous distribution of taxa throughout the sites, rare species with low occurrence and abundance presented extremely low contributions to β diversity. This may be a reflection of factors such as: individual characteristics of each species, dispersion abilities or even their synergistic interaction with abiotic variables (Siegloch et al., 2018), which does not necessarily reflect a specialized habit of the species. Moreover, the absence of these in almost all sites indicates homogenous distribution, in the same way for species which occur at all the sites. Therefore, strong contributors were primarily influenced by the number of occupied sites, regardless of the niche characteristics provided by land uses. We conclude that modified areas can contribute highly to β diversity as they have a distinct assemblage of species, but this is due to different levels of impact in these areas, and that different factors at local and catchment scale may be important in explaining this pattern. Although these areas may be major contributors to β diversity patterns at the local scale, different patterns can be observed at the landscape scale. In mixed landscapes, with mosaics composed of modified and pristine areas, the effect of distinct processes can contribute to the maintenance or increase of dissimilarity between communities (β). We

high within the treatment, i.e., between streams, but not within the stream section. Thus, the pasture area ends up presenting an environmental configuration of mixed characteristics and its streams can be potentially occupied by communities of different composition. For this reason, different disturbance regimes may make some sites large contributors (high LCBD) to total β diversity, leading them to also present high β diversity values (Arroyo-Rodríguez et al., 2013). Assessing the factors that influence LCBD is a complex task, as the processes that govern this contribution are difficult to predict, and evidence in the literature is still relatively scarce, especially for modified sites, since this metric is relatively new (Legendre and De Cáceres, 2013). However, some of this contribution can be predicted by the local environmental conditions, as observed in our results and also in other studies (Tonkin et al., 2016; Pajunen et al., 2017; Tolonen et al., 2018). This is justified because the variation in habitat characteristics provides for the establishment of different sets of species between localities. Since many species have distinct ranges of environmental tolerance and preference, an intrinsic variation occurs in species composition between sites (Leibold et al., 2004). Thus, high variation in the environmental characteristics between sites leads to greater uniqueness along a gradient’s edge, consequently increasing the number of unique species occurring at these sites (Pajunen et al., 2017). However, although these taxa are rare regionally, they may be opportunistic and tolerant to variations. For example, Astyanax bimaculatus (Linnaeus, 1758) and Nannostomus beckfordi Günther, 1872 were found only in one and two streams, respectively, that drain pasture areas, but are known to have strong opportunistic habits (Andrian et al., 2001; Carvalho et al., 2009). Among the local scale variables, two were important in explaining the contribution of streams. Thalweg depth presented a positive relationship with LCBDab, while the volume of large woody debris presented a negative relationship with LCBDab and LCBDp/a. This indicates that β diversity received a higher contribution from streams with greater thalweg depth and less large woody debris in the riverbed. Many studies have discussed the importance of riparian vegetation in the provision of structural components in streams, such as large woody debris (e.g. De Paula et al., 2011; Larson et al., 2018). In areas with more preserved riparian vegetation, higher input of trunks and branches into the stream occurs due to falling trees, landslides and other slope processes. The removal of this vegetation causes a reduction in the input of these structures into the stream, which results, among other consequences, in increased thalweg depth (Dias and Thomaz, 2011), a condition that can be particularly associated with the pasture areas in our study. Considering that large woody debris also provides shelter for fish (Crook and Robertson, 1999), its reduction considerably affects the composition and characteristics of the assemblages (Howson et al., 2012). Therefore, the greater uniqueness observed in pasture areas may be the result of changes in riparian vegetation, and consequently, in the stream dynamics. Land use and cover within the catchment (PCA1) were important in explaining LCBDab. Using this ordination axis as a predictor variable, these results indicate an increase in LCBD corresponding with increased forest area, with this variable responsible for most of the variation in the data. In addition, the results indicate a reduction in LCBD corresponding with increased pasture and exposed soil within the catchment. Although these results seem to contradict that which was initially exposed, since the streams that drain pasture areas showed greater contribution, they can be explained by the presence of forests in many areas of the catchments. That is, the sum of the contributions of forest, reduced-impact logging and conventional logging areas outweighs the contribution of pasture areas alone, as together they possess the highest percentage of forest in the catchment. Environmental heterogeneity was important in explaining both LCBDab and LCBDp/a, where a positive relation was observed. This indicates that the greatest contribution of streams within pasture areas should be related to the high environmental heterogeneity observed in this area. In heterogeneous environments, species with different 8

Ecological Indicators 111 (2020) 106039

H. Leão, et al.

suggest the development of monitoring programs that use new biodiversity assessment tools, such as LCBD and SCBD, coupled with commonly used traditional tools such as species richness, abundance and composition, or other promising tools such as functional beta diversity. We believe that this approach, by including different assessment metrics and therefore being more comprehensive, will provide more complete answers that will help identify priority areas for both conservation and recovery in regions with high biodiversity and increasing deforestation, as is the case of the Amazon.

Arantes, C.C., Winemiller, K.O., Petrere, M., Castello, L., Freitas, C.E., Hess, L.L., 2017. Relationships between forest cover and fish diversity in the Amazon River floodplain. J. Appl. Ecol. 55, 386–395. https://doi.org/10.1111/1365-2664.12967. Arroyo-Rodríguez, V., Rös, M., Escobar, F., Melo, F.P., Santos, B.A., Tabarelli, M., Chazdon, R., 2013. Plant β-diversity in fragmented rain forests: testing floristic homogenization and differentiation hypotheses. J. Ecol. 101, 1449–1458. https:// doi.org/10.1111/1365-2745.12153. Astorga, A., Death, R., Death, F., Paavola, R., Chakraborty, M., Muotka, T., 2014. Habitat heterogeneity drives the geographical distribution of beta diversity: the case of New Zealand stream invertebrates. Ecol. Evol. 4, 2693–2702. https://doi.org/10.1002/ ece3.1124. Barona, E., Ramankutty, N., Hyman, G., Coomes, O.T., 2010. The role of pasture and soybean in deforestation of the Brazilian Amazon. Environ. Res. Lett. 5, 024002. https://doi.org/10.1088/1748-9326/5/2/024002. Batáry, P., Báldi, A., Szél, G., Podlussány, A., Rozner, I., Erdõs, S., 2007. Responses of grassland specialist and generalist beetles to management and landscape complexity. Divers. Distrib. 13, 196–202. https://doi.org/10.1111/j.1472-4642.2006.00309.x. Bojsen, B.H., Barriga, R., 2002. Effects of deforestation on fish community structure in Ecuadorian Amazon streams. Freshw. Biol. 47, 2246–2260. https://doi.org/10.1046/ j.1365-2427.2002.00956.x. Borcard, D., Gillet, F., Legendre, P., 2018. Numerical Ecology with R. Springer. Bourassa, A.L., Fraser, L., Beisner, B.E., 2017. Benthic macroinvertebrate and fish metacommunity structure in temperate urban streams. J. Urban Ecol. 3, 1–14. https://doi. org/10.1093/jue/jux012. Brejão, G.L., Hoeinghaus, D.J., Pérez-Mayorga, M.A., Ferraz, S.F., Casatti, L., 2018. Threshold responses of Amazonian stream fishes to timing and extent of deforestation. Conserv. Biol. 32, 860–871. https://doi.org/10.1111/cobi.13061. Callisto, M., Alves, C.B.M., Lopes, J.M., Castro, M.A., 2014. Condições ecológicas em bacias hidrográficas de empreendimentos hidroelétricos. Cemig, Belo Horizonte. Calvão, L.B., Nogueira, D.S., Montag, L.F.A., Lopes, M.A., Juen, L., 2016. Are Odonata communities impacted by conventional or reduced impact logging? For. Ecol. Manage. 382, 143–150. https://doi.org/10.1016/j.foreco.2016.10.013. Carvalho, L.N., Zuanon, J., Sazima, I., 2009. Natural history of Amazon fishes. In: Del Claro, K., Oliveira, P.S., Rico-Gray, V., Ramirez, A., Barbosa, A.A.A., Bonet, A., Scarano, F.R., Consoli, F.L., Garzon, F.J.M., Nakajima, J.N., Costello, J.A., Vinicius, M. (Eds.), Encyclopedia of Life Support System. Eolss Publishers & UNESCO, Oxford, pp. 1–24. Casatti, L., Langeani, F., Silva, A.M., Castro, R.M.C., 2006. Stream fish, water and habitat quality in a pasture dominated basin, southeastern Brazil. Braz. J. Biol. 66, 681–696. https://doi.org/10.1590/S1519-69842006000400012. Castello, L., McGrath, D.G., Hess, L.L., Coe, M.T., Lefebvre, P.A., Petry, P., Macedo, M.N., et al., 2013. The vulnerability of Amazon freshwater ecosystems. Conserv. Lett. 6, 217–229. https://doi.org/10.1111/conl.12008. Ceschin, F., Bini, L.M., Padial, A.A., 2018. Correlates of fish and aquatic macrophyte beta diversity in the Upper Paraná River floodplain. Hydrobiologia 805, 377–389. https:// doi.org/10.1007/s10750-017-3325-x. Clarke, A., Mac Nally, R., Bond, N., Lake, P.S., 2008. Macroinvertebrate diversity in headwater streams: a review. Freshw. Biol. 53, 1707–1721. https://doi.org/10.1111/ j.1365-2427.2008.02041.x. Crook, D.A., Robertson, A.I., 1999. Relationships between riverine fish and woody debris: implications for lowland rivers. Mar. Freshw. Res. 50, 941–953. https://doi.org/10. 1071/MF99072. Curto, J.D., Pinto, J.C., 2010. The corrected VIF (CVIF). J. Appl. Stat. 38, 1499–1507. https://doi.org/10.1080/02664763.2010.505956. Dala-Corte, R.B., Sgarbi, L.F., Becker, F.G., Melo, A.S., 2019. Beta diversity of stream fish communities along anthropogenic environmental gradients at multiple spatial scales. Environ. Monit. Assess. 191, 288. https://doi.org/10.1007/s10661-019-7448-6. Definiens, A.G., 2009. Developer User Guide. Definiens AG, Munich, Germany. De Paula, F.R., de Barros Ferraz, S.F., Gerhard, P., Vettorazzi, C.A., Ferreira, A., 2011. Large woody debris input and its influence on channel structure in agricultural lands of Southeast Brazil. Environ. Manage. 48, 750. https://doi.org/10.1007/s00267-0119730-4. Dolédec, S., Chessel, D., Gimaret-Carpentier, C., 2000. Niche separation in community analysis: a new method. Ecology 81, 2914–2927. https://doi.org/10.1890/00129658(2000) 081[2914:NSICAA]2.0.CO;2. Dias, W.A., Thomaz, E.L., 2011. Influência de escombros lenhosos na dinâmica de riachos em área de faxinal. Terr@Plural 5, 229–248. https://doi.org/10.5212/TerraPlural.v. 5i2.0006. Dirzo, R., Young, H.S., Galetti, M., Ceballos, G., Isaac, N.J., Collen, B., 2014. Defaunation in the Anthropocene. Science 345, 401–406. https://doi.org/10.1126/science. 1251817. Dray, S., Legendre, P., Peres-Neto, P., 2006. Spatial modelling: a comprehensive framework for principal coordinate analysis of neighbour matrices (PCNM). Ecol. Model. 196, 483–493. https://doi.org/10.1016/j.ecolmodel.2006.02.015. Dray, S., Legendre, P., Blanchet, G., 2016. packfor: Forward Selection with permutation (Canoco p.46). R package version 0.0-8/r136. https://R-Forge.R-project.org/ projects/sedar/. Dray, S., Dufour, A.-B., 2018. Analysis of Ecological Data: Exploratory and Euclidean Methods in Environmental Sciences. R package version 1.7-13. https://cran.r-project. org/web/packages/ade4/ade4.pdf. Dray, A.S., Blanchet, G., Borcard, D., Clappe, S., Guenard, G., Jombart, T., Larocque, G., Legendre, P., Madi, N., Wagner, H.H., 2017. adespatial: Multivariate Multiscale Spatial Analysis. R package version 0.0-9. https://CRAN.R-project.org/package= adespatial. Dudgeon, D., Arthington, A.H., Gessner, M.O., Kawabata, Z., Knowler, D.J., Lévêque, C., Naiman, R.J., et al., 2006. Freshwater biodiversity: Importance, threats, status and

CRediT authorship contribution statement Híngara Leão: Conceptualization, Methodology, Validation, Formal analysis, Investigation, Data curation, Writing - original draft, Project administration. Tadeu Siqueira: Conceptualization, Methodology, Writing - review & editing. Naiara Raiol Torres: Formal analysis, Writing - review & editing. Luciano Fogaça de Assis Montag: Conceptualization, Methodology, Resources, Writing - review & editing, Supervision, Funding acquisition. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements We would like to thank 33 Forest Capital, Cikel Ltda., Instituto Floresta Tropical (IFT) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq; project 449314/2014-2) for providing financial and logistic support. We thank the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) and Empresa Brasileira de Pesquisa Agropecuária (EMBRAPA) (process 88882.157125/2017-01) for granting scholarships grants for HL. We also thank the Programa Nacional de Cooperação Acadêmica (PROCAD) of the CAPES for grants provided to HL (process 88881.068425/2014-01) during the development of this paper in academic cooperation between the Universidade Federal do Pará (UFPA) and Universidade Estadual Paulista (UNESP). Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.ecolind.2019.106039. References Akaike, H., 1981. Likelihood of a model and information criterion. J. Econom. 16, 3–14. https://doi.org/10.1016/0304-4076(81)90071-3. Andrian, I.F., Silva, H.B.R., Peretti, D., 2001. Dieta de Astyanax bimaculatus (Linnaeus, 1758) (Characiformes, Characidae), da área de influência do reservatório de Corumbá, Estado de Goiás, Brasil. Acta sci. Biol. sci. 23, 435–440. Allan, J.D., 2004. Landscapes and riverscapes: the influence of land use on stream ecosystems. Annu. Rev. Ecol. Evol. Syst. 35, 257–284. https://doi.org/10.1146/annurev. ecolsys.35.120202.110122. Allard, L., Popée, M., Vigouroux, R., Brosse, S., 2016. Effect of reduced impact logging and small-scale mining disturbances on Neotropical stream fish assemblages. Aquat. Sci. 78, 315–325. https://doi.org/10.1007/s00027-015-0433-4. Almeida, O.T, Uhl, C., 1998. Planejamento do Uso do Solo do Município de Paragominas/ Oriana Trindade e Christopher Uhl. Série Amazônia N° 09 – Belém: Imazon. pp. 46. Almeida, C.A., Coutinho, A.C., Esquerdo, J.C.D.M., Adami, M., Venturieri, A., Diniz, C.G., Dessay, N., Durieux, L., Gomes, A.R., 2016. High spatial resolution land use and land cover mapping of the Brazilian Legal Amazon in 2008 using Landsat-5/TM and MODIS data. Acta Amaz 46, 291–302. https://doi.org/10.1590/18094392201505504. American Veterinary Medical Association, 2013. AVMA Guidelines for the Euthanasia of Animals: 2013 Edition 38, Schaumburg, Illinois. Available at: https://www.avma. org/KB/Policies/Documents/euthanasia.pdf. Anderson, M.J., 2006. Distance-based tests for homogeneity of multivariate dispersions. Biometrics 62, 245–253. https://doi.org/10.1111/j.1461-0248.2006.00926.x.

9

Ecological Indicators 111 (2020) 106039

H. Leão, et al.

instream habitats. Land. Ecol. 31, 1725–1745. https://doi.org/10.1007/s10980-0160358-x. Legendre, P., De Cáceres, M., 2013. Beta diversity as the variance of community data: dissimilarity coefficients and partitioning. Ecol. Lett. 16, 951–963. https://doi.org/ 10.1111/ele.12141. Legendre, P., Legendre, L., 2012. Numerical Ecology-Developments in Environmental Modelling, 3rd ed. Elsevier Science BV, Amsterdam. Legendre, P., 2014. Interpreting the replacement and richness difference components of beta diversity. Glob. Ecol. Biogeogr. 23, 1324–1334. https://doi.org/10.1111/geb. 12207. Leibold, M.A., Holyoak, M., Mouquet, N., Amarasekare, P., Chase, J.M., Hoopes, M.F., Holt, R.D., Shurin, J.B., Law, R., Tilman, D., Loreau, M., Gonzalez, A., 2004. The metacommunity concept: a framework for multi-scale community ecology. Ecol. Lett. 7, 601–613. https://doi.org/10.1111/j.1461-0248.2004.00608.x. Lopes, P.M., Bini, L.M., Declerck, S.A.J., Farjalla, V.F., Vieira, L.C.G., Bonecker, C.C., Lansac-Toha, F.A., Esteves, F.A., Bozelli, R.L., 2014. Correlates of zooplankton beta diversity in tropical lake systems. PLoS One 9, e109581. https://doi.org/10.1371/ journal.pone.0109581. Mac Nally, R., Fleishman, E., Bulluck, L.P., Betrus, C.J., 2004. Comparative influence of spatial scale on beta diversity within regional assemblages of birds and butterflies. J. Biogeogr. 31, 917–929. https://doi.org/10.1111/j.1365-2699.2004.01089.x. Marini, L., Bertolli, A., Bona, E., Federici, G., Martini, F., Prosser, F., Bommarco, R., 2013. Beta-diversity patterns elucidate mechanisms of alien plant invasion in mountains. Glob. Ecol. Biogeogr. 22, 450–460. https://doi.org/10.1111/geb.12006. Miller, S.D., Goulden, M.L., Hutyra, L.R., Keller, M., Saleska, S.R., Wofsy, S.C., Figueira, A.M.S., Rocha, H.R., Camargo, P.B., 2011. Reduced impact logging minimally alters tropical rainforest carbon and energy exchange. Proc. Natl. Acad. Sci. 108, 19431–19435. https://doi.org/10.1073/pnas.1105068108. Mimouni, E.A., Pinel-Alloul, B., Beisner, B.E., 2015. Assessing aquatic biodiversity of zooplankton communities in an urban landscape. Urban Ecosyst. 18, 1353–1372. https://doi.org/10.1007/s11252-015-0457-5. Montag, L.F.A., Winemiller, K.O., Keppeler, F.W., Leão, H., Benone, N.L., Torres, N.R., Prudente, B.S., Begot, T.O., Bower, L.M., Saenz, D.E., Lopez-Delgado, E.O., Quintana, Y., Hoeinghaus, D.J., Juen, L., 2018. Land cover, riparian zones and instream habitat influence stream fish assemblages in the eastern Amazon. Ecol. Freshw. Fish. https:// doi.org/10.1111/eff.12455. Monteiro, M.A., Coelho, M.C.N., Barbosa, E.J., 2009. A área de estudo no contexto regional. In: Monteiro, M.A., Coelho, M.C.N., Barbosa, E.J.S. (Eds.), Atlas socioambiental: municípios de Tome-Acu, Aurora do Pará, Ipixuna do Pará, Paragominas e Ulianópolis. Núcleo de Altos Estudos Amazônicos, Belém ISBN, 987-85. Myers, J.A., Chase, J.M., Crandall, R.M., Jiménez, I., 2015. Disturbance alters beta-diversity but not the relative importance of community assembly mechanisms. J. Ecol. 103, 1291–1299. https://doi.org/10.1111/1365-2745.12436. Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., O’Hara, R.B., Simpson, G.L., Stevens, M.H.H., Wagner, H., 2011. Vegan: community ecology package. Version 1.17-11. Disponível em: http://vegan.r-forge.r-project.org. Pajunen, V., Luoto, M., Soininen, J., 2017. Unravelling direct and indirect effects of hierarchical factors driving microbial stream communities. J. Biogeogr. 44, 2376–2385. https://doi.org/10.1111/jbi.13046. Peck, D.V., Herlihy, A.T., Hill, B.H., Hughes, R.M., Kaufmann, P.R., Klemm, D.J., Lazorchak, J.M., McCormick, F.H., Peterson, S.A., Ringold, P.L., Magee, T., Cappaert, M.R., 2006. Environmental Monitoring and Assessment Program: Surface Waters Western Pilot Study – Field Operations Manual for Wadeable Streams. EPA 620/R06/003. Environmental Protection Agency, Washington, DC: US. Peel, M.C., Finlayson, B.L., McMahon, T.A., 2007. Updated world map of the KoppenGeiger climate classification. Hydrol. Earth Syst. Sci. 11, 1633–1644. https://doi.org/ 10.5194/hess-11-1633-2007. Phillips, H.R., Newbold, T., Purvis, A., 2017. Land-use effects on local biodiversity in tropical forests vary between continents. Biodivers. Conserv. 26, 2251–2270. https:// doi.org/10.1007/s10531-017-1356-2. Piroli, E.L., 2010. Disciplina de geoprocessamento: práticas em Idrisi – versão Taiga. Unesp/Campus Experimental de Ourinhos, 56. Prudente, B.S., Pompeu, P.S., Juen, L., Montag, L.F.A., 2017. Effects of reduced-impact logging on physical habitat and fish assemblages in streams of Eastern Amazonia. Freshw. Biol. 62, 303–316. https://doi.org/10.1111/fwb.12868. Putz, F.E., Zuidema, P.A., Synnott, T., Peña-Claros, M., Pinard, M.A., Sheil, D., Vanclay, J.K., Sist, P., Gourlet-Fleury, S., Griscom, B., Palmer, J., Zagt, R., 2012. Sustaining conservation values in selectively logged tropical forests: the attained and the attainable. Conserv. Lett. 5, 296–303. https://doi.org/10.1111/j.1755-263X.2012. 00242.x. R Core Team. 2016. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. https://www. R-project.org/. Ripley, B., Venables, B., Bates, D.M., Hornik, K., Gebhardt, A., Firth, D., Ripley, M.B., 2015. Package ‘MASS’. https://cran.r-project.org/web/packages/MASS/index.html. Schneider, K.N., Winemiller, K.O., 2008. Structural complexity of woody debris patches influences fish and macroinvertebrate species richness in a temperate floodplain-river system. Hydrobiologia 610, 235–244. https://doi.org/10.1007/s10750-008-9438-5. Siegloch, A.E., Silva, A.L.L., Silva, P.G., Hernández, M.I.M., 2018. Local and regional effects structuring aquatic insect assemblages at multiple spatial scales in a MainlandIsland region of the Atlantic Forest. Hydrobiologia 805, 61–73. https://doi.org/10. 1007/s10750-017-3277-1. Silva, P.G., Hernández, M.I.M., 2014. Local and regional effects on community structure of dung beetles in a mainland-island scenario. PLoS One 9, e111883. https://doi.org/ 10.1371/journal.pone.0111883. Silva, R.J., Coletti, F., Costa, D.A., Vaz-de-Mello, F.Z., 2014. Rola-bostas (Coleoptera: Scarabaeidae: Scarabaeinae) de florestas e pastagens no sudoeste da Amazônia

conservation challenges. Biol. Rev. 81, 163–182. https://doi.org/10.1017/ S1464793105006950. Dunn, O.J., 1964. Multiple comparisons using rank sums. Technometrics 6, 241–252. Edge, C.B., Fortin, M.J., Jackson, D.A., Lawrie, D., Stanfield, L., Shrestha, N., 2017. Habitat alteration and habitat fragmentation differentially affect beta diversity of stream fish communities. Landsc. Ecol. 32, 647–662. https://doi.org/10.1007/ s10980-016-0472-9. Edwards, D.P., Tobias, J.A., Sheil, D., Meijaard, E., Laurance, W.F., 2014. Maintaining ecosystem function and services in logged tropical forests. Trends Ecol. Evol. 29, 511–520. https://doi.org/10.1016/j.tree.2014.07.003. Ekroos, J., Heliölä, J., Kuussaari, M., 2010. Homogenization of lepidopteran communities in intensively cultivated agricultural landscapes. J. Appl. Ecol. 47, 459–467. https:// doi.org/10.1111/j.1365-2664.2009.01767.x. ESRI, 2014. Environmental Systems Research Institute – ESRI. ArcGIS 10.1. Environmental Systems Research Institute Inc., Redlands, CA, USA. Fearnside, P.M., Barbosa, R.I., 1998. Soil carbon changes from conversion of forest to pasture in Brazilian Amazonia. For. Ecol. Manage. 108, 147–166. https://doi.org/10. 1016/S0378-1127(98)00222-9. Ferreira, M.C., Begot, T.O., Prudente, B.S., Juen, L., Montag, L.F.A., 2018. Effects of oil palm plantations on habitat structure and fish assemblages in Amazon streams. Environ. Biol. Fishes 101, 547–562. https://doi.org/10.1007/s10641-018-0716-4. Flohre, A., Fischer, C., Aavik, T., Bengtsson, J., Berendse, F., Bommarco, R., Emmerson, M., 2011. Agricultural intensification and biodiversity partitioning in European landscapes comparing plants, carabids, and birds. Ecol. Appl. 21, 1772–1781. Fox, J., Weisberg, S., Adler, D., Bates, D. et al. 2015. Package ‘car’. https://cran.rproject. org/web/packages/car/index.html. Fugère, V., Kasangaki, A., Chapman, L.J., 2016. Land use changes in an afrotropical biodiversity hotspot affect stream alpha and beta diversity. Ecosphere 7, e0135. Göthe, E., Wiberg-Larsen, P., Kristensen, E.A., Baattrup-Pedersen, A., Sandin, L., Friberg, N., 2015. Impacts of habitat degradation and stream spatial location on biodiversity in a disturbed riverine landscape. Biodivers. Conserv. 24, 1423–1441. https://doi. org/10.1007/s10531-015-0865-0. Gurevitch, J., Padilla, D.K., 2004. Are invasive species a major cause of extinctions? Trends Ecol. Evol. 19, 470–474. https://doi.org/10.1016/j.tree.2004.07.005. Gutiérrez-Cánovas, C., Millán, A., Velasco, J., Vaughan, I.P., Ormerod, S.J., 2013. Contrasting effects of natural and anthropogenic stressors on beta diversity in river organisms. Glob. Ecol. Biogeogr. 22, 796–805. https://doi.org/10.1111/geb.12060. Hawkins, C.P., Mykrä, H., Oksanen, J., Vander Laan, J.J., 2015. Environmental disturbance can increase beta diversity of stream macroinvertebrate assemblages. Glob. Ecol. Biogeogr. 24, 483–494. https://doi.org/10.1111/geb.12254. Heino, J., Grönroos, M., 2017. Exploring species and site contributions to beta diversity in stream insect assemblages. Oecologia 183, 151–160. https://doi.org/10.1007/ s00442-016-3754-7. Heino, J., Bini, L.M., Andersson, J., Bergsten, J., Bjelke, U., Johansson, F., 2017. Unravelling the correlates of species richness and ecological uniqueness in a metacommunity of urban pond insects. Ecol. Indic. 73, 422–431. https://doi.org/10. 1016/j.ecolind.2016.10.006. Hill, R., 2007. PCI Geomatics Inc. PCI Geomatica version 10.1. Ontario – Canadá. Howson, T.J., Robson, B.J., Matthews, T.G., Mitchell, B.D., 2012. Size and quantity of woody debris affects fish assemblages in a sediment-disturbed lowland river. Ecol. Eng. 40, 144–152. https://doi.org/10.1016/j.ecoleng.2011.12.007. Hutchinson, M.F., 1957. Concluding remarks. Cold Spring Harb. Symp. Quant. Biol. 22, 415–427. https://doi.org/10.1101/SQB.1957.022. Jackson, D.A., 1993. Stopping rules in principal components analysis: a comparison of heuristical and statistical approaches. Ecology 74, 2201–2214. https://doi.org/10. 2307/1939574. Jaramillo-Villa, U., Caramaschi, E.P., 2008. Índices de integridade biótica usando peixes de água doce: uso nas regiões tropical e subtropical. Oecol. Bras. 12, 442–462. https://doi.org/10.4257/oeco.2008.1203.06. Juen, L., Cunha, E.J., Carvalho, F.G., Ferreira, M.C., Begot, T.O., Andrade, A.L., Shimano, Y., Leão, H., Pompeu, P.S., Montag, L.F.A., 2016. Effects of oil palm plantations on the habitat structure and biota of streams in Eastern Amazon. River Res. Appl. 32, 2081–2094. https://doi.org/10.1002/rra.3050. Karp, D.S., Rominger, A.J., Zook, J., Ranganathan, J., Ehrlich, P.R., Daily, G.C., 2012. Intensive agriculture erodes β-diversity at large scales. Ecol. Lett. 15, 963–970. https://doi.org/10.1111/j.1461-0248.2012.01815.x. Kaufmann, P.R., Levine, P., Robison, E.G., Seeliger, C., Peck, D.V., 1999. Quantifying Physical Habitat in Wadeable Streams. U.S Environmental Protection. Knop, E., 2016. Biotic homogenization of three insect groups due to urbanization. Glob. Chang. Biol. 22, 228–236. https://doi.org/10.1111/gcb.13091. Landeiro, V.L., Magnusson, W.E., Melo, A.S., Espírito-Santo, H., Bini, L.M., 2011. Spatial eigenfunction analyses in stream networks: do watercourse and overland distances produce different results? Freshw. Biol. 56, 1184–1192. https://doi.org/10.1111/j. 1365-2427.2010.02563.x. Landeiro, V.L., Franz, B., Heino, J., Siqueira, T., Bini, L.M., 2018. Species-poor and lowlying sites are more ecologically unique in a hyperdiverse Amazon region: evidence from multiple taxonomic groups. Divers. Distrib. 24, 966–977. https://doi.org/10. 1111/ddi.12734. Landis, J.R., Koch, G.G., 1977. The measurement of observer agreement for categorical data. Biometrics 33, 159–174. https://doi.org/10.2307/2529310. Larson, D.M., Dodds, W.K., Veach, A.M., 2018. Removal of woody riparian vegetation substantially altered a stream ecosystem in an otherwise undisturbed grassland watershed. Ecosystems 1–13. https://doi.org/10.1007/s10021-018-0252-2. Leal, C.G., Pompeu, P.S., Gardner, T.A., Leitão, R.P., Hughes, R.M., Kaufmann, P.R., Zuanon, J., de Paula, F.R., Ferraz, S.F.B., Thomson, J.R., MacNally, R., Ferreira, J., Barlow, J., 2016. Multi-scale assessment of human-induced changes to Amazonian

10

Ecological Indicators 111 (2020) 106039

H. Leão, et al.

and environmental drivers of stream fish communities in protected, agricultural and urban landscapes. Ecol. Indic. 101, 742–748. https://doi.org/10.1016/j.ecolind. 2019.01.063. Tritsch, I., Sist, P., Narvaes, I.D.S., Mazzei, L., Blanc, L., Bourgoin, C., Cornu, G., Gond, V., 2016. Multiple patterns of forest disturbance and logging shape forest landscapes in Paragominas. Brazil. Forests 7, 315. https://doi.org/10.3390/f7120315. Uieda, V.S., Castro, R.M.C., 1999. Coleta e fixação de peixes de riachos. In: In: Caramaschi, E.P., Mazzoni, R., Peres-Neto, P.R. (Eds.), Ecologia de Peixes de Riachos, Série Oecologia Brasiliensis, vol. VI. PPGE-UFRJ, Rio de Janeiro, pp. 1–22. Veloso, H.P., Rangel Filho, A.L.R., Lima, J.C.A., 1991. Classificação da Vegetação Brasileira, Adaptada a um Sistema Universal. Ministério da Economia, Fazenda e Planejamento, Fundação Instituto Brasileiro de Geografia e Estatística, Diretoria de Geociências, Departamento de Recursos Naturais e Estudos Ambientais. Vilmi, A., Karjalainen, S.M., Heino, J., 2017. Ecological uniqueness of stream and lake diatom communities shows different macroecological patterns. Divers. Distrib. 23, 1042–1053. https://doi.org/10.1111/ddi.12594. Watrin, O.D.S., da Rocha, A.M.A., 1992. Levantamento da Vegetação Natural e do Uso da Terra no Município de Paragominas (PA) Utilizando Imagens TM/Landsat. Belém, EMBRAPA/CPATU, fev. 40 p. (EMBRAPA/CPATU, Boletim de Pesquisa, 124). Weibull, A.C., Bengtsson, J., Nohlgren, E., 2000. Diversity of butterflies in the agricultural landscape: the role of farming system and landscape heterogeneity. Ecography 23, 743–750. https://doi.org/10.1111/j.1600-0587.2000.tb00317.x. Wood, A., Stedman-Edwards, P., Mang, J., 2013. The Root Causes of Biodiversity Loss. Routledge, New York. Zorzal-Almeida, S., Bini, L.M., Bicudo, D.C., 2017. Beta diversity of diatoms is driven by environmental heterogeneity, spatial extent and productivity. Hydrobiologia 800, 7–16. https://doi.org/10.1007/s10750-017-3117-3. Zuur, A.F., Ieno, E.N., Walker, N.J., Saveliev, A.A., Smith, G.M., 2009. GLM and GAM for Count Data. In Mixed Effects Models and Extensions in Ecology with R. Springer, New York, NY, pp. 209–243.

brasileira: Levantamento de espécies e guildas alimentares. Acta Amaz 44, 345–352. https://doi.org/10.1590/1809-4392201304472. Silva, P.G., Hernández, M.I.M., Heino, J., 2018. Disentangling the correlates of species and site contributions to beta diversity in dung beetle assemblages. Divers. Distrib. 24, 1674–1686. https://doi.org/10.1111/ddi.12785. Simões, N.R., Nunes, A.H., Dias, J.D., Lansac-Tôha, F.A., Velho, L.F.M., Bonecker, C.C., 2015. Impact of reservoirs on zooplankton diversity and implications for the conservation of natural aquatic environments. Hydrobiologia 758, 3–17. https://doi.org/ 10.1007/s10750-015-2260-y. Socolar, J.B., Gilroy, J.J., Kunin, W.E., Edwards, D.P., 2016. How should beta-diversity inform biodiversity conservation? Trends Ecol. Evol. 31, 67–80. https://doi.org/10. 1016/j.tree.2015.11.005. Solar, R.R.C., Barlow, J., Ferreira, J., Berenguer, E., Lees, A.C., Thomson, J.R., Louzada, J., et al., 2015. How pervasive is biotic homogenization in humanmodified tropical forest landscapes? Ecol. Lett. 18, 1108–1118. https://doi.org/10.1111/ele.12494. Sor, R., Legendre, P., Lek, S., 2018. Uniqueness of sampling site contributions to the total variance of macroinvertebrate communities in the Lower Mekong Basin. Ecol. Indic. 84, 425–432. https://doi.org/10.1016/j.ecolind.2017.08.038. Tews, J., Brose, U., Grimm, V., Tielbörger, K., Wichmann, M.C., Schwager, M., Jeltsch, F., 2004. Animal species diversity driven by habitat heterogeneity/diversity: the importance of keystone structures. J. Biogeogr. 31, 79–92. https://doi.org/10.1046/j. 0305-0270.2003.00994.x. Tolonen, K.E., Leinonen, K., Erkinaro, J., Heino, J., 2018. Ecological uniqueness of macroinvertebrate communities in high-latitude streams is a consequence of deterministic environmental filtering processes. Aquatic Ecol. 52, 17–33. https://doi. org/10.1007/s10452-017-9642-3. Tonkin, J.D., Heino, J., Sundermann, A., Haase, P., Jähnig, S.C., 2016. Context dependency in biodiversity patterns of central German stream metacommunities. Freshw. Biol. 61, 607–620. https://doi.org/10.1111/fwb.12728. Tóth, R., Czeglédi, I., Kern, B., Erős, T., 2019. Land use effects in riverscapes: Diversity

11