Research Article |
Corresponding author: Fabrício B. Teresa ( fabricioteresa@yahoo.com.br ) Academic editor: Adriano S. Melo
© 2018 Milton P. Ávila, Rafaela Neves Carvalho, Lilian Casatti, Juliana Simião-Ferreira, Lorrane Faria de Morais, Fabrício B. Teresa.
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation:
Ávila MP, Carvalho RN, Casatti L, Simião-Ferreira J, de Morais LF, Teresa FB (2018) Metrics derived from fish assemblages as indicators of environmental degradation in Cerrado streams. Zoologia 35: 1-8. https://doi.org/10.3897/zoologia.35.e12895
|
The development of effective monitoring tools depends on finding sensitive metrics that are capable of detecting the most important environmental impacts at a given region. We assessed if metrics derived from stream fish assemblages reflect physical habitat degradation and changes in land cover. We sampled the ichthyofauna and environmental characteristics of 16 stream sites of first and second order in the Upper Tocantins River basin. The streams were classified according to their environmental characteristics into reference (n = 5), intermediate (n = 4), and impacted (n = 7). A total of 4,079 individuals in five orders, 12 families, and 30 species were collected. Of the 20 metrics tested, eight were non-collinear and were tested for their performance in discriminating among groups of streams. Three metrics were sensitive to the gradient of degradation: Berger-Parker dominance index, percentage of characiform fish, and percentage of rheophilic individuals. Some commonly used metrics did not reflect the disturbances and many others were redundant with those that did. These results indicate that the metrics derived from fish assemblages may be informative for identifying the conservation status of streams, with the potential to be used in biomonitoring.
Bioindicators, Brazilian savannah, fish fauna, multimetric index
Anthropogenic impacts, such as pollution, riparian zone logging, and landscape (including network) fragmentation (
Biological communities can provide valuable information on ecosystem functioning and environmental quality (
Environmental quality has often been measured and monitored using metrics that have not been thoroughtly evaluated, particularly with respect to how they respond to the disturbance gradient (
Fish are excellent indicators of the conditions of aquatic environments (
We studied the response of headwater fish assemblages according to contrasting conditions of physical habitat degradation associated with the deforestation of the riparian zone. We evaluated whether metrics based on fish assemblages can be used to differentiate streams according to their conservation status.
We described fish assemblage of headwater sites in relation to potential indicators (metrics) and tested whether these data would allow us to distinguish among streams with different levels of environmental degradation. Stream sites were classified according to their environmental characteristics, representing a gradient of land use and habitat degradation.
The studied streams are part of the Upper Tocantins basin, Santa Tereza sub-basin (northern Goiás state) (Fig.
We sampled fish (license: ICMBio #34144-1) from 16 stream sites (sampling reach length of 80 meters) of 1st and 2nd order (sensu
Streams were characterized based on their environmental variables: 1) depth, 2) wetted width, 3) water velocity, 4) percentage of stable (hard) substrate (rocks and pebbles), 5) riparian forest width, 6) proportion of trees and grass in the riparian zone (30 m from the stream channel in each margin), 7) percentage of native vegetation in a buffer of 100- and 500-m upstream of the sampling sites, and 8) percentage of native vegetation in the entire catchment. These variables describe the natural environmental gradient (variables 1 to 3) and the level of stream degradation (variables 4 to 8) (
The physical habitat variables were assessed using a protocol adapted from
We tested 20 metrics to describe fish assemblages. These metrics describe patterns of diversity and dominance (Shannon diversity index, Berger-Parker dominance index, and Pielou’s evenness index), taxonomic composition (percentage of individuals in each family or order), size distribution (W statistic for ABC curve summarizing abundance and biomass accumulated among species), trophic ecology, and habitat use (percentage of individuals in each guild).
In order to identify trophic guilds, fish diets were evaluated. Ten individuals of each species had their stomachs removed and analyzed under a stereomicroscope. Food items were separated and identified in broad categories: detritus, terrestrial invertebrates, aquatic invertebrates, plant material (seeds and leaf fragments), and filamentous algae. The alimentary index (IAi) was calculated for each item following
The relationship between species abundance and biomass within an assemblage was summarized by ABC curves (
Our data anaysis can be summarized in three steps. In step one, we removed colinear variables from environmental and fish-based metric data, performing Pearson pairwise correlations in each matrix and removing variables that were strongly correlated (R > |0.75|). When two or more metrics were redundant with one another, we retained the metric that could more be objectively identified or recorded by a technician. In step two, streams were classified as reference (minimally disturbed), intermediate, or impacted, according to their environmental variables. Only variables that describe the gradient of environmental degradation in agroecosystems (variables 4-8 in sampling section), and which are useful to explain the variation in the structure and composition of fish assemblages were considered for stream classification (
Two variables (width of riparian zone and the percentage of trees in the riparian zone) were removed prior to the clustering analysis because they were collinear with the percentage of grass in the riparian zone. Cluster analysis based on environmental variables associated with dirturbance resulted in four major groups (Fig.
Environmental variables (mean ± standard deviation) from reference, intermediate and impacted stream groups.
Variables | Reference | Intermediate | Impacted |
---|---|---|---|
Depth (cm) | 13.73 ± 1.98 | 13.82 ± 4.39 | 10.62 ± 2.41 |
Wet width (m) | 3.16 ± 0.79 | 2.56 ± 0.78 | 2.18 ± 0.64 |
Water velocity (m/s) | 0.22 ± 0.11 | 0.13 ± 0.08 | 0.22 ± 0.14 |
Hard substrate (%) | 65.78 ± 10.10 | 41.11 ± 8.73 | 30.09 ± 20.03 |
Grass in the riparian zone (%) | 18.00 ± 19.32 | 23.75 ± 20.16 | 72.92 ± 15.69 |
Native vegetation (% in 100 m buffer upstream) | 94.23 ± 12.90 | 89.49 ± 11.20 | 45.38 ± 29.37 |
Native vegetation (% in 500 m buffer upstream) | 95.45 ± 4.24 | 72.27 ± 14.82 | 81.85 ± 12.32 |
Native vegetation (% in the whole catchment) | 88.25 ± 15.48 | 63.92 ± 8.88 | 57.20 ± 25.96 |
A total of 4,079 individuals in five orders, 12 families, and 30 species were sampled. The species accumulation curves reached the asymptote, indicating that the species sampling was representative (see Suppl. material
Of the 20 metrics, only eight non-collinear metrics were retained for the assessment of monitoring potential (Table
Performance of the metrics in distinguishing streams according to their conservation status, with the corresponding F statistic and p-value. For each sensitive metric, the redundant metrics (R > |0.75|) are shown. (+) positive correlation with the sensitive metric; (–) negative correlation with the sensitive metric.
Metrics | Redundant metrics | F(2,13) | p-value |
---|---|---|---|
Berger-Parker dominance index | W-statistic (–) | 10.78 | 0.001 |
Shannon diversity (–) | |||
Pielou evenness (–) | |||
% omnivorous individuals (+) | |||
% individuals of Characiformes | % individuals of Siluriformes (–) | 3.82 | 0.05 |
% individuals of Characidae (+) | |||
% individuals of Loricariidae (–) | |||
% benthic individuals (–) | |||
% rheophilic individuals | % individuals of Characidae (–) | 7.24 | 0.001 |
% individuals of Loricariidae (+) | |||
% water column individuals (–) | |||
% benthic individuals (+) | |||
% nektobenthic individuals (+) | |||
% aquatic invertivorous individuals (+) | |||
% omnivorous individuals (–) | |||
% detritivorous/periphytivorous individuals (+) | |||
Abundance | 1.15 | NS | |
% individuals of Gymnotiformes | 2.52 | NS | |
% individuals of Heptapteridae | 0.02 | NS | |
% individuals of Characiformes + Siluriformes | 1.17 | NS | |
% terrestrial invertivorous | 0.36 | NS |
We found that at least three metrics were capable of distinguishing streams in contrasting conditions. Accordingly, they can be useful for identifying the conservation status of streams. Thus, our study assists with the fundamental step of incorporating bioindicators in environmental assessment frameworks, which is the identification of sensitive metrics (
The stream groups identified by us represent the typical physionomies present in agroecosystems (
Despite the high proportion of minimally disturbed landscapes (
In streams with steep slopes, like those studied here, runs and riffles are prevalent along the longitudinal gradient. Thus, an abundance of rheophilic species is expected. Most of these species are dependent upon hard substrates, where they find suitable microhabitat and food resources (
Our sensitive metrics are complementary, as they allow to distinguish different stream conditions along the gradient of perturbation. Two metrics (dominance and percentage of Characiformes) are capable of distinguishing more impacted streams. However, at intermediate levels of disturbance these metrics should be uninformative. In these cases, the percentage of rheophilic individuals would allow the discrimination of reference from intermediate conditions. Therefore, these three metrics can be used jointly to improve the discriminatory power of assessment indices when the goal is to classify regional streams according to their environmental condition. In fact, the capacity to discriminate different conditions along a gradient of environmental perturbation is a desirable feature of bioindicators (
Recently a new fish-based multimetric index has been developed for streams in the Brazilian savannah (
We are grateful to the Laboratório de Ciências Biológicas (Universidade Estadual de Goiás/Porangatu) and the Laboratório de Biogeografia e Ecologia Aquática (Universidade Estadual de Goiás/Anápolis) for data collection; to Fernando R. Carvalho for fish identification; to Fundação de Apoio à Pesquisa do Estado de Goiás (201210267000703 and AUXPE 2036/2013) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (482185/2012-0) for research funding. MPA and LFM received scholarships from FAPEG; FBT and LC receive research grants from CNPq (302158/2015-4, 301755/2013-2 respectively).
Table S1. Species identity, total number of individuals and species classification according to trophic guilds and habitat use. Terins: terrestrial invertivorous; Aquins: aquatic invertivorous; Det-Per: detritivorous/periphytivorous; Pis: piscivorous; Omni: omnivorous; WC: water column; Ben: benthic; Nectb: nectobenthic; Bank: bank-dwelling species; Rheo: rheophilic.
Figure S1. Species accumulation (black) and rarefaction curve (gray) based on samples.
Data type: measurement