The protected areas system in Brazil as a baseline condition for wetlands management and fish conservancy : the example of the Pantanal National Park

Considering the need for the Brazilian government to develop tools for environmental monitoring for biodiversity conservancy purposes in the national protected areas system, this paper focuses on determining reference site metrics (or baselines) for adapting the Index of Biotic Integrity (IBI) based on the fish assemblages in the Pantanal National Park (PNP). The habitats in the PNP were grouped into four categories: main rivers, corixos (channels connecting the floodplain), permanent bays, and temporary bays. Fish samplings were performed at 12 points during the dry season (Oct-Nov 2010 and 2011). 146 fish species were identified from the total 18,954 individuals collected with standardized fishing gear. There was no association between the structure of the fish assemblage and categories, suggesting a theory on homogeneity of habitats. The final IBIPNP consists of nine metrics, most of them were framed in excellent class, some in fair, and none in poor. There was no significant difference in IBIPNP scores between the two sampled years. This approach provides a direct application for wetland management purposes.


Introduction
Human disturbances interact in a complex way with aquatic ecosystems, and their effects can rarely be evaluated using only physical or chemical variables as indirect measures of environmental integrity.Assessing the response of the aquatic biota (e.g., freshwater fish) often provides a more integrated view of environmental impacts (Fausch et al., 1990).These are the principles of the original idea of the Index of Biotic Integrity (IBI) (Karr, 1981;Karr et al., 1986) that combined 12 metrics (or attributes) of fish assemblages to measure and predict the outcomes of anthropogenic environmental disturbance through a given environmental gradient.

e170041[2]
The applications of the IBI in the U.S. and Europe demonstrated the effectiveness of the method and highlighted the need to modify or adapt the index to particularities of regions and systems under study (Hughes, Oberdorff, 1999;Simon, Sanders, 1999;Roset et al., 2007).
In Brazil, the first adaptation of the IBI based on fish assemblages was carried out in 1998 (Araújo et al., 2003) in one of the most critical segments of the highly polluted Paraíba do Sul River, which runs through the main industrial region of the country.In the Upper Paraná River, small streams (Ferreira, Casatti, 2006;Casatti et al., 2009) and reservoirs (Petesse et al., 2007;Terra et al., 2010) were also investigated by adapting the IBI to these systems.
According to the original proposition (Karr, 1981;Karr et al., 1986), the intensity of environmental degradation should be evaluated by comparing the modified habitats with pristine environments in the same or a nearby region (such as a protected area) taken as reference conditions or baselines.However, pristine sites are frequently not found, explained by an environmental and cultural "illness" known as SBS -Shifting Baseline Syndrome (Pauly, 1995).In those cases, some authors have considered the site with the lowest degree of modification to represent reference conditions (Hughes et al., 1986;McDonough, Hickman, 1999;Pont et al., 2006).From this moment on, the baseline becomes shifted in a contagious and probably irreversible process.One generation later (25-30 yrs.), there will last no environmental collective memory of the prior baseline.
In the face of rapid degradation of aquatic environments around the world, it is argued that adaptations of IBI should also be undertaken for the few environments where human modifications are still not apparent.For these environments, reference conditions can and should be established before human modification becomes evident and irreversible.This idea is, again, in line with the SBS concept: after an anthropogenic impacts are disseminated, there are no reference points that could be used as a base for future evaluation of ongoing stresses on ecosystem functionality (Pauly, 1995).The problem arises because there are few situations in which a monitoring tool is implemented before the impact is present.Nonetheless, efforts should be made to change this scenario, at least for the few ecosystems that still have pristine conditions, so that the reference point could drive policy makers in proposing environmental and conservation programs.
Thus, the present study aims to adapt the IBI for the fish community of the Pantanal National Park, a large wetland area in Brazil.The most remarkable characteristic of wetlands of the Pantanal is seasonal variation in water levels, which controls ecosystem dynamics.The Pantanal is a Biosphere Reserve by UNESCO and can be considered a biodiversity refuge among Brazilian wetlands.

e170041[3]
Sampling design and fish collection.Based on cartographic charts of habitat distribution and photographic records, four regions, totaling 12 sampling sites (named P1 through P12), were pre-established and visited in the field for confirmation.Strata were defined based on water flow (lotic or lentic) and connectivity with main river channel (permanent or temporary connection).Category A (lotic environments represented by P4 and P6) was characterized by two major rivers, Cuiabá and Paraguay, in the Pantanal floodplain.These rivers are permanent water bodies with well-defined channels, and, henceforth, this category is designated as main channel sites.Category B (lentic environments; P3, P7, and P8) is composed of floodplain lakes with permanent connection to the main river channel throughout the hydrological cycle (Burro, Três Bocas, and Morro bays).This category is identified as permanent bays.With the exception of Três Bocas Bay, which is deeper, the other two are shallower and are about 50 cm deep during the dry season.Category C (lentic environments; P5, P10, and P11) is represented by smaller bays (Nove, Morro, and Biguerinho river bays), called baiotas, with intermittent connection with the main river and average depths ranging from 1 to 2 meters during the high-water season.These river bays can dry out completely in the low water season (except in years with higher flows) and, hereinafter, are referred to as temporary bays.Lastly, category D (lotic environments; P1, P2, P9, and P12) is represented by connected channels linking the main rivers and other elements of the floodplain, such as bays, and are referred locally as corixos (Boca do Ricardo, Morro do Caracara, Caracarazinho do Morro river, and Biguerinho river).As they link the main rivers to the floodplain, the current is weaker than in the main channel, and they have no delimited headwater and mouth.However, they may exhibit very well defined channels.When the channels are not well defined, they are called vazante (ebb), where the water simply overflows.This type of body is identified as corixos.
In a preliminary recognition campaign in the PNP area (Sep-2009), these 12 sampling points were selected out of 40 other potential locations based on: a) typology of habitat structure, b) cartographic analysis, and c) access by boat, considering that the core area of the PNP is unreachable.The samples were taken during the dry season of 2010 and 2011 because, during wet season, the four categories are indistinguishable and fish capture is less efficient due to high water level.
The sampling effort was characterized by 30 sequential hauls, separated by a minimum distance of 3 m along the marginal vegetation.Medium to large fish were collected with two cast nets (with 2 and 8 cm between opposite knots and with 15 hauls of each cast net in each sampling point) and gillnets (set of 7 nets with mesh ranging from 4 to 16 cm between opposite knots that were set for 2 hours in the water during sunlight).The sequence of nets was set at random.The small-sized fish species, usually associated with macrophytes, were collected with sieves built from net shading (nylon 1.5 mm opposite knots mesh size) that had dimensions of 1.5 m x 1.5 m and 0.7 m depth.
Metrics selection and description.The IBI PNP metrics were adapted from Karr et al. (1986), Ganasan, Hughes (1998), Araújo et al. (2003), Bozzetti, Schulz (2004), Petesse et al. (2007), andFerreira (2011).34 candidate metrics were selected and grouped into four categories (Tab. 1) considering the following traits of the fish community structure: (i) richness and species assemblages (8 metrics); (ii) fish abundance (8 metrics); (iii) composition of feeding groups (10 metrics) and; (iv) indicators of environmental health at community level (8 metrics).All metrics were calculated in numbers of fish, as the measurement of biomass is difficult to obtain in the field due to the amount of fish collected.With exception of metrics 32 (number of introduced species) and 33 (number of hybrid species) only native species of the Pantanal ecosystem were considered.
In order to select appropriate metrics to compose the IBI PNP , 3 tests were conducted (i) a range test to assess the magnitude and variability of the metrics in the sampling sites, (ii) a responsiveness test to evaluate the response of each metric to habitat modification, and (iii) a redundancy test to exclude highly correlated metrics of modification (Stoddard et al., 2008;Petesse et al., 2014).Metrics with ranges equal or smaller than 4 or with 75% of identical values (range test) were excluded, since they would be unable to detect differences among the sampling locations (Whittier et al., 2007) and therefore would have low discriminatory power.The metric responses to environmental modification (responsiveness test) were verified by the Spearman correlation coefficient (r S ) between each remaining metric and the following abiotic characteristics: macrophyte cover-MC, macrophyte integrity-MI, presence of communal bird nests-PN, air temperature-AT, water temperature-WT, water transparency-WTp, depth-D, electrical conductivity-EC, dissolved oxygen-DO, and oxidation/reduction potential-ORP (Polaz, 2013).Metrics that showed significant correlation (p<0.05)passed the test.The Spearman correlation was also calculated to identify pairs of redundant metrics (redundancy test).For pairs of metrics that had a correlation coefficient above 0.7 (Whittier et al., 2007), the ones that exhibited the lowest number of significant correlations in the responsiveness test were excluded (McCormick et al., 2001;Whittier et al., 2007).
Criteria for scoring and classes of integrity.As in most applications of IBI, ordinal scores were used to standardize metrics in different scales considering the categories: impacted biota (1), fair or intermediate (3), and not impaired (5).
The choice of a scale with discrete values (1, 3, and 5) is mainly justified by facilitating the communication of results to non-specialists that is a desirable feature for the management of protected areas open to the public.

e170041[4]
In order to exclude the effect of outliers, only the interval between the 5 th and 95 th percentiles were considered.For the remaining metrics, the 25 th percentile of the distribution as the inferior limit was assumed.Values above the 25 th percentile were considered unimpaired (score 5), and values below the 25 th percentile were bisected.Those in the top half received a score of 3, and those in the bottom half received a score of 1. Gerritsen et al. (2003) recommended this procedure for scoring when most sites are undisturbed.This score attribution was inverted for metrics that increase with disturbance intensity such as the number of introduced species.Finally, the scores were summed to obtain the final IBI, which was classified as "poor", "fair", or "excellent" classes of biotic integrity.A two-way Analysis of Variance was used to test the effects of water body categorization and year of sampling on the final IBI PNP (Zar, 2010).
Tab. 1. Candidate metrics, expected responses to disturbance and results of criteria tests (1, 2 and 3) to compose the IBI PNP .The sign (-) indicates that the measure did not pass the test.
Metrics selection.Nine of the 34 candidate metrics passed the criteria established by the range, responsiveness, and redundancy tests (Tab.1).The number of native orders (metric 2) ranged from 3 to 6, and, therefore, they did not pass the range test (range < 4).Introduced species or hybrids were not sampled (metrics 32 and 33, respectively), and only 4 of the 24 samples exhibited had single individual anomaly (metric 34).In all these cases, the metrics did not meet the first criteria, as more than 75% of the values were zero.
Ten of the remaining 30 metrics (3, 6, 8, 9, 10, 11, 12, 13, 25, and 28) passed the second test (responsiveness), as they were correlated with at least one environmental variable.Most metrics (3, 8, 9, 25, and 28) showed a significant correlation with only one of the abiotic factors, one metric (10) with two factors, three metrics (6, 12, and 13) with three factors, and one metric (11) with four factors.Among the abiotic variables, the macrophyte cover (MC) and the oxidation-reduction potential (ORP) showed the highest number of metrics correlation.In particular, four metrics were correlated with MC, while there were four others with ORP.A test was proposed for the communal bird nests (locally known as ninhais) as a metrics because, especially in the reproductive season, they can provide a relevant feeding microhabitat for the fish community.Nonetheless, this factor and water depth were not correlated with any metrics.
From the 10 metrics accepted by the second test, the pairwise correlation matrix was calculated using the Spearman correlation coefficient (see more in supplementary material), resulting in only one redundant pair (metrics 11 and 12, r S = -0.863).It was decided to keep metric 11 as it had the greater number of significant correlations (four correlations) with the environmental variables (responsiveness test).Therefore, only metric 12 did not pass in the test of redundancy.
Metrics scoring and IBI PNP calculation.All nine selected metrics are expected to decrease their values with increasing intensities of disturbance.Therefore, lower scores indicate worse conditions, and the highest scores indicate the best conditions.As there are nine metrics, the final IBI PNP for a site could range from 9 (the worst condition in which all of the metrics are given the value 1) to 45 (the best condition in which all metrics are given the value 5) (Tab.3).
Tab. 3. Score intervals and ranges of values of the metrics for calculating the IBI PNP .Sites with excellent biotic integrity (ranging from 35 to 45) represented the best-observed situations.The community composition includes the most intolerant forms in all ranges of age category and sex ratio.They have a diverse and balanced trophic structure, the presence of large species and migratory fish, and the expected fish diversity for the biome.Fair conditions (ranging from 22 to 34) are those sites in which the number of species is close the expected richness, having a relatively balanced trophic structure but with some species decreasing their relative abundance and size below the values expected for the best condition.This fish community begins to show some signs of stress and needs to be monitored.The poor sites (ranging from 9 to 21) have low diversity, simplified trophic structure, dominance of tolerant and introduced species (non-native and/or exotic ones), and may show a high percentage of individuals with anomalies and parasites.

Discussion
The PNP is a large area within the Pantanal biome representing the wetlands of Central-West Brazil.This national park is closed for public visitation as of this publication.However, tourism will soon be allowed, which may result in additional sources of environmental impact if this activity is poorly managed.The building of small reservoirs in the headwaters of its main rivers may also lead to changes in the hydrological cycle and consequently modify the ecosystem dynamics.The multimetric indices are best suited to assess the impact on aquatic environments from multiple sources, as they take into account different ecological attributes of communities (Seegert, 2000).Thus, this attempt to conduct the first implementation of an index of biotic integrity to PNP is aimed to provide a monitoring tool and a reference condition for assessing future environmental impairment within the park and its surroundings.
Applying the IBI to a preserved area has the unique possibility of gathering information with confidence on the distribution of reference values.As the metrics are measured before the beginning of disturbance processes and virtually all sites can be considered in good or excellent biotic integrity, there is no need to model sample values for baseline conditions in this case.A logical difficulty, however, is that in the absence of human impacts, the IBI PNP values become difficulty to validate.Considering scenarios of strong disturbance, some responses are easily predicted, such as the reduction of species richness, migratory fishes, and intolerant species.On the other hand, predicting the responses in advance of these impacts may be difficult and challenging when the disturbances are subtle.Because the PNP is a conservation unit not severely impacted, small values for a metric distribution were not expected.Score 1 should represent clear evidence of human impact, but within the range of observations on hand, sites classified as 1 are considered as far as possible from the best potential condition (score 5) for a determined metric.This is expected due to natural variation, even when analyzing pristine environments (Gerritsen et al., 2003).The challenge of adapting an IBI to an undisturbed area is just as difficult as adapting it to a heavily disturbed one, where natural conditions are no longer available.In both cases, not all classes of disturbance-integrity are present, and the researchers have only a partial view of the range of possibilities for the metrics distribution.
Due to this limitation, the choice for candidate metrics on IBI's adapted for other environments was investigated and the appropriate metrics were selected based on the three established criteria: range test, responsiveness, and redundancy (Whittier et al. 2007, Petesse et al., 2014).The second criteria, for example, attempted to identify metrics that were sensitive to environmental conditions, expecting that these responses will exacerbate in the presence of anthropogenic impacts.In addition, following Gerritsen et al. (2003), the choice for the kind of metric scoring reflected the fact that the observed values will act as reference for future impacts within the region.
Significant differences among the types of water bodies and between years on the mean values of IBI PNP were not detected.This suggests that despite being clearly heterogeneous, the categories of water bodies analyzed have similar integrities, and these patterns do not change significantly from one year to another.In other applications of the IBI, temporal variability have been generally less important than spatial variability (Petesse et al., 2007;Casatti et al., 2009;Terra, Araújo, 2010;Ferreira, 2011), being the reason why Casatti et al. (2009) reinforced the importance of investing in the spatial variability instead of seasonality in early adaptations of multimetric indices.Terra, Araújo (2010), for instance, reported that seasonality had no effect on Reservoir Fish Assemblage Index (RFAI; an adaptation of IBI for reservoirs) in a transitional riverdam stretch of the Paraíba do Sul River that covers the Funil reservoir.
Considering this question, the current temporal study and sampling did not investigate the variation during the hydrological cycle (drought, flood, ebb, and flow), which l is the most important factor controlling community dynamics in the Pantanal.It was expected that large changes in fish communities during hydrological cycle would occur, but the similarity found in the IBI PNP between the dry seasons of 2010 and 2011 suggests that the fish populations are adapted to such phenomena.When year to year variability is not important, Karr et al. (1986) recommended that fish sampling should be conducted in a season that minimizes variation in fish community, due to discrete events such as fish recruitment.Several factors contribute to variations in the observed values for the IBI of a particular place and time because the index is entirely dependent on the composition and abundance of the collected fish population.Some pattern of movements and the collectors experience, for example, interfere decisively in the species and individuals sampled (e.g., larger or smaller) (Fore et al., 1994).This makes the multimetric indices behave as a random variable, adding a stochastic component upon the estimate of biotic integrity.In the Pantanal, recruitment takes place mainly during flood periods.As such, more variation in the community composition during flow and flood periods due to fish migrations and recruitment can be expected.Fish sampling is also less efficient during these seasons due to the higher water volume.In the dry season, the aquatic environments are more constrained, and sampling is more efficient, yielding a better characterization of fish communities (Junk et al., 1989;Junk, 2001).These facts provide us a better picture of how fish community structures are organized, which is a practical concern if the IBI PNP is established as a monitoring tool.
Nine out of the 34 candidate metrics that were selected were related to fish richness and composition (3 metrics), abundance (4 metrics), composition of feeding 12 e170041 [12] groups (1 metric), and indicators at the community levels (1 metric).Four metrics handle with the patterns observed at higher taxonomic levels such as order (total number of Siluriformes species or abundance of Characiformes and Gymnotiformes) or family (total number of Cichlidae species).As an advantage, these taxonomic levels are easily recognized by non-specialists.Other metrics, such as total number of individuals or number of species comprising 80% of total abundance, are easily calculated.For example, there is no need to determine each species when counting how many species comprise 80% of total population.In addition, Petesse et al. (2007) concluded that calculating the index in terms of fish biomass or fish numbers generates similar results in a previous application of a multimetric index to reservoirs.If the same applies for the IBI PNP , the application in numbers would require less field effort because it would be impracticable to weigh the high numbers of fish captured in the field (mainly the small ones).
The IBI PNP values are subject to the environmental characteristics observed within the park.In this case, it should be considered that the type of gear employed tended to catch the fish living in shallow habitats (<2m depth).Thus, there is a need to standardize the sampling method for future comparisons.
By combining indicators into an overall multimetric index to summarize biological condition, it is reinforced that the presented results in metrics combination always carried to the same baseline: PNP sites can be considered as references.Although the IBI PNP can be used as a reference for future adaptations, further studies are needed on the biology of each species and how they interact with their habitats.It is also suggested that wetland areas outside of the park with evident signs of anthropogenic degradation be selected to validate our results and to provide new opportunities to test the IBI PNP .
In addition, research that has the opportunity to investigate pristine areas should be encouraged and supported by governments and international organizations because information in these areas, even if already available, cannot be easily transferred to multimetric indices.Given that budget and time to conduct additional studies is always a limitation, in addition to considering the degree of human impact on aquatic environments, such information will be of great value to substantiate management decisions.
area.The Pantanal National Park (PNP) is located in the extreme southwest of the Mato Grosso State in the municipality of Poconé at the junction of the Paraguay and Cuiabá rivers (Fig. 1).This unique Park retains a considerable sample of the original lowland with a total area of 1,350 km 2 , equivalent to 1% of the Brazilian Pantanal biome (Instituto Brasileiro do Meio Ambiente e dos Recursos Naturais Renováveis, 2003).Founded in September 1981, this national protected area (Ministério do Meio Ambiente, 2000) has been managed since 2007 by the Instituto Chico Mendes de Conservação da Biodiversidade, an agency of the Brazilian government.

Fig. 2 .
Fig. 2. Distribution of the IBI PNP in the strata sampled for the Pantanal National Park during the dry season (Oct-Nov, 2010 and 2011).PB -permanent bays, TB -temporary bays, CR -corixos, MR -main rivers.

Fig. 3 .
Fig. 3. Distribution of the IBI PNP for the years of 2010 and 2011.