NCOND1500004  Research letter  Supplementary Material
Data analysis
The Shannon diversity index for water velocity (H’_{vel}) was calculated combining channel depth and water velocity data. Each combination of depth and water velocity was considered as a point in a Cartesian system, in which the x axis represented the depth measurements and the y axis represented the water velocity measurements. The axes intersection occurred at the median values. Each quadrant of this Cartesian system represented a combination of depthvelocity: deepfast (DF), deepslow (DS), shallowslow (SS) and shallowfast (SF). The number of occurrences in each quadrant was used to compose H’_{vel}, in which the maximum richness was four (four quadrants) and the maximum abundance was the total number of measurements made in each site.
The abundance of Isbrueckerichthys epakmos, described by a Poisson distribution, was used as the response variable in different generalized linear models (GLM), which predictor variables consisted of combinations of the percentage of coarse substrate (cs), percentage of pools and runs and the Shannon diversity index for velocity (H’_{vel}). In these Poisson GLM, the systematic part is given by E(N_{i}) = e^{b0 + b1 x cs + b2 x pr + b3 x Hvel} (eq. 1) (Zuur et al. 2009). This is the full model, where E(Ni) is the expected abundance of I. epakmos and b0 to b3 are the model parameters. Nested submodels of eq. 1 were tested setting different combination of the parameters b1 to b3 to zero and let the others to be estimated. For instance, when all parameters except b0 are set to zero, the abundance of I. epakmos is described by the reduced model E(N_{i}) = e^{b0}. In total, eight different model combinations were tested, including the reduced and the full models (Table 1). The model parameters were estimated by using the function glm of the package stats (R Development Core Team 2011).
To choose the best predictor model we used the Akaike second order Information Criterion (AICc), which can be interpreted as a measure of the distance between each model and a “true” model, which is not necessarily known. The key idea is that the best model is the one with the shortest distance to this hypothetical model (Anderson 2008), that is, the one with lowest AICc value (function aictab of the package AICcmodavg, R Development Core Team 2011). The probability ratio between the best and reduced model (wi) checked the number of times that the chosen model was more parsimonious than the reduced one. The value of wi was calculated using the formula: _{ }
Where wi = probability of model i, Δi = relative difference between the AICc of the chosen model and the reduced model and R = number of models to be compared (function evidence of the package AICcmodavg, R Development Core Team 2011).
Results
We captured 1006 individuals among 26 species (Table S1). The order Characiformes was the most representative with 431 specimens (43%) from nine captured species (Table S1). Next was Siluriformes with 238 specimens (24%) from ten species, Cyprinodontiformes had 189 specimens (19%) from a single species, Gymnotiformes had 119 specimens (12%) from two species, Perciformes had 28 specimens (3% ) from two species and, finally, Cypriniformes had a single specimen of an exotic species, Misgurnus anguillicaudatus (Cantor, 1842). Moreover, among the orders Characiformes, Siluriformes and Perciformes, we identified five species endemic to the Ribeira de Iguape River basin (Table S1).
In the analyses of rarity (Fig. S1), the species with the lowest frequency of occurrence were Astyanax ribeirae, Bryconamericus microcephalus, Misgurnus anguillicaudatus, and Pimelodella transitoria and, together with Neoplecostomus yapo, these were also nonrepresentative in terms of abundance. With the exception of P. transitoria, these species were also nonrepresentative in biomass, as was Neoplecostomus ribeirensis. The most frequently occurring species were Characidium pterostictum, Gymnotus pantherinus, Isbrueckerichthys epakmos, Phalloceros reisi and Rhamdia quelen. With the exception of R. quelen, these species were very representative in terms of abundance, together with Astyanax sp.1 and Deuterodon iguape. The most abundant species by biomass were Astyanax sp.1, D. iguape, G. pantherinus, I. epakmos and R. quelen.
References
Anderson DR, 2008. Modelbased inference in the life sciences: a primer on evidence. New York: Springer.
Nelson JS, 2006. Fishes of the world. New Jersey: John Wiley & Sons, Inc.
Oyakawa OT et al., 2006. Peixes de riachos da Mata Atlântica nas Unidades de Conservação do Vale do Rio Ribeira de Iguape. São Paulo: Neotrópica.
R Development Core Team, 2011. R: A language and environment for statistical computing. Austria: R Foundation for Statistical Computing.
Zuur AF et al., 2009. Mixed effects models and extensions in Ecology with R. New York: Springer.
Table S1  Sampled species and their corresponding families and orders, according to the classification of Nelson (2006). Nsp: species abundance, DZSJRP: voucher number in the Fish Collection, Department of Zoology, São José do Rio Preto. ¹Exotic species, ²undescribed species, identified based on Oyakawa et al. (2006), ³endemic to the Ribeira de Iguape de Iguape River basin.
Order/Family/Species

N_{sp}

DZSJRP

Cypriniformes



COBITIDAE



¹Misgurnus anguillicaudatus (Cantor, 1842)

1



Characiformes



CRENUCHIDAE



Characidium lanei Travassos, 1967

17

13680

Characidium pterostictum Gomes, 1947

173

13659

CHARACIDAE



³Astyanax ribeirae Eigenmann, 1911

1

15338

²Astyanax sp.1

122

13672

²Astyanax sp.2

7

13671

Bryconamericus microcephalus (Miranda Ribeiro, 1908)

1

15335

³Deuterodon iguape Eigenmann, 1907

80

13700

Hyphessobrycon anisitsi (Eigenmann, 1907)

25

13652

ERYTHRINIDAE



Hoplias malabaricus (Bloch, 1794)

5

13683

Siluriformes



TRICHOMYCTERIDAE



Ituglanis proops (Miranda Ribeiro, 1908)

7

13669

Trichomycterus zonatus (Eigenmann, 1918)

34

13668

LORICARIIDAE



Harttia kronei Miranda Ribeiro, 1908

5

13697

Hisonotus sp.

6

13688

Hypostomus ancistroides (Ihering, 1911)

12

13692

³Isbrueckerichthys epakmos Pereira & Oyakawa, 2003

153

13658

³Neoplecostomus ribeirensis Langeani, 1990

2

13696

Neoplecostomus yapo Zawadzki, Pavanelli & Langeani, 2008

3

13651

HEPTAPTERIDAE



Pimelodella transitoria Miranda Ribeiro, 1907

1

13704

Rhamdia quelen (Quoy & Gaimard, 1824)

15

13685

Gymnotiformes



GYMNOTIDAE



Gymnotus pantherinus (Steindachner, 1908)

117

13684

Gymnotus sylvius Albert & FernandesMatioli, 1999

2

13657

Cyprinodontiformes



POECILIIDAE



Phalloceros reisi Lucinda, 2008

189

13699

Perciformes



CICHLIDAE



Australoheros facetus (Jenyns, 1842)

3

13654

³Geophagus iporangensis Haseman, 1911

25

13660

Fig. S1  Distribution data of average number of individuals (N'i), average biomass (P'i) and frequency of occurrence (Fo_{i}) of the species Astyanax sp.1 (Asp1), Astyanax ribeirae (Arib), Bryconamericus microcephalus (Bmic), Characidium pterostictum (Cpte), Deuterodon iguape (Digu), Gymnotus pantherinus (Gpan), Isbrueckerichthys epakmos (Iepa), Misgurnus anguillicaudatus (Mang), Neoplecostomus ribeirensis (Nrib), Neoplecostomus yapo (Nyap), Phalloceros reisi (Prei), Pimelodella transitoria (Ptra) and Rhamdia quelen (Rque).
