Carvalho 2014

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Ecology of Freshwater Fish 2014 Ó 2014 John Wiley & Sons A/S.

Ó 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd

ECOLOGY OF
FRESHWATER FISH

Environmental and spatial processes: what


controls the functional structure of fish
assemblages in tropical rivers and headwater
streams?
Rodrigo A. Carvalho1, Francisco L. Tejerina-Garro2
1
Departamento de Ecologia, Programa de Pos-graduacß~ao em Ecologia e Evolucß~ao, Universidade Federal de Goias, Instituto de Ci^encias Biol ogicas, Rodovia
Goi^ania-Ner
opolis km 5, Campus II Itatiaia, Goi^ania, 74001-970, GO Brasil
2
Centro de Biologia Aquatica, Pontifıcia Universidade Catolica de Goias – Campus II, Av. Engler s/n, Jardim Mariliza, Goi^ania, 74605-010, GO Brasil

Accepted for publication April 27, 2014

Abstract – In this study, we investigated functional structure patterns of tropical headwater and river fish
assemblages. We hypothesised that environmental conditions are primarily structuring headwater streams leading to
functionally clustered assemblages, whereas processes that favour functional overdispersion would guide river
assemblages. For 27 headwater streams and 22 rivers, we used eight functional traits for calculating two functional
indexes: mean pairwise distance (MPD) and net relatedness index (NRI). We performed linear regressions between
indexes and species richness, a multiple regression between NRI and eight environmental variables and a variation
partitioning to disentangle the role of environment and space on NRI. Our findings indicate that fish assemblages of
headwaters are structured by environmental conditions as most assemblages in this habitat displayed a tendency to
clustering and MPD/NRI were not correlated with species diversity, whereas the opposite pattern was observed for
river habitat. Four environmental variables (channel depth, water velocity, dissolved oxygen and turbidity) explain
56% of functional structure variation. These variables seem to function as selective filters in headwaters, whereas
channel depth may be determinant for functional overdispersion of river fish assemblages. Components associated
with space are also influencing the functional structure. Limitations of species dispersal through space (between
both habitat types) appear as a possible cause to this. In this sense, both environmental conditions and processes
linked with space are capable of influencing the functional structure of tropical headwater streams and river fish
assemblages.

Key words: Central Brazil; mean pairwise distance; net relatedness index; Tocantins-Araguaia basin; variation partitioning

independent of their traits and/or interactions, would


Introduction
be another possibility as proposed by the neutral the-
A central issue of ecological research relies on under- ory of Hubbell (2001). It has been demonstrated that
standing which processes underlie the diversity pat- all three processes may act at the same time (Helmus
terns of biological communities. A wide range of et al. 2007) or sequentially – in the case of environ-
explanations has been proposed to explain these pat- mental gradients – to determine diversity patterns
terns in aquatic systems, including abiotic factors (Mason et al. 2007). Therefore, the current challenge
(Poff & Allan 1995; Peres-Neto 2004; Hoeinghaus no longer seems to be finding a valid mechanism but
et al. 2007; S
uarez et al. 2007) and biotic interactions comprehending which one most strongly influences
among species (Winston 1995; Taylor 1996, 1997). on local community organisation (Mouchet et al.
The co-occurrence of species and their persistence, 2010). For freshwater fish assemblages, some studies

Correspondence: Rodrigo A. Carvalho, Departamento de Ecologia, Universidade Federal de Goias – UFG, Caixa Postal 131, CEP, Goi^ania 74001-970, GO, Brasil.
E-mail: rodrigoassiscarvalho@gmail.com

doi: 10.1111/eff.12152 1
Carvalho & Tejerina-Garro
investigated such mechanisms and different findings Grenouillet et al. 2004). Due to high variability in
have been presented. Poff & Allan (1995) found that their environment, these low-order streams frequently
the presence of fish species with generalist or specialist support low-diversity and less-structured assemblages
feeding strategies is directly associated with hydrologi- (Jackson et al. 2001) and a few hardy species tend to
cal variability of streams. Peres-Neto (2004) showed persist (Rahel & Hubert 1991). On the other hand,
that habitat features are of greater importance for deter- there is an increase in habitat diversity and environ-
mining species co-occurrence than species interactions. mental stability further downstream along fluvial gra-
On the other hand, Winston (1995) demonstrated that dients (Meffe & Minkley 1987; Jackson et al. 2001).
competition is a powerful force in determining species This fact frequently allows the persistence of fish
co-existence and Gilliam & Fraser (2001) found out assemblages with a higher number of species (Mat-
that predation threat is capable of influencing the thews 1986). Given the potential influence of abiotic
movement of fish species along streams. factors as selective filters at headwater streams and
Traditionally, environmental conditions have been the higher stability of downstream areas (high-order
used to explain the variation of diversity patterns in rivers), it is possible to derive predictions about the
biological communities (Sharma et al. 2011). The functional structure of fish assemblages in these habi-
role of environmental filters on freshwater fish tat types. In headwater streams, where environmental
assemblages was explored by Smith & Powell (1971) filtering seems to be more preponderant, and fish
who proposed the filter model, which was later modi- assemblages are composed of few hardy species, we
fied to include species traits (Poff 1997). According expect the coexistence of functionally similar species
to this model, fish species from the regional pool (functional clustering). By contrast, we may expect
must pass through a series of environmental filters to the coexistence of more functionally dissimilar spe-
establish themselves locally (Tonn et al. 1990). cies (functional overdispersion) in downstream areas,
Therefore, species from a local fish assemblage are where the environment is more stable and diverse,
expected to respond similarly to environmental fac- and fish assemblages tend to be richer.
tors and probably share many traits (Mouillot et al. In this study, we used a set of tropical headwater
2007). In this context, environmental filters can be streams and rivers to understand functional diversity
determinant factors in driving the composition of and functional structure patterns of their fish assem-
local fish assemblages. blages, trying to elucidate which ecological processes
Despite the role of environment in structuring fish drive their functional structure patterns. First, we
assemblages, ecologists have found evidence that the investigated how the functional diversity and the
spatial structure of fish assemblages is also a result of functional structure of fish assemblages are related to
spatial patterns. Sharma et al. (2011) found out that species diversity in both habitat types (headwater
assemblage composition of native and non-native fish streams and rivers). After that, we evaluated how the
species in temperate lakes is driven by different eco- variation of fish assemblages’ functional structure is
logical rules. While native fish assemblages are struc- related to different environmental variables. Finally,
tured by environmental conditions suggesting we explored the variation of fish assemblages’ func-
environmental filtering, non-native fish assemblages tional structure between headwater streams and riv-
are structured by human-mediated dispersal. Assess- ers. We hypothesise that headwater streams are
ing the role of environmental and spatial components primarily structured by environmental variables, lead-
through the use of variation partitioning (Borcard ing to functionally clustered assemblages, whereas
et al. 1992; Peres-Neto & Legendre 2010) combined rivers would be less influenced by the environment
with spatial statistical techniques [Moran’s Eigenvec- and would be structured by processes that favour the
tors Maps (MEM) Dray et al. 2006; Asymmetric existence of functionally overdispersed assemblages,
Eigenvectors Maps (AEM) Blanchet et al. 2008] is such as competition or species dispersal ability. Here,
now recognised as a key step for understanding the we tested how much of the variation in the functional
ecological processes that rule community composi- structure of fish assemblages is explained by multiple
tion patterns. components, namely environment, spatially structured
In freshwater systems, headwater streams can be environment and space. This kind of approach allows
considered as the most upstream areas in a river sys- us to identify which processes are more likely to
tem, and they are generally classified from first order structure fish assemblages, as the spatial component
up to third order (Vannote et al. 1980). Abiotic fac- and the spatially structured environment variation are
tors in headwaters – such as water velocity, substrate assumed to express the importance of neutral dynam-
type, dissolved oxygen, water temperature and trans- ics (Gilbert & Lechowicz 2004). On the other side,
port of particulate organic matter (Poff 1997) – may the environmental component may express the occur-
predominate over biotic interactions in structuring rence of important niche-based processes, such as
local fish assemblages (Ostrand & Wilde 2002; environmental filtering.

2
Functional structure of rivers and streams
modification of Horton’s scale (Petts 1994), and we
Material and methods
checked it using a geographical information system
map (1:250,000) available on the website of the
Study area
Sistema Estadual de Estatıstica e de Informacß~ ao
We conducted our study in the Tocantins-Araguaia Geografica de Goias (SIEG 2013).
River Basin (hereafter called Tocantins-Araguaia All headwater streams and rivers were sampled
Basin) that drains Goias State territory, Central Bra- during the dry season when the flows are low and
zil. The Tocantins-Araguaia Basin has a drainage fishes are captured more efficiently (Pease et al.
area of approximately 767,000 km2 with a mean 2012). Furthermore, the relationship between fish
annual discharge of 11,000 m³s 1 (Costa et al. assemblages’ structure and habitat structure in dry
2003). It has a well-defined dry season that occurs season is stronger (Willis et al. 2005). All samplings
between May and October and also a wet season that were conducted between May and September 2008,
extends between November and April (Albrecht & and the location of the sites was defined according to
Pellegrini-Caramaschi 2003; Quesada et al. 2004). In the possibility of access. Sampling locations were
this region, the knowledge about aquatic communities marked with an equipment of Global Position System
and its biodiversity is still very incipient (Tejerina- (GPS; Garmin 12).
Garro 2008). Fish species from headwater streams were sampled
with a seine net (4 m 9 1.5 m 9 1 cm of mesh
between opposing knots), while species from rivers
Data and sampling protocol
were collected with gillnets of different sizes of
In this paper, we used a data set relative to 27 head- opposing knots (10 m 9 1.5 m 9 12–70 mm).
water streams and 22 rivers located in the Tocantins- Seine nets are an efficient sampling technique for col-
Araguaia Basin (Fig. 1) at an average altitude of 450 lecting fish fauna when the focus is on ecological
and 390 m, respectively. We defined a headwater patterns of species richness and composition (Medei-
stream by its location in the river system, order ros et al. 2010), but the presence of rocks and woody
(1st–3rd order; Vannote et al. 1980) and drainage debris in the bottom substrate, as found in rivers,
area (10 < drainage area < 1000 km²; Meybeck often limits the application of this method for captur-
et al. 1996). We defined a watercourse as a river ing fish species (Tejerina-Garro & Merona 2000).
by its location in the river system, order (4th–6th These authors indicate also that the use of different
order) and drainage area (1000 < drainage area < mesh size of gillnets in repeated batteries is a useful
100,000 km2; Meybeck et al. 1996). We determined method that gives a reliable picture from the fraction
headwater stream and river orders using Strahler’s of the fish community sampled in rivers.

Fig. 1. Spatial distribution of headwater streams (circles) and rivers (triangles) sampled in the upper section of the Tocantins-Araguaia
River Basin, Central Brazil. The Tocantins-Araguaia basin is represented in the small map.

3
Carvalho & Tejerina-Garro
For each headwater stream, a reach of 50 m was we were not able to find information for a species, we
defined following Imhof et al. (1996), and the sam- extrapolated the data for genus or family level when-
pling effort was two persons per 50 m per 10 foot- ever it was possible. If extrapolation was not possible
steps. For rivers, the reach size was delimited as (a single species in the genus), species trait was con-
1000 m. All specimens collected were stored in iden- sidered as not available (NA).
tified plastic bags and were put in plastic drums with Afterwards, we constructed an environmental vari-
formaldehyde 10%. The specimens were separated ables matrix (local sites vs. environmental variables)
and identified using taxonomic keys and expert opin- with altitude (m), channel depth (m), water velocity
ion. Finally, specimens were stored in alcohol 70% at (cms 1), channel width (m), turbidity (NTU),
the Aquatic Biology Center of the Pontifıcia dissolved oxygen (mgl 1), pH, conductivity
Universidade Catolica de Goias and the Laboratory (lScm 1) and water temperature (°C). In each reach,
of Ichthyology of the Museum of Pontifıcia Univer- three transversal transects were delimited where each
sidade Cat olica do Rio Grande do Sul. environmental variable could be measured. Altitude
was measured only in the first transect with a GPS
(Garmin 12). In all transects, the following variables
Data matrices
were measured: channel width with a proper device
To perform the analyses in this paper, we used four to measure the distance (Bushnell Yardage Pro 500,
data matrices. First, we constructed a presence– Bushnell Company, Lenexa, KS, USA, for large riv-
absence matrix (species vs. local sites) and a func- ers), dissolved oxygen and temperature with an oxi-
tional traits matrix (species vs. traits). Although we meter (Lutron YK-22DO, Lutron Eletronic Entreprise
had species abundance, this kind of data can be rela- Co., Ltd., Taipei, Taiwan), pH with a pH meter
tively sensitive to different sampling techniques and (Lutron PH-208, Lutron Eletronic Enterprise Co.,
effort (Poff & Allan 1995). Therefore, to avoid any Ltd., Taipei, Taiwan), conductivity with a conductiv-
bias generated by different sampling techniques used ity meter (WTW 315i, Xylem Group, White Plains,
in headwater and river habitats, we did not include NY, USA), water velocity with a flowmeter (Flow-
abundance data in our analyses. We used eight traits meter Model 2030, General Oceanics Inc., Miami,
to construct our functional traits matrix: average body FL, USA) and channel depth with a measuring rope.
mass, trophic guild, water column position, parental Turbidity was measured only in the middle transect
care, foraging location, substrate preference, foraging of headwater streams and all transects of rivers, both
method and migration (Tables S1 and S2). Body mass with a digital turbidimeter (LaMotte 2020, LaMotte
was obtained directly from collected fish specimens, Company, Cherstertown, MD, USA). Each variable
and it was calculated as the average body mass found was measured in one location inside the transect.
for all specimens of the species collected. Trophic Finally, we constructed a site-by-edge matrix (local
guild was determined according to information pres- vs. link edges) that indicates the links among sites
ent in Melo (2011). This author identified the main according to a directional processes (for more details
component of diet for each species and then classified see Blanchet et al. 2008). We built our sites-by-edges
it in one of the following categories: algivorous, matrix according to the direction of the flow
omnivorous, carnivorous, herbivorous, detritivorous, (upstream–downstream) and considering that this
invertivorous and iliophagous. For species without flow direction is used by fish species for larval dis-
information in Melo (2011), we searched for the main persion (Agostinho et al. 2007).
component of species’ diet using an online database
(FishBase; Froese & Pauly 2014) and specialised liter-
Analyses
ature. Data for parental care, water column position,
foraging method and migration were taken from spec- For each fish assemblage, we calculated an index of
ialised literature and an online database for fishes functional diversity using the mean pairwise distance
(FishBase; Froese & Pauly 2014; see Table S1 for (henceforth MPD, Webb et al. 2002). Despite its
more details of traits categories). To determine sub- usual use in the calculation of phylogenetic diversity,
strate preference, we looked for all locations where a phylogeny and a functional dendrogram have simi-
the species were collected and identified local sub- lar structures and properties (Pavoine & Bonsall
strate(s). Then, in the species traits matrix, we marked 2010); therefore, this measure can be used to calcu-
all substrates where the species has occurred. Forag- late functional diversity substituting the phylogenetic
ing location was determined according to water velo- distance matrix by a functional distance matrix (Kem-
city. Watercourses with water velocity > 20 cms 1 bel et al. 2010; Pavoine & Bonsall 2010). Then, for
were considered as running waters (riffles and runs), calculating functional diversity using MPD, we
while those under this limit were considered as stand- substituted the phylogenetic distance matrix by a
ing waters or slower pools (see Rezende 2007). When functional distance matrix in the analysis (Kembel

4
Functional structure of rivers and streams
et al. 2010; Pavoine & Bonsall 2010) and used the null community has a NRI value, and all 999 values
following protocol: (i) conversion of species traits will be used to create a null distribution of NRI val-
matrix to a distance matrix, (ii) clustering distance ues expected at random. Then, the observed NRI
matrix into a functional dendrogram, (iii) transform- value can be compared with the null distribution. We
ing the functional dendrogram into a phylogeny, that considered as our regional pool the set of fish species
is, converting it into an object with phylogenetic found in all rivers and streams sampled.
properties, (iv) extracting distance matrix (‘cophenet- We performed two simple linear regressions for
ic matrix’) and (v) calculating functional diversity as each habitat type to examine the relationship of spe-
a measure of mean functional distances among all cies diversity with functional diversity and commu-
taxa that occur in a sample. We obtained the distance nity functional structure: (i) species richness versus
matrix using a modification of Gower’s distance as MPD and (ii) species richness versus NRI. MPD and
we had both qualitative and quantitative traits NRI were considered as response variables, while S
together (Pavoine et al. 2009) and the functional den- was the predictor variable. All analyses were per-
drogram using the unweighted pair group method formed in the R software (R Core Team 2013).
with arithmetic averages (UPGMA) as the clustering We performed a multiple linear regression between
method. NRI and environmental variables to explore whether
For each fish assemblage, we also calculated an environmental features are predicting the variation in
index of community functional structure using the net the functional structure among sites. We used a for-
relatedness index (henceforth NRI; Webb 2000; ward stepwise selection of the environmental vari-
Webb et al. 2002). The NRI quantifies the distribu- ables to reduce the cases of co-linearity (Oberdorff
tion of taxa in a sample relative to a species pool et al. 1995). We ran the multiple linear regression in
(Webb 2000) and gives the standardised effect of the R software (R Core Team 2013).
observed distances among species (Kembel & To test the contribution of the environment and
Hubbell 2006). The NRI for each locality can be space for NRI variation, we performed a variation
measured as 1[(MPD MPDnull)/SDMPDnull ], where partitioning analysis (Borcard et al. 1992; Peres-Neto
the MPD is the observed mean pairwise distance & Legendre 2010) using spatial descriptors and envi-
among species in the locality, MPDnull is the mean ronmental variables matrices as predictors and the
MPD for that locality in 999 null communities and NRI values as our response variable. To obtain the
SDMPDnull is the standard deviation of MPD for that spatial descriptors, we used the AEM method pro-
locality in 999 null communities (Webb et al. 2002). posed by Blanchet et al. (2008). It takes into account
This index is calculated by measuring the mean dis- the distance among watercourses and the direction of
tance of each species to all other species in the same the flow to represent dispersal patterns among sam-
community. Positive NRI values (P < 0.05) indicate pled sites. Based on our sites-by-edges matrix and
that co-occurring species in a community are more the spatial coordinates of sampled sites, we created a
functionally similar than expected by chance (func- connexion diagram linking the sites to one another
tional clustering), while negative values (P > 0.95) according to the asymmetric processes influencing
point to species less functionally similar than our response variable. Then, the connexion diagram
expected (Kembel & Hubbell 2006). was used to calculate the spatial descriptors. Given
To obtain the NRI, we used the functional traits the different nature and scales of our environmental
matrix to calculate a distance matrix with a modifica- variables, we performed a log-transformation to
tion of Gower’s distance (Pavoine et al. 2009) and increase linearity and avoid extreme values (Sharma
UPGMA to develop the functional dendrogram. Here, et al. 2011), excepting for pH values.
we used the same approach for MPD to convert the When a single variable is used as a predictor, the
functional dendrogram into a phylogeny and extract partition of the variance is based on a linear regres-
the ‘cophenetic’ matrix. Based on the ‘cophenetic’ sion, and it uses the adjusted R2 to estimate the
matrix, we utilised the function ses.mpd (Kembel unique and combined effects of environmental and
et al. 2010) available in R software (R Core Team spatial predictors (Borcard et al. 1992; Peres-Neto &
2013) to create a series of null communities for each Legendre 2010). The adjusted R2 values are capable
fish assemblage. Then, the observed degree of clus- to provide unbiased estimates on the real contribution
tering was compared with those expected by chance. of explanatory variables to the response variable (Oh-
We generated a set of 999 null communities for tani 2000; Peres-Neto et al. 2006). All these analyses
each fish assemblage using the independent swap were performed in R environment (R Core Team
algorithm (Gotelli & Entsminger 2001). This algo- 2013) using AEM package for Asymmetric Eigen-
rithm randomises the community data matrix main- vectors Maps (Blanchet & Legendre 2013) and var-
taining constant species occurrence frequency and part function present in vegan package for partition
species richness of each community. In the end, each analysis (Oksanen et al. 2013).

5
Carvalho & Tejerina-Garro
overdispersion (lower NRI values) of river fish
Results
assemblages are partially explained (48%) by an
Fish assemblages from headwater streams and riv- increase in species richness suggesting that functional
ers contained 150 fish species. Of these, 121 spe- patterns in this habitat type may also be associated
cies were found in rivers and 72 in headwater with other factors. For headwater streams, linear
streams. Thirty-one species occurred only in rivers, regressions indicated that S is not significantly corre-
while 12 species only in headwater streams. The lated with MPD (Fig. 3c; adjusted R2 = 0.23,
mean average of species per local site was 11.75 P = 0.22) or NRI (Fig. 3d; adjusted R2 = 0.20,
species considering both rivers and headwater P = 0.30).
streams. Taking into account only rivers, the mean The multiple regression demonstrated that environ-
average of species per local site was 14, while the mental variables explain 56% of the total variation in
mean average for headwater streams was 9.9 spe- NRI values (Table 1; adjusted R2 = 0.56,
cies per local site. P = 0.0001). Based on the forward stepwise selec-
The spatial patterns of functional grouping found tion, four environmental variables were indicated as
in this study showed that most rivers evaluated had influencing NRI variation: channel depth, water
negative NRI values (18 out of 22), whereas most of velocity, dissolved oxygen and turbidity (Table 1).
headwater streams (25 out of 27) had positive NRI Variation partitioning revealed that neither the
values (Fig. 2 and Table S3). According to our null local environment alone (fraction a, Table 2) nor the
model, eight of the 22 rivers (36.4%) presented a sig- spatially structured environment (fraction b, Table 2)
nificant functional grouping (P < 0.05), and six of is the best predictor to explain NRI variation. The
them (27.3%) had fish assemblages with functional space alone presented the higher fraction of explana-
overdispersion and two (9.1%) with functional clus- tion for NRI variation (fraction c, Table 2), but the
tering (Table S3). On the other hand, 10 of the 27 unexplained fraction of variation was also high
headwater streams (34.5%) presented a significant (fraction d, Table 2).
functional grouping (P < 0.05), with functional clus-
tering of fish assemblages (Table S3).
Discussion
For rivers, linear regressions showed that S is not
correlated with MPD (Fig. 3a; adjusted R2 = 0.30, Despite some recent efforts (Villeger et al. 2010,
P = 0.16), but it is negatively correlated with NRI 2012; Pease et al. 2012), functional diversity and
(Fig. 3b; adjusted R2 = 0.48, P = 0.02). This nega- functional structure of tropical fish assemblages
tive association shows that patterns of functional remain poorly explored. For headwater streams, our

Fig. 2. Spatial pattern of the NRI values of headwater streams (circles) and rivers (triangles) sampled in the Tocantins-Araguaia basin. Col-
ours of the circles and triangles correspond to significant (grey) and nonsignificant (white) cases of functional grouping at a 0.05 level
under the null model.

6
Functional structure of rivers and streams
(a) Headwater streams (b) Rivers
0.70 0.8

0.65
0.7
0.60
0.6
0.55
MPD

0.50 0.5

0.45
0.4
0.40
R 2 adj = 0.23; p = 0.22 0.3 R 2 adj = 0.30; p = 0.16
0.35

0.30 0.2
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 40 45
S S
(c) (d) 7
6
6
5
R 2 adj = 0.20; p = 0.30 5 R 2 adj = –0.48; p = 0.02
4 4
3
NRI

3
2
2 1
1 0
–1
0
–2
–1 –3
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 40 45
S S
Fig. 3. Linear regression results between species richness and two functional indexes: MPD (a–b) and NRI (c–d). Linear regressions were
performed for headwater (a–c) and river (b–d) fish assemblages. R2 adj refers to adjusted R2 of the linear regression.

Table 1. Statistics of the multiple linear regression between the functional thermore, functional structure patterns showed that
structure index (NRI) and environmental variables (adjusted R² = 0.56, 34.5% of the headwater streams sampled had a sig-
P < 0000).
nificant functional clustering, but almost all of the
streams exhibited positive raw values for NRI indi-
Functional Variable entered in the model
structure index (sign of the effect) P N
cating a tendency of fish assemblages to clustering.
Taking into account that headwater streams habitats
NRI Channel depth ( ) 0.0002 49 are highly affected by local environmental conditions
Water velocity (+) 0.0000 49 (Ostrand & Wilde 2002; Grenouillet et al. 2004),
Dissolved oxygen ( ) 0.0005 49
Turbidity ( ) 0.0010 49 such as channel depth and water velocity (Suarez
et al. 2007), it seems quite reasonable to think that
The sign of the effect indicates a positive (+) or negative ( ) effect of the environmental conditions drive the functional compo-
variable on the functional structure index.
sition patterns of local fish assemblages in this study.
If environmental conditions are leading to function-
Table 2. Results of variation partitioning on NRI values. ally clustered assemblages, then they are probably
acting as filters to fish species traits. In this case,
a b c d even assemblages with more species would be com-
posed of functionally similar species, and an increase
Fraction (%) 12.9 17.5 34.9 34.7
in species richness would not be accompanied by an
Fractions correspond to the variation explained by environment alone (a); increase in functional diversity or functional structure
environment spatially structured (b); space alone (c); and residuals (d). values. For rivers, functional trends were opposite to
Fractions are based on adjusted R squares (R2 adj) of each component. those observed in headwaters streams. Although only
27.3% of the rivers sampled displayed fish assem-
findings revealed that neither functional diversity nor blages with significant functional overdispersion,
functional structure of their fish assemblages is almost all of them presented negative raw values for
significantly correlated with species diversity. Fur- NRI suggesting a tendency to overdispersion. This

7
Carvalho & Tejerina-Garro
trend was partially explained by species diversity, strate type and mobility, organic matter input, flood
thus rivers with a higher number of species tend to intensity, extreme temperatures, etc. Future investiga-
present a higher degree of functional overdispersion. tions should explore how the functional structure of
In general, habitat diversity and stability tend to headwater streams fish assemblages is related to these
increase further downstream reaches (Meffe & Mink- environmental parameters. For rivers, the association
ley 1987; Jackson et al. 2001), and greater habitat between functional structure and environmental vari-
structural complexity can be associated with greater ables seems not to be clear as it is for headwaters.
morphological diversity (Willis et al. 2005). There- The tendency for functional overdispersion observed
fore, the functional overdispersion observed in rivers in rivers may be a consequence of the increase in
may be a result of the increase in the number of spe- species traits diversity due to a higher habitat struc-
cies with different functional traits due to habitat tural complexity (Willis et al. 2005). Given that
structural complexity. Nevertheless, we suggest that channel depth is usually considered as a good indica-
this proposal should be evaluated more carefully in tor of habitat structural complexity (Suarez et al.
future investigations as the average species richness 2007), we suggest that this variable could also be
found in this study for rivers was not much higher important for determining species co-occurrence in
than in headwaters. Moreover, species richness rivers.
explained only 48% of NRI trends suggesting that However, differences between headwater streams
functional structure in this type of habitat is probably and rivers related to environmental parameters are
affected by other factors. If both habitat types often associated to an expressive altitudinal differen-
(streams and rivers) are likely to be affected by dis- tial. In this study, sampled sites in both habitats have
tinct ecological factors, the main question here is: similar levels of altitude, suggesting that the relation-
Which processes are most strongly affecting their ship between the functional structure of fish assem-
functional structure? blages and the environment can be even stronger
Several studies have pointed out a major role of when the altitudinal gradient is more evident. In this
abiotic factors in shaping fish assemblages’ structure. case, we could expect a higher level of functional
Peres-Neto (2004) showed that habitat features are clustering in headwater streams.
more relevant than species interactions to determine Despite the significant role of the environment on
species co-occurrence. Poff & Allan (1995) demon- the functional structure of headwater streams and riv-
strated that trophic specialist fishes are more associated ers, it is worth noting that variation partitioning
with fish assemblages of stable watercourses, while results indicated a lower contribution of the environ-
trophic generalists are more numerous at sites with mental component. Furthermore, both spatial and
higher hydrological variability, whereas Hoeinghaus spatially structured environment components showed
et al. (2007) revealed that both habitat type and stabil- some contribution to functional structure variation.
ity have a powerful effect on fish assemblages and Together, the spatial component and the spatially
their functional organisation. Our results pointed to structured environment are supposed to express the
an important role of environmental variables on func- influence of neutral dynamics on community struc-
tional structure variation in headwater streams and ture (Gilbert & Lechowicz 2004; Diniz-Filho et al.
rivers. More specifically, four of the eight variables 2012). A neutral process guided by species dispersal
considered were strongly correlated with patterns in limitations through the space is likely to occur
functional structure: channel depth, dissolved oxygen, between headwater and river habitats. For example,
turbidity and water velocity. Water velocity and oxy- several species from rivers are not capable of reach-
gen have been presented as possible selective filters ing headwaters, whereas headwaters are known to
for species traits (Poff 1997). For headwater streams, support fish species that do not occur anywhere else
channel depth and water velocity are expected to in the river system (Paller 1994; Meyer et al. 2007).
affect species co-occurrence because they restrict Moreover, fish assemblages sampled in this study are
events of colonisation and influence the persistence located inside a region that is characterised by the
of species in this kind of habitat (Suarez et al. 2007). occurrence of many fish species with restricted spa-
Based on this background, we suggest that a potential tial range (Nogueira et al. 2010), in which dispersion
ecological process that might lead headwater stream probably occurs at short distances. Finally, although
fish assemblages to functional clustering is environ- the focus of this study is on a local scale, the entire
mental filtering, and at least three environmental vari- set of our sampled sites is distributed between two
ables are good candidates to act as selective filters watersheds (Tocantins and Araguaia). Even if they
for species functional traits: channel depth, water are capable of actively moving throughout different
velocity and dissolved oxygen. According to Poff habitats (Agostinho & Zalewski 1995), fish species
(1997), there are other environmental conditions that often have limited rates of dispersal between distinct
may act in the selection of species traits such as: sub- watersheds, which is imposed by the drainage area

8
Functional structure of rivers and streams
itself or shifts in environmental conditions (Matthews streams and rivers may be the explanation, although
& Robison 1988; Jackson et al. 2001). Under these the spatial structure of environmental variables may
circumstances, space also appears to play an impor- affect these patterns too. In this sense, the functional
tant role for the functional structure of headwater structure patterns observed for headwater streams and
streams and river fish assemblages. Limitations of rivers are probably structured by both environment
spatial species dispersion between headwaters and and space. Finally, we found, excepting for a partial
rivers seem to be a potential cause for this. It is worth contribution in the functional structure of river
to highlighting that the lower contribution of the assemblages, that species diversity is not related to
environmental component on variation partitioning functional diversity and functional structure trends in
can be a consequence of the fact that variables are both headwater stream and river fish assemblages.
spatially structured. Local variables as channel depth, These findings are relevant for conservation of these
water velocity, turbidity and/or dissolved oxygen systems. If sites with high species diversity do not
may present different levels for headwater streams harbour high functional diversity, the use of a conser-
and rivers (for example, rivers tend to have a deeper vation strategy based on a single aspect of diversity
channel than headwater streams). as a cure-all should be avoided.
It is necessary to consider the influence of two pos- This study opens the possibility for further and
sible biases on the results. First, differences observed challenging investigations, such as: (i) Are there
between rivers and headwater streams may be influ- other environmental variables that may act as selec-
enced by the different sampling methods used because tive filters in headwater streams? (ii) Are environ-
of the distinct remarkable features of the two habitats mental variables selecting the same functional traits
sampled. This may facilitate plausible fish species in different fish assemblages, that is, are fish assem-
selectiveness and differences on total number of indi- blages with functional clustering occupying the same
viduals sampled with potential consequences on spe- region of the functional traits space? (iii) Which func-
cies and/or functional richness. The second bias is tional traits are more related to these selective envi-
associated with the use of field data for fulfilling ronmental variables? (iv) Which traits are more
information about the functional data set, particularly related to the habitat structural complexity of rivers?
substrate preference and foraging location traits, that (v) How the altitudinal gradient may influence func-
may increase the association between functional indi- tional patterns between headwater and river habitats?
ces and the environmental component. However, we Answering these questions will help us to understand
call attention that these are the best data available for more completely how different processes are acting
the fish assemblages sampled as inventories of the fish on fish assemblages and how species traits are related
fauna for the sampled region are incomplete (Tejeri- to them in both freshwater habitats.
na-Garro 2008) and far from being well documented
(Vari 1988; Revenga & Kura 2003), as observed for
freshwater ichthyofauna in other regions of the world Acknowledgements
(Lev^eque et al. 2005; Dudgeon et al. 2006). We are grateful to all students, technicians and researchers
Our findings suggest that there is a co-existence of from Centro de Biologia Aquatica (CBA) that helped during
functionally similar species in fish assemblages of the fieldwork, especially Waldeir Francisco de Menezes. We
headwater streams, whereas a co-existence of func- also thank Dr. Frabricio Barreto Teresa for helping us with
tionally dissimilar species is more common in fish relevant suggestions in earlier versions of the manuscript; Dr.
assemblages of rivers. The environment seems to F. Guillaume Blanchet who helped with the understanding of
play an important role in driving these functional pat- the AEM framework and the construction of some R func-
tions; and Dr. Thiago Santos for several debates and ideas
terns and environmental filtering emerges as a poten-
about spatial analysis and R software. We thank to researchers
tial ecological process in determining functional from Pontifıcia Universidade Cat olica do Rio Grande do Sul
clustering in headwater habitats. This selective force who assisted in species identification. We thank anonymous
would be guided by, at least, three main environmen- reviewers for their contributions to improve the initial draft of
tal variables: channel depth, water velocity and dis- this study. To the National Council of Technological and Sci-
solved oxygen. For rivers, we argued that habitat entific Development – CNPq for the financial support given
structural complexity might be the environmental for the project (CNPq No. 471283/2006-1). RAC research is
condition driving the co-existence of dissimilar spe- supported with a scholarship by Coordenacß~ao de Aperfeicßoa-
cies as heterogeneous habitats may support species mento de Pessoal de Nıvel Superior – CAPES.
with different traits. Nevertheless, our results revealed
that spatial components also play a key role in deter-
References
mining the functional trends in both habitat types.
We suggest that a neutral dynamic process linked to Agostinho, A.A. & Zalewski, M. 1995. The dependence of
species dispersal limitations between headwater fish community structure and dynamics on floodplain and

9
Carvalho & Tejerina-Garro
riparian ecotone zone in Parana River, Brazil. Hydrobiologia structure: inferences based on taxonomic vs. functional
303: 141–148. groups. Journal of Biogeography 34: 324–338.
Agostinho, A.A., Marques, E.E., Agostinho, C.S., Almeida, Hubbell, S.P. 2001. The unified neutral theory of biodiversity
D.A., Oliveira, R.J. & Melo, J.R.B. 2007. Fish ladder of La- and biogeography. Princeton: Princeton University Press.
jeado Dam: migrations on one-way routes? Neotropical Ich- 448 p.
thyology 5: 121–130. Imhof, J.G., Fitzgibbon, J. & Annable, W.K. 1996. A hierar-
Albrecht, M.P. & Pellegrini-Caramaschi, E. 2003. Feeding quical evaluation system for characterizing watershed eco-
ecology of Leporinus taeniofasciatus (Characiformes: Anos- systems for fish habitat. Canadian Journal of Fisheries and
tomidae) before and after the installations of a hydroelectric Aquatic Sciences 53: 312–326.
plant in the upper rio Tocantins, Brazil. Neotropical Ichthy- Jackson, D.A., Peres-Neto, P.R. & Olden, K.D. 2001. What
ology 1: 53–60. controls who is where in freshwater fish communities – the
Blanchet, F.G. & Legendre, P. 2013. AEM: tools to construct roles of biotic, abiotic, and spatial factors. Canadian Journal
Asymmetric eigenvector maps (AEM) spatial variables. R of Fisheries and Aquatic Sciences 58: 157–170.
package version 0.4-5/r110. Kembel, S.W. & Hubbell, S.P. 2006. The phylogenetic struc-
Blanchet, F.G., Legendre, P. & Borcard, D. 2008. Modelling ture of a neotropical forest tree community. Ecology 87:
directional spatial processes in ecological data. Ecological S86–S99.
Modelling 215: 325–336. Kembel, S.W., Cowan, P.D., Helmus, M.R., Cornwell, W.K.,
Borcard, D., Legendre, P. & Drapeau, P. 1992. Partialling out Morlon, H., Ackerly, D.D., Blomberg, S.P. & Webb, C.O.
the spatial component of ecological variation. Ecology 73: 2010. Picante: R tools for integrating phylogenies and ecol-
1045–1055. ogy. Bioinformatics 26: 1463–1464.
Costa, M.H., Botta, A. & Cardille, J.A. 2003. Effects of large- Lev^eque, C., Balian, E.V. & Martens, K. 2005. An assessment
scale changes in land cover on the discharge of the Tocan- of animal species diversity in continental waters. Hydrobio-
tins River, Southeastern Amazonia. Journal of Hydrology logia 542: 39–67.
283: 206–217. Mason, N.W.H., Lanoiselee, C., Mouillot, D., Irz, P. & Argil-
Diniz-Filho, J.A.F., Siqueira, T., Padial, A.A., Rangel, T.F., lier, C. 2007. Functional characters combined with null mod-
Landeiro, V.L. & Bini, L.M. 2012. Spatial autocorrelation els reveal inconsistency in mechanisms of species turnover in
analysis allows disentangling the balance between neutral lacustrine fish communities. Oecologia 153: 441–452.
and niche processes in metacommunities. Oikos 121: 201– Matthews, W.J. 1986. Fish faunal “breaks” and stream order
210. in the eastern and central United States. Environmental Biol-
Dray, S., Legendre, P. & Peres-Neto, P.R. 2006. Spatial mod- ogy of Fishes 17: 81–92.
elling: a comprehensive framework for principal coordinate Matthews, W.J. & Robison, H.W. 1988. The distribution of
analysis of neighbour matrices (PCNM). Ecological Model- the fishes of Arkansas: a multivariate analysis. Copeia 2:
ling 196: 483–493. 358–374.
Dudgeon, D., Arthington, A.H., Gessner, M.O., Kawabata, Z.-I., Medeiros, E.S.F., Silva, M.J., Figueiredo, B.R.S., Ramos,
Knowler, D.J., Lev^eque, C., Naiman, R.J., Prieur-Richard, T.P.A. & Ramos, R.T.C. 2010. Effects of fishing technique
A.-H., Soto, D., Stiassny, M.L.J. & Sullivan, C.A. 2006. on assessing species composition in aquatic systems in
Freshwater biodiversity: importance, threats, status and con- semi-arid Brazil. Brazilian Journal of Biology 70: 255–262.
servation challenges. Biological Reviews 81: 163–182. Meffe, G.K. & Minkley, W.L. 1987. Persistence and stability
Froese, R. & Pauly, D., eds. 2014. FishBase. World Wide of fish and invertebrate assemblages in a repeatedly dis-
Web electronic publication. Available in: www.fishbase.org, turbed Sonoran Desert stream. American Midland Naturalist
version (04/2014). 117: 177–191.
Gilbert, B. & Lechowicz, M.J. 2004. Neutrality, niches, and Melo, T.L. 2011. Avaliacß~ao espacial das variaveis ambientais
dispersal in a temperate forest understory. Proceedings of e da estrutura trofica da ictiofauna de tributarios da bacia
the National Academy of Sciences of the United States of Tocantins-Araguaia, Brasil Central. S~ao Carlos: Unpublished
America 101: 7651–7656. Ph.D. dissertation, Universidade Federal de S~ao Carlos.
Gilliam, J.F. & Fraser, D.F. 2001. Movement in corridors: Meybeck, M., Friedrich, G., Thomas, R. & Chapman, D.
enhancement by predation threat, disturbance, and habitat 1996. Rivers. In: Chapman, D., ed. Water quality assess-
structure. Ecology 82: 258–273. ments: a guide to use of biota, sediments and water in envi-
Gotelli, N.J. & Entsminger, G.L. 2001. Swap and fill algo- ronmental monitoring. Paris: Unesco/Who/Unep, pp. 241–
rithms in null model analysis: rethinking the Knight’s Tour. 320.
Oecologia 129: 281–291. Meyer, J.L., Strayer, D.L., Wallace, J.B., Eggert, S.L., Helf-
Grenouillet, G., Pont, D. & Herisse, C. 2004. Within-basin fish man, G.S. & Leonard, N.E. 2007. The contribution of head-
assemblage structure: the relative influence of habitat versus water streams to biodiversity in river networks. Journal of
stream spatial position on local species richness. Canadian the American Water Resources Association 43: 86–103.
Journal of Fisheries and Aquatic Sciences 61: 93–102. Mouchet, M.A., Villeger, S., Mason, N.W.H. & Mouillot, D.
Helmus, M.R., Savage, K., Diebel, M.W., Maxted, J.T. & 2010. Functional diversity measures: an overview of their
Ives, A.R. 2007. Separating the determinants of phyloge- redundancy and their ability to discriminate community
netic community structure. Ecology Letters 10: 917–925. assembly rules. Functional Ecology 24: 867–876.
Hoeinghaus, D.J., Winemiller, K.O. & Birbaun, J.S. 2007. Mouillot, D., Dumay, O. & Tomasini, J.A. 2007. Limiting
Local and regional determinants of stream fish assemblage similarity, niche filtering and functional diversity in coastal

10
Functional structure of rivers and streams
lagoon fish communities. Estuarine, Coastal and Shelf Sci- Rahel, F.J. & Hubert, W.A. 1991. Fish assemblages and habitat
ence 71: 443–456. gradients in a Rocky Mountain – Great Plains stream: biotic
Nogueira, C., Buckup, P.A., Menezes, N.A., Oyakawa, O.T., zonation and additive patterns of community change. Trans-
Kasecker, T.P., Neto, M.B.R. & Silva, J.M.C. 2010. actions of the American Fisheries Society 120: 319–332.
Restricted range fishes and the conservation of Brazilian Revenga, C. & Kura, Y. 2003. Status and trends of biodiver-
freshwaters. PLoS One 5: e11390. sity of inland water ecosystems. Montreal, Canada: Secre-
Oberdorff, T., Guegan, J.-F. & Hugueny, B. 1995. Global tary of the Convention on Biological Diversity, Technical
scale patterns of fish species richness in rivers. Ecography Series no. 11.
18: 345–352. Rezende, C.F. 2007. Estrutura da Comunidade de Macroinver-
Ohtani, K. 2000. Bootstrapping R2 and adjusted R2 in regres- tebrados associados ao folhicßo submerso de remanso e cor-
sion analysis. Economic Modelling 17: 473–483. renteza em igarapes da Amaz^ onia Central. Biota Neotropica
Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, O., Minchin, 7: 301–305.
P.R., O’Hara, R.B., Simpson, G.L., Solymos, P., Henry, M., Sharma, S., Legendre, P., Caceres, M. & Boisclair, D. 2011.
Stevens, H. & Wagner, H. 2013. Vegan: community ecology The role of environmental and spatial processes in structur-
package. R package version 2.0-8. Available in: http:// ing native and non-native fish communities across thousands
CRAN.R-project.org/package=vegan of lakes. Ecography 34: 762–771.
Ostrand, K.G. & Wilde, G.R. 2002. Seasonal and spatial vari- SIEG. 2013. Sistema Estadual de Estatıstica e Informacß~ oes
ation in a prairie stream-fish assemblage. Ecology of Fresh- Geograficas de Goias. Available in: http://www.sieg.go.gov.
water Fish 11: 137–149. br
Paller, M.H. 1994. Relationships between fish assemblage Smith, C.L. & Powell, C.R. 1971. The summer fish communi-
structure and stream order in South Carolina coastal plain ties of Brier Creek, Marshall County, Oklahoma. American
streams. Transactions of the American Fisheries Society Museum Novitates 2458: 1–30.
123: 150–161. S
uarez, Y.R., Valerio, S.B., Tondato, K.K., Ximenes, L.Q.L.
Pavoine, S. & Bonsall, M.B. 2010. Measuring biodiversity to & Felipe, R.T.A. 2007. Determinantes ambientais da
explain community assembly: a unified approach. Biological ocorr^encia de especies de peixes em riachos de cabeceira da
Reviews 86: 792–812. bacia do rio Ivinhema, alto rio Parana. Acta Scientiarum
Pavoine, S., Vallet, J., Dufour, A.-B., Gachet, S. & Daniel, H. Biological Sciences 29: 145–150.
2009. On the challenge of treating various types of vari- Taylor, C.M. 1996. Abundance and distribution within a guild
ables: application for improving the measurement of func- of benthic stream fishes: local processes and regional pat-
tional diversity. Oikos 118: 391–402. terns. Freshwater Biology 36: 385–396.
Pease, A.A., Gonzalez-Dıaz, A.A., Rodiles-Hernandez, R. & Taylor, C.M. 1997. Fish species richness and incidence pat-
Winemiller, K.O. 2012. Functional diversity and trait–envi- terns in isolated and connected stream pools: effects of
ronment relationships of stream fish assemblages in a large pool volume and spatial position. Oecologia 110: 560–
tropical catchment. Freshwater Biology 57: 1060–1075. 566.
Peres-Neto, P.R. 2004. Patterns in the co-occurrence of fish Tejerina-Garro, F.L. 2008. Biodiversidade e impactos ambi-
species in streams: the role of site suitability, morphology entais no estado de Goias. In: Rocha, C., Tejerina-Garro,
and phylogeny versus species interactions. Oecologia 140: F.L., Pietrafesa, J.P., eds. Cerrado, Sociedade e Meio Ambi-
352–360. ente: Desenvolvimento Sustentavel em Goias. Goi^ania: Edi-
Peres-Neto, P.R. & Legendre, P. 2010. Estimating and con- tora da UCG, pp. 15–47.
trolling for spatial structure in the study of ecological com- Tejerina-Garro, F.L. & Merona, B. 2000. Gill net sampling
munities. Global Ecology and Biogeography 19: 174–184. standardisation in large rivers of French Guiana (South
Peres-Neto, P., Legendre, P., Dray, S. & Borcard, D. 2006. America). Bulletin Francais de la P^eche e de la Psiculture
Variation partitioning of species data matrices: estimation 357/358: 227–240.
and comparison of fractions. Ecology 87: 2614–2625. Tonn, W.M., Magnuson, J.J., Rask, M. & Toivonen, J. 1990.
Petts, G.E. 1994. Rivers: dynamic components of catchment Intercontinental comparison of small-lake fish assemblages:
ecosystems. In: Calow, P., Petts, G.E., eds. The river hand- the balance between local and regional processes. The
book, Vol. 2. Oxford: Blackwell Scientific, pp. 3–22. American Naturalist 136: 345–375.
Poff, N.L. 1997. Landscape filters and species traits: towards Vannote, R.L., Minshall, G.W., Cummings, K.W., Sedell, J.R.
mechanistic understanding and prediction in stream ecology. & Cushing, C.E. 1980. The river continuum concept.
Journal of the North American Benthological Society 16: Canadian Journal of Fisheries and Aquatic Sciences 37:
391–409. 130–137.
Poff, N.L. & Allan, J.D. 1995. Functional organization of Vari, R.P. 1988. The Curimatidae, a lowland neotropical fish
stream fish assemblages in relation to hydrological variabil- family (Pisces: Characiformes); distribution, endemism, and
ity. Ecology 76: 606–627. phylogenetic biogeography. In: Vanzolini, P.E., Heyer,
Quesada, C.A., Miranda, A.C., Hodnett, M.G., Santos, A.J.B., W.R., eds. Proceedings of a Workshop on Neotropical Dis-
Miranda, H.S. & Breyer, L.M. 2004. Seasonal and depth tribution Patterns. Rio de Janeiro: Academia Brasileira de
variation of soil moisture in a burned open savanna (campo Ci^encias. pp. 343–377.
sujo) in Central Brasil. Ecological Applications 14: 33–41. Villeger, S., Miranda, J.R., Hernandez, D.F. & Mouillot, D.
R Core Team 2013. R: A language and environment for statis- 2010. Contrasting changes in taxonomic vs. functional
tical computing. Vienna, Austria: R Foundation for Statisti- diversity of tropical fish communities after habitat degrada-
cal Computing. Available in: http://www.R-project.org/. tion. Ecological Applications 20: 1512–1522.

11
Carvalho & Tejerina-Garro
Villeger, S., Miranda, J.R., Hernandez, D.F. & Mouillot, D. Supporting Information
2012. Low functional b-diversity despite high taxonomic b-
diversity among tropical estuarine fish communities. PLoS Additional Supporting Information may be found in
One 7: e40679. the online version of this article:
Webb, C.O. 2000. Exploring the phylogenetic structure of Table S1. Fish traits used to calculate mean pair-
ecological communities: an example for rain forest trees. wise distance (MPD) index of fish assemblages
The American Naturalist 156: 145–155. located in headwater streams and rivers, Tocantins-A-
Webb, C., Ackerly, D., McPeek, M. & Donoghue, M. 2002. raguaia River Basin, Brazil.
Phylogenies and community ecology. Annual Review of Table S2. Description of the eight functional traits
Ecology and Systematics 33: 475–505. and their respective categories used to calculate func-
Willis, S.C., Winemiller, K.O. & Lopez-Fernandez, H. 2005. tional diversity (MPD) and functional structure (NRI)
Habitat structural complexity and morphological diversity of
of headwater stream and river fish assemblages.
fish assemblages in a Neotropical floodplain river. Oecolo-
gia 142: 284–295.
Table S3. Results found during NRI calculation
Winston, M.R. 1995. Co-occurrence of morphologically simi- via null model approach for fish assemblages of
lar species of stream fishes. The American Naturalist 145: headwaters streams and rivers of the Tocantins-Ara-
527–545. guaia basin.

12

You might also like