Hydrological connectivity drives patterns of macroinvertebrate biodiversity in
floodplain rivers of the Australian wet/dry tropics
Aquatic macroinvertebrates in floodplain rivers of the wet/dry tropics
1. Catherine Leigh (corresponding author)
Australian Rivers Institute, and the Griffith School of Environment
Nathan, Queensland 4111
2. Fran Sheldon
Australian Rivers Institute and the Griffith School of Environment
1. Floodplain rivers in Australia’s wet/dry tropics are regarded as being among the
most ecologically intact and bio-diverse lotic ecosystems in the world, yet there have
been relatively few community-based studies of their aquatic fauna.
2. To investigate relationships between hydrological connectivity and biodiversity in
the region, macroinvertebrates were collected from sites within two contrasting
floodplain rivers, the ‘tropical’ Gregory River and ‘dryland’ Flinders River systems,
during the dry season and analysed at various spatial scales. A subset of sites was re-
sampled in the following dry season to explore temporal variation. The fauna
consisted of 124 morphotaxa, dominated by gatherers and the Insecta.
3. As predicted, hydrological connectivity (the lotic or lentic status of waterbodies)
had a major influence on macroinvertebrate assemblage structure and diversity, both
in space and time. Assemblages from waterbodies with similar connection histories
were most alike, and beta-diversity between assemblages was greatest between lotic
and lentic waterbodies, tending to increase with increasing spatial separation.
4. At smaller spatial scales, a number of within-waterbody, habitat and water quality
characteristics were important for explaining variation (61 %) in the taxonomic
organisation of assemblages, and characteristics associated with primary productivity
and habitat diversity were important for explaining variation (45 %) in the functional
organisation of assemblages. However, much of the small-scale environmental
variation across the study region appeared to be related to broad-scale variation in
hydrological connectivity, which had both direct and indirect effects on
5. Conservation of the biodiversity in Australia’s wet/dry tropics may depend on
conserving the natural variation in hydrological connectivity and the unregulated flow
of floodplain rivers.
Keywords: beta-diversity, dryland rivers, functional feeding groups, multiple scales,
Understanding spatiotemporal variation in patterns of biota and their relationships
with the environment is a key theme of riverine ecology (Ward, 1989; Poff, 1997;
Ward & Tockner, 2001; Thorp, Thoms & Delong, 2006) including that of floodplain
rivers (Ward, Tockner & Schiemer, 1999; Amoros & Bornette, 2002). For example,
variation in patterns of biodiversity within unregulated floodplain rivers is related to
the complex hydro-geomorphology of such systems and their changing connection
and disconnection through space and time (Ward et al., 1999). We can describe this
variation in terms of beta-diversity or the turnover in biotic composition (taxonomic
or functional) between any two habitats (within or between rivers). Maximum beta-
diversity theoretically occurs at some intermediate level of connectivity between
habitats (Ward et al., 1999; Ward & Tockner, 2001), as may result from different
states of hydrological connection in space and time. However, different types and
levels of variation may be associated with different types of river systems.
In dryland rivers with large and active floodplain zones, differentiation among biotic
assemblages (beta-diversity) can be explained by the ‘connectivity potential’ between
habitats, a combination of spatial separation and the historical frequency of
hydrological disconnection (Marshall et al., 2006). For tropical rivers with active
floodplains, however, our understanding of biodiversity patterns stems from concepts
developed specifically for, and from, these river types (e.g. Ezcurra De Drago,
Marchese & Wantzen, 2004). The most influential of these is the Flood Pulse Concept
(Junk, Bayley & Sparks, 1989). This model predicts that biodiversity in floodplain
waterbodies is greater than that of main channels due to greater variability within
floodplain habitats. In neo-tropical rivers, this has indeed been demonstrated to occur
(Ezcurra De Drago et al., 2004). In addition, assemblage structure appears to persist in
these systems throughout wet and dry seasons (Melo & Froehlich, 2001). Despite
these insights, however, it is uncertain how the interaction between spatial scale (e.g.
Boyero & Bailey, 2001) and hydrological connectivity (e.g. Dos Santos & Thomaz,
2007) influences beta-diversity in tropical floodplain systems.
Flow regime plays a major role in structuring patterns of biotic composition and
diversity in riverine ecosystems (Poff et al., 1997; Puckridge et al., 1998). Tropical
rivers generally have regular flow regimes (Latrubesse, Stevaux & Sinha, 2005),
whereas dryland rivers are characterised by flow variability (Puckridge et al., 1998).
Therefore, biodiversity patterns in tropical rivers may not be influenced by variation
in hydrological connectivity, and the ‘connectivity potential’ between habitats, to the
same extent as patterns in dryland rivers. This has implications for regions that
include both ‘tropical’ and ‘dryland’ river types. The wet/dry tropics in northern
Australia is one such region: many of the floodplain rivers here have flow regimes
that can be described as typically ‘tropical’ (more permanent with regular flow
regimes) or ‘dryland’ (more ephemeral with greater flow variability) (Leigh &
Sheldon, 2008). As such, it is likely that spatiotemporal patterns of variation in the
biotic assemblages of these systems will differ between the contrasting river types.
Floodplain river systems in Australia’s wet/dry tropics are regarded as among the
most ecologically intact and bio-diverse lotic ecosystems in the world (Woinarski et
al., 2007), yet they have been the focus of relatively few community-based studies
(Marchant, 1982; Outridge, 1988; Paltridge et al., 1997; Erskine et al., 2005). To
increase our understanding of biodiversity patterns and hydrological connectivity
within these systems, macroinvertebrates were collected from two contrasting
floodplain rivers in the southern Gulf of Carpentaria (the ‘tropical’ Gregory River and
‘dryland’ Flinders River systems) across consecutive dry seasons and analysed at
various spatial scales. Specifically, we predicted that the two river systems (‘tropical’
versus ‘dryland’) would be associated with different patterns of variation in
assemblage structure and diversity; that assemblages from waterbodies with different
states of hydrological connection (lotic versus lentic waterbodies) would show more
differentiation in structure and diversity than those from waterbodies with the same
flow status (lotic versus lotic, lentic versus lentic); and that beta-diversity in
floodplain habitats (‘off-channel’ waterbodies) would be much greater than in main
channels and would be associated with greater habitat diversity in the floodplain.
More generally, we also explored: (a) the effect of spatial scale on beta-diversity; (b)
relationships between assemblage structure and environmental conditions at spatial
scales smaller than catchment, and; (c) temporal change in assemblage structure
between the two dry seasons.
Study area and design
Australia’s wet/dry tropics are located north of the Tropic of Capricorn and are
comprised of savannah and dry forest (Fig. 1). Floodplain catchments in this region,
including those in the Gulf of Carpentaria drainage division in Australia’s northeast,
experience an annual cycle of monsoonal rains, high flows and flooding in the wet
season (c. Nov-Apr) followed by a dry season (c. May-Oct) of low flows and virtually
no rainfall. Hydrological analysis of rivers here suggests there are two dominant flow
patterns—the more regular ‘tropical’ rivers, and the more ephemeral ‘dryland’ rivers
(Leigh & Sheldon, 2008). During the dry season, flow tends to cease in the ‘dryland’
rivers, with both channels and floodplains becoming a mosaic of disconnected
waterholes. This occurs to a much lesser extent in the floodplains of the ‘tropical’
rivers. From within these broad groups, we studied two large river systems in the
Nicholson (52 300 km2) and Flinders (109 400 km2) catchments (Fig. 1), in southern
Gulf of Carpentaria. The clear-flowing, perennial and aquifer-fed Gregory River and
Beames Brook in the Nicholson catchment are more typically ‘tropical’, whereas the
turbid Flinders and Cloncurry Rivers in the Flinders catchment are more typically
Eleven waterbodies were sampled across the lower freshwater sections of the remote
Gregory and Flinders river systems during the 2005 dry season (August) (Fig. 1).
Waterbodies were classified as either lotic or lentic, representing their hydrological
connection or disconnection at the time of sampling (hereafter ‘flow status’). Codes
were used to represent the catchment (Gregory = G, Flinders = F), river section
(downstream = D, mid = M, upstream = U) and lateral position (main channel = m,
off-channel = o) of each site (Table 1). Lotic waterbodies tended to have long runs
either side of a deep pool, whereas lentic waterbodies were typically reduced to
shallow pools. However GDm was a long run, and both GDm and GUm had riffles at
their downstream ends. Four of the 11 waterbodies (two in each catchment) were re-
sampled in the dry season of 2006 (in September) (Fig. 1). Thus, the study design
included one temporal scale (2005 versus 2006 dry season) along with four spatial
scales (catchment, waterbody, within waterbody and within habitat) that were
described by environmental conditions (see below). Under this design, differences
between the Gregory and Flinders river systems could not be attributed to categorical
differences between all ‘tropical’ and ‘dryland’ river types. However, the design
enabled our predictions about patterns of biodiversity and hydrological connectivity in
the study region to be explored and allowed us to formulate testable hypotheses about
our study systems and about others with similar flow regimes (cf. Hargrove &
Macroinvertebrates were sampled from all habitat types present at each waterbody
(e.g. bare littoral, snags, leaf litter, aquatic macrophytes) using a (semi-quantitative)
patch-weighted composite-habitat protocol (Marshall et al., 2006). The littoral
distance covered by each habitat type was estimated and proportional distances for
sampling were then allocated to each habitat type. These distances were swept with a
500 µm dip-net and produced samples that represented an entire waterbody, allowing
comparison among samples and waterbodies. Three samples were collected at each
waterbody, to give 45 samples in total (three samples each from 11 waterbodies in
2005; three samples each from four waterbodies in 2006). Samples were preserved in
70 % aqueous methanol for later identification in the laboratory.
All macroinvertebrates (aquatic invertebrates > 500 µm) were sorted from detritus and
sediment under a dissecting microscope, identified according to taxonomic and
functional feeding group (FFG) classifications (Cummins & Klug, 1979; Merritt &
Cummins, 1996; Hawking, 2000) and counted. Identification was performed to the
lowest taxonomic level practicable, given keys, life-history stage and condition. This
was most often to genus or species. However, where keys were incomplete or not
specific to the study region, or individuals were too small (e.g. tiny Zygoptera) or had
lost vital parts (e.g. many mayfly larvae had broken legs and antennae), identification
was to a higher taxonomic level (but representative of morphotaxa at lower levels of
resolution where practicable). Voucher specimens of all taxa were retained as a
reference collection at Griffith University.
Hydrographs, produced using mean daily flow data (megalitres per day standardised
by upstream catchment area, ML d-1 km-2) from gauging stations in and around the
study region (DNRM, 2005), were first used to compare flow regimes and likely
connection histories among waterbodies and catchments. Although continuous daily
flow data were available only until the late 1980s, the purpose was to assess typical
patterns in the flow regime (sensu Puckridge et al., 1998) rather than assess flow
records that corresponded directly with macroinvertebrate sampling times.
Secondly, hydro-geomorphological and biophysical measures were used to describe
waterbodies at spatial scales smaller than the catchment. These were visually
estimated or taken by direct or remote survey and included waterbody (seven
variables), within-waterbody (eight variables) and macroinvertebrate habitat (12
variables) scale measures (Appendix S1). Additionally, three replicate samples each
of littoral zone sediment and biofilm were collected and analysed for chlorophyll a
concentration using standard methods (APHA, 1989). Concentrations were converted
to median areal values, and these were included in the set of macroinvertebrate habitat
variables (14 in total) (Appendix S1).
Water quality characteristics (15 in total) were described using a number of variables
(Appendix S1). Conductivity, salinity and pH were measured from a mid-channel
location at each waterbody using a multi-parameter sonde (YSI 600XLM in 2005 and
YSI 6920 in 2006, Yellow Springs, Ohio, USA). Dissolved oxygen concentration was
also recorded but data were unreliable due to a faulty probe. Light irradiance (E,
photosynthetic radiation at 400 – 700 nm) was measured as a function of depth (z)
with a 2-pi sensor and light meter (Li-cor Li-1400, Lincoln, NE, USA) to determine
the euphotic zone depth (equivalent of 1 % of surface irradiance). Light attenuation
(k) was first determined by fitting a regression to the measured irradiance and depth
data, using the exponential equation: ln (Ez) = -k(z) + ln (E0), where Ez is the
irradiance at depth z, and E0 is the irradiance at the surface of the waterbody (Kirk,
2003). The euphotic depth (ED) was then calculated by substituting 1 % of the surface
irradiance value for Ez. Three samples of surface water from the mid-channel location
were collected and analysed for median concentrations of chlorophyll a, total and
dissolved nitrogen and phosphorus (N and P), and organic and inorganic fractions of
suspended solids using standard methods (APHA, 1989).
The approach used in this study was similar to that of previous studies exploring
relationships between spatial and temporal patterns of biotic assemblages and
environmental factors (e.g. Marshall et al., 2006). Three types of pattern were
explored: assemblage composition (based on taxonomic abundances), functional
composition (based on the proportionate representation of FFGs calculated from
abundance data), and diversity (based on diversity measures calculated from
abundance data). For individuals classified by more than one FFG (e.g. elmid beetles
were considered both grazers and gatherers), their abundance was shared equally
among these groups (e.g. one elmid beetle = 0.5 grazer + 0.5 gatherer) before
calculating FFG proportions (Dudgeon, 1994). Diversity measures included richness
(S), abundance (N), Margalef’s index of richness [D = (S-1)/ln N], the Berger-Parker
index of dominance [BP = Nmax/N, where Nmax = the number of individuals in the
most abundant taxon] and a simple measure of beta-diversity [ = (S/Sav) – 1, where
Sav is the average richness of the sample units used to calculate S]. Margalef’s index
(D) was used to incorporate evenness and richness into one measure, and the Berger-
Parker index (BP) provided an indication of the unevenness between richness and
abundance within a sample (Magurran, 1988). Beta-diversity () was calculated
within-waterbody (between-sample) and between-waterbodies. Maximum occurs
when no taxa are shared amongst samples, and minimum (= 0) occurs when all
sample units share all the same taxa (McCune, Grace & Urban, 2002).
Multivariate analyses were used to explore patterns of variation in taxonomic and
functional composition of assemblages and included analysis of similarities and
similarity percentages (ANOSIM and SIMPER), clustering (unweighted pair group
method with arithmetic mean, UPMGA), ordination (non-metric multidimensional
scaling, MDS) and correlation (BIOENV) techniques in the PRIMER-5 software
package (PRIMER-E, 2002). Patterns of variation in diversity were explored using
univariate techniques (analysis of variance, ANOVA) in SAS (SAS Institute, 2002).
Spatial patterns were explored using a two-way factorial ANOSIM (with up to 999
permutations) to test for differences in assemblage composition between groups
within a priori-defined factors (‘catchment’: Gregory versus Flinders; ‘flow status’:
lotic versus lentic). Groups were based on Bray-Curtis dissimilarity matrices of
log10x-transformed data for both taxonomic abundances (x +1) and FFG proportions
(x). Patterns of variation among groups were visualised using MDS with default
settings and 100 random starts. Ordination solutions were displayed in two
dimensions when stress was low (< 0.2) and accompanied by dendrograms produced
from agglomerative, hierarchical cluster analyses based on group-averaged Bray-
Curtis dissimilarity scores (UPGMA) (McCune et al., 2002). SIMPER was used to
identify key taxa contributing to the average dissimilarity between groups that were
significantly different (ANOSIM p < 0.05).
Analysis of variance (ANOVA) was used to determine the effect of the a priori-
defined factors on macroinvertebrate diversity measures (S, N, D and BP). All factors
were included in each ANOVA model, and non-significant factors were removed
stepwise until the most parsimonious and significant model (lowest P-value) was
achieved. Data were transformed as necessary (e.g. log and arcsine square root
transformations) to comply with ANOVA assumptions (normality and homogeneity
of variance) and least squares means were used due to unequal replication among
groups within factors. Where multiple comparisons were made between pairs of
sample groups or factors (in ANOVA and ANOSIM), their significance was tested
using the Bonferonni t-test, which controlled the experiment-wise error rate across all
paired comparisons (Neter, Wasserman & Kutner, 1985; Montgomery, 2001).
Two issues narrowed the breadth of analyses performed. Firstly, the multivariate
interaction effect between catchment and flow status was not tested, as this requires a
balanced number of samples within groups. Secondly, nested (hierarchical) designs
are common in ecological studies, and for this study, the factor ‘lateral position’ (main
or off-channel waterbody) was nested within catchment. Anderson (2001) provides a
method of nonparametric multivariate analysis of variance (NPMANOVA), based on
Bray-Curtis dissimilarity scores, that can be used to examine nested designs and
effects of main and interaction terms. However, if flow status was found to have a
significant effect, this method could not be used due to main and off-channel
waterbodies being inconsistently lotic or lentic. Thus, the effect of lateral position
could not be partitioned from that of flow status. However, there was enough
consistency within the subset of lentic Flinders waterbodies to test for differences
between main (FUm and FMm) and off-channel (FUo and FDo) locations in a
balanced design using a one-way ANOSIM and SIMPER. Additionally, as ANOVA
can cope with unbalanced designs, the effect of lateral position on diversity measures
could be tested within lotic waterbodies in the Gregory catchment, as well as lentic
waterbodies in the Flinders.
Variation in macroinvertebrate assemblages across a hierarchy of spatial scales
(within and between waterbodies, both within and between catchments) was explored
using pair-wise Bray-Curtis dissimilarity scores (as a measure of beta-diversity) based
on log-transformed abundance and FFG proportion data (Marshall et al., 2006).
Sample data were averaged across waterbodies before calculating between-waterbody
dissimilarity scores. There is no simple test available to compare Bray-Curtis
dissimilarity scores at different scales particularly with low level replication (n = 3) at
the base level (within-waterbody) (cf. Underwood & Chapman, 1998; Marshall et al.,
2006). Thus, differences among scales were interpreted by examining ranges of Bray-
Curtis dissimilarity scores with box-plots.
Relationships between assemblages and their environment at spatial scales smaller
than catchment (waterbody, within-waterbody, macroinvertebrate habitat, and water
quality characteristics) were investigated using generalised Mantel tests with Monte
Carlo randomisations (BIOENV). However, as the number of environmental variables
within each scale was large (Appendix S1), Spearman’s rank correlation coefficients
(rs) were calculated prior to the BIOENV analyses to avoid variable redundancy.
Variables with the greatest potential ecological importance acted as surrogates for
those variables with which they were highly correlated (rs 0.9) (Clarke & Warwick,
2001). The BIOENV analysis described the association between the assemblage data
matrices (log-transformed abundance or proportional FFG data subjected to the Bray-
Curtis dissimilarity measure) and environmental matrices (range-standardised
variables within each scale, transformed if necessary to reduce skew below 1, and
subjected to the normalised Euclidean distance measure) (Clarke & Warwick, 2001).
The amount of variation in assemblage patterns explained by combinations of
environmental variables was estimated as the square of the BIOENV correlation
coefficient (rs) (cf. Marshall et al., 2006). Additionally, the combinations of variables
at each environmental scale that explained the most variation in the biotic datasets
(highest BIOENV rs > 0.4) were combined for an overall BIOENV analysis.
Temporal variation was explored by comparing spatial patterns in assemblage
composition seen in the 2005 (all waterbodies, and the smaller subset of GUm, GUo,
FUm and FUo) and 2006 datasets (GUm, GUo, FUm and FUo), using analyses
detailed above. Additionally, ‘year’ was included as an a priori-defined factor (2005
versus 2006) in ANOSIM and ANOVA analyses. Interaction terms were not tested
due to insufficient replication within groups, which also affected the ability to test for
differences in assemblages between main and off-channel locations across years,
particularly in the presence of a significant effect of flow status (see above).
The most obvious differences between waterbodies in their environmental
characteristics were associated with broad-scale hydro-geomorphology and the
Gregory and Flinders Rivers’ flow regimes. Flow records from FDm, the only lotic
Flinders waterbody at the time of sampling (also known as Walkers Bend), showed
that dry season periods of zero flow are usual at this site and follow a similar pattern
as experienced at those immediately upstream of the study region (at the Cloncurry
River at Canobie and Flinders River at Etta Plains gauging stations) (Fig. 2). Indeed,
the only time in the period of continuous daily flow records from Walkers Bend (1969
– 89) when flow was experienced in the same month as the majority of sampling
conducted for this study (August), was in 1988 for 22 days. Although the gauging
station at FDm is the only one within the Flinders sampling area, the similarity
between flow patterns at this waterbody and the nearby upstream stations suggested
that macroinvertebrate assemblages from all the sampled Flinders waterbodies were
likely to have experienced a similar connection history (featuring variable lengths of
disconnection most dry seasons). Additionally, this history would be in strong
contrast to that of the lotic Gregory waterbodies, which permanently connected. In
fact, over the same 16 year time period of continuous daily flow records (30/9/72 –
30/9/88), Walkers Bend had zero flow days 59.4 % of the time, Canobie 72.5 % and
Etta Plains 69.5 % (Fig. 2). This was in contrast to no zero flow days (0 %)
experienced at Gregory River at Gregory Downs, the gauging station immediately
upstream of the Gregory sampling area (Fig. 2).
At the waterbody and within-waterbody scales of resolution, a number of variables
also appeared related to the broader-scale influence of hydrological connection and
disconnection. This included depth, many water quality variables, and
macroinvertebrate habitat variables such as macrophyte presence or absence. In
general, lotic Gregory waterbodies were deeper and appeared to have lower
concentrations of nutrients, suspended solids (but greater % organic solids) and algal
biomass in the water column, lower pH, and to have more abundant and diverse
vegetation in littoral zones than lentic and Flinders waterbodies, which appeared to
have greater proportions of bare sediment and snags. These general trends were
observed in the 2005 dry season (as displayed by a number of variables used in the
BIOENV analyses below) and in 2006 (Table 2).
In total, 48 669 individuals were identified from the 2005 and 2006 dry season
samples, representing 124 morphotaxa (Appendix S2). These included a total of 35
664 individuals and 119 morphotaxa from the 33 samples collected from 11
waterbodies in the 2005 dry season. Within these samples, Insecta dominated the
abundance (45 %) and richness (79 %). Crustacea made up 38 % of the total
abundance, followed by Mollusca (12 %), with these latter two groups comprising 8
and 9 % of the total richness, respectively. Among the Insecta, Diptera were the most
abundant (65 %) and Coleoptera the most diverse (33 %); however, most families
within Coleoptera were identified to a lower level of taxonomic resolution than
Diptera, such that their richness may have been underestimated (Appendix S2). All
FFGs were represented, with abundance dominated by gatherers (42 %), and richness
by predators (50 %). Shredders were the least abundant (< 1 %) and taxon rich (5 %)
The twelve 2006 dry season samples (three samples each from GUm, GUo, FUm and
FUo) represented 88 morphotaxa, from which 13 005 individuals were identified,
compared with 13 053 individuals from 74 morphotaxa from the same waterbodies in
2005 (Appendix S2). Insecta dominated the abundance (43 %) and richness (74 %) of
the 2006 samples, which was also the case in 2005 for the same four waterbodies
(47 % for abundance and 72 % for richness). All functional feeding groups were
represented in samples collected in 2006, showing similar patterns as in 2005.
Spatial variation in taxonomic composition, 2005 dry season
Catchment and flow status both had significant effects on the taxonomic composition
of assemblages across the study region (ANOSIM, P = 0.008 and 0.004, respectively;
Table 3). The effects of these factors were visible on the agglomerative dendrogram
and MDS ordination of sample assemblages, with the separation between lotic
Gregory waterbodies and all remaining waterbodies, including GUo (the lentic
waterbody in the Gregory catchment) clearly evident (Figs 3a & 4a). Additionally, a
key group of species were associated with the difference between groups within the
factors (SIMPER; Table 3). The abundances of cladocerans (Simocephalus sp.),
ostracods and bivalves (Corbicula sp.) were particularly important in contributing to
the difference between catchments (Gregory versus Flinders) and flow states of
waterbodies (lotic versus lentic), making up the first 10 % of the difference between
groups within both factors. The abundances of Simocephalus sp. and ostracods were
higher in samples from the Flinders than from the Gregory. However, Simocephalus
sp. abundance was greater in samples from lotic waterbodies than lentic, whereas the
opposite was the case for ostracods. The abundance of Corbicula sp. was greater in
samples from Gregory and lotic waterbodies than from Flinders and lentic
waterbodies, respectively. This pattern was also the case for Thiara (Plotiopsis) sp.,
the fourth most important taxon to contribute to the differences within each factor.
Indeed, this taxon was completely absent from both Flinders and lentic waterbodies.
Although catchment and flow status both had significant effects on assemblage
structure, differences between main and off-channel waterbodies were not as clear.
Within lentic Flinders waterbodies, a weak but significant difference was found
between main and off-channel locations (one-way ANOSIM on lateral position, P =
0.045; Table 3). This effect was apparent on the dendrogram within groups of the
main separation between lotic Gregory waterbodies and all others (Fig. 3a). However,
taxa associated with this difference primarily consisted of gatherers, grazers and
predators, all of which appeared equally present in both locations (SIMPER, Table 3).
The most obvious difference between the two groups of taxa was that main channel
assemblages were characterised by greater abundances of two filter feeding taxa
(Simocephalus sp. and Tanytarsini) (SIMPER, Table 3).
Ranges of pair-wise Bray-Curtis dissimilarity scores within groups of samples at
increasing spatial scales of resolution suggested a ‘faunal differentiation by distance’
across the study region (cf. Marshall et al., 2006) (Fig. 5a). Sample assemblages
within waterbodies were similar (low pair-wise dissimilarity scores); but, when
abundances of taxa were averaged over samples to produce waterbody centroids, pair-
wise dissimilarity scores increased (greater between-waterbody variability in
assemblages than within-waterbody). Additionally, between-waterbody variation in
assemblages was greater (in range and marginally in median score) when catchment
boundaries were disregarded (greater between-waterbody variation at the scale of the
whole study region than within catchments).
Environmental influences at scales smaller than the broad-scale effects of catchment,
flow status and position in the floodplain also related to spatial patterns of assemblage
composition. Multivariate correlations between structural assemblage and
environmental variable dissimilarity matrices (BIOENV) indicated relationships at the
waterbody and within-waterbody scales, and with habitat and water quality
characteristics (Table 4). Combinations of habitat characteristics tended to explain the
most variation in assemblage patterns compared with other types of variables, and the
proportions of aquatic vegetation (macrophytes plus algae), leaf litter and snags in
macroinvertebrate habitats gave the best combination of variables within any one
dataset (explaining 39 % of the variation in assemblages). However, the best
correlation between environment and assemblage composition (highest rs) was found
using a combination of variables from the within-waterbody scale (presence or
absence of macrophytes and undercuts) plus a number of habitat (proportions of
aquatic vegetation and leaf litter) and water quality characteristics (concentration of
ammonium-N and pH) (61 % of the variation in assemblage patterns explained).
Many of these features could be associated with the difference in conditions between
lotic Gregory waterbodies and all other waterbodies. In particular, lotic waterbodies in
the Gregory catchment were characterised by the presence of undercuts and
macrophytes, higher proportions of aquatic vegetation and lower pH (Table 2).
Spatial variation in functional composition, 2005 dry season
In terms of functional feeding groups, a number of general trends in their
proportionate representation were observed among waterbodies. Overall, gatherers
and filterers tended to dominate the macroinvertebrate FFGs found in samples (Fig.
6). However, shredders were relatively most abundant in Gregory waterbodies,
gatherers in lentic waterbodies and filterers in lotic waterbodies. Both grazers and
predators appeared to make up at least a quarter of the FFG abundances in most
waterbodies, and main channel waterbodies, except for FUm, appeared to have greater
proportions of filterers than their corresponding off-channel waterbodies (Fig. 6).
Additionally, some waterbodies appeared to have site-specific differences in FGG
proportions. Gatherers occurred in comparatively high proportions in assemblages of
GUo, FDo, FUm and FUo (waterbody means 63 %). Grazer proportions were
comparatively low at FDm, FUm and FUo (waterbody means 7 %). FDm had the
greatest proportion of predators (29 ± 3 %) and GDo the greatest proportion of
shredders (5 ± 1 %).
In addition to these general trends, there were statistically significant differences in
the functional composition of assemblages between catchments and states of flow
(ANOSIM R = 0.275 and 0.703, P = 0.029 and 0.001, respectively). As seen in the
taxonomic composition of assemblages, these differences were apparent on the MDS
ordination plot based on FFG proportions (Fig. 4c). SIMPER analysis showed that
within groups of waterbodies, gatherers made up a greater proportion in lentic and
Flinders samples, filterers and grazers in lotic and Gregory samples. Interestingly,
flow status had a much stronger influence than catchment (greater R and lower P-
values) on functional differences between groups within these a priori-defined
factors, than on taxonomic differences (cf. Table 3). The effect of lateral position on
assemblage composition was also different between taxonomic- and functional-based
analyses; despite the association of filterers with main channel waterbodies (see
above), lateral position did not have a significant effect on variation in the
representation of functional feeding groups (one-way ANOSIM between main and
off-channel locations for lentic Flinders waterbodies only, R = 0.231, P = 0.082).
However, faunal differentiation by distance in the functional composition of
assemblages was similar to that seen in taxonomic composition, whereby pair-wise
variation between assemblages (Bray-Curtis dissimilarity scores and their ranges)
increased with spatial scale (Fig. 5b). In contrast to taxonomic composition, however,
the variation in functional composition between waterbodies was similar at both the
within- and across-catchment levels of spatial resolution (Fig. 5).
Within sets of environmental variables at spatial scales smaller than the catchment,
relationships between patterns of environmental characteristics and the functional
composition of assemblages were strongest for habitat and water quality
characteristics (BIOENV, Table 4). However, water quality variables explained a
greater amount of the pattern (32 %). In contrast to patterns based on taxonomic
composition, within-waterbody scale features were poor at describing functional
patterns ( 11 % of the variation explained); but combinations of variables across
datasets still explained the most variation. The best combination (highest rs) explained
45 % of the variation in the functional assemblage dataset and included, (i) the
number of different macroinvertebrate habitats, (ii) total nitrogen concentration, (iii)
the proportion of organic suspended solids, (iv) the areal amount of chlorophyll a and
proportion of silt in the littoral-zone sediment, (v) canopy cover, and (vi) waterbody
depth (Table 4). Total nitrogen concentration was highly correlated with euphotic
depth and concentrations of total phosphorus and suspended solids (redundancy
analysis; Spearman’s rs > 0.9). Together, the importance of these features suggested a
strong influence of primary productivity, along with macroinvertebrate habitat
diversity, on the functional organisation of waterbody assemblages.
Spatial variation in diversity, 2005 dry season
Assemblages varied among waterbodies in terms of their calculated measures of
diversity (Fig. 7). High abundances of macroinvertebrates were found in FDm, FMm
and FUo, probably due to the large numbers of ostracods collected at these
waterbodies. Within waterbodies, the variation in abundances between samples was
greatest for GMmB (1076 ± 8210); one sample from this site contained a
comparatively high number of cladocerans (the filterer, Simocephalus sp.). As a
result, this site had the highest within-waterbody beta-diversity ( = 0.58; Fig. 7),
which could be seen in the MDS ordinations on the taxonomic and functional
compositions of assemblages (Fig. 4). However, beta-diversity scores were low for all
waterbodies ( < 1), which indicated sample assemblages within waterbodies were
similar. Richness (S) was highest in GDo (50 ± 3) and lowest in FDo (20 ± 1). In
terms of evenness, Margalef’s D values suggested that macroinvertebrate assemblages
in lotic and Gregory waterbodies were more even (higher means) than in lentic and
Flinders waterbodies (lower means) (Fig. 7). Berger-Parker (BP) scores suggested that
assemblages from lentic, off-channel waterbodies were more dominated by one taxon
than other waterbodies (GUo, FDo, FUo all had mean scores 0.45 and had large
numbers of ostracods compared with abundances of other taxa) (Fig. 7).
Off-channel waterbodies did not appear to have more diverse macroinvertebrate
habitats than main channels. Rather, main channel waterbodies tended to have similar
or greater numbers of different macroinvertebrate habitats than their corresponding
off-channels (within a reach) and as stated above, the greatest within-site beta-
diversity () was at GMmB, a main channel site (Fig. 7). However, the lowest beta-
diversity between waterbodies within a reach was for the two main channel sites,
GMmB and GMmG ( = 0.73), which were both lotic. The highest beta-diversity
scores were found between lentic off-channels and lotic-main channels within reaches
(GUm and GUo, = 0.97; FDm and FDo, = 1.13), and intermediate beta-diversity
scores were found between main and off-channel waterbodies that were either lentic
(FUm and FUo, = 0.82) or lotic (GDm and GDo, = 0.93) (Fig. 7).
Flow status had a statistically significant effect on assemblage richness, evenness and
dominance (least squares means ANOVA, P < 0.05; Table 5). For evenness
(Margalef’s D, log-transformed), the effect of flow status interacted with that of
catchment (P < 0.0001); assemblages in lotic waterbodies in the Gregory catchment
were more even than assemblages in the lotic Flinders waterbody (FDm), the lentic
Gregory waterbody (GUo) and the lentic waterbodies in the Flinders catchment. For
richness and dominance, there was no interaction effect and the only significant main
effect (P < 0.05) was from flow status (not catchment). Richness (S) was greater in
assemblages from lotic waterbodies than lentic, whereas dominance (BP) was greater
in lentic waterbodies. Sample abundance (N, log-transformed) and within-waterbody
-diversity were not affected by flow status or catchment (P > 0.05), indicating
waterbodies from different catchments or states of flow supported similar numbers of
biota and had similar levels of within-waterbody variability (despite the
comparatively high variation observed among GMmB samples). Within lotic
waterbodies in the Gregory catchment only, a significant difference was detected in
richness (S) and evenness (D, log-transformed), which were both greater in samples
from off-channel waterbodies than from main channels. Within lentic waterbodies in
the Flinders catchment only, evenness (D, log-transformed) was greater in samples
from main channels than off-channels, and these off-channel samples were less
dominated by one taxon, as was indicated by their significantly lower BP scores.
Temporal variation in taxonomic composition
Similar patterns of variation in taxonomic composition of assemblages were seen
among all waterbodies sampled in 2005 and those sampled in 2006 (MDS ordination,
Fig. 4a,b), supporting the clear separation between assemblages from lotic
waterbodies in the Gregory catchment and those from all other waterbodies. However,
when patterns of variation in assemblage structure were examined among re-sampled
waterbodies only (GUm, GUo, FUm, FUo in 2005 and 2006 dry seasons), more
variation could be attributed to flow status of waterbodies, rather than to other broad-
scale factors, such as year or catchment (UPGMA dendrogram, Fig. 3b). This was
confirmed by ANOSIM: although year and catchment both had significant effects on
assemblage patterns, the effect of flow status was stronger, having higher ANOSIM R
and lower P-values than other factors (Table 3). Additionally, a similar group of taxa
was associated with the difference in assemblage structure between lotic and lentic
waterbodies (SIMPER) in both the spatial (11 waterbodies in 2005 only) and the
temporal dataset (four waterbodies in both 2005 and 2006) (Table 3). The main
difference between these groups of taxa was due to the relative absence of
Simocephalus sp. from the 2006 samples, such that this species only made an
important contribution to the difference between lotic and lentic assemblages within
the 2005 dataset.
Temporal variation in assemblage structure between the same two waterbodies
appeared greater than spatial variation within waterbodies. This was demonstrated by
comparing ranges of pair-wise Bray-Curtis dissimilarity scores between groups of
sample assemblages (Fig. 5c). Although dissimilarly increased with distance
(between-waterbody greater than within-waterbody) in both 2005 and 2006, the
dissimilarity between the same waterbodies across the two dry seasons was much
higher than that within waterbodies.
The diversity of macroinvertebrate fauna collected from the study region (124 taxa in
total across 11 locations and two dry seasons) appeared similar to that found
elsewhere in Australia’s wet/dry tropics, in both numbers and types of taxa present
(Marchant, 1982; Outridge, 1988). Different collection methods and levels of
taxonomic resolution make comparisons problematic, but diversity in the study region
also appeared greater than that found in floodplain-rivers in the neo-tropics (see
Ezcurra De Drago et al., 2004) and in dryland Australia (in Cooper Creek; Marshall et
Spatial variation and hydrological connectivity
Hydrological connectivity (the lotic or lentic nature of waterbodies and their
connection history) can be considered the key driver of spatial patterns of
macroinvertebrate diversity and structural composition (both taxonomic and
functional) in the study region. This driver was stronger than catchment in its
influence on macroinvertebrates, contributing to major differences between
assemblages sampled from lotic Gregory waterbodies and all others (all lentic and
Flinders waterbodies). Indeed, a key group of taxa was associated with the difference
between lotic and lentic waterbodies. In particular, the gastropod Thiara (Plotiopsis)
sp., which is typically associated with flowing waters in Australian rivers (Hawking &
Smith, 1997; Tsyrlin & Gooderham, 2002), was strongly indicative of this split, being
found in lotic waterbodies within the Gregory catchment only. Functionally, gatherers
tended to be more abundant in lentic and Flinders waterbodies, with filterers and
grazers more abundant in lotic and Gregory waterbodies. Additionally, assemblages
from lotic waterbodies were more rich and even than those from lentic waterbodies,
which were often dominated by a single taxon (although not consistently the same
taxon in each waterbody).
Thus, our findings support our prediction that variation in hydrological connectivity
has a major influence on macroinvertebrate assemblages in the study region. More
specifically, we found greater differentiation between lotic and lentic waterbodies
than between waterbodies of the same flow status. In addition, the two river systems
(the ‘tropical’ Gregory versus the ‘dryland’ Flinders) were associated with different
patterns of variation in assemblage composition and diversity. Indeed, differences in
the taxonomic composition of assemblages were more apparent in the ‘tropical’
Gregory catchment, in which lentic waterbodies in the floodplain disconnect from
permanently flowing channels each dry season, than in the Flinders, in which
waterbodies tend to share similar connection histories.
The effect of hydrological connectivity on assemblages was so strong that the
influence of lateral position in the catchment (main versus off-channel location) could
not be fully explored, and differences between main and off-channel locations for
waterbodies of the same flow status within a catchment were either weak (for
taxonomic composition) or non-significant (for proportionate FFG composition).
However, differences in diversity were more obvious, and showed opposite trends for
lotic waterbodies in the Gregory compared with lentic waterbodies in the Flinders.
Off-channel assemblages were richer and more even than were main channel
assemblages within the former group of waterbodies, but were less even within the
latter group. This suggests that connection between waterbodies by flow, as found in
the Gregory catchment, supported more diverse assemblages in off-channel locations
than main channels. Alternatively, lack of flow and disconnection between
waterbodies, as found in the Flinders catchment, may have lead to the dominance of
assemblages in off-channel locations by particular taxa suited to stable lentic habitats
(usually gatherers). Results also suggest that, if a difference in flow status between
main and off-channel waterbodies exists, lotic waterbodies would be likely to contain
more taxa and be less dominated by particular taxa than lentic waterbodies, regardless
of their location (because the strong effect of flow status on assemblage diversity
would dominate over that of lateral position).
Our study indicates that concepts of connectivity and biotic diversity developed for
dryland rivers (Walker, Sheldon and Puckridge, 1995; Sheldon, Boulton and
Puckridge, 2002; 2003; Marshall et al., 2006; Sheldon & Thoms, 2006) are well
suited to river systems in Australia’s wet/dry tropics. In dryland rivers, prolonged
flow disconnection among waterbodies (as also occurs in Australia’s wet/dry tropics
each dry season) is associated with comparatively depauperate assemblages in lentic
rather than in lotic waterbodies, regardless of their position in the catchment. Even in
Australia’s tropical north, a decline in richness over the course of the dry season has
been observed for macroinvertebrate assemblages in lentic waterbodies both on and
off main channels (Marchant, 1982; Outridge, 1988). In general, these changes in
assemblage richness would therefore be expected to affect between-waterbody beta-
diversity, such that the greatest beta-diversity would occur between waterbodies of
different states of connectivity and thus connectivity potential (sensu Marshall et al.,
2006; Sheldon & Thoms, 2006). This phenomenon appears to have occurred in the
southern Gulf of Carpentaria study region, where the greatest beta-diversity was
found within pairs of lotic and lentic waterbodies (as represented by a simple
conceptual diagram; Fig. 8). This relates most readily to the modified telescoping
ecosystem model of Ward and Tockner (2001) for waterbodies of floodplain rivers,
whereby flooding creates uniformity among habitats (maximum connectivity) and
drying re-establishes heterogeneity (maximum individually), with maximum diversity
occurring at some intermediate level between these two states (Ward et al., 1999).
Although this model was initially developed from temperate floodplain rivers, it is
appropriate for describing biodiversity patterns in dryland systems (Sheldon et al.,
2002; Marshall et al., 2006), and, based on our findings, we propose that it also has
application in tropical contexts (see also Thomaz, Bini and Bozellii, 2007).
In addition, beta-diversity (as represented by the change in assemblage composition
between pairs of samples or waterbodies in the study region) was seen to increase
with increasing spatial scale in a similar fashion to the ‘faunal differentiation by
distance’ trend observed for macroinvertebrate assemblages in dryland rivers
(Marshall et al., 2006). These authors related spatial proximity between waterbodies
to antecedent hydrology, such that waterbodies closest to each other within regions
would share the most similar connection histories. For the study region, it was clear
that waterbodies within the ‘dryland’ Flinders catchment (more ephemeral and
variable flow regime) share similar connection histories despite differences in flow
status at the time of sampling. For the ‘tropical’ Gregory catchment (more permanent
and regular flow regime), however, the lentic waterbody (GUo) clearly has a different
connection history to its nearby lotic waterbodies. These lotic waterbodies flow
permanently throughout the year whereas GUo becomes disconnected each dry season
(personal communication with landholders). Indeed, GUo assemblages were more
similar to those in the Flinders sampling region than in the Gregory, a pattern
supported by ordination and agglomerative clustering analyses. Therefore, although
similar ‘differentiation by distance’ trends have been observed in both arid-zone,
dryland rivers and in the study region, we suggest that for the ‘tropical’ Gregory river,
assemblages in disconnected (lentic) waterbodies may be substantially different from
those in connected (lotic) waterbodies, even when these waterbodies are
comparatively close together.
At smaller spatial scales, a number of within-waterbody, habitat and water quality
characteristics were important for explaining patterns of assemblage structure based
on taxonomic composition, particularly those associated with habitat availability and
composition. For patterns based on proportions of functional feeding groups,
waterbody scale (rather than within-waterbody scale), habitat and water quality
characteristics were important, particularly those variables associated with potential
primary productivity and habitat diversity. These patterns were not surprising, as it is
well established that small-scale biophysical and chemical variation influences
macroinvertebrate composition (Townsend et al., 2003; 2004) and that taxonomic and
functional composition may show different relationships with environmental variables
(Feld & Hering, 2007; Heino et al., 2007).
However, it is possible that much of the small-scale environmental variation across
the study region was a secondary affect of broad-scale variations in hydrological
connectivity. For example, many water quality characteristics in the study region
appeared to be associated with the flow status of waterbodies. Similarly, the flow
status of waterbodies had a major influence on macroinvertebrate assemblages. Thus,
water quality characteristics would be expected to explain much of the variation in the
biotic datasets, particularly as the influence of local environmental conditions, such as
water quality and habitat characteristics, is expected to increase in importance at
decreasing spatial scales (Mykra, Heino & Muotka, 2007). However, only 21 to 32 %
of the variation in taxonomic abundance and functional feeding group proportion data
could be explained by the water quality dataset alone. This suggests that the broader-
scale influence of hydrological connectivity, as represented by waterbody flow status,
was acting directly on biotic assemblages as well as on water quality; water quality
then directly (but secondarily) adds to the variation in biotic assemblages. This may
also have been the case for within-waterbody and habitat characteristics of
waterbodies, as many of these features, such as the presence of aquatic vegetation,
were also associated with hydrological connectivity and waterbody flow status (lotic
Therefore, although natural water quality, habitat and waterbody characteristics at
both within- and between-waterbody scales are important for macroinvertebrate
structure and diversity in the study region, the broad-scale effect of hydrological
connectivity (sensu Sheldon & Thoms, 2006) was the main driver, influencing biotic
assemblages both directly and indirectly. Despite this, smaller-scale effects are
undoubtedly important contributors to assemblage patterns, and environmental factors
that act at different spatial scales have been shown to have a combined influence on
patterns of variation in aquatic assemblages elsewhere (e.g. in Swedish streams and
lakes) (Johnson, Goedkoop & Sandin, 2004). Thus, maintaining the natural
characteristics of waterbodies is likely to be important for their continued function. Of
overriding importance for the study region, however, is the maintenance of the key
aspects of the natural flow regime (permanence and regularity for ‘tropical’ rivers;
flow variability and ephemerality for ‘dryland’ rivers; and annual wet/dry seasonality)
(Leigh & Sheldon, 2008). This will maintain functional and diverse macroinvertebrate
assemblages in Australia’s wet/dry tropics, and is a concept well established in
riverine ecology (Poff et al., 1997; Benke, 2001).
Patterns of variation in assemblage structure were generally consistent across the two
dry seasons and confirmed the importance of hydrological connectivity on assemblage
patterns. Indeed, the effect of flow status was stronger than differences attributable to
catchment or year of sampling and a similar group of species was associated with the
difference between lotic and lentic waterbodies in both years. In addition, the amount
of temporal variation observed in both the ‘dryland’ Flinders and the ‘tropical’
Gregory river systems did not appear to have been as great as that observed in dryland
rivers, where major changes have been observed through time (Marshall et al., 2006).
This may be due to the greater regularity of flow found in rivers in the wet/dry tropics,
as a consequence of annual monsoons (Douglas, Bunn & Davies, 2005), in
comparison with rivers in arid climates. Monsoonal rainfall produces consistent
changes between seasons in both ‘dryland’ and ‘tropical’ rivers of Australia’s north:
each dry season, the number of zero flow days increases in ‘dryland’ rivers and flow
magnitudes decrease in ‘tropical’ rivers; each wet season, flow magnitudes increase in
both river types (Leigh & Sheldon, 2008). Indeed, recolonisation of
macroinvertebrates within the lower reaches of Magela Creek, in Australia’s wet/dry
tropics, has been attributed to downstream drift from its upper reaches each wet
season during high flow periods (Paltridge et al., 1997). Such regular cycles would
probably assist in the regular and predictable dispersal of invertebrates via aquatic and
aerial pathways, so that a similar, but seasonal, fauna persists in these systems through
time (Malmqvist, 2002; Robinson, Tockner & Ward, 2002).
Beta-diversity, hydrological connectivity and the Flood Pulse Concept
If the increased diversity often associated with floodplain waterholes (as proposed by
the Flood Pulse Concept and demonstrated in the neo-tropics) is due to increased
habitat diversity across the floodplain environment (Junk et al., 1989; Ezcurra De
Drago et al., 2004), then increased richness and beta-diversity would be expected
within off-channel waterbodies with comparatively high habitat diversity. However,
off-channel waterbodies in the study region did not tend to have increased habitat
diversity. Although this observation does not refute the contention that increased beta-
diversity is related to increased habitat diversity in the floodplain, particularly given
the limited number of comparisons involved, it emphasises the more apparent
correlation in the study region that was found between beta-diversity and hydrological
connectivity (see above), a phenomenon that can act in all directions and across small
to large scales.
Synthesis and prospects
Dynamic hydrological connectivity among waterbodies in space and time is important
for maintaining high biodiversity and function in large floodplain rivers (Tockner et
al., 1999; Ward et al., 1999; Ward & Tockner, 2001; Amoros & Bornette, 2002;
Robinson et al., 2002; Sheldon et al., 2002; Thorp et al., 2006). In the lower,
floodplain reaches of the Gregory and Flinders Rivers in the southern Gulf of
Carpentaria, northeast Australia, a close relationship was found between hydrological
connectivity and macroinvertebrate assemblage composition and diversity during the
dry season. Although differences in biodiversity patterns were apparent between the
two river systems, macroinvertebrate assemblages showed similar responses in both:
assemblages from waterbodies with similar connection histories were most alike, and
beta-diversity between assemblages was greatest between lotic and lentic waterbodies,
tending to increase with increasing spatial separation.
The studied rivers systems fall into two major classes of flow regime in Australia’s
wet/dry tropics (‘dryland’ and ‘tropical’), and the main exception to the above
patterns was found in the ‘tropical’ Gregory River. In this system, lotic and lentic
waterbodies may be spatially close but have distinct connection histories (temporally
distant), resulting in distinct macroinvertebrate communities. Although, our study can
not be used to infer categorical differences between all ‘dryland’ and ‘tropical’ rivers
across Australia’s wet/dry tropics, we propose that this phenomenon may be less
likely to occur in the ‘dryland’ rivers in this region, as these do not have perennially
flowing channels, and thus their waterbodies, spatially close or distant, would be more
likely to share a similar connection history. In the wet season, when monsoons
produce widespread flooding, particularly in the ‘dryland’ rivers, we also expect the
close relationship between assemblages and hydrological connectivity to remain;
maximum beta-diversity would occur between flooded (highly connected) and non-
flooded (disconnected) waterbodies within wet seasons, and between lotic (connected)
and lentic (disconnected) waterbodies between wet and dry seasons. We recommend
the continued study of the Gregory and Flinders river systems, the inclusion of
additional sites, greater within-site replication and extension to other ‘tropical’ and
‘dryland’ systems in Australia’s wet/dry tropics in order to explore the widespread
applicability of these proposals.
Like most rivers in Australia’s north, the studied Gregory and Flinders systems
generally flow unimpeded. However, there is much interest in developing the region,
including water resource development options like abstraction and regulation
(Woinarski et al., 2007). This will undoubtedly threaten the region’s freshwater
biodiversity (Pringle, 2001; Dudgeon et al., 2006) including that of aquatic
macroinvertebrates (Miller, Wooster & Li, 2007). This study has highlighted the
importance of spatiotemporal variation in levels of hydrological connectivity in
contributing to the diversity, structure and function of macroinvertebrate communities
in two of Australia’s northern floodplain rivers. Flow regulation and abstraction
reduce variation in connectivity by tending to increase or decrease connection,
respectively, in space and for prolonged periods of time (Sheldon & Thoms, 2006;
Leigh & Sheldon, 2008). Thus, conservation of the high levels of macroinvertebrate
biodiversity in Australia’s wet/dry tropics, along with the functional processes in
which they take part (e.g. biomass contribution to higher trophic levels, subsidies to
terrestrial food webs and resource processing), may depend on conserving the natural
variation in hydrological connectivity and the flow regimes that the unregulated
floodplain rivers in the region currently possess.
Research funding was provided by Land & Water Australia (Project code GRU35)
with additional support from the Australian School of Environmental Studies. Flow
data for Gulf of Carpentaria rivers were provided by the Queensland Department of
Natural resources and Mines (DNRM, 2005), which gives no warranty in relation to
the data (including accuracy, reliability, completeness or suitability) and accepts no
liability (including without limitation, liability in negligence) for any loss, damage or
costs (including consequential damage) relating to any use of the data. We thank
colleagues at the Australian Rivers Institute for comments on manuscript drafts and
advice on field and laboratory methods and preparation. We thank local Indigenous
communities, pastoral leaseholders and station managers for granting permission and
access to field sites, and Erica Alacs, Ben Cook, James Fawcett, Joel Huey, Jim
McGuire, Tim Page, Terry Reis, Brett Taylor, and Matthew Vickers (Southern Gulf
Catchments) for assistance in the field. Comments from an anonymous reviewer,
David Dudgeon and Alan Hildrew provided substantial improvement to the
Amoros C. & Bornette G. (2002) Connectivity and biocomplexity in waterbodies of
riverine floodplains. Freshwater Biology, 47, 761-776.
Anderson M.J. (2001) A new method for nonparametric multivariate analysis of
variance. Austral Ecology, 26, 32-46.
APHA (American Public Health Association) (1989) Standard Methods for the
Examination of Water and Wastewater (20th edn). APHA-AWWA-WEF,
Washington, D. C.
Benke A.C. (2001) Importance of flood regime to invertebrate habitat in an
unregulated river–floodplain ecosystem. Journal of the North American
Benthological Society, 225–240.
Boyero L. & Bailey R.C. (2001) Organization of macroinvertebrate communities at a
hierarchy of spatial scales in a tropical stream. Hydrobiologia, 464, 219 - 225.
Clarke K.R. & Warwick R.M. (2001) Change in marine communities: an approach to
statistical analysis and interpretation, 2nd edition. PRIMER-E, Plymouth.
Cummins K.W. & Klug M.J. (1979) Feeding ecology of stream invertebrates. Annual
Review of Ecology and Systematics, 10, 147-172.
DNRM (Department of Natural Resources and Mines) (2005) Discharge Data for Gulf
of Carpentaria Rivers. Available online at www.nrw.qld.gov.au/watershed/
(Accessed May 2005). Department of Natural Resources and Mines,
Dos Santos A.M. & Thomaz S.M. (2007) Aquatic macrophytes diversity in lagoons of
a tropical floodplain: the role of connectivity and water level. Austral Ecology,
Douglas M.M., Bunn S.E. & Davies P.M. (2005) River and wetland food webs in
Australia's wet-dry tropics: general principles and implications for
management. Marine and Freshwater Research, 56, 329-342.
Dudgeon D. (1994) The influence of riparian vegetation on macroinvertebrate
community structure and functional organisation in six New Guinea streams.
Hydrobiologia, 294, 65-85.
Dudgeon D., Arthington A.H., Gessner M.O., Kawabata Z., Knowler D., Lévêque C.,
Naiman R.J., Prieur-Richard A.-H., Soto D. & Stiassny M.L.J. (2006)
Freshwater biodiversity: importance, threats, status, and conservation
challenges. Biological Reviews, 81, 163-182.
Erskine W.D., Saynor M.J., Erskine L., Evans K.G. & Moliere D.R. (2005) A
preliminary typology of Australian tropical rivers and implications for fish
community ecology. Marine and Freshwater Research, 56, 253-267.
Ezcurra De Drago I., Marchese M. & Wantzen K.M. (2004) Benthos of a large
neotropical river: spatial patterns and species assemblages in the Lower
Paraguay and its floodplains. Archiv Fur Hydrobiologie, 160, 347-374.
Feld C.K. & Hering D. (2007) Community structure or function: effects of
environmental stress on benthic macroinvertebrates at different spatial scales.
Freshwater Biology, 52, 1380-1399.
Hargrove W.W. & Pickering J. (1992) Pseudoreplication: a sine qua non for regional
ecology. Landscape Ecology, 4, 251-258.
Hawking J.H. & Smith F.J. (1997) Colour Guide to Invertebrates of Australian Inland
Waters. Cooperative Research Centre for Freshwater Ecology, Thurgoona,
Hawking J.H. (2000) Key to Keys: A Guide to Keys and Zoological Information to
Identify Invertebrates from Inland Waters (2nd edn). Cooperative Research
Centre for Freshwater Ecology, Albury.
Heino J., Mykra H., Kotanen J. & Muotka T. (2007) Ecological filters and variability
in stream macroinvertebrate communities: do taxonomic and functional
structure follow the same path? Ecography, 30, 217-230.
Johnson R.K., Goedkoop W. & Sandin L. (2004) Spatial scale and ecological
relationships between the macroinvertebrate communities of stony habitats of
streams and lakes. Freshwater Biology, 49, 1179-1194.
Junk W.J., Bayley P.B. & Sparks R.E. (1989) The flood pulse concept in river-
floodplain systems. Special Publication of the Canadian Journal of Fisheries
and Aquatic Sciences, 106, 110-127.
Kirk J.T.O. (2003) The vertical attenuation of irradiance as a function of the optical
properties of the water. Limnology and Oceanography, 48, 9–17.
Latrubesse E.M., Stevaux J.C. & Sinha R. (2005) Tropical rivers. Geomorphology,
Leigh C. & Sheldon F. (2008) Hydrological changes and ecological impacts
associated with water resource development in large floodplain rivers in the
Australian tropics. River Research and Applications. DOI: 10.1002/rra.1125.
Magurran A.E. (1988) Ecological Diversity and its Measurement. Croom Helm,
Malmqvist B. (2002) Aquatic invertebrates in riverine landscapes. Freshwater
Biology, 47, 679-694.
Marchant R. (1982) Seasonal variation in the macroinvertebrate fauna of billabongs
along Magela Creek, Northern Territory. Australian Journal of Marine and
Freshwater Research, 33, 329-342.
Marshall J.C., Sheldon F., Thoms M. & Choy S. (2006) The macroinvertebrate fauna
of an Australian dryland river: spatial and temporal patterns and
environmental relationships. Marine and Freshwater Research, 57, 61-74.
McCune B., Grace J.B. & Urban D.L. (2002) Analysis of Ecological Communities.
MjM Software Design, Geneden Beach, Oregon.
Melo A.S. & Froehlich C.G. (2001) Macroinvertebrates in neotropical streams:
richness patterns along a catchment and assemblage structure between 2
seasons. Journal of the North American Benthological Society, 20, 1-16.
Merritt R.W. & Cummins K.W. (1996) Trophic relations of macroinvertebrates. In:
Methods in Stream Ecology (Eds. F.R. Hauer & G.A. Lamberti), pp. 453-474.
Academic Press, San Diego.
Miller S.W., Wooster D. & Li J. (2007) Resistance and resilience of
macroinvertebrates to irrigation water withdrawals. Freshwater Biology, 52,
Montgomery D.C. (2001) Design and Analysis of Experiments (5th edn). John Wiley
& Sons, New York.
Mykra H., Heino J. & Muotka T. (2007) Scale-related patterns in the spatial and
environmental components of stream macroinvertebrate assemblage variation.
Global Ecology and Biogeography, 16, 149-159.
Neter J., Wasserman W. & Kutner M.H. (1985) Applied Linear Statistical Models:
Regression, analysis of variance, and experimental designs (2nd edn). Irwin,
Outridge P. (1988) Seasonal and spatial variations in benthic macroinvertebrate
communities of Magela Creek, Northern Territory. Marine and Freshwater
Research, 39, 211-223.
Paltridge R.M., Dostine P.L., Humphrey C.L. & Boulton A.J. (1997)
Macroinvertebrate recolonization after re-wetting of a tropical seasonally-
flowing stream (Magela Creek, Northern Territory, Australia). Marine and
Freshwater Research, 48, 633-645.
Poff N.L. (1997) Landscape filters and species traits: towards mechanistic
understanding and prediction in stream ecology. Journal of the North
American Benthological Society, 16, 391-409.
Poff N.L., Allan J.D., Bain M.B., Karr J.R., Prestegaard K.L., Richter B.D., Sparks
R.E. & Stromberg J.C. (1997) The natural flow regime: a paradigm for river
conservation and restoration. Bioscience, 47, 769-784.
PRIMER-E (2002) PRIMER 5 for Windows Version 5.2.9. PRIMER-E, Plymouth.
Pringle C.M. (2001) Hydrologic connectivity and the management of biological
reserves: a global perspective. Ecological Applications, 11, 981-998.
Puckridge J.T., Sheldon F., Walker K.F. & Boulton A.J. (1998) Flow variability and
the ecology of large rivers. Marine and Freshwater Research, 49, 55-72.
Robinson C.T., Tockner K. & Ward J.V. (2002) The fauna of dynamic riverine
landscapes. Freshwater Biology, 47, 661-677.
SAS Institute (2002) SAS Propriety Software Release, Version 8 Edition. SAS
Institute, Cary, North Carolina.
Sheldon F., Boulton A.J. & Puckridge J.T. (2002) Conservation value of variable
connectivity: aquatic invertebrate assemblages of channel and floodplain
habitats of a central Australian arid-zone river, Cooper Creek. Biological
Conservation, 103, 13-31.
Sheldon F., Boulton A.J. & Puckridge J.T. (2003) Variable connection structures
invertebrate composition in dryland rivers. Records of the South Australian
Museum Monograph Series, 7, 119-130.
Sheldon F. & Thoms M.C. (2006) Relationships between flow variability and
macroinvertebrate assemblage composition: data from four Australian dryland
rivers. River Research and Applications, 22, 219–238.
Thomaz S.M., Bini L.M. & Bozelli R.L. (2007) Floods increase similarity among
aquatic habitats in river-floodplain systems. Hydrobiologia, 579, 1-13.
Thorp J.H., Thoms M.C. & Delong M.D. (2006) The riverine ecosystem synthesis:
biocomplexity in river networks across space and time. River Research and
Applications, 22, 123-147.
Tockner K., Pennetzdorfer D., Reiner N., Schiemer F. & Ward J.V. (1999)
Hydrological connectivity, and the exchange of organic matter and nutrients in
a dynamic river-floodplain system (Danube, Austria). Freshwater Biology, 41,
Townsend C.R., Doledec S., Norris R., Peacock K. & Arbuckle C. (2003) The
influence of scale and geography on relationships between stream community
composition and landscape variables: description and prediction. Freshwater
Biology, 48, 768-785.
Townsend C.R., Downes B.J., Peacock K. & Arbuckle C.J. (2004) Scale and the
detection of land-use effects on morphology, vegetation and macroinvertebrate
communities of grassland streams. Freshwater Biology, 49, 448-462.
Tsyrlin E. & Gooderham J. (2002) The Waterbug Book : A guide to the freshwater
macroinvertebrates of temperate Australia. CSIRO Publishing, Melbourne.
Underwood A.J. & Chapman M.G. (1998) A method for analysing spatial scales of
variation in composition of assemblages. Oecologia, 117, 570-578.
Walker K.F., Sheldon F. & Puckridge J.T. (1995) A perspective on dryland river
ecosystems. Regulated Rivers: Research and Management, 11, 85-104.
Ward J.V. (1989) The four-dimensional nature of lotic ecosystems. Journal of the
North American Benthological Society, 8, 2-7.
Ward J.V., Tockner K. & Schiemer F. (1999) Biodiversity of floodplain river
ecosystems: ecotones and connectivity. Regulated Rivers: Research and
Management, 15, 125-139.
Ward J.V. & Tockner K. (2001) Biodiversity: towards a unifying theme for river
ecology. Freshwater Biology, 46, 807-819.
Woinarski J., Mackey B., Nix H. & Traill B. (2007) The Nature of Northern
Australia: Natural values, ecological processes and future prospects. ANU E
Table 1: Codes used for waterbodies sampled during the 2005 dry season, with detail
of catchment, reach, lateral position in relation to the main channel, and flow status at
the time of sampling.
Waterbody code Catchment Reach
GDm Gregory Downstream
FDm Flinders Downstream
FDo Flinders Downstream
FMm Flinders Mid
Waterbodies re-sampled during the 2006 dry season are underlined.
Table 2: Environmental conditions of waterbodies in the dry season of 2005, described by variables used in the correlation analyses with
macroinvertebrate assemblage data (BIOENV), with water quality data for waterbodies re-sampled in the 2006 dry season.
GDm GDo GMmB GMmG GUm GUm
CW 5 20 30 40 30
WW 5 15 25 30 15
LS 80 1125 750 800 450
D 0.8 1.5 5.7 3.9 1.5
WV 800 25310 106880 93600 10130
CC 1.00 0.92 0.68 0.77 0.60
BS 3 1 2 1 1
MP 0 1 1 1 1
MA 0 1 1 1 1
ma 1 0 0 0 1
UC 1 1 1 1 0
Hab 4 4 2 5 5
ChlS 7.1 6.6 1.7 7.1 4.2
ChlB 23.0 4.4 2.4 13.2 72.4
%Sn 0.00 0.05 0.00 0.10 0.00
%LL 0.05 0.35 0.00 0.20 0.10
%MA 0.00 0.20 0.30 0.20 0.05
%AV 0.05 0.60 0.80 0.40 0.30
%Si 0.70 0.70 0.70 0.70 0.15
pH 8.2 8.3 8.4 8.2 8.3 8.5
TN 0.24 0.20 0.19 0.17 0.12 0.18
NOx 0.09 0.04 0.01 0.01 0.02 0.02
NH4 0.05 0.05 0.02 0.02 0.05 0.02
PO4 0.006 0.007 0.009 0.007 0.010 0.003
N:P 57 34 7 12 10 32
%OSS 0.34 0.40 0.51 0.28 0.39 0.57
Waterbody scale characteristics: CW, channel width (m); WW, wetted width (m); LS, maximum length of straight section (m); D, depth (mid-channel location) (m); WV, wetted
volume (m3); CC, canopy cover (mid-channel location) (0 to 1); BS, bank slope (1, 2 or 3 in increasing levels of steepness). Within-waterbody scale characteristics: MP, macrophytes
presence/absence (0 or 1); MA, macroalgae presence/absence (0 or 1); ma, microalgae presence/absence (0 or 1); UC, undercuts presence/absence (0 or 1). Macroinvertebrate habitat
characteristics: Hab, number of different macroinvertebrate habitats; ChlS, median biomass chlorophyll a in littoral-zone sediment (g m2); ChlB, median biomass of chlorophyll a in
littoral-zone biofilm (g m2); %Sn, proportion snags; %LL, proportion leaf litter; %MP, proportion macrophytes; %AV, proportion all aquatic vegetation; %Si, proportion silt. Water
quality characteristics: TN, median total nitrogen concentration (mg N L-1); NOx, median nitrate/nitrite concentration (mg N L-1); NH4, median ammonium concentration (mg N L-1);
PO4, median phosphate concentration (mg P L-1); N:P, median dissolved molar N:P ratio; %OSS, median proportion organic suspended solids.
7 45 34 34 61 29 11
0.29 0.52 0.26 0.28 0.49 0.22 0.21
Table 3: Results of ANOSIM on assemblage composition (based on Bray-Curtis dissimilarities using log-transformed abundance data) between groups
within a priori-defined factors. Results are presented with taxa identified by SIMPER as contributing to more than 50 % of the difference between
statistically different groups.
ANOSIM Factor Pair-wise comparison ANOSIM
Significant taxa (SIMPER)
33 samples from 11 waterbodies (2005 dry season only)
Catchment Gregory v. Flinders 0.316
Simocephalus sp., Corbicula sp., Thiara (Plotiopsis) sp.pa, Baetidae, Gyraulus sp., Orthocladinae, Angrobia sp., Caridina spp.,
Tasmanocoenis sp., Cyclopoida (greater abundances in the Gregory catchment); Ostracoda, Ferrissia sp., Micronecta spp.,
Oligochaeta, Tanytarsini, Ceratopogoninae, Nematoda, Hydracarina, Tanypodinae (greater abundances in the Flinders
Simocephalus sp., Corbicula sp., Thiara (Plotiopsis) sp. pa, Tanytarsini, Baetidae, Orthocladinae, Angrobia sp., Caridina spp.,
Gyraulus sp., Tasmanocoenis sp., Cyclopoida, Tanypodinae (greater abundances in lotic waterbodies); Ostracoda, Ferrissia sp.,
Oligochaeta, Micronecta spp., Nematoda, Ceratopogoninae, Hydracarina, Hydroglyphus sp. (greater abundances in lentic
Oligochaeta, Ferrissia sp., Tasmanocoenis spp., Simocephalus sp., Tanytarsini, Wundacaenis sp. pa, Hydroglyphus sp.,
Hydrophilidae (indeterminate adult sp.), Tanypodinae (greater abundances in main channel locations); Micronecta spp.,
Hydracarina, Baetidae, Gyraulus sp. (greater abundances in off-channels)
Flow status Lotic v. Lentic 0.365
Main v. off-channel
24 samples from four waterbodies (2005 and 2006 dry seasons)
Flow status Lotic v. Lentic
Catchment Gregory v. Flinders 0.267
Corbicula sp., Angrobia sp., Tanytarsini, Thiara (Plotiopsis) sp.pa, Orthocladinae, Austrolimnius sp., Gyraulus sp., Libellulidae
(tiny)pa, Micronecta spp., Tanypodinae, Cyclopoida, Economidae (greater abundances in lotic waterbodies); Ostracoda,
Oligochaeta, Austrogomphus spp. pa, Nematoda, Hydroglyphus sp., Ferrissia sp. (greater abundances in lentic waterbodies)
Tanytarsini, Micronecta spp., Hydroglyphus sp., Tanypodinae, Corbicula sp., Baetidae, Hydracarina, Orthocladinae (greater
abundances in 2005); Ostracoda, Oligochaeta, Nematoda, Gyraulus sp., Angrobia sp., Ferrissia sp., Cyclopoida, Neoplea sp.,
Ceratopogoninae (greater abundances in 2006)
Corbicula sp. pa, Gyraulus sp. pa, Tanytarsini, Nematoda, Oligochaeta, Angrobia sp. pa, Cyclopoida, Hydroglyphus sp.,
Orthocladinae, Cyclestheria sp. pa, Ferrissia sp., Baetidae, Triplectides spp. (greater abundances in the Gregory catchment);
Ostracoda, Micronecta spp., Tanypodinae, Hydracarina (greater abundances in the Flinders catchment)
Year 2005 v. 2006 0.534
Catchment Gregory v. Flinders 0.362
Flow status Lotic v. Lentic 0.784
Year 2005 v. 2006 see above
* P < 0.05; ** P < 0.01; NS p is non-significant; † Significance of P-values adjusted according to the Bonferroni t-test, which corrects for the possibility of multiple tests (catchment x
flow status; catchment x year; flow status x year) being simultaneously correct. This was used instead of a three-way factorial ANOSIM, and reduced the significance level, , to
0.017 (0.05/3). pa indicates contribution of taxa to the difference between groups is due to presence/absence rather than abundance.
Table 4: Correlations between composition of macroinvertebrate samples (based on taxonomic abundances or functional feeding group proportions,
‘function’) and combinations of environmental variables (BIOENV results) for the study region during the 2005 dry season.
Taxonomic composition: best variable combination (rs) Functional composition: best variable combination (rs)
one variable two variables
D, CC, WV
UC, MP, A
TN, NH4, pH
> three variables
D, CC, WV, BS
UC, MP, A, MA
TN, NH4, pH,
UC, MP, %LL,
%AV, NH4, pH
CC, D, LS
MP, MA, A
Hab, ChlS, %AV
> three variables
CC, D, LS, CW
MP, MA, A, UC
Hab, ChlS, MA,
TN, %OSS, PO4,
%Si, CC, D
TN, %OSS, PO4
each set with rs >
D, depth mid-channel; CC, canopy cover mid-channel; WV, wetted volume of waterbody; BS, bank slope of waterbody; LS, length of straight section of waterbody; CW, channel
width. UC, presence or absence of undercuts; MP, presence or absence of macrophytes; A, presence or absence of micro-algae; MA, presence or absence of macro-algae. %AV,
proportion of aquatic vegetation (algae + macrophytes); %LL, proportion of leaf litter; %Sn, proportion of snags; %MA, proportion of macroalgae; Hab, number of different
macroinvertebrate habitats present; ChlS, chlorophyll a on the sediment surface; %Si = proportion of silt. TN, concentration of total particulate nitrogen; NH4, concentration of
ammonium-N; PO4, concentration of phosphate-P; %OSS, proportion of organic suspended solids. Bold rs values indicate the highest scores within one, two, three or more than three
variable combinations for non-combined environmental sets of variables. Italicised rs values indicate those combinations within each set of environmental variables with the highest
scores. Variable codes in bold indicate the combination with the highest rs value.
UC, %LL, %AV
Hab, TN, %OSS
Table 5: Mean values of diversity measures for groups within a priori-defined factors with significant differences (ANOVA), based on
macroinvertebrate abundance data for samples collected in the study region during the 2005 dry season.
Factor Diversity measure Group within factor Mean (S.E.) Group within factor Mean (S.E.) F statistic (P-value)
Catchment x flow status (interaction) Dab 16.66 (<0.0001)****
Lotic Gregory 6.1 (0.3)
Lotic Gregory 6.1 (0.3)
Lotic Gregory 6.1 (0.3)
Lotic 39 (2)
Lotic 0.27 (0.03)
Main channel 38 (1)
Main channel 5.8 (0.3)
Main channel 4.3 (0.4)
Main channel 0.27 (0.02)
Lotic Flinders 4.3 (0.3)
Lentic Flinders 3.8 (0.3)
Lentic Gregory 3.7 (0.2)
Lentic 27 (2)
Lentic 0.38 (0.03)
Off-channel 50 (3)
Off-channel 7.5 (0.5)
Off-channel 3.3 (0.1)
Off-channel 0.46 (0.02)
Lateral position (lotic Gregory waterbodies only)
Lateral position (lentic Flinders waterbodies only)
N = abundance; S = richness; D = Margalef’s index; BP = Berger-Parker index. a D was log-transformed prior to ANOVA to meet test assumptions; b t-tests for multiple comparisons
of group means within interaction terms corrected using the Bonferonni method; * P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001.
Figure legends Download full-text
Fig. 1: The Nicholson and Flinders catchments in the Gulf of Carpentaria drainage
division with detail of waterbodies sampled in the lower Gregory and Flinders river
systems during the 2005 dry season. Waterbodies re-sampled during the 2006 dry
season are underlined. Waterbody abbreviations refer to the catchment (Gregory = G,
Flinders = F), river section (downstream = D, mid = M, upstream = U) and lateral
position (main channel = m, off-channel = o) of each site (see Table 1 for more detail).
Fig. 2: Historical hydrographs of mean daily flow standardised by upstream catchment
area (ML d-1 km-2) at gauging stations () near waterbodies () sampled in the
Flinders and Gregory study regions (see map insets). Flinders River at Walkers Bend is
at FDm. Flinders River at Etta Plains, Cloncurry River at Canobie, and Gregory River at
Gregory Downs stations are approximately 80, 200 and 20 km upstream by middle
thread distance from FMm, FUm, and GUm, respectively. Broken arrows indicate
direction and ephemerality of flow connection among gauging stations in the Flinders
study region. Flows are all > 0 ML d-1 km-2 between wet season peaks for the Gregory
River at Gregory Downs station.
Fig. 3: Agglomerative dendrogram of macroinvertebrate samples collected from the
study region in the dry season of 2005 only for 11 waterbodies (a), and in both 2005 and
2006 for the four re-sampled waterbodies only (b), based on group-average linking on
Bray-Curtis sample dissimilarities from log-transformed abundance data.
Fig. 4: MDS ordinations on the Bray-Curtis dissimilarity matrix of log-transformed
abundance (a and b) and FFG proportion data (c), using the spatial dataset from 2005
alone [33 samples from 11 waterbodies in (a) and (c)] and together with the temporal
dataset of four waterbodies from the 2006 dry season [12 additional samples in (b)].
Samples are represented as waterbody centroids (mean ordination co-ordinates for n = 3
samples) with ± 1 standard error bars. Lotic waterbodies are represented by open
centroids, lentic by closed.
Fig. 5: Spatial and temporal variation among assemblages at different scales of
resolution, measured by pair-wise Bray-Curtis dissimilarities [based on log-transformed
abundance data (a and b), and FFG proportion data (c)] within and between waterbodies
and between years, for the 11 waterbodies sampled in the 2005 dry season (a and b) and
for the four waterbodies sampled in both the 2005 and 2006 dry seasons (c).
Fig. 6: Mean relative abundances of taxa within functional feeding groups for
waterbodies sampled in the 2005 dry season (presented with -1 standard error bars).
Fig. 7: Diversity measures (mean +1 standard error bars), based on macroinvertebrate
abundance data and habitat types for waterbodies sampled from the study region in the
2005 dry season. N = abundance; S = richness; D = Margalef’s index; BP = Berger-
Parker index; = beta-diversity (single value).
Fig. 8: A conceptual diagram of dry season beta-diversity between macroinvertebrate
assemblages of waterbodies in the study region, in relationship to the hydrological
connectivity potential between any two waterbodies. Developed and modified from
ideas presented in previous studies and reviews (Ward et al., 1999; Ward & Tockner,
2001; Sheldon et al., 2003; Sheldon & Thoms, 2006).