Date post: | 24-Nov-2023 |
Category: |
Documents |
Upload: | independent |
View: | 1 times |
Download: | 0 times |
Pathways for algal recolonization in seasonally-flowingstreams
BELINDA J. ROBSON, TY G. MATTHEWS, PETER R. LIND AND NIGEL A. THOMAS
School of Life and Environmental Sciences, Deakin University, Warrnambool, Vic., Australia
SUMMARY
1. In semi-arid climates, seasonally-flowing streams provide most of the water required
for human use, but knowledge of how water extraction affects ecological processes is
limited. Predicted alterations in stream flows associated with the impacts of climate
change further emphasize the need to understand these processes. Benthic algae are an
important base for stream food webs, but we have little knowledge of how algae survive
dry periods or respond to altered flow regimes.
2. We sampled 19 streams within the Grampians National Park, south-eastern Australia
and included four components: a survey of different drought refuges (e.g. permanent
pools, dry biofilm on stones and dry leaf packs) and associated algal taxa; a survey of
algal regrowth on stones after flows recommenced to determine which refuges contributed
to regrowth; reciprocal transplant experiments to determine the relative importance of
algal drift and regrowth from dry biofilm in recolonization; direct measurement of algal
drift to determine taxonomic composition in relation to benthic assemblage composition.
3. Algae showed little specificity for drought refuges but did depend on them; no
species were found that were not present in at least one of the perennial pool, dry biofilm
or leaf pack refuges. Perennial pools were most closely correlated with the composition of
algal assemblages once flows resumed, but the loss or gain of perennial pools that might
arise from stream regulation is unlikely to affect the composition of algal regrowth.
However, regulated streams were associated with strong increases in algal density in dry
biofilm, including increased densities of Cyanobacteria.
4. A model for algal recolonization in seasonally-flowing streams identified three
pathways for algal recolonization (drift-dependent, dry biofilm-dependent and
contributions from both), depending on whether streams are diatom-dominated or
dominated by filamentous algae. The model predicted the effects of changes to stream
flow regimes on benthic algal recolonization and provides a basis for hypotheses
testable in streams elsewhere.
Keywords: climate change, drought refuges, freshwater algae, intermittent streams, temporarystreams
Introduction
In Mediterranean and semi-arid climate regions,
seasonally-flowing streams are the most abundant
stream type and provide most of the water required
for human use (Gasith & Resh, 1999). Despite their
economic importance, ecological studies of intermit-
tent or seasonally flowing streams are scarce in
comparison to studies of perennial streams, so we
have little understanding of the effects of water
extraction. Water extraction may prolong dry periods
in streams or increase their frequency and this may
affect the survival of stream biota. Some recent studies
have identified different types of drought refuge for
Correspondence: Belinda J. Robson, School of Life and Envi-
ronmental Sciences, Deakin University, PO Box 423 Warrnam-
bool, Vic. 3280, Australia. E-mail: [email protected]
Freshwater Biology (2008) 53, 2385–2401 doi:10.1111/j.1365-2427.2008.02061.x
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd 2385
both algae (e.g. Robson & Matthews, 2004) and
animals (e.g. Magalhaes et al., 2002). However, the
recolonization dynamics of biota in these streams is
still largely unknown.
The stream canopy in Mediterranean and semi-arid
regions is typically more open than in temperate
areas, allowing greater light penetration to the stream
bed and creating ideal conditions for algal growth.
Benthic algae are an important food source in many
streams, including those in arid regions (e.g. Benenati,
Shannon & Blinn, 1998; Bunn, Davies & Winning,
2003), so it is important to understand their response
to impacts such as altered flow regimes. However,
there is little knowledge of algal recolonization pro-
cesses following periods without surface flow. Studies
of algal responses to flow regulation have found that
rapid wetting and drying leads to increased growth of
Cyanobacteria and reduced growth of green algae and
diatoms that are more palatable to grazers (e.g. Blinn
et al., 1995; Benenati et al., 1998; Robson, 2000; Ryder,
2004). Under conditions of rapid drying, algal taxa
respond differently depending on their ability to
retain moisture (Benenati et al., 1998) and the presence
or absence of surface water during dry periods may
affect the density of subsequent algal growth (Robson
& Matthews, 2004).
The rate of water loss when streams dry out is also
likely to influence the source of recolonization (in situ
biofilm regrowth versus recolonization from external
sources, Stanley, Fisher & Jones, 2004). Field experi-
ments examining the impacts of desiccation on algal
recolonization show that when drying occurs rapidly
(i.e. stones removed from water and placed higher on
the bank) there is a devastating effect on algal
recovery and regrowth from dry biofilm (Dodds et al.,
1996; Benenati et al., 1998; Mosisch, 2001). In contrast,
when drying occurred naturally and thereby more
slowly, regrowth from dry biofilm is apparent (Rob-
son, 2000; Robson & Matthews, 2004); similar results
have also been observed with cultured algae (Peter-
son, 1996a and references therein). Peterson (1987)
showed that there may be an interaction between the
current speed to which a developing assemblage is
exposed and its subsequent desiccation resistance;
assemblages sheltered from current showed lower
desiccation resistance. This suggests that, in regulated
streams, prolonged periods of slow flow may encour-
age development of an algal assemblage that is
less resistant to subsequent desiccation than an
assemblage that develops in unregulated streams
with fast currents. Benenati et al. (1998) concluded
that desiccation-induced changes in algal composition
would have strong effects on the quality of the algal
food base of the Colorado River and, hence, on animal
communities. Therefore, alterations to flow regimes of
seasonally-flowing streams may affect the contribu-
tion of primary productivity to stream food webs.
It is difficult to predict the effects of flow modifi-
cation on algal growth without an understanding of
the recolonization process. Stanley et al. (2004) sug-
gested that primary productivity should be examined
at the landscape scale, but their model assumed that
the connectivity of streamflow was uninterrupted
across the landscape. In many catchments where
intermittency predominates, flow patterns are discon-
nected by weirs, pipelines and inter-basin transfers
and the use of streams as irrigation channels (Hughes,
2005). Therefore, to determine effects of these artificial
disconnections and re-connections on algal growth, it
is necessary to understand recolonization processes at
a smaller spatial scale than whole catchments. The
present study aimed to determine the pathways for
recolonization in seasonally-flowing streams and to
evaluate effects of flow regulation on recolonization
and productivity (growth).
The bed of seasonally-flowing streams may provide
a range of microhabitats suitable as refuges for algae
during periods without surface flow (Fig. 1). Many of
these streams retain perennial pools, where algae
continue to grow and proximity to these pools may
influence recolonization (Dodds et al., 1996). Algae
regrow from dry biofilm that remains on stream
stones, once flows resume (Robson, 2000). There are
also places that may be cooler and retain more
moisture than the exposed stone surfaces, such as
patches of moss, accumulations of leaves, seeps and
woody debris. Dry sediments and leaf litter may also
be suitable drought refuges (Davis, 1972). These
refuges differ in their vulnerability to flow regulation:
perennial pools are more likely to disappear with
increased water extraction (Hughes, 2005) whereas
dry biofilm is invulnerable. Conversely, stream regu-
lation may create perennial (weir) pools in streams
that previously dried out completely. An important
aspect of the concept of a refuge is that it must make a
significant contribution to recolonization of the main
habitat areas post-disturbance, otherwise it is merely a
remnant. We therefore identified refuges that are
2386 B. J. Robson et al.
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd, Freshwater Biology, 53, 2385–2401
important for algal recolonization and whether dif-
ferent taxa showed specific requirements for particu-
lar refuge types.
For recolonization to occur from these refuges once
flows resume, it is assumed that algae will drift in the
current and attach to stones downstream of their
source (Peterson, 1996a). Algal drift is rarely mea-
sured directly in intermittent streams, but is recog-
nized as an important recolonization mechanism in
perennial streams (McCormick & Stevenson, 1991;
Biggs, 1996). Therefore, we measured algal drift
directly to substantiate it as a viable mechanism for
algal regrowth in seasonally-flowing streams.
In an experiment where the dry biofilm was
removed prior to the resumption of flow, Robson &
Matthews (2004) showed that dry biofilm made a
strong contribution to regrowth, but that this contri-
bution was also an order of magnitude more important
in streams with perennial surface water (pools) than in
streams that dried out completely. This suggested an
interaction between two potential drought refuges;
dry biofilm and perennial pools (Fig. 1). This interac-
tion may be due to increased entrapment of algae
drifting from pools, or increased density of cells in the
dry biofilm in streams with perennial surface water, or
both (Robson & Matthews, 2004). Peterson (1987)
suggested that, by stimulating mucilage production,
desiccation could enhance subsequent entrapment of
drifting cells. Similarly, coating substrata with agar led
to a twofold increase in diatom immigration from the
drift (Stevenson, 1983). This is one mechanism by
which the density of cells in re-wetted dry biofilm
might affect the contribution of drift to algal recol-
onization (Fig. 1) creating an interaction between
refuges. To evaluate the relative contribution of algal
drift and regrowth from dry biofilm, we did three
reciprocal transplant experiments using stream stones
containing established dry biofilm.
In summary, we determined pathways for algal
recolonization in seasonally-flowing streams by
addressing the following questions: first, do algae
depend on refuges (Stanley et al., 2004) and, if so, do
different taxa use particular refuges and do particular
refuges contribute more to subsequent regrowth?
Secondly, do algae disperse from drought refuges
into the drift once flows resume? Thirdly, is the arrival
of drifting algae onto stones a substantial contributor
to algal regrowth and is this contribution affected by
the taxa drifting or by the assemblage already on
stream stones?
Methods
Study area
We sampled 19 intermittent streams within the
Grampians ⁄Gariwerd National Park, western Victoria,
Australia [Table 1; Fig. 2: see descriptions in Robson
(2000), Robson & Matthews (2004) and Cowell,
Matthews & Lind (2006)]. The streams are stony, with
boulder-cobble substrata of Grampians sandstone ⁄siltstone. Vegetation is typical Australian dry sclero-
phyll forest and woodland, dominated by Eucalyptus
species. The climate is semi-arid (400–600 mm average
rainfall per annum) and most streams are seasonal,
flowing from June to December each year.
Study design
The study comprised four components: a survey of
algal taxa present in different drought refuge types; a
survey of algal regrowth on stones after flows recom-
menced to determine which refuge types contributed
to regrowth; reciprocal transplant experiments to
drift
In situ regrowth and/or entrapment of propagules in re-establishing biofilm
stone tops
Dry
leaves
pools sand
wood
Flowing
Fig. 1 Diagram summarizing possible recolonization mecha-
nisms. Left hand side shows potential drought refuges for algae
when stream beds are dry. It is not known which of these ref-
uges contribute most to algal regrowth, nor which taxa use each
refuge type. Right hand side describes three recolonization
methods: (i) drift; (ii) in situ regrowth from dry biofilm and (iii)
enhanced entrapment of drifting algae by regrowing biofilm. It
is not known whether algae enter the drift immediately after
flows resume, or only after in situ regrowth on stones from dry
biofilm followed by active drift. Lastly, neither relative contri-
butions to recolonization by each of these mechanisms are
known, nor is the role of taxonomic composition understood.
Pathways for algal recolonization 2387
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd, Freshwater Biology, 53, 2385–2401
determine the relative importance of algal drift and
regrowth from dry algal biofilm in the recolonization
processes; and, lastly, direct measurement of algal drift
to determine taxonomic composition in relation to
benthic assemblage composition. For each component,
data from different streams were used (Table 1; Fig. 2).
This was partly due to the desire randomly to select
streams for each component, thereby ensuring inde-
pendence among components and maximize the
generality of our results. In some cases, it was also
performed to balance sampling designs for our analy-
ses. In all cases algae were quantified by counts of live
cell units per unit area or volume (see Robson, 2000).
Drought refuge survey
Our previous studies have shown that dry biofilm on
stones is the predominant drought refuge in several
Grampians streams (Robson, 2000; Robson & Mat-
thews, 2004; Cowell et al., 2006). However, we recog-
nized that other potential drought refuges may have
also existed. So we surveyed 19 streams during March
& April, 2004 for the frequency of six other potential
drought refuges: leaf packs, woody debris, seeps,
pools, dry sediment and patches of moss. Two people
intensively searched 100 m of streambed counting the
frequency of each refuge type. A further 400 m of
streambed was surveyed upstream for the presence of
pools or perennially flowing sections. We found no
seeps, and moss and woody debris were also rare or
absent from many stream sections, and therefore these
refuges are not considered further. Living algae were
randomly sampled from submerged rocks in six
streams that contained permanent surface water
(Table 1) using methods described in Robson &
Matthews (2004). In addition to pool samples, six
Table 1 Streams used in this study, with their latitude and longitude
Regulated R F M T D Unregulated R F M T D
+ Pools
Reservoir Creek mainstem * * Boggy Creek *
37.347�S, 142.616�E 37.107�S, 142.274�E
2nd Wannon Creek * * *B Bovine Creek * * * *F
37.317�S, 142.541�E 37.239�S, 142.538�E
Dairy Creek * * * Barneys Creek * * * *F
37.209�S, 142.533�E 37.222�S, 142.582�E
Fyans Creek * Mt William Creek * * * *B
37.251�S, 142.536�E 37.322�S, 142.619�E
Brown Creek * Reservoir Creek – South Fork† * *B
37.458�S, 142.19�E 37.347�S, 142.611�E
No. 1 Creek
37.357�S, 142.217�E
) Pools
1st Wannon Creek * * * Muline Creek
37.307�S, 142.540�E 37.207�S, 142.262�E
Cultivation Creek
37.25�S, 142.28�E
Middleton Creek *
37.225�S, 142.532�E
Unnamed Creek * *F
37.259�S, 142.619�E
Bomjinna Creek * * * *F
37.277�S, 142.619�E
Stockyard Creek * * *
37.329�S, 142.519�E
Grevillea Creek * * * *B
37.228�S, 142.547�E
All 19 streams were sampled to determine frequency of drought refuge types. A subset of streams (*) were used for: R, drought refuge
survey; F, stream survey after 1 week of flow; M, Mantel tests of relationship between drought refuges and streams post-flow; T,
reciprocal transplant experiments; D, drift sampling (*B = bacillariophyte-dominated streams; *F = filamentous-dominated streams).
Dairy Creek was not sampled 1 week after flow (i.e. n = 18 for flow survey after 1 week of flow). Stream names marked ‘†’ were
sampled upstream of weirs marked in Fig. 2. Stream names without ‘†’ were sampled downstream of those weirs.
2388 B. J. Robson et al.
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd, Freshwater Biology, 53, 2385–2401
randomly chosen leaf packs and six randomly chosen
dry rocks with dry biofilm were also collected and
taken back to the laboratory and rewetted to culture
any dormant algae from within the dry biofilm.
Samples were taken only from the thalweg to ensure
that the selected stones were inundated during the
previous winter. Leaf packs and rocks containing dry
biofilm were cultured in distilled water for 48 h before
being sampled and preserved for later counting
(Robson, 2000). This short culture period enabled
algal taxa to be identified and ensured that relative
abundances reflected those of the dry biofilm. A
longer culture period might have allowed interactions
between taxa to alter relative abundances and
increased the risk of contamination from other
sources. There was no evidence of such contamination
(such as novel taxa on laboratory stones) during the
study.
We aimed to sample each refuge type from four
different stream categories (Table 1). The presence or
absence of pools in each stream was determined by
searching 500 m upstream from each study site during
the dry season. Where water was found, sites were
revisited in late autumn to determine whether it had
persisted throughout the dry period. Note that it is
possible with our design that some streams designated
as lacking permanent surface water did have perma-
nent pools at the top of the catchment beyond our
500 m search range. However, we had decided
in an earlier study (Robson & Matthews, 2004) that
pools more than 500 m upstream from sampling
sites were unlikely to influence algal recolonization.
N
To GlenelgRiver
Muline
Cultivation
No.1
LakeBellfield
Bovine
Barneys
Grevillea
Dairy
Middleton
Brown
To WannonRiver
Wannon River
Stockyard
2nd Wannon
1st Wannon
Fyans
Reservoir
Reservoir
south fork
mainstem
Interbasin transfer
Pools present
Pools absent
Perennial
Bomjinna
Unnamed
Mt William
Fig. 2 Schematic diagram showing presence and absence of pools or permanent surface water in 18 of the 19 streams sampled.
Boggy Creek not shown as it was in an isolated catchment. Second-order streams are shown as forked lines, first-order streams are
simple lines. Weirs are indicated with curved solid lines.
Pathways for algal recolonization 2389
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd, Freshwater Biology, 53, 2385–2401
Unfortunately, only one regulated stream without
pools was found among the 19 streams surveyed (1st
Wannon Creek; Table 1). Therefore, only the three
remaining stream types were used for univariate ana-
lyses of potential drought refuges (see Data analyses).
Stream survey after 1 week of flow
We used a broad-scale survey of benthic algae in 18
streams to provide an overview of the variation that is
possible among seasonally-flowing streams (Table 1).
Two hypotheses were tested; first, that algal density
and ⁄or algal assemblage structure differ between
regulated and unregulated streams (both with per-
manent surface water) and, secondly, that mean algal
density or algal assemblage structure differ between
unregulated streams with and without permanent
surface water.
To test these hypotheses, three streams were
randomly selected from each of the three stream
categories (nine streams in total) for univariate anal-
yses and two sites were sampled on each of these
streams (Table 1). Six replicate samples were taken
from the tops of individual stones at each site, as
described by Robson & Matthews (2004). In addition,
data from nine streams were used for comparison
with the drought refuge data. That is, data from the
same nine streams (Table 1) were used to determine
the degree of association between the pres-
ence ⁄absence of algal taxa and the assemblage com-
position in drought refuges and on stream stones once
flows resumed.
Reciprocal transplant experiments
Three pairs of streams were used. While streams were
dry, six marked stones were transplanted between Mt
William Creek (unregulated, with pools) and Bomj-
inna Creek (unregulated, no pools), six were trans-
planted between Grevillea Creek (unregulated, no
pools) and Barney Creek (unregulated, with pools)
and six were transplanted between Grevillea Creek
and Dairy Creek (regulated, with pools). In addition,
six stones from each stream were transported to the
laboratory (same algal culture method as drought
refuge samples). Once streams had resumed flow for
1 week, laboratory stones and marked stones were
sampled along with six unmarked, randomly chosen,
non-transplanted stones from each stream (methods
as above). Therefore, for each stream, it was possible
to compare assemblage composition from dry biofilm
in the absence of any external colonization (=labora-
tory stones), with stones exposed to both recoloniza-
tion by drift and from local dry biofilm (=in situ
stones) and, finally, with stones exposed to recolon-
ization by a drift flora that differed from the dry
biofilm on stones (=transplanted stones).
Algal drift
Eight streams were determined a priori to be domi-
nated by either diatoms (four streams) or filamentous
algae (four streams) (Table 1). Drift samples were
collected from each of these streams with a 35 lm
plankton net (0.038 m2 net opening) supported by a
steel frame. While some small stream algae would
pass through a 35 lm mesh net, a finer mesh creates
too much resistance in the stream flow. A wooden
0.045 m block and U-bolt connected the lower tie
down strap to the net frame, providing 5 cm clearance
above the bed (Culp, Scrimgeour & Beers, 1994). The
net was fitted with a threaded flange allowing
attachment of a 120 mL sample container and posi-
tioned in the thalweg for sampling. Placement of the
driftnet ensured avoidance of chutes, boulder clusters,
woody debris, leaf packs and depths <0.08 m. Net
opening height out of the water was measured (to
enable net surface area calculation) and the depth of
the net underwater was recorded when the net was
full. The net was lifted at the end of each sampling
time and the inside was washed into the collection pot
with 500 mL of stream water. After each sample
collection, the pot was removed and the inside of the
net was again flushed with water.
A MiniAir2 flow meter (Schiltknecht Messtechnik
AG, Switzerland), was used to measure water velocity
at the net mouth. Velocity was measured in the net
opening at half net water depth for the first and last
30 s of sample time (10 min – determined by a pilot
study) providing a mean measurement for each 30-s
interval. The net entrance velocity consisted of the
mean of these two 30-s readings. The total volume of
water sampled was calculated as [(net entrance veloc-
ity, m s)1) · (cross-sectional area of net entrance in
m2) · (sample interval in s)] (Faulkner & Copp, 2001).
Streams were randomly allocated for sampling
either in the morning or afternoon (09:00–12:00 hours
and 13:00–16:00 hours) because a pilot study revealed
2390 B. J. Robson et al.
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd, Freshwater Biology, 53, 2385–2401
no clear diel rhythm of drift. Six random drift and six
random epilithic samples (methods as above) were
collected from a randomly selected 50 m stream
length, working upstream. All epilithic and drift
sampling equipment was flushed with distilled water
between use at different streams, to prevent cross-
contamination of samples. Algae from both benthic
and drift samples were preserved, counted and
identified using the methods described above.
Data analyses
All univariate analyses were performed using SYSTATSYSTAT
version 10. Assumptions of ANOVAANOVA (normality and
homoscedasticity) were checked by inspecting resid-
ual plots and data were log-transformed if necessary
(Quinn & Keough, 2002). The Peritz procedure was
used for post hoc pairwise comparisons of significant
treatments after ANOVAANOVA (Quinn & Keough, 2002).
Multivariate analyses (with data square-root transfor-
mations to reduce the influence of very abundant
taxa) used Bray–Curtis similarities using PRIMERPRIMER
software (version 5; Clarke & Gorley, 2001).
Drought refuge survey
The dependent variables total density, species rich-
ness and the total density of Cyanobacteria, chloro-
phytes and bacillariophytes were compared across the
three stream categories (fixed, three levels: regulated
streams with pools; unregulated streams with pools;
and unregulated streams without pools) for two of the
three refuge types (i.e. omitting the pool refuge type
which did not occur in all three stream categories).
Secondly, we compared the dependent variables
across all three drought refuge types in regulated
and unregulated streams with pools. We used a
partly-nested mixed-model ANOVAANOVA with the main
effects: stream category (fixed, two levels: regulated
streams with pools versus unregulated streams with
pools), refuge type (fixed, three levels: leaf packs,
pools and dry biofilm) and Stream (stream category)
(random, three streams nested within each stream
category).
Algal community composition was analysed in two
ways: comparison of algal assemblages across refuge
types by pooling data across stream categories (10
streams including 1st Wannon Creek); and compari-
son of algal assemblages in refuge types across
regulated and unregulated streams with pools (i.e.
streams without pools excluded = six streams rather
than 10) using a combined factor (refuge type · stream
category; see Robson & Matthews (2004) for method).
Stream Survey after 1 week of flow
The same dependent variables used for the drought
refuge survey were compared across sites, streams
and stream categories using a three-factor fully nested
ANOVAANOVA. The three factors were stream category (fixed,
three levels: regulated streams with pools; unregu-
lated streams with pools; unregulated streams with-
out pools), Stream (stream category) (random, three
streams nested within each stream type) and Site
[Stream (stream category)] (random, two sites nested
within each of three streams nested within each of
three stream categories, two levels) (Table 1).
Ordination of samples from 18 of the 19 streams
was used to compare assemblage composition across
all four stream categories (Dairy Creek did not flow
for long enough for algae to be sampled). A hypoth-
esis test of differences in assemblage composition
between stream categories was performed using the
same nine streams as for the univariate tests.
Relationship between assemblage composition of drought
refuges and streams after 1 week of flow
The composition of algae in each of the three dom-
inant drought refuge types (dry biofilm, leaf packs
and pools) was related to community composition of
algae after 1 week of flow using Mantel tests
(RELATE function in PRIMERPRIMER) on square-root trans-
formed data for nine streams (Table 1). The weighted
Spearman rank correlation was used (Clarke & War-
wick, 1994), with 999 permutations.
Reciprocal transplant experiment
Single factor ANOSIMANOSIM was used to analyse algal
assemblage composition data from the three
reciprocal transplant experiments. Univariate analy-
ses were not used for these data, because the purpose
of these experiments was to identify the dominant
colonization source based on taxonomic composition.
Total abundances would have differed among treat-
ments as a simple product of the experimental design
(i.e. stones cultured in the laboratory would always
Pathways for algal recolonization 2391
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd, Freshwater Biology, 53, 2385–2401
have had fewer individuals). We had no a priori
reason to believe that the number of species would
vary among any of the treatment levels.
For each experiment (pair of streams), the factor
‘treatment’ was analysed with six levels: laboratory
stones stream 1, laboratory stones stream 2, in situ
stones stream 1, in situ stones stream 2, stones
transplanted into stream 1 and stones transplanted
into stream 2. All three experiments showed signifi-
cant differences among treatments, so pairwise com-
parisons were used to compare treatment levels.
Relationship between algal drift and benthic algal
composition
Differences between the composition of algal drift and
epilithic assemblages were examined using a Bray–
Curtis similarity matrix. Presence ⁄absence data were
used to eliminate the effect of systematic differences
in abundances between drift and epilithic samples.
Differences in taxonomic composition between drift
and epilithic samples were examined by two-way
ANOSIMANOSIM. Factors used were stream and sampling
method (drift ⁄epilithic).
Results
Relative frequency of occurrence of refuge types
Only three drought refuges were common and exten-
sively available to algae, including dry biofilm on
stones, dry biofilm on leaf packs and living algal
biofilm on stones in pools. No species were found in
the post-flow stone sampling that were not found in
one of these three refuges. Therefore, only these three
refuge types were considered further.
Drought refuge survey
Patterns of total algal density, total Cyanobacteria
density and total bacillariophyte density were similar
for both the pools and leaf pack refuge types (Figs 3 &
4). These three dependent variables showed signifi-
cant stream-to-stream variation in both pool (density:
F4,30 = 16.7, P < 0.01; Cyanobacteria: F4,30 = 26.2,
P < 0.01; Bacillariophyta: F4,30 = 7.5, P < 0.01) and
leaf pack refuge types (density: F4,30 = 13.5, P < 0.01;
Cyanobacteria: F4,30 = 3.4, P < 0.01; Bacillariophyta:
F4,30 = 12.2, P < 0.01), but did not vary significantly
0
1
2
3
4
5
6
7
Unregulated + poolsUnregulated –pools
Regulated + pools
0
0.5
1
1.5
2
2.5
3
3.5
40
1
2
3
4
5
60
2
4
6
8
10
12
140
1
2
3
4
5
6
7 Dry biofilmLeaf pack
Lo
g10
(T
ota
l alg
al d
ensi
ty)
Lo
g10
(Cya
no
bac
teri
a)L
og
10 (C
hlo
rop
hyt
es)
Lo
g10
(Bac
illar
iop
hyt
es)
Sp
ecie
s ri
chn
ess
Fig. 3 Log-transformed [log10(x)] total algal densities, species
richness, Cyanobacteria, Chlorophyta and Bacillariophyta sam-
pled from the three stream categories for two of the three
drought refuge types (i.e. pool refuge type was omitted leaving
dry biofilm and leaf packs refuge types, each bar represents a
single stream, n = 6, units = no. individuals per 36.3 cm2). All
data represent least square means ± 1 SE.
2392 B. J. Robson et al.
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd, Freshwater Biology, 53, 2385–2401
across the three stream categories. The only significant
variation across the three stream categories existed for
both total algal densities (F1,4 = 8.7, P < 0.05) and the
Cyanobacteria in the dry biofilm (F1,4 = 13.1,
P < 0.01). Total algal densities in the dry biofilm were
higher in regulated streams with pools than in the
unregulated stream types, which did not differ (Peritz
Pairwise comparison) (Figs 3 & 4). Total cyanobacte-
rial density differed across all three stream category
combinations (Peritz Pairwise comparisons) and was
highest in regulated streams with pools and lowest in
unregulated streams with pools (Fig. 3). Total chloro-
phyte density was not significantly influenced by
stream category, but did show stream-to-stream var-
iation in pools (F4,30 = 3.8, P < 0.05) and in dry biofilm
(F4,30 = 4.1, P < 0.01; Fig. 4).
The predominant taxa sampled during the drought
refuge survey were the diatoms Fragilaria spp. and
Gomphonema spp. and some unidentified Cyanobacte-
ria taxa (Table 2). Variation in algal composition was
clearly lower among leaf pack samples than from the
dry biofilm or living biofilm on pool stones (Fig. 5)
and there were significant differences in algal com-
position across the three refuge types (Global
R = 0.29, P = 0.001). The greatest differences were
between leaf packs and pools (R = 0.39, P = 0.001)
and leaf packs and dry biofilm (R = 0.37, P = 0.001).
Differences in algal composition between pools and
dry biofilm were small, but significant (R = 0.09,
P = 0.004).
Algal composition differed across regulated and
unregulated streams with pools for each of the three
refuge types (Global R = 0.28, P = 0.001; pools,
R = 0.225, P = 0.002; leaf packs, R = 0.561, P = 0.001;
dry biofilm, R = 0.15, P = 0.02; Table 2). There was
small, but significant variation in algal composition
(data pooled across the three drought refuges) across
the three replicated stream categories: between unreg-
ulated and regulated streams with pools (R = 0.11,
P = 0.001) and unregulated streams with and without
pools (R = 0.09, P = 0.005), but no difference was
detected between regulated streams with pools and
unregulated streams without pools (R = 0.04,
P = 0.06).
Stream survey after 1 week of flow
None of the dependent variables (total algal density,
species richness or Cyanobacteria, Chlorophyta and
0
1
2
3
4
5
6
7
Regulated + poolsUnregulated + pools
0
0.5
1
1.5
2
2.5
3
3.5
40
1
2
3
4
5
60
2
4
6
8
10
12
14
16
180
1
2
3
4
5
6
7 Dry biofilmLeaf packsPools
Lo
g10
(T
ota
l alg
al d
ensi
ty)
Lo
g10
(Cya
no
bac
teri
a)L
og
10 (C
hlo
rop
hyt
es)
Lo
g10
(Bac
illar
iop
hyt
es)
Sp
ecie
s ri
chn
ess
Fig. 4 Log-transformed [log10(x)] total algal densities, species
richness, Cyanobacteria, Chlorophyta and Bacillariophyta sam-
pled from the three drought refuge types in regulated and
unregulated streams with pools (each bar represents a single
stream, n = 6, units = no of individuals per 36.3 cm2). All data
represent least square means ± 1 SE. Note: unregulated streams
without pools had to be omitted due to the absence of pool
refuges in these streams.
Pathways for algal recolonization 2393
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd, Freshwater Biology, 53, 2385–2401
Bacillariophyta densities) differed across the three
stream categories after flows had resumed (Table 2;
Fig. 6). There was significant stream-to-stream vari-
ability for species richness (F6,9 = 10.1, P < 0.01),
chlorophytes (F6,9 = 8.0, P < 0.01) and the Cyano-
bacteria (F6,9 = 3.9, P < 0.05), and significant site-to-
site variation for total density (F9,90 = 2.1, P < 0.05),
chlorophytes (F9,90 = 2.1, P < 0.05) and the Cyano-
bacteria (F9,90 = 2.5, P < 0.05) in some streams (Fig. 6).
The predominant taxa sampled after flow
resumed were similar to those sampled during the
survey of drought refuges (Table 2). Differences in
algal composition were detected across the four
stream categories using 18 streams (Fig. 7), but
hypothesis tests showed that these differences were
small (e.g. Global R = 0.08, P = 0.001) regardless of
whether 1st Wannon Creek was included or
excluded from analyses.
Relationship between algal composition in drought
refuges and in-stream algal composition after 1 week
of flow
Neither leaf pack nor dry biofilm refuges were
strongly correlated with the assemblage composition
(species · abundance) that developed a week after
flows resumed in any of the stream categories (leaf
pack q = 0.184, P = 0.001; dry biofilm q = 0.178,
P = 0.001). In addition, algal assemblage composition
in these two refuge types was strongly correlated
(q = 0.446, P = 0.001) indicating substantial similarity
in assemblage composition. In contrast, algal assem-
blage composition of pool refuges was more strongly
correlated with assemblages that grew on stones in
streams after flows resumed (q = 0.332, P = 0.001).
Reciprocal transplant experiments
For the experiment in Mt William Creek and Bomjinna
Creek, treatments differed significantly (Global
R = 0.484, P < 0.001). Differences in assemblage
composition occurred between laboratory stones
Leaf packsPoolDry biofilm
Stress: 0.13
Fig. 5 Non-metric dimensional scaling (NMDS) ordination
comparing algal assemblages sampled from leaf packs, dry
biofilm and pools (pooled across stream categories, n = 6 sam-
ples of each refuge type within each stream). These multivariate
data include 1st Wannon Creek, the only regulated stream
without pools available for sampling (i.e. inclusion of all 10
streams, Table 1).
Table 2 Dominant algal taxa collected during drought refuge study (all 10 streams) and 1 week after flow (all 18 streams)
Drought Survey Stream category Dominant algal taxa
Leaf packs Regulated with pools Fragilaria spp., Gomphonema spp., Cyanobacteria sprout., unknown sp. 1,
unknown Cyanobacteria sp. 1.
Unregulated with pools Fragilaria spp., Gomphonema spp., Tabellaria sp., unknown sp.,
Cyanobacteria sprout
Pools Regulated with pools Fragilaria spp., unknown Cyanobacteria sp. 1, Gomphonema spp.
Unregulated with pools Unknown sp. 2, Fragilaria spp., Gomphonema spp. Stigonema sp.
Dry biofilm Regulated with pools Unknown Cyanobacteria sp. 1, Stigonema sp., unknown sp. 2, Gomphonema spp.
unknown sp. 1
Unregulated with pools Fragilaria spp., unknown sp. 2, unknown sp. 1, Gomphonema spp.
Cyanobacteria sprout
1 week after flow Regulated with pools Fragilaria spp., Gomphonema spp., unknown Cyanobacteria sp. 1
Unregulated with pools Fragilaria spp., Gomphonema spp., unknown Cyanobacteria sp. 1
Regulated without pools Unknown Cyanobacteria sp. 1, Gomphonema spp., Fragilaria spp.
Unregulated without pools Fragilaria spp., Gomphonema spp., unknown Cyanobacteria sp. 1, Stigonema sp.,
unknown sp. 1
Taxa are listed in order of decreasing contribution to within sample variation. Only those contributing to >5% are shown.
2394 B. J. Robson et al.
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd, Freshwater Biology, 53, 2385–2401
(R = 0.435, P = 0.006), with each stream having a
distinct assemblage dominated by diatoms. In both
streams, diatom assemblages on laboratory stones
differed from those left in situ in each stream (Mt
William R = 0.624, P = 0.002; Bomjinna R = 0.367,
P = 0.004). This is likely to have resulted from the
different lengths of time each treatment was exposed
to inundation, as the laboratory stones were only
cultured for 48 h and had lower abundances of algae
than the in situ stones, which were exposed to flow for
a week. Interestingly, algal assemblages that grew on
transplanted stones in both streams did not differ from
those on in situ stones for that stream (Mt William
R = 0.046, P = 0.247; Bomjinna R = 0.128, P = 0.123).
That is, transplanted stones grew a diatom assemblage
characteristic of local conditions, not resembling the
assemblage that originated from their dry biofilm.
In Grevillea Creek and Barney Creek, experimental
treatments also differed significantly (Global R =
0.432, P < 0.001). Algal assemblages on laboratory
stones differed between streams (R = 0.507, P = 0.006),
with Barney Creek stones being dominated by fila-
mentous chlorophytes whereas Grevillea Creek stones
were diatom-dominated. When compared to in situ
stones, Barney Creek laboratory stones did not differ
from in situ in Barney Creek (R = 0.161, P = 0.067),
Unregulated + poolsRegulated + poolsUnregulated – poolsRegulated – pools
Stress: 0.1
Fig. 7 Non-metric dimensional scaling (NMDS) ordination
comparing algal assemblages after 1 week of flow across the
four stream categories. All 18 streams are included in this
example (n = 6 at each of two sites per stream).
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1
2
3
4
5
6
0
1
2
3
4
5
6
0
1
2
3
4
5
6
1 2 3 4 5 6 7 8 9Unregulated + pools Unregulated – pools Regulated + pools
0
1
2
3
4
5
60
Lo
g10
(T
ota
l alg
al d
ensi
ty)
Lo
g10
(Cya
no
bac
teri
a)L
og
10 (C
hlo
rop
hyt
es)
Lo
g10
(Bac
illar
iop
hyt
a)L
og
10 (S
pec
ies
rich
nes
s)Fig. 6 Log-transformed [log10(x)] total algal densities, species
richness, Cyanobacteria, Chlorophyta and Bacillariophyta sam-
pled after 1 week of flow from regulated streams with pools,
unregulated streams with pools and unregulated streams with-
out pools (each bar represents a site, each paired bar represents a
stream, n = 6 rocks sampled from each of two sites per stream,
units = no. individuals per 36.3 cm2). All data represent least
square means ± 1 SE. Note: there was only one regulated stream
without pools available for sampling (1st Wannon Creek,
Table 1), so it was omitted from the analysis. Note: scale on y-
axis differs for species richness.
Pathways for algal recolonization 2395
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd, Freshwater Biology, 53, 2385–2401
whereas Grevillea Creek laboratory stones did differ
from their in situ counterparts (R = 0.35, P = 0.011), as
in the experiment above. However, when in situ stones
from Grevillea Creek were compared to stones trans-
planted from Grevillea Creek to Barney Creek, there
was only a small, though significant, difference in the
algal assemblage (Barney Creek stones R = 0.235, P =
0.019) and the transplanted stones differed strongly
from those they were surrounded by in Barney Creek
(R = 0.563, P = 0.002). Therefore, in Barney Creek
(which was dominated by filamentous chlorophytes),
the in situ stones grew filamentous chlorophytes
similar to laboratory stones. The transplanted stones
placed in Barney Creek from Grevillea Creek grew a
diatom assemblage similar to that in Grevillea Creek,
indicating that regrowth of algae on these stones was
also dependent on the dry biofilm and that there was
little or no recolonization from the drift in Barney
Creek. In contrast, in Grevillea Creek, diatom drift was
the dominant pathway because the stones trans-
planted from Barney Creek to Grevillea Creek did
not differ from those in situ in Grevillea Creek
(R = 0.039, P = 0.316) and differed strongly from those
in Barney Creek (R = 0.68, P = 0.002), showing no
sign of the filamentous chlorophyte assemblage that
originated from Barney Creek dry biofilm.
Lastly, treatments for the Grevillea Creek and Dairy
Creek experiment also differed significantly (global
R = 0.676, P < 0.001). Both streams were diatom-
dominated, with distinct assemblages (laboratory
stones from both streams: R = 0.989, P = 0.002). Lab-
oratory stones and in situ stones from both streams
showed significant differences in their algal assem-
blages (Dairy Creek R = 0.35, P = 0.004; Grevillea
Creek R = 0.35, P = 0.011), probably because of the
different length of immersion, with notably lower
abundances on the laboratory stones. However, algal
assemblages on transplanted stones from the two
streams did not differ from each other (R = 0.276,
P = 0.05), but their in situ assemblages did differ
(Dairy Creek in situ compared with stones trans-
planted to Grevillea Creek R = 0.794, P = 0.002; Gre-
villea Creek in situ stones compared with stones
transplanted to Dairy Creek R = 0.659, P = 0.004),
showing that sufficient recolonization from algal drift
was occurring to create an intermediate assemblage
on the transplant stones that was a mixture of
regrowth from dry biofilm and arrivals from the drift.
Interestingly, while this intermediate assemblage on
transplanted stones differed markedly from the in situ
stones in Dairy Creek (R = 0.887, P = 0.002), it did not
differ from the in situ stones in Grevillea Creek
(R = 0.29, P = 0.058) or from the laboratory stones
from Grevillea Creek (R = 0.29, P = 0.05), because it
was more variable than the other treatments. There-
fore, diatom recolonization was occurring from both
the drift and dry biofilm in both streams, but diatom
taxa prevalent in Grevillea Creek were more success-
ful colonists than those from Dairy Creek, regardless
of the recolonization pathway.
Relationship between algal drift and benthic algal
composition
Individual drift and epilithic sample densities ranged
from 1.53 to 64.71 algal units per litre (n = 48) and
1870–1 788 480 algal units per sample (n = 48), respec-
tively. A total of 25 individual morphogenera from
four algal divisions were identified (Table 3), com-
prising similar taxa to the earlier components of the
study. Bacillariophyta was the dominant division
within both drift and epilithic samples, as well as
being represented in all samples collected from the
eight creeks (Table 1). Diatoms were also observed
‘rafting’ on both chlorophyte and Cyanobacteria fila-
ments, and also on leaf fragments. Chlorophyta was
the second most dominant division, with an uniden-
tified chlorophyte (sp. 11) present in both sampling
methods across all creeks. Green algae also contri-
buted a greater proportion of the algal units present in
the drift than their proportion of the epilithic assem-
blage. Cyanobacteria were a minor component of both
drift and epilithic samples. Stigonema sp. was the most
common cyanobacterial taxon, being absent only from
drift samples in 2nd Wannon Creek (Table 3).
Stream drift assemblages from the eight streams
showed significant differences (Global R = 0.792,
P = 0.001). Pairwise comparisons of drift assemblages
found that 2nd Wannon Creek was significantly
different to all other creeks (all R > 0.864, all
P = 0.002). There was a range of differences between
the bacillariophyte-dominated streams with Grevillea
Creek and Mt William Creek being significantly
different (R = 0.988, P = 0.002) whereas Mt William
Creek and Reservoir South Creek were more similar
(R = 0.327, P = 0.009). Grevillea Creek and an
Unnamed Creek were the only two that were not
clearly different (R = 0.259, P = 0.059).
2396 B. J. Robson et al.
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd, Freshwater Biology, 53, 2385–2401
Algal assemblages from epilithic surfaces differed
from drift assemblages (Global R = 0.144, P = 0.001).
The taxonomic composition of epilithic and drift
samples differed within each stream (R = 0.2– 0.6,
all P = 0.001) with the exception of Bomjinna Creek,
where the composition of the drift samples resembled
epilithic samples closely (R = )0.02, P > 0.05). In six
of the eight streams, densities of drifting algae
reflected epilithic algal densities, but drift densities
in the remaining two streams were either lower
(Reservoir Creek) or higher (2nd Wannon Creek) than
expected (Fig. 8).
Discussion
Drought refuges for stream algae
All taxa present on stream stones after 1 week of flow
were also found in at least one of the three drought
refuges (i.e. leaf packs, dry epilithic biofilm and living
biofilm in pools). Therefore, other potential drought
refuges (woody debris, moss and dry sediment) were
probably unimportant in algal recolonization because
they covered a much smaller area of the streambed
than leaf packs or epilithic biofilm. However, these
other drought refuges may be prominent in other
rivers; for example, woody debris has been shown to
be important in the lowland Murrumbidgee River,
Australia, and may provide a refuge under conditions
of moderately frequent fluctuations in river level
(Ryder, 2004).
The composition of assemblages in leaf pack and
dry biofilm refuges correlated poorly with assem-
blages that developed a week after flows resumed.
However, there was a stronger correlation with the
pool flora. This was possibly the product of higher cell
densities on pool stones compared to stones or leaves
cultured for 48 h, because there was a large overlap
between the flora of pools and cultured dry biofilm.
However, assemblages in these refuges were not
Table 3 Taxa identified in epilithic (E) and drift (D) samples
Taxa 2nd Wannon Bovine Grevillea Barneys Bomjinna Unnamed Mt William Reservoir South
Chlorophyta
Chlorophyta sp. 1 ED ED D ED ED ED ED ED
Chlorophyta sp. 2 ED D D ED E D E ED
Chlorophyta sp. 4 ED – – E ED – – D
Chlorophyta sp. 5 D D – ED ED E D –
Bulbochaete sp. E – – D ED – – D
Closterium sp. D D – – ED – – –
Chlorophyta sp. 10 ED D D D E D – –
Chlorophyta sp. 11 ED ED ED ED ED ED ED ED
Chlorophyta sp. 12 D – – – – D – –
Chlorophyta sp. 13 D ED ED ED – ED ED ED
Chroodactylon sp.* – – – – – – E D
Bacillariophyta
Achnanthidium sp. ED ED ED ED ED ED ED ED
Gomphonema sp. 1 ED ED D ED ED – ED ED
Gomphonema sp. 2 D ED – ED ED – D E
Fragilaria sp. 1 E D ED ED ED ED ED ED
Fragilaria sp. 2 ED ED D ED ED ED ED ED
Fragilaria sp. 3 ED D – E – D ED –
Navicula sp. – ED – ED ED – ED ED
Cyanobacteria
Stigonema sp. E ED ED ED ED ED ED ED
Hapalosiphon sp. E ED – E – D – –
Cyanobacteria sp. 1 E D D – – D – –
Loefgrenia anomala Gomont – – – – D E – D
Cyanobacteria sp. 2 – – – – – D D –
Cyanobacteria sp. 3 – – – – – D – –
Nostoc sp. – – – – – D D D
Bold letters indicates species found either in drift or in epilithic samples; –, not collected.
*Rhodophyta.
Pathways for algal recolonization 2397
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd, Freshwater Biology, 53, 2385–2401
particularly distinct, nor were they very different to
what grew after flows resumed, so it appears that
these taxa showed relatively little specificity for
different refuges. The predominance of diatoms in
leaf packs was interesting, but probably arose more
from this relatively unstable, short-lived refuge being
a poor substratum for filamentous algae than to any
‘preference’ shown by diatoms.
The role of permanent surface water (pools) in these
streams was less significant than expected, since few
taxa were found only in pools. Furthermore, pools
appeared to have limited influence on the density or
composition of regrowth once flows resumed across
the 18 streams analysed. Therefore, we found limited
evidence to support the conservation of pools (either
natural or artificial weir pools) as refuges. However,
pools are a vital refuge for some stream-dwelling
animals.
The effect of regulation
Robson & Matthews (2004) showed a strong positive
relationship between the presence of permanent
surface water and the density of algal regrowth after
a week of flow. However, the role of regulation was
unclear, as both streams with permanent surface
water were also regulated. Results from the present
drought refuge survey of a larger number of streams
confirm that higher cell densities within dry biofilm
coincide with stream regulation and that permanent
pools have a limited influence on density in epilithic
biofilm. Unregulated streams did not differ in mean
dry algal density, regardless of the presence or
absence of permanent pools, but regulated streams
with permanent pools did have higher densities.
Therefore, it is unlikely that the contribution of
drifting cell units washed from permanent pools is
responsible for the increased algal regrowth observed
by Robson & Matthews (2004). Rather, it appears that
permanent pools in regulated streams alter the role of
epilithic dry biofilm refuges by increasing their
quality (density of cells) and hence, capacity for
regrowth and possibly the rate of entrapment (Peter-
son, 1987).
This was particularly apparent for Cyanobacteria,
which were most dense in regulated streams with
pools and least dense in unregulated streams with
pools; unregulated streams without pools had inter-
mediate dry biofilm densities. Cyanobacteria are
known to be very resilient to frequent and rapid
drying (Blinn et al., 1995) and to be strongly reliant on
their persistence in the dry biofilm for regrowth
(Robson, 2000). They are also less prone to senescence
(and sloughing, Peterson et al., 1994), thereby increas-
ing the persistence of cyanobacterial mats. Therefore,
it is likely that part of the increased cell density in dry
biofilm in regulated streams is often, but not always,
due to Cyanobacteria responding positively to rapid
declines in water depth, as opposed to other taxa that
cannot tolerate rapid fluctuations in water levels.
Other algal taxa can also be abundant in the dry
biofilm of regulated streams, so this is not the only
mechanism responsible for increased cell densities in
regulated streams.
Several other mechanisms might explain increased
biofilm cell densities in regulated streams. Increased
reliance on dry biofilm could be induced by frequent
cessation and resumption of flows resulting from
5
10
15
20
25
30
0 200 000 400 000 600 000 800 000 1 000 000 1 200 000
Epilithic algal densities (units sample–1)
Algal density in drift (units. L–1)
1. 2nd Wannon 2. Bovine 3. Grevillea 4. Barneys 5. Unnamed 6. Bomjinna 7. Mt William 8. Reservoir South
1
7
3
4
2
68
5
Fig. 8 Average drift densities versus
average epilithic algal densities per stream
(±1 SE).
2398 B. J. Robson et al.
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd, Freshwater Biology, 53, 2385–2401
regulation (Dairy Creek and First and Second Wan-
non Creeks; B.J. Robson and T.G. Matthews, pers.
obs.). Regulation may also alter the onset of drying in
these streams: stream drying is likely to occur earlier
toward the end of the flow season for regulated
streams (spring–early summer) rather than later in the
summer for unregulated streams. This is likely to
reduce rates of desiccation and, hence, improve
successful regrowth from the dry biofilm in regulated
streams following the recommencement of flow.
Furthermore, regulated streams generally have a
lower discharge and the effects of small-medium
spates are damped compared to unregulated streams
(Hughes, 2005). This may reduce losses to scour
during the flow season, leading to higher cell densities
at cessation of flow. Matthaei, Guggelberger & Huber
(2003) showed that local disturbance history had a
greater effect on epilithic algal biomass than did
substratum particle size, water depth or near-bed
velocities. Peterson (1987) found that diatom-domi-
nated algal assemblages grew more densely in areas
sheltered from current in a regulated river, suggesting
that low, constant regulated flows might increase dry
biofilm biomass. However, Peterson (1987) also
showed that assemblages that developed in slower
flows were less resistant to both short-term desicca-
tion and scour.
In the past, it has been assumed that prolonged
periods of drying in seasonally-flowing streams that
result from water abstraction will not be a problem,
because biota will be adapted to this drying. This
appears to be true for algae since they are able to use a
range of refuge types, are adapted to desiccation and
produce a dry biofilm that will be viable in the
following year. However, as mentioned previously,
there are many instances where reduced flows in
regulated streams appear to favour the development
of a dry biofilm with a higher biomass and proportion
of Cyanobacteria. This is of concern since Cyano-
bacteria are less palatable to grazers than most algae
(e.g. Benenati et al., 1998; Peterson & Boulton, 1999).
Sources of recolonization
The reciprocal transplant experiment showed differ-
ent sources of colonists in different streams. In part,
this could be related to the structural composition of
the biofilm assemblage, as also shown by Peterson
(1996b). For example, recolonization in Barney Creek
occurred entirely from the dry biofilm on both in situ
and transplanted stones. This may be explained by
Barney Creek being dominated by filamentous green
algae and very few diatoms, so that drift, at least in
the first week of flow, was minimal. In contrast,
other streams were diatom-dominated and showed
either dependence on drifting propagules or a
mixture of taxa arriving from the drift or regrowing
from biofilm.
Sampling drifting cells later in the flow season
showed that propagule numbers were proportional
to benthic density in most streams, but both epilithic
and drift densities were lower than those reported in
previous studies of perennial streams (e.g. Peterson
et al., 1994; Peterson, 1996b). Cyanobacteria were
equally rare on the bed and in the drift, whereas
they were more common in drought refuges and
also during the first week of flow. This is consistent
with the ability to colonize by gliding (Peterson,
1996a) and to resist both scour and desiccation
(Robson, 2000; Stanley et al., 2004). It also suggests a
pattern of passive, rather than active drift (Peterson,
1996b) and, hence, Cyanobacteria appear to be more
dependent on recolonization from dry biofilm than
drift.
Both chlorophytes and diatoms were common in
winter drift, and chlorophytes also drifted more
frequently than their benthic densities suggested.
Filaments rather than spores dominated, and were
probably broken from parent plants by the current.
However, these broken filaments cannot reattach to
surfaces and so cannot colonize stones. Therefore, it
seems likely that most initial regrowth of filamentous
chlorophytes is from dry biofilm. In contrast, diatoms
were shown both to regrow from biofilm and to
colonize from the drift. The process of recolonization
used by diatoms may depend on the specific taxa
present or on characteristics of past or present flow
regimes.
Drift sampling showed that the composition of
algal drift differed from benthic composition in most
streams. This indicates that only a subset of the algal
assemblage is drifting under normal (non-spate)
conditions and, as there was no consistent difference
in drift rates between diatom-dominated and fila-
mentous-algae dominated streams, this subset of taxa
is a mixture of algal divisions. Therefore, this
suggests that, although there is little specificity
among these algae for particular drought refuges,
Pathways for algal recolonization 2399
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd, Freshwater Biology, 53, 2385–2401
there are differences in colonization pathways among
taxa. Different diatom species have been shown to
differ in their patterns of active drift (Peterson,
1996b).
A model for pathways of recolonization in intermittent
streams
Together, results from the four components of this
study provide a model for the recolonization path-
ways for epilithic algae in seasonally-flowing streams
including modifying effects of stream regulation
(Fig. 9). Our results clearly show that stream algae
depend on refuges, but there appears to be consider-
able flexibility with respect to their use during
droughts. Of the three refuges studied, leaf packs
appear to contribute least to post-flow algal growth.
The relative importance of in situ regrowth and
recolonization via drift varies between streams. Recol-
onization does occur immediately from refuges (dry
biofilm or pools) after flows resume, due to ‘wash
out’, but this is mainly diatoms, not other algal
divisions. Arrival of these drifting algae onto stones
makes a substantial contribution to algal regrowth in
some streams. For example, drifting algae, rather than
regrowth from the biofilm, was the dominant recol-
onization source in Bomjinna, Mt William and Gre-
villea Creeks (Fig. 9). Why some diatom-dominated
streams show a stronger dependence on drift than
others (e.g. Dairy Creek) is less clear. However, Dairy
Creek is a regulated stream with a particularly erratic
flow regime (B.J. Robson & T.G. Matthews, pers. obs.),
so the increased dependence on dry biofilm that
appears to result from stream regulation (discussed
above) may lead to the apparent mixture of diatom
regrowth observed from both drift colonists and dry
biofilm.
By identifying the pathways of recolonization, we
have constructed a model that can be tested
elsewhere (Fig. 9). Although we sampled a large
number of streams, one weakness of our sampling
was that we did not sample over more than one
dry-to-flow transition. Knowledge of recolonization
processes is fundamental to understanding how
algae will adjust to changes in flow regime, includ-
ing those arising from climate change and increased
abstraction. For example, in regions where the
climate is predicted to become drier and where
rainfall is predicted to occur in shorter, more intense
bursts, an increased reliance on regrowth from dry
biofilm may be anticipated because such changes
will be similar to those that have already occurred in
regulated streams. Further challenges remain,
including testing of the general applicability of this
model and to determine whether these changes in
colonization processes have subsequent impacts on
assemblages of stream animals.
Acknowledgments
This research was funded by the Glenelg-Hopkins
Catchment Management Authority and conducted
with appropriate permits from Parks Victoria. We
would like to thank Amanda Cowell and Brett
Downey for assistance in the field and Gerry Quinn
for helpful comments on the manuscript.
Dry UnregulatedFlowing
RegulatedFlowing
leaves
dry biofilmon stonetops
pools
driftingdiatoms
driftingdiatoms
Driftingdiatoms
regrowth
regrowth regrowth
in situregrowth insitu
Fig. 9 Model for the pathways of recolonization for epilithic algae in seasonally-flowing streams. When streams are dry, three
refuge types are sufficient to ensure the survival of all algal taxa subsequently growing on stones. Once streams resume flow,
both drifting algae and regrowth from dry biofilm assist recolonization, the relative magnitude of each depends on the
composition of the algal assemblage in each stream. Pools and dry biofilm are the main source of drifting colonists. Flow
regulation increases the dependence on regrowth from dry biofilm. Thicker arrows and bold type indicate stronger pathways.
2400 B. J. Robson et al.
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd, Freshwater Biology, 53, 2385–2401
References
Benenati P.L., Shannon J.P. & Blinn D.W. (1998) Desic-
cation and recolonization of phytobenthos in a regu-
lated desert river: Colorado River at Lees Ferry,
Arizona, USA. Regulated Rivers: Research and Manage-
ment, 14, 519–532.
Biggs B.J.F. (1996) Patterns in benthic algae of streams. In:
Algal Ecology: Freshwater Benthic Ecosystems (Eds R.J.
Stevenson & M.L. Bothwell), pp. 31–56. Academic
Press, San Diego, CA.
Blinn D.W., Shannon J.P., Stevens L.E. & Carder J.P.
(1995) Consequences of fluctuating discharge for lotic
communities. Journal of the North American Benthological
Society, 14, 233–248.
Bunn S.E., Davies P.M. & Winning M. (2003) Sources of
organic carbon supporting the food web of an arid
zone floodplain river. Freshwater Biology, 48, 619–635.
Clarke K.R. & Gorley R.N. (2001) PRIMER v5: User
Manual ⁄Tutorial. PRIMER-E, Plymouth.
Clarke K.R. & Warwick R.M. (1994) Change in Marine
Communities: An Approach to Statistical Analysis and
Interpretation. Plymouth Marine Laboratory, Plymouth.
Cowell A.L., Matthews T.G. & Lind P.R. (2006) Effect of
fire on benthic algal assemblage structure and recol-
onization in intermittent streams. Austral Ecology, 31,
696–707.
Culp J.M., Scrimgeour G.J. & Beers C.E. (1994) The effect
of sample duration on the quantification of stream
drift. Freshwater Biology, 31, 165–173.
Davis J.S. (1972) Survival records in the algae, and the
survival role of certain algal pigments, fat and muci-
laginous substances. The Biologist, 54, 52–93.
Dodds W.K., Hutson R.E., Eichem A.C., Evans M.A.,
Gudder D.A., Fritz K.M. & Gray L. (1996) The
relationship of floods, drying, flow and light to
primary production and producer biomass in a prairie
stream. Hydrobiologia, 333, 151–159.
Faulkner H. & Copp G.H. (2001) A model for accurate
drift estimations in streams. Freshwater Biology, 46, 723–
733.
Gasith A. & Resh V.H. (1999) Streams in Mediterranean
climate regions: abiotic influences and biotic responses
to predictable seasonal events. Annual Review of Ecology
and Systematics, 30, 51–81.
Hughes D.A. (2005) Hydrological issues associated with
the determination of environmental water require-
ments of ephemeral rivers. River Research and Applica-
tions, 21, 899–908.
Magalhaes M.F., Beja P., Canas C. & Collares-Pereira M.J.
(2002) Functional heterogeneity of dry-season fish
refugia across a Mediterranean catchment: the roles
of habitat and predation. Freshwater Biology, 47, 1919–
1934.
Matthaei C.D., Guggelberger C. & Huber H. (2003) Local
disturbance history affects patchiness of benthic river
algae. Freshwater Biology, 48, 1514–1526.
McCormick P.V. & Stevenson R.J. (1991) Mechanisms of
benthic algal succession in lotic environments. Ecology,
72, 1835–1848.
Mosisch T.D. (2001) Effects of desiccation on stream
epilithic algae. New Zealand Journal of Marine and
Freshwater Research, 35, 173–179.
Peterson C.G. (1987) Influences of flow regime on
development and desiccation response of lotic diatom
communities. Ecology, 68, 946–954.
Peterson C.G. (1996a) Response of benthic algal commu-
nities to natural physical disturbance. In: Algal Ecology:
Freshwater Benthic Ecosystems (Eds R.J. Stevenson &
M.L. Bothwell), pp. 375–403. Academic Press, San
Diego, CA.
Peterson C.G. (1996b) Mechanisms of lotic microalgal
colonization following space-clearing disturbances act-
ing at different spatial scales. Oikos, 77, 417–435.
Peterson C.G. & Boulton A.J. (1999) Stream perma-
nence influences microalgal food avilability to
grazing tadpoles in arid-zone springs. Oecologia,
118, 340–352.
Peterson C.G., Weibel A.C., Grimm N.B. & Fisher S.G.
(1994) Mechanisms of benthic algal recovery following
spates: comparison of simulated and natural events.
Oecologia, 98, 280–290.
Quinn G. & Keough M. (2002) Experimental Design and
Data Analysis for Biologists. Cambridge University
Press, Cambridge.
Robson B.J. (2000) Role of residual biofilm in the
recolonization of rock intermittent streams by benthic
algae. Marine and Freshwater Research, 51, 725–732.
Robson B.J. & Matthews T.G. (2004) Drought refuges
affect algal recolonization in intermittent streams. River
Research and Applications, 20, 753–763.
Ryder D.S. (2004) Response of epixylic biofilm metabo-
lism to water level variability in a regulated floodplain
river. Journal of the North American Benthological Society,
23, 214–223.
Stanley E.H., Fisher S.G. & Jones J.B. (2004) Effects of
water loss on primary production: a landscape-scale
model. Aquatic Sciences: Research Across Boundaries, 66,
130–138.
Stevenson R.J. (1983) Effects of current and conditions
simulating autogenically changing microhabitats on
benthic diatom immigration. Ecology, 64, 1514–1524.
(Manuscript accepted 13 June 2008)
Pathways for algal recolonization 2401
� 2008 The Authors, Journal compilation � 2008 Blackwell Publishing Ltd, Freshwater Biology, 53, 2385–2401