Fluvial biotopes influence macroinvertebrate biodiversity in South-East Asian tropical streams

Fluvial biotopes inﬂuence macroinvertebrate biodiversity in
South-East Asian tropical streams
Kꢁꢂꢃ Bꢁkꢃꢄ,1,† Mꢅꢆꢇꢁꢃꢈ A. Cꢇꢁꢉꢊꢅꢆk,1 Rꢁfꢇꢅꢁꢇ Kꢁꢇꢁꢄ,2
Zꢋꢇꢄꢁꢇ Hꢁꢌꢅ Sꢍꢈꢁꢅꢎꢁꢏ,2 ꢁꢏꢉ Rꢋꢉzꢁꢐ A. Wꢁꢇꢁꢑ2
1Department of Geography, King’s College London, London WC2R 2LS UK
2Institute for Biodiversity and Environmental Research, Universiti Brunei Darussalam,
Bandar Seri Begawan, Brunei Darussalam BE1410 Brunei
Citation: Baꢒer, K., M. A. Chadwicꢒ, R. Kahar, Z. H. Sulaiman, and R. A. Wahab. 2016. Fluvial biotopes inﬂuence macroinvertebrate biodiversity in South-East Asian tropical streams. Ecosphere 7(12):e01479. 10.1002/ecs2.1479
Abstract. Given the widespread degradation oꢓ aquatic systems caused by land-use changes associated with palm oil production in South-East Asia, it is imperative to identiꢓy and study the remaining undisturbed rivers and streams. Stream macroinvertebrates are reliable indicators oꢓ environmental health.
Linꢒing the community structure oꢓ these organisms to natural hydraulic and geomorphic conditions
(categoriꢔed as biotopes) is vital ꢓor the conservation and restoration oꢓ streams. This study characteriꢔes the eﬀects oꢓ biotopes on macroinvertebrate community structure in three streams within Ulu Temburong National Parꢒ in northern Borneo. Biotopes within these streams were categoriꢔed as either bedrocꢒ
(waterꢓalls and cascades) or mixed substrate (riﬄes and pools). In total, 119 taxa were collected ꢓrom all sampled biotopes, but not all taxa were collected ꢓrom each stream. Biotopes were statistically distinct in terms oꢓ taxonomic richness, but not mean individual density or average community biomass. There were diﬀerences in community structure between waterꢓalls, cascades, pools, and riﬄes. The survey suggests that pool and riﬄe biotopes were more vulnerable to scouring ﬂows and had similar community structure, while waterꢓalls and cascades liꢒely experienced lower sheer stress during ﬂoods and had similar macroinvertebrate communities. This study has ꢓound that classiﬁcation and mapping oꢓ macroinvertebrates with biotope theory in pristine, tropical streams is a useꢓul ꢓrameworꢒ ꢓor simpliꢓying the many linꢒages between ecology, geomorphology, and hydrology. These natural paꢕerns increase our understanding oꢓ tropical streams and can be used to assess the impacts oꢓ ꢓorest degradation.
Key words: biodiversity; biotopes; macroinvertebrates; palm oil plantations; tropical streams.
Received 10 June 2016; accepted 21 June 2016. Corresponding Editor: D. P. C. Peters.
Copyright: © 2016 Baꢒer et al. This is an open access article under the terms oꢓ the Creative Commons Aꢕribution
License, which permits use, distribution and reproduction in any medium, provided the original worꢒ is properly cited.
† E-mail: ꢒate.baꢒer@ꢒcl.ac.uꢒ
IntroductIon
Borneo, an island that is home to one oꢓ the oldest rain ꢓorests in the world. A recent study suggests
Human activities are increasing the urgency that approximately 80% oꢓ Malaysian Borneo rain
ꢓor investigating the basic tropical stream ecol- ꢓorests have been severely impacted by deꢓorogy (Dolný et al. 2011, Dudgeon 2015, Lewis estation and conversion to palm oil plantations et al. 2015, Ramíreꢔ et al. 2015). This is particu- (Bryan et al. 2013). This land-use change and the larly apparent in South-East Asia, where rising subsequent loss oꢓ aquatic biodiversity limit the world demand ꢓor palm oil is driving deꢓoresta- ability to study the properties oꢓ natural systems. tion. In spite oꢓ this phenomenon, large areas oꢓ
One approach to studying the tropical stream the South-East Asia tropics are not being actively ecology is the classiﬁcation and mapping oꢓ studied (Ramíreꢔ et al. 2015). This includes invertebrates associated with geomorphic and ꢀvꢀ
1
December 2016ꢀvꢀVolume 7(12)ꢀvꢀArticle e01479 BAKER ET AL.
Table 1.ꢖFlow type descriptions used to identiꢓy the physical biotopes present in the ﬁeld (based on Newson and Newson 2000 and Parasiewicꢔ 2007).
Associated biotope Flow type Description
Waterꢓall Free ꢓall Water ꢓalls vertically and without obstruction ꢓrom a distinct
ꢓeature, generally more than 1 m high and oꢗen across the ꢓull channel width
Cascade Chute Fast ﬂow with a smooth boundary and turbulent ﬂow over boulders or bedrocꢒ. Flow is in contact with the substrate and exhibits upstream convergence and downstream divergence
Pool Scarcely perceptible ﬂow
Surꢓace ꢓoam appears to be stationary and reﬂections are not distorted. A sticꢒ ﬂoated on the water’s surꢓace will remain still
Riﬄe Unbroꢒen standing waves
Undular standing waves in which the crest ꢓaces upstream without “breaꢒing” hydraulic conditions. This technique has pro- process by which they were ꢓormed and position vided a robust evaluation oꢓ the importance in the channel (Bisson et al. 1982). Many tropioꢓ hydraulics, sediment dynamics, and geo- cal headwaters experience ﬂash ﬂoods and are morphology on the temperate stream habitats categoriꢔed as relatively unpredictable systems
(Bunn and Arthington 2002, McManamay et al. (Boulton et al. 2008). This range in conditions can
2014, Villeneuve et al. 2015) and is operation- result in biotopes, especially those with mixed ally reꢓerred to as “biotope theory” (Dahl 1908, substrates (i.e., pools and riﬄes), exhibiting a Townsend and Hildrew 1994, Newson and continuum oꢓ conditions, which may result in
Newson 2000). At its core, biotope theory is based two distinct environments. For example, during on observable environmental conditions (Joweꢕ low ﬂows, tropical streams are complex systems
1993, Wadeson 1995, Padmore 1998, Newson and exhibiting a mix oꢓ ﬂow biotopes (i.e., pools, riꢓ-
Newson 2000, Cliﬀord et al. 2006). As such, bio- ﬂes, and cascades) and ꢓunctional habitats (i.e., topes reꢓer to the abiotic environment; in streams wood debris, leaꢓ liꢕer, cobbles, and gravel; sensu and rivers, these are typically observed as sur- Harper et al. 1995, Harvey et al. 2008); however,
ꢓace ﬂow ꢓeatures (i.e., ﬂow biotopes), such as riꢓ- during a ﬂood event, these streams become
ﬂes, pools, and waterꢓalls. These biotopes reﬂect homogeneous as water rises to ꢓorm a uniꢓorm the combinations oꢓ substrate type, depth, and ﬂood biotope. For naturally disturbed systems, velocity, which ultimately inﬂuence macroinver- ﬁxed habitat ꢓeatures create reꢓuge space ꢓor tebrate biodiversity (Newson and Newson 2000, macroinvertebrates during high ﬂows (Bond and Parasiewicꢔ 2007; Table 1).
Downes 2000), suggesting that some biotopes
Few studies conducted in the tropics have and habitat ꢓeatures may have a disproportionate strictly employed biotopes as a sampling ꢓrame- importance on the maintenance oꢓ biodiversity worꢒ (Furtado 1969, St Quentin 1973, Dudgeon (Buendia et al. 2014).
1994, Yule 1996, Ramíreꢔ et al. 1998, Principe
It is vitally important to increase our under-
2008). However, other studies have modiﬁed the standing oꢓ tropical stream ecosystems in order biotope theory to assess the longitudinal assem- to assess and mitigate the impacts oꢓ ꢓorest blage structure oꢓ tropical rivers (Bishop 1973, modiﬁcation and destruction on biodiversity
Rundle et al. 1993, Greathouse et al. 2005). Not (Dolný et al. 2011). Streams ﬂowing through Ulu surprisingly, there is still much to learn about the Temburong National Parꢒ in northern Borneo are mechanisms by which the structure, composi- still surrounded by unlogged primary rain ꢓortions, and paꢕerns oꢓ biotopes can aﬀect the mac- est, with no roads (Sheldon 2011). This provides roinvertebrate biodiversity in the tropics (Bisson a unique opportunity to study the importance et al. 1982, Ramíreꢔ and Pringle 1998, Cheshire oꢓ biotopes in preserving the macroinvertebrate et al. 2005, Md Rawi et al. 2014). biodiversity. This study aimed to evaluate the The conﬁguration and hydraulic properties oꢓ macroinvertebrate biodiversity and community biotopes are highly variable and depend on the structure among three study streams in ꢓour
ꢀvꢀ
2
December 2016ꢀvꢀVolume 7(12)ꢀvꢀArticle e01479

BAKER ET AL.
Fig. 1.ꢖBrunei is situated in the north oꢓ Borneo. The country is split into two contiguous regions, with Ulu
Temburong National parꢒ located within the Temburong District. Kuala Belalong Field Study Centre (KBFSC) and the study reaches including Lower Apan, Esu, and Threelan (highlighted by asterisꢒs) are all within the national parꢒ. types oꢓ biotopes: pools, riﬄes, cascades, and climate paꢕern, daily weather in the Ulu waterꢓalls. This study speciﬁcally evaluated the Temburong National Parꢒ is very erratic. Most importance oꢓ biotopes, rather than streams or rain originates as convection cells; as the cells rise reaches, ꢓor the operational scale oꢓ biodiversity. over Buꢒit Belalong and Buꢒit Pagon, they condense, producing heavy rainꢓall (Cranbrooꢒ and Methods
Edwards 1994). Dyꢒes (1997) has argued that no month can be considered dry as every month oꢓ the year receives an average oꢓ over 200 mm oꢓ
Study sites
This project was conducted in Ulu Temburong rainꢓall.
National Parꢒ in the Temburong District oꢓ
Three streams situated near the Kuala Belalong
Brunei, northern Borneo (Fig. 1). The national Field Study Centre (KBFSC) were the ꢓocus oꢓ parꢒ has sharp topography; the elevation oꢓ this study: Sungai Lower Apan, Sungai Esu, and Kuala Belalong is 30 m. a.s.l., but rises to moun- Sungai Apan Threelan (Fig. 1). All three streams tain peaꢒs oꢓ 1850 m. a.s.l. at Buꢒit Pagon and are tributaries oꢓ Sungai Belalong or Sungai
913 m. a.s.l. at Buꢒit Belalong (Dyꢒes 1994). The Temburong and were chosen because they each area is composed oꢓ deep V-shaped valleys with contain a mixture oꢓ biotopes. Further, these no ﬂoodplains, and many waterꢓalls occur along streams are uninﬂuenced by anthropogenic ꢓacthe tributaries that drain the mountains. The tors, and their natural water quality is high geology is characteriꢔed by sedimentary rocꢒs (Sheldon 2011), which is important because the with some sandstone pebbles that have been variation in water quality impacts the biodivertransported ꢓrom the headwaters in the south- sity paꢕerns (Everaert et al. 2014). Lower Apan east. Brunei has a tropical climate, which is has the longest reach (90 m), exceeding those oꢓ weaꢒly inﬂuenced by the South-East Asia mon- Esu (70 m) and Apan Threelan (75 m). For each soon season (Dyꢒes 1996). Despite the annual stream, the survey locations started just beꢓore
ꢀvꢀ
3
December 2016ꢀvꢀVolume 7(12)ꢀvꢀArticle e01479

BAKER ET AL. the ﬁrst waterꢓall upstream ꢓrom each conﬂuence depth. Stream velocity was measured using an with the main rivers (Belalong or Temburong). electromagnetic ﬂowmeter (Valeport model 801;
Esu and Apan Threelan had waterꢓalls higher Valeport Ltd., Totnes, UK). Benthic substrates
(approximately 6 m high) than those oꢓ Lower were assessed visually and categoriꢔed accord-
Apan (approximately 3 m high). Sampling ing to the percentage gravel, cobble, boulder, and locations began 360 m upstream ꢓrom the bedrocꢒ. The presence or absence oꢓ ꢓunctional conﬂuence oꢓ Apan Threelan and Temburong, habitats was recorded including wood debris
157 m upstream ꢓrom the conﬂuence oꢓ Esu and (large and small), leaꢓ liꢕer, and moss; trailing
Belalong, and 60 m upstream ꢓrom the conﬂu- roots in all biotopes were also recorded. ence oꢓ Lower Apan and Temburong (Fig. 1).
Sampling was conducted during April 2013.
Benthic macroinvertebrates were sampled in each biotope using a Surber sample (0.10 m2;
250-μm mesh). Decapods are not eﬀectively sampled by Surber sampling (Jacobsen et al. 2008)
Field methods
Biotopes (i.e., pools, riﬄes, cascades, and and thereꢓore were not included in this study. waterꢓalls) were mapped in each oꢓ the study Because oꢓ low macroinvertebrate densities, reaches by observing river surꢓace ꢓeatures at three samples were composited ꢓor each biotope. baseﬂows (Newson and Newson 2000,
Parasiewicꢔ 2007). For the Lower Apan reach, 14 Laboratory methods biotopes were sampled: ﬁve pools, two riﬄes,
Owing to the requirements oꢓ specimen export
ﬁve cascades, and two waterꢓalls. For the Esu permits, macroinvertebrate samples were proreach, 10 biotopes were sampled: ﬁve pools, one cessed under (10×) magniﬁcation at KBFSC and riﬄe, two cascades, and two waterꢓalls. For the preserved in 70% ethanol. Once exported to the Threelan reach, 11 biotopes were sampled: six UK, macroinvertebrates were identiﬁed to the pools, two riﬄes, one cascade, and two water- lowest practical taxonomic level and enumer-
ꢓalls. Across the entire study 16 pools, ﬁve riﬄes, ated; the total body lengths were measured to the eight cascades, and six waterꢓalls were sampled. nearest 0.5 mm. The macroinvertebrate diversity
Features oꢓ each biotope habitat were mea- oꢓ Borneo is still mostly undescribed; thereꢓore, sured. Large habitat ꢓeatures can ꢓorm biotopes, the identiﬁcations were made using the ꢓew ꢒeys such as boulders and tree trunꢒs that dam the available, including Dudgeon (1999) and Yule water ﬂow (Fig. 2). Conversely, habitat ꢓeatures, and Yong (2004) as well as open source identiﬁsuch as leaꢓ liꢕer, can occur within biotopes. cation methods. Most specimens were identiﬁed
Physical conditions oꢓ the biotopes were mea- to the genus level or morphotyped to a similar sured with surveying tapes and meter sticꢒs and level. However, some taxa, such as Coleoptera included weꢕed and banꢒꢓull width and channel and Diptera specimens, could only be identiﬁed to the ꢓamily level (Yule 2004; J. Manꢓred, personal communication).
Taxa-speciﬁc ash-ꢓree dry mass (AFDM) was calculated using length–mass regressions (Benꢒe et al. 1999, Sabo et al. 2002, McNeely et al. 2007).
When no taxon-speciﬁc equations were available, estimates were made using the equations
ꢓrom taxa with similar body shapes (Ramíreꢔ and Pringle 1998). Where only dry mass (DM) estimates were available, the values were converted to AFDM ꢓollowing Waters (1977).
Data analysis
Macroinvertebrate biodiversity, richness, den-
Fig. 2.ꢖExtensive debris dam at waterꢓall on Sungai sity, and biomass (AFDM) were quantiﬁed ꢓor all
Esu. A man is highlighted in white circle to indicate oꢓ the biotopes in each oꢓ the tributaries. scale.
Comparisons among tributaries and biotopes
ꢀvꢀ
4
December 2016ꢀvꢀVolume 7(12)ꢀvꢀArticle e01479 BAKER ET AL. were carried out via a two-way ANOVA ꢓollowed R (R Core Team 2013). The “envﬁt” ꢓunction uses by Tuꢒey’s post hoc tests. Richness and density the mixed environmental data including both conmet the required statistical assumptions (i.e., nor- tinuous variables and categorical data (Oꢒsanen mally distributed, homoscedastic residuals), but 2016). Only the statistically signiﬁcant environbiomass was square-root-transꢓormed in order to mental variables (i.e., P 0.05) were ﬁꢕed and are minimiꢔe the deviations ꢓrom normality and independent oꢓ the NMDS ordination. homoscedasticity. Similarity percentage analysis
(SIMPER) was used to identiꢓy the taxa that con- results tributed most to the average dissimilarity among biotopes. Analysis oꢓ similarities (ANOSIM; Site description
Clarꢒe 1993) was used to test ꢓor the diﬀerences in
Many biotopes in Lower Apan were unconabundance and composition oꢓ macroinverte- strained laterally, transitioning directly ꢓrom the brates among the biotopes. The global R statistic, stream to the rain ꢓorest, whereas Esu and Apan which ranges ꢓrom −1 to +1, measures the distinc- Threelan were constrained by riparian bedrocꢒ, tiveness oꢓ the grouping according to ANOSIM. resulting in narrower banꢒꢓull widths. Thus, the Values close to 1 indicate high similarity among Lower Apan had more trailing roots and terrestrial groups, 0 indicates that there is no relationship in vegetation at the margins oꢓ the stream compared composition among the groups, and −1 indicates with the other study reaches. All oꢓ the study that the samples are distinct to each group. reaches exhibited evidence oꢓ landslides, and large
Abundance data were used ꢓor both SIMPER and wood debris was oꢗen ꢓound to be lodged between
ANOSIM, and both oꢓ these tests use the Bray– waterꢓalls, sometimes creating dams. Many oꢓ
Curtis index, a popular dissimilarity index ꢓor these dams were quite large; ꢓor example, Fig. 2 ecological data (Borcard et al. 2012).
Macroinvertebrate assemblage structures were shows a large debris dam at a waterꢓall on Esu.
Esu had the highest baseﬂow discharge examined among biotopes using a hierarchical (0.92m3/s)comparedwithLowerApan(0.62m3/s) cluster analysis carried out using Bray–Curtis and Apan Threelan (0.18 m3/s; Table 2). However, index values (Thomas et al. 2013). Bray–Curtis dis- banꢒꢓull width (F2,29 = 1.84, P = 0.18) and weꢕed similarity matrices were calculated, summariꢔing width (F2,29 = 1.23, P = 0.30) did not diﬀer among the compositional dissimilarity oꢓ sites based on tributaries. Average depths diﬀered signiﬁcantly the density oꢓ taxa at each site. Non-metric multi- among tributaries (F2,29 = 15.79, P 0.001) with dimensional scaling (NMDS) analysis was used to Apan Threelan having the shallowest biotopes test the robustness oꢓ groups deﬁned by the clus- (0.16 m), ꢓollowed by Lower Apan (0.25 m) ter analysis. NMDS is a ﬂexible statistical tool with and then Esu (0.37 m). Average velocities were
ꢓew statistical assumptions. The stress value was higher along Lower Apan (0.39 m/s) than along
0.16, which indicates a good ordination (Thomas Esu (0.37 m/s) and Apan Threelan (0.20 m/s; et al. 2013). Environmental data were ﬁꢕed to the F2,29 = 4.66, P 0.05). ordination using the “envﬁt” ꢓunction oꢓ the vegan
Banꢒꢓull widths diﬀered among biotopes pacꢒage in the statistical computing environment (F3,29 = 3.56, P 0.05), with waterꢓalls (7.07 m)
Table 2.ꢖAverage physical conditions including depth, weꢕed and banꢒꢓull width, velocity, and discharge oꢓ the three study streams (pooled across all biotopes) and oꢓ the ꢓour biotopes (pooled across all study reaches).
Sites Average depth (m) Weꢕed width (m) Banꢒꢓull width (m) Velocity (m/s)
Discharge (m3/s)
Lower Apan 0.26 5.58 8.79 0.39 0.62
Esu 0.37 4.88 7.65 0.37 0.92
Threelan 0.16 3.76 6.78 0.20 0.18
Pool 0.47 5.47 7.15 −0.12 –
Riﬄe 0.14 3.42 11.89 0.48 –
Cascade 0.06 5.02 7.23 0.60 –
Waterꢓall 0.07 3.90 7.07 1.01 –
ꢀvꢀ
5
December 2016ꢀvꢀVolume 7(12)ꢀvꢀArticle e01479

BAKER ET AL. habitats compared with the other biotopes with
88% oꢓ pools having leaꢓ liꢕer (Table 3). Cascades and waterꢓalls had the lowest percentage presence oꢓ ꢓunctional habitats with moss occurring in the highest percentage (Table 3).
Biodiversity of the study reaches
In total, 119 taxa were collected in this study.
Aꢗer pooling all the biotopes in each study reach, total richness was lowest at Lower Apan (71 taxa),
ꢓollowed by Esu (77 taxa) and then Apan Threelan
(81 taxa; F2,29 = 9.02, P 0.001; Fig. 4; Appendix
S1). Average biomass was also lowest at Lower
Apan (65 mg/m2), ꢓollowed by Esu (176 mg/m2) and then Apan Threelan (65 mg/m2; F2,29 = 9.46,
P 0.001; Fig. 4). There was no diﬀerence in the macroinvertebrate density among the tributaries
(F2,29 = 2.59, P = 0.07). A Tuꢒey post hoc test showed that richness and biomass at Lower Apan were signiﬁcantly lower than at Apan Threelan and Esu. ANOSIM showed an overall diﬀerence in the macroinvertebrate community structure
Fig. 3.ꢖPercentage substrate (gravel, cobbles, boulders, and bedrocꢒ) oꢓ the three study streams (pooled across all biotopes) and oꢓ the ꢓour biotopes (pooled across all study reaches). having the lowest average values and riﬄes hav- among the three tributaries (global R = 0.31; ing the highest (11.89 m). Weꢕed widths also P = 0.03). These diﬀerences were illustrated by the diﬀered among biotopes (F3,29 = 33.95, P 0.05), SIMPER analysis, which revealed that the averwith the lowest values occurring at riﬄes and age similarity between taxa was highest ꢓor Apan waterꢓalls ( 4 m) and the highest values at cas- Threelan (51%), ꢓollowed by Esu (47%) and then cades and pools ( 5 m). Biotope depths diﬀered Lower Apan (41%; Table 4).
(F3,29 = 55.14, P 0.001), with average values
Pooling together all benthic macroinvertebrates lowest ꢓor the waterꢓalls and cascades ( 0.10 m) revealed that Diptera (38%) was the most abundant and highest ꢓor the pools ( 0.40 m). There was order, with the highest number oꢓ individuals sama diﬀerence in velocity among the biotopes pledꢓromthethreestreams. Otherdominantorders
(F3,29 = 80.91, P 0.001), with the lowest average included Coleoptera (21%), Ephemeroptera (20%), velocity in the pools (−0.12 m/s) and the highest Trichoptera (9%), and Plecoptera (5%; Appendix in the waterꢓalls (1.01 m/s). Pools and riﬄes con- S1). In addition to these biodiversity measuretained a mix oꢓ gravel, cobbles, and boulders, ments, there were some ﬁrst recordings oꢓ aquatic while cascades and waterꢓalls were dominated insects ꢓrom Borneo including Compsoneuriella by bedrocꢒ ( 80% coverage; Fig. 3). Pools had sp. (Ephemeroptera: Baetidae), (M. Sartori and the highest percentage presence oꢓ ꢓunctional J. C. Gaꢕolliat, personal communication) and new
Table 3.ꢖPercentage presence oꢓ ꢓunctional habitats oꢓ the three study streams (pooled across all biotopes) and oꢓ the ꢓour biotopes (pooled across all study reaches).
Sites Large wood debris Small wood debris Leaꢓ liꢕer Moss Trailing roots
Lower Apan 25 38 63 44 44
Esu 44 56 50 38 6
Threelan 31 50 63 31 13
Pool 56 81 88 25 38
Riﬄe 19 31 31 19 13
Cascade 625 38 31 6
19 Waterꢓall 619 38 6
ꢀvꢀ
6
December 2016ꢀvꢀVolume 7(12)ꢀvꢀArticle e01479

BAKER ET AL.
Threelan (F3,29 = 3.97, P 0.05). However, no diꢓ-
ꢓerence in density (F3,29 = 0.50, P = 0.69) or biomass (F3,29 = 2.54, P = 0.08) was observed among the other biotopes. For the overall community structure, ANOSIM showed a strong diﬀerence among biotopes (global R = 0.71; P = 0.01). This result was supported by the SIMPER analysis, which showed that the average similarity between macroinvertebrates was highest in the pools (51%), ꢓollowed by riﬄes (43%), waterꢓalls
(44%), and cascades (19%; Table 5).
Community ordination analysis using individual taxon densities showed community structure among the biotopes (Fig. 5; stress = 0.16;
Clarꢒe and Warwicꢒ 2001). Ordination axis
1 liꢒely represented a gradient oꢓ both substrate and velocity, in which higher velocities and increased bedrocꢒ substrate were associated with waterꢓalls and cascades. In addition to gradients in velocity and substrate, axis 2 strongly reﬂected the paꢕerns in taxa richness.
Speciﬁcally, sites AWF1, AWF2, and AC3 (each with less than six taxa) all clustered at the top oꢓ the plot. According to the analysis perꢓormed with envﬁt, the environmental ꢓactors that were most strongly correlated with biological variables were velocity, gravel, cobbles, and bedrocꢒ (P 0.001), along with depth, small wood debris, and moss (P 0.05). As expected, pools and riﬄes were associated with increased depths and areas oꢓ deposition, with a strong association with small wood debris, gravel, and cobbles. Bedrocꢒ and high ﬂow velocities were associated with waterꢓalls and cascades.
Fig. 4.ꢖRichness, density, and biomass (ash-ꢓree dry mass; AFDM) oꢓ macroinvertebrates ꢓor Sungai
Apan Threelan, Sungai Esu, and Sungai Lower Apan as well as ꢓor each biotope (cascade, waterꢓall, riﬄe, and pool). Error bars represent standard deviations
(Lower Apan, n = 14; Esu, n = 10; Apan Threelan, n = 11; waterꢓall, n = 6; cascade, n = 8; riﬄe, n = 5; and pool, n = 16). Taxa richness (F2,29 = 9.02, P 0.001) and biomass (F2,29 = 9.46, P 0.001) diﬀered among tributaries. Taxa richness diﬀered among biotopes
(F3,29 = 3.97, P 0.05).
The hierarchical cluster analysis supports the results oꢓ the ordination analysis. There was a Bray–Curtis dissimilarity oꢓ 0.8 between the rocꢒ biotopes (cascades and waterꢓalls) and mixed substrate (riﬄes and pools; Fig. 6). However, two waterꢓalls (TWF2 and EWF2) were separated
ꢓrom the rocꢒ biotopes, which may be explained by the higher richness oꢓ these waterꢓalls (25 individuals) compared with other rocꢒ biotopes. recordings ꢓrom Brunei with Pelthydrus elongatu lus (Coleoptera: Hydrophilidae; Schonmann 1995),