The aquatic plant communities of the Pilbara region of Western Australia: a region of arid zone wetland diversity

Introduction

Decision making in conservation is a knowledge-dependent process. Decisions associated with sustainable economic development, conservation estate design, land management, and management of threatened species and ecosystems are most effective when informed by adequate data describing spatial and temporal patterns in biodiversity, its ecological dependencies, and responses to threatening processes (Margules et al. 2002; Buxton et al. 2021; Hoffmann 2022).

Well-designed systematic and rigourous surveys of biodiversity provide internally consistent and taxonomically broad datasets for conservation planning at regional scales, that can then be contributed to larger data repositories at national (Belbin et al. 2021) and international (The Global Biodiversity Information Facility 2024) scales, expanding access to more comprehensive datasets on species diversity and distribution for conservation planning (Reichman et al. 2011). This is especially important in areas of competing land uses and where there are broadscale threats to biodiversity requiring informed management.

The Pilbara Region of north-western Western Australia constitutes one such area, with globally significant mineral resources and a recent history of extensive pastoralism (Booth et al. 2021). What little was known of the region’s biodiversity suggested a rich and endemic biota, but with a rapidly expanding resources sector it was also understood that knowledge of the region’s biodiversity was inadequate for regional conservation planning and environmental impact assessment (McKenzie et al. 2009). To inform impact assessment and landscape scale conservation planning, a comprehensive biological survey of the region was undertaken between 2003 and 2006 (McKenzie et al. 2009). This survey aimed to provide systematic survey data on the distribution of the terrestrial and aquatic biota across the region and use these data to identify gradients in community composition, the environmental factors related to these gradients, and to assess the adequacy of the existing reserve system (Gibson et al. 2015).

The Pilbara is an arid region with an abundance of aquatic habitats, many associated with discharge from numerous and extensive groundwater aquifers and with significant topographic variation. Wetlands in arid and semi-arid regions, with rainfall <500 mm per year, are frequently data deficient due to their remoteness from population centres, perceptions that they are not as threatened as wetlands in more populated temperate regions and because many such wetlands have water only temporarily. Australia is a continent dominated by semi-arid and arid landscapes, which support a wide diversity and very high number of wetlands, ranging from permanent springs and river pools, ephemeral rock holes and claypans to vast episodically filled salt lakes in palaeodrainages (Roshier and Rumbachs 2004; Duguid et al. 2005).

Wetlands are amongst the most threatened ecosystems globally (Dudgeon et al. 2006, Davidson and Finlayson 2018) partly due to their position in the landscape, which means they accumulate impacts from the broader catchments, and also because of their immense value for other human uses. Semi-arid and arid wetlands and their flora communities may be particularly sensitive to disturbance (Parra et al. 2021; Khelifa et al. 2022) due to climate-change effects in already dry regions (Scholes 2020), factors such as sedimentation from pastoral grazing in areas of of easily erodible catchment soils (Reid et al. 2017) and competition for limited and unreliable water (Jenkins et al. 2005).

Whilst many studies of the biotas of Australian arid zone wetlands have been undertaken at catchment to regional scales for a variety of biotic groups (e.g. Fensham et al. 2004; Kingsford et al. 2010; Casanova 2015; Timms 2022), regional scale studies of aquatic flora in Australia’s arid zones, which would assist in broad conservation planning and environmental impact assessment, are limited (Duguid et al. 2005; Hunter and Lechner 2018; Rossini et al. 2018) or of limited taxonomic scope (e.g. Costelloe et al. 2005).

This paper aims to document the diversity and major patterns in the aquatic flora of Pilbara wetlands and rivers. The current study complements previous analyses of the riparian plant communities and aquatic invertebrate fauna from the survey (Pinder et al. 2010; Lyons 2015). Importantly, the current study includes charophytes in addition to vascular plants; a group that has not been included in regional scale surveys in Western Australia previously.

Study area

The Pilbara region, as defined in this study represents an area of approximately 180,000 km², bounded to the south by the Ashburton River and to the east by the De Grey-Oakover River system. A small number of wetlands were also sampled in the Karlamilyi National Park to the east of the main study area (Fig. 1). The study area includes the entire Pilbara biogeographic region (Thackway and Cresswell 1994) and small areas of the adjoining Gascoyne and Little Sandy Desert regions.

Fig. 1.

Maps of the Pilbara study area showing the 98 wetland sites sampled. (a) The Pilbara biogregion and its four sub-regions (shaded grey with full names) (IBRA6.1): (1) Roebourne Plains (eastern and western); (2) Chichester; (3) Fortescue; and (4) Hamersley. Drainage basin boundaries are outlined in black and labelled with abbreviated names: ASH, Ashburton; OC, Onslow Coast; FOR, Fortescue; PHC, Port Hedland Coast; DGO, De Grey-Oakover; GSD, Great Sandy Desert. (b) Main rivers (blue lines) and two major wetlands (grey shaded) of the Pilbara region (taken from Lyons 2015). The sampled wetland sites are numbered as described by Lyons (2015).

The Pilbara region includes two bioclimatic zones: semi-desert tropical in western coastal areas and desert in the central and eastern area (Beard 1990). Rainfall in the region averages 290 mm, dominated by summer precipitation associated with tropical low-pressure systems and is spatially and inter-annually highly variable. Major cold fronts from the south produce winter rainfall concentrated along the coast and central ranges but rarely extending to the east of the region. At least one cyclone traverses the region in a normal summer and others pass close by along the coast. They supply half of the average annual rainfall.

The Pilbara region is characterised by extensive rocky landscapes including the dominant Hamersley and Chichester Ranges. Major aquifer systems are associated with these landscapes, discharging to form springs and supporting permanent pools in rivers, gorges, and creeks. The Western Coastal Plain and central and upper Fortescue River Valley lowlands support an array of claypans, clay flats and turbid creeks.

Mining and pastoral industries dominate the region’s land use. Given the regions extensive aquifer systems, mining typically requires pit dewatering and associated discharge, with associated risks to springs and river pools from dewatering, and disturbance to hydroperiods where discharges occur within rivers and creeks (Eckersley et al. 2024). Grazing pressure and soil disturbance from trampling is a known risk to riparian zone condition (Lyons 2015). Water quality is also impacted from cattle that increase nutrient levels, degrade riparian vegetation and increase turbidity. Wetlands are particularly vulnerable in dry seasons where cattle congregate at a more restricted set of natural water sources.

Materials and methods

Results

Site classification

Six site groups were recognised from the cluster analysis with a single site forming an artifactual ‘group’. The classification revealed a primary division between mostly riverine (river pools, gorge creeks, springs, ephemeral creeks), and non-riverine wetlands (claypans, marshes, and turbid creek pools) (Fig. 2). The frequency of occurrence of the a priori wetland types for each of the floristic groups is in Table 2.

Fig. 2.

Dendrogram showing classification of 87 survey sites (UPGMA) at the six-group level. Dotted line denotes an artifactual site.

Table 2.Frequency of the a priori wetland types for each of the floristic groups.

Floristic group	Clay flat	Claypan	Ephemeral creek	Gorge creek	Clear river pool	Rock pool	Salt marsh	Spring	Turbid creek pool
1. Rocky springs and ephemeral creeks			2					2
2. Gorges, springs and river pools with rocky substrates				2	16			8
3. Permanent and near permanent river pools			1		18			4
4. Claypans, clay flats and turbid creek pools	2	9				1			2
5. Miscellaneous seasonal wetlands	1		1		1	1			1
6. Salt marshes, claypans and linear claypans		2			1		3		3

Group 1. Rocky springs and ephemeral creeks

Small shallow pools in rocky shallow springs and a single ephemeral creek, characterised by emergent taxa such as Eleocharis geniculata, Lobelia arnhemiaca, Typha domingensis, Eleocharis geniculata, Schoenus dissacanthus, and Fimbristylis sieberiana, with the near absence of submerged aquatics. Average species richness was 10 species per site with an average permanence score of 3.2.

Group 2. Gorges, springs and river pools with rocky substrates

This group included many sites with submerged aquatics distinguishing the group from Group 1. Significant taxa from the SIMPER analysis included Schoenus subulatus, Typha domingensis, Najas marina, Potomogeton tepperi, Chara zeylanica, and Cyperus vaginatus. The group includes a discrete cluster of six springs and spring fed gorge creeks including Palm Spring at Millstream, Fortescue Falls, and Hamersley Gorge. Average species richness was seven taxa per site with an average permanence score of 3.7.

Group 3. Permanent and near permanent river pools

A large group of 23 river pools with Schoenus subulatus, Chara wightii, Najas marina, Nitella hyalina, and Nitella congesta as significant taxa from the SIMPER analysis. The low abundance of wetland fringing taxa including Typha domingensis and Cyperus vaginatus distinguishes the group from Group 3 above (average group dissimilarity 67.05). Average species richness was 11 species per site with an average permanence score of 3.2.

Group 4. Claypans, clay flats, and turbid creek pools

A group of ephemeral and seasonal wetlands with Eriachne benthamii s.l. and Marsilea spp. as distinguishing taxa. The group includes a variety of claypan types that typically fill seasonally (average permanence class 2.1). The sites were often turbid with low average species richness (2.6 taxa per site).

Group 5. Miscellaneous seasonal wetlands

A heterogeneous group of seasonal wetlands (average permanence score 2.1) that included pools in seasonal creeks, shallow river pools and non-turbid claypans/flats. Average site similarity for the group was low (23.66). Najas graminea s.l., Chara porteri, and Marsilea spp. were significant taxa defining the group. The group includes a distinctive group of non-turbid claypans in the east of the study area with almost complete beds of Marsilea with Chara porteri as the dominant aquatic. Average species richness was 7.9 taxa per site.

Group 6. Salt marshes, claypans and linear claypans

A very heterogeneous group of wetlands with low average group similarity (16.53). Significant species were Nitella heterophylla and Nitella micklei. Average species richness was 5.6 taxa per site. The group includes two sites sampling the flooded margins of Fortescue Marsh.

Community composition and categorical variables

Wetland types

Anosim analysis of both site and sample ordinations for wetland type (Figs 3 and 4) showed significant differences between several wetland types (samples: global R = 0.503, P < 0.001; sites: global R = 0.55, P < 0.001). For both seasonal and site data, clear river pools, springs, and to a lesser extent ephemeral creeks were separated from claypans, rock pools, turbid pools, and salt marshes. Claypans were the most compositionally heterogeneous group albeit species poor. For both sites and samples, pairwise anosim results between wetland types provided little support (low r-values) for compositional differences between turbid pools and claypans, river pools and springs, and ephemeral creeks and springs.

Fig. 3.

Two-dimensional non-metric multidimensional scaling ordination (2D nMDS) 87 sites overlain with a priori wetland types.

Fig. 4.

Two-dimensional non-metric multidimensional scaling ordination (2D nMDS) of 179 samples (seasonal) overlain with a priori wetland types.

Hydrology

Anosim analysis and interpretation of nMDS ordination plots revealed differences in community composition with assigned hydroperiod/permanence classes for both site and sample resemblance matrices (Figs 5 and 6) (site: r = 0.368; sample: r = 0.384, P < 0.001). Pairwise tests for both site and sample data showed very similar results, with differences largely driven by significant differences (P < 0.001) between near permanent/permanent classes, compared to ephemeral/seasonal classes. Anosim results revealed no significant compositional difference between ephemeral and seasonal classes (r-values < 0.05, P > 0.2), although near-permanent and permanent classes showed significant yet weak compositional separation (r-values < 0.17, P < 0.001). Since hydroperiod class was strongly aligned with wetland types (e.g. claypans were all ephemeral or seasonal), and river pools and springs were largely near-permanent or permanent, hydroperiod class and composition was examined within the major groups identified from the anosim analysis of wetland types (see above). For riverine sites and samples (excluding turbid creek pools), compositional differences were relatively weak but significant between seasonal and permanent classes (site: pairwise r = 0.407; sample: pairwise r = 0.588, P < 0.001). Combined analysis of claypans, turbid pools and salt marshes showed no compositional relationship with hydroperiod classes.

Fig. 5.

Two-dimensional non-metric multidimensional scaling ordination (2D nMDS) of 87 sites overlain with permanence class (1, ephemeral; 2, seasonal episodic; 3, near permanent; 4, permanent).

Fig. 6.

Two-dimensional non-metric multidimensional scaling ordination (2D nMDS) of 179 samples (seasonal) overlain with permanence class (1, ephemeral; 2, seasonal episodic; 3, near permanent; 4, permanent).

Seasonality

Within wetland types, anosim tests of autumn vs spring samples revealed little compositional separation with season. Of the wetland types, only river pools showed significantly different richness in spring (7.0 ± 2.8) vs autumn (5.4 ± 2.6) (P<0.05).

Comparison of the resemblance matrices (using the RELATE routine in Primer) for autumn vs spring samples (for sites visited in both seasons) showed a significant rank correlation (Spearman) between the seasons (rho = 0.595, P < 0.001) suggesting that sites are relatively compositionally stable between seasons.

Discussion

Our survey was one of the few to have systematically documented aquatic plants across an entire region, especially when combined with a study of the region’s riparian flora (Lyons 2015). We recorded 111 aquatic and amphibious taxa in the current study with charophytes representing a third of species documented. The flora includes taxa with extensive Australian tropical and arid-zone distributions (e.g. Vallisneria annua, Najas tenuifolia, Chara zeylanica) and taxa with very broad Australian distributions (e.g. Myriophyllum verrucosum, Najas marina, Nitella hyalina). These distributions largely exclude the sandy deserts that occur to the north and east of the study area, in which suitable aquatic habitats are rare or absent. The Pilbara also represents a large yet disjunct part of the distributions of numerous taxa (e.g. Najas spp.). Other taxa recorded represent major range disjunctions from the Kimberley, and elsewhere in northern Australia (e.g. Pseudoraphis spinescens, Ipomoea aquatica, Eriocaulon cinereum). Myriophyllum decussatum represents one of the few taxa with a south-western semi-arid distribution, with collections made during the survey representing a major northerly range extension. The record of Sporobolus ramigerus (syn. Eragrostis australasica) at Mulga Downs Outcamp claypan (PSW040) is one of three known occurrences in the Pilbara bioregion. The record of Aponogeton queenslandicus from PSW036 (Munreemya Billabong) represents a new record for Western Australia with the closest populations occurring along the gulf coast of the Northern Territory and isolated records from south-east Queensland with its core distribution in wetlands along the coastal ranges of Queensland. Other vascular taxa are of taxonomic interest. Thyridea sp. northern saline floodplains (A. Markey & S. Dillon FM 10730) is a recently recognised putatively novel entity recorded from the margins of the Fortescue Marsh. The species flowers as the wetland dries following a submerged phase. An additional putatively novel taxon, Vallisneria sp. Weelarrana (M. N. Lyons & S. D. Lyons 3050) was collected opportunistically from Weelarrana Lake following a cyclonic rainfall event and is listed as a conservation priority species. Collections of Potamogeton × salicifolius (Potamogeton lucens × P. perfoliatus) from Gregory Gorge and Millstream represent the only occurrence outside Eurasia. With only one of the parent taxa present in Australia and neither in Western Australia the origin and alien status of these populations are unknown (Kaplan et al. 2019).

Within the Characeae several taxa collected during the survey are putatively new to science. Notably, Genus et sp. nov. (M. N. Lyons and D. A. Mickle 3080), recorded exclusively from Weelaranna Salt Lake (PSW043), has attributes significantly different from all previously described Characeae (Casanova 2013). This was confirmed by DNA analysis (K. Karol, pers. comm.). Formal description awaits the collection of additional material from this episodic wetland. Similarly, Chara sp. nov. (aff. globularis) (M. N. Lyons & D. A. Mickle 3218) and Nitella sp. nov. ‘verrucate’ (M. T. Casanova PBS7) are significantly different from other members of their respective genera.

Although all formal taxonomic descriptions have not been completed it also evident that the Pilbara includes a segregate charophyte flora. Collections from the current study have also provided material for the description of novel charophytes including Nitella mickleii currently endemic to the Pilbara (Casanova and Porter 2013; Casanova and Karol 2023a, 2023b). This highlights the value of systematic surveys in poorly sampled regions to make major contributions to biodiversity knowledge and provide valuable material for taxonomic studies, particularly for poorly collected groups, including the discovery of novel taxa. Notably only a single non-native aquatic plant (Ceratopteris thalictroides) was recorded from Pilbara wetlands adjacent to a major pastoral settlement. Elsewhere in north-west Australia, this taxon is considered as native (see Keighery 2013). The lack of weedy aquatics is unsurprising as many occurrences elsewhere in Australia are human-mediated, the product of accidental introductions from discarded ornamental material, an event that is unlikely in the sparsely populated Pilbara. However, the potential for the introduction of wetland associated (amphibious) weedy grasses and sedges remains real in the light of increasing movements of fodder (potentially contaminated) produced under irrigation (Sudmeyer et al. 2016; Galloway et al. 2022).

Comparable systematic surveys of wetland aquatic plants are limited in Australia, particularly at the regional scale. This is particularly true of arid zone wetlands. Earlier studies have included surveys of submerged macrophytes in salt lakes within the south-west of Western Australia, with Ruppia, Althenia and Lamprothamnium taxa systematically surveyed in relation to wetland salinity (Brock and Lane 1983). Vascular macrophytes were also included in an extensive wetland survey of the same region (Lyons et al. 2004). Wetland aquatic and amphibious plants were also included in the survey of the southern Carnarvon Basin of Western Australia, in a study that focused primarily on riparian vegetation (Gibson et al. 2000). In contrast to the Pilbara, the Carnarvon Basin is a relatively subdued landscape dominated by seasonal and ephemeral wetlands with few rocky uplands supporting permanent pools and springs, and with only six submerged aquatics recorded (Gibson et al. 2000), noting they did not record charophytes. In Australia, the most comparable wetland survey to the current study was undertaken in arid central Australia (c. southern half of the NT) in an area with a comparable diversity of wetland types to the Pilbara (Duguid et al. 2005).

Duguid et al. (2005) did not document the charophyte flora to species level in their study of 324 sites. They list 39 other taxa as being aquatic and semi-aquatic and provide details of other groups of taxa associated with wetlands. The classification of the aquatic/semi-aquatic status of taxa differed slightly between the studies. For example, we have included inundation tolerant shrubs and trees in the amphibious class, which does not directly equate to the semi-aquatic grouping of Duguid et al. (2005). In aligning our data with the scheme of Duguid et al. (2005) we recorded 38 aquatic and semi-aquatic taxa, suggesting the Pilbara is of comparable diversity to the flora of the southern Northern Territory. Notably the southern Northern Territory is significantly richer for Nymphaea (five taxa), a genus with tropical affinities.

Wetland type and associated hydrological regime were the primary drivers of compositional patterns of aquatic plant communities in the Pilbara. Permanent and seasonal sites (gorge pools and many spring feed river pools) showed the highest species richness, and such sites are likely to act as refugia as many occur as isolated pools along river channels during the dry season. Many river pools are likely to be groundwater dependent and maintain water during the dry season as isolated waterbodies along the length of major river channels. The importance of water permanence as refugia has been highlighted elsewhere in arid Australia (Brim Box et al. 2008). These sites may become increasingly important given predictions for significant temperature increases for the Pilbara region under climate change with rainfall likely to remain largely unchanged to 2090 (Sudmeyer 2016).

Claypans and other shallow ephemeral and intermittent wetlands were generally species poor and compositionally distinct from other Pilbara wetlands. At regional scales they are a compositionally heterogeneous group with occurrences concentrated on the western coastal plain and Fortescue valley. More detailed sampling within the upper Fortescue Valley revealed finer scale patterning in claypan composition with large interconnected systems capturing a greater diversity of plant and invertebrate communities (Pinder et al. 2017). Unique plant assemblages were also recorded from a cluster of claypans at the upper reaches of the Fortescue valley at Jigalong Creek flood plain and downstream of the Fortescue Marsh (Pinder et al. 2017). Except for the reservation of Fortescue Marsh itself, floodplain wetlands such as claypans are very poorly reserved across the Pilbara, with most reserves focussed on upland areas (Gibson et al. 2015). This and subsequent studies (Pinder et al. 2017) suggest broad compositional differences between claypans of the coastal plain, central Fortescue valley and eastern wetlands associated with the Jigalong Creek flood plain. This provides a broad framework to address reservation inadequacies for claypan wetlands across the region.

For all wetland types the compositional patterns in aquatic plants reported here are concordant with broad patterns in other aquatic biota sampled across the Pilbara by Lyons (2015) and Pinder et al. (2010). At broad scales this is unsurprising given the major drivers of community composition (water permanence and seasonality, substrate and wetland morphology) are common across groups. For aquatic plants floristic composition in part reflects the major functional plant types at an individual wetland which is driven by a similar suite of factors determining wetland type (Moor et al. 2017). Importantly, this concordance across groups provides a framework for conservation planning with synergies available to maximise the capture of wetland and river biodiversity across multiple biotic groups.

Mining and pastoral activities impact on wetland and river conservation values across the Pilbara, with direct impacts from alterations to hydrology associated with mine dewatering and its discharge and disturbance associated with the grazing of herbivores. These land uses are entrenched across the region and limit the capacity to establish new reserves managed primarily for conservation. An emerging threat to Pilbara rivers and springs is the introduction of redclaw crayfish, Cherax quadricarinatus (von Martens, 1868) (Pinder et al. 2019). A native of Queensland and the Northern Territory, it was first detected in Harding Dam, a drinking water reservoir 23 km south-east of Roebourne. Subsequently it has become widespread across the region. These populations are likely the result of illegal translocations to provide for recreational fishing. The introduction of other decapods has been shown to have significant impacts on macrophyte biomass with subsequent cascading effects on invertebrate communities (Harper et al. 2002; Lodge et al. 2005). This represents a major threat to the rich aquatic plant communities of Pilbara river pools.