Dry season habitat use of fishes in an Australian tropical river

Article metrics

Abstract

The modification of river flow regimes poses a significant threat to the world’s freshwater ecosystems. Northern Australia’s freshwater resources, particularly dry season river flows, are being increasingly modified to support human development, potentially threatening aquatic ecosystems and biodiversity, including fish. More information is urgently needed on the ecology of fishes in this region, including their habitat requirements, to support water policy and management to ensure future sustainable development. This study used electrofishing and habitat survey methods to quantify the dry season habitat use of 20 common freshwater fish taxa in the Daly River in Australia’s wet-dry tropics. Of twenty measured habitat variables, water depth and velocity were the two most important factors discriminating fish habitat use for the majority of taxa. Four distinct fish habitat guilds were identified, largely classified according to depth, velocity and structural complexity. Ontogenetic shifts in habitat use were also observed in three species. This study highlights the need to maintain dry season river flows that support a diversity of riverine mesohabitats for freshwater fishes. In particular, shallow fast-flowing areas provided critical nursery and refuge habitats for some species, but are vulnerable to water level reductions due to water extraction. By highlighting the importance of a diversity of habitats for fishes, this study assists water managers in future decision making on the ecological risks of water extractions from tropical rivers, and especially the need to maintain dry season low flows to protect the habitats of native fish.

Introduction

Freshwater ecosystems are threatened around the world by flow modification1. Flow modification can be attributed to a number of human-related activities including urbanisation, industrialisation, mining and agriculture arising from anthropogenic structures such as dams, reservoirs, levee’s and channelization1,2,3. Flow is a major driver of physical habitat in rivers and streams; affecting the biotic composition, distribution and diversity of all aquatic life4. Alterations to river flow can therefore affect the availability of habitat quality and quantity, and in turn can negatively impact the assemblage composition and abundance of aquatic species4,5.

Stream fishes are closely associated with physical habitat attributes at a range of spatial scales6,7. Variations in water depth and velocity have been hypothesised by various studies as the key factors influencing habitat use of fishes, with physical attributes such as substrate composition, submerged vegetation, wood and root masses also being important8,9,10. The availability and types of physical habitat attributes can vary within and between river systems, and can influence species composition, distribution and abundances, due to differences in ecological requirements for food, spawning sites and/or refuge11,12,13. Previous studies have also classified groups of species into habitat use guilds to summarise the relationships among taxa, and to explore the influence of habitat on assemblage dynamics, not just individual species (e.g. Grossman & Freeman14; Leonard & Orth15; Aadland16; Lamouroux & Souchon17). These studies often incorporate water depth, velocity and physical habitat features (e.g. substrate composition) in assigning taxa to certain habitat guilds. For some species, habitat use can also change ontogenetically (e.g. Leonard & Orth15; Aadland16); thus life stage may be considered as an important factor in assessment of species-habitat associations.

Hydrological and water quality variability can influence the availability and type of aquatic habitats present for fish to use. In tropical regions, seasonality and interannual variability of the flow regime may be the most important temporal influence on fish distribution and abundance18,19,20. In the wet-dry tropics of Northern Australia, this effect of flow is particularly evident at the end of the dry season when water levels are at their lowest due to reduced rainfall, and factors such as competition and predation can strongly influence fish community structure21,22. In these systems, aquatic organisms have developed different modes of adaptation (e.g. life history, behaviour, morphology) in response to the seasonal timing and predictability of flow events4,23. Several studies indicate that the combination of persistent, predictable flows and habitat complexity are important in promoting and maintaining species specialisation and diversification in tropical rivers10,24,25.

Northern Australian rivers are recognised both nationally and internationally for their high ecological and cultural value26,27, and support higher aquatic species diversity than temperate Australian rivers28,29,30. However, recent efforts to develop northern Australia (e.g. expanded agricultural and mining interests) may result in increased demand for water resources, potentially reducing the diversity and abundance of aquatic habitats and associated biota, including fish18,31,32. Deleterious impacts on subsistence fisheries are of particular concern to the numerous Indigenous traditional owner groups of the region33. Knowledge of fish habitat requirements are common considerations in management and assist in amelioration actions such as environmental flow rules34,35,36. However, our knowledge of many aspects of the ecology of northern Australian fishes, including habitat requirements, is currently sparse; and therefore our ability to inform water policy and management about fish requirements is limited.

This study aims to explore the dry season habitat use of freshwater fishes in the Daly River, Northern Territory; a perennial river with a largely unmodified flow regime in Australia’s wet-dry tropics. The study uses a data-rich, multi-year dataset to quantify fish-habitat associations during the dry season. It was hypothesised that (i) fish taxa would be strongly associated with flow-related characteristics such as water depth and current velocity, (ii) there are distinct fish habitat guilds associated with specific mesohabitat features (e.g. riffles, pools) and (iii) ontogenetic habitat changes would occur for some species. The implications of our findings are discussed with regards to future water management in tropical rivers.

Methods

Study area and focal reach

The Daly River catchment (mean annual streamflow 8653 GL37) is located in the wet-dry tropical climate region of northern Australia (Fig. 1), covers ~53,000 km2, and has a largely intact savannah ecosystem throughout the catchment38. The catchment’s population is 10,000 people, of whom 28% are indigenous39. The Katherine (largest tributary) and the Daly Rivers are considered to be in relatively good ecological condition38,40. All rivers in the catchment are unregulated (no dams or weirs), but some groundwater extraction occurs for agriculture and water supply and this is likely to increase in the future18. The rivers have a characteristic wet-dry tropical flow regime, with predictable and large wet season flows with high interannual variability41. Mean annual rainfall averages 1070 mm42 across the catchment and is highly seasonal, with the majority falling during the wet season months (November-April), and negligible rainfall during the dry season (May–October). Flow in the Daly River and some tributaries is perennial, with significant groundwater inputs coming from two underlying aquifers, resulting in a lengthy period of continuous and stable dry season base flows (May-November). The Daly and Katherine rivers are sand-bed rivers that contain dispersed bedrock outcrops and gravel bars; river banks are typically steep, rising 15–20 m above the river bed42,43. Riparian vegetation in the catchment is largely intact and natural, consisting of Eucalyptus woodlands, Melaleuca forests and closed monsoon rainforests44.

Figure 1
figure1

Location of fish sampling locations in the Daly River catchment. Fish sampling sites are indicated with site identification numbers (1–6). Refer to Table 1 for site details. The inset shows the location of the study area in northern Australia.

This study occurred along a 120 km long reach of river from near Katherine township downstream to Oolloo crossing (Fig. 1). Six sites that were accessible by boat were selected; these sites represent the available habitat types for fish within the reach (Table 1).

Table 1 Details of six sampling sites in the Daly River catchment.

Fish and habitat sampling

Fish were surveyed biannually in both the early and late dry season, over a 10-year period from 2006–2015 (n = 89 sampling events in total). A detailed description of fish sampling methods can be found in Stewart-Koster et al.19 and Chan et al.18, but is briefly summarised here. Within each sampling site (500–1000 m reach length, 5–100 m wetted width), fish were sampled at discrete multiple locations (shots or replicates) using a boat or backpack electrofishing (pulsed DC), depending on water depth. Electrofisher settings were adjusted to maximise efficiency of collecting fish with minimum power. At least 15 electrofishing shots of five minute elapsed duration were undertaken for each sampling event, and shots were stratified to ensure each available habitat type within each site was sampled at least once. Prior examination of sampling efficiency has revealed that 15 shots yields an accurate estimate of species composition and assemblage structure within each site (Kennard et al. unpubl. data). Electrofishing shots were conducted in as homogenous area as possible and averaged 77 m (±48 SD) in length (range = 5–263 m). At the completion of each electrofishing shot, fish captured were counted and identified to species, measured (standard length (SL) in mm) and returned alive to the point of capture. The total number of fish from each species observed and reliably identified but not caught during each electrofishing shot was also recorded. Catch data were converted to catch per unit effort (CPUE—total number of individuals caught and observed per electrofishing shot). Some species captured during sampling were separated into juvenile or adult age classes (Table 2), where the minimum length used for juvenile determination for these species was <150 mm SL, except Lates calcarifer, where the minimum length was <300 mm SL.

Table 2 Fish taxa recorded during the 2006–2015 study period including age classes used in habitat analyses. Two taxa (Ambassis sp. and Neosilurus hyrtlii) were excluded since there was little explanatory power for these taxa (see Suppl. Table 1 in Supplementary Material).

A total of 20 variables describing fish habitat characteristics was measured within the area of each discrete fishing location (sampling shot) by a second field team. A full description of the methods employed in describing habitat structure is available in Kennard et al.45. Briefly, five replicate measures within each electrofishing shot were measured for depth (initially using a staff, then a Garmin 150 Fish Finder) and mean water velocity (Swoffer™ 2000, and Hach™ FH950 flow meter, measured as 20% total depth). The total percentage contribution of each mesohabitat type (i.e. riffles, runs, pools) and substrate composition (modified Wentworth scheme- see Pusey et al.46) was visually estimated within each shot (summed to 100%). The proportional contributions of a range of submerged structures (including wood, macrophytes, root masses etc. Table 3) were also visually estimated for each shot (contributions of these variables were not required to add up to 100%). The fish collected within each electrofishing shot were assigned the habitat characteristics of that shot and these data were used to examine the local habitat use of each taxon.

Table 3 Habitat variables estimated or measured in each electrofishing shot location over the 2006–2015 study period used in analyses (adapted from Pusey et al. 2004).

Data analysis

The data were filtered to remove 26 rare taxa (i.e. those taxa present in <5% of samples and/or with <50 individuals collected in total). Neoarius graeffei, N. bernei and N. midgleyi were combined to genus level (i.e. Neoarius spp.), due to either low counts or unreliable identification to species. This resulted in a total of 22 taxa suitable for analysis with six taxa separated into juvenile and adult age classes (Table 2). For the entire study period, 54,739 taxa-habitat data points were available for analysis.

Fish habitat variables were assessed for their potential to influence the abundance of each fish taxon across electrofishing shots using boosted regression trees (BRTs). Modelling can improve ecological understanding and allow evaluation of generalizability47, and thus has advantages over traditional approaches to quantifying habitat use (e.g. by comparing habitat availability to frequency of use to infer habitat selection or preference)48,49. Tree-based modelling approaches such as boosted regression trees (BRTs) are widely used for developing predictive species-environment relationships50 and have been applied to quantify fine-scale fish habitat use (e.g. Sharma et al.51, Greenwood52). BRTs are a decision tree based method which uses a boosting technique to combine large numbers of relatively simple tree models adaptively to optimize predictive performance53. Some advantages of BRTs include their ability to model complex and nonlinear relationships and to accommodate missing data and large numbers of potential predictor variables by ignoring non-informative variables53. BRTs also perform better than other regression techniques under conditions of high multicollinearity, especially when a large number of variables are analysed.

We assumed a Poisson distribution for the response variable in all models since our data is composed of individual fish counts in each shot. The best combination of parameters required by BRTs (learning rate, tree complexity and bag fraction) was identified for every model using cross-validation52. Cross-validation was automatically repeated for learning rates from 0.001 to 0.05 (steps of 0.002), tree complexities of 1–3 and bag fractions of 0.5 and 0.75, which span the range of likely optimal values52. The combinations that generated the lowest mean cross-validation deviances, calculated from at least 1000 trees, were used for the final models. Following the derivation of full models with all variables, models were investigated to establish whether irrelevant predictors could be removed (procedure as detailed in Elith et al.52).

The percentage relative contributions of each habitat variable, including the effect of season (early or late dry season) were calculated from the BRTs for each taxon. The relative contribution of a variable describes the proportion of variation in the data explained by that variable relative to all other variables in the model, scaled to 10053. For each taxon, partial-dependence functions were fitted for the top two most influential variables to visualise relationships between fish abundances and each variable. The fitted function for a given variable incorporates the average effect of all other variables (Elith et al. 2008), with positive fitted function values suggesting an increase in abundance, while negative values suggest a decrease. Following initial BRT analyses, upon examination of percent deviance explained for each model, two taxa (Ambassis spp. and Neosilurus hyrtlii) with low calculated percentage deviance (<20%) were removed from further consideration (Table 4). Hence BRTs for a total of 20 taxa with five taxa separated into adult and juvenile categories are reported, noting that each model has different percent deviance and hence explanatory power (Table 2).

Table 4 Number of trees and the percentage deviance explained as calculated for fish taxa from the Boosted Regression Tree models.

Hierarchical cluster analysis with complete linkage distance was used to group fish taxa abundance and habitat variables into habitat guilds. The number of classes (guilds) in the clustering dendrogram was determined visually in a way that allowed for a biological interpretation. Pairwise comparisons were conducted between each taxon and all habitat variables using Pearson’s correlation coefficient. All analyses were performed in R version 3.3.254 using the ‘gbm’ package version 2.1.355 and the ‘dismo’ package version 1.1–456 for the BRT models plus customised code written and described by Elith et al.52, and the ‘hclust’ function of ‘vegan’ package version 2.4-457 for the cluster analysis.

Ethical Statement

All methods were carried out in accordance with ethical guidelines and regulations and approved by Charles Darwin University’s Committees for Animal Ethics and Human Research Ethics. Fish sampling was also conducted under NT Fisheries S17 Fisheries Act Permit.

Results

Twelve habitat variables were identified as being the top two most important contributors to the BRT models for the 20 taxa (Table 5). Seasonal variation in habitat use was minimal, with this factor having a low relative influence in BRT models for all taxa (Table 5). Water depth was identified as the top contributor for 16 taxa (Fig. 2, Table 5). Taxa that were most abundant in shallow habitats (0–100 cm) included L. unicolor, juvenile H. fuliginosus, M. mogurnda, juvenile O. selheimi, C. stramineus, and Glossogobius sp. In contrast, both C. stercusmuscarum and P. ordensis were more likely to occur in deeper habitats (>200 cm). The remaining taxa, A. percoides, adult S. butleri, adult H. fuliginosus, juvenile O. lineolata, Neoarius spp., G. aprion, juvenile and adult L. calcarifer, M. australis, T. chatareus, N. erebi, N. ater, and S. kreffti, were most abundant in moderate to deep habitats (100–200 cm).

Table 5 Percentage relative contribution of each variable for all taxa as calculated from the Boosted Regression Tree models.
Figure 2
figure2

(an) Partial dependence plots showing relationships between taxa abundance and depth as the first or second most important variable in the BRT models, plots grouped into families. Note different y axis ranges. Refer to Table 4 for the percentage relative contribution of each variable for all taxa. Taxa were grouped into family categories for comparison between species and ontogeny (juvenile/adult). Dotted line highlights zero on the fitted function axis. Positive fitted function values above the dotted line shows high abundance and low values below this line show the opposite. Juvenile and adult H. fuliginosus (b) are separated from other Terapontids for clarity.

Velocity was the second most important contributor to the BRT models, occurring as one of the top two most important contributor for 12 taxa overall, and was the top habitat contributor for three taxa (Fig. 3, Table 5). Taxa that were most abundant in low velocity habitats (0–30 cm/s) included adult O. lineolata C. stramineus, G. aprion, M. australis, M. mogurnda, T. chatareus and adult S. butleri. Both juvenile H. fuliginosus and P. ordensis were most abundant in fast flowing habitats (>90 cm/s). The remaining taxa, juvenile S. butleri, N. ater and Neoarius spp., were most abundant in habitats with moderate to high velocities (30–90 cm/s).

Figure 3
figure3

(ai) Partial dependence plots showing relationships between taxa abundance and velocity as the first or second most important variable in the BRT models, plots grouped into families. Note different y axis ranges. Refer to Table 4 for the percentage relative contribution of each variable for all taxa. Taxa were grouped into family categories for comparison between species and ontogeny (juvenile/adult). Dotted line highlights zero on the fitted function axis. Positive fitted function values above the dotted line shows high abundance and low values below this line show the opposite.

Submerged wood, root mass, leaf litter, rock, sand, fine gravel, coarse gravel, bedrock, filamentous algae and overhanging vegetation were the next most important contributors to the BRT models (Fig. 4, Table 5). The abundance of both juvenile and adult L. calcarifer, adult O. selheimi, and N. erebi, were all higher in habitats with over 50% submerged wood. Similarly, the abundance of both juvenile and adult O. lineolata, as well as juvenile O. selheimi was higher in habitats with over 40% root mass. Likewise, the abundance of L. unicolor and juvenile S. butleri was also higher in habitats with over 15% leaf litter and 40% rock, respectively. The abundance of L. triramus was higher in habitats with over 80% sand and fine gravel, respectively. Similarly, the abundance of Glossogobius sp. and adult H. fuliginosus was higher in habitats with 50–75% and over 40% coarse gravel, respectively. In contrast, the abundance of A. percoides, C. stercusmuscarum, S. kreffti and adult O. selheimi was lower in habitats with over 15% leaf litter, 60% bedrock, 30% aquatic vegetation and 10% overhanging vegetation, respectively.

Figure 4
figure4

(am) Partial dependence plots showing relationships between taxa abundance and other variables which were identified as the first or second most important variable in the BRT models, plots grouped into families. Note different y axis ranges. Refer to Table 4 for the percentage relative contribution of each variable for all taxa. Taxa were grouped into family categories for comparison between species and ontogeny (juvenile/adult). Dotted line highlights zero on the fitted function axis. Positive fitted function values above the dotted line shows high abundance and low values below this line show the opposite.

Cluster analysis of the habitat data revealed four distinct habitat-use guilds (Fig. 5a). We interpreted these groups qualitatively, using information on taxon body size (Table 2) and the taxon correlations with each habitat variable (Fig. 5b). The four habitat use guilds comprised: Guild I: fishes occupying deep pools containing root masses and undercut banks, Guild II: large-bodied fishes occupying deep pools containing wood, Guild III: a mixed habitat use guild of small-bodied fishes, and Guild IV: small-bodied fishes occupying shallow riffles with high water velocities and coarse substrates.

Figure 5
figure5

Hierarchical cluster analysis with pairwise comparisons between taxa and habitat variables as used in the BRT models, displaying (a) four distinct habitat-use guilds and (b) the strength of the positive (blue) and negative (red) associations with each habitat variable, using the Pearson’s correlation coefficient. Habitat Guilds (a): Green = fishes occupying deep pools containing root masses and undercut banks (Guild I), Light blue = large-bodied fishes occupying deep pools containing wood (Guild II), Red = a mixed habitat use guild of small-bodied fishes (Guild III), Dark blue = small-bodied fishes occupying shallow riffles with high water velocities and coarse substrates (Guild IV); A = Adult, J = Juvenile.

Taxa associated with Guild I had a range of body sizes and included both juveniles and adults of O. lineolata, M. australis, G. aprion, adult S. butleri, N. erebi, N. ater, Neoarius spp. and T. chatareus (Fig. 5b). Taxa within Guild II were large-bodied fishes including both juvenile and adult L. calcarifer, adult H. fuliginosus, adult O. selheimi, S. kreffti and P. ordensis. Guild III consisted of six small-bodied taxa with a range of habitat associations. This included L. triramus which was associated with shallow habitats and higher densities of fine gravel and sand, M. mogurnda which was associated with higher densities of coarse gravel, both C. stercusmuscarum and juvenile O. selheimi which were associated with higher bedrock cover, undercut banks and root masses; C. stramineus was associated with increasing densities of aquatic vegetation and filamentous algae, and A. percoides was associated with shallow habitats with increasing bedrock cover. Guild IV consisted of small-bodied fishes occupying shallow riffles with high water velocities and coarse gravel and/or rock substrates (Fig. 5b). This included juvenile H. fuliginosus, juvenile S. butleri, Glossogobius sp. and L. unicolor.

Discussion

This study used a multi-year and data-rich dataset to quantify the dry season habitat use of freshwater fishes in an Australian wet-dry tropical river. Although many taxa displayed distinct associations with the measured habitat variables, depth and velocity were the key habitat variables for most of the taxa examined (16 and 12 taxa, respectively). Other important habitat variables included submerged wood, root masses, vegetation and sediment type. The analysis also identified four distinct fish habitat guilds according to body size and habitat associations. Whilst it is important to also consider taxon-specific differences in habitat use, the use of fish habitat guilds here may have advantages for habitat management in taxa rich systems such as tropical rivers, and also for forecasting habitat use in other systems or similar taxa (e.g. Leonard & Orth15).

Depth and velocity are often identified as important factors in determining fish assemblage structure in rivers (e.g. this study; Jackson et al.8; Kennard et al.58). These factors characterise the geomorphological complexity of rivers59 and hence the availability of habitats for fish; they are also likely to influence fish behavioural characteristics, such as swimming type, reproductive and feeding behaviours. Not surprisingly then, the depth and velocity requirements of fish are common considerations in river habitat and flow management actions, such as environmental flow determinations (e.g. Bunn & Arthington4; Poff et al.60) and habitat restoration activities61. In this study, 15 taxa were associated with moderately deep and low velocity habitats or pools. Pool habitats may have higher species diversity relative to other habitat types, because: (i) they are likely to be more effectively buffered against changes in habitat structure, water level and water quality58,62, (ii) they contain higher structural diversity (submerged wood, undercut banks, root masses, etc.) for feeding and shelter63, and (iii) they usually cover a greater surface area compared to other mesohabitats, such as riffles and runs9.

Four taxa had higher abundance in riffle habitats, including L. unicolor, Glossogobius spp. and juveniles only of S. butleri and H. fuliginosus (Guild IV). Juveniles of the terapontids H. fuliginosus and S. butleri demonstrated ontogenetic habitat shifts, with juveniles using riffle habitats before moving to deeper, slow-flowing pool habitats as adults (deep, structured pools habitat guild, see below). These findings confirm previous research and observations for these two species18,19,29,64. Riffle habitats may be providing a suitable refuge habitat for these four small-bodied taxa from piscivorous and avian predators, and/or discrete feeding or reduced competition with larger individuals (e.g. Rosenfeld & Boss65). Both Glossogobius spp. and L. unicolor were grouped into the riffle-dwelling habitat guild, due to their association with shallow depths and coarse gravel. This supports in part previous research by Rayner et al.24 that suggested that Glossogobius sp. 1 (identified in their study) was a fluvial specialist, preferring high velocity, shallow water habitats. Our analysis also provides further evidence that L. unicolor prefers shallow habitats and coarse gravel substrates, but with little velocity preference, thus this species may utilise broader habitat types with various velocities compared to the three riffle-dwelling species grouped in this guild.

In this study, the juveniles and adults of L. calcarifer, adult O. selheimi and adult H. fuliginosus, S. kreffti and P. ordensis, showed a positive association with deep, densely structured habitats composed of submerged wood (Guild II). A number of previous studies have shown that riverine fish assemblages are influenced by the physical structural complexity of river channels, including the presence of large submerged woody debris and instream vegetation66. Pettit et al.63 examined the distribution and movement of large woody debris and its importance as a fish habitat in the Daly River. Using a sub-set of the same fish data analysed in the current study, they demonstrated that fish species richness, diversity and abundance were not correlated with the proportion of wood present at a reach scale. However, juveniles and adults of L. calcarifer and adults of H. fuliginosus were strongly associated with wood cover63. Large submerged wood is considered to be an important contributor to habitat heterogeneity, through providing overhead cover which reduces predation risk, and also camouflage for predators to ambush their prey, among other functions66.

Fishes associated with pools that contain root masses and undercut banks (Guild I) included a range of taxa and life stages, including O. lineolata, G. aprion, N. ater, N. erebi, M. australis, T. chatareus and adult S. butleri. These taxa were all positively associated with a high density of both root masses and undercut banks, except Neoarius spp., which was more strongly associated with sandy substrates and moderate depths. The presence of both undercut banks and root masses serve as important cover for O. lineolata29, which was found to have a strong association with this habitat type in this study. Rainbowfish (Melanotaenia sp.), have previously been recorded as abundant in habitats composed of both root masses and undercut banks, as well as leaf litter46. Mouth almighty, G. aprion, a small-bodied predatory fish, may be associated with this habitat type in order to hide and ambush its prey22,67. Mouth almighty have previously been classified as a microhabitat generalist24, but has also been described as having a close association with root masses, leaf litter and aquatic vegetation22,67. Undercut banks have also been suggested as important daytime resting habitat for taxa within the Plotosidae and Terapontidae families68, which supports the findings for N. ater and adult S. butleri in this study. In contrast, the presence of overhanging vegetation may be more important for T. chatareus, since surface-dwelling and terrestrial invertebrate prey which occur in this vegetation type feature heavily in the diet of this species67. The presence of root masses, undercut banks and vegetation are suggested to function as an important refuge and/or foraging habitat for some species46,68. Indeed, many of the taxa in this habitat guild are regular prey of L. calcarifer69, which supports the hypothesis that the taxa in this guild are using these habitats as a predator refuge. Studies from other systems have also shown that species of small-bodied fishes (e.g. cyprinids) rarely co-exist in the same habitat type as larger predatory species, such as northern pike, Esox lucius70, and smallmouth bass, Micropterus dolomieu71. This further supports the predator refuge hypothesis.

The fourth habitat guild (Guild III) grouped six small-bodied taxa into a mixed habitat guild. Both M. mogurnda and C. stercusmuscarum, which had low associations with all habitat features in this study, have previously been shown to occur across varying substrate types67, while C. stercusmuscarum has been described as a habitat generalist24. Likewise Oxyeleotris sp. is described as a habitat generalist with a preference for abundant cover29,46, and juveniles of O. selheimi in this study were found to have a slight positive association with both root masses and undercut banks. Amniataba percoides was found to have a strong association with bedrock substrate in this study, but has been previously described as associated with a variety of substrates67. In comparison, L. triramus in this study was positively associated with both sandy and fine gravel substrates, this supports previous observations of the species being associated with sandy slackwater habitats29. Craterocephalus stramineus has previously been described as a riffle-dwelling species22, however we found that this species was associated with aquatic vegetation and filamentous algae.

The habitat use differences observed in this study may be due to differences among rivers and over time, or potentially this study’s coarser assessment of habitat use across a 5 minute shot length; i.e. fish may be captured in a rarer, smaller habitat type within the whole habitat sampling shot, and therefore specific associations may be overlooked. Indeed, this is an important limitation of our study and we recommend that future studies should focus on more detailed microhabitat assessments if finer resolution is important, as well as to confirm habitat use or preference. Furthermore, it is important to note that this study only examined day-time patterns of habitat use. Thus a range of complementary methods which may be used to determine spatial and temporal habitat use include snorkelling, underwater video surveys, and/or biotelemetry (e.g. Winter72; Lucas & Baras73).

Ontogenetic habitat shifts by fishes often occur in rivers (e.g. Scott & Nielsen74; Schiemer et al.75; King76), presumably because the size of an individual influences its ability to respond to velocity, seek shelter from predation and use other resources77. In this study, three species (H. fuliginosus, S. butleri, and O. selhemi) demonstrated distinct habitat changes from juvenile to adult life stages. As previously discussed, juvenile H. fuliginosus and S. butleri utilised riffle habitats and then shifted to deeper, slow flowing pools as adults. Similar ontogenetic shifts in habitat use have been observed in other species including brown trout, Salmo trutta, where young-of-the-year trout inhabited fast-flowing riffles, while larger, older trout occupied deeper and slower flowing habitats78. In this study, juvenile O. selheimi was associated with habitats dominated by aquatic vegetation, undercut banks and root masses (mixed habitat guild, see above), before switching to habitats with increased levels of submerged wood as adults (deep, structured pools habitat guild, see above). Conversely, our findings showed no change in habitat use between juvenile and adult life stages for L. calcarifer or O. lineolata. Whilst the number of species demonstrating ontogenetic habitat changes could be viewed as low (only three out of five species where life stages were categorised), we did not sample larval or early juvenile life stages in this study, and further research on the habitat use of these important life stages in wet-dry tropical rivers is needed.

The patterns of habitat use revealed in this study improve our understanding of the habitat requirements of a range of fish taxa during the dry season in northern Australian rivers. The study quantified fish-habitat use associations for a large number of common taxa, and identified four distinct habitat use guilds. The fish-habitat guild approach is useful in summarising and simplifying communication of fish habitat use requirements to managers and the broader community. After further exploration of the proposed fish-habitat guilds, the approach may also be useful for predicting habitat requirements in similar systems with little ecological knowledge.

While a range of habitat variables were important habitat descriptors for individual taxa and size classes, unsurprisingly water depth and velocity were key factors influencing dry season habitat use of fish in this system. Many studies have emphasised the maintenance of natural seasonal flow patterns for the persistence of freshwater fish assemblages and the diversity of riverine mesohabitats4,58,79,80. In perennial wet-dry river systems, the prolonged low water period during the dry season (~May-October), maintains critical riverine longitudinal connectivity and a range of shallow and deeper habitats that sustain aquatic ecosystems (this study31,63,81). Shallow habitats such as riffles are likely to be particularly vulnerable to natural or artificial reductions in water levels18,31, and hence changes in flow or dewatering of these habitats during the dry season could reduce connectivity, constrain fish movement and have particularly strong impacts on specific riffle dwellers such as L. unicolor, Glossogobius spp., and juvenile H. fuliginosus and S. butleri (this study18). These results, coupled with data on the extent of riffle habitat area at various water levels, would enable predictions to be made about the extent of riffle habitat at different water levels, and hence predict the impact on riffle-dwelling species. Incorporation of this type of information into water planning is critical for future decision making regarding the management of environmental water in these systems.

Data Availability

Data is available through Charles Darwin University’s Research webportal (https://researchers.cdu.edu.au/).

References

  1. 1.

    Vörösmarty, C. J. et al. Global threats to human water security and river biodiversity. Nature 467, 555, https://doi.org/10.1038/nature09440 (2010).

  2. 2.

    Meybeck, M. Global analysis of river systems: from Earth system controls to Anthropocene syndromes. Philosophical Transactions of the Royal Society B: Biological Sciences 358, 1935–1955, https://doi.org/10.1098/rstb.2003.1379 (2003).

  3. 3.

    Ward, J. V. & Stanford, J. A. Ecological connectivity in alluvial river ecosystems and its disruption by flow regulation. Regulated Rivers: Research & Management 11, 105–119, https://doi.org/10.1002/rrr.3450110109 (1995).

  4. 4.

    Bunn, S. E. & Arthington, A. Basic principles and ecological consequences of altered flow regimes for aquatic biodiversity. Environmental Management 30, 492–507 (2002).

  5. 5.

    Rosenfeld, J. Assessing the habitat requirements of stream fishes: an overview and evaluation of different approaches. Transactions of the American Fisheries Society 132, 953–968, https://doi.org/10.1577/T01-126 (2003).

  6. 6.

    Grossman, G. D., Ratajczak, R. E., Crawford, M. & Freeman, M. C. Assemblage organization in stream fishes: effects of environmental variation and interspecific interactions. Ecological Monographs 68, 395–420, https://doi.org/10.1890/0012-9615 (1998).

  7. 7.

    Webb, M., Reid, M. & Thoms, M. The influence of hydrology and physical habitat character on fish assemblages at different temporal scales. River Systems 19, 283–299 (2011).

  8. 8.

    Jackson, D. A., Peres-Neto, P. R. & Olden, J. D. What controls who is where in freshwater fish communities - the roles of biotic, abiotic, and spatial factors. Canadian Journal of Fisheries and Aquatic Sciences 58, 157–170, https://doi.org/10.1139/f00-239 (2001).

  9. 9.

    Jowett, I. G. A method for objectively identifying pool, run, and riffle habitats from physical measurements. New Zealand Journal of Marine and Freshwater Research 27, 241–248, https://doi.org/10.1080/00288330.1993.9516563 (1993).

  10. 10.

    Gorman, O. T. & Karr, J. R. Habitat structure and stream fish communities. Ecology 59, 507–515, https://doi.org/10.2307/1936581 (1978).

  11. 11.

    Benson, B. J. & Magnuson, J. J. Spatial heterogeneity of littoral fish assemblages in lakes: relation to species diversity and habitat structure. Canadian Journal of Fisheries and Aquatic Sciences 49, 1493–1500, https://doi.org/10.1139/f92-165 (1992).

  12. 12.

    Arrington, D. A. & Winemiller, K. O. Habitat affinity, the seasonal flood pulse, and community assembly in the littoral zone of a Neotropical floodplain river. Journal of the North American Benthological Society 25, 126–141, https://doi.org/10.1899/0887-3593 (2006).

  13. 13.

    Pusey, B. J., Arthington, A. H. & Read, M. G. Species richness and spatial variation in fish assemblage structure in two rivers of the Wet Tropics of northern Queensland, Australia. Environmental Biology of Fishes 42, 181–199, https://doi.org/10.1007/bf00001996 (1995).

  14. 14.

    Grossman, G. D. & Freeman, M. C. Microhabitat use in a stream fish assemblage. Journal of Zoology 212, 151–176, https://doi.org/10.1111/j.1469-7998.1987.tb05121.x (1987).

  15. 15.

    Leonard, P. M. & Orth, D. J. Use of habitat guilds of fishes to determine instream flow requirements. North American Journal of Fisheries Management 8, 399–409, https://doi.org/10.1577/1548-8675 (1988).

  16. 16.

    Aadland, L. P. Stream habitat types: their fish assemblages and relationship to flow. North American Journal of Fisheries Management 13, 790–806, https://doi.org/10.1577/1548-8675 (1993).

  17. 17.

    Lamouroux, N. & Souchon, Y. Simple predictions of instream habitat model outputs for fish habitat guilds in large streams. Freshwater Biology 47, 1531–1542, https://doi.org/10.1046/j.1365-2427.2002.00880.x (2002).

  18. 18.

    Chan, T. U. et al. Bayesian network models for environmental flow decision making in the Daly River, Northern Territory, Australia. River Research and Applications 28, 283–301, https://doi.org/10.1002/rra.1456 (2012).

  19. 19.

    Stewart-Koster, B. et al. Fish response to the temporal hierarchy of the natural flow regime in the Daly River, northern Australia. Journal of Fish Biology 79, 1525–1544 (2011).

  20. 20.

    Winemiller, K. O. Spatial and temporal variation in tropical fish trophic networks. Ecological Monographs 60, 331–367, https://doi.org/10.2307/1943061 (1990).

  21. 21.

    Magoulick, D. D. Spatial and temporal variation in fish assemblages of drying stream pools: The role of abiotic and biotic factors. Aquatic Ecology 34, 29–41, https://doi.org/10.1023/a:1009914619061 (2000).

  22. 22.

    Pusey, B. J., Kennard, M. J., Douglas, M. & Allsop, Q. Fish assemblage dynamics in an intermittent river of the northern Australian wet–dry tropics. Ecology of Freshwater Fish, https://doi.org/10.1111/eff.12325 (2016).

  23. 23.

    Lytle, D. A. & Poff, N. L. Adaptation to natural flow regimes. Trends in Ecology &. Evolution 19, 94–100, https://doi.org/10.1016/j.tree.2003.10.002 (2004).

  24. 24.

    Rayner, T. S., Pusey, B. J. & Pearson, R. G. Seasonal flooding, instream habitat structure and fish assemblages in the Mulgrave River, north-east Queensland: towards a new conceptual framework for understanding fish-habitat dynamics in small tropical rivers. Marine and Freshwater Research 59, 97–116, https://doi.org/10.1071/MF07129 (2008).

  25. 25.

    Willis, S. C., Winemiller, K. O. & Lopez-Fernandez, H. Habitat structural complexity and morphological diversity of fish assemblages in a Neotropical floodplain river. Oecologia 142, 284–295, https://doi.org/10.1007/s00442-004-1723-z (2005).

  26. 26.

    Pusey, B. J. Aquatic Biodiversity in Northern Australia: Patterns, Threats and Future. (Charles Darwin University Press, 2011).

  27. 27.

    Jackson, S. E. et al. “We Like to Listen to Stories about Fish”: Integrating Indigenous Ecological and Scientific Knowledge to Inform Environmental Flow Assessments. Ecology and Society 19, 43, https://doi.org/10.5751/es-05874-190143 (2014).

  28. 28.

    Pusey, B. & Kennard, M. Species richness and geographical variation in assemblage structure of the freshwater fish fauna of the wet tropics region of northern Queensland. Marine and Freshwater Research 47, 563–573, https://doi.org/10.1071/MF9960563 (1996).

  29. 29.

    Pusey, B. J. et al. Freshwater fishes of northern Australia. Zootaxa 4253, 1–104 (2017).

  30. 30.

    Pusey, B. J. et al. In Aquatic Biodiversity in Northern Australia: Patterns, Threats and Future (ed. Pusey, B. J.) 71–92 (Charles Darwin University Press, 2011).

  31. 31.

    King, A. J., Townsend, S. A., Douglas, M. M. & Kennard, M. Implications of water extraction on the low-flow hydrology and ecology of tropical savannah rivers: an appraisal for northern Australia. Freshwater Science 34, 741–758 (2015).

  32. 32.

    Stoeckl, N. et al. An integrated assessment of financial, hydrological, ecological and social impacts of ‘development’ on Indigenous and non-Indigenous people in northern Australia. Biological Conservation 159, 214–221, https://doi.org/10.1016/j.biocon.2012.12.007 (2013).

  33. 33.

    Jackson, S., Finn, M. & Featherston, P. Aquatic resource use by indigenous Australians in two tropical river catchments: the Fitzroy River and Daly River. Human Ecology 40, 893–908, https://doi.org/10.1007/s10745-012-9518-z (2012).

  34. 34.

    Dutta, V., Sharma, U. & Kumar, R. Assessment of river ecosystems and environmental flows: Role of flow regimes and physical habitat variables. Climate Change and Environmental Sustainability 5, 20–34 (2017).

  35. 35.

    Poff, N. L. et al. The natural flow regime: A paradigm for river conservation. BioScience 47/11, 769–784 (1997).

  36. 36.

    Dudgeon, D. et al. Freshwater biodiversity: importance, threats, status and conservation challenges. Biological Reviews 81, 163–182, https://doi.org/10.1017/S1464793105006950 (2006).

  37. 37.

    CSIRO. Water in the Timor Sea Drainage Division. A report to the Australian Government from the CSIRO Northern Australia Sustainable Yields Project. 273–361 (CSIRO, 2009).

  38. 38.

    Schult, J. & Townsend, S. In A Report to the Daly River Management Advisory Committee. Report Vol. Report No. 03/2012D (Department of Natural Resources, Environment, The Arts and Sport, Darwin, Northern Territory, 2012).

  39. 39.

    Larson, S. & Alexandridis, K. Socio-economic profiling of tropical rivers. (Land & Water Australia, 2009).

  40. 40.

    Ganf, G. G. & Rea, N. Potential for algal blooms in tropical rivers of the Northern Territory, Australia. Marine and Freshwater Research 58, 315–326 (2007).

  41. 41.

    Kennard, M. J. et al. Classification of natural flow regimes in Australia to support environmental flow management. Freshwater Biology 55, 171–193, https://doi.org/10.1111/j.1365-2427.2009.02307.x (2010).

  42. 42.

    Erskine, W., Saynor, M. & Lowry, J. Application of a new river classification scheme to Australia’s tropical rivers. Singapore Journal of Tropical Geography 38, 167–184, https://doi.org/10.1111/sjtg.12196 (2017).

  43. 43.

    Wasson, R. J. et al. Sediment sources and channel dynamics, Daly River, Northern Australia. Geomorphology 114, 161–174, https://doi.org/10.1016/j.geomorph.2009.06.022 (2010).

  44. 44.

    Lamontagne, S., Cook, P. G., O’Grady, A. & Eamus, D. Groundwater use by vegetation in a tropical savanna riparian zone (Daly River, Australia). Journal of Hydrology 310, 280–293, https://doi.org/10.1016/j.jhydrol.2005.01.009 (2005).

  45. 45.

    Kennard, M. J., Pusey, B. J., Perna, C., Burrows, D. & Douglas, M. Field manual–Including protocols for quantitative sampling of fish assemblages, habitat, water quality and sample preservation. (Charles Darwin University, Northern Territory, Available at, http://www.track.org.au/publications/registry/track973 [Accessed 25 November 2016], 2011).

  46. 46.

    Pusey, B. J., Kennard, M. & Arthington, A. H. Freshwater Fishes of Northeastern Australia. (CSIRO Publishing, 2004).

  47. 47.

    Houlahan, J. E., McKinney, S. T., Anderson, T. M. & McGill, B. J. The priority of prediction in ecological understanding. Oikos 126, 1–7, https://doi.org/10.1111/oik.03726 (2017).

  48. 48.

    Manly, B., McDonald, L., Thomas, D. L., McDonald, T. L. & Erickson, W. P. Resource selection by animals: statistical design and analysis for field studies. (Chapman and Hall, 2002).

  49. 49.

    Koehn, J. & Kennard, M. J. “Habitat” In Ecology of Australian freshwater fishes (eds Paul Humphries & Keith Walker) 81–103 (CSIRO Publishing, 2013).

  50. 50.

    Elith, J. & Leathwick, J. R. Species distribution models: ecological explanation and prediction across space and time. Annual Review of Ecology, Evolution, and Systematics 40, 677–697, https://doi.org/10.1146/annurev.ecolsys.110308.120159 (2009).

  51. 51.

    Sharma, S., Legendre, P., Boisclair, D. & Gauthier, S. Effects of spatial scale and choice of statistical model (linear versus tree-based) on determining species–habitat relationships. Canadian Journal of Fisheries and Aquatic Sciences 69, 2095–2111, https://doi.org/10.1139/cjfas-2011-0505 (2012).

  52. 52.

    Greenwood, M. Distribution, spread, and habitat predictability of a small, invasive, piscivorous fish in an important estuarine fish nursery. Fishes 2, 6, https://doi.org/10.3390/fishes2020006 (2017).

  53. 53.

    Elith, J., Leathwick, J. R. & Hastie, T. A working guide to boosted regression trees. Journal of Animal Ecology 77, 802–813, https://doi.org/10.1111/j.1365-2656.2008.01390.x (2008).

  54. 54.

    R Core Team. R: A language and environment for statistical computing. (Vienna, Austria, 2016).

  55. 55.

    Ridgeway, G. gbm: Generalized Boosted Regression Models, https://cran.r-project.org/web/packages/gbm/index.html (2017).

  56. 56.

    Hijmans, R. J., Phillips, S., Leathwick, J. & Elith, J. dismo: Species Distribution Modeling, https://cran.r-project.org/web/packages/dismo/index.html (2017).

  57. 57.

    Oksanen, J. et al. vegan: Community Ecology Package, https://cran.r-project.org/web/packages/vegan/index.html (2017).

  58. 58.

    Kennard, M. J., Olden, J. D., Arthington, A. H., Pusey, B. J. & Poff, N. L. Multiscale effects of flow regime and habitat and their interaction on fish assemblage structure in eastern Australia. Canadian Journal of Fisheries and Aquatic Sciences 64, 1346–1359, https://doi.org/10.1139/f07-108 (2007).

  59. 59.

    Newson, M. D. & Newson, C. L. Geomorphology, ecology and river channel habitat: mesoscale approaches to basin-scale challenges. Progress in Physical Geography: Earth and Environment 24, 195–217, https://doi.org/10.1177/030913330002400203 (2000).

  60. 60.

    Poff, L. N. et al. The ecological limits of hydrologic alteration (ELOHA): a new framework for developing regional environmental flow standards. Freshwater Biology 55, 147–170, https://doi.org/10.1111/j.1365-2427.2009.02204.x (2010).

  61. 61.

    Gore, J. A. & Shields, F. D. Can large rivers be restored? BioScience 45, 142–152, https://doi.org/10.2307/1312553 (1995).

  62. 62.

    Beesley, L. S. & Prince, J. Fish community structure in an intermittent river: the importance of environmental stability, landscape factors and within-pool habitat descriptors. Marine and Freshwater Research 61, 605–614, https://doi.org/10.1071/MF09137 (2010).

  63. 63.

    Pettit, N. E. et al. Dynamics of in-stream wood and its importance as fish habitat in a large tropical floodplain river. River Research and Applications 29, 864–875, https://doi.org/10.1002/rra.2580 (2012).

  64. 64.

    Pusey, B. J. & Kennard, M. J. In Northern Australia Land and Water Science Review full report Vol. 3–59 (2009).

  65. 65.

    Rosenfeld, J. S. & Boss, S. Fitness consequences of habitat use for juvenile cutthroat trout: energetic costs and benefits in pools and riffles. Canadian Journal of Fisheries and Aquatic Sciences 58, 585–593, https://doi.org/10.1139/f01-019 (2001).

  66. 66.

    Crook, D. A. & Robertson, A. I. Relationships between riverine fish and woody debris: implications for lowland rivers. Marine and Freshwater Research 50, 941–953, https://doi.org/10.1071/MF99072 (1999).

  67. 67.

    Bishop, K., Allen, S., Pollard, D. & Cook, M. G. Ecological studies on the freshwater fishes of the Alligator Rivers Region, Northern Territory (2001).

  68. 68.

    Pusey, B. J. & Arthington, A. H. Importance of the riparian zone to the conservation and management of freshwater fish: a review. Marine and Freshwater Research 54, 1–16, https://doi.org/10.1071/MF02041 (2003).

  69. 69.

    Morgan, D. L., Rowland, A. J., Gill, H. S. & Doupé, R. G. The implications of introducing a large piscivore (Lates calcarifer) into a regulated northern Australian river (Lake Kununurra, Western Australia). Lakes & Reservoirs: Research & Management 9, 181–193, https://doi.org/10.1111/j.1440-1770.2004.00247.x (2004).

  70. 70.

    Robinson, C. L. K. & Tonn, W. M. Influence of environmental factors and piscivory in structuring fish assemblages of small Alberta lakes. Canadian Journal of Fisheries and Aquatic Sciences 46, 81–89, https://doi.org/10.1139/f89-012 (1989).

  71. 71.

    MacRae, P. S. D. & Jackson, D. A. The influence of smallmouth bass (Micropterus dolomieu) predation and habitat complexity on the structure of littoral zone fish assemblages. Canadian Journal of Fisheries and Aquatic Sciences 58, 342–351, https://doi.org/10.1139/f00-247 (2001).

  72. 72.

    Winter, J. D. In Fisheries techniques (eds Brian R. Murphy & David W. Willis) 555–590 (American Fisheries Society Bethesda, Maryland, 1996).

  73. 73.

    Cooke, S. J. et al. A moving target—incorporating knowledge of the spatial ecology of fish into the assessment and management of freshwater fish populations. Environmental Monitoring and Assessment 188, 239, https://doi.org/10.1007/s10661-016-5228-0 (2016).

  74. 74.

    Scott, M. T. & Nielsen, L. A. Young fish distribution in backwaters and main‐channel borders of the Kanawha River, West Virginia. Journal of Fish Biology 35, 21–27, https://doi.org/10.1111/j.1095-8649.1989.tb03042.x (1989).

  75. 75.

    Schiemer, F., Spindler, T., Wintersberger, H., Schneider, A. & Chovanec, A. Fish fry associations: Important indicators for the ecological status of large rivers. SIL Proceedings, 1922-2010 24, 2497–2500, https://doi.org/10.1080/03680770.1989.11899997 (1991).

  76. 76.

    King, A. J. Ontogenetic patterns of habitat use by fishes within the main channel of an Australian floodplain river. Journal of Fish Biology 65, 1582–1603, https://doi.org/10.1111/j.0022-1112.2004.00567.x (2004).

  77. 77.

    Werner, E. E. & Gilliam, J. F. The ontogenetic niche and species interactions in size-structured populations. Annual Review of Ecology and Systematics 15, 393–425 (1984).

  78. 78.

    Ayllón, D., Almodóvar, A. & Nicola, G. G. & Elvira, B. Ontogenetic and spatial variations in brown trout habitat selection. Ecology of Freshwater Fish 19, 420–432, https://doi.org/10.1111/j.1600-0633.2010.00426.x (2010).

  79. 79.

    Godfrey, P. C., Arthington, A. H., Pearson, R. G., Karim, F. & Wallace, J. Fish larvae and recruitment patterns in floodplain lagoons of the Australian Wet Tropics. Marine and Freshwater Research, https://doi.org/10.1071/MF15421 (2016).

  80. 80.

    Donaldson, J. A., Ebner, B. C. & Fulton, C. J. Flow velocity underpins microhabitat selection by gobies of the Australian Wet Tropics. Freshwater Biology 58, 1038–1051, https://doi.org/10.1111/fwb.12107 (2013).

  81. 81.

    Warfe, D. M. et al. Productivity, disturbance and ecosystem size have no influence on food chain length in seasonally connected rivers. Plos One 8, e66240, https://doi.org/10.1371/journal.pone.0066240 (2013).

Download references

Acknowledgements

We are indebted to a large number of people for their assistance and hard work in the field over the period of study. This study was funded through various Australian Commonwealth Government organisations, including the Tropical River and Coastal Knowledge programme and National Environmental Science Program - Northern Australia Environmental Resources Hub. Funding and in-kind support was also received from the Northern Territory Government. We also gratefully acknowledge the Wagiman, Wardaman, and Jawoyn people, traditional indigenous owners of that part of the Daly River in which this study took place. We also want to highlight the ongoing collaboration with Wagiman Traditional Owners, particularly Mona Liddy and Elizabeth Sullivan, through a series of research agreements and steering committees. Wagiman Traditional Owners have been involved in project planning, field data collection and the sharing of cultural knowledge and protocols as part of their cultural responsibilities to managing country and to ensure safe travels for researchers on Wagiman land and waters.

Author information

M.M.D., S.J., M.J.K. and B.J.P. initiated and designed the original research; A.J.K., J.B.B., D.A.C. and S.A.T. proposed current research questions and initial analyses, K.K. and O.J.L. conducted the statistical analyses, K.K. led the writing of the manuscript, all authors conducted the fieldwork, writing and/or editing of the manuscript.

Correspondence to K. Keller.

Ethics declarations

Competing Interests

The authors declare no competing interests.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.