Frogs and flows: Using life‐history traits and a systematic review to establish water‐dependent functional groups for stream frogs in New South Wales, Australia

Hydrological alteration has contributed to the global decline of stream frogs. Flows support stream frog reproduction, juvenile development, food resources, and maintain habitats for all life stages. At present, there is a lack of information regarding the specific water requirements necessary for the conservation of stream frogs. To address this gap, we developed a traits‐based approach that serves as a valuable tool for grouping and prioritising water‐dependent stream frog species to inform future research priorities and environmental flow design. In this study, we focussed on 53 Australian frog species and analysed eight species traits to develop water‐dependent functional groups for stream frogs. We classified frogs based on their level of water dependency using an agglomerative hierarchical clustering analysis and a systematic review of water requirements and water management threats. The distinguishing traits that determined functional groups were: tadpole body type, egg clutch type, stream breeding habitat type, and documented association with flowing water and water permanence. Our study identified two distinct water‐dependent groups: facultative stream spawners, capable of reproducing in both stream and non‐stream habitats, and obligate stream spawners, restricted solely to stream habitats. Importantly, we highlight that the obligate stream spawners are the most sensitive group to within‐channel flow alteration and should be prioritised for water management decisions in lotic environments. This study represents the first comprehensive overview of the importance of hydrology for stream frogs and identifies the critical need for additional research and validation to enhance our understanding of stream frog responses to flows regimes.

Ensuring that freshwater ecosystems are sustainably managed to meet the water needs of humans and the environment is a major global challenge (Poff et al., 1997;Poff & Zimmerman, 2010).Increasing pressure on water resources as a result of a changing climate (Prosser et al., 2021), population growth and habitat destruction are expected to exacerbate biodiversity declines and species losses in freshwater ecosystems (Canessa & Parris, 2013;Reid et al., 2019;Sofi et al., 2020).
One of the main management strategies for restoring freshwater ecosystems is to establish environmental flows (e-flows) for river conservation (Arthington et al., 2010;Poff et al., 2017).E-flows aim to improve or protect key components of flow regimes such as their frequency, magnitude, duration and timing (Richter et al., 1996;Yarnell et al., 2020).Understanding the intricate relationship between flow and ecological responses is often a pre-requisite for designing e-flows, including the significance of individual flow components on various aspects such as habitats, abiotic conditions (Shenton et al., 2012), dispersal, reproduction, and subsequent effects on target populations and communities (Horne et al., 2017;Kupferberg et al., 2012;Yarnell et al., 2020).However, for numerous aquatic organisms like frogs, these connections often represent substantial knowledge gaps.
Frogs have declined globally, with over 30% of species threatened with extinction (Geyle et al., 2021).At greatest risk are the stream frogs, those species affiliated with flowing water that breed in or occupy lotic habitats and which represent 46% (2,650 species) of the global amphibian diversity (Stuart et al., 2004).The alteration of natural hydrological regimes has been linked to the decline of frog populations (Kupferberg et al., 2012;Littlefair et al., 2021;Railsback et al., 2016), reduced genetic diversity (Peek et al., 2021) and changes to the composition of frog communities (Wassens & Maher, 2011).Until recently, there has been relatively little effort to understand the mechanisms by which flow and other hydrological components (e.g.inundation) effects frogs in order to guide e-flows (Kupferberg, 1996;Ocock, 2013;Railsback et al., 2016;Wassens et al., 2008).However, much of this research has primarily targeted wetland species and has not been focussed on 'true' stream frogs that only use in-channel habitats (e.g.Littlefair et al., 2021;McGinness et al., 2014;Ocock et al., 2014).
Whilst targeted flow or inundation response research for stream frog populations is limited, there is substantial trait information available that provides insight into the importance of stream habitats and some hydrological components to a species (e.g.Anstis, 2017;Knowles et al., 2015).One method to identify environmental requirements across a group of similar species is to group them based on shared traits, which may be life-history (e.g.number of offspring) or other species traits (e.g.spawning habitat), hereafter called species traits.Developing water-dependent groups based on the association of a species trait with flow or inundation is a powerful approach often used in freshwater ecosystems, particularly for fish (Growns, 2004;Stocks et al., 2020;Wang et al., 2019).This trait information, combined with any more detailed literature can be used to inform e-flows, then tested and refined by targeted monitoring programmes (Davies et al., 2014;Poff & Zimmerman, 2010).However, a method to develop flow or water-dependent functional groups for stream frogs remains a priority knowledge gap.
This study had two main aims: firstly, to develop a new traitsbased approach that can classify water-dependent functional groups for stream frogs using species traits and a systematic review, and secondly, to use this approach to develop conceptual models and waterdependence hypotheses for each group.This information is needed in order to target future research through testable water-dependence hypotheses, assist with e-flow development, and guide water resource conservation actions to target the needs of declining stream frog species and populations at risk from alteration to hydrological regimes in Australia and internationally (Bondi et al., 2011;Horne et al., 2017).
We have applied this approach to stream frogs that occur in the State of New South Wales, Australia.

| MATERIALS AND METHODS
We followed the three-step framework for using species trait data established by Gallagher et al. (2021).This includes identifying the information needed for the conservation goal, trait prioritisation and identification of the limitations.Our approach can be broken up into three key steps: 1. Compile species traits with a hypothesised link to stream flow, inundation, or stream habitats to prioritise the most relevant traits.
2. Determine water-dependent clusters based on trait similarities.
3. Review of literature for available water, flow, habitat requirements, and water management threats for each species to refine group composition and to develop conceptual models and testable water-dependence hypotheses for each group.

| Study area
The study area is the entire state government region of New South Wales, Australia.This region includes the Great Dividing Range, a mountainous region extending from the north to the south of the

| Frog trait dataset
We compiled information on eight species traits (Table 1) that we a priori hypothesised were linked to hydrology for an initial list of 90 frog species that occur within NSW, Australia (Tables 1 and S1).
Trait data were either verbatim from Anstis (2017) or categorised where appropriate.Egg clutch type, egg clutch size, egg stage duration and tadpole duration were grouped into categories using the median, 25th and 75th percentile for each trait (Table 1).We assigned categories for these traits, as there was a large discrepancy in the accuracy of a species' numerical value.For example, some species (e.g.Litoria daviesae) had estimates of clutch size from one individual compared to other species that had an average and range taken from multiple individuals for the same trait (e.g.Litoria lesueurii).Species described after publication of Anstis (2017) were not included in this analysis as minimal data were available.
We calculated an additional trait which represented the frog calling distance from streams.Most frog calls are from males advertising readiness to breed (Köhler et al., 2017) from sites within breeding habitats (Rowley et al., 2019).We used this information as a measure of breeding habitat distance to streams to provide a trait that quantifies how close a species breeding habitat is to a stream.This was calculated using frog spatial records from the NSW state database BioNet Atlas (NSW DPE, 2023) and the NSW Strahler stream order layer (NSW DPE, 2022) as the streamline.We applied several filters to this dataset in order to calculate the median calling distance from streams for each species.The original dataset included 340,655 frog records.We then applied the following filters to the dataset: exclude records before the year 2000 (resulted in 283,469 records), exclude uncertain identifications with source codes <= 5 (resulted in 277,358 records), include only calling records (resulted in 62,370 records) and then a final filter to exclude species with less than 30 records.The final dataset included 62,165 records of calling frogs.
When information was missing for one of a species' traits, we estimated the value using the average for the genus or the value of a closely related species as indicated by Anstis (2017).Hereafter, we refer to these estimations as gap-filled traits.Species with more than one gap-filled trait were excluded from further analysis.This reduced the original list of 90 species to 61 species.There were an additional eight species which either bred terrestrially (e.g.Assa darlingtoni which is a pouch brooder) or Anstis (2017) suggested only bred in flooded T A B L E 1 Species traits for frog species used to develop water-dependent groups.Different stream breeding habitats have different inundation or flow associations which may be significant to specific species.Hydrolines were assessed against the Strahler stream order layer with creeks relating to 74% of waterways and had a median stream order of 4 (Q90 = 2, Q10 = 6), whilst rivers, hereafter referred to streams had a median stream order of 6 (Q90 = 4, Q10 = 9), suggesting they were larger and more permanently waterways.

Distance to stream
Median distance (m) of available frog records from a stream.Species with less than 30 records were excluded from further analysis Species which frequently call closer to stream habitats are more likely to depend on streams for breeding and other resources.
Note: See Table S1 for more detailed associated hypothesis related to the rationale (Anstis, 2017).
ditches, grasslands, ponds and dams.These habitats were not considered to have a meaningful link to in-channel stream flow and were removed from the trait analysis.The final list of frog species include 53 of the original 90 species, with nine species having one trait gapfilled (Table S2).

| Trait analysis
All analyses were conducted in R version 3.6.2(R Core Team, 2020) using the 'cluster' (Maechler et al., 2021), 'ggplot2' (Wickham, 2016:2) and 'factoextra' packages (Kassambara & Mundt, 2020).Gower's dissimilarity measure was used to calculate trait similarity among species as it can incorporate both categorical and numerical data.The dissimilarity measure was converted into a matrix and visualised to identify clustering patterns (Figure S1).We used the partitioning around medoids (PAM) clustering algorithm and the silhouette width method to determine the optimal number of clusters.The cluster solution with the highest silhouette width was considered the optimal number of clusters.
An agglomerative hierarchical clustering (AHC) analysis was performed using the Ward method (Ward, 1963) to generate a cluster dendrogram using the optimal number of clusters.The Ward2 method was used as it is the most accurate representation of the original Ward method (Murtagh & Legendre, 2014).Silhouette widths within each cluster were then assessed to identify any species that stood out as a poor fit within a cluster.The list of species within each cluster and the associated traits were extracted from the dendrogram and summarised to develop the water-dependent functional groups (Table S2).

| Systematic review of water requirements and sensitivity to hydrological alteration
A systematic review of peer-reviewed and grey literature was conducted to collate relevant information on (1) the broad water requirements of each species and (2) the water management threats  Australia, using Jowett (1993), Rosgen (1994), Thompson (2018) and Cowley et al. (2018).The hydrological components (e.g. a specific flow class like low flows) were modified from conceptual models for perennial and non-perennial rivers (Allen et al., 2020), floodplains (Opperman et al., 2010), upland swamps (Cowley et al., 2018) and the functional flows approach (Yarnell et al., 2020).We developed conceptual models for dominant breeding habitats and did not attempt to create conceptual models for all potential breeding habitats and stream cross-sections that occur within the state of NSW, Australia.

| Developing functional groups and waterdependence hypotheses
The  and S3).The key traits that distinguished each cluster are shown in Figure 2 and Tables 2 and S2.
Cluster 1 was the most taxonomically diverse cluster, including species from four families and seven genera: Litoria, Adelotus, Cyclorana, Platyplectrum, Lechriodus, Limnodynastes and Heleioporus.The average Gower's similarity among species within the cluster was the lowest of the four clusters: 0.67.Cluster 2 had the second highest average similarity of 0.75 and was the largest cluster with 20 species from the genera Litoria, Crinia, Neobatrachus, Paracrinia and Uperoleia species.
Cluster 3 included 13 species and contained all species in the study from the genus Mixophyes and some species of the genus Litoria (Figure 1).This cluster had an average similarity of 0.72.Cluster 4 was the smallest and least diverse with five Pseudophryne species and one Philoria species and had the highest average similarity (0.85).

| Key attributes of stream frog groups
The key traits that distinguished each cluster are shown in Figure 2 and Tables 2 and S2.Cluster 1 included species with larger males F I G U R E 2 Heat map showing the percentage of stream frog species within each level of the traits (male size and distance to stream were excluded from the heat map, as it was a continuous variable).Darker blue represents a higher proportion of species in a cluster fall within that trait category.
T A B L E 2 Summary of the eight traits for the stream frog clusters derived from the AHC analysis.Note: The categorical data are summarised as the most common trait category based on the percent of species with that trait.The numerical data are represented by the median (Q1, Q3).
(median = 63 mm) and a greater calling distance from streams than the other clusters.The cluster was also defined by an aquatic floating egg clutch type (93%), larger clutches (median = 979) and lentic tadpole body types (100%).The egg stage for this cluster is short (3-4 days) whilst the median tadpole development duration is 14 weeks, which is the longest of any of the clusters.Cluster 2 shared many similar traits with Cluster 1; greater calling distance from streams, tadpole duration and egg duration are all similar.However, male size (median = 33 mm) and clutch size are much smaller, and all species have submerged clutches.Cluster 3 was unique as it was composed of species that had lotic tadpole body types (100%), short tadpole durations (8-10 weeks), and preferred to breed in flowing streams (61%), permanent streams (8%) and stream pools (31%).Cluster 4 differed from the other clusters due to the small male sizes (27-30 mm), the unique composition of boggy seep spawners (100%), high proportion of clutch placement near water (100%), small clutch sizes and long egg stages (Figure 2 and Table 2).

| Systematic review results
Our systematic review returned 653 articles from the Web of Science and 15,857 articles from Google Scholar.We identified 158 relevant articles or grey literature reports with some form of information on stream habitat dependence, water requirements, or water management threats for 46 of the 53 frog species (Figure 3 and Table S3).The literature ranged from detailed flow-ecology studies The obligate stream spawners or true stream frogs only breed in stream habitats and are composed entirely of species from the AHC Cluster 3. The group is defined by the species' association with flowing streams with permanent aquatic habitats, lotic tadpole body types, and similar water management threats (Table 3 and Figures 3 and 4a).
This group of stream spawners are considered the most sensitive to changes to within channel flows based on the proportion of species with literature detailing general flow alteration, low flow alteration, and impacts of flow regulation by dams as a threat (Figure 3b).
Alteration to low flows, including unseasonal high flows due to river regulation are considered the key threats.The group was split into two OSS subgroups based on tadpole duration, body size and literature to support different water dependence (Table 3).
OSS subgroup 1 contains 11 species of Litoria.All species in this group have short to long tadpole stages (7-15 weeks) and a small to medium body size (<48 mm) (Tables 3 and S2).The smaller body size may relate to a shorter life span and reduced dispersal capability (Liu et al., 2021;Nali et al., 2014).Low flows may be important for protecting flowing habitats and minimising periods of zero flow (Table 3 and Figure 4).The significance of higher flows is unknown; however, they may be required to reduce sedimentation in riffles and maintain stream habitats.
The OSS subgroup 2 is composed entirely of the Mixophyes genus which have a distinctly long tadpole duration (≥21 weeks), large body F I G U R E 3 Summary of the systematic review results showing the literature categorised into different (a) broad water requirements categories and (b) water management threat categories.For Figure 3a, the values represent the proportion of species within each cluster with literature referencing a broad water requirement.For Figure 3b, the values represent the proportion of literature within each cluster referencing one or more of the water management threats.
T A B L E 3 Summary of the final stream water-dependent groups, water-dependence summary statements and testable water-dependent hypotheses.
Water size (>65 mm) and strong links to permanent streams with breeding sites in or near flowing riffles (Hunter & Gillespie, 2011;Knowles et al., 2015;Lewis & Rohweder, 2005;Stratford et al., 2010).Reduced flows and increased zero flow events during the breeding and egg development phase would result in egg desiccation and failed reproduction opportunities (Knowles et al., 2015).Therefore, this subgroup may require permanent pools all year and minimum flows during the breeding period to protect flowing water habitats.They are also expected to benefit from pulse events in the system during the breeding period (Stratford et al., 2010) and higher flows to maintain clean riffle breeding habitats.

| Facultative stream spawners (FSS)
The facultative stream spawning group includes species capable of breeding in stream habitats (temporary and permanent), but spawning is not restricted to this habitat.In many cases, streams were secondary habitats, with floodplain, wetlands and other habitats more important for FSS species.The group included AHC Cluster 1, 2 and 4 and is represented by three FSS subgroups based on life-history traits and the species' association with non-flowing habitats (subgroup 1), flowing streams (subgroup 2), and headwater habitats (subgroup 3) (Tables 3, S2 and S3).The FSS group may be sensitive to altered hydrology (Figure 3b), but this is more likely to be from any form of flow manipulation that impacts higher flows (e.g.Litoria raniformis), those that inundate backwaters, benches, off-river pools, wetlands and lagoons (Figure 4b,c).
FSS subgroup 1 contains 30 species from nine different genera.It includes all 14 species from AHC Cluster 1 and 16 species from ACH Cluster 2. The species traits within this group are highly variable.
However, the subgroup is distinct, as the species' are associated with non-flowing stream habitats and can breed in both stream and nonstream habitats (Table 3 and Figure 3).The subgroup includes species that occupy and breed in both permanent and non-permanent stream habitats such as coastal sedgelands (e.g.Litoria olongburensis, Lewis & Goldingay, 2005), isolated pools (e.g.Litoria revelata, Williams & Hero, 1998), and both floodplain and in-stream wetlands (e.g.L. raniformis, Heard et al., 2008;Wassens et al., 2010) Note: Each hypothesis is supported by suggested hydrological components and characteristics, as well as species-specific considerations, for example, differences in breeding seasons within a group or subgroup.FSS, facultative stream spawners; OSS, obligate stream spawning specialists.
frequency and/or extent) of key habitats appears to be the crucial factor for this subgroup (Table 3).This may be related to direct rainfall contribution or derived from different river flows inundating appropriate habitats.
FSS subgroup 2 is composed of four Litoria species, with very similar species traits to FSS subgroup 1, except that these species have a strong association with lotic habitats and can breed in both flowing and non-flowing permanent habitats (Table S3).The significance of stream flow for this subgroup is unclear, as there was very little detailed literature available for all species.However, they are likely to benefit from the maintenance of flowing water, but their ability to breed in nonflowing habitats suggests it is not crucial.Protecting the persistence of stream habitats for amplexus, egg development, tadpole development, and the adult active period may be important (Table 3).
The FSS subgroup 3 is composed of five species of Pseudophryne and one species of Philoria, which are small frog species with small clutches, long egg stages, terrestrial tadpole body types, and requires saturated seepage lines and boggy seeps within permanent headwater habitats (Table 2 and 3 and Figure 4c).We found no specific information suggesting the need for flowing water for this subgroup even though the subgroup had the closest calling distance to streams.
There was a clear requirement for moist terrestrial habitats for nests, permanent water filled cavities or pooling within burrows in stream banks to support tadpole development (Heard et al., 2021;Knowles et al., 2004).Hydrological alteration, climate change and the associated drying of habitats are listed as threats to this group.Based on these threats, breeding habitat associations, and tadpole development requirements, the permanence of moist terrestrial habitats and small pools for tadpoles may be required to support this subgroup (Table 3).

| DISCUSSION
Trait-based approaches are frequently employed as an alternative to taxonomy, enabling the examination of species or species groups' hypothesised responses along environmental gradients The diverse non-flowing habitats that the FSS subgroups 1 and 2 utilise (c) A crosssection of upland and headwater habitats important to FSS subgroup 3 and some species from FSS subgroup 1. (Chessman, 2013;Keck et al., 2014).Trait-based approaches have been used to develop flow-based or stream-dependent functional groups for a variety of biota including native fish (Baumgartner et al., 2013;Stocks et al., 2020), mussels (Gates et al., 2015), macroinvertebrates (Usseglio-Polatera et al., 2000) and riparian vegetation (Merritt et al., 2010).Our study is the first to demonstrate the importance of utilising species traits as a valuable tool to enhance our understanding of stream frog environmental water requirements and to determine priority species for conservation amidst threats posed by hydrological alteration (Martini et al., 2021).Whilst our initial clustering based solely on traits resulted in four groups, we found that the final water-dependent groups (comprising two main groups and five subgroups) required the incorporation of information from targeted literature searches to define the groups.Unfortunately, there was inadequate data regarding flowecology relationships within the traits and systematic review to suggest specific water requirements for each group or subgroup.Nonetheless, it is crucial to emphasise that traits remain a vital component within each group as they elucidate water-related characteristics such as optimal timing for crucial life-history events, like spawning.

| Limitations when using species traits
This study is subject to several limitations.One important consideration is the varying dependence of species traits on water or stream flow, which we have treated equally for all traits in this study, despite the likelihood of variations.For instance, tadpole duration is likely to have a much stronger dependence on river flow or inundation than egg clutch size, as longer tadpole stages require extended periods in aquatic habitats, and increased drying can result in premature metamorphosis, smaller metamorph size, or even mortality (Laurila & Kujasalo, 1999;Maciel & Juncá, 2009).There may also be unidentified traits that play a role in a species dependence on either flow or stream habitats.For example, geographic range size is an important trait for consideration in threat assessment (Liu et al., 2021;Mahony et al., 2022) and extinction risk (Geyle et al., 2021) and may impact a species susceptibility to threatening processes like flow alteration.
Another possible limitation is the effect of phylogenetically linked traits, which can group species by taxonomy, although the a priori hypothesised selection of water-sensitive traits in this study reduces the relevance of phylogeny (Menezes et al., 2010).Further, the addition of our systematic review as a second step provides a form of validation to support and refine the trait-based groupings.Overall, despite selecting traits based on their theoretical water-dependency, there remains significant research gaps for stream frogs that cannot be addressed using traits alone.

| The significance of flows for stream frogs remains a major research gap
Frog diversity and abundance are impacted by a myriad of threats, with alteration to stream flows considered just one key threat (Mathwin et al., 2021).Our study highlights that there remains little targeted flow-ecology research or flow-impact studies for frogs in NSW, Australia, particularly for lotic stream species like the obligate spawners identified in this study.Most of the research has focussed on lentic breeding frogs and the importance of how long and how frequently an area is inundated for reproduction (Littlefair et al., 2021;Salice, 2012;Seigel et al., 2006;Wassens et al., 2008;Wassens & Maher, 2011).However, stream species linked to lotic habitats may be associated with different hydrological components and related hydraulic conditions (Kupferberg, 1996).For example, tadpoles of the endangered Purple frog (Nasikabatrachus sahyadrensis) in southern India prefer higher velocities (Thomas et al., 2019), whilst eggs and tadpole survival in the lotic-breeding Foothill yellow-legged frog (Rana boylii) in North America is reduced during extreme flow velocities (Kupferberg et al., 2011;Rose et al., 2021) as it has very specific breeding site requirements (Kupferberg, 1996).This highlights the need for information on the links and underlying mechanisms between hydrological variables and stream frog ecology as a priority for future research.
From the 53 species in our study, 13 were identified as obligate stream spawning specialists that primarily inhabit permanent flowing streams.This group also had the largest number of primary literature articles referencing flow alteration as a threat and should be considered the obligate stream frogs in this region (NSW, Australia).Despite this, there is a lack of specific information regarding the environmental water requirements for these stream frogs, with most references offering only simplified explanations about the significance of flow.
For example, the importance of flowing riffles for breeding in Mixophyes balbus and Mixophyes fleayi is well documented but not quantified using any hydraulic or hydrological metrics (Hines et al., 1999;Hunter & Gillespie, 2011;Knowles et al., 2015;Lemckert & Brassil, 2000;Stratford et al., 2010).Another example is the South African Table mountain ghost frog (Heleophryne rosei) that requires permanent flowing waters to allow population persistence (Ebrahim et al., 2020).The lack of detail on the significance of specific flow components (e.g.low flows) on population dynamics and distribution of any of these species provides opportunities for targeted research to establish linkages between flow components and flow metrics to identify the specific requirements for stream frogs.

| Water management implications
We suggest that additional focus on the water management requirements (i.e.flow regulation) is needed for stream frogs.Particularly, the breeding period for many stream frogs coincides with warmer months when the competition for water between humans and the environment is often greatest.Based on our systematic review, we consider river regulation by large dams and water extraction from unregulated rivers, two of the major water management threats to stream frogs in this region.The regulation and delivery of large water volumes from dams for downstream use often results in reduced flow variability, higher velocities (Kupferberg et al., 2011), and commonly colder water temperatures during a period where species should naturally experience lower flows and warmer water temperatures for egg and tadpole development (Catenazzi & Kupferberg, 2013;Hunter & Gillespie, 2014).The second threat is the diversion or extraction of water from unregulated rivers during warmer periods, which has the potential to reduce stream flows, triggering very low flow and zero flow conditions more frequently (Smakhtin, 2001), which is listed as a key threat for many species (Stratford et al., 2010;Threatened Species Scientific Committee, 2021).Whilst we acknowledge a need for further research on stream frogs, there is adequate evidence at present to suggest that streams frogs should be considered during the development of e-flows or other related water management decisions.

| CONCLUSIONS
In this study, we provide a new approach to using species' traits in combination with a systematic review to develop water-dependent functional groups with the aim to guide the conservation and management of stream frog species.We highlight the significant gaps in quantitative flow-ecology studies for stream frogs and develop testable water-dependency hypothesis for different functional groups in NSW, Australia, to guide future research and improve environmental water management for this group of animals.This information should be incorporated in the decision process to better manage and integrate environmental flows to support key river frog life stages like egg deposition and tadpole development.Future investment in targeted research projects (i.e.ecological validation) will improve our understanding of stream frog responses to hydrological regimes, which will subsequently allow the fine-tuning of environmental water requirements and management options.
state which separates the western flowing rivers from the eastern flowing rivers.The western flowing rivers form part of the Murray Darling Basin.The headwaters of the Murray and Darling Rivers and its tributaries rise in the Great Dividing Range to flow over great floodplains and lowlands to the southwest.The eastern flowing river catchments flow from the Great Dividing Range to the east.They vary in size and geomorphology from large multi-catchment river systems arriving at the coast in broad coastal floodplains, or large drowned river valley estuaries, to small high relief headwater streams that flow directly to the Pacific Ocean.The climatic conditions range from subtropical in the north of the state, semi-arid in the west, to temperate in the southeast.Rainfall also varies across the region, with cycles of drought and flood driving the hydrology of rivers across the state.As a result, the region has a variety of ephemeral, intermittent and perennial rivers important to stream frogs.

(
see supporting information for details).The water requirements of species were split into four broad categories based on evidence linking any part of a species life history with water persistence or flowing water needs.These groups are (1) temporary stream habitats without flowing water, (2) temporary or permanent stream habitat without flowing water, (3) permanent stream habitats without flowing water or (4) permanent stream habitats with flowing water.Any threatening process that related to hydrological alteration was documented and categorised into five groups based on recurring themes.These groups were (1) enhanced drying of stream habitats, (2) alteration to low flows, (3) regulation of flows, (4) alteration to high flows and inundation, and (5) any flow alteration.The review results were summarised as the proportion of species within each AHC cluster could be linked to a broad water requirement or water management threat category.

2. 5 |
Developing stream conceptual modelsConceptual models of stream cross-sections and/or longitudinal profiles representing the key habitats used for breeding (amplexus, egg deposition and tadpole development) and their link to stream flow, inundation or other water dependence were developed using the key breeding habitats and water requirement literature identified by the systematic review.The generalised cross-sections and/or longitudinal profiles were developed using different morphological classifications and cross-sections relevant to key breeding habitats in NSW, AHC clusters were used in combination with the systematic review results to develop the final water-dependent functional groups for stream frogs.Species were included in a functional group based on the original AHC clusters unless the systematic review of water requirements and water management threats demonstrated a clear distinction in the species-specific water needs, in which case the species were divided into subgroups.The key habitats used by each subgroup were identified within the stream conceptual models to visualise the dependence of different life stages on flow or other hydrological components.For each water-dependent group or subgroup, a waterdependence summary statement was developed based on the species traits and literature for species within each group.This was then used to form water-dependence hypotheses and associated hydrological components (e.g.specific flow class), hydrological characteristics (e.g.duration), and species-specific considerations (e.g.breeding season) to guide future testing of each hypothesis.3| RESULTS3.1 | Trait-based grouping of stream frogsStream frog species were divided into four clusters based on the optimal number of clusters (average silhouette width of 0.39) and the traits-based AHC analysis (Figures 1, S2

F
I G U R E 1 Dendrogram showing the hierarchical structure between the four clusters from the trait-based AHC analysis.The dendrogram represents the hierarchical relationship among the clusters of species based on their individual species traits with the lower branches of the tree indicating species that are more similar.The solid-coloured lines indicate hierarchical structures that clustered together.
(e.g.Mathwin et al., 2023) to general information on habitat preferences (e.g.Hunter & Gillespie, 2014), with the majority of literature in the latter category.Based on our categorisation of the literature, Clusters 1, 2 and 4 include a range of species that have varied water permanence requirements (e.g.temporary and permanent habitats) but do not require flowing water for reproduction.In contrast, Cluster 3 was represented by a high proportion of species preferring permanent stream habitats with flowing water and a high proportion of literature relating to flow alteration, particularly changes to low flows.3.4 | Water-dependent functional groups 3.4.1 | Obligate stream spawners (OSS)

F
I G U R E 4 Conceptual models for the water-dependent functional groups summarising the water requirement of key habitats used for breeding, including egg deposition and tadpole development and their link with hydrological components.(a) A longitudinal profile showing the link between flowing habitats like riffles and stream pools which are important water requirements for the obligate stream spawners (OSS) and a subset of facultative stream spawners (FSS) subgroup 2. (b)