Soil seed bank can complement restoration efforts in a coastal freshwater creek

Riparian vegetation is a keystone ecosystem element heavily impacted by livestock grazing. Historically, regeneration approaches of the riparian vegetation use either active (planting) or passive (natural regeneration) strategies. Frameworks based on an understanding of the soil seed bank are needed to help guide the approach adopted. This study compared the soil seed bank composition to the extant riparian vegetation to assess the potential for natural regeneration to supplement active plantings, following livestock exclusion on the south-coast of Australia. At each of four sites, we surveyed the extant vegetation, and collected five soil samples from each of three zones (instream, slope, bank). Our results show that the proportions of species nativeness, growth forms, and life cycle was similar for both communities. However, while there were more terrestrial species recorded in the standing vegetation, there were more wetland specialists in the soil seed bank. A persistent native seed bank, combined with other basic management strategies, could potentially compete and limit the abundance of non-native or invasive grasses. The high abundance of native grass-types (Cyperaceae, Juncaceae, Juncaginaceae, Asparagaceae, Poaceae, Typhaceae), can be relied upon to regenerate the understorey to complement active planting of the trees and shrubs. As such, resources could be prioritised to sites where native species abundance in the seed bank is low. This study recommends that land managers should investigate the composition of the soil seed bank prior to allocating resources for active planting activities to ensure the most efficient use of resources across the landscape.


Introduction
As the interface between aquatic and terrestrial ecosystems, the riparian ecotones generally hold higher species diversity and plant productivity than adjacent uplands, having complementary environmental processes from both terrestrial and aquatic biomes (Kauffman and Krueger 1984;Naiman and Decamps 1997).Despite occupying a relatively small area in the landscape, riparian areas have been recognised as having a disproportionate large number of ecosystem services (Hunter 2017).These include nutrient recycling, riverbank stabilisation, flood attenuation, groundwater recharge, water purification and regulation of water flows and temperatures (Richardson Vol:. (1234567890) et al. 2007), which is becoming even more important with the predicted more frequent extreme climate events (Isaak et al. 2012;Leigh et al. 2015).As such, they act as 'blue-green' corridors that are vital for biota movement, supporting food webs, and influencing the microclimate such as providing shade (Gregory et al. 1991;Beschta 1997;Baxter et al. 2005;Peterson et al. 2011).
However, human disturbances such as clearing and grazing along waterways have detrimental impacts on these services.Livestock grazing in riparian zones is an example of disturbance that can result in either chronic (e.g., Beever et al. 2003) or acute (e.g., Walker 1995) ecosystem impairment, depending on the timing, duration and intensity of grazing (Sternberg et al. 2003;McInnis and McIver 2009).Grazing animals reduce or eliminate the standing vegetation and remove potential seed plants.This results in decreased vigour, biomass and root strength of riparian plants of the standing vegetation, as well as changing species composition and diversity (Pond 1961;Platts 1979;Bryant et al. 1982;Robertson and Rowling 2000).Foliage removal increases the amount of bare ground, thus also allowing increased soil temperatures and evaporation (Kauffman and Krueger 1984;Trimble and Mendel 1995;Beschta and Ripple 2012).Ungulates also elevate soil loss and compaction that cause reduced soil productivity, increased runoff, and decreased water availability to plants (Beschta et al. 2013).Along with livestock urine and faeces that add nitrogen to soils, which favour nonnative species (Asner et al. 2004;Beschta and Ripple 2012), these factors all compromise support for native plant communities.After decades of anthropogenic modifications, the native riparian vegetation communities will require more active support to restore key ecological processes (González et al. 2017).
Restoration approaches to accelerate the desired successional trajectories of riparian zones may include re-establishing hydrogeomorphic processes, active plant and seed introductions, exotic species control, conversion to natural floodplain, grazing and herbivory control, and others (e.g., topsoil removal, limitation or enhancement of fertilisation) (González et al. 2015).Re-establishing hydrogeomorphic processes includes re-introducing natural flow of surface water, groundwater and sediment, as a local-scale approach.Conversion to natural floodplain is a more broader landscape approach including removing and setting back flood defenses (Dadson et al. 2017).Regeneration of the riparian vegetation can be addressed using several strategies: planting, direct seeding (active strategies) or natural regeneration from seed bank (passive strategy).Active planting includes planting cuttings, seedlings, or sapling poles, while active seeding includes the provision of natural seed fall, or seeds collected from soil banks and other propagules (González et al.2015).In passive restoration, vegetation is allowed to regrow naturally from existing seed banks (ter Heerdt and Drost 1994; Brock and Rogers 1998;De Steven et al. 2006;Marchante et al. 2011;Cui et al. 2013;Lisboa et al. 2021;Erfanzadeh et al. 2023).Studies have suggested that in conjunction with suitable above-ground management strategies, the seed bank may harbour native species that are able to establish (Roche et al. 1997;Crosslé and Brock 2002;Sarr 2002;Vosse et al. 2008;Marchante et al. 2011;Ruwanza et al. 2013;Sarneel et al. 2014).An obvious benefit of natural regeneration once livestock are excluded, is the low-cost and affordability.Nevertheless, many studies suggests that the longer duration of disturbance, the more degraded the soil seed bank will be (Bakker et al. 1996;Thompson 2000;Chang et al. 2001a, b).Therefore, understanding the composition and viability of the soil seed bank could inform restoration planning and resource management.
While many studies have investigated the impacts of livestock and other human disturbances, such as flow regulation and invasion of exotic species, on wetlands and riparian soil seed bank (e.g., Belsky et al. 1999;Navie et al. 1996;Nicol et al. 2007;Williams et al. 2008;Kelleway et al. 2020), and the trajectories of change following restoration (Hale et al. 2018;McKergow et al. 2016;Robertson and Rowling 2000;Sievers et al. 2017), they are mainly focused on large inland river systems, particularly in arid and semi-arid regions (Capon and Reid 2016;Jensen et al. 2001;Powell et al. 2000).There is less research on agricultural landscape in coastal catchments that are generally small, humid, and heavily impacted with high-density stocking (Thornton and Elledge 2021).In Australia, to our best knowledge, there is no previous research on the potential of soil seed bank for restoration efforts on heavily degraded coastal streams.In south-eastern Australia, coastal streams are heavily impacted by land clearing, grazing and development of housing and urban settlements.The New South Vol.: (0123456789) Wales (NSW) coastal regions are under particular stress with over 85% of the population residing along the coastal zone.Recently, government programs in NSW have provided funding for landholders to exclude livestock from streams and riparian zones on the south coast.As part of this management strategy, the local government agencies have supplied native plants for planting inside the fenced riparian zone.However, anecdotal observations from both landholders and researchers indicated that due to the extended drought (2018-2020) there was a risk that these plantings may perish.Along with the active plantings, understanding the availability of native species in the soil seed bank for natural regeneration would provide the community and land managers with the confidence that their investment will achieve desired restoration outcomes.
For this study, we wanted to answer the question of whether the composition of soil seed bank is sufficient to renaturalise the riparian zone with representative native species from all vegetation stratum, when livestock were excluded from the sites.We did this by comparing the standing vegetation in Autumn and Spring of 2020, with the seed banks collected in Autumn 2020 at four sites in south-eastern Australia.The results of this study could inform future resource allocation for coastal streams.

Regional setting
Our case study site is on Victoria Creek, a major tributary flowing into Tilba Tilba Lake (-36.3281°S,150.1156°E,Fig. 1), on the south coast of NSW, Australia.In comparison with other nearby coastal lagoon systems (e.g., Wallaga Lake, Lake Mummuga), the Victoria Creek catchment has largely been cleared of native vegetation.Clearing occurred for various purposes including mining, forestry, dairy and beef farming, improved pasture, building and road constructions, and began as early as the 1850s (Hope et al. 2007).This has altered the natural catchment processes and impacted the freshwater and estuarine ecosystems (Hetherington 1983).By February 2021, approximately 62% of the of woodland or forest vegetation in the catchment had been cleared (Pinto et al. under review).
Most remnant native vegetation cover is located at higher elevations in the catchment and includes areas of rainforest, wet sclerophyll forest, and dry sclerophyll forest (Tozer et al. 2010;Roff et al. 2022).Before clearing, mid to lower areas of the catchment were likely comprised primarily of wet sclerophyll forest and grassy woodland.The riparian zone is thought to have been occupied by gully rainforests, including a tall eucalypt stratum dominated by Eucalyptus botryoides, a dense rainforest mid stratum of rainforest trees, shrubs and vines, and an understorey including ferns grasses and forbs (Roff et al. 2022).The creek system begins from the historic volcanic mountain Gulaga and passes through Central Tilba township.Although there are no dams on the system, watershed slopes are generally steep and extensively modified before entering the Tilba Tilba Lake system.This has posed some unique challenges to the hydrologic regime of the creek system.For example, Pinto et al. (2023) reported that in the current set up rainfall over 20 mm is likely to generate overland flow potential for habitat smothering for downstream macroinvertebrates and fish communities following rainfall events and lack of riparian buffer creates turbulence and vertical mixing affecting nutrient dynamics.Very little is known about the composition of the in-stream aquatic and semi-aquatic vegetation along Victoria Creek and surrounding streams prior to clearing and development.

Study sites
The first recorded history of the dairy farming in the Tilba Tilba region is in 1840 where a small-scale dairy farming occurred (Gibbney and Hoyer 1983).Over the next few decades, the area attracted more people for gold mining, and improved-pasture grazing of beef and dairy cattle.As such, the area had a history of various mixed land use patterns,, currently, the study sites were purely used for beef cattle farming.
All sites were grazed with beef cattle up until the time of fencing.The grazing densities and histories of the sites are not known in detail.However, it is likely that the sites were grazed for beef cattle farming for at least 20 years.
The stream form at our survey sites (~ 100 m reach) included an incised channel with pool and riffle sequences at baseflow conditions.Prior to clearing, the vegetation at all sites likely consisted of South Coast Temperate Gully Rainforest extending out from the stream banks, and Southern Lower Floodplain Freshwater Wetland in the stream and at the stream-edge (Roff et al. 2022).At the time of the survey all sites consisted of a highly modified pasture and riparian vegetation community including a mix of native and non-native grasses, forbs and ferns.Only a very few remnant native trees and shrubs (Eucalyptus and Acacia spp.) were present at sites VC1 and VC2.Sites VC3 and VCR1 had no remnant trees or shrubs.
At each site, a fixed and random plot (20 × 20 m) was surveyed for extant vegetation and 15 random soil samples were collected for the soil seed bank germination experiment.

Soil seed bank collection and germination of seeds
A total of 60 (15 cores per site × 4 sites) surface sediment samples (to 10 cm depth) were collected on 24-25 February 2020 from the four sites to determine the seed bank composition.To ensure the soil samples were representative of the riparian zone, we collected five replicate sediment cores (60 mm diameter, 100 mm depth) randomly from three zones (A: instream or channel, B: riparian slope, C: top of bank) (Fig. 2) within a 100 m stretch of the reach.As the water level was low due to the drought, we defined Zone A as the centre of the channel and with at least 20 cm of standing water, Zone B was the riparian slope of at least 45 degrees, and Zone C was the top of the bank.The 5 replicate cores from each zone (at each site) were pooled in a plastic bag and returned to the laboratory for processing.
In the laboratory, each composite soil sample was handled according to the seedling-emergence method of ter Heerdt et al. (1996): washing each sample with water on a coarse sieve (4.0 mm mesh width) to remove rocks, roots and other organics, and then washed through a fine sieve (0.212 mm mesh width) to remove clay and silt.Compared with unconcentrated samples, this method increases the number of species and individuals emerging from the samples and the germination rates (ter Heerdt et al. 1996).
A total of 72 buckets (20 cm height, 20 cm diameter) were prepared with vermiculite (Brunnings 5L Vermiculite), perlite (Brunnings 5L Perlite Potting Mix) and seed potting mix (Brunnings 15L Coir seed Raising Potting Mix Block).The 72 buckets contained 3 replicates from each site (4), zone (3) and watering treatment (2) (Fig. 2).An additional three buckets containing only soil preparation mixture and reverse osmosis (RO) water were prepared as controls (one buckets per treatment) to detect possible plant germinations from wind-blown propagates or impurities in the soil mixture.
To maximise germination conditions for the seeds we applied three watering treatments: Rain, where soil was wetted daily and allowed to drain; Waterlogged, where the water level was maintained at soil surface; or Flooded, where the water level was maintained 100 mm above the soil surface.The waterlogged watering regime has been shown to provide the highest species richness in germination trials (Kelleway et al. 2020) and was thus used for all experimental treatments.With this in mind, the Flooded treatment was only used for the Zone A (instream or channel) samples, and Rain treatments were only used for Zone B and C (Fig. 2).For this paper, we were only interested in what species germinated from the seed bank, to better understand the relationship between the extant vegetation and the soil seed bank.As such, the data from zone and watering treatment were pooled into site.
Each bucket was then covered with plastic food wrap to reduce evaporation and create a greenhouse effect.The glasshouse lights were set for 12 h during the day and the temperature was kept constant at 26 °C.Along with checking water levels and ◂ identification of plants, photos were taken at least fortnightly for the first six months, and then monthly for the next four months (a total of 300 days).Once counted and identified, the entire seedling was removed to avoid self-seeding, and preserved as a dried voucher specimen.After 300 days, any remaining plants were left for plants to seed to confirm identifications.

Standing vegetation survey
Surveys of the extant vegetation to record the species composition were conducted in late summer (24-25 February) and mid-spring (19-21 October 2020).A fixed and random plot (20 × 20 m) was surveyed at three sites (V1, V2, V3) in both seasons to facilitate species identification and recording.Random plots were chosen using a random number table to allocate the beginning of the plot along a 100 m stretch of the stream.The control site (VCR1 -not fenced) was only surveyed in the late summer of 2020.Species presence and cover were recorded in each plot.
Assignment of growth forms and High Threat Weeds (HTW) or invasive exotics for the NSW southcoast followed the Biodiversity Assessment Method (BAM) calculator (https:// www.envir onment.nsw.gov.au/ bamat oolapp/, accessed Nov 2019).Grouping of wetland plant indicator list (WPIL) scores followed Ling et al. (2019) where the degree of each species' association with water is quantified on a continuous 'status' scale ranging from 1 to 5. A score of 1 designates a species with a strong association with wet conditions (WI: wetland indicator species), while a score of 5 designates the species that has a weak association with water (NWI: non-wetland indicator).These values were analysed in three groups to characterise WI species (codes 1-2.9), potential wetland indicator species (PWI: codes 3-3.9), and NWI species (codes 4-5).A voucher specimen was taken of most species.Naming conventions for all plant species follow the NSW PlantNet standard (https:// plant net.rbgsyd.nsw.gov.au, accessed July 2022).Asterisks are used throughout the manuscript to identify nonnative species (*) and invasive or high threat nonnative species (**).

Statistical analysis
To investigate the relationship between the extant vegetation and the soil seed bank we explored different indices (species richness; numbers of non-native, native and high-threat weeds (HTW); growth forms: woody, grass-types, and forbs, wetland indicator classes) and life cycle groupings (perennial, annual, biennial) in each community at each site.To compare these indices between the two communities, the data were converted to proportions of the total number of species for each sample.
Univariate analysis of variance (ANOVA) was used to calculate statistical differences between the means of each index for each site within each community (extant vegetation or seed bank) and where significant, we tested for post hoc differences between individual sites with Tukey's HSD test.Univariate dispersion tests (Levene's Test) showed that homogeneity was met for all indices between groups (Sites) (p > 0.05) for the standing vegetation data and all except NWI for the seed bank data.None of the transformations (square-root, cube-root, 4-squareroot, log10) could coerce the NWI data to satisfy the assumption of homogeneity of variance or normality and no further analyses were completed.
Besides the above univariate comparisons of community index, we also used non-metric multidimensional scaling (nMDS), an indirect gradient analysis approach (Clarke 1993), to compare the species composition between seed banks and extant vegetation.As the abundance measure is not the same for extant vegetation (percentage cover) and seed bank (no of individuals), ordination was conducted with the presence-absence based Sørensen index, which is equivalent to the Bray-Curtis distance in binary form (Legendre and Gallagher 2001).Permutational multivariate analysis of variance (PERMANOVA) was then used to determine whether the difference in species composition between seed bank and extant vegetation is significant.All statistics were conducted in R programming environment (R Core Team 2020), the functions metaMDS and adonis2 of the R package vegan (Oksanen et al. 2009) was used for NMDS ordinations and PERMANOVA, respectively.

Results
A total of 121 plant species from 52 families were identified in the vegetation survey and seedling emergence trial, of which 117 were recorded in the extant vegetation (Autumn and Spring 2020) and 39 germinated from the soil seed bank (Appendix A).Of these species, four species were unique to the seed bank, 82 were recorded only in the extant vegetation, and 35 (29%) were common to both.Overall, 60 species were native, 50 were non-native, and 11 were highthreat weeds (HTW) or invasive non-natives for the NSW south coast.
Across all growth forms, forbs dominated the communities (65 species) followed by the grass-types (36 species).In the extant vegetation, ten woody (trees and shrubs) species were recorded, with the invasive weeping willow (Salix sp.**)only recorded at sites VC3 and VCR1.From the soil seed bank, the only woody species germinated was the black wattle (Acacia mearnsii) from VC1.
Vol:. ( 1234567890) Relationship of the seed bank ratios to the extant vegetation Since the relationship between the composition of the soil seed bank and the standing vegetation was the key question of this study, we compared the proportions of different functional groups between the two communities: nativeness, growth forms, life cycles, and water dependencies of the species (Table 1).

Nativeness
Despite the difference in the total number of species recorded for extant vegetation (117) and seed bank (39) samples, the overall ratios of native, nonnative, and HTW or invasive species were similar between both communities for native (~ 49:49), non-native (~ 43:41), and HTW (~ 9:11) (Table 2).Of the non-native species, ten of those recorded in the extant vegetation are classed as HTW or invasive non-natives for the south coast.Four HTW species also germinated from the soil seed bank samples: Axonopus fissifolius** (carpet grass), Cenchrus clandestinus** (kikuyu), Juncus articulatus** (jointed rush), Senecio madagascariensis** (fireweed).However, while the extant cover of all sites was dominated by C. clandestinus** in 2020, the germinated seed numbers were dominated by

Growth and life cycle forms
The majority of both communities (extant vegetation and seed bank) were dominated by forbs (Table 1), and this was reflected in both the native (~ 21:24) and non-native (~ 34:35) proportions.Likewise for the grass-types in each community, the proportions of native (~ 14:14) and non-native (~ 14:16) were also reflected in the SSB.The native grass-type species (in families Cyperaceae, Juncaceae, Juncaginaceae, Asparagaceae, Poaceae, Typhaceae) were also coded as strongly associated with water (WI) in both communities.The only native grass designated as terrestrial, Cynodon dactylon (WPIL code 5) was recorded in the extant vegetation.Similarly, the proportions of ferns in the seed bank also reflected those in the extant vegetation (~ 5:5).Across all sites, the life cycles groupings of the species were also reflected in both communities, with both communities dominated by perennial species (~ 91:87).The similarity between the extant and seed bank communities was further mirrored in the perennial grass-types (~ 28:32), both for native (~ 15:16) and non-native (~ 14:16) grass-type species, as well as for the forb species (~ 55:60), both native (~ 21:24) and non-native (~ 34:35) species.

Water dependant species
In contrast to the other indices, the proportions of wetland indicator species were opposite for the two communities.That is, in the extant vegetation there were less water-dependant (WI and PWI) than terrestrial species (NWI) (~ 44:56), yet in the soil seed bank samples more water-dependant species germinated than terrestrial species (~ 62:38).In both communities, however, water dependant species were dominated by natives (~ 36% and 49%, respectively).
More species diversity at site VC2 Overall, site VC2 was more diverse than the other sites.In both the extant vegetation and seed bank samples the site had statistically more species richness than the other sites, including more native and non-native species (Fig. 3).For the extant vegetation, the total number of species recorded at VC2 in 2020 was 83 species, compared to the site with the lowest species richness at VCR1 of 33 species.The number of species recorded in the 20 × 20 m plots at VC2 ranged from 50 -37 species, with 54 -48% being native species.Of all the native species, 12 species only occurred at VC2, of which eight are considered wetland or potential wetland indicators: Isachne globose, Rumex brownii, Carex sp., Cyperus imbecillis, Carex longebrachiata, Tetragonia tetragonioides, Plectranthus parviflorus, and Sigesbeckia orientalis.Similarly, for the seed bank data, the species richness was statistically more diverse at VC2 than VC3 and VCR1 (TukeyHSD, p adj = 0.018 and 0.003 respectively, Fig. 3b), with a total of 28 species identified, compared to the site with the lowest  species richness in the soil seed bank at VCR1 (16 species).Of the native species, two germinated from the soil samples from VC2 that did not occur at any other sample: Calochlaena dubia and Persicaria decipiens, which are both wetland or potential wetland indicators.
In the extant vegetation plots, VC2 also had statistically more woody, fern (except VC1) and non-wetland indicator species than the other sites (Fig. 4a).Of the 10 woody species recorded in the extant vegetation on Victoria Creek, eight of these occurred at VC2, of which six are native species (Eucalyptus botryoides, Acacia mearnsii, Eucalyptus robusta, Corymbia maculata, Acacia sp., Breynia oblongifolia).In contrast, VC3 and VCR1 only had one woody species, and this was the invasive weeping willow (Salix spp.).The only woody plant that germinated from the soil seed bank was Acacia mearnsii (black wattle) from VC1 (Zone B).This wattle was also recorded in the standing vegetation.Of the six fern species that were recorded in 2020 extant vegetation, five of these species occurred at VC2 (Adiantum aethiopicum, Azolla filiculoides, Calochlaena dubia, Hypolepis muelleri, Pteridium esculentum), with VC1 also recording Blechnum nudum but not C. dubia.This similarity in VC1 and VC2 is obvious in the boxplots of the ferns detected in Tukey's post-hoc test for pairwise comparisons (Fig. 4a).The total numbers of non-wetland indicators or terrestrial species were also highest at the VC2 site with 47 species, compared with 30 terrestrial species recorded at VC1.
From the seed bank analyses, VC2 also germinated statistically more forb species and wetland indicator species than the other sites (Fig. 4).
All other indices were not statistically different (p > 0.05).
Comparing seed bank and standing vegetation composition and assemblages (presence-absence data) The PERMANOVA results showed species assemblages were significantly different between seedbank and extant vegetation (global R = 0.5429, P = 0.0001,Fig.5).Assumptions of homogeneity of multivariate dispersion were met for species assemblage data (ANOSIM statistic R = 0.15 and 0.06, P > 0.0001).All samples grouped closely by seed bank and extant vegetation with the exception of the 7 replicates samples from site VC3 that only had one seed germinate from each replicate.

Discussion
Despite the riparian zone being dominated by the invasive non-native Cenchrus clandestinus (kikuyu), and subjected to more than 150 years of clearing and grazing, our study found that native plant species were not depleted from the soil seed bank.We found a clear relationship between the soil seed bank and the extant riparian vegetation for nativeness and growth form at each site.That is, the overall ratios of native, non-native, and HTW or invasive species were similar between both communities, and both communities were dominated by forbes and this was reflected in both the native and non-native proportions.With 97% of the species recorded in the seed bank occurring in the extant vegetation, the seed bank could best be described as a subset of the extant vegetation, rather than a distinctly different community.Similar to other studies in inland rivers (Williams et al. 2008), we found that 28% of the recorded species were common to both standing and seed bank communities.
Our results showed that the seed bank at the coastal Victoria Creek held slightly more species (~ 31%) in the extant vegetation than other similar studies (20-25%) for inland systems (Wilson et al. 1993;Beismann et al. 1996;Ling et al. 2022).Despite the extended period of grazing history along Victoria Creek (i.e., since the 1850s.Hope et al. 2007), our study demonstrated that the riparian zone still contains viable soil seed banks, which have higher percentage of native species than the extant vegetation communities, providing direct benefits to the natural regeneration of these sites.For example, species such as Juncus usitatus, J. bufonius and Alisma plantagoaquatica were highly abundant in the seed bank, with the native Juncus spp.accounting for ~ 48% of the seedlings.Juncus spp.are well-reported for their large, persistent soil seed banks even when absent from the standing vegetation (Wisheu and Keddy 1991;Lunt 1997;Greet et al. 2012).These species could provide important bank stability and shading functions, both instream and the stream banks up to the fencing, to reduce the cover of the currently dominating exotic species.Importantly, none of the species recorded in the seed bank were part of the replanting scheme.These native species can add further structural complexity to the understorey layer and complement the tree and shrub layer.These native emergent riparian communities are particularly important, given the vital role of rooted aquatic macrophytes in the structure and functioning of shallow freshwater ecosystems to provide ecosystem services such as floodwater retention and water purification (Jeppesen et al. 1997;Zedler 2000).This structural layer also supports vital habitat for aquatic fauna such as waterbirds, frogs, fish and macroinvertebrates that rely on the aquatic and fringing non-woody vegetation Passive restoration or natural regeneration studies that rely on the seed bank have mixed results.Some studies have found lower native cover and species richness in passive restoration compared to active restoration strategies (Gornish et al. 2017).Others have found good recovery where the densities of invasive species were low (Reinecke et al. 2008;Galatowitsch and Richardson 2005), but poor recovery where the standing vegetation community was dominated by non-natives (Blanchard and Holmes 2008), with some studies finding no recruitment of native species (Ruwanza et al. 2013).Other studies have suggested that passive restoration may be preferred over any active techniques and may be necessary to restore stream function (Kauffman et al. 1997;Roper et al. 1997;Tullos et al. 2009).Clearly, from these studies, many factors influence recruiting success at any site including time in the season, flooding history before seed bank replenishment, and species-specific dormancy conditions.
Interestingly, three species (Elatine gratioloides, Juncus bufonius and J. articulatus**) that germinated from the soil seed bank were not recorded from any of the 2019-2022 vegetation surveys conducted in this area (unpublished data).However, both species are known to produce large numbers of minute seeds that can persist in the seed bank for at least 12 years (McIntyre 1985, Chamber et al. 1995, Albrecht 2002), and for J. articulatus** up to several decades (Bakker et al. 1996).Other studies have suggested that the absence of species found in the seed bank from the extant vegetation could be due to a lack of the ideal conditions for germination such as the dependency of some wetland species to germinate with exposure to unfiltered sunlight (Cresswell and Grime 1981;Keddy and Reznicek 1982;Smith and Kadlec 1983;Change et al. 2001).The presence of the native Elatine in the seed bank at all sites suggests that it has the potential to contribute to the native species diversity of the instream or channel vegetation given the right conditions for germination.

Nativeness relationships
Our results showed that the seed bank and the extant vegetation had similar community indices for nativeness.Few studies reported the proportions of native and exotic species when investigating the relationship between the seed bank and the extant riparian vegetation in freshwater wetland ecosystems, even though they highlighted the importance of the native/ Fig. 5 A nMDS plot for standing vegetation plots (green) and seed bank (red) for species assemblages (presence-absence) (Stress = 0.1655).PER-MANOVA showed statistically significant differences between the vegetation plots and seed bank communities (global R = 0.5429, P = 0.0001) exotic ratio is important for any restoration effort (Hopfensperger 2007).We calculated that a study by Davies et al. (2013) also found similar proportions of native (~ 69:71) and exotic (~ 31:29) species between the extant vegetation and the seed bank impacted by grazing on Kangaroo Island in South Australia.
In our highly degraded sites, nearly half (~ 49%) of the species for both communities were native, although this was site dependant.While Greet et al. (2013) found similar results (~ 44% natives), when we calculated the proportion of native seeds in other seed bank studies in Australia, < 63% native species is considered low (Lunt 1997;Williams et al. 2008;Brock et al. 2011;Casanova 2012;Davies et al. 2013;Greet et al. 2013).

Growth form and life cycle relationships
The life cycles (perennial, annual and biennial) and growth forms of the species were also highly similar in both seed bank and extant communities.This was similar to other studies in coastal NSW (Ling et al. 2022) and Victoria (Williams et al. 2008) where most of the seeds that germinated (~ 62%) were perennial monocotyledons such as grasses, reeds, rushes, and sedges.This contrasted to studies in semi-arid regions of Australia where viable seed banks were dominated by annuals or biennial dicotyledons, especially in degraded sites (Navie et al. 1996;Greet et al. 2013;Capon and Reid 2016;Kelleway et al. 2020).Considering the average understory cover of the standing vegetation in our study is dominated by perennial grass-types (Cenchrus clandestinus* ~ 84%, Cynodon dactylon ~ 13%, Typha domingensis ~ 8%, Schoenoplectus validus ~ 8%) this was not unexpected especially since these species produce large quantities of seeds.For example, Typha sp. can produce 20,000-700,000 fruits per inflorescence (Ahee et al. 2015;Bansal et al. 2019).Only one woody plant germinated from the soil seed bank (Acacia mearnsii), which produces copious numbers of small seeds that are not dispersed actively but often buried by ants and can remain viable for up to 50 years (O'Neill Seeds 2022).These seed bank germination results reflect other studies in Australia, with few shrubs or trees emerging (Williams et al. 2005;Capon and Reid 2016), the abundances dominated by grass-types, and diversity dominated by forbs (McIvor and Gardner 1994;Williams et al. 2008).
Studies have suggested that in repeatedly disturbed communities, there is a high similarity in the species composition between the standing vegetation and the seed bank (Chang et al. 2001a, b).In later successional stages, the standing vegetation is typically comprised of perennial species, while the seed bank is typically composed of short-lived annuals (Chambers 1993).Our data concur with these authors given both the extant vegetation and the seed bank communities were highly similar in the degrees of nativeness, growth forms and life cycles, and given the extended history of grazing disturbance at our sites.

Water dependant species
The target species for riparian restoration are those water-dependant or wetland specialist species that could renaturalise the wetter zones.The dominance of wetland specialists in the seed bank (~ 63%) with nearly 50% being native indicates the regeneration potential for future restoration.In contrast, the extant vegetation had less than half of the species being wetland specialists (~ 45%) and only 36% of these being native.This potential for wetland specialists in our sites is contrary to other studies of degraded riparian zones that suggest reduced numbers of wetland species at sites with high grazing intensities (Eldridge and Lunt 2010;Dawson et al. 2017).
Most of the terrestrial species were non-natives from both the standing vegetation (63%) and the seed bank (79%), which is slightly higher than the proportions found from intermittent wetland seed banks that range from 60-70% terrestrial non-natives (Casanova and Brock 2000;Greet et al. 2013).

The challenge of invasive and non-natives species in the seed bank
With over 51% of the species in the seed bank classed as exotic or invasive, some authors suggest there is an inherent risk that invasive species will outcompete any native species in soil seed bank for re-naturalisation (James et al. 2007;Williams et al. 2008;Tererai and Wood 2014;Grewell et al. 2019).There is also the risk that the non-native species could out-compete native species with similar environmental requirements.For example, the non-native Isolepis prolifera* and the native I. inundata share similar zones in the channel and often occur together, however I.
prolifera* is typical of degraded wetland conditions (Sieben et al. 2017;Rebelo et al. 2018) and may act as an indicator of poor water quality.Of the nine invasive species recorded in the standing vegetation, three of these also germinated from the soil seed bank samples, plus the additional J. articulatus** that was only recorded from the seed bank.As typical invasive species, they are characterised as fast-growing to the flowering stage, have vegetative propagation and non-specialized pollination systems and germination requirements, and are prolific seeders with seeds spread by livestock (A.fissifolius**, C. clandestinus**, and S. madagascariensis**), wind (S. madagascariensis**), and water flow (A.fissifolius** and J. articulatus**) (Lake and Leishman 2004;Tropical Forages 2020;Lucidcentral.org 2022).Indeed, the ability to form a large seed bank is one of many traits that have contributed to the success of many invasive species (Pyšek and Richardson 2010).
The most dominant invasive species present in the standing vegetation sites were, unsurprisingly, the pasture grasses in the riparian zone, especially the densely stoloniferous C. clandestinus** (Pinto et al. 2021).Research by Bunn et al. (1998) suggest that riparian shading may be a highly effective means of controlling invasive grasses in disturbed streams and some pasture grasses (Norton et al.1990;Chauhan 2013;Casanova-Lugo et al. 2022).The literature suggests that while the pasture grass C. clandestinus is a highly tolerant of salinity, waterlogging, drought, frost, fire and day-length (Tainton 1998), a study in Australia has shown that they can still produce a maximum yield in 42% shade (Samarakoon et al. 1990).Similarly, other pasture grasses such as Axonopus sp.* (carpet grasses) and Stenotaphrum secundatum* (buffalo grass), can also thrive in shaded conditions up to 59% (Samarakoon et al. 1990).Therefore, we suggest multiple strategies to address this challenge, such a combination of livestock exclusion, active planting, and seed bank growth.Firstly, along with the exclusion of livestock, the active planting of trees and shrubs may not eliminate all pasture grasses from the riparian zone.Nevertheless, they could reduce pasture grass productivity and, thus density to allow the natural regeneration of native species.Secondly, the natural regeneration of the native rushes (especially J. usitatus and J. bufonius) and sedges, ferns, and forbs will assist in outcompeting the understory space to create a more diverse habitat, stabilise the sediment and contribute to the sediment composition and seed bank.Together, these approaches could limit the growing conditions for pasture grass germination.This approach of multiple strategies is supported by the literature where more recent attention has been given to how to utilise riparian seed banks better to support the rehabilitation of vegetation and riparian zone (Middleton 2003;Nishihiro et al. 2006;Boudell and Stromberg 2008;Jensen et al. 2008;Vosse et al. 2008;Meli et al. 2013;Hough-Snee et al. 2013).Other management strategies available might also include (but are not limited to): the removal of exotic species, the application of germination promoters such as smoke and related extracts, disturbance of topsoil, and the alteration of inundation or watering regimes (Roche et al. 2008;González et al. 2015).
There are a number of factors that may have affected the observed relationships between the standing vegetation and the seed banks that have not been accounted for.High numbers of non-native terrestrial species could be due to the drought conditions prior to February 2020, since non-native species are found to be more sensitive to increased inundation than native species in riparian zone (Tabacchi 1995).Also, while our seedling emergence technique is a popular and adequate method for assessing species richness and abundance in wetland seed banks (Brock et al. 1994), it is recognised that any method is unlikely to address all the germination requirements for all species.As such, the lower diversity in the seed bank compared to the standing vegetation was expected.Species that did not emerge from the trial cannot automatically be presumed as absent from the seedbank.Not only may sampling and germination methods may not have been appropriate for all species but it is unknown whether the lower species richness is due to other factors such as absence of a persistent SSB, low seed densities (Chong and Walker 2005;Capon and Brock 2006), seed removal and predation (especially by ants: e.g.Andersen and Ashton 1985;Reader 1993;Yates et al. 1995), or because some wetland emergent species, such as Typha and Schoenoplectus, characteristically reproduced clonally.
Riparian seed banks are derived from various sources, including seed rain from the surrounding environment and hydrochory (the transport and deposition of seeds by water) (Chambert and James 2009).This study found that areas with high diversity in extant vegetation and the soil seed bank, may require less active management once livestock are excluded.These sites could have the seed bank potential to renaturalise the riparian vegetation, although this may not be in the timeframe required by land managers.However, active planting of trees, especially native trees and shrubs, would be the appropriate management strategy for sites with low diversity in both the extant vegetation and the seed bank.
Across all sites, we found 18 native species from various growth forms (perennial forbs, rushes, ferns, sedges and grasses) germinating from the soil seed bank.With the combination of reliable rainfall since April 2020, the exclusion of livestock, the active planting of native shrubs and trees, and the viable native soil seed bank, we suggest that the successful restoration of the riparian vegetation of Victoria Creek is highly likely.At these sites, the high abundance of an early successional target species, such as Juncus usitatus and J. bufonius in the seed bank could act as an indicator of successful renaturalisation, when the standing vegetation has a reasonable diversity of native trees and shrubs.
Future seed bank studies in the catchment will establish whether the planted species have further contributed to the soil seed bank at the restoration sites, compared to new control and reference sites.

Conclusions
Soil seed banks reflect the past and extant vegetation, and in the riparian zones they also reflect the upstream diversity.Despite an extended history of livestock grazing, the ratios of nativeness (native/exotic) and growth form types (woody/grass-types/forbs) in the seed bank reflected the standing vegetation.Our study suggests that the natural regeneration of native grass-type species from the soil seed bank will complement restoration approaches such as livestock exclusion and plantings of tree and shrub species along the riparian corridor.
Active plantings will increase species richness and accelerate recovery trajectories at all restoration sites.However, prioritising planting at sites that are species-poor (especially of native trees and shrubs) could be an efficiency option.Our findings have important implications for land managers in not only their selection of appropriate species for restoration, but also to extend their funding resources further to more sites across the landscape.We therefore recommend, as part of the pre-restoration (Before) activities, that seed bank composition is conducted to better understand what native species are likely to naturally recruit following exclusion of livestock, and which non-native and invasive species might pose an issue.

Fig. 1 a
Fig. 1 a Location of Tilba Tilba Lake at south of Sydney, NSW, Australia.b Catchment boundary of Victoria Creek.c The four riparian vegetation survey sites (VC1, VC2, VC3, VCR1) along Victoria Creek.The relics of native vegetation communities are mapped as Plant Community Type (PCT) downloaded from the SEED data portal https:// datas ets.seed.nsw.gov.au/ datas et/ nsw-state-veget ation-type-map.Accessed May 2022

Fig. 2
Fig. 2 Experimental design for the soil seed bank germination experiment

Fig. 3
Fig. 3 Box plots show the median, quartiles, and ranges for species richness, native species, and non-native species at each site for a standing vegetation and b seed bank data.The letters above each box indicate the statistical difference between sites

Fig. 4
Fig. 4 Box plots show statistical differences between VC2 and other sites calculated for woody species for a standing vegetation and b seed bank data.'a' and 'b' inside the plots highlight Tukey's post-hoc test (p < 0.05) for pairwise comparisons fluctuation-responders-morphologically plastic; ATe, amphibious fluctuationtolerators-emergent; ATl, amphibious fluctuation-tolerators-low growing; ATw, amphibious