Fish and crustaceans provide early indicators of success in wetland restoration

Early indicators of restoration success can inform adaptive management strategies and maintain community interest and financial investment. Coastal wetland restoration projects prioritize monitoring the succession of habitat forming communities. However, these communities often expand slowly. In contrast, fish and crustaceans can quickly occupy newly available habitats and therefore, may be early indicators of restoration success. Here, we compared the short‐term responses of fish and crustacean communities and landcover area to restoration actions at a restoring wetland and three reference wetlands in the Maroochy River in eastern Australia. Fish and crustacean communities and landcover area were surveyed every 6 months between Spring 2017 and Spring 2019 (n = 5), with two sampling events conducted before restoration actions commenced in May 2018 and three after. Fish and crustaceans were surveyed using baited underwater video stations and fyke nets. Landcover was remotely sensed using a combination of Worldview‐2 satellite imagery, object‐based image analysis, and random forest classification. Fish and crustaceans make viable indicators for early restoration success, as our findings show the composition, diversity, and abundance of species targeted in fisheries of these communities can resemble assemblages at reference sites within 1 year of restoration actions. The area of tidal inundation significantly increased overtime, but the recovery of mangroves and saltmarsh communities did not. These findings support the notion that coastal wetland restoration can promote immediate recruitment by fish and crustacean communities, thus making these taxa useful indicators for communicating early success of restoration projects.

• The fast response of fish and crustaceans to restoration is likely due to numerous species having high mobility, generalist diets, broad habitat needs, and physiological tolerances allowing them to exploit the diversity of foraging and sheltering opportunities available within the restoring wetlands.

Introduction
Humans have fundamentally transformed most ecosystems globally, and this has led to widespread reductions in many of the services and functions they provide, including the provision of animal habitat, carbon storage, and nutrient cycling (Worm et al. 2006;Newbold et al. 2015;Miller & Hutchins 2017).Consequently, many degraded ecosystems are now the focus of intensive and expensive restoration projects which aim to reverse detrimental effects of human modification to flora and fauna in terrestrial, aquatic, and marine environments (Rey Benayas et al. 2009;Deal et al. 2012;Gerner et al. 2018).The success of ecosystem restoration projects is underpinned by the goals and objectives set, strategic planning of monitoring events, and the accuracy and frequency of assessments delivered for a variety of ecosystem services (Hobbs & Norton 1996;Loke et al. 2015;Gann et al. 2019).The scope of monitoring for many restoration initiatives is, however, often limited to describing outcomes for ecosystem components and processes that are often slow to recover, and instead are used as proxies for other potential success indicators (McCauley 2006;Bullock et al. 2011).For example, monitoring succession in focal vegetation communities is a primary objective for most restoration initiatives, often under the assumption that change in these indicators will deliver additional benefits for biodiversity and ecosystem health (Catterall 2018).However, this approach frequently fails to index responses of key fauna, ecological functions, or ecosystem services (Palmer et al. 1997(Palmer et al. , 2016;;Catterall 2018).This can be detrimental to the support, longevity, and success of some initiatives because vegetation is typically slow to respond to restoration, and this makes it difficult to maintain social interest and economic investment without other indicators of early restoration success (Field 1999;Clewell & Aronson 2006;Shoo et al. 2016).
Animals perform many important ecological functions that underpin the delivery of ecosystem services and help to support ecological resilience (Noss 1991;Lindenmayer et al. 2008;Nagelkerken et al. 2015).Animal assemblages are dynamic and can respond rapidly to both impacts from perturbations or recovery following management intervention.This can make them suitable as indicators for monitoring biodiversity and ecosystem health (Warren et al. 2001;Feary et al. 2007;Siddig et al. 2016).These potential rapid responses of animals to ecosystem changes are, however, often overlooked in restoration monitoring programs because of funding limitations, or uncertainty in indicator selection (McAlpine et al. 2016;Catterall 2018;Cross et al. 2019).Restoration projects that have monitored faunal recovery report a range of positive outcomes, including increases in biodiversity, rates of ecological functions (e.g.pollination, seed dispersal, herbivory, and predation), and the delivery of ecosystem services (e.g.carbon storage, nutrient cycling, fisheries catches) (Clewell & Aronson 2006;Hagger et al. 2017;Gilby et al. 2018).Quantifying these biodiversity and functional benefits can help to inform adaptive management decisions, and might be significant in maintaining community interest and financial investment, particularly when positive outcomes are communicated in the early phases of ecosystem recovery (Wortley et al. 2013;Barton & Moir 2015;Bayraktarov et al. 2020).
The construction of coastal cities, and the agricultural transformation of adjacent river catchments, has resulted in the extensive and widespread fragmentation of wetlands globally (Waycott et al. 2009;Waltham & Connolly 2011;Atwood et al. 2017).It is for this reason that many coastal wetlands are now the targets of concentrated restoration efforts which aim to promote the recovery of mangroves, saltmarshes, and seagrasses and their related ecosystem services (Lewis 2005;De Groot et al. 2013;Stewart-Sinclair et al. 2021).Coastal wetland restoration initiatives are becoming more widespread, but many still fall short of desired goals and objectives due to funding and design limitations (e.g.poor site selection, overlooking suitable success indicators), which can impinge on project scope, longevity, and community optics (De Groot et al. 2013;Bayraktarov et al. 2016;Duarte et al. 2020).Restoration can enhance biodiversity, and the nutrient buffering, erosion control, and carbon storage capacity of focal wetlands, but objectives for these aspirations are often poorly defined and performance is rarely tested with empirical data (Weinstein et al. 2001;Barbier 2013).The provision of nursery habitats for a diversity of juvenile marine organisms, particularly fish and crustaceans and supporting fisheries productivity, is perhaps the most widely appreciated ecosystem service derived from mangrove, saltmarsh, and seagrass ecosystems (Olds et al. 2013;Nagelkerken et al. 2015).It is, therefore, not surprising that most wetland restoration projects also seek to promote the diversity and abundance of species that underpin fisheries (Able et al. 2008;Boys & Pease 2017).Evidence for the ecological benefits of restoration for fish and crustaceans is plentiful and geographically widespread across most oceans and landmasses (Able et al. 2012;Boys & Williams 2012;Lechene et al. 2018).For example, restoration has promoted the diversity and abundance of fish and crustacean communities over small timescales (2-5 years) (Boys & Williams 2012;Moreno-Mateos et al. 2012;Kitchingman et al. 2023).These communities are rarely used as success indicators in wetland restoration projects and instead are overlooked in studies that focus on documenting changes to the area of target ecosystems (i.e.mangroves and saltmarshes) (Cadier et al. 2020).Fish and crustaceans can be quick to occupy recently inundated and newly available habitats in estuaries and might, therefore, be useful as indicators of early restoration success, particularly if the pace of their habitat use, diversification, and abundances exceeds that of focal vegetation communities.
This study examined the responses of estuarine flora and fauna to the initial phases (<2 years) of restoration at the Yandina Creek wetlands, a 200 ha restoring coastal wetland in eastern Australia (Fig. 1).Our objective was to contrast patterns in the timing of recovery between fish and crustacean assemblages with change in the mangrove and saltmarsh ecosystems that provide habitat for aquatic organisms, and which were also primary targets for restoration.The Yandina Creek wetlands was converted to agriculture land in the 1950s when tidal inundation was blocked and the site was planted with pasture grass and sugar cane (Brown et al. 2018;Iram et al. 2022).Restoration began at the site in autumn 2018 when tidal obstructions were removed to reconnect the site with the downstream waters of the Maroochy River, and promote passive recovery of wetland vegetation, biodiversity, and nutrient cycling (Iram et al. 2022).We tested the hypothesis that fish and crustaceans would respond rapidly to the reintroduction of tidal waters, and tested their utility as potential early indicators of restoration success by comparing changes in both floral and faunal assemblages across multiple reference wetlands.

Study Design
We adopted a biannual sampling approach to monitor early changes in the composition of fish and crustacean assemblages at the restoration site, and to contrast potential restoration effects with natural variation at three reference wetlands (Fig. 1).We surveyed these wetland sites (i.e. one restoration site and three control sites) which are located between 10 and 20 km upstream of the Maroochy River mouth.All chosen sites experience similar semi-diurnal tidal regimes (to a maximum tidal height of 1.4 m) and salinity levels (an annual range of 0-30 ppt) (Olds et al. 2018;EHMP 2022).The three reference wetlands also encompass a diversity of potential endpoints for restoration of the Yandina Creek wetlands.One site is a prescribed conservation area since 2004 (Coolum Creek), one is a local environment reserve (Twin Ridges Creek), and one abuts a small urban development (Rocky Creek), which therefore makes them a suitable suite of controls for indexing possible pathways toward restoration success.Surveys were conducted at the wetlands in spring and autumn each year from Spring 2017 to 2019 (n = 5).
Two events were completed prior to start of restoration (i.e.À6 and 0 months), and three were completed after restoration actions had commenced (i.e. 6, 12, and 18 months).The timing of all survey events was standardized to occur during monthly spring tides to maximize both the extent and duration of wetland inundation.On each event, we surveyed fish and crustaceans with baited cameras and nets, and measured the area and distribution of vegetation communities (i.e.mangrove, saltmarsh, mangrove fern, mixed she-oak and paperbark, and common reed) and the extent of tidal inundation at each wetland location.We sampled fish and crustacean assemblages using both baited remote underwater video stations (BRUVS) and fyke nets at the restoration site and three reference wetlands.These are standard techniques for the survey of estuarine fish and crustaceans and their use in combination provides a comprehensive assessment of assemblages in a diversity of wetland habitats (Weaver et al. 1993;Hollingsworth & Connolly 2006;Gladstone 2012).
On each sampling event, BRUVS were deployed at 16 sites in the Yandina Creek wetlands, and at 10 sites in each reference wetland (n = 46 per event).BRUVS were comprised of a high-definition GoPro Hero-5 camera and filmed at 1,080 pixels, 30 frames per second, and with a wide field of view (14 mm) and  then mounted on a 5 kg weight, with a bait bag filled with 500 g of pilchards (Sardinops sagax) that was held 0.5 m in front of the camera on a PVC pipe (Olds et al. 2018;Borland et al. 2022a).All BRUVS deployments lasted for 1 hour and were completed within 2 hours either side of spring high tides to standardize for potential effects of tidal variation.BRUVS were positioned in mangrove aligned subtidal creeks at approximately 2 m deep to record the diversity and abundance of larger fish and crustacean assemblages where fyke nets are not appropriate (Fig. 1).Fish and crustacean assemblage composition, species richness, and the abundance of harvestable species from video footage were quantified to the bait bag to standardize for water clarity using the standard MaxN statistic for all species recorded on each deployment (Murphy & Jenkins 2010).
On each sampling event, fyke nets were deployed at five sites in the Yandina Creek wetlands, and at up to five sites in each reference wetland (n = 15 per event).All fyke nets had a 6 mm mesh fixed to four (0.7 Â 0.7 m) frames and two (3 Â 0.7 m) wings (Faunce & Serafy 2006).Fyke nets were deployed at the mouth of intertidal creeks, on the spring high tide, at approximately 0.5 m water depth, with the opening facing upstream to target small and juvenile fish and crustacean assemblages.All captured fish and crustaceans were identified, counted, and released (Olds et al. 2012).Anything that was difficult to identify was photographed and released.

Quantifying Temporal and Spatial Variation of Estuarine Land Cover
We used a combination of remote sensing processes to describe changes in the area and distribution of estuarine vegetation communities and the extent of tidal inundation at the restoration site and three reference wetlands (Fig. 2) Remote sensing approaches are commonly used methods to detect large-scale shifts in the coverage and distribution of coastal and particularly blue carbon ecosystems (Pham et al. 2019(Pham et al. , 2019)).We followed the remote sensing approach by Rummell et al. (2022), which is designed to track the response of target ecosystems to wetland restoration.Briefly, this approach used a combination of multispectral and high spatial resolution Worldview-2 imagery, an object-based image analysis and random forest algorithm to classify estuarine land cover including, mangroves, saltmarsh, mangrove fern (Acrostichum speciosum), common reed (Phragmites australis), and water bodies.These classes encompassed target ecosystems and areas of open water, at the restoration site and control wetlands, and variations in their distribution and area, which are often associated indicators of restoration success and with changes in the habitat function of wetlands for estuarine fish and crustaceans (Olds et al. 2012;Zhao et al. 2016;Gilby et al. 2021).These maps were found to be highly accurate consistently achieving overall accuracies between 90 and 95% accuracy and were generated in QGIS after random forest classification via confusion matrices (Rummell et al. 2022).

Data Analysis
Permutational multivariate analysis of variance (PERMANOVA) was used to test for potential early effects of wetland restoration on the composition of fish and crustacean assemblages (Anderson 2001).This analysis included interactions between the factors site (fixed factor, four levels) and time (fixed factor, five levels: À6, 0, 6, 12, and 18 months).Significant effects identified by pairwise tests following PERMANOVA were displayed using centroids on nonmetric multidimensional scaling (NMDS) ordinations (Clarke et al. 2008).PERMANOVA and NMDS analyses were calculated on square-root-transformed data and Bray-Curtis dissimilarity measures, which emphasize differences in both species abundance and species diversity (Anderson et al. 2011).All multivariate analyses were conducted using PRIMER with the PERMANOVA+ addon (Anderson et al. 2008).
Generalized linear models (GLMs) in the R statistical framework (R Core Team 2022) were then used to test for potential effects of wetland restoration on: (1) fish and crustacean diversity (species richness); (2) the total abundance of fisheries target species (harvestable abundance) (Froese & Pauly 2018); (3) the abundance of each fish and crustacean species; and (4) the area of vegetation communities (i.e.mangroves, saltmarsh, mangrove fern, mixed she-oak and paperbark, and common reed) and the extent of tidal inundation.GLMs used the same model structure as assemblage-level analyses (i.e.site and time), except time was included as a continuous variable.All models were checked for the distribution of residuals, homogeneity of variance, and overdispersion, and were fitted with either Poisson, negative binomial (if overdispersion was present) or Gaussian distributions with log link functions using the MASS package (Ripley et al. 2013).Significant interactions between site and time were then plotted using ggplot2 and ggpubr (Wickham 2011; Kassambara & Kassambara 2020).

Results
The restoration site (Yandina Creek wetlands) initially supported a depauperate assemblage of fish and crustaceans, which differed from those recorded at the three reference wetlands before restoration actions commenced (i.e. at times À6 and 0 months) (Tables 1 & S1; Fig. 3).These differences were no longer as apparent after restoration begun, with assemblages at the restoration site (Yandina Creek wetlands) resembling those at one reference location (Rocky Creek wetlands) after 12 months of monitoring (Table S1; Fig. 3).Fish and crustacean assemblages at the restoration site were not distinct from those at the three reference locations (i.e.Rocky, Coolum, and Twin Ridges Creek wetlands) after 18 months of monitoring (Fig. 3).This positive effect of restoration on fish and crustacean assemblages was present in data collected from both BRUVS and fyke net surveys (Fig. 3).
There were no significant temporal changes in the area of mangroves ( p = 0.541), saltmarsh ( p = 0.851), mangrove fern ( p = 0.057), or mixed she-oak and paperback communities ( p = 0.21) between all sites and over time (Table S4; Fig. 6).There were significant temporal changes in the area of common reed communities ( p ≤ 0.001) and open water ( p ≤ 0.001) between all sites and over time (Table S4).
Summary statistics of fish and crustacean surveys and also areas estimates across the restoration and all reference sites are available in supplementary materials (Tables S5 & S6).

Discussion
Many restoration projects focus monitoring on changes in vegetation communities toward target ecosystem conditions.However, assemblages of habitat-forming species are often slow to recover during restoration, and this can limit the likelihood of detecting successes in the early phases of interventions (McAlpine et al. 2016;Cross et al. 2019).Our findings show that fish and crustaceans can respond rapidly to the reconnection of degraded, isolated wetlands with downstream coastal waters.Here, assemblages changed to the point of being indistinct from three natural reference wetlands within 1 year of restoration commencing.The occupancy and diversification of fauna tracked the rate and extent of tidal inundation and the reduction in common reed communities but occurred before significant recovery of focal mangrove and saltmarsh communities that provide nursery habitats for many of these species.The early responses of multiple taxa that are targeted in local fisheries (e.g.yellowfin bream, bay prawns) is particularly encouraging because supporting fisheries, and enhancing fish catches, are common goals in wetland restoration, but are still rarely measured with empirical data in restoration monitoring programs (Gilby et al. 2018).Identifying and reporting ecosystem service benefits beyond the recovery of habitat-forming species can be critical for maintaining social enthusiasm and long-term economic support for restoration (Bullock et al. 2011;Taylor et al. 2018;Gann et al. 2019).Our results highlight the role that monitoring of biodiversity and fisheries species might have in providing early indicators of restoration success.
The high mobility, generalist diets, broad habitat needs, and wide physiological tolerances of many estuarine fish and crustaceans shape their ability to rapidly occupy new space, habitats, and ecological niches, which are a common characteristic of restoring coastal wetlands (Peterson & Bayley 1993;Henderson et al. 2020).The reinstatement of tidal flows provides a diversity of feeding and sheltering opportunities for opportunistic species which use restoring wetlands as intermittently inundated habitats or occupy newly connected drainage channels and pools throughout the tidal cycle (Krzton-Presson et al. 2018;Bennett et al. 2020).Faunal diversity is typically low in the early phases of wetland restoration, and expanding estuarine food webs are comprised of species that are able to tolerate high variation in salinity, while also deriving their nutrition from flows of detritus and carrion that originate on land, or from decaying vegetation Fauna responses indicate early restoration success and animal carcasses which have succumbed to tidal inundation (Kwak & Zedler 1997;Matthews & Endress 2010;Vincent et al. 2015).The establishment and subsequent expansion of estuarine habitat forming species like mangroves and saltmarshes can significantly enhance the habitat functions of restoring wetlands for estuarine communities by diversifying the range of carbon sources, and the heterogeneity of physical habitat structure (Hollingsworth & Connolly 2006;Valentine-Rose & Layman 2011).Mangroves and some saltmarsh species, however, can be slow growing (e.g.15-25 years to maturity for many mangrove species) and their colonization of new locations is often constrained by the effects of tides, currents, and rainfall which shape the recruitment of propagules to new restoration sites (Lugo & Snedaker 1974;Zedler et al. 2010;Van Wijnen et al. 2014).Our findings show that the occupancy and diversification of estuarine fauna can occur quickly, and before substantial changes in the cover and condition of target wetland ecosystems.This is likely an important indicator that restoration is progressing in an upward trajectory and toward project goals and objectives (Gann et al. 2019;Gilby et al. 2019;Stewart-Sinclair et al. 2021).
The early occupancy and diversification of fish and crustaceans at the restoration site was largely driven by the recruitment of small fish and crustaceans and the juveniles of several species that are targeted in local fisheries.Four species, including yellowfin bream, estuary perchlet, bay prawn, and flathead gudgeon were indicators of the positive trajectory in biodiversity and fisheries benefits at the restoration site.These species are common in local estuaries, where they use inundated wetland ecosystems as feeding sites, sheltering areas, or juvenile nursery habitats (Meynecke et al. 2007;Davis et al. 2014;Brook et al. 2018), and have been reported to respond positively to restoration elsewhere (Beesley et al. 2012;Boys et al. 2012).They encompass a range of functional groups, including detritivores, zooplanktivores, and omnivores, and occupy a diversity of estuarine habitats in natural, urban, and restoring environments (Curley et al. 2013;Whitfield 2017;Henderson et al. 2020).Additionally, both yellowfin bream and bay prawn are highly sought after targets in local commercial and recreational fisheries (Courtney & Masel 1997;Lowry et al. 2017).These findings show that wetland restoration is having a positive effect on the composition and diversity of local food webs, with potential to benefit several local fisheries.These benefits have occurred within only 18 months of restoration actions commencing, and in the absence of substantial changes to the area of mangrove and saltmarsh habitats.As restoration progresses, we expect the extent and frequency of tidal inundation to increase alongside the colonization and expansion of fish habitats.Therefore, secondary and tertiary changes to the composition, abundance and diversity of fish and crustaceans may occur in the future (Boys & Williams 2012;Boys & Pease 2017).
The response of target ecosystems (i.e.mangroves and saltmarsh) to restoration initiatives are commonly measured through cheap and effective long-term indicators for the regeneration of ecosystem services (e.g.carbon storage and nutrient cycling) and ecosystem functions (e.g.food web dynamics and nesting locations) (Zhao et al. 2016;Pham et al. 2019;Billah et al. 2022).Tidal reinstatement increased the area of saltmarsh and water, but saltmarsh communities experienced high variability prior to restoration actions and then stabilized afterward.This initial variability relates to the attributes and composition of saltmarsh communities at the site which were comprised of many small remnant patches and a mix of salt-tolerant and intolerant species that expanded and senesced with shifts in local growing conditions (i.e.temperature, season, salinity, rainfall) (Baldwin et al. 2001;Palinkas & Engelhardt 2016;Grieger et al. 2022).Tidal reinstatement stabilized the extent of saltmarsh communities by reducing the volatility of growing conditions and the diversity of these communities to feature slow growing salt-tolerant species such as saltwater couch (Sporbolus virginicus) (Laegdsgaard 2002;Winning & MacFarlane 2010;Rankin et al. 2022).Increases to water cover were clearly linked to the reintroduction of tidal flows, but interestingly continued to expand overtime.This is likely due to a combination of tidal waters pushing further into the upstream reaches of the site and also remaining in low lying intertidal areas for longer in each tidal cycle and tidal flows incrementally eroding exposed soils and saltwater intrusion over time causing the dieback of salt intolerant communities such as common reed (Karberg et al. 2018).The stabilization of saltmarsh communities and the reinstatement of tidal flooding can produce significant restoration benefits by limiting greenhouse gas production (i.e.CO 2 and CH 4 ) and improving access to an important basal food sources that structure food webs in estuarine seascape (Poffenbarger et al. 2011;Kroeger et al. 2017;Whitfield 2017).Therefore, measuring the response of target ecosystems to restoration actions can also provide stakeholders and managers with positive outcomes early in the restoration trajectory despite little expansion to target ecosystems communities.We expect as tidal influences continue to progress and mangrove and saltmarsh communities recruit, the area of target communities further expand.
We successfully detected changes to the composition, species richness and harvestable abundance of fish and crustacean assemblages with both BRUVS and fyke nets.This is a promising result because both sampling gears are commonly used in wetland research to monitor different components of the fish and crustacean assemblages (Hollingsworth & Connolly 2006;Olds et al. 2012;Borland et al. 2022b).For example, BRUVS target large and often less abundant fish and crustacean communities in deeper subtidal areas and fyke nets target small and juvenile fish and crustacean assemblages in shallow intertidal areas (Weaver et al. 1993;Gladstone 2012).BRUVS, however, were not effective at measuring changes to the abundance of individual fish and crustacean species and this is likely because large fish and crustaceans are less abundant in the early stages of wetland restoration projects (Boys & Pease 2017).By contrast, fyke nets were more effective for detecting early shifts to small and juvenile fish and crustacean assemblages which are typically highly abundant in the early stages of wetland restoration projects (Kowalski et al. 2014;Lechêne et al. 2018;Baker et al. 2020).However, we experienced high variability in individual species abundance which contributed to large confidence intervals and may become problematic for identifying trends over large temporal and spatial scales.We recommend the combined use of both BRUVS and fyke nets if fish and crustaceans are incorporated as a success indicator in future coastal wetland restoration projects.
Our findings show that coastal fish and crustacean assemblages can respond quickly to wetland restoration initiatives, with recovery tracking improvements to tidal connectivity and the extent of wetland inundation.Increase in biodiversity and the abundance of fisheries species were significant and occurred rapidly (2-5 times over 1 year) and before any discernible expansion in the area of fish nursery habitats, such as mangroves and saltmarshes.This is a significant outcome for local management efforts, and for restoration planning decisions more broadly because many monitoring programs neglect to measure whether, and how, changes in faunal populations are linked to the recovery of restoring habitats, despite this being a central goal for many interventions.Wetland restoration can also quickly increase the area of water and stabilize the area of saltmarsh communities, which can have significant implications for limiting greenhouse gas emissions.Our results suggest that indexing early responses of fish and crustaceans in addition to the coverage and condition of some target ecosystems might help to demonstrate a diversity of complimentary cobenefits from wetland restoration, and this is often essential for maintaining long-term community engagement and financial support for restoration.

Figure 1 .
Figure 1.Distribution of sampling locations at the restoration site (Yandina Creek wetlands) and the three control sites in the Maroochy River catchment in eastern Australia.

Figure 2 .
Figure 2. Nonmetric multidimensional scaling ordinations (NMDS) displaying relationships among fish and crustacean assemblages sampled with baited remote underwater video stations (BRUVS) (A), and fyke nets (B).Each point is a centroid representing mean values from each location at each point in time.Ellipses visualize statistically similar groupings of wetland sites at each point in time (e.g.assemblages from Yandina Creek wetlands differed from those at all reference sites at À6 months, but not after 0 months).Arrows indicate the trajectory of changes in assemblage composition over time.

Figure 3 .
Figure3.Generalized linear models (GLMs) visualizing the effects of time (AE95% CI) on species richness and the abundance of harvestable fish and crustacean species at all sites, in samples collected with baited remote video stations (BRUVS) and fyke nets.

Figure 4 .
Figure 4. Generalized linear models (GLMs) visualizing the effects of time (AE95% CI) on the abundance of yellowfin bream, estuary perchlet, bay prawn, and flathead gudgeon at all sites in samples collected with fyke nets.

Figure 5 .
Figure 5. Changes in the area and composition of vegetation communities, and the extent of tidal inundation, at the restoration site (Yandina Creek wetlands) and control sites (Coolum Creek, Rocky Creek, and Twin Ridges Creek wetlands) between Autumn 2017 (À6 months) and Spring 2019 (18 months).

Figure 6 .
Figure 6.Generalized linear models (GLMs) visualizing the effects of time (AE95% CI) on land-cover classes including water, common reed, mangroves, saltmarsh, and mangrove fern at all sites.

Table 1 .
Summary of the permutational analysis of variance (PERMANOVA) testing for changes in the composition of fish and crustacean assemblage over time at all wetland sites.