Succession of fish and crustacean assemblages following reinstatement of tidal flow in a temperate coastal wetland

Ecological Engineering 49 (2012) 221–232

Contents lists available at SciVerse ScienceDirect

Ecological Engineering journal homepage: www.elsevier.com/locate/ecoleng

Succession of fish and crustacean assemblages following reinstatement of tidal flow in a temperate coastal wetland Craig A. Boys a,∗ , Robert J. Williams b a b

New South Wales Department of Primary Industries, Port Stephens Fisheries Institute, Locked Bag 1, Nelson Bay, NSW 2315, Australia New South Wales Department of Primary Industries, Cronulla Fisheries Institute, P.O. Box 21, Cronulla, NSW 2230, Australia

a r t i c l e

i n f o

Article history: Received 31 January 2012 Received in revised form 5 July 2012 Accepted 10 August 2012 Available online 4 September 2012 Keywords: Tidal restriction Culverts Estuarine rehabilitation

a b s t r a c t Fish and decapod crustacean assemblages were sampled from two manipulated creeks in which tidal flow had been increased through culvert removal. Assemblages were compared to those of two control creeks where culverts remained and two reference creeks without culverts. The presence of culverts reduced the richness and abundance of estuarine–marine dwelling species. Successional changes over a 16 year period (2 years prior to culvert removal and 14 years after) showed an immediate response at one manipulated creek but many years to attain reference condition. The other manipulated creek showed a completely different trajectory: oscillation between control and reference status with no clear intermediary condition. We concur with other investigators that short-term studies may give only a partial indication of response to rehabilitation efforts in coastal wetlands and longer-term studies (5–10 years) are recommended. There is some evidence that over longer-time frames, the presence of distinct groups of fish and/or decapods may indicate stages in the maturation of a rehabilitated wetland. © 2012 Elsevier B.V. All rights reserved.

1. Introduction An underlying principle of rehabilitation ecology is that the removal of stressors can reinstate ecological processes, thereby moving a degraded system towards a more natural or unstressed state (Simenstad et al., 2006). This progression of ecological recovery through time has been described as a pathway (trajectory) towards equivalency in structure or function to some reference or desirable state (Kentula et al., 1992; Hobbs and Norton, 1996; Simenstad and Thom, 1996). Timeframe, pathway and endpoint need to be considered when determining whether rehabilitation goals are being met and sustained (Simenstad et al., 2006). Coastal wetlands provide important ecosystem functions, many of which have been significantly compromised by human populations (Vitousek et al., 1997; Gedan et al., 2009). The restriction of tidal flushing onto marsh surfaces and intersecting creeks by roads, pipelines, floodgates and culverts has negatively impacted wetland productivity and detrimentally impacted bird, fish and plant assemblages (Roman et al., 1984; Raposa and Roman, 2001; Howe et al., 2009). Accordingly, there has been increasing effort in recent decades to ‘passively rehabilitate’ (sensu Simenstad et al., 2006) coastal wetlands by removing tidally restrictive structures (e.g. Streever, 1998; Teal and Weishar, 2005; Gedan et al., 2009).

∗ Corresponding author. Tel.: +61 2 4916 3851; fax: +61 2 4982 2265. E-mail address: [email protected] (C.A. Boys). 0925-8574/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ecoleng.2012.08.006

The fundamental premise of this passive approach is that through time, the alleviation of a key stressor such as a tidal restriction will reinstate natural hydraulic and ecological processes which in turn rehabilitate ecosystem structure and function. The extensive application of coastal wetland rehabilitation has been paralleled by growing literature on ecosystem responses to these works, typically involving pair-wise comparisons between manipulated and nearby natural reference systems (Zedler and Callaway, 2009). Such studies have been met with varying results, with some reporting improvements in ecosystem structure over relatively rapid time periods (Raposa, 2008; Thelen and Thiet, 2009; Boys et al., 2012), whilst others studies have failed to detect a response (Raposa and Roman, 2003; Boys et al., 2012). Response variability has been in part attributed to site specific differences in habitat and disturbance, and it has also been suggested that short term studies (2–3 years) are not long enough to determine whether an ecosystem (or part thereof) has reached maturity and/or equivalency with the natural reference system (Zedler and Callaway, 2000). The lack of long-term datasets in the literature hampers the ability of wetland managers to assess the success of rehabilitation efforts (Simenstad and Thom, 1996) and is particularly problematic when policies require rehabilitation targets be set on how closely a site may be expected to attain a reference state and how long this may take to happen (Zedler and Callaway, 1999). Greater insight into rehabilitation responses can be gained by longer-term studies, particularly in terms of ecological succession (e.g., Zedler and Callaway, 1999, 2000, 2009; Walker et al.,

222

C.A. Boys, R.J. Williams / Ecological Engineering 49 (2012) 221–232

Fig. 1. Location of the study area on Kooragang Island in the lower Hunter River.

2007). Several such rehabilitation studies have been conducted in North American coastal wetlands in conjunction with passive activities such as removing tidal restrictions to reinstate tidal flushing (Simenstad and Thom, 1996; Zedler and Callaway, 1999; Able et al., 2008) or more ‘active’ approaches including wetland construction (Morgan and Short, 2002). Even in long-term studies, responses have been variable and often dependent on what components of the ecosystem are monitored (Moreno-Mateos et al., 2012). Whilst changes in some structural attributes, such as the assemblage of resident and transient fauna (e.g. fish), can be both rapid (often immediate) and sustained (Simenstad and Thom, 1996; Able et al., 2008), other ecosystem functions such as primary production, organic matter accumulation and sediment transport can be significantly slower to respond (e.g. 5–12 years) (Morgan and Short, 2002; Craft et al., 2003), or lack a consistent directional response towards a reference state throughout the entire study (Zedler and Callaway, 1999). While all these studies have been long-term (7–12 years) and well-designed in respect to having comparable natural reference sites, they have lacked comparison to control sites where rehabilitation activities had not been carried out, thus making it impossible to rule out change could have occurred irrespective of the rehabilitation actions (Underwood, 1990). Furthermore, these studies have been concentrated around a small number of estuaries in North America, and given that coastal wetland degradation and rehabilitation is a global issue (Lotze et al., 2006), the generality of the findings should be tested on assemblages and ecosystems in other parts of the world. In this study we initially evaluated differences in constricted and unconstricted tidal creeks and then examined successional changes in fish and decapod assemblages in tidal creeks of a degraded wetland in south-eastern Australia, to determine if assemblage structure followed a clear trajectory of recovery once tidally restrictive culverts were removed. We evaluated one of the longest datasets (16 years) published to date on wetland rehabilitation, utilising a robust design that compared manipulated creeks to both un-manipulated control creeks and natural reference creeks. Of particular interest was how long it took assemblages to respond

following culvert removal, whether assemblages reached structural equivalency with reference creeks, whether the responses were sustained, whether responses varied across creeks, and what species were responsible for change at different times.

2. Methods 2.1. Study area The Kooragang Wetland Rehabilitation Project was initiated in the early 1990s (Streever, 1998) initially to improve degraded fish habitat but over time facilitated other studies on nutrient cycling (Howe et al., 2009), vegetation change (Streever and Genders, 1997; Howe et al., 2010) and shore bird roosting (Kingsford et al., 1998). The study area, Kooragang Island (32◦ 51 52 S, 151◦ 42 15 E), is a 26 km2 tidal wetland situated in a mature barrier estuary (Roy et al., 2001) ∼10 km upstream of the mouth of the Hunter River, New South Wales (NSW) Australia (Fig. 1). The river’s catchment is the third largest in NSW, draining an area in excess of 22,000 km2 (Williams et al., 2000) Tidal flow across Kooragang Island is hydrologically complex with numerous creeks feeding mangrove and at higher elevations salt marsh. Many decades ago farmers placed culverts at the junctions of the island’s smaller tidal creeks to enhance access to grazing land. A combination of small diameter culverts (400 mm or 600 mm) at a high invert level severely restricted upstream flow even though spring tide range in the Hunter River at creek entrances is of the order of 1200 mm. Five of the six creeks chosen for study are situated on the South Arm of the Hunter River. The two manipulated locations, Fish Fry Creek and Crabhole Creek, are ∼10 and 14 km from the mouth of the Hunter River, respectively, and two of the control creeks and one of the reference creeks are also along this stretch of river (Fig. 2). As no other suitable reference creek was located on the South Arm, the sixth creek was chosen on the North Arm. At spring tide these creeks ranged in width from two to five meters and in

C.A. Boys, R.J. Williams / Ecological Engineering 49 (2012) 221–232

223

creeks and for 8 out of 14 years following culvert removal (see Table S2 and S3 in the supplementary material for further details on the spatial and temporal designs). 2.3. Fish and crustacean sampling

Fig. 2. Control (C; tidally restricted throughout study), reference (R; unrestricted throughout study) and manipulated (M; tidal restriction removed throughout study) creeks in Kooragang Island. CHC (M) = Crabhole Creek, CC (R) = Cobbans Creek, MC (R) = Mosquito Creek, DC (C) = Dead Mangrove Creek, WC (C) = Wader Creek, FFC (M) = Fish Fry Creek.

depth from 400 mm to 1500 mm. After the second summer of sampling, culverts at the manipulated creeks were replaced with 7 m free standing bridges. Key habitat characteristics (vegetation, channel cross section and salinity) are given for all creeks in the before and after phases in Table S1, along with the dimensions of culverts. Tidal range within the control creeks was of the order of 300–400 mm due to culvert diameter and high invert level, and the latter also produced a pronounced tidal lag. In the reference creeds tide range was 1200–1500 mm. The mangrove Avicennia marina was found in all creeks. Mangrove trunks and pneumatophores were readily avoided during net hauls in the reference creeks but care was needed when randomising net sets around these natural obstructions in the control and manipulated creeks. After the culverts were removed the density of mangrove upstream of the sampling sites in Fish Fry Creek increased considerably (Howe et al., 2010 and Table S1). There was only limited increase in mangrove pneumatophore at Crabhole Creek. The cross-sectional dimensions of control and reference creeks did not change throughout the study, but channel scour and creek deepening and widening occurred in the two manipulated creeks following culvert removal (Table S1), with the response being most significant in Fish Fry Creek, resulting in the collapse of a bridge in 2010. 2.2. Experimental design In the first phase of the study we examined assemblages of small fishes and decapods at six creeks: four creeks that were tidally restricted by culverts and two creeks not under the influence of culverts. In the second and major phases of study we compared the response at two of the four restricted creeks when their culverts were removed to response of the remaining restricted creeks throughout the entire study (controls) as well as to the two natural creeks (references) (Fig. 2). The hypothesis tested was that if responses were to be attributed to culvert removal, manipulated creeks would undergo a trajectory of change to become similar to the reference creeks, without similar changes occurring in the control creeks (Grayson et al., 1999). All creeks were surveyed repeatedly during 10 austral summers between 1993/94 and 2008/09. Each creek was sampled randomly on spring tides three times between December and February for 2 consecutive years prior to removal of culverts from manipulated

Fish and decapod crustaceans were collected during spring tide in daylight hours from late spring to summer to coincide with the summer recruitment period. All sites within a sampling month were visited within one week. Tidal creeks were sampled at three replicate sites using seine net hauls (10 m headline × 1.5 m drop × 3 mm stretch mesh), performed in a ‘U’-shape and pursed onto the shore. These sites were spaced no closer than ∼20 m apart. A separate study utilizing up to eight seine haul sites was performed on various occasions and this verified that three hauls were adequate in capturing the vast majority of species present, with species seldom added with subsequent hauling (R.J. Williams, unpublished data). Other studies have also found that three seine haul sites in a locality captured 86% of species present (Kroon and Ansell, 2006). Seine netting was commenced during daylight hours shortly before and after spring high tide to coincide with maximum depth and minimum velocity. Seine netting was known to be biased toward the capture of small-bodied species or juveniles of larger commercially important species. To in part overcome this bias, microfilament floating gill nets of various mesh sizes (two panels 10 m in length with a 2 m drop, each of 25, 50, 75 and 100 mm mesh) were also used to target larger fish in Fish Fry Creek (it was too shallow to use gill nets in Crabhole Creek). Panels were set haphazardly from alternate banks with the upstream order of mesh sizes randomised, and at 45◦ angles across the channel into the incoming tide with one end secured to the bank and the other weighted at mid channel. Gill nets were set approximately 1 h before the turn of high tide and allowed to soak for approximately 2 h. Fish and decapod crustaceans were placed in buckets of estuarine water prior to identification to species level; fork length or as necessary total length was recorded for fish and carapace length for decapods. After measurement, all live specimens were returned to the water. 2.4. Statistical analysis Analyses of assemblage change were performed on zeroadjusted Bray–Curtis similarity matrices (Clarke et al., 2006) derived from fourth-root transformed abundance data. The fourthroot transformation reduces the ‘swamping effect’ of a few very abundant species on the composition of an assemblage (Clarke and Green, 1988). Zero-adjusting with a ‘dummy value’ did not affect the normal functioning of the Bray–Curtis coefficient, but ensured that samples with denuded assemblages could be incorporated into analyses to generate meaningful ordinations which would otherwise ‘collapse’ (Clarke et al., 2006). Fish and decapod data obtained by seine netting were analysed as the same assemblage. Because gill nets targeted only a few large-bodied species and were inconsistently applied between manipulated sites, gill net data were excluded from the multivariate analyses, and instead, species caught in large enough numbers were analysed separately using univariate calculations of means for each creek or treatment at each year of sampling. For the multivariate analysis, non-metric multidimensional scaling (nMDS; Kruskal and Wish, 1978) ordinations were created from the similarity matrices. These ordinations were based on the centroids for each manipulated creek and centroids of combined control and combined reference creeks for each year of sampling. That is, we incorporated the three site and three annual survey data

224

C.A. Boys, R.J. Williams / Ecological Engineering 49 (2012) 221–232

sets for each manipulated creek into a single Creek × Time centroid, whereas for reference and control creeks we incorporated the three site/three annual survey data from both creeks into a single Treatment x Time centroid (reference and control). Prior to combining the two reference creeks and the two control creeks (by centroiding), an ordination was created separately plotting all reference and control creeks (see Supplementary material Fig. S1). Examination of this ordination confirmed that the variability between the individual reference and control creeks was sufficiently low to allow further simplification of the ordinations by creating Treatment × Time centroids without compromising the interpretability of the data. Centroiding the data in this way greatly reduced the number of data points on the ordination and enabled better interpretation of trajectories of assemblage change in manipulated creeks against the ‘average’ response of both control and reference creeks. To further aid interpretation, each manipulated creek was presented as a separate ordination along with the control and reference creeks. Simple agglomerative hierarchical clustering and similarity profiles (SIMPROF; Clarke et al., 2008) were performed on each of the ordinated matrices to identify statistically significant groupings and changes in assemblage composition over time. The groupings generated have been numbered according to decreasing similarity in which they split from the previous group and do not reflect chronological order. Of particular interest within the context of our hypothesis was whether assemblages in manipulated creeks became similar to those of reference creeks at any time, whilst control locations persisted with a unique assemblage. These assemblage groupings were compared with Similarity Percentages analyses (SIMPER; Clarke, 1993) to identify which species contributed most to between group dissimilarity, thus identifying the species driving successional changes within creeks. Only those species exceeding an arbitrary threshold value of dissimilarity of 3% were interpreted (Terlizzi et al., 2005). Finally, mean species richness was calculated corresponding to each centroid within each of the assemblage groupings. This was done on both the whole assemblage and the following salinity guilds: Estuarine–marine (saltwater species that are primarily estuarine–marine dwelling as adults) and Freshwater-estuarine (euryhaline species equally well adapted to saline or freshwater habitats as adults) (after Pollard and Hannan (1994)). Freshwater species (those dwelling entirely in freshwater) were not analysed as they were too few (one species) to be of use. 3. Results 3.1. Catch summary A total of 900 seine and gill net samples were collected across all six creeks over the 16 years that the study took place. These samples netted 212,744 fish and 86,720 decapods (49 and 21 species, respectively) (Table 1 and Tables S4 and S5 in supplementary material). 3.2. Response at Fish Fry Creek In years one and two, when culverts were in place at Fish Fry Creek the assemblage was similar to that of the other creeks with culverts (control sites), but differed significantly from reference (unculverted) creeks (Fig. 3a, Table 2: Group II similarity 40.79, p = 0.001). Estuarine–marine (E–M) species typified the difference, as more of these species were found in reference creeks (Fig. 4: Group II versus V) and when present were typically more abundant (Table 3 : Group II versus V). In particular, species such as glass goby, Port Jackson glassfish, flat-tail mullet, yellow-finned

Fig. 3. nMDS of centroids for (a) Fish Fry Creek and (b) Crabhole Creek and the combined reference and control creeks at each year of sampling showing the trajectories in fish and decapod assemblage change (based on seine net data only). Statistically significant assemblage groupings are shown in Roman numerals, with order relating to decreasing similarity of grouping splits (see SIMPROF, Table 2). Numbers correspond to years (1–2 pre-culvert removal and 3–16 post-culvert removal).

bream, largemouth goby, pink shrimp, school prawn (all E–M), striped shrimp and Tamar River goby (both F–E) were consistently more abundant in reference creeks when compared to the control creeks and Fish Fry Creek prior to culvert removal. In contrast, the control (culverted) creeks as well as Fish Fry Creek prior to culvert removal contained greater abundances of mangrove goby, bluespot goby, half-bridled goby, bridled goby and checkered mangrove goby (all E–M) than the reference creeks (Table 3: Group II versus V). Fish Fry Creek and the reference and control creeks showed a change in fish and decapod assemblages through time (Fig. 3a). For the first 10 years of the study, assemblages at the control and reference creeks remained distinct and relatively unchanged when compared to the significant shift that occurred at Fish Fry Creek immediately following culvert removal (Fig. 3a: after year 2). The change at the latter was sustained for at least the next 5 years (Fig. 3a: Fish Fry Creek years 3–8, Table 2: Group III similarity 37.02, p = 0.001), demonstrating a clear and sustained response to culvert removal. Demersal goby species (mangrove, blue-spot, half-bridled, bridled and checkered mangrove) that had characterised Fish Fry Creek in its culverted condition (and control creeks) became immediately less abundant and were replaced by a reciprocal increase in abundance of glass goby, flat-tail mullet, fan-tail mullet, sandy sprat, pink shrimp (all E–M), Tamar River goby and sea mullet (both F–E) (Table 3: II versus III). Whilst these reciprocal changes in species abundance and species replacement

C.A. Boys, R.J. Williams / Ecological Engineering 49 (2012) 221–232

225

Table 1 Catch summary of fish and decapods from all creeks (seine and gill nets). Abundances for the three most numerous species in each habitat are shown. Treatment Number of species Fish Non C/R species C/R species E–M species F–E species F species Decapods Non C/R species C/R species E–M species F–E species F species Abundance Fish Non C/R species C/R species Glass goby (Gobiopterus semivestitus) Swan River goby (Pseudogobius olorum) Flat-tail mulleta (Liza argentea) Decapods Non C/R species C/R species Pink shrimp (Acetes sibogae australis) Striped shrimp (Macrobrachium intermedium) School prawna (Metapenaeus macleayi)

Total

Control

Reference

Manipulated

49 29 20 32 16 1 21 17 4 19 2 0

32 20 12 21 11 0 18 14 4 16 2 0

43 25 18 31 11 1 16 12 4 15 1 0

36 21 15 24 11 1 17 13 4 16 1 0

212,744 204,126 8618 176,402 11,795 4147 86,720 84,739 1981 52,596 31,705 1475

29,038 27,645 1393 14,427 8413 752 11,860 11,632 228 415 11,032 77

62,341 60,978 1363 55,764 549 352 43,223 41,766 1,457 39,953 1637 1356

121,365 115,503 5862 106,211 2833 3043 31,637 31,341 296 12,228 19,036 42

C/R = of commercial or recreation importance, E–M = estuarine–marine, F–E = freshwater–estuarine, F = freshwater. a C/R species.

significantly changed the composition of the assemblage at Fish Fry Creek, it equated to no net change in the number of estuarine–marine or freshwater–estuarine–marine species (Fig. 4: Group II versus III). It was not until sometime after year 8 (6 years after culvert removal) that the assemblage at Fish Fry Creek became equivalent in composition to the reference creeks (Fig. 3a: Fish Fry Creek years 10–12, Table 2: Group V similarity 32.54, p = 0.001). This shift was driven by changing species abundances and a net increase in the number of estuarine–marine species (Table 3 Group III versus V and Fig. 4). School prawn, Port Jackson glassfish, pink shrimp, yellow-finned bream and large mouth goby (all E–M) became more abundant in Fish Fry Creek during this time whereas the abundance of striped shrimp, glass goby, sandy sprat and the three mullet species: fan-tail, flat-tail and sea mullet all fell (Table 3: Group III versus V). Importantly, in the years that Fish Fry Creek changed to become equivalent in composition to the reference creeks, no such change was seen in the control creeks (Fig. 3a), providing strong evidence that the response was due to culvert removal rather than unexplained environmental variation.

At some point after year 10 in the control creeks and year 12 in the reference creeks the assemblages at these locations converged (Fig. 3a). Sometime after year 12 the assemblage at Fish Fry Creek also underwent a significant change, albeit on a different trajectory from the references and controls (Fig. 3a, Table 2: Group I similarity 50.87, p = 0.001). During this time species richness declined (Fig. 4: Group I versus V and IV versus V) and species such as pink shrimp, striped shrimp, glass goby, flat-tail mullet, sandy sprat, Tamar River goby and large mouth goby all decreased in abundance in the reference as well as Fish Fry Creek, whereas blue-spot goby and school prawn increased in abundance (Table 3: Group I versus V and IV versus V). In control creeks, similar reductions in abundance were observed for striped shrimp, pink shrimp and glass goby, in addition to those goby species which characterised culverted controls in previous years (i.e. mangrove, blue-spot, half-bridled, bridled and checkered mangrove) (Table 3: Group II versus IV). Opposite responses were seen for Port Jackson glassfish, Tamar River goby and large mouth goby which increased in control creeks in the latter 2 years (Table 3: Group II versus IV). The net result of these changes in abundance in reference and control creeks was that their

Table 2 Statistically significant groupings (SIMPROF) of creek assemblages across space and time based upon Bray–Curtis similarity. Group a

Contains creeks and timesb

Fish Fry Creek versus combined references and controls FFC Yr 16 Group I (FFC Yr 1–2) (Cs Yr 1–10) Group II FFC Yr 3–8 Group III (Cs Yr 12–16) (Rs Yr 16) Group IV Group V (FFC Yr 10–12) (Rs Yr 2–12) Crabhole Creek versus combined references and controls Rs Yr 2–12 Group VI (CHC Yr 1–2, 4, 6–10) (Cs Yr 1–10) Group VII (CHC Yr 3, 5, 12–16) (Cs Yr 12–16) (Rs Yr 16) Group VIII a b

SIMPROF test Split from group(s)

At similarity



Prob.

II–V III–V IV–V V

50.87 40.79 37.02 32.54

4.34 3.86 2.32 2.07

0.001 0.001 0.001 0.001

VII–VIII VIII

38.96 34.45

3.08 2.60

0.001 0.001

For referencing, groups are labelled with Roman numerals as they appear across the ordinations in Fig. 3. Yr = Year, FFC = Fish Fry Creek, CHC = Crabhole Creek, Rs = combined reference creeks, Cs = combined control creeks.

226

C.A. Boys, R.J. Williams / Ecological Engineering 49 (2012) 221–232

Table 3 Results of SIMPER analyses showing the species that contributed most to the dissimilarity between statistically significant creek assemblage groupings identified by SIMPROF (Table 2). Fourth-root transformed average abundance. Only relevant group comparisons are shown. Fish Fry Creek versus combined references and controls Av. Abuna

Group I versus Group IV Average dissimilarity: 54.08 Glass goby Port Jacks. glassfish Blue-spot goby Striped shrimp School prawn Flat-tail mullet Common toadfish Largemouth goby Bridled goby Half-bridled goby Tamar River goby

Gobiopterus semivestitus Ambassis jacksoniensis Pseudogobius olorum Macrobrachium intermed. Metapenaeus macleayi Liza argentea Tetractenos hamiltoni Redigobius macrostoma Arenigobius bifrenatus Arenigobius frenatus Afurcagobius tamarensis

Group I

Group IV

Av. Diss

Sim/SD

Contrib%

Cum.%

0.48 1.70 0.42 0.13 0.45 0.77 0.49 0.00 0.11 0.00 0.31

2.02 0.65 1.36 1.03 0.85 0.29 0.04 0.42 0.51 0.36 0.68

8.34 5.72 5.11 4.84 3.44 2.68 2.42 2.29 2.17 2.03 1.90

4.07 3.65 3.30 2.44 1.17 1.32 4.55 7.16 3.03 1.48 1.20

15.42 10.57 9.45 8.94 6.37 4.95 4.47 4.24 4.01 3.75 3.51

15.42 26.00 35.44 44.38 50.75 55.70 60.17 64.41 68.41 72.16 75.68

Group I versus Group V

Av. Abun

Average dissimilarity: 56.59

Group I

Group V

Av. Diss

Sim/SD

Contrib%

Cum.%

0.48 0.39 0.13 0.42 0.00 0.31 0.11 0.77 1.70 0.49

3.13 2.5 1.37 1.20 0.79 0.94 0.65 1.03 1.23 0.19

10.43 8.50 4.88 3.13 3.11 2.68 2.11 1.86 1.83 1.80

4.13 3.06 3.16 2.10 2.32 1.61 2.76 2.22 1.33 3.04

18.43 15.01 8.62 5.53 5.50 4.73 3.74 3.30 3.24 3.18

18.43 33.45 42.07 47.61 53.11 57.84 61.58 64.87 68.11 71.29

Glass goby Pink shrimp Striped shrimp Blue-spot goby Largemouth goby Tamar River goby Bridled goby Flat-tail mullet Port Jacks. glassfish Common toadfish

Gobiopterus semivestitus Acetes sibogae australis Macrobrachium intermed. Pseudogobius olorum Redigobius macrostoma Afurcagobius tamarensis Arenigobius bifrenatus Liza argentea Ambassis jacksoniensis Tetractenos hamiltoni

Group II versus Group III

Av. Abun

Average dissimilarity: 49.04

Group II

Group III

Av. Diss

Sim/SD

Contrib%

Cum.%

2.38 2.00 0.39 0.54 0.01 2.36 1.28 1.52 0.06 1.06 0.63 0.17 0.73

5.36 3.61 1.81 1.63 1.18 1.29 0.39 0.63 0.79 0.38 0.94 0.72 0.14

7.57 4.11 3.66 2.94 2.89 2.70 2.38 2.28 1.90 1.80 1.74 1.68 1.48

3.06 2.01 3.20 1.42 1.74 3.71 1.43 2.26 1.56 1.62 1.47 1.02 1.47

15.44 8.37 7.46 6.00 5.89 5.50 4.86 4.64 3.88 3.66 3.55 3.43 3.01

15.44 23.81 31.27 37.27 43.17 48.67 53.53 58.17 62.05 65.71 69.26 72.69 75.70

Glass goby Striped shrimp Flat-tail mullet Pink shrimp Sandy sprat Blue-spot goby Half-bridled goby Mangrove goby Fan-tail mullet Bridled goby Tamar River goby Sea mullet Check. mang. goby

Gobiopterus semivestitus Macrobrachium intermed. Liza argentea Acetes sibogae australis Hyperlophus vittatus Pseudogobius olorum Arenigobius frenatus Mugilogobius paludis Paramugil georgii Arenigobius bifrenatus Afurcagobius tamarensis Mugil cephalus Mugilogobius stigmaticus

Group II versus Group IV

Av. Abun

Average dissimilarity: 43.39

Group II

Group IV

Av. Diss

Sim/SD

Contrib%

Cum.%

1.52 2.36 2.00 1.28 0.17 0.73 0.60 0.12 2.38 1.06 0.63 0.54 0.37

0.41 1.36 1.03 0.36 0.85 0.02 0.02 0.65 2.02 0.51 0.68 0.53 0.42

3.95 3.53 3.49 3.38 2.68 2.47 2.20 2.15 2.05 2.03 1.67 1.34 1.30

2.95 3.32 2.00 1.48 1.16 1.83 1.07 3.26 1.54 1.71 1.48 1.39 1.91

9.11 8.14 8.04 7.78 6.17 5.69 5.08 4.96 4.72 4.68 3.85 3.10 3.00

9.11 17.25 25.29 33.08 39.24 44.94 50.01 54.97 59.69 64.37 68.22 71.32 74.32

Mangrove goby Blue-spot goby Striped shrimp Half-bridled goby School prawn Check. mang. goby Eastern king prawn Port Jacks. glassfish Glass goby Bridled goby Tamar River goby Pink shrimp Largemouth goby

Mugilogobius paludis Pseudogobius olorum Macrobrachium intermed. Arenigobius frenatus Metapenaeus macleayi Mugilogobius stigmaticus Penaeus plebejus Ambassis jacksoniensis Gobiopterus semivestitus Arenigobius bifrenatus Afurcagobius tamarensis Acetes sibogae australis Redigobius macrostoma

Group II versus Group V

Av. Abun

Average dissimilarity: 47.51 Pink shrimp Mangrove goby Blue-spot goby Port Jacks. glassfish Glass goby Half-bridled goby

Acetes sibogae australis Mugilogobius paludis Pseudogobius olorum Ambassis jacksoniensis Gobiopterus semivestitus Arenigobius frenatus

Group II

Group V

Av. Diss

Sim/SD

Contrib%

Cum.%

0.54 1.52 2.36 0.12 2.38 1.28

2.5 0.19 1.20 1.23 3.13 0.37

5.76 3.88 3.38 3.32 2.84 2.76

2.50 3.47 2.63 2.39 1.33 1.49

12.13 8.16 7.12 6.98 5.97 5.81

12.13 20.29 27.42 34.40 40.37 46.18

C.A. Boys, R.J. Williams / Ecological Engineering 49 (2012) 221–232

227

Table 3 (Continued) Group II versus Group V

Av. Abun

Average dissimilarity: 47.51 Striped shrimp Flat-tail mullet Tamar River goby Check. mang. goby School prawn Largemouth goby Yellow-finned bream Bridled goby

Macrobrachium intermed. Liza argentea Afurcagobius tamarensis Mugilogobius stigmaticus Metapenaeus macleayi Redigobius macrostoma Acanthopagrus australis Arenigobius bifrenatus

Group II

Group V

Av. Diss

Sim/SD

2.00 0.39 0.63 0.73 0.17 0.37 0.11 1.06

1.37 1.03 0.94 0.13 0.71 0.79 0.66 0.65

2.05 2.00 1.82 1.73 1.72 1.69 1.62 1.45

1.32 1.48 1.35 1.51 1.34 1.57 1.71 1.46

Contrib% 4.32 4.20 3.83 3.65 3.62 3.56 3.40 3.06

Cum.% 50.50 54.70 58.54 62.18 65.80 69.36 72.76 75.83

Group III versus Group V

Av. Abun

Average dissimilarity: 40.05

Group III

Group V

Av. Diss

Sim/SD

Contrib%

Cum.%

3.61 5.36 1.63 1.18 1.81 0.54 0.79 0.94 0.72 0.14 0.54 0.50

1.37 3.13 2.50 0.38 1.03 1.23 0.07 0.94 0.35 0.71 0.66 0.79

5.53 5.44 2.95 2.14 2.03 1.91 1.81 1.57 1.52 1.48 1.16 1.15

2.52 2.00 1.41 1.36 1.51 1.50 1.55 1.44 1.13 1.38 1.55 1.42

13.80 13.57 7.36 5.34 5.07 4.76 4.52 3.92 3.80 3.70 3.10 3.00

13.80 27.37 34.73 40.07 45.14 49.90 54.42 58.34 62.13 65.83 66.93 69.93

Striped shrimp Glass goby Pink shrimp Sandy sprat Flat-tail mullet Port Jacks. glassfish Fan-tail mullet Tamar River goby Sea mullet School prawn Yellow-finned bream Largemouth goby

Macrobrachium intermed. Gobiopterus semivestitus Acetes sibogae australis Hyperlophus vittatus Liza argentea Ambassis jacksoniensis Paramugil georgii Afurcagobius tamarensis Mugil cephalus Metapenaeus macleayi Acanthopagrus australis Redigobius macrostoma

Group IV versus Group V

Av. Abun

Average dissimilarity: 38.80

Group IV

Group V

Av. Diss

Sim/SD

Contrib%

Cum.%

0.53 2.02 0.29 0.85 0.65 0.68 1.03 0.42 0.00 1.36 0.32

2.50 3.13 1.03 0.71 1.23 0.94 1.37 0.79 0.38 1.20 0.66

6.71 3.63 2.62 2.39 2.13 1.78 1.76 1.51 1.27 1.25 1.24

2.7 1.51 1.73 1.37 1.65 1.59 1.58 2.17 1.03 1.51 1.29

17.28 9.37 6.75 6.17 5.50 4.59 4.52 3.88 3.26 3.21 3.21

17.28 26.65 33.39 39.56 45.07 49.65 54.18 58.06 61.32 64.53 67.74

Group VI

Group VII

Av. Diss

Sim/SD

Contrib%

Cum.%

2.61 0.18 1.17 1.14 3.32 0.86 1.51 0.92 1.02 0.33 1.04 0.71 0.16

0.40 1.50 0.10 2.07 2.60 0.13 2.11 0.36 0.73 0.91 0.70 0.20 0.62

6.42 3.80 3.17 2.81 2.77 2.11 1.93 1.81 1.81 1.74 1.73 1.57 1.51

2.68 2.95 2.37 1.98 1.41 1.71 1.36 1.68 1.36 1.80 1.46 1.70 1.35

14.20 8.41 7.01 6.21 6.14 4.67 4.26 4.02 4.00 3.85 3.82 3.47 3.35

14.20 22.61 29.62 35.83 41.97 46.64 50.90 54.92 58.92 62.76 66.58 70.05 73.40

Pink shrimp Glass goby Flat-tail mullet School prawn Port Jacks. glassfish Tamar River goby Striped shrimp Largemouth goby Sandy sprat Blue-spot goby Yellow-finned bream

Acetes sibogae australis Gobiopterus semivestitus Liza argentea Metapenaeus macleayi Ambassis jacksoniensis Afurcagobius tamarensis Macrobrachium intermed. Redigobius macrostoma Hyperlophus vittatus Pseudogobius olorum Acanthopagrus australis

Crabhole Creek versus combined references and controls Av. Abun

Group VI versus Group VII Average dissimilarity: 45.19 Pink shrimp Mangrove goby Port Jacks. glassfish Blue-spot goby Glass goby School prawn Striped shrimp Largemouth goby Tamar River goby Half-bridled goby Flat-tail mullet Yellow-finned bream Check. mang. goby a

Acetes sibogae australis Mugilogobius paludis Ambassis jacksoniensis Pseudogobius olorum Gobiopterus semivestitus Metapenaeus macleayi Macrobrachium intermed. Redigobius macrostoma Afurcagobius tamarensis Arenigobius frenatus Liza argentea Acanthopagrus australis Mugilogobius stigmaticus

Average abundance refers to mean number per replicate seine haul.

assemblages merged in the latter years so that little discernable difference was noted between them (Fig. 3a, Table 2: Group IV similarity 32.54, p = 0.001). This convergence is discussed further below. Whilst the general nature of these latter changes across both un-manipulated creeks and Fish Fry Creek suggests that factors occurring at scales beyond that of the manipulation were responsible, there did appear to be some interaction between location and background environmental change. This is demonstrated by the different trajectory that Fish Fry Creek took from the control and

reference sites in year 16. This change was driven by the fact that the reductions in species abundance observed in both reference and control creeks occurred to a larger extent in Fish Fry Creek so that at year 16, Fish Fry Creek had lower abundances of these species than the reference (Table 3: Group I versus IV). Furthermore, blue-spot goby and common toadfish increased at Fish Fry Creek but not any of the controls or references (Table 3: Group I versus IV). The gill nets placed in Fish Fry Creek, the reference and control creeks caught fewer species than the seine nets but targeted larger size ranges: sea mullet (86–442 mm), flat-tail mullet

228

C.A. Boys, R.J. Williams / Ecological Engineering 49 (2012) 221–232

Fig. 5. Mean abundance of sea mullet and flat-tail mullet caught in multi-panel gill nets at Fish Fry Creek and the combined control and combined reference creeks. Years 1–2 were before culvert removal at Fish Fry Creek and years 3–16 are postculvert removal.

(85–305 mm), fan-tail mullet (93–134 mm) and yellow-finned bream (180–260 mm), compared to seine nets that targeted smaller individuals of these species (<125 mm, <203 mm, <92 mm and <185 mm, respectively). Of these species, only sea mullet and flat-tail mullet were caught in large enough numbers for further consideration. The abundance of sea mullet greater than 86 mm increased significantly in Fish Fry Creek in the first year following culvert removal (year 3), without any reciprocal increase in the control or reference creeks (Fig. 5). This implied a clear initial response to culvert removal, but it was not sustained and no further responses were noted in later years. The abundance of flat-tail mullet (>85 mm) was far more variable between years (Fig. 5). Increases in abundance observed in reference creeks in the first half of the study were not seen in Fish Fry Creek and the controls. Although flat-tail mullet numbers increased in Fish Fry Creek in year 12 and this was seen again in year 16, a similar increase was observed in the control creeks. As such, this response cannot be attributed to culvert removal. Fig. 4. Mean (±S.E.) number of all species, estuarine–marine species and freshwater-estuarine species for the different creek assemblage groupings identified by SIMPROF (Table 2).

3.3. Response at Crabhole Creek As at Fish Fry Creek, the assemblage at Crabhole Creek prior to culvert removal was similar to the control creeks and different

C.A. Boys, R.J. Williams / Ecological Engineering 49 (2012) 221–232

to the reference creeks (Fig. 3b, Table 2: Group VI defined from VII to VII at similarity 38.96, p = 0.001). More estuarine–marine (E–M) species were found in reference creeks (Fig. 4: Group VI versus VII) and they were generally more abundant (Table 3: Group VI versus VII). Like Fish Fry Creek, changes in the assemblage at Crabhole Creek were observed at different times after culvert removal (Fig. 3b). However, unlike Fish Fry Creek, no consistent trajectory towards reference condition was observed in any of the years that would suggest a sustained response to culvert removal. At various times (years 3 and 5), the assemblage at Crabhole Creek differed from that of the controls, however, this was never maintained and the assemblages between Crabhole Creek and the control creeks resumed significant similarity in alternative years (Fig. 3b). Notably, in years 12 and 16, the assemblage at Crabhole Creek became more similar to that of reference locations in year 16 (Fig. 3b: Table 2: Group XIII similarity 34.45, p = 0.001), however, this shift also occurred at control creeks in these years and is likely due to broader-scale assemblage changes occurring across Kooragang Island beyond the influence of the manipulations.

4. Discussion 4.1. Assemblage differences between restricted and non-restricted creeks A comparison between control (culverted) and reference (unculverted) creeks at Kooragang Island suggests that the installation of culverts many decades ago significantly changed the upstream assemblage of fish and decapods in tidal creeks. One of the more conspicuous ecological changes appears to have been the exclusion of many estuarine–marine species upstream of culverts, which resulted in a reduction in species richness and abundance. Numerous species were absent or consistently less abundant in tidally-restricted creeks, including: glass goby, Port Jackson glassfish, flat-tail mullet, yellow-finned bream, largemouth goby, pink shrimp, school prawn (all estuarine–marine dwelling), striped shrimp and Tamar River goby (both freshwater-estuarine dwelling). However, it is important to note that not all estuarine–marine dwelling species appeared disadvantaged by the presence of tidally restrictive culverts. This has been noted in studies elsewhere (Raposa and Roman, 2003; Eberhardt et al., 2011), where restricted wetlands that maintain some degree of tidal connectivity can support viable assemblages of estuarine–marine species. In fact, the habitat and inter-specific conditions created in these disturbed environments may be of direct advantage to certain species of fish (Raposa, 2008). At Kooragang Island, mangrove goby, bluespot goby, half-bridled goby, bridled goby and checkered mangrove goby were consistently more abundant in creeks with culverts. This may be in part due to their demersal nature (Cole and Shapiro, 1995) and preference for muddy substrates (Allen et al., 1989), the latter being typical of tidally restricted creeks (Raposa, 2008). The fact that other demersal species were not similarly benefited by habitat changes suggests that other factors may also be contributing to assemblage differences. Another possible explanation of the greater abundance of the gobies is that tidal creeks function as important refugia for small fish to avoid larger predatory fish (Rozas and Odum, 1988; Ruiz et al., 1993; Paterson and Whitfield, 2000). Culverts may enhance the refuge function for small-bodied species such as gobies: we found yellow-finned bream (a piscivore) was more abundant in the absence of culverts. However, the prey–refuge hypothesis is questionable here, as no significant increase in predatory fish species paralleled the observed decrease in mangrove goby, blue-spot

229

goby, half-bridled goby, bridled goby and checkered mangrove goby following culvert removal. 4.2. Successional changes in tidal creek assemblages following culvert removal Succession is a sequence of directional changes in the composition of a community (or assemblage of species) towards a stable condition (sensu Shugart, 2001). In this study we were able to observe clear changes in the fish and decapod assemblage of Fish Fry Creek over 16 years in response to culvert removal. Change at Crabhole Creek did not proceed in a straightforward manner. The first change at Fish Fry Creek occurred immediately and persisted for at least 6 years following culvert removal. Although there was no net increase or decrease in the mean number of species inhabiting the creek, there were significant changes in assemblage composition. The goby species previously mentioned as dominant in tidally restricted creeks became less abundant and a reciprocal increase in species such as flat-tail mullet, fantail-mullet and sea mullet was observed. Mullet are highly mobile and have been shown to quickly colonise habitats once passage is improved (Kroon and Ansell, 2006; Boys et al., 2011). They move immediately into tidal creeks once floodgates are opened and frequently use newly constructed fishways and widened culverts in coastal streams (Boys et al., 2011). Because mullet feed predominately on benthic organic matter and other detritus (Blaber, 1977) they are important primary consumers in estuarine foodwebs. Studies using stable isotope analysis (Deegan and Garritt, 1997) and bioenergetic modelling (Kneib, 2003) show that fish play a major role in the exchange of energy and nutrients between coastal wetlands and an estuary as well as offshore waters. By feeding in tidal creeks and subsequently returning to downstream environments where they interact with predators and prey, species such as mullet may play an important role in the export of energy and nutrients out of wetlands (Kneib, 1986; Deegan and Garritt, 1997) This concept has been defined as ‘trophic relay’ (Kneib, 2003). Given that rehabilitated wetlands such as Kooragang Island can act as important carbon sinks (Howe et al., 2009), improved passage rates of nektonic species may help translate these energy stores into increased ecosystem and fisheries productivity. Assemblage succession at Fish Fry Creek was potentially driven by a change in habitat. Culvert removal allowed significant increases in tidal height and hence tidal exchange that resulted in increased velocities and extensive widening and deepening of the channel through erosion (Howe et al., 2010). Although it was not quantified in this study, these changes in velocity and geomorphology would have scoured soft benthic muds and made the sites sampled less desirable for the previously mentioned demersal goby species. There may have been a migration of muddy habitat further upstream away from the initial sites that had been randomly sampled throughout the study. In future studies sampling should be conducted at extended distances along all selected creeks to better account for spatial shifts. A second shift in the assemblage of Fish Fry Creek was observed after year 8 (6 years after culvert removal). At this time its assemblage appeared to have taken a maturational step as it appeared equivalent to that of reference creeks. Most importantly, this change was not seen at control creeks, providing strong evidence that culvert removal was responsible. Notable changes in Fish Fry Creek at this time involved a reduction in the abundance of the mullet species (fan-tail, flat-tail and sea mullet), that had been present immediately after culvert removal. The reduction of these detrital feeding species may reflect changes in foraging conditions that occurred as the tidal creek became deeper and wider channel

230

C.A. Boys, R.J. Williams / Ecological Engineering 49 (2012) 221–232

(Howe et al., 2010). Other species to significantly decrease in abundance at this time at Fish Fry Creek were sandy sprat, glass goby and striped shrimp. No such marked changes in channel width, depth or velocity were noted at Crabhole Creek during the study. Whilst there were notable declines in abundance of some species in Fish Fry Creek, many others increased and overall there was a net gain in the number of estuarine–marine species during this secondary maturation stage. School prawn, Port Jackson glassfish, pink shrimp, yellow-finned bream and large mouth goby increased in abundance. Other studies at the Macleay and Clarence River estuaries (Kroon and Ansell, 2006; Boys et al., 2012) have observed pink shrimp in greater abundance in reference creeks relative to creeks with floodgates. This species is an important food source for carnivorous and omnivorous species co-habiting manipulated creeks, such as yellow-finned bream and large mouth goby (Xiao and Greenwood, 1993). Thus an increase in prey may be driving reciprocal increases in other species. Another shift was observed across all manipulated, control and reference creeks sometime after year 10 for control sites and after year 12 for reference and manipulated sites. Typically, this change manifested in a reduction in species richness and abundance, and surprisingly the reductions were greatest at Fish Fry Creek. One suggested reason for the change at the latter site was the collapse of the bridge across the creek in June 2007 during a La Nina cycle between the ninth and tenth samples (years 12 and 16), which may have significantly impacted on the passage of fish and decapods. However, the general nature of the reduction across all creeks strongly suggests it was caused by unknown broader scale effects across the whole of Kooragang Island, in the Hunter estuary, or at even larger scales. The use of control as well as reference sites is further validated due to this occurrence. Speculating about the cause of broader scale effects on fish and decapod populations across Kooragang Island or in the Hunter estuary was outside the scope of this study.

It is possible that the different responses between Fish Fry Creek and Crabhole Creek were mediated by inherent differences in habitat. As mentioned, the width and depth of the former were greater than the latter, and culvert removal at the former led to significant increases in tidal exchange that resulted in increased velocities that further widened and deepened the channel (Howe et al., 2010). In comparison, Crabhole Creek underwent little change in water velocity or creek depth. A slight increase in pneumatophore density at Crabhole Creek made sampling more difficult as there was greater refuge for small bodied fish and decapods, and this increase in density would presumably have offered greater shelter from predators (Sasekumar et al., 1992). This may explain why predatory fish such as yellow-finned bream did not utilise Crabhole Creek after culvert removal as much as they did in the deeper and more open habitats afforded by the reference creeks or Fish Fry Creek following its remediation. In trying to understand why some locations respond differently to rehabilitation than others, consideration has been given to either pre-settlement processes (such as the ability of individuals to colonise new habitats, Bell et al., 1988; Ford et al., 2010), or post-settlement processes such as competitive and predator–prey interactions between species (e.g. Raposa and Roman, 2003). The closer proximity (4 km) of Fish Fry Creek to the mouth of the Hunter River may be sufficient to create the difference in colonisation rates between the two locations and therefore different trajectories of rehabilitation response. Colonisation rates tend to be faster the closer a habitat is to a good supply of post-settlement individuals (although most research to date concerns artificial reefs (e.g. Hueckel et al., 1989; Golani and Diamant, 1999). Within estuaries, it has been shown that fish densities in some habitats are higher the closer they are to the estuary mouth, possibly reflecting a greater supply of larvae and juvenile entering from offshore habitats (Bell et al., 1988; Hannan and Williams, 1998; Ford et al., 2010). 4.4. Summary and conclusion

4.3. Variability in rehabilitation trajectories Unlike Fish Fry Creek, where the response trajectory to culvert removal was readily identified, change at Crabhole Creek was erratic. Significant changes in the assemblage, reflecting a reduction in the abundance and number of species were detected irregularly following culvert removal (years 3 and 5), but were not observed in years 4, 6 and 8 when the assemblage was akin to the control sites. These findings suggest that ecological responses to wetland rehabilitation can be location-specific, even to degree that creeks within the same wetland respond differently. Location specificity has been noted in other wetland remediation studies (Raposa, 2008; Boys et al., 2012). Similarly, weak directional response to rehabilitation is commonly reported for some indicators of ecosystem functioning but not others (Bishel-Machung et al., 1996; Simenstad and Thom, 1996; Minello and Webb, 1997; Zedler and Callaway, 2000; Moreno-Mateos et al., 2012). A review of wetland rehabilitation has shown that fish and crustacean assemblages can take 6–8 years to respond to improved tidal flushing (Moreno-Mateos et al., 2012), but that a large amount of variability in responses are observed, possibly driven by differences in the amount of upstream habitat which is created between studies. Such variability was found in this study and reiterates why caution must be exercised when generalising ecological responses across different rehabilitation projects and demonstrates that it is important for rehabilitation efforts to have some degree of pre- and post-manipulation evaluation. This is particularly relevant for wetland rehabilitation in disturbed environments, where a multitude of other stressors may be constraining responses (Grayson et al., 1999).

This study has provided irrefutable evidence that culverts, by creating a passage constriction and/or by reducing tidal flow, can reduce the richness and change the composition of fish and decapod assemblages of tidal creeks. These changes were driven by a reduction in the number and abundance of estuarine–marine dwelling species. Most importantly, it has been demonstrated that culvert removal can lead to clear successional changes and a trajectory of improvement in fish and decapod assemblages. Whilst the assemblages of manipulated creeks that have had culverts removed can become equivalent to unculverted reference creeks, this may take many years and is not experienced at all locations. The presence of distinct groups of fish and/or decapods may indicate stages in the maturation of a rehabilitated wetland and should be further investigated. Speculating about the cause of broader scale effects on fish and decapod populations across Kooragang Island or in the Hunter estuary is outside the scope of this study, but the observed changes highlight the importance of incorporating suitable reference locations into rehabilitation studies. We have shown that rehabilitated wetlands can develop along complex trajectories that may be difficult to predict. Without references it would be impossible to ascertain what unforeseen changes can be attributed to the rehabilitation intervention and what may be purely due to apparently unpredictable disturbances that can be common in wetlands due to ˜ ˜ natural circumstances such as El Nino/La Nina-Southern Oscillation (ENSO) or due to changes in land use. We concur with others (e.g., Zedler and Callaway, 2000; Choi, 2004) that it is difficult to predict ecological assemblages once a rehabilitation has been initiated, as multiple response trajectories

C.A. Boys, R.J. Williams / Ecological Engineering 49 (2012) 221–232

are possible and may be location-specific (even within the same wetland). Long-term studies (5–10 years, but not necessarily annually) are needed to track responses to rehabilitation outcomes. Whilst manipulated sites can show an improvement in the nature of their assemblages, the time taken for manipulated wetlands to reach equivalency with reference states may exceed conventional short term (3–5 years) monitoring periods of the past. Further, manipulated sites may never fully replace natural systems in composition or function. Acknowledgements We are grateful to all staff of the Department of Primary Industries and Kooragang Wetland Rehabilitation Project (KWRP) and all volunteers in the field over the 16 years of this study. Staff included J. Hannan, F. Watford, V. Balachov, D. Sullings, B. Louden, A. Genders, I. Thiebaud, B. McCartin, B. Rampano, T. Fowler, G. Reilly, B. Kearney and A. Creese. We are thankful to the leaseholders of Kooragang Island who allowed us access to the wetlands. This research was initiated in 1993 by a grant from Port Waratah Coal Services Pty. Ltd and finalised by a grant from the KWRP via the Hunter-Central Rivers Catchment Management Authority. These funding bodies had no role in the study design or the collection, analysis and interpretation of data. We would like to acknowledge the efforts of C. Copeland who was the initial instigator of the project. We are particularly thankful to P. Svoboda, R. Henderson, B. Creese and (the late) S. Rostas for their support throughout this study. B. Creese and two anonymous reviewers provided advice which significantly improved this manuscript. The collection of all animals in this project was in accordance with the appropriate animal care and ethics research authority (98/11) and a Section 37 research permit in accordance with the NSW Fisheries Management Act 1994. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ecoleng.2012.08.006. References Able, K.W., Grothues, T.M., Hagan, S.M., Kimball, M.E., Nemerson, D.M., Taghon, G.L., 2008. Long-term response of fishes and other fauna to restoration of former salt hay farms: multiple measures of restoration success. Rev. Fish Biol. Fisher. 18, 65–97. Allen, G.R., Midgley, S.H., Allen, M., 1989. Freshwater Fishes of Australia. T.F.H. Publications, Sydney. Bell, J.D., Steffe, A.S., Westoby, M., 1988. Location of seagrass beds in estuaries: effects on associated fish and decapods. J. Exp. Mar. Biol. Ecol. 122, 127–146. Bishel-Machung, L., Brooks, R.P., Yates, S.S., Hoover, K.L., 1996. Soil properties of reference wetlands and wetland creation projects in Pennsylvania. Wetlands 16, 532–541. Blaber, S., 1977. The feeding ecology and relative abundance of mullet (Mugilidae) in Natal and Pondoland estuaries. Biol. J. Linn. Soc. 9, 259–275. Boys, C.A., Glasby, T., Kroon, F.J., Baumgartner, L.J., Wilkinson, K., Reilly, G., Fowler, T., 2011. Case Studies in Restoring Connectivity of Coastal Aquatic Habitats: Floodgates, Box Culvert and Rock-Ramp Fishway. NSW Department of Primary Industries, Cronulla. Boys, C.A., Kroon, F.J., Glasby, T., Wilkinson, K., 2012. Improved fish and crustacean passage in tidal creeks following floodgate remediation. J. Appl. Ecol. 49, 223–233. Choi, Y.D., 2004. Theories for ecological restoration in changing environment: toward ‘futuristic’ restoration. Ecol. Restor. 19, 75–81. Clarke, K.R., 1993. Non-parametric multivariate analyses of changes in community structure. Aust. J. Ecol. 18, 117–143. Clarke, K.R., Green, R.H., 1988. Statistical design and analysis for a ‘biological effects’ study. Mar. Ecol. Prog. Ser. 46, 213–226. Clarke, K.R., Somerfield, P.J., Chapman, M.G., 2006. On resemblance measures for ecological studies, including taxonomic dissimilarities and a zero-adjusted Bray–Curtis coefficient for denuded assemblages. J. Exp. Mar. Biol. Ecol. 330, 55–80.

231

Clarke, K.R., Somerfield, P.J., Gorley, R.N., 2008. Testing of null hypotheses in exploratory community analyses: similarity profiles and biota–environment linkage. J. Exp. Mar. Biol. Ecol. 366, 56–69. Cole, K.S., Shapiro, D.Y., 1995. Social facilitation and sensory mediation of adult sex change in a cryptic, benthic marine goby. J. Exp. Mar. Biol. Ecol. 186, 65–75. Craft, C., Megonigal, P., Broome, S., Stevenson, J., Freese, R., Cornell, J., Zheng, L., Sacco, J., 2003. The pace of ecosystem development of constructed Spartina alterniflora marshes. Ecol. Appl. 13, 1417–1432. Deegan, L.A., Garritt, R.H., 1997. Evidence for spatial variability in estuarine food webs. Mar. Ecol. Prog. Ser. 147, 31–47. Eberhardt, A.L., Burdick, D.M., Dionne, M., 2011. The effects of road culverts on Nekton in New England salt marshes: implications for tidal restoration. Rest. Ecol. 19, 776–785. Ford, J.R., Williams, R.J., Fowler, A.M., Cox, D.R., Suthers, I.M., 2010. Identifying critical estuarine seagrass habitat for settlement of coastally spawned fish. Mar. Ecol. Prog. Ser. 408, 181–193. Gedan, K.B., Silliman, B., Bertness, M., 2009. Centuries of human-driven change in salt marsh ecosystems. Annu. Rev. Mar. Sci. 1, 117–141. Golani, D., Diamant, A., 1999. Fish colonization of an artificial reef in the Gulf of Elat, northern Red Sea. Environ. Biol. Fish. 54, 275–282. Grayson, J.E., Chapman, M.G., Underwood, A.J., 1999. The assessment of restoration of habitat in urban wetlands. Landsc. Urban Plan. 43, 227–236. Hannan, J., Williams, R.J., 1998. Recruitment of juvenile marine fishes to seagrass habitat in a temperate Australian estuary. Estuaries 21, 29–51. Hobbs, R.J., Norton, D.A., 1996. Towards a conceptual framework for restoration ecology. Restor. Ecol. 4, 93–110. Howe, A.J., Rodríguez, J.F., Spencer, J., MacFarlane, G.R., Saintilan, N., 2010. Response of estuarine wetlands to reinstatement of tidal flows. Mar. Freshw. Res. 61, 702–713. Howe, A.J., Rodríguez, J.R., Saco, P.M., 2009. Surface evolution and carbon sequestration in disturbed and undisturbed wetland soils of the Hunter estuary, southeast Australia. Estuar. Coast. Shelf Sci. 84, 75–83. Hueckel, G.J., Buckley, R.M., Benson, B.L., 1989. Mitigating rocky habitat loss using artificial reefs. Bull. Mar. Sci. 44, 913–922. Kentula, M.W., Brooks, R.P., Gwin, S.E., Holland, C.C., Sherman, A.D., Sifneos, J.C., 1992. An Approach to Improving Decision Making in Wetland Restoration and Creation. Island Press, Corvallis. Kingsford, R.T., Ferster Levy, R., Geering, D., Davis, S.T., Davis, J.S.E., 1998. Rehabilitating Estuarine Habitat on Kooragang Island for Waterbirds, Including Migratory Wading Birds (May 1994–May 1997). NSW National Parks and Wildlife Service, Hurstville. Kneib, R.T., 1986. The role of Fundulus heteroclitus in salt marsh trophic dynamics. Am. Zool. 26, 259. Kneib, R.T., 2003. Bioenergetic and landscape considerations for scaling expectations of nekton production from intertidal marshes. Mar. Ecol. Prog. Ser., 264. Kroon, F.J., Ansell, D.H., 2006. A comparison of species assemblages between drainage systems with and without floodgates: implications for coastal floodplain management. Can. J. Fish. Aquat. Sci. 63, 2400–2417. Kruskal, J.B., Wish, M., 1978. Multidimensional Scaling. Sage Publications, London. Lotze, H.K., Lenihan, H.S., Bourque, B.J., Bradbury, R.H., Cooke, R.G., Kay, M.C., Kidwell, S.M., Kirby, M.X., Peterson, C.H., Jackson, J.B.C., 2006. Depletion, degradation, and recovery potential of estuaries and coastal seas. Science 312, 1806. Minello, T., Webb Jr., J., 1997. Use of natural and created Spartina alterniflora salt marshes by fishery species and other aquatic fauna in Galveston Bay, Texas, USA. Mar. Ecol. Prog. Ser. 151, 165–179. Moreno-Mateos, D., Power, M.E., Comín, F.A., Yockteng, R., 2012. Structural and functional loss in restored wetland ecosystems. PLoS Biol. 10, e1001247. Morgan, P.A., Short, F.T., 2002. Using functional trajectories to track constructed salt marsh development in the Great Bay estuary, Maine/New Hampshire, USA. Restor. Ecol. 10, 461–473. Paterson, A.W., Whitfield, A.K., 2000. Do shallow-water habitats function as refugia for juvenile fishes? Estuar. Coast. Shelf Sci. 51, 359–364. Pollard, D.A., Hannan, J.C., 1994. The ecological effects of structural flood mitigation works on fish habitats and fish communities in the lower Clarence river system of south-eastern Australia. Estuaries 17, 427–461. Raposa, K.B., 2008. Early ecological responses to hydrologic restoration of a tidal pond and salt marsh complex in Narragansett Bay, Rhode Island. J. Coast. Res. 55, 180–192. Raposa, K.B., Roman, C.T., 2001. Seasonal habitat-use patterns of nekton in a tide-restricted and unrestricted New England salt marsh. Wetlands 21, 451–461. Raposa, K.B., Roman, C.T., 2003. Using gradients in tidal restriction to evaluate nekton community responses to salt marsh restoration. Estuar. Coast. 26, 198–205. Roman, C.T., Niering, W.A., Warren, R.S., 1984. Salt marsh vegetation change in response to tidal restriction. Environ. Manage. 8, 141–149. Roy, P.S., Williams, R.J., Jones, A.R., Yassini, I., Gibbs, P.J., Coates, B., West, R.J., Scanes, P.R., Hudson, J.P., Nichol, S., 2001. Structure and function of south-east Australian estuaries. Estuar. Coast. Shelf Sci. 53, 351–384. Rozas, L.P., Odum, W.E., 1988. Food, predation risk, and microhabitat selection in a marsh fish assemblage. Ecology 69, 1341–1351. Ruiz, G.M., Hines, A.H., Posey, M.H., 1993. Shallow water as a refuge habitat for fish and crustaceans in non-vegetated estuaries: an example from Chesapeake Bay. Mar. Ecol. Prog. Ser. 99, 1–16. Sasekumar, A., Chong, V.C., Leh, M.U., D’Cruz, R., 1992. Mangroves as a habitat for fish and prawns. Hydrobiologia 247, 195–207.

232

C.A. Boys, R.J. Williams / Ecological Engineering 49 (2012) 221–232

Shugart, H.H., 2001. Phenomenon of succession. In: Levin, S.A. (Ed.), Encyclopedia of Biodiversity, vol. 5. Academic Press, San Diego, pp. 541–552. Simenstad, C., Reed, D., Ford, M., 2006. When is restoration not? Incorporating landscape-scale processes to restore self-sustaining ecosystems in coastal wetland restoration. Ecol. Eng. 26, 27–39. Simenstad, C.A., Thom, R.M., 1996. Functional equivalency trajectories of the restored Gog–Le–Hi–Te estuarine wetland. Ecol. Appl. 6, 38–56. Streever, W.J., 1998. Kooragang Wetland Rehabilitation Project: opportunities and constraints in an urban wetland rehabilitation project. Urban Ecosyst. 2, 205–218. Streever, W.J., Genders, A.J., 1997. Effect of improved tidal flushing and competitive interactions at the boundary between salt marsh and pasture. Estuar. Coast. 20, 807–818. Teal, J.M., Weishar, L., 2005. Ecological engineering, adaptive management, and restoration management in Delaware Bay salt marsh restoration. Ecol. Eng. 25, 304–314. Terlizzi, A., Bernedetti-Cecchi, L., Bevilacqua, S., Fraschetti, S., Guidetti, P., Anderson, M.J., 2005. Multivariate and univariate assymetrical analyses in environmental impact assessment: a case study of Mediterranean subtidal sessile assemblages. Mar. Ecol. Prog. Ser. 289, 27–42.

Thelen, B.A., Thiet, R.K., 2009. Molluscan community recovery following partial tidal restoration of a New England estuary, USA. Rest. Ecol. 17, 695–703. Underwood, A.J., 1990. Experiments in ecology and management: their logics, functions and interpretations. Aust. J. Ecol. 15, 365–389. Vitousek, P.M., Mooney, H.A., Lubchenco, J., Melillo, J.M., 1997. Human domination of Earth’s ecosystems. Science 277, 494–499. Walker, L.R., Walker, J., Hobbs, R.J., 2007. Linking Restoration and Ecological Succession. Springer-Verlag, New York, p. 190. Williams, R.J., Watford, F.A., Balashov, V., 2000. Kooragang Wetland Rehabilitation Project: History of Changes in Estuarine Wetlands of the Lower Hunter River. NSW Fisheries, Cronulla. Xiao, Y., Greenwood, J.G., 1993. The biology of Acetes (Crustacea; Sergestidae). Oceanogr. Mar. Biol. 31, 259–444. Zedler, J.B., Callaway, J., 2009. Tidal wetland restoration. Ecology 5, 291–300. Zedler, J.B., Callaway, J.C., 2000. Evaluating the progress of engineered tidal wetlands. Ecol. Eng. 15, 211–225. Zedler, J.G., Callaway, J.C., 1999. Tracking wetland restoration: do mitigation sites follow desired trajectories? Restor. Ecol. 7, 69–73.