Feral pig exclusion fencing provides limited fish conservation value on tropical floodplains

Efforts to protect and restore tropical wetlands impacted by feral pigs (Sus scrofa) in northern Australia have more recently included exclusion fences, an abatement response proposing fences improve wetland condition by protecting habitat for fish production and water quality. Here we tested: 1) whether the fish assemblage are similar in wetlands with and without fences; and 2) whether specific environmental processes influence fish composition differently between fenced and unfenced wetlands. Twenty-one floodplain and riverine wetlands in the Archer River catchment (Queensland) were surveyed during post-wet (June-August) and late-dry season (November-December) in 2016, 2017 and 2018, using a fyke soaked overnight (~14-15hrs). A total of 6,353 fish representing twenty-six species from 15 families were captured. There were no multivariate differences in fish assemblages between seasons, years and for fenced and unfenced wetlands (PERMANOVA, Pseduo-F <0.58, P<0.68). Late-dry season fish were considerably smaller compared to post-wet season: a strategy presumably to maximise rapid disposal following rain. At each wetland a calibrated Hydrolab was deployed (between 2-4 days, with 20min logging) in the epilimnion (0.2m), and revealed distinct diel water quality cycling of temperature, dissolved oxygen and pH (conductivity represented freshwater wetlands) which was more obvious in the late-dry season survey, because of extreme summer conditions. Water quality varied among wetlands, in terms of the daily amplitude, and extent of daily photosynthesis recovery, which highlights the need to consider local site conditions rather than applying general assumptions around water quality conditions for these types of wetlands examined here. Though many fish access (fenced and unfenced) wetlands during wet season connection, the seasonal effect of reduced water level conditions seems to be more over-improvised compared to whether fences are installed or not, as all wetlands supported few, juvenile, or no fish species because they had dried completed regardless of whether fences were present or not.


51
Wetlands (palustrine and lacustrine) that are located on floodplains away from riverine 52 channels support rich aquatic plant and fauna communities [1][2][3]. However, some point after 53 peak flood connection, aquatic organisms occupying these wetlands begin to face a moving 54 land-water margin, until connection is broken, at which point the remaining wetland 55 waterbodies typically support a non-random assortment of species, including fish [4,5]. The 56 duration, timing and frequency that off channel wetlands maintain lateral pulse connection to 57 primary rivers is an important determining factor in broader contribution to coastal fisheries 58 production [6][7][8][9]. In addition to connection, environmental conditions become important 59 including water quality [10], access to shelter to escape predation and available food resources 60 [11,12]. Efforts by managers to restore wetland ecosystem values is increasing, nevertheless 61 access to data establishing success of these programs are limited, which becomes important 62 when attempting to establish biodiversity returns for the funding investment made by 63 government or private investor organisations [13][14][15].

65
After floodplain wetlands begin receding and progressively disconnect from the main river 66 channel, they become smaller and shallower [16] because of water loss via evaporation, 67 groundwater recharge, or consumption by wildlife [17,18]. In tropical north Australia,68 seasonal off channel wetlands are more pronounced owing to high evaporation rates, loss to 69 groundwater [19], and in many situations waters quickly retract away from the banks and 70 riparian shade [16]. At that point, it is thought that they become more prone to reduced water 71 quality conditions -most notably reduced water depth [18], and high water temperatures [10, 72 20]. This increases aquatic fauna exposure risks to acute and chronic thresholds [21,22]. In 73 the late-dry season, fish confined to isolated wetlands on floodplains therefore have very 74 limited avoidance options [10], unlike other fauna such as the freshwater crab also occupying 75 seasonal tropical rivers in northern Australia, that will employ terrestrial re-location and access 76 burrows when confronted with thermoregulation [23]. Fish must exploit available ephemeral 77 aquatic habitats [24,25], which can be specific to each wetland depending on orientation and 78 location [26], depth and vegetation cover in the landscape [20], in order to survive until 79 monsoonal rain reconnects overbank river networks again. gives rise to soil erosion and benthic sediment re-suspension, reduced water clarity and 87 eutrophication which becomes particularly critical late-dry season. The fact that limited data 88 exits on the impact that feral pigs contribute to wetlands [31][32][33][34], places a strain on the ability 89 for land managers to quantify the consequences of pig destruction [35]. Conversely, a lack of 90 baseline data means quantifying success following expensive mitigation efforts is difficult. The aims were twofold: first to determine whether the model of non-randomness of fish stands 105 here in wetlands, and secondly whether specific environmental conditions influence fish 106 composition in wetlands with and without fences. These data are important and necessary 107 given increasing government funding investment underway and planned in northern Australia 108 for restoration of wetlands impacted by feral animals (including pigs) [10].

116
The Archer River catchment is located on Cape York Peninsula, north Queensland (Fig 1).

117
The head waters of the river rise in the McIlwraith range on the eastern side Cape York, where 118 it flows and then enters Archer Bay on the western side of the Gulf of Carpentaria; along with 119 the Watson and Ward Rivers. The catchment area is approximately 13,820 km 2 , which 120 includes approximately 4% (510 km 2 ) of wetland habitats 121 (https://wetlandinfo.des.qld.gov.au/wetlands/facts-maps/basin-archer/), such as estuarine   The characteristics of each wetland are summarised in Table S1. Here, sampling focused on 152 two periods: 1) immediately following the wet season after disconnection between the river 153 and wetlands (hereafter referred to as post wet season); and 2) late-dry season (hereafter late-    were assessed to determine those with a high overall contribution to tree node split, with the 201 best overall classifier being given a relative importance of 100%.

203
Kolmogorov-Smirnov (K-S) two-sample tests determined differences in the overall shape of 204 fish body size distribution using a Bonferroni correction for multiple comparisons. K-S tests 205 take into account differences between the location, skew, and kurtosis of frequency 206 distributions; but do not identify which of these parameters are driving distributional 207 differences. Therefore, we report the following characteristics of each body size distribution to 208 further describe any differences found: mean, standard deviation (sd), minimum value (min), 209 maximum value (max), the range of values, skewness, and kurtosis.

213
Wet season rainfall totals in the Archer River catchment were low during the study period 214 compared to the preceding years (Fig. 2), with rainfall within the 10 th percentile for historical 215 recordings held by the Australian Bureau of Meteorology. This means that some caution is 216 necessary with interpretation of these data; namely that floodplain connectivity under higher 217 rainfall years is likely to have a longer duration when compared to lower connection duration 218 under the current rainfall conditions. 219 220 A full summary of water quality data are provided in Supplementary files (S1). In summary, 221 water temperatures during the study period were generally about 26 o C (Table 1)  The water column exhibited pronounced diel temperature periodicity; one or two hours after 225 sunrise each day. Near-surface water temperatures began to rise at an almost linear rate for a 226 period of 8.0 ± 0.5 hours, generally reaching daily maxima during the middle of the afternoon.

227
The mean daily temperature amplitude was 6.2ºC (highest daily amplitude 9.6ºC, lowest 4.4 o C). For the remaining 16 hours of each day, near-surface water temperatures gradually 229 declined reaching minimum conditions shortly after sunset.

231
The electrical conductivity (EC) was very low (Table S1) during the post wet season surveys, 232 while the late-dry season conductivity was generally higher in wetlands, a consequence of 233 evapo-concentration. The lowest wetland in the catchment (AR08 located on the coastal 234 floodplain) recorded the highest conductivity, suggesting connection with tidal water from the 235 nearby estuary at some stage.

237
There was evidence of cyclical daily DO fluctuations supporting the contention that biological 238 diel periodicity processes were probably not significantly inhibited in all wetlands (Fig. 3). Daily 239 minimum DO concentrations were low enough to suggest there was enough respiratory oxygen 240 consumption to measurably affect water quality, particularly so at the pig impacted wetlands, but   (Table 1).

261
The most common species was the freshwater glassfish (Ambassis sp., 51% total catch),   The low species richness in wetlands relative to the main Archer River channel might be a 307 consequence of the frequency and duration of connection between wetlands and the main 308 Archer river channel. The wet season rainfall immediately prior, and during this survey, was 309 within the 10 th percentile for historical records. In research elsewhere, a longer connection 310 duration was shown to result in more fish present post wet season, and conceivably more 311 species present late-dry season [6,47]. Examples exist where longer connection between main 312 river channels and wetlands contributes positively to fish growth rates and higher abundance 313 and diversity of fish [24,26,48]. It is also possible that the field methods used here confound 314 our ability to determine the full species composition in wetlands -this could be overcome by 315 using additional survey techniques, including multi-panel gill nets, traps or electrofishing 316 (though we attempted to electrofish these wetlands however, conductivity was too low to 317 effectively use that method).   The cultural and ecological value of coastal wetlands means that management intervention is 770 increasingly necessary to ensure they remain productive and viable habitat (Creighton et al., 771 2015). These data support a model that damage to wetlands from pig activities not only 772 contributes to reduced aquatic habitat, through loss of aquatic vegetation communities, but also 773 probably has secondary impacts including water temperature and asphyxiation risks for many 774 hours each day, that are higher than when compared to fenced wetlands ( Figure 6). However, 775 fish occupying fenced and unfenced wetlands here were similar, particularly in the late-dry 776 season where those remaining few species were juveniles ready for wet season re-distribution.

777
On this basis, installing fences to both floodplain and riverine wetlands that were not on the 778 main flow channels, but rather were on anabranches and flood channels that connect to the main channels only during high flow conditions, seems to offer little additional habitat value 780 for fish. Where wetlands are largely ephemeral and will dry anyway, or where wetlands remain 781 until the next seasons rain connection; species abundance and/or diversity is not improved by 782 restricting feral pig access. Further research is necessary to examine climate change resilience