Introduction

With the global population expected to reach nine billion inhabitants by 2050, global food demand is expected to increase by 35–56% by 2050 from 2010 levels, consequently exacerbating tensions between land use for agriculture and habitat for biodiversity (van Dijk et al. 2021). Agricultural expansion and intensification can result in the loss of ecosystems and biodiversity, the eutrophication and sedimentation of aquatic ecosystems, increased greenhouse gas emissions, and altered hydrology (Tilman et al. 2001, 2002; Awuchi et al. 2020; Gaugler et al. 2020). Wetlands, including rivers and estuaries following RAMSAR convention (Ramsar Convention Secretariat 2016), are among the ecosystems most impacted by agricultural activities given their vulnerability to nutrient, sediment, and pesticide runoff (Buck et al. 2012; Ostrowski et al. 2021), water abstraction (Acreman et al. 2000), and to drainage and reclamation (Coleman et al. 2008). Global floodplain wetland loss is estimated at 95 km2/yr, though this loss is unlikely to slow or reduce under global human population predictions over the coming few decades (Coleman et al. 2008; Davidson 2014), with climate change set to exacerbate wetland loss and fragmentation across the globe (Segan et al. 2016). Despite this, approximately 10% of global animal biodiversity is found in freshwater ecosystems, which occupy <1% of Earth’s surface (Dudgeon 2019), with wetlands providing many ecosystem services (Barbier 2019; Davidson et al. 2019; Xu et al. 2020). There is, however, growing recognition of the need for the transformation of agricultural landscapes with both wetlands restoration and the application of agroecological principles, such as regenerative agricultural practices (LaCanne and Lundgren 2018; van Coppenolle and Temmerman 2019; Gliessman 2020). Large-scale wetlands construction and restoration are critical to advancing progress towards the UN Sustainable Development Goals (SDGs), including the universal provision of clean water access, the sustainable management of land and water ecosystems, and climate action (United Nations General Assembly 2015). Agroecology uses ecological principles to design and manage sustainable food systems, seeking to reconcile economic, environmental, and social dimensions (Gliessman 2020).

Like the rest of the world, Australia faces a legacy of degraded freshwater ecosystems, despite a small population and a relatively short 200 years of colonial urban, industrial, and agricultural development (Creighton et al. 2016). For example, between 2001 and 2017, the Great Barrier Reef (GBR) catchment experienced net loss of 740 ha (0.55% of 2001 extent) of coastal floodplain wetland. While contemporary loss has been relatively small, since European arrival, approximately 77,500 ha of palustrine wetland has been lost (~21.2% loss of pre-clear extent), much of which is from freshwater floodplain systems. Artificial/highly modified wetlands, however, have increased substantially between 2001 and 2017, increasing by 21,690 ha, much of which was created through the construction of tidal barrages (8299 ha) (Environmental Protection Agency 2005; Department of Environment and Science 2019). This loss of natural habitat, and the degradation of remaining habitat, is also reducing the GBR’s resilience to pressures from on-going pollutant runoff (Waterhouse et al. 2016; MacNeil et al. 2019; Adame et al. 2019a), and reducing habitat availability for species with freshwater life stages (Arthington et al. 2015; Adame et al. 2019a). Not only do coastal freshwater floodplain wetlands support diverse biological communities, but they also form part of important, broader, connected ecosystems, providing habitat for migratory species, flow regulation, and reduce sedimentation and nitrogen runoff (Bainbridge et al. 2009; Brodie and Waterhouse 2012; Waterhouse et al. 2016). While artificial/highly modified wetlands have been largely constructed for agricultural benefit, they still provide habitat for many wetland-dependent species and may be mitigating impacts of natural wetland loss (Canning and Waltham 2021).

Funding wetlands restoration and ensuring that it does not come at the expense of food production are two of the biggest barriers to large-scale wetland restoration (Waltham et al. 2020; Canning et al. 2021). While wetlands have been shown to improve food security directly through providing harvestable resources (e.g., fish, shellfish, and plants) (Cunningham 2015), food production can be improved indirectly through increased water security for irrigation and improved catchment drainage (Shennan and Bode 2002; Chen and Wong 2016). Leveraging on situations where wetland restorations also provide agricultural benefits (i.e., win-win scenarios) will be essential for realizing large-scale wetland restoration. In many situations, the financial benefits arising from wetland restoration need to outweigh the costs of restoring and maintaining the wetlands, and lost income. Where wetlands do not result in sufficient on-farm profit returns, financial incentive schemes, such as those paying for the provision of ecosystem services in the form or carbon sequestration, nutrient removal, or habitat provision, may help to ensure they are profitable at the farm scale (Banerjee et al. 2013; Sapkota and White 2020; Canning et al. 2021).

Instances where restored wetlands provide both agricultural and biodiversity benefits have primarily been due to the use of wetlands for growing rice, supporting livestock grazing, and improved water storage for irrigation (Verhoeven and Setter 2010; McIntyre et al. 2011; Peh et al. 2014). However, the benefits from using wetlands for drainage and flow regulation to improve both crop production and biodiversity simultaneously are rarely documented (Brander et al. 2013; Kadykalo and Findlay 2016).

On the floodplains of Australia’s wettest catchment (Tully-Murray), sugarcane farmers have grappled with the challenges of growing sugarcane on frequently flooded land, while supporting the health of the downstream Great Barrier Reef ecosystems. As part of Queensland’s $40 M (2001 AUD) Sugar Industry Infrastructure Program (hereafter ‘SIIP’), funded by the Queensland Government and the Australian Government, the Riversdale-Murray Valley Water Management Scheme (hereafter ‘the Riversdale-Murray Scheme’), completed in 2004, financially incentivised sugarcane farmers to construct wetlands across the Tully-Murray floodplain (Ernst and Young 2001). The Riversdale-Murray Scheme aimed to reduce cane inundation by flood waters and increase cane production across the Tully-Murray floodplain through the creation of arterial drainage and lagoon wetlands. Reduced inundation of mid-catchment cane land was achieved by directing water via preferential flow paths and draining it into ‘sump’ lagoons where the increased residence time then improved flow regulation and reduced flooding of downstream areas (Merrin; Karim et al. 2012, 2014). Lagoons were to have 100 m2 of surface area per hectare of their contributing catchment and a depth of up to 3 m to provide adequate detention time. Spoil excavated from a lagoon could be spread on adjacent low-lying areas to improve their productivity. Lagoons were also designed to be steep-sided to reduce weed growth, vegetated, and well-connected to the downstream river to provide on-farm fish habitat. As a result, a series of lagoons were created via the Riversdale-Murray Scheme for the specific purpose of providing both agricultural and biodiversity benefits (Merrin).

With over 15 years of wetland maturation and stakeholder hindsight, the Riversdale-Murray Scheme provides an opportunity to evaluate the impacts of catchment-integrated wetlands, designed primarily for improved drainage and flow regulation, on sugarcane profitability, fish biodiversity, and an array of ecosystem services more broadly. In this study, we aimed to:

  1. (1)

    Quantify the fish biodiversity, and associated water quality, provided by scheme-funded wetlands;

  2. (2)

    Estimate the return on investment and the benefit to cost ratio from the perspective of a representative landholder investing in wetland construction via the Riversdale-Murray Scheme; and

  3. (3)

    Identify the range of final ecosystem services potentially provided by the Scheme-funded wetlands.

Methods

Our study involved a mixed-method approach, whereby ecological and economic analyses were carried out concurrently. The ecological assessments were primarily informed by field surveys and laboratory analysis; while the economic analyses were primarily informed by financial records, landholder surveys, and climatic and satellite remote sensing data.

Study Area

The Riversdale-Murray scheme occurred across the Tully-Murray floodplain (Queensland), approximately 140 km south of Cairns and 200 km north of Townsville. Tully is often the wettest region of Australia, with an annual average rainfall of greater than 4000 mm (Fig. 1). The Tully and Murray rivers begin with headwaters in mountainous tropical rainforest protected as several National Parks, approximately 50 km (Euclidian distance) from the coastline. The lowland reaches drain flat agricultural land that primarily supports sugarcane farming. Across the Tully River—Murray River Floodplains there is approximately 43.3 km2 of DIWA nationally important wetlands, largely comprised of palustrine and riverine systems, covering 30.9 km2 and 9.7 km2, respectively (Environmental Protection Agency 2005; Department of Environment and Science 2019).

Fig. 1
figure 1

Locations of created wetlands sampled within the Tully-Murray catchment, Queensland. Topographic base map sourced from State of Queensland (Department of Resources) (2021)

Full size image

Fish Biodiversity and Water Quality

Fish assemblages were surveyed at twelve lagoons in October 2019 and repeated at nine lagoons in June 2020 (Fig. 1), to assess community composition and structure, from which species richness was estimated. The GPS locations and landholder details are withheld to respect privacy, though dimensions summarized in Table 1. Fish assemblages were surveyed using a Smith-Root 2.5 GPP generator boat-mounted electrofishing unit. Surveys involved a single pass navigation of the entire edge habitat (where most fish reside) and zig-zagging through the middle until no new species were observed after 30 min of sampling. Pusey et al. (1998) demonstrated, in the nearby Johnstone and Mary Rivers, that most species can be detected within a single pass. The electrofishing unit supplied a pulsed direct current at 30–60 Hz with a duty cycle between 10 and 25%, settings were adjusted depending on a site’s electrical conductivity, depth and species composition. After assessing electrical conductivity, the output was gradually increased until fish were mildly stunned without incurring physical harm. Fish surveys were completed in accordance with the Queensland Animal Care and Protection Act 2001, and JCU animal ethics permit number A2178. All fish were measured (standard length in mm) and identified according to Allen et al. (2002). Sampling was non-destructive with all fish returned to the water, apart from non-native species which were retained and euthanised in accordance with Queensland’s Biosecurity Act 2014.

Table 1 Descriptive statistics of scheme-funded lagoons as catalogued in Riversdale-Murray Valley Water Board archival data, along with details for a mock-landholder considered representative of the participants involved used for analysis purposes
Full size table

At each site, grab samples were collected to assess for an array of nutrients, sediment, and toxicants from three random locations (Tables S1 & S2). At the same time, a calibrated multi-parameter sensor (Quanta, Hydrolab Corporation) was used to profile (at 0.25 m intervals, down to 3.50 m where possible) the pH, electrical conductivity (EC), temperature and dissolved oxygen (Table S3). Collection was made with the vessel facing into the current, at 15–30 cm subsurface, with sterile containers and gloves. Care was taken to ensure that the bottom sediment was not disturbed and that surface films were not collected, in accordance with the Queensland sampling manual (Department of Environment and Science 2018). While animal movement may affect bottom water quality surveys, having three random replicate samples enables some encapsulation of variability. Water samples for filterable nutrients were syringe-filtered on site with an unused disposable plastic 60 mL syringe, 0.45μm Sartorius minisart filters, and were kept on ice until late freezing within 8 h, and eventually processed at a NATA accredited laboratory (American Public Health Association (APHA) et al. 2005; DERM 2009).

A calibrated multi-probe data logger (Hydrolab HL7, OTT HydroMet Ltd) was deployed in the near-surface water layer (0.2 m below the surface) at wetland sites to measure diel periodicity (cycling) of these physico-chemical parameters (water temperature, dissolved oxygen, conductivity, pH) at 20 min intervals (Table S4). All loggers remained at a site overnight to measure diel patterns.

Nutrient, sediment, and toxicant samples were compared against the relevant water quality objectives prescribed by the Environmental Protection (Water and Wetland Biodiversity) Policy 2019: Tully River, Murray River and Hinchinbrook Island Basins Environmental Values and Water Quality Objectives, or the Australian and New Zealand Environment and Conservation Council (2000) trigger values, where available, for parameters not covered by the policy. While dissolved oxygen concentrations were assessed against the north Australian default guideline values by Butler and Burrows (2007).

Landholder Financial Assessment

Here we evaluate the return on investment that a mock-landholder, considered representative of the participants, obtained from the costs incurred in constructing a representative scheme-subsidised lagoon on a medium-sized cane farm. Using a mock-landholder allows for a generally applicable financial assessment while avoiding the need to disclose real landholder financial information, following Human Ethics approval requirements. Prior to the scheme, frequently flooded land only supported light cattle grazing or sugarcane production with a maximum of two ratoon crops. Return on the private investment in lagoon construction is calculated from the estimated financial benefits of improved drainage across the farm as a whole, and the increased gross margins under two scenarios arising from elevating adjacent land with dredge spoil: (1) conversion of cattle grazing to cane growing, or (2) lengthening of the existing cane production cycle from two to four ratoon crops after the initial plant cane crop.

The cost incurred by the landholder comprises 33% of the construction cost of their on-farm drainage lagoon (the remaining 67% of construction costs being covered by the scheme subsidy), plus any on-going maintenance costs. Landholder’s costs were incurred as a combination of in-kind and cash costs. Data on construction and maintenance costs were sourced from archived Riversdale-Murray Valley Water Board lagoon and drainage designs, farm compliance plans, and from semi-structured stakeholder interviews (N = 5, Note S1). Benefits, in terms of gross margin, were estimated using the online Farm Economic Analysis Tool (FEAT) developed by the Queensland Department of Agriculture and Fisheries (State of Queensland Department of Agriculture and Fisheries 2020).

Construction and Maintenance Costs

Construction costs for individual cost elements in farm-scale drainage designs were extracted (in 2003 AUD $) from archived Riversdale-Murray Valley Water Board farm compliance plans for 44 drainage features (drainage lagoons and excavated silt traps) on 16 properties. Data extracted was used to estimate a function for the total cost (Scheme subsidy plus farmer contribution, with the farmer contribution reported in 2003 AUD $ as a combination of in-kind and cash) of constructing farm-scale drainage lagoons using the linear regression model in below equation:

$$Cost_i = \alpha + \beta _1Vol\_Excavated_i + \beta _2Vol\_Excavated\_Sq_i + \beta _3SurfaceArea_i + \varepsilon _i$$
(1)

where the subscript i is an index for lagoon; α is a constant term; β1, β2 and β3 are coefficients to be estimated for the cost drivers: volume excavated (m3) as linear and squared variables, (Vol_Excavated and Vol_Excavated_Sq) and lagoon surface area in (m2) (SurfaceArea), respectively. ε1 is the error term which is assumed to be normally distributed. Equation (1) was then used to estimate the cost of constructing a representative lagoon with a size of 0.3 ha, (Table 1), escalating the estimated cost to 2019 AUD $ using the Quarterly Consumer Price Index for Brisbane, Queensland (Australian Bureau of Statistics 2020).

Maintenance costs were estimated from the semi-structured farmer interviews (N = 5), in which farmers were asked to quantify the operations undertaken and the annual costs incurred in maintaining their scheme-funded wetland (Note S1). Maintenance costs for the representative wetland were based on these responses, again escalated to 2019 AUD ($) using the Quarterly Consumer Price Index for Brisbane, Queensland.

Drainage

The extent and frequency of cane inundated by water was derived using Digital Earth Australia’s ‘Water observations from Space’ dataset from 1986 to 2019 (Geoscience Australia; Mueller et al. 2016). This dataset reports the number of times over a specified period that 25 m × 25 m pixels have been (i) observed and (ii) deemed to have been covered by water when viewed with multi-band imaging from Landsat satellites every 16 days. Cane land extent (13,606 ha) was derived from the Queensland Land Use Mapping Program (QLUMP) dataset across the Riversdale-Murray Valley Drainage Scheme area (Queensland Government 2015). This dataset was split into pre- (1986–2004) and post-drainage scheme (2005–2019) segments to estimate the proportion of cane land in the scheme area for which inundation time reduced after implementation of the Riversdale-Murray Scheme. For each segment, the frequency of pixel observation, and the frequency of those observations in which pixels were ‘wet’ were calculated. The frequency of both observations was then normalised per year to adjust for the different durations of the pre- and post-drainage datasets. The percentage of wet observations per pixel, pre- and post-scheme establishment, was then used to estimate the area of cane land across which inundation duration reduced or increased. As potential changes in rainfall may have influenced inundation patterns, we also used a Welch two-sample t-test to compare mean annual rainfall between the two periods. Rainfall data (1986–2019) was sourced from the Bureau of Meteorology (BOM) rainfall gauge at Tully Sugar Mill (monitoring station number 32042), approximately 10 km north-east of the Scheme area (Figure S1). Previous hydrological modelling has also demonstrated the efficacy of the lagoons and drainage network in altering flood regimes (Karim et al. 2012, 2014).

Economic Benefit from Reduced Duration of Inundation

The original business case for the Riversdale-Murray Scheme estimated the planned drainage improvements would reduce average cane losses from inundation of cane land by between 1.7 and 3.4 tonnes per hectare (Merrin). The increase in gross margin per hectare arising from a given increase in cane yield can be estimated using below equation (Canegrowers 2020a):

$$GM_{inc} = \left( {\left( {P_s \times 0.009 \times \left( {CCS - 4} \right)} \right) + C_c - C_h} \right) \times Y_{inc}$$
(2)

where GMinc is the increment to gross margin ($/ha), Ps is the market price of sugar ($/tonne of sugar), CCS is the percentage sugar content per tonne of cane harvested (expressed as a %, i.e., ‘13’ denotes 13%), Cc is the constant term in the cane price formula ($/tonne of cane), Ch is the harvesting cost ($/tonne of cane), and Yinc is the increment to yield (tonnes/ha) (Canegrowers 2020b). Equation (2) was applied to calculate the increment to gross margin for a medium-sized cane farm (150 ha) in the Murray catchment for cane yield increments of 1.7 and 3.4 tonnes per hectare, using the default FEAT market price of sugar, constant term, and cane harvesting cost for 2015 (State of Queensland 2016), escalated to 2019 AUD ($) using the Quarterly Consumer Price Index for Brisbane.

Farm Economic Analysis Tool (FEAT)

The economic benefit from the two productivity improvement scenarios is calculated from the increased gross margin that follows from either: (i) extension of the cane rotation from plant cane plus two ratoons to plant cane plus four ratoons, or (ii) from the switch from cattle fattening to cane production on a plant cane plus four ratoon production cycle. Gross margins from cane are estimated using FEAT (State of Queensland Department of Agriculture and Fisheries 2020), starting from the default 2015 activity and cost profile for cane production in the Tully catchment (the Riversdale-Murray Valley scheme was directly adjacent to this catchment) (State of Queensland 2016), but using cane yields predicted by the APSIM (Agricultural Production Systems Simulator) cane production simulation software for the climate zone, soil type and soil permeability surrounding a representative Riversdale-Murray Valley lagoon (Keating et al. 2003). In estimating the gross margins from cane production, the FEAT default region-specific assumptions and costs (in 2015 AUD) incurred in implementing the full suite of cane production activities were used initially (Table S5), before escalating the resulting gross margin to 2019 AUD ($) using the Quarterly Consumer Price Index for Brisbane. Gross margins from cattle fattening on poorly draining alluvial loam soil in the Tully-Murray catchment were drawn from Roebeling et al. (2007), escalated to 2019 AUD ($) using the Quarterly Consumer Price Index for Brisbane.

Return on Investment

The return on investment a representative landholder would be expected to obtain from the costs they incurred in constructing and maintaining a representative scheme-subsidised lagoon on their cane farm was calculated using discounted cash flow analysis over the 15-year period 2019 to 2034 inclusive. An annual real discount rate of 5% per annum is assumed, in common with much of the literature e.g., Alluvium (2019). For this analysis, all costs and benefits were expressed in 2019 AUD ($). The net present value achieved over the analysis period is given by below equation:

$$NPV = \mathop {\sum}\limits_{i = 0}^{15} {\frac{{B_t}}{{\left( {1 + r} \right)^t}}} - \mathop {\sum}\limits_{t = 0}^{15} {\frac{{C_t}}{{\left( {1 + r} \right)^t}}}$$
(3)

where NPV denotes net present value, t denotes the year within the 15-year analysis time span (commencing in 2019), Bt denotes the monetary benefit arising in year t, Ct denotes the monetary cost incurred in year t, and r denotes the real discount rate (5% per annum). The benefit to cost ratio (BCR) and internal rate of return (IRR) are calculated. The IRR delivered by the representative farmer’s investment in lagoon construction and maintenance was calculated as that discount rate which when inserted in Eq. (3) reduces NPV to zero. IRR reports the return on investment delivered by lagoon construction.

Identification of Final Ecosystem Services

Final ecosystem services are the benefits arising from an ecosystem that flow directly to and are directly used by humans (Boyd and Banzhaf 2007; Johnston and Russell 2011; Mace et al. 2012). Final ecosystem services, and their classification systems, are used to allow more accurate and consistent definitions of ecosystem services, improve communication, and allow more seamless integration with national accounting (Johnston and Russell 2011; Wong et al. 2015; Finisdore et al. 2020). Using the Common International Classification of Ecosystem Services framework (Haines-Young and Potschin 2012), we sought to identify other ecosystem services that are potentially supported by the scheme-funded lagoons that remain as hypotheses with various levels of examination. This exercise only aimed to hypothesize potential services and did not attempt to quantify the benefit of each service. We also assigned pedigree scores for each hypothesized service to indicate our confidence in a service being provided, ranging from 1 (low confidence) to 4 (total confidence), in line with those proposed by Costanza et al (Costanza et al. 1992).

Results

Fish Biodiversity and Water Quality

Across all survey events, 36 native fishes and 4 non-native fishes were observed (Table 2). The mean species richness observed at each survey was 8.2 (range = 5–15; Fig. 2), while each survey caught, on average, 222 individuals (range = 35–838).

Table 2 Fish observed in scheme-funded wetlands surveyed in 2019 and 2020 using electric fishing
Full size table
Fig. 2
figure 2

The maximum number of fish observed in each of the 11 SIIP-constructed lagoons under the Riversdale-Murray Scheme, surveyed using boat electric fishing in 2019 and 2020

Full size image

Most water quality samples exceeded the total nitrogen, ammonia and chlorophyll-a objectives, and all samples exceeded the total phosphorus objective (Table 3; Table S1). While the total dissolved nitrogen objective was largely met (small exceedances) and the total dissolved phosphorus objective was always met. All toxicants (pesticides and herbicides) tested were well within set objectives or guideline trigger values (Table S2), except for one sample at one site which exceeded the diuron trigger value (Note: the ANZECC (2000) disclaims the diuron trigger value as having high uncertainty). Of the 20 diurnal oxygen cycles assessed, all but three events had dissolved oxygen saturations that exceeded the default acute trigger value guideline of 30% (Tables S3 & S4; Butler and Burrows 2007).

Table 3 Summary statistics for water quality parameters measured at 11 SIIP constructed lagoons as grab samples in 2019 and 2020
Full size table

Landholder Financial Assessment

Changes in Farm Practice

During the semi-structured interviews (N = 5, Note S1), cane farmers reported two forms of productivity improvement following elevation of fields adjacent to the drainage lagoon with excavated spoil. Two farmers reported that they had been able to grow cane on these areas prior to elevating them with excavated spoil. However, the typical cane production cycle for these areas prior to elevation comprised two ratoon harvests in addition to the plant cane harvest. Following elevation, this extended to four or five ratoon harvests following the plant cane harvest. A further three farmers reported that they had been unable to grow cane adjacent to the lagoon site before these areas were elevated, previously using them for cattle fattening. After elevation, these areas could support cane production with a cane cycle covering four ratoons following plant cane.

Construction and Maintenance Costs

Using the regression from Eq. (1) (R2 = 0.94, N = 44, Table S6), the total construction cost of a representative 0.3 ha lagoon (depth 3.5 m, excavated volume 10,500 m3) was estimated to be $20,053 in 2003 AUD, (equivalent to $29,900 in 2019 AUD$; Table 1). After the Riversdale-Murray Scheme paid 67% of lagoon construction cost, a representative farmer would have contributed approximately $6684 in 2003 AUD ($9967 in 2019 AUD$) towards this cost as a combination of in-kind and cash.

Farmers indicated, via the semi-structured interviews (N = 5), that annual maintenance of the drainage lagoon was largely limited to one annual application of glyphosate herbicide to control aquatic weeds, particularly the introduced invasive species Hymenachne amplexicaulis and hybrids. Treatment took 2–5 h annually and typically cost between $100 and $1000 (in 2020 AUD$ at the time of the interviews; equivalent to $99 to $990 in 2019 AUD$). Based on this information, we assumed a mid-range annual maintenance cost of $545 (in 2019 AUD$) for a representative lagoon. Other one-off maintenance actions, un-costed here, include maintaining steep edges and planting trees to reduce grass and Hymenachne growth and minor sediment removal.

Drainage

Overall, the scheme resulted in greater water aggregation, with large areas becoming drier, while some areas became wetter (Fig. 3). Specifically, 1192 ha (i.e., 14.5% of cane land in the scheme area or 31,868 grid cells) recorded fewer wet observations, while 58 ha (i.e., 0.4% of cane area or 929 cells) showed increased inundation, after implementation of the Riversdale-Murray Valley Scheme (Fig. 3). There was no significant difference in average annual rainfall for the duration of the ‘Water observations from space’ datasets pre- and post-implementation of the drainage scheme (p = 0.244, Welch 2-sample t test; Fig. S1).

Fig. 3
figure 3

Spatial plots of 25 m × 25 m grid cells from Digital Earth Australia’s ‘Water observations from space’ dataset for which the integer-rounded percentage of ‘wet’ observations (a) decreased [green] and (b) increased [blue] after implementation of the Riversdale-Murray Valley Drainage Scheme. Cane land in the Scheme area is shaded light grey

Full size image

Benefit from Productivity Improvement on Elevated Land

Increasing from a two-ratoon to a four-ratoon production cycle across elevated land increased the estimated gross annual margin from $777/ha to $953/ha, an increase of $176/ha (all gross margins in 2019 AUD$; Tables 4 and 5). Roebeling et al. (2007) estimated that the gross margin from cattle fattening in the Tully-Murray catchment on poorly drained loam soils of alluvial origin could be maximized at a fertiliser application rate of 80 kgN/ha and a stocking density of 1.75 animal units/ha (where one animal unit equals 400 kg live weight). The resulting maximized annuity gross margin is $151/ha (in 2019 AUD$) (Roebeling et al. 2007; reported as $115/ha in 2007 AUD$). Converting from cattle fattening to 4-ratoon cane would, therefore, increase gross margins by $802/ha (in 2019 AUD$). Interviews with farmers indicated that land adjacent to the lagoon was typically elevated by 0.2 m–1.5 m when construction spoil was added. An elevation of 0.84 m is assumed this analysis (with 10,500 m3 of excavated spoil spread over 1.25 ha of land).

Table 4 Predicted gross margins under two and four-ratoon cane production system for a moderate permeability Mari soil in SILO climate zone 1800_14585 under Six Easy Steps fertiliser applications and legume fallow
Full size table
Table 5 Return on investment from a representative farmer’s contribution to the construction and maintenance of a representative scheme-funded lagoon (Table 2) and soil spreading parcel (Table S5)
Full size table

Net Financial Benefit

Reduced inundation from improved drainage across 14% of the area of a representative medium-sized farm was estimated to increase the annual gross margin by $984–1967 for the farm as a whole (in 2019 AUD$; Table 4). When the benefit from reduced inundation across 14% of the farm area is added to the benefit from converting from two-ratoon cane to four-ratoon cane on a representative 1.25 ha elevated soil parcel, less the costs that the farmer incurs in lagoon construction and maintenance, an internal rate of return ranging from 0.0 to 14.5% is achieved, with corresponding benefit:cost ratios of between 0.8:1 and 1.5:1 evaluated at a 5% discount rate over a 15-year timeframe. When the elevated land can be converted from cattle fattening to four-ratoon cane (in addition to the benefit from reduced inundation across the representative farm), the corresponding internal rate of return ranges from 11.9 to 23.5%, with corresponding benefit:cost ratios between 1.3:1 and 2.0:1 evaluated at a 5% discount rate over a 15-year timeframe (Table 5).

Final Ecosystem Services

We identified 22 final ecosystem services, as per the Common International Classification of Ecosystem Services (Haines-Young and Potschin 2012), that are potentially supported by the Riversdale-Murray Scheme wetlands (Table 5). Of these, we have moderate or high confidence that nine of the benefits of these potential services are being realised. These benefits include: provision of wild animals (i.e., fish) for nutrition; water flow regulation; providing habitat for nursery populations and biodiversity; the promotion of both active and passive activities for recreation and mental health; supporting scientific inquiry; improving farm aesthetics; and having bequest value for future generations (Table 6).

Table 6 Final ecosystem services estimated to be provided by wetlands created as part of the Riversdale-Murray Scheme
Full size table

Discussion

Providing for the needs of agriculture and biodiversity in the same landscape is a significant challenge the world over, with heated debate over approaches for balancing competing demands. Here we observed that the integration of constructed lagoons throughout an intensive sugarcane dominated catchment in north Queensland to reducing flooding not only improved sugarcane profitability but provided habitat for freshwater biodiversity and potentially provide numerous other ecosystem services.

We found that the constructed lagoons provided sufficient habitat, water quality and connectivity to support high native fish diversity (Pearson et al. 2013; Karim et al. 2014), including commercially valuable species such as mangrove jack (Lutjanus argentimaculatus) and barramundi (Lates calcarifer), and iconic species such as the saltwater crocodile (Crocodylus porosus). Mangrove jack, for example, spawn near the outer reef and continental shelf, then migrate as juveniles to the shoreline, inhabiting mangrove roots, snags and rocks, gradually moving upriver and into lagoons as they mature. Once mature, after 2–11 years, adults begin to migrate back to spawning areas where they may reside for up to 40 years (Waltham et al. 2019). Barramundi, a diadromous fish, will ingress into coastal freshwater wetlands during wet season floods, to access important nursery habitat and forage in wetlands. As the lagoons were designed to be close to, and well connected to, the mainstem of the Tully River (Karim et al. 2014), migratory species are able to access and use the created habitat. The lagoons were also designed to be steep-sided and at least 1 m deep, which has helped to reduce macrophyte weed growth and provide habitable dissolved oxygen concentrations (Butler and Burrows 2007). As nutrient and phytoplankton (indicated by chlorophyl a) concentrations were high, reducing nutrient runoff into the lagoons may further improve the habitable condition and food web (Dodds and Smith 2016). Despite being artificial and draining intensive agricultural land, the wetlands support freshwater biodiversity and may provide alternative habitat to that lost from the draining of natural wetlands (Canning and Waltham 2021). Further work is required to examine the contribution of the lagoons towards providing habitat and food for aquatic birds and insects and delineating the habitats that support the greatest diversity for these groups.

In addition to the lagoons (and associated drains) supporting freshwater biodiversity, we estimated that they also improved on-farm profitability (from a landholder’s perspective) in two ways: (1) increased cane yield from improved drainage and flow regulation across a portion of the farm, and (2) increased cane yield from land that was elevated using wetland excavation spoil. Improved drainage and flow regulation reduced cane yield loss from waterlogging. Elevation of land with excavation spoil permitted two potential land use change scenarios to arise. The first was the conversion of cattle fattening land to production of sugarcane on a plant cane plus four ratoons cycle. The second was extension of the existing sugarcane production cycle from plant cane plus two ratoons to plant cane plus four ratoons.

It remains unexamined whether the scheme resulted in a net increase or decrease in nitrogen loading to the downstream Great Barrier Reef (GBR). An expansion and intensification of cane farming, along with increased drainage, may result in greater nitrogen leaching and runoff (Thorburn et al. 2011; Fraser et al. 2017). However, the constructed lagoon has the potential to denitrify runoff and offset any increased nitrogen losses (Land et al. 2016; Adame et al. 2019b; Wallace and Waltham 2021), while the increased cane growth, reduced tillage from longer cane cycles (increased ratooning) and reduced fertiliser loss (from flooding) may contribute to lower nitrogen losses (Webster et al. 2012; Skocaj et al. 2013; Thorburn et al. 2017). As the Great Barrier Reef requires substantial reductions in nitrogen loading to improve its ecological health (Brodie et al. 2012; Kroon et al. 2012; Wooldridge et al. 2015), if the approach resulted in a net increase in nitrogen loading to the reef, then decision-makers would need to consider a values trade-off. The health of the Great Barrier Reef, and the associated economic benefits from tourism and fisheries, would be pitted against improved sugarcane profitability and freshwater biodiversity. A net nitrogen reduction, however, would benefit the reef and could help attract funding that improves financial viability.

The extent to which nitrogen is removed by wetlands depends on many factors, including: the concentration and speciation nitrogen inflows; hydraulic loading rate, residence time and efficiency; temperature; wetland size and shape; composition of the ecological community (particularly the vegetation type and density); sediment type and composition; and oxygen concentrations and redox potential (Land et al. 2016; Alldred and Baines 2016; Vymazal 2017). Denitrification occurs in anaerobic conditions (negative oxidation-reduction (redox) potential) and when nitrate is used by denitrifying bacteria in respiration. Denitrifying bacteria are those with either the nirS, nirK, and nosZ genes, and use oxidised nitrogen compounds as a terminal electron acceptor in the absence of oxygen. Complete denitrification occurs in optimal conditions and released nitrogen gas as a by-product, whereas sub-optimal conditions lead to incomplete denitrification that releases N2O (a potent greenhouse gas) or NO2 as by-products (Burgin and Hamilton 2007; Martínez-Espinosa et al. 2021; Pinto et al. 2021). Complete denitrification is more probable when C:N ratios are high (e.g., >15–20) and anoxic conditions are persistent, if carbon becomes scarce or the environment becomes oxygenated then incomplete denitrification can occur (Klemedtsson et al. 2005). It is, therefore, essential that constructed wetlands with the goal of reducing nitrogen runoff are designed to have high hydrological residence time, persistent anoxic conditions and high carbon supply if it is to have low nitrous oxide emissions (Land et al. 2016; Oertel et al. 2016; Jahangir et al. 2016; Maucieri et al. 2017). Anoxic conditions can arise when soils are water-logged with minimal mixing, have minimal disturbance (such as mechanical ploughing and animal grazing) (Drewry et al. 2008), and when plant oxygenation rates are low as aquatic plants often oxygenate the soils proximal to their roots via aerenchyma transport (Oertel et al. 2016; Jahangir et al. 2016; Maucieri et al. 2017). Root oxygenation rates differ with functional guilds, growth stage, root density and depth, and the abundance of aerenchyma in tissue (Sorrell and Brix 2013; Alldred and Baines 2016). Given that the Tully-Murray catchment has substantial nitrogen runoff to the GBR, further research is recommended to evaluate the efficacy of the constructed wetlands in denitrifying runoff and whether improved design and management of the lagoons could lead to greater nutrient removal or retention. Given that denitrification requires systems to be anoxic, designing wetlands to support this function would require a values trade-off as anoxic conditions are not conducive to supporting biodiversity. It may be more appropriate for any future PES schemes to incentivise wetlands for denitrification in some instances and incentivise wetlands for biodiversity in other instances (Canning et al. 2021).

While these lagoons were profitable from a landholder’s perspective when a farm-wide benefit from reduced inundation and the Scheme’s 67% subsidy were included alongside the benefits from improved productivity on the elevated land adjacent to the lagoon, future schemes seeking to create wetlands in a similar way will also likely need to subsidize works. For lagoon creation to be profitable for a landholder, payments would likely need to ensure farmers achieve a benefit to cost ratio of at least 1:1 over the evaluation period (15 years in this study). To achieve a benefit to cost ratio of 1:1 over 15 years without the subsidy for lagoon construction and without including benefits from reduced inundation, but still accounting for the improved productivity on land elevated with excavated spoil, payments would need to be $8076/ha of wetland (in 2019 AUD$) when elevated land can be converted from cattle fattening to four-ratoon cane rotation, or $10,684/ha of wetland (in 2019 AUD$) when elevated land is converted from two-ratoon to four-ratoon cane rotation. Achieving a benefit to cost ratio of 1:1 over a shorter timeframe would be advantageous. It may be possible for future schemes to cover these costs by securing payments for ecosystem services, particularly if the payment scheme has sufficient flexibility to support multiple services, including those benefits that are non-rival and non-excludable (Canning et al. 2021). However, the annual payment rates required to achieve a benefit to cost ratio of 1:1 over a 15-year evaluation timeframe are much higher than the annual gross margins achieved from the land uses that preceded wetland conversion (by a factor of 14 with 2 ratoon cane as the prior land use, or by a factor of 54 with cattle fattening as the prior land use).

While we anticipate the lagoons providing, with various levels of confidence, at least 22 final ecosystem services (Haines-Young and Potschin 2012), not all services provide easily quantifiable benefits to clearly identifiable beneficiaries. Examples include the removal of nutrients and sediment, improvements to physical and mental health, and harvests from transient fisheries. Challenges in quantifying benefits and clearly attributing beneficiaries can make it difficult for these services to be recognized and rewarded through market-like schemes (Costanza et al. 2021). Having a scheme, that funds investment into wetland creation/restoration that does not rely heavily on benefit quantification, can accommodate multiple, bundled ecosystem services (including non-excludable and non-rival services), and is viable long-term to support ongoing maintenance, such as that facilitated by a common asset trust, may be the best option going forward (Canning et al. 2021; Costanza et al. 2021).

The results obtained here indicate a significant need to consider the retention and restoration of well-connected wetlands within the sustainable development of sugarcane landscapes. Wetlands have substantial ability to improve agricultural flood resilience, while providing wildlife habitat and ecosystem services. Recently, Saunders et al. (2022) identified barriers to the uptake of nature-based solutions, such as wetlands restoration, within Australia, along with key actions to address barriers. Recommendations included: (1) developing fit-for-purpose permitting processes for ecological restoration; (2) improving integrated mapping and classification of coastal ecosystems; (3) conducting research into the effective and risks of using restoration as nature-based solutions; (4) developing national-scale restoration guidelines that can cascade to state and local levels, including guidelines to support climate-resilient restorations; (5) develop decision-support models to help inform which actions to take under what circumstances; and (6) adapt the Restoration Opportunities Assessment Methodology (ROAM) to inform a systematic approach towards prioritization of restoration (Saunders et al. 2022). With respect to restoring wetlands for flood control, further work would be required to identify and prioritize locations where wetlands could be restored and yield a positive return on investment. If returns included accounting for other ecosystem services, then there may also be opportunities for funding from payment for ecosystem service schemes (Canning et al. 2021).

While this study demonstrates the potential benefits of an integrated catchment-scale wetlands restoration scheme, benefits may not be readily transferrable to other catchments without further assessment. Future schemes should use catchment-scale multi-property hydrological modelling to determine the ideal wetland sizes and positions on low-value agricultural land for regulating flood flows. Multi-criteria analysis could then be used to inform site selection by weighting locations and sizes that support the provision of other ecosystem services, such as carbon sequestration, nitrogen removal and supporting biodiversity.

In addition to using hydrological modelling to inform wetland design and connectivity for flood regulation, consideration should also be given to the hydrological connectivity needs of native fish assemblages. Across the studied lagoons, the temporal variation in fish assemblages is significantly influenced by lagoon connectivity with the downstream river, distance from the coast and flood dynamics (Karim et al. 2014; Arthington et al. 2015; Godfrey et al. 2016). Future schemes, particularly those in the Wet Tropics region, should still ensure there is a maintenance of seasonal patterns of flow and connectivity. Furthermore, wetlands should be designed to prevent the growth of exotic ponded-pasture grasses (such as Hymenachne and Brachiaria mutica) as these habitats supported the lowest fish species richness compared with other lagoon habitats (Arthington et al. 2015). This would include ensuring depth is greater than the tolerance of ponded-pasture grasses, riparian zones are well shaded with vegetation, and frequent flushing flows are maintained.

Conclusion

Damage from flooding is a major risk for crop production, particularly in locations where agricultural land is established on previously drained wetland with high rainfall, and where rainfall is expected to become more extreme with climate change. Nature-based solutions, such as wetlands restoration, provide an alternative to conventional flood protection, such as establishing levee banks, as an avenue for achieving reduced flooding risk while supporting other values, such as biodiversity. In the present study, we demonstrate how the use of 44 lagoons and excavated silt traps over 16 properties substantially reduced crop flooding across Australia’s wettest catchment. Reduced flooding increased landholder gross margins for sugarcane cropping by allowing for a longer sugarcane cycle and the conversion of low-intensity cattle grazing areas into sugarcane. Further, we observed the wetlands providing for biodiversity by supporting 36 native fish species, along with the potential provision of 22 final ecosystem services. By rewarding the provision of ecosystem services in PES schemes, we show how the strategic restoration of wetlands for catchment-scale flood control can be a profitable nature-based solution. Those seeking similar benefits in other locations should make use of hydrological modelling to appropriately size and position wetlands across the landscape to achieve flood reduction goals and provide for the connectivity and flood regime needs of fish fauna. Further work is needed to quantify other benefits, such as carbon sequestration and water quality improvement, are recommended as they may support the development of viable supplemental income pathways under emerging PES schemes.