Isotopic Evidence for Multi-Decadal Shifts

Home , Burnett River, Fish scale, Hucho taimen

Received: 2 June 2018 Revised: 21 September 2018 Accepted: 24 September 2018 DOI: 10.1002/rra.3369

SPECIAL ISSUE PAPER

Looking to the past to ensure the future of the world's oldest living vertebrate: Isotopic evidence for multi‐decadal shifts in trophic ecology of the Australian lungfish

Julian D. Olden1,2 | Stewart J. Fallon3 | David T. Roberts4 | Tom Espinoza5 | Mark J. Kennard2

1 School of Aquatic and Fishery Sciences, University of Washington, Seattle, Abstract Washington, United States Meeting the conservation challenges of long‐lived animal species necessitate long‐ 2 Australian Rivers Institute, Griffith term assessments of trophic ecology. The use of dietary proxies, such as ratios of University, Brisbane, Queensland, Australia naturally occurring stable isotopes in animal tissues demonstrating progressive 3 Research School of Earth Sciences, The Australian National University, Canberra, growth, has shown considerable promise to reconstruct trophic histories of long‐ Australian Capital Territory, Australia lived organisms experiencing environmental change. Here, we combine innovative 4 Queensland Bulk Water Supply Authority (Seqwater), Brisbane, Australia radiocarbon scale‐ageing techniques with stable isotope analysis of carbon and nitro- 5 Queensland Department of Natural gen from cross sections of scale to reconstruct the trophic ecology of Australian lung- Resources, Mines and Energy, Queensland fish (Neoceratodus forsteri) across its remaining global distribution. Over a 65‐year Government, Brisbane, Queensland, Australia δ13 δ15 Correspondence period, we found pronounced temporal shifts in the C and N isotopic ratios J. D. Olden, School of Aquatic and Fishery of lungfish that coincided with a period of hydrological modification by dams and Sciences, University of Washington, Seattle, ‐ WA, USA 98105. land use intensification associated with agriculture and livestock grazing. In the Email: [email protected] Brisbane and Burnett Rivers, whose hydrology is substantially regulated by large Funding information dams, lungfish showed consistent trends of δ13C depletion and δ15N enrichment over American Philosophical Society; Australian Research Council, Grant/Award Number: time. This may indicate anthropogenic changes in background isotopic levels of basal LP130100118; Mohamed bin Zayed Species energy sources and/or that additional seston exported downstream from impound- Conservation Fund; National Geographic Society ments represent a carbon source that was previously unavailable, thus shifting lungfish diet from benthic‐dominated primary production typical of unmodified river systems, to pelagic carbon sources. By contrast, δ13C ratios of lungfish in the unreg- ulated Mary River were more stable through time, whereas δ15N ratios increased during a period of dairy industry expansion and increased application of nitrogen fertilization and then subsequently decreased at the same time that rates of pasture development declined and nutrient inputs presumably decreased. In conclusion, we provide evidence for human‐caused alterations in background isotopic levels and potential changes in availability of benthic versus pelagic energy resources supporting Australian lungfish and demonstrate how detectable trophic signals in long‐lived fish scales can reveal long‐term anthropogenic changes in riverine ecosystems.

KEYWORDS

impoundment, long‐term studies, nutrients, river regulation, scale, stable isotope analysis

1 | INTRODUCTION microchemistry (Campana, 2005), investigations using fish scales have proved useful to document both age and life history patterns (Tzadik Large, long‐lived freshwater fishes face tremendous conservation chal- et al., 2017). Scales share several properties with otoliths such as lenges to ensure their future persistence (Hogan, Moyle, May, Vander incremental growth and the incorporation of chemical constituents Zanden, & Baird, 2004; Jensen et al., 2009; Olden, Hogan, & Vander from the surrounding environment within each increment. In addition, Zanden, 2007). Characterized by slow growth, late maturity, and thus scale growth in most fish species is coupled to somatic growth, low rates of population growth, long‐lived species are intrinsically more resulting in growth cycles being recorded as concentric circuli of calci- vulnerable to environmental change and, ultimately, extinction (Pimm, fied collagen corresponding to age (Hutchinson & Trueman, 2006; Jones, & Diamond, 1988). For example, the Australian lungfish, Trueman & Moore, 2007). Studies have shown that stable isotope Neoceratodus forsteri (Krefft, 1870), listed as “vulnerable” under federal values of slower turnover tissues such as fish scale collagen, correlate legislation, is one of only six surviving lungfish species in the world and well with other calcified and noncalcified tissues (Jardine et al., 2012; may be the oldest living vertebrate on the planet (Tokita, Okamoto, & Fincel, Vandehey, & Chipps, 2012), and have the distinct advantage of Hikida, 2005). Despite ongoing threats and repeated calls to fill critical being a nonlethal alternative to otolith removal. knowledge gaps, very little is known about the implications of long‐term Stable isotope analysis of carbon and nitrogen in fish scales can environmental change for the Australian lungfish (hereafter “lungfish”; provide a chronological repository of long‐term ecosystem changes. Arthington, 2009; Curtis, Dennis, McDonald, Kyne, & Debus, 2012). Previous studies have used isotopic changes to provide evidence for Riparian degradation and nutrient addition associated with agricul- catchment nitrogen loading (Roussel et al., 2014), ecosystem eutrophi- tural land use, and hydrologic alteration from impoundments, modify the cation and reoligotrophication (Grey, Graham, Britton, & Harrod, availability and quality of basal food resources in river ecosystems (e.g., 2009), climate cycles (Satterfield & Finney, 2002), hydrologic change Anderson & Cabana, 2006; Boëchat, Krüger, Giani, Figueredo, & Gücker, (Delong et al., 2011), changing prey availability (Pruell, Taplin, & 2011; Delong, Thorp, Thoms, & McIntosh, 2011; Kennedy et al., 2016). Cicchelli, 2003), interactions with invasive species (Mercado‐Silva, Agricultural land use subject to livestock grazing and cropping practices Helmus, & Vander Zanden, 2009), and fisheries overexploitation increase run‐off of sediments and nutrients leading to anthropogenic (Wainright, Fogarty, Greenfield, & Fry, 1993). Stable isotope analysis nutrient enrichment and altered delivery of allochthonous carbon (e.g., for trophic reconstruction is most often measured by homogenizing Buck, Niyogi, & Townsend, 2004; Hagen, McTammany, Webster, & entire scales or subsampling the scale's growing edge to provide either Benfield, 2010). By intercepting flooding events, dams often reduce a combined lifetime measure, or just the most recent diet proxy, lateral inundation of riparian and floodplain areas downstream, thus respectively (Tzadik et al., 2017). By contrast, cross‐sectional sampling limiting allochthonous inputs to riverine environments (Tockner & across a scale's longitudinal axis has rarely been examined (e.g., Seeley, Stanford, 2002). Similarly, the increased hydrologic stability below dams Miller, & Walther, 2015) but offers new opportunities to sample dis- can lead to excessive growth of autochthonous sources including mac- crete time periods over the life of a fish. The reason being is that rophytes, periphyton, and phytoplankton (Pingram, Collier, Hamilton, well‐calcified layers accrete radially towards the leading edge as new David, & Hicks, 2012). Large impoundments are also a significant layers overlay older layers (Hutchinson & Trueman, 2006). Provided source of planktonic production, which ultimately provides food subsi- scales are thick enough to allow targeted cross‐sectional sampling, dies to downstream aquatic consumers (Doi et al., 2008; Growns, assays of the elemental composition across growth increments can Chessman, Mitrovic, & Westhorpe, 2014). provide a lifetime profile of trophic ecology and previous water chem- Human‐induced environmental change has been shown to modify istry encountered by individual fish (Seeley et al., 2015; Tzadik et al., isotope ratios of actively cycled elements such as carbon and nitrogen 2017). This is the case for lungfish where the large elasmoid scales (Bowen, 2010) and alter the availability of benthic versus pelagic and (in excess of 50 mm from primordia to leading edge) are retained for allochthonous versus autochthonous basal energy resources life, laying down successive new layers of thick collagen material as supporting riverine food webs (Growns et al., 2014; Robertson, Bunn, they increase in body size (Kemp, 2012). Furthermore, by combining Boon, & Walker, 1999; Shannon et al., 2001; Sheldon & Thoms, 2006). isotopic profiles with bomb‐curve radiocarbon ageing (Uno et al., Such shifts in isotope signatures and dominant basal energy sources 2013), the chronological record of lungfish trophic ecology can be can translate to changes in the quantity and quality of resources avail- reconstructed over extended time periods (James, Fallon, McDougall, able to higher trophic level consumers (Delong et al., 2011; Delong & Espinoza, & Broadfoot, 2010; Fallon et al., 2015, n.d.). Thoms, 2016). However, anticipating species responses to long‐term In this study, we combine innovative radiocarbon scale‐ageing environmental change is challenged by the need to accurately quantify techniques with stable isotope analysis of carbon and nitrogen to patterns in trophic ecology over appropriately long time scales. This is reconstruct the multi‐decadal trophic ecology of lungfish. Cross‐ particularly relevant for lungfish; a sedentary and extremely long‐lived sectional analyses of the large lungfish scales were used to derive con- species that can reach ages up to 80 years in the wild (Fallon et al., n. tinuous transects of elemental concentrations beginning from early life d.). Fortunately, the use of dietary proxies has emerged as a powerful (juvenile) extending through to present‐day (adult). Lungfish are basal approach to reconstruct diet histories on ecological and evolutionary trophic level consumers, therefore, any environmental changes that time scales for long‐lived organisms (Davis & Pineda‐Munoz, 2016). influence the isotopic composition of primary producers are likely to Elemental analysis of calcified structures in fish have increasingly translate to changes in the isotopically‐distinct resources incorporated been used to estimate lifetime diet variability of individuals. into lungfish tissues. Trends in isotopic signatures of lungfish over the Although otoliths comprise the majority of studies on chronological past 65 years are examined for the Burnett, Mary and Brisbane Rivers, OLDEN ET AL. 3 respectively, presenting the core of the species' remaining global distribution. Our expectation is that the precise nature of isotopic change over time will be both complex and specific to the different histories of hydrologic alteration and land‐use in these watersheds. Currently, a considerable proportion of the annual flow of the Brisbane and Burnett Rivers is altered by impoundments (large weirs and dams), and all three watersheds have experienced a long but varying history of land‐use change for agricultural and livestock practices (Arthington, 2009). We predict that trends in river impoundment and anthropogenic nutrient enrichment have caused changes in background isotopic levels and a shift in the dominant sources of carbon and nitrogen supporting lungfish populations. Reconstructing trophic histories can help inform management actions seeking to ensure the long‐term persistence of the Australian lungfish.

2 | METHODS

2.1 | Study system

Historically, multiple lungfish species were distributed throughout the central‐eastern portion of the Australian continent, but at present, native populations of the Australian lungfish occur only in the Burnett and Mary Rivers (south‐eastern Queensland) with additional populations from intentional translocations to the Brisbane FIGURE 1 Lungfish sample sites in Burnett, Mary, and Brisbane River and surrounding catchments (Brooks & Kind, 2002; Hughes Rivers, Australia. Also shown are the locations of major reservoirs, et al., 2015). Our study focused on these three rivers, located in the dams, and weirs. Inset displays the location of the study region in eastern fastest growing region in Australia with a population of more than 3 Australia [Colour figure can be viewed at wileyonlinelibrary.com] million people (Figure 1). The Burnett and Brisbane Rivers are highly regulated with a number of large dams and weirs constructed for 2.2 | Lungfish trophic ecology and field collection agricultural and urban water supply and for flood control (Arthington et al., 2000; Marshall, Espinoza, & McDougall, 2015). Collectively, Evidence from wild individuals suggests that lungfish are benthic these impoundments store 196% and 134% of mean annual run‐off omnivores (Pusey, Kennard, & Arthington, 2004). Although no quanti- in the Burnett and Brisbane Rivers, respectively (Table 1). The rate tative dietary data are available in the published literature, bivalves, of water infrastructure development increased progressively from gastropods, and organic detrital material are frequently observed in 1965 in the Burnett River catchment and doubled in the Brisbane the faeces of freshly caught live lungfish. Anecdotal observations also River catchment with the construction of Wivenhoe Dam in 1985 suggest that lungfish diet changes relative to early developmental (Figure 2). The Mary River is less regulated with only a barrage at stages and changes in dentition (Kemp, 1977), possibly becoming ben- the tidal interface, and few smaller dams and weirs on tributary thic generalists early in life. Lungfish in riverine locations predomi- streams (total storage capacity 9% of mean annual discharge). nantly occupy areas of complex instream habitat, such as log jams, All three catchments are affected to similar degrees by nutrient macrophyte beds, and areas of overhanging riparian vegetation (Kind, loading from agricultural land use (Table 1; Kemp, Olley, Ellison, & 2011). They are considered a sedentary species, exhibiting strong site McMahon, 2015). The Burnett River has a significant proportion of fidelity and restricted home ranges, rarely venturing beyond a single its catchment area devoted to agricultural production, involving mostly pool or, occasionally between adjacent connected pools in riverine grazing of native vegetation and modified pastures, cropping (sugar- systems (Brooks & Kind, 2002). cane) with some horticulture. Land use in the Mary and Brisbane river Thirty‐five lungfish were collected from eight main channel loca- catchments is similar, though with more intensive dairy production tions in the Mary, Burnett, and Brisbane Rivers (Table 1; Figure 1) and horticulture, respectively, compared with the Burnett River. Key during the period April–October (2014) using non‐destructive boat pollutants include nutrients and chemicals from the horticulture and electrofishing. All locations were in the lower river reaches below sugar industries, sediment and particulate nutrients from grazing, solu- any major impoundments when present. A single undamaged, ble nutrients from dairy, and nutrient and organic matter pollutants nonregenerated, scale was removed from the ventral/lateral side for from cattle grazing within the stream channel and riparian zone. age determination and stable isotope analysis. Following scale Management of nutrient run‐off from farms has improved markedly removal, these lungfish were released alive and unharmed at the point in the past decade after the termination of the Australian Dairy of capture. In addition, four frozen lungfish obtained from an earlier Pasture Subsidy Scheme in the late 1970s. However, cattle grazing study (D. Roberts, unpublished; size range: 258–890 mm; weight within streams and riparian zones still occurs in some areas. range: 100–4740 g) were utilized for a comparative analysis of tissue 4 OLDEN ET AL.

TABLE 1 Catchment environmental attributes, anthropogenic threats, and lungfish specimen summaries for the Burnett, Mary, and Brisbane Rivers, Australia

River Attribute Unit Burnett Mary Brisbane Catchment Catchment area km2 33,273 9,510 13,654 Mean annual discharge (MAQ)a GL/year 803 1,515 1,401 Anthropogenic threats Cumulative storage capacity % MAQ 196 9 134 Natural environmentsb % Area 8.9 19.6 15.5 Agriculture and plantationsc % Area 47.6 38.9 41.3 Intensive usesd % Area 1.1 5.5 9.4 Total nitrogen concentratione mg L−1 (mean ± SD) 0.62 ± 0.24 0.42 ± 0.17 0.63 ± 0.30 Lungfish Sample size Number 6 5 24 Body length mm (mean, range) 776 (610–923) 1,168 (730–1,340) 792 (495–1,128) Age (estimate) Years (mean, range) 11 (5–18) 39 (18–53) 23 (4–54) aModelled mean annual discharge (MAQ) for the period 1970–2008. Source: Stein, Hutchinson, and Stein (2014). bLand used primarily for conservation purposes, based on the maintenance of the essentially natural ecosystems present. Source: ABARES (2016). cLand used mainly for primary production, based on dryland farming systems (includes grazing modified pastures, cropping, and nonintensive horticulture). Source: ABARES (2016). dLand subject to extensive modification, generally in association with closer residential settlement, commercial, or industrial uses (includes intensive horticulture and animal husbandry). Source: ABARES (2016). eMean (±SD) nutrient concentration data (mg L−1) from periodic (approximately quarterly) samples collected in the downstream main channel of each river during the period 1998–2015. Source: Department of Natural Resources and Mines Water Monitoring Information Portal (https://water‐monitoring.information.qld.gov.au/). types to evaluate the association between carbon and nitrogen been reported previously (James et al., 2010; Fallon et al., 2015, n.d.) isotopic ratios derived from scale, fin, and muscle tissues. and is summarized here. A single undamaged whole scale for each specimen was dried and the position of the primordium (origin of growth) was assigned from agreement between two readers and 2.3 | Age determination from scales

Traditional methods for ageing fish have shown limited utility for lungfish. Lungfish demonstrate highly variable growth rates depending on environmental conditions and growth slows considerably with age; therefore, body length is not a reliable indicator of age (Kemp, 1986). Scales of lungfish exhibit apparent growth lines, but these lines vary considerably in thickness and spacing (Kemp, 2005), and become indeterminable as age increases (Brooks & Kind, 2002). Lungfish exhibit a crystalline otolith structure that appears to possess layers, however, these incremental growth lines are not thought to coincide with growth increments according to Gauldie, Dunlop, and Tse (1986). Similarly, tooth plates show growth lines in dentine and enamel but wear continually from the occlusal surface (Kemp, 2003), so the earlier lines are lost with age. Skull bones can contain lines of bone deposition, varying in thickness and intensity, but lines are diffi- cult to distinguish in the dense areas where the bone was first laid down (Kemp, 1999). In addition to not being suitable for ageing, the use of all these calcified structures require samples from a euthanized fish, which is undesirable with a threatened species. This study capitalizes on recent advances in radiocarbon dating FIGURE 2 River regulation by impoundments (dams and weirs) over of lungfish scales. Specifically, this approach utilizes the pulse of time in the Burnett River (squares and blue dashed line), Mary River 14 atmospheric radiocarbon ( C) produced by nuclear weapons testing (circles and black solid line), and Brisbane River (diamonds and red and incorporated into the scale structure to determine the lifespan dotted line), Australia [Colour figure can be viewed at of each individual. The specific methodology and age validation have wileyonlinelibrary.com] OLDEN ET AL. 5 marked on a scanned image of the scale. A Dremel® drill with a dia- Britton, 2015; Kelly, Hagar, Jardine, & Cunjak, 2006; Trueman & Moore, mond encrusted bit was used to remove the outer layers of the scale 2007). To confirm these correlations for lungfish, we conducted stable along the growth axis, isolating the elasmodin from the possible over- isotope analysis (following the methods described above) for replicate print of the heavily mineralized squamulae (Fallon et al., 2015). Once samples of scale, fin, and muscle tissues obtained from two captive the outer later was removed, the sample was rinsed with MilliQ™ raised, and two wild caught individuals. Scale samples were collected water to soften the elasmodin, and a scalpel used to slice transverse from the outer edge of the scale to ensure the most recent material grooves as thinly as possible to achieve an end sample size ~0.5 mm was compared with fin and muscle samples with presumed shorter in width and approximating ~1–1.5 mg dry weight. Samples were turnover rates. Tissue samples were rinsed with distilled water and placed in clean 6‐mm‐diameter quartz tubes with CuO and Ag wire then dried in an oven at 60°C for at least 24 hr before grounding the and then combusted at 900°C for 4 hr. Conversion to graphite was samples to a fine powder in an electric ball‐mill grinder (Retsch achieved in the presence of Fe powder and H2 gas, where the water MM200, Haan, Germany). A quantity (1–2 mg) of the powdered sample 2 is removed during reaction with Mg (ClO4) . Samples were run on was weighed out into small tin capsules and analysed for C and N the single stage accelerator mass spectrometer at the Research School isotopes with a Micromass Isoprime EuroVector EA300 mass spec- of Earth Sciences, The Australian National University. trometer at the Australian Rivers Institute Stable Isotope Laboratory. As a first approximation of age and growth rate, F14C was plotted against distance from the outer edge (James et al., 2010). From each | scale, 1 mm slices were taken from the outer edge to the primordium 2.5 Statistical analysis and ~10 of these samples from throughout the scale were measured Time series of δ13C and δ15N concentrations of sectioned lungfish 14 from each scale. Previous research has shown that the scale F C scales were visually inspected for temporal patterns, and linear‐mixed 14 13 reproduces the shape of the atmospheric F CCO2 curve but does effects models (LMMs) were used to assess overall changes in δ C not have the same absolute values (James et al., 2010). Therefore, a and δ15N ratios over time using the lme4 R package (Bates, Maechler, radiocarbon reference curve was produced using prebomb values from Bolker, & Walker, 2015). In these analyses, river and year were 14 old fish, the highest F C value measured in all fish and the outer included as a fixed factors, and individual lungfish were included as a 14 sample from known age fish (Fallon et al., n.d.). The scale F C values random factor. This analysis controls for the effect of different sample were then calibrated using this reference curve and the radiocarbon periods represented by each fish specimen to evaluate the overall calibration programme (OxCal 4.1; Bronk Ramsey, 2013). The resulting linear trends in δ13C and δ15N ratios through time for each river. calendar ages were used in conjunction with the von Bertalanffy Nonparametric Mann–Kendall tests were also employed to detect growth function to provide an age/year for each fish using the monotonic trends for each specimen (Mann, 1945). Significance following equation (von Bertalanffy, 1938): testing was performed using block bootstrapping (i.e., sampling within ‐ 5 year time windows) in order to account for the serial correlation −ktðÞ−t LtðÞ¼L∞ 1 − e 0 ; present in the isotope time series. Mann–Kendall τ statistic (reported as the standardized Z‐statistic) is defined as the proportion of up‐ movements against time versus the proportion of down‐movements where L(t) is total length along growth axis at time t (age in year), L∞ ‐ the maximum length, k the growth constant that describes the initial (across all possible pairwise time differences), and values falling outside the 95% confidence interval according to 5,000 bootstrap slope, and t0 the theoretical year at zero length. samples were considered statistically significant. All analyses were conducted in the R programming environment, version 3.4.4. 2.4 | Stable isotope analysis

Stable isotope sample preparation and analysis of all lungfish scales followed standard protocols (see Fry, 2006; Jardine et al., 2012). A 3 | RESULTS 0.5 mg subsample from the 0.5–1 mm radiocarbon slices of scale was weighed, transferred to a tin capsule, and analysed for C and N Isotopic composition of scale edges, fin, and muscle tissues were isotopes. Isotope analyses were performed via combustion and mass positively associated for lungfish. Corroborating previous studies for spectrometry using a Sercon Europa EA‐GSL inlet with a Sercon Hydra a variety of fish species, δ13C of scales were found to be slightly 20–22 isotope ratio mass spectrometer at the Radiocarbon Laboratory, enriched over fin and muscle tissue by approximately 1–4‰, with The Australian National University. Ratios of stable isotopes (13C/12C smaller differences observed between scale and fin tissues (Figure 3 and 15N/14N) were calculated as parts per thousand (‰) relative to a,b). By contrast, δ15N of scales show smaller differences when com- international standards (Pee Dee Belemnite carbonate and atmo- pared with fin and muscle tissues (Figure 3c,d). Isotopic composition spheric nitrogen) and expressed as “δ.” Lipid correction was not per- of replicate samples from individual fish was highly consistent formed on δ13C values, as most samples had C/N ratios <4, indicative within each tissue type, showing little variability across different body of low lipid content (Jardine et al., 2012). Machine measurements were locations where the tissues were taken. precise, within ±0.1 ‰ SD for δ13C, and ± 0.1 ‰ SD for δ15N. Multidecadal trends in isotopic composition of lungfish scales Previous studies have found that the δ13C and δ15N of fish scale were evident across all three river systems. In the Brisbane River, collagen are correlated with fin and muscle tissues, indicating that overall δ13C ratios of lungfish showed a significant decrease over the scales can be used to interpret dietary sources (e.g., Busst, Bašic, & last half‐century (LMM estimated slope = −0.054‰ per year, 6 OLDEN ET AL.

FIGURE 3 Relationship between δ13C(a and b) and δ15N (c and d) derived from scale edge, fin, and muscle tissues for four lungfish (different symbols). Symbols and bars represent the mean and SD from multiple samples for each tissue type, including scale edge (back, middle, and front), fin (pectorial, pelvic, and caudal [top, middle, and bottom]), and muscle (back, middle, and front of fish)

t = 9.850, p < 0.001; Figure 4a). Lungfish δ13C showed some evidence to 0.80, all p < 0.05) were detected in each of two oldest individuals. of a step decrease in the mid‐1990s, when ratios consistently fell Similar to the Brisbane River, lungfish from the Burnett River below −20‰ and then declined subsequently to between −24‰ showed a consistent δ15N enrichment of 3–4‰ (LMM estimated and −26‰ in present‐day. Negative trends were detected for 17 slope = 0.178‰, t = 4.956, p < 0.0001); this increase was more out of 24 individuals, of which 12 specimens demonstrated statisti- evident in the period after 2005 (Figure 4d). Statistically significant cally significant monotonic trends over time (Mann–Kendall Z‐statis- positive trends were detected for four of the six individuals tic = −0.31 to −0.89, all p < 0.05). The remaining seven individuals (Mann–Kendall Z‐statistic = 0.66 to 0.82, all p < 0.05). showed no evidence for monotonic trends over time, and largely rep- Lungfish from the Mary River also demonstrated a decrease in resented younger fish with depleted δ13C ratios that reflect the δ13C ratios over time (Figure 4e), although the rate of this decline present‐day values of older fish. A significant increasing overall trend was the lowest of all river systems and significantly less than in in δ15N of lungfish was also evident in the Brisbane River (LMM esti- the Brisbane River (LMM estimated slope = −0.025‰ per year, mated slope = 0.041‰ per year, t = 7.137, p < 0.001), demonstrating t = −3.273, p < 0.01). All individuals, with the exception of one showed gradual enrichment of 3–4‰ over this same time period (Figure 4b). no statistical evidence for monotonic trends over time (Mann–Kendall Significant positive trends were detected in eight out of 24 individuals Z‐statistic = −0.47 to 0.08, all p > 0.05). In contrast to the Brisbane and (Mann–Kendall Z‐statistic = 0.42 to 0.94, all p < 0.05), whereas only Burnett Rivers, the Mary River showed a significant linear decrease in four individuals demonstrated significant negative trends (Mann–Ken- δ15N ratios over time (LMM estimated slope = −0.040‰, t = −8.686, dall Z‐statistic = −0.44 to −0.61, all p < 0.05). p < 0.001). However, there were some clear non‐linear temporal Broadly similar patterns in isotopic composition of lungfish were patterns, with an increase evident during the period 1960–1980, found in the Burnett River, where progressive depletion of δ13C ratios followed by a marked decrease from levels commonly exceeding were apparent over the past 20 years (Figure 4c), and this rate of 16‰ to around 13‰ in the present‐day (Figure 4f). This strong decrease was significantly greater than in the Brisbane River (LMM non‐linearity in δ15N ratios over time lead to all individuals, with the estimated slope = −0.160‰ per year, t = −3.975, p = <0.001). There exception of one showing no statistical evidence for monotonic trends was some indication that δ13C ratios shifted to being relatively more (Mann–Kendall Z‐statistic = −0.28 to 0.21, all p > 0.05). depleted over the life span of the two oldest individuals, whereas δ13C of all young fish were depleted relative to the values of younger samples from the late 1990s. However, this conclusion is based on a 4 | DISCUSSION more limited sample size (n = 6) and smaller time period for aged lungfish. Significant negative trends (Mann–Kendall Z‐statistic = −0.60 to Despite being globally recognized as a scientific icon and listed as a −0.69, all p < 0.05) and positive trends (Mann–Kendall Z‐statistic = 0.50 “vulnerable” species under the Australian Environment Protection and OLDEN ET AL. 7

FIGURE 4 Temporal changes in δ13C and δ 15N ratios of individual lungfish (lines) in the Brisbane River (a and b), Burnett River (c and d), and Mary River (e and f), Australia [Colour figure can be viewed at wileyonlinelibrary.com]

Biodiversity Conservation Act 1999, the long‐term persistence of the share many of the same limitations associated with conducting tempo- Australian lungfish remains uncertain. Here, we show that hydrological ral isotope analyses of museum specimens (e.g., Delong & Thoms, 2016; modification by dams and land‐use changes associated with agricul- Grey et al., 2009; Pruell et al., 2003; Roussel et al., 2014). Perhaps most ture and livestock grazing may have left a distinct trophic fingerprint importantly is the fact that accurate estimates of temporal baseline var- on the last remaining native populations of Australian lungfish. This iation are critical for distinguishing whether changes in consumer stable study is among the first to use cross‐sectional sampling of carbon isotope ratios are due to dietary shifts or simply to variation in and nitrogen stable isotopes across known time periods in scale isoscapes due to changing biogeochemical processes that control isoto- growth increments of a long‐lived species. pic fractionation (Solomon, Carpenter, Rusak, & Vander Zanden, 2008). Our approach to measuring long‐term temporal changes in trophic For instance, a decrease in consumer δ15N over time could indicate a dynamics provides unique advantages over traditional short‐term food shift by that consumer to feeding at a lower trophic position, or it might web analyses by integrating large scale processes over sufficiently long indicate a change in the inorganic N source utilized by primary pro- time periods (Hadwen, Spears, & Kennard, 2010). However, it does ducers. Unfortunately, assembling complete time series of baseline 8 OLDEN ET AL. stable isotope data is often extremely challenging or impossible for to a pelagic‐based diet following dam construction. For example, trends long‐term studies (including this study) because of limited representa- in fish δ13C over different periods of hydrological modification for five tion of aquatic primary consumers in many museum collections. In addi- major river basins of the United States was highly variable (Delong & tion, hindcasting isotopic values based on simple environmental proxies Thoms, 2016). A progressive shift in carbon isotope ratios in response risks oversimplifying the complex hydrological and biological processes to environmental change is somewhat expected given the likely slow that influence the dynamics of stable isotope fractionation (Hanson, isotopic turnover of tissue in relation to changing resource signatures. Jones, & Harris, 2018). In summary, the most parsimonious explanation Although subtle, it is interesting to note that δ13C of lungfish showed for our results is that anthropogenic changes in background isotopic some evidence of a step decrease in the mid‐1990s, when values con- levels of basal energy sources and changes in their availability are sistently fell below −20‰ and then declined further to between −24‰ reflected in changes in carbon and nitrogen stable isotope ratios of and −26‰ in present‐day. This pattern corresponds to the fully lungfish. operational Wivenhoe Dam, where reservoir pelagic plankton δ13C In the Brisbane and Burnett Rivers, where hydrology is substan- values range between −24‰ to −31‰ (D. Roberts, unpublished data). tially regulated by dams, lungfish showed consistent trends of δ13C In the less regulated Mary River, δ13C of lungfish were more sta- depletion and δ15N enrichment over time. An extensive body of liter- ble through time, potentially reflecting the limited water infrastructure ature has shown that hydrologic alteration and impoundment due to development and more natural carbon sources of this catchment. It is dams can modify the transportation and production of both allochtho- also notable that lungfish δ13C in the Mary River were more depleted nous and autochthonous carbon sources (Hoeinghaus, Winemiller, & overall in comparison to the two regulated rivers. This may reflect Agostinho, 2007). First, the attenuation of high flows below dams catchment‐specific mechanisms operating to influence isotopic frac- can influence patterns of autochthonous primary production by mod- tionation during carbon assimilation by primary producers (e.g., local ifying the exchange of organisms, nutrients, and detritus between carbon supply, photosynthetic rates, and water turbulence) and pri- floodplain and river (Cross et al., 2011; Kennedy et al., 2016). Second, mary consumers (e.g., molluscs and crustaceans; Finlay, 2003; Finlay, considerable recycling of both carbon and nitrogen occur within the Power, & Cabana, 1999; Hadwen et al., 2010), which are the major anoxic stratified layers of impoundments, resulting in the fractionation food sources of lungfish. Together with potential differences in the of stable isotopes producing depleted δ13C (Schilder et al., 2017) and availability of these food resources, these factors collectively may con- enriched δ15N (Burford et al., 2011) values. Recent measurements tribute to the between‐river differences observed in δ13C of lungfish. from impoundments within the study region demonstrate they pro- In addition to the potential for impoundments to alter riverine car- + duce considerable quantities of methane and NH4 , from the anoxic bon and nitrogen signatures, agricultural land use (e.g., grazing and sediments and hypolimnion (Musenze, Werner, Grinham, Udy, & Yuan, cropping) and associated riparian degradation can increase run‐off of 2014), confirming the potential for carbon depletion and nitrogen sediments and nutrients leading to anthropogenic nutrient enrichment enrichment processes to influence stable isotope values of food webs (Buck et al., 2004; Hagen et al., 2010). Anthropogenic N inputs can be in these systems. Third, studies have shown that enhanced autotro- assimilated by primary producers and transferred to higher trophic phic production in large reservoirs can lead to considerable quantities levels, resulting in elevated δ15N of aquatic biota (Roussel et al., of additional planktonic primary production and organic matter 2014). There is limited quantitative information on temporal changes (seston) being exported downstream via releases or spilling events in agricultural land use and nutrient inputs in the three study catch- (Marty, Smokorowski, & Power, 2009; Smokorowski et al., 2011). ments, however, all have similarly high relative proportions of land Living and detrital phytoplankton and zooplankton production from devoted to agriculture and intensive land uses (including horticulture impoundments may represent a carbon source to downstream food and animal husbandry; Table 1), although the Mary River had a higher webs that was previously unavailable and typically exhibit a character- proportion of land dedicated to dairy cattle production. In addition, all istic δ13C signature that is more depleted than terrestrial primary three catchments have relatively high measured total nitrogen concen- production and benthic primary production (de Junet et al., 2009; trations (>0.4 mg L−1; Table 1), which approach and often exceed Solomon et al., 2011), and these signatures can be detected in down- water quality guideline thresholds for ecosystem protection in lowland stream food webs (e.g., Chen & Jia, 2009; Kaymak et al., 2015; south‐eastern Australian rivers (ANZECC & ARMCANZ, 2000). Smokorowski et al., 2011). Taken together, our results for the Bris- We observed gradual but distinct enrichment of δ15N(3–4‰) bane and Burnett Rivers suggest that reduced autochthonous carbon over time in lungfish scales from the Brisbane and Burnett Rivers, contributions below dams, coupled with increased availability of which are likely due to increasing water infrastructure development reservoir‐derived autochthonous carbon sources released from dams, and agricultural land use intensification over time. Previous studies may have caused lungfish to increasingly rely upon pelagic‐dominated suggest that dams are net exporters of nitrogen and have demon- from benthic‐dominated food sources over time. strated δ15N enrichment of benthic algae (Growns et al., 2014) and Marked trends in lungfish isotope ratios were evident over a period higher consumers downstream of dams (Beutel, 2006; Duda, Coe, of increased hydrologic modification in the Brisbane River, including Morley, & Kloehn, 2011). By contrast, δ15N of lungfish from the Mary the construction of two large dams: Somerset Dam (completed 1953, River gradually increased during the period 1960–1980, followed by a fully operational 1958) and Wivenhoe Dam (completed 1984, fully marked decrease to present‐day. This transition coincided with the operational in 1988). Carbon isotope ratios of lungfish decreased con- peak in the dairy industry in the Mary River catchment and likely sistently over time, which is supported by other long‐term studies indi- corresponds to high rates of N fertilizer application for pasture (Rees, cating a gradually, but notably variable, shift from a benthic‐based diet Minson, & Kerr, 1972). Subsequently, the termination of the Australian OLDEN ET AL. 9

Dairy Pasture Subsidy Scheme in the late 1970s saw declines in the Beutel, M. W. (2006). Inhibition of ammonia release from anoxic profundal rate of pasture development, reductions in the application of nitrogen sediments in lakes using hypolimnetic oxygenation. Ecological Engineer- ing, 28, 271–279. fertilizers, and smaller dairy cattle populations (Mannetje, 1984). Boëchat, I. G., Krüger, A., Giani, A., Figueredo, C. C., & Gücker, B. (2011). ‐ Collectively, reduced agricultural run off, improvements in wastewater Agricultural land‐use affects the nutritional quality of stream microbial treatment, and less organic inputs from cattle accessing stream corri- communities. FEMS Microbiology Ecology, 77, 568–576. dors have likely resulted in reduced nutrient inputs to aquatic food Bowen, G. J. (2010). Isoscapes: Spatial pattern in isotopic biogeochemistry. webs and changing background nitrogen isotopic levels that is Annual Review of Earth and Planetary Sciences, 38, 161–187. reflected in lower δ15N for lungfish in recent decades. Bronk Ramsey, C. (2013). Recent and planned developments of the program OxCal. Radiocarbon, 55, 720–730. Meeting the conservation challenges of long‐lived species requires Brooks, S. G., & Kind, P. K. (2002). Ecology and demography of the ecological investigations that span appropriately long time scales. By Queensland lungfish (Neoceratodus forsteri) in the Burnett River, combining innovative radiocarbon scale‐ageing techniques with stable Queensland with reference to the impacts of Walla Weir and future isotope analysis of carbon and nitrogen from cross‐sectional sampling water infrastructure development. Queensland Department of Primary of scales, this study demonstrates pronounced temporal shifts in the Industries, Fisheries Division, Brisbane, Report No. QO02004. ‐ dominant energy sources available to lungfish over the last half‐ Buck, O., Niyogi, D. K., & Townsend, C. R. (2004). Scale dependence of land use effects on water quality of streams in agricultural catchments. century. The implications of these dietary shifts and changing carbon Environmental Pollution, 130, 287–299. ‐ and nitrogen sources for the long term persistence of Australian lung- Burford, M. A., Revill, A. T., Palmer, D. W., Clementson, L., Robson, B. J., & fish are still uncertain and further research is needed to reveal what Webster, I. T. (2011). River regulation alters drivers of primary produc- the implications of long‐term shifts in basal energy sources may pose tivity along a tropical river‐estuary system. Marine and Freshwater Research, 62, 141–151. for a large bodied, slow growing, low trophic level consumer. Busst, G. M. A., Bašic, T., & Britton, J. R. (2015). Stable isotope signatures and trophic‐step fractionation factors of fish tissues collected as ACKNOWLEDGEMENTS non‐lethal surrogates of dorsal muscle. Rapid Communications in Mass Spectrometry, 29, 1535–1544. We thank Steven Brooks for providing fish scales from the Burnett Campana, S. E. (2005). Otolith science entering the 21st century. In Marine River and comments on the manuscript, Andrew McDougall and and Freshwater Research, 56, 485–495. Sharon Marshall for assistance with lungfish sampling in the Mary Chen, F., & Jia, G. (2009). Spatial and seasonal variation in δ13C and δ15N River, and Jane Hughes, Dan Schmidt, Peter Kind, and Nick Bond for of particulate organic matter in a dam controlled subtropical river. River advice and participation in the broader project team. Special thanks Research and Applications, 25, 1169–1176. to Ryan Burrows and Ben Stewart‐Koster for assistance with statisti- Cross, W. F., Baxter, C. V., Donner, K. C., Rosi‐Marshall, E. J., Kennedy, T. … cal analyses. Comments from two anonymous reviewers improved A., Hall, R. O., Rogers, R. S. (2011). Ecosystem ecology meets adap- tive management: Food web response to a controlled flood on the the final manuscript. Lungfish sampling was conducted in accordance Colorado River, Glen Canyon. Ecological Applications, 21, 2016–2033. with the provisions of Queensland General Fisheries Permit 174232 Curtis, L. K., Dennis, A. J., McDonald, K. R., Kyne, P. M., & Debus, S. J. S. and the Griffith University Animal Ethics committee (project reference (2012). Queensland's threatened animals. Collingwood: CSIRO Publishing. ENV/17/14/AEC). Davis, M., & Pineda‐Munoz, S. (2016). The temporal scale of diet and dietary proxies. Ecology and Evolution, 6, 1883–1897. ORCID de Junet, A., Abril, G., Guérin, F., Billy, I., & de Wit, R. (2009). A multi‐ tracers analysis of sources and transfers of particulate organic matter Julian D. Olden http://orcid.org/0000-0003-2143-1187 in a tropical reservoir (Petit Saut, French Guiana). River Research and Applications, 25, 253–271. REFERENCES Delong, M. D., & Thoms, M. C. (2016). Changes in the trophic status of fish ABARES—Australian Bureau of Agricultural and Resource Economics and feeding guilds in response to flow modification. Journal of Geophysical Sciences. (2016). Land use of Australia 2010‐11. Available at: http:// Research – Biogeosciences, 121, 949–964. www.agriculture.gov.au/abares/aclump/Pages/land‐use/data‐down- Delong, M. D., Thorp, J. H., Thoms, M. C., & McIntosh, L. M. (2011). Tro- load.aspx phic niche dimensions of fish communities as a function of historical Anderson, C., & Cabana, G. (2006). Does δ15N in river food webs reflect hydrological conditions in a Plains river. River Systems, 19, 177–187. the intensity and origin of N loads from the watershed? Science of the Doi, H., Chang, K.‐H., Ando, T., Imai, H., Nakano, S., Kajimoto, A., & Katano, Total Environment, 367, 968–978. I. (2008). Drifting plankton from a reservoir subsidize downstream food ANZECC & ARMCANZ. (2000). Australian and New Zealand guidelines for webs and alter community structure. Oecologia, 156, 363–371. fresh and marine water quality. Australian and New Zealand Environ- Duda, J. J., Coe, H. J., Morley, S. A., & Kloehn, K. K. (2011). Establishing ment and Conservation Council and the Agriculture and Resource spatial trends in water chemistry and stable isotopes (δ15N and Management Council of Australia and New Zealand, Canberra. δ13C) in the Elwha River prior to dam removal and salmon recoloniza- Available at: http://www.agriculture.gov.au/water/quality/guidelines tion. River Research and Applications, 27, 1169–1181. Arthington, A. H. (2009). Australian lungfish, Neoceratodus forsteri, threat- Fallon, S. J., McDougall, A., Espinoza, T., Roberts, D. T., Brooks, S., & Kind, ened by a new dam. Environmental Biology of Fishes, 84, 211–221. P. (2015). Updated methods to age the Australian lungfish: Reply to – Arthington, A. H., Brizga, S. O., Choy, S. C., Kennard, M. J., Mackay, S. J., Kemp (2015). Radiocarbon, 57, 195 196. McCosker, J. L., & Zalucki, J. M. (2000). Environmental flow require- Fallon, S. J., McDougall, A. J., Espinoza, T., Roberts, D. T., Brooks, S., Kind, ments of the Brisbane River downstream from Wivenhoe Dam. South P. K., … Hughes, J. (n.d. in revision). Age structure of the Australian East Queensland Water Corporation, Brisbane, and Centre for Catch- lungfish (Neoceratodus forsteri). PLoS One ‐ ment and In Stream Research, Griffith University. ISBN 0 86857 989 0 Fincel, M. J., Vandehey, J. A., & Chipps, S. R. (2012). Non‐lethal sampling of Bates, D., Maechler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed‐ walleye for stable isotope analysis: A comparison of three tissues. effects models using lme4. Journal of Statistical Software, 67,1–48. Fisheries Management and Ecology, 19, 283–292. 10 OLDEN ET AL.

Finlay, J. C. (2003). Controls of stream water dissolved inorganic carbon Kemp, A. (2003). The ultrastructure of developing tooth plates in the dynamics in a forested watershed. Biogeochemistry, 62, 231–252. Australian lungfish, Neoceratodus forsteri. Tissue & Cell, 35, 401–426. Finlay, J. C., Power, M. E., & Cabana, G. (1999). Effects of water velocity on Kemp, A. (2005). New insights into ancient environments using dental algal carbon isotope ratios: Implications for river food web studies. characters in Australian Cenozoic lungfish. Alcheringa, 29, 123–149. Limnology and Oceanography, 44, 1198–1203. Kemp, A. (2012). Formation and structure of scales in the Australian lung- Fry, B. (2006). Stable isotope ecology. New York: Springer. fish, Neoceratodus forsteri. Journal of Morphology, 273, 530–540. Gauldie, R. W., Dunlop, D., & Tse, J. (1986). The remarkable lungfish oto- Kemp, J., Olley, J. M., Ellison, T., & McMahon, J. (2015). River response to lith. New Zealand Journal of Marine and Freshwater Research, 20,81–92. European settlement in the sub‐tropical Brisbane River, Australia. Grey, J., Graham, C. T., Britton, J. R., & Harrod, C. (2009). Stable isotope Anthropocene, 11,48–60. analysis of archived roach (Rutilus rutilus) scales for retrospective study Kennedy, T. A., Muehlbauer, J. D., Yackulic, C. B., Lytle, D. A., Miller, S. W., of shallow lake responses to nutrient reduction. Freshwater Biology, 54, Dibble, K. L., … Baxter, C. V. (2016). Flow management for hydropower 1663–1670. extirpates aquatic insects, undermining river food webs. BioScience, 66, Growns, I., Chessman, B., Mitrovic, S., & Westhorpe, D. (2014). The effects 561–575. of dams on longitudinal variation in river food webs. Journal of Freshwa- Kind, P. K. (2011). The natural history of the Australian lung!sh – ter Ecology, 29,69 82. Neoceratodus forsteri (Krefft 1870). In J. M. Jorgensen & J. Joss Hadwen, W. L., Spears, M., & Kennard, M. J. (2010). Temporal variability of (Eds.), ‘The Biology of Lung!shes’ (pp. 61–96). En!eld, NH: CRC Press. δ13 benthic algal C signatures influences assessments of carbon flows in Krefft, G. (1870). Description of a giant amphibian allied to the genus – stream food webs. Hydrobiologia, 651, 239 251. Lepidosiren, from the Wide Bay District, Queensland. Proceedings of Hagen, E. M., McTammany, M. E., Webster, R., & Benfield, E. F. (2010). the Zoological Society of London, 221–224. Shifts in allochthonous input and autochthonous production in streams Mann, H. B. (1945). Nonparametric tests against trend. Econometrica, 13, – along an agricultural landuse gradient. Hydrobiologia, 655,61 77. 245–259. Hanson, N., Jones, E. L., & Harris, R. N. (2018). Multi‐decadal and ontoge- Mannetje, L.'t. (1984). Pasture development and animal production in netic trophic shifts inferred from stable isotope ratios of pinniped Queensland since 1960. Tropical Grasslands, 18,1–18. teeth. Oikos, 127, 134–146. Marshall, S. M., Espinoza, T., & McDougall, A. J. (2015). Effects of water Hoeinghaus, D. J., Winemiller, K. O., & Agostinho, A. A. (2007). Landscape‐ level fluctuations on spawning habitat of an endangered species, the scale hydrologic characteristics differentiate patterns of carbon flow in Australian lungfish (Neoceratodus forsteri). River Research and large‐river food webs. Ecosystems, 10, 1019–1033. Applications, 31, 552–562. Hogan, Z. S., Moyle, P. B., May, B., Vander Zanden, M. J., & Baird, I. G. Marty, J., Smokorowski, K., & Power, M. (2009). The influence of fluctuat- (2004). The imperiled giants of the Mekong. American Scientist, 92, ing ramping rates on the food web of boreal rivers. River Research and 228–237. Applications, 25, 962–974. Hughes, J. M., Schmidt, D. J., Huey, J. A., Real, K. M., Espinoza, T., Mercado‐Silva, N., Helmus, M. R., & Vander Zanden, M. J. (2009). The McDougall, A., … Roberts, D. T. (2015). Extremely low microsatellite effects of impoundment and non‐native species on a river food web diversity but distinct population structure in a long‐lived threatened in Mexico's central plateau. River Research and Applications, 25, species, the Australian lungfish Neoceratodus forsteri (Dipnoi). PLoS 1090–1108. One, 10(4), e0121858. Musenze, R. S., Werner, U., Grinham, A., Udy, J., & Yuan, Z. (2014). Meth- Hutchinson, J. J., & Trueman, C. N. (2006). Stable isotope analyses of col- ane and nitrous oxide emissions from a subtropical estuary (the lagen in fish scales: Limitations set by scale architecture. Journal of Fish Brisbane River estuary, Australia). Science of the Total Environment, Biology, 69, 1874–1880. 472, 719–729. James, K. M., Fallon, S. J., McDougall, A., Espinoza, T., & Broadfoot, C. (2010). Assessing the potential for radiocarbon dating the scales of Olden, J. D., Hogan, Z. S., & Vander Zanden, M. J. (2007). Small fish, big ‐ Australian lungfish (Neoceratodus forsteri). Radiocarbon, 52, 1084–1089. fish, red fish, blue fish: Size biased extinction risk of the world's freshwater and marine fishes. Global Ecology and Biogeography, 16, 694–701. Jardine, T. D., Pettit, N. E., Warfe, D. M., Pusey, B. J., Ward, D. P., Douglas, M. M., … Bunn, S. E. (2012). Consumer‐resource coupling in wet‐dry Pimm, S. L., Jones, H. L., & Diamond, J. (1988). On the risk of extinction. – tropical rivers. Journal of Animal Ecology, 81, 310–322. The American Naturalist, 132, 757 785. Jensen, O. P., Gilroy, D. J., Hogan, Z., Allen, B. C., Hrabik, T. R., Weidel, B. C., Pingram, M. A., Collier, K. J., Hamilton, D. P., David, B. O., & Hicks, B. J. … Vander Zanden, M. J. (2009). Evaluating recreational fisheries for an (2012). Carbon sources supporting large river food webs: A review of endangered species: A case study of taimen, Hucho taimen, in Mongolia. ecological theories and evidence from stable isotopes. Freshwater – Canadian Journal of Fisheries and Aquatic Sciences, 66, 1707–1718. Reviews, 5,85 103. Kaymak, N., Winemiller, K. O., Akin, S., Altuner, Z., Polat, F., & Dal, T. Pruell, R. J., Taplin, B. K., & Cicchelli, K. (2003). Stable isotope ratios in (2015). Stable isotope analysis reveals relative influences of seasonal archived striped bass scales suggest changes in trophic structure. hydrologic variation and impoundment on assimilation of primary pro- Fisheries Management and Ecology, 10, 329–336. duction sources by fish in the Upper Yesilırmak River, Turkey. Pusey, B. J., Kennard, M. J., & Arthington, A. H. (2004). Freshwater fishes of Hydrobiologia, 753, 131–147. north‐eastern Australia (p. 684). Collingwood: CSIRO Publishing. ISBN: Kelly, M. H., Hagar, W. G., Jardine, T. D., & Cunjak, R. A. (2006). Nonlethal 0643069666. sampling of sunfish and slimy sculpin for stable isotope analysis: How Rees, M. C., Minson, D. J., & Kerr, J. D. (1972). Relation of dairy productiv- scale and fin tissue compare with muscle tissue. North American Journal ity to feed supply in the Gympie district of south‐eastern Queensland. of Fisheries Management, 26, 921–925. Australian Journal of Experimental Agriculture and Animal Husbandy, 12, Kemp, A. (1977). The pattern of tooth plate formation in the Australian 553–560. lungfish, Neoceratodus forsteri, Krefft. Zoological Journal of the Linnean Robertson, A. I., Bunn, S. E., Boon, P. I., & Walker, K. F. (1999). Sources, – Society London, 60, 223 258. sinks, and transformations of organic carbon in Australian floodplain Kemp, A. (1986). The biology of the Australian lungfish, Neoceratodus rivers. Marine and Freshwater Research, 50, 813–829. – forsteri (Krefft 1870). Journal of Morphology, 1(Supplement), 181 198. Roussel, J. M., Perrier, C., Erkinaro, J., Niemelä, E., Cunjak, R. A., Huteau, D., Kemp, A. (1999). Ontogeny of the skull of the Australian lungfish & Riera, P. (2014). Stable isotope analyses on archived fish scales reveal Neoceratodus forsteri (Osteichthyes: Dipnoi). Journal of Zoology, 248, the long‐term effect of nitrogen loads on carbon cycling in rivers. 97–137. Global Change Biology, 20, 523–530. OLDEN ET AL. 11

Satterfield, F. R., & Finney, B. P. (2002). Stable isotope analysis of Pacific Tockner, K., & Stanford, J. A. (2002). Riverine flood plains: Present state salmon: Insight into trophic status and oceanographic conditions over and future trends. Environmental Conservation, 29, 308–330. – the last 30 years. Progress in Oceanography, 53, 231 246. Tokita, M., Okamoto, T., & Hikida, T. (2005). Evolutionary history of African Schilder, J., Van Hardenbroek, M., Bodelier, P., Kirilova, E. P., Leuenberger, lungfish: A hypothesis from molecular phylogeny. Molecular Phyloge- M., Lotter, A. F., & Heiri, O. (2017). Trophic state changes can affect netics and Evolution, 35, 281–286. ‐ the importance of methane derived carbon in aquatic food webs. Pro- Trueman, C. N., & Moore, A. (2007). Use of the stable isotope composition ceedings of the Royal Society B: Biological Sciences, 284, 20170278. of fish scales for monitoring aquatic ecosystems. Terrestrial Ecology, 1, Seeley, M., Miller, N., & Walther, B. (2015). High resolution profiles of ele- 145–161. ‐ ments in Atlantic tarpon (Megalops atlanticus) scales obtained via cross Tzadik, O. E., Curtis, J. S., Granneman, J. E., Kurth, B. N., Pusack, T. J., ‐ sectioning and laser ablation ICP MS: A literature survey and novel Wallace, A. A., … Stallings, C. D. (2017). Chemical archives in fishes – approach for scale analyses. Environmental Biology of Fishes, 98, 2223 2238. beyond otoliths: A review on the use of other body parts as chronolog- Shannon, J. P., Blinn, D. W., McKinney, T., Benenati, E. P., Wilson, K. P., & ical recorders of microchemical constituents for expanding O'Brien, C. (2001). Aquatic food base response to the 1996 test flood interpretations of environmental, ecological, and life‐history changes. below Glen Canyon Dam, Colorado River, Arizona. Ecological Applica- Limnology and Oceanography: Methods, 15, 238–263. – tions, 11, 673 685. Uno, K. T., Quade, J., Fisher, D. C., Wittemyer, G., Douglas‐Hamilton, I., Sheldon, F., & Thoms, M. C. (2006). In‐channel geomorphic complexity: Andanje, S., … Cerling, T. E. (2013). Radiocarbon dating of modern tusks, The key to the dynamics of organic matter in large dryland rivers? hair, and teeth. (ed., Vol. 110) (pp. 11736–11741)Proceedings of the Geomorphology, 77, 270–285. National Academy of Sciences. Smokorowski, K. E., Metcalfe, R. A., Finucan, S. D., Jones, N., Marty, J., von Bertalanffy, L. (1938). A quantitative theory of organic growth Power, M., … Steele, R. (2011). Ecosystem level assessment of environ- (inquiries on growth laws II). Human Biology, 10, 181–213. mentally based flow restrictions for maintaining ecosystem integrity: A Wainright, S. C., Fogarty, M. J., Greenfield, R. C., & Fry, B. (1993). Long‐ comparison of a modified peaking versus unaltered river. Ecohydrology, term changes in the Georges Bank food web: Trends in stable isotopic – 4, 791 806. compositions of fish scales. Marine Biology, 115, 481–493. Solomon, C. T., Carpenter, S. R., Clayton, M. K., Cole, J. J., Coloso, J. J., Pace, M. L., … Weidel, B. C. (2011). Terrestrial, benthic, and pelagic resource use in lakes: Results from a three‐isotope Bayesian mixing model. Ecology, 92, 1115–1125. How to cite this article: Olden JD, Fallon SJ, Roberts DT, Solomon, C. T., Carpenter, S. R., Rusak, J. A., & Vander Zanden, M. J. Espinoza T, Kennard MJ. Looking to the past to ensure the (2008). Long‐term variation in isotopic baselines and implications for future of the world's oldest living vertebrate: Isotopic evidence estimating consumer trophic niches. Canadian Journal of Fisheries and for multi‐decadal shifts in trophic ecology of the Australian Aquatic Sciences, 65, 2191–2200. lungfish. River Res Applic. 2018;1–11. https://doi.org/ Stein, J. L., Hutchinson, M., & Stein, J. (2014). A new stream and nested catchment framework for Australia. Hydrology and Earth System 10.1002/rra.3369 Sciences, 18, 1917–1933.