Coral skeletons reveal the history of nitrogen cycling in the coastal Great Barrier Reef

Anthropogenic nutrient discharge to coastal marine environments is commonly associated with excessive algal growth and ecosystem degradation. However in the world’s largest coral reef ecosystem, the Great Barrier Reef (GBR), the response to enhanced terrestrial nutrient inputs since European settlement in the 1850’s remains unclear. Here we use a 333 year old composite record (1680–2012) of 15N/14N in coral skeleton-bound organic matter to understand how nitrogen cycling in the coastal GBR has responded to increased anthropogenic nutrient inputs. Our major robust finding is that the coral record shows a long-term decline in skeletal 15N/14N towards the present. We argue that this decline is evidence for increased coastal nitrogen fixation rather than a direct reflection of anthropogenic nitrogen inputs. Reducing phosphorus discharge and availability would short-circuit the nitrogen fixation feedback loop and help avoid future acute and chronic eutrophication in the coastal GBR.


Supplementary Note 1 -Isotope mixing modelwater column
The δ 15 N of total dissolved N in coastal water after the flooding event of early 2019 Where: [TOT] is the total measured TDN concentration in the flood plume samples [TR] is the concentration of terrestrial N in the river runoff [CW] is the concentration of TDN in the coastal water prior to flooding. ). This fractionation value varies between 0% and -2‰ (Erler et al. 2015;Wang et al. 2016;Ren et al. 2017). The model assumes that there are two major sources of N to the inshore GBR reefs, terrestrial runoff and N2 fixation (Equations 3 and 4).
By fixing the δ 15 N of terrestrial runoff, and matching the calculated δ 15 N-AvailN against measured CS-δ 15 N, we were able to constrain the contribution of terrestrial N and N2 fixation to the N available to the corals during different time periods (Equation 5). The input of N from precipitation and upwelling are considered to be relatively minor for inshore corals (Furnas et al. 2011). fixation to the available N pool were 17% (δ 15 N-TR of 7‰) and 55% (δ 15 N-TR of 9.2‰) for ε values ranging between 0‰ and -2‰ respectively. Using the average δ 15 N-TDN of 8.1‰, the contribution of N2 to new N inputs to the coastal GBR ranged between 27% and 50% (for ε values ranging between 0‰ and -2‰ respectively).
For the first 100 years of the CS-δ 15 N record (i.e. pre-European settlement) we repeated the calculations using the average CS-δ 15 N between 1680 and 1780 of 6.4‰. The same δ 15 N-TR from the modern GBR were used for the pre-European calculation. Given that δ 15 N-TR has likely risen since European settlement, the contribution of N2 fixation to the coastal ocean N pool in the first 100 years of the record is an overestimation. The δ 15 N of N produced through N2 fixation was assigned a value of -1‰. With the endmember δ 15 N values set, and using high and low estimates of δ 15 N-AvailN (i.e. ε varying between 0‰ and -2‰), the contribution of terrestrial N and N2 fixation to the N being assimilated by the corals could be calculated (Figure 4 in main text).

Supplementary Note 3 -Estimating N inputs from N2 fixation
The model calculations presented above estimate the fractional contribution of N from N2 fixation to new N entering the coastal GBR, which is then recorded as CS-δ 15 N. To convert this to an actual amount of N from N2 fixation added to the water column each year we need to know the total amount of N that is available to the corals before and after

Equation 6
Where: N-N2 = annual N from N2 delivered to the coastal GBR relative to the N discharge from the Burdekin and Herber Rivers (tonnes yr -1 ) TotalN = the annual N discharge from the Burdekin and Herber Rivers (tonnes yr -1 ) The annul inputs of N from N2 fixation presented here are calculated as a fraction of the terrestrial N exports. Previous measurements of N2 fixation in the coastal GBR are given as areal rates (tonnes km 2 , or mmol m -2 d -1 ) (Bell et al. 1999;Furnas et al. 2011), or volumetric rates (nmol L -1 d -1 ) (Messer et al. 2017). To convert these to an annual N input value we estimated the surface area of the inshore GBR (~ 20 km from land) between the Burdekin and Herbert Rivers (3872 km 2 ) and multiplied the areal rates by this area. For the volumetric conversion we assumed an average depth of 10 m.
Comparing our annual rates with the scaled up literature values is problematic because it assumes that the river discharge affects the same area as is used to scale up the areal rates. However river discharge may extend further than the area enclosed by the dashed line in Fig. 1 (main text). In this case the areal rates would need to be multiplied by a higher km 2 area value.

Supplementary Figures
Supplementary Figure 1

Supplementary Tables
Supplementary Table 1. Results of Pearson correlation tests between CS-δ 15 N from Havannah Island and Pandora Reef, and the reconstructed Burdekin River flow of Lough et al. (2015). The coral records are a composite of the different coral cores for that reef (n = 2).
Annual data has not been filtered. Other data have had different frequency bands removed, <10 means that data was filtered to remove periods of less than 10 years, <25 means that data was filtered to remove periods of less than 24 years, and <50 means that data was filtered to remove periods of less than 50 years.