This study was conducted in northern New Zealand. The city of Auckland is situated on an isthmus between two inlets, the Waitemata Harbour to the north-east and the Manukau Harbour to the south-west. Even though these two inlets are only about 1.6 km apart at their closest point on the isthmus, there is no direct connection between the two, with the Waitemata opening to the Hauraki Gulf and the Manukau to the Tasman Sea. The third main site sampled was Mangawhai harbour located 81 km north of Auckland. The sites were selected because they are in close proximity to anthropogenic activity hot spots (e.g., water treatment plant) and they covered a wide demographic and geographic region. Also, in the present study, an attempt was made to locate mangroves growing in relatively pristine conditions, by obtaining samples from nature reserves on Great Barrier and Motu Kaikoura Islands, located approximately 90 km offshore from central Auckland.

Initially, mangrove leaf samples were collected in 2013 from 18 sites throughout Manukau Harbour, Auckland and nine sites in the Mangawhai Harbour Estuary, Mangawhai. In 2014, leaves were collected from 12 sites in Waitemata Harbour, Auckland. A larger collection was made in 2015, with 30 sites throughout Manukau Harbour, 29 sites in Waitemata Harbour, 10 sites in Mangawhai Harbour and five sites in the Great Barrier Island archipelago. In all years, collections were made during the winter season (April–September) so that the leaves would be mature following the peak summer growing season. Sampling points were recorded with geographic coordinates and are listed in Supplementary Table 1. A total of 10 leaves per tree were sampled from 10 trees spaced approximately 10 m apart at each sampling site. For consistency, the leaves were selected to be fully mature, but not senescent. All leaves were oven dried at 65°C for 3 days.

To compare contemporary and past leaf nutrients levels, historical herbarium mangrove leaves were obtained from the AM specimen collections and compared to contemporary samples from the same locations. When no samples from the same locations could be obtained or exact information about the herbarium sample location was missing, the closest geographical location was sampled or the mean value of the entire site was used. These leaves were mainly from Waitemata Harbour (18 samples), Manukau Harbour (3 samples), and one sample from Great Barrier Island (Supplementary Table 2). The collection dates for the herbarium material ranged from 1863 to 1990. The leaves were analyzed for nitrogen content and nitrogen stable isotope ratio. Total phosphorus contents were not measured in herbarium mangrove leaves, since the amount of leaf sample available from the museum was limited.

Dried composite leaf samples were ground to a fine powder with a ball mill (PM 100, Retsch, © Retsch GmbH) and sifted through a 200 μm pore size sieve. Analysis for total phosphorus (%P) was conducted using wet digestion in concentrated HNO₃ (McQuaker et al., 1979), followed by quantification using an inductively coupled plasma atomic emission spectrometer (ICP-AES, Varian Liberty AX Series II, USA). A National Institute of Standards and Technology (NIST) Peach Leaf standard reference material (SRM1547) was used as a quality control in all analytical batches. A total of 0.1–0.5 grams (g) of each composite leaf sample was sent to the Waikato University Stable Isotope Unit for total nitrogen (%N) and stable isotope (δ¹⁵N) analyses.

Herbarium samples from the AM were brought to the laboratory and ground to a fine powder in the presence of liquid nitrogen. These samples were analyzed using the same procedures as mentioned above. Only one leaf per herbarium voucher could be analyzed, which could have caused some additional variation.

Leaf %N, %P, N/P ratio, and δ¹⁵N values are reported as mean ± standard error (SE). To test for differences between polluted and less polluted harbors using contemporary leaves, ANOVA was used to compare %N, %P, N/P ratio, and δ¹⁵N values between the three harbor locations in the main trial year (2015). Residuals of ANOVA were normal and homogeneous so that no transformations were considered necessary. ANOVAs were followed by Tukey’s HSD tests when significant differences were detected. To test for changes in leaf nutrients over time by comparing contemporary and historical samples, %N, and δ¹⁵N of Herbarium and inter-annual variability samples were tested for normality and t-tests and Wilcoxon rank-sum tests were used accordingly. Inter-annual variability was assessed using t-test, since all variables were normal. For all statistical analyses and plotting, the R software version 3.2.1 was used (R Core Team, 2015).

When comparing parameters measured during the preliminary study years (2013, 2014) with the data from the same locations gathered in the main sampling year (2015), measurements did not differ greatly, except for the nitrogen stable isotope (δ¹⁵N) ratios between Manukau samples (p-value < 0.05, t-test or Wilcoxon test, Table 1).

The highest nitrogen stable isotope value in the historical museum samples was found in the Manukau Harbour at Ihumatao Street point (18.2‰) in the mid-1980s. The highest total nitrogen content was found in Purewa Bush point (3.1%) in the mid-1950s.

In general, nutrient parameters of mangrove leaves found in the present study were lower than historical values, except for two sampling points at the Manukau site (Figure 4). The mean total nitrogen content of herbarium samples was significantly higher at 2.6 ± 0.1%, compared to 2.1 ± 0.1% in the 2015 samples (p-value < 0.001, t-test). The stable nitrogen isotope ratio in the herbarium samples were also significantly higher (7.9 ± 0.4‰) in comparison with values found in the 2015 samples (6.2 ± 0.2‰) (p-value = 0.001, Wilcoxon test).

In general for a variety of plants growing under pristine natural conditions in various areas of the world, the δ¹⁵N of tree leaves ranges from -8 to +3‰ (Peterson and Fry, 1987) and leaf values lying within this range have been observed in A. marina mangroves (1.6‰, 2.2‰) growing on an off-shore island in Australia (Costanzo et al., 2001). In contrast, in the present study, the three studied sites in northern New Zealand had mean δ¹⁵N values ranging from +5 to +10‰, and no single site had a δ¹⁵N mean value within the natural range.

Two potential sources of the nitrogen which may have caused these enhanced δ¹⁵N values are agricultural practices and human sewage. With respect to agriculture, New Zealand has the third highest use of nitrogenous fertilizer in the OECD countries after South Korea and Japan at 27 kg/ha (OECD, 2015). Excess deposition of urea on land from either fertilization or by livestock urination results in an increased probability of kinetic isotope fractionation either via ammonia volatilisation or by microbial denitrification due to system “leakiness” associated with the excess nitrogen (Heaton, 1986; Fry et al., 2003). Thus, both in New Zealand and internationally, high levels of dairy and animal farming are associated with high levels of δ¹⁵N in nitrate produced by microbial processes (Komor and Anderson, 1993; Harrington et al., 1998). Plants, in turn, absorb this ¹⁵N enriched nitrate, leading to elevated δ¹⁵N values in plant tissues. For example in a study of δ¹⁵N and % N in Rhizophora mangrove leaves in Florida, the highest δ¹⁵N and %N values were observed where canals draining agricultural lands deliver high-nitrate waters to fringing mangrove marshes. Mangroves growing adjacent to agricultural canals had δ¹⁵N values in the range +11 to +16‰, whereas mangroves growing at a more pristine site had values that ranged from -5 to +2‰ (Fry et al., 2000). A similar elevation in δ¹⁵N values has been associated with proximity to human sewage sources in both aquatic plants in general (McClelland and Valiela, 1998) and A. marina mangrove stands in particular (Costanzo et al., 2001). While direct measurements of the ¹⁵N signature of water running into the harbors is technically feasible there would be a high intra-day and inter-day variability in such data due to changes in temporal events such as rainfall and land use. The advantage of using mangroves as ecological indicators is that the mangrove acts as a continuous sampler, integrating, and storing ¹⁵N over the lifetime of the leaf. In addition mangroves leaves are more easily accessible than other potential indicators (such as macroalgae) and the trees themselves have long lives, potentially allowing sampling from the same plant over several decades.

Based on the δ¹⁵N values observed at the three sites, it appears that Mangawhai at 5.2‰ is receiving mildly elevated inputs of anthropogenic nitrogen at +2‰ above the upper natural background level of +3‰ (Peterson and Fry, 1987), while the Manukau at +9.9 ‰ is receiving strong anthropogenic inputs at almost +7‰ above the natural background level. The Waitemata Harbour lies between these two values at +6.4‰. The main source of anthropogenic nitrogen at Mangawhai is likely to be from farm animals since Mangawhai has a relatively small human population of 1,329 (Census 2013). In contrast, the city of Auckland, where the Waitemata and Manukau harbours are located, is New Zealand’s largest city with a population of 1.4 million within the wider Auckland region (Census 2013). The highly significant difference in δ¹⁵N values between the two Auckland sites (6.4 vs. 9.9‰) is, therefore, a potential indicator that the Manukau Harbour receives greater anthropogenic nitrogen inputs.

The most likely source for this difference is the main Auckland sewage treatment plant which is located adjacent to the Manukau Harbour and discharges an effluent output of c. 120,000,000 m³ of treated sewage per year into the harbor (with an average total nitrogen concentration of 8.3 gm^-3 in winter and 6.8 gm^-3 in summer; Watercare Service Limited, 2013). Also, the Manukau Harbour has a higher proportion of adjacent farmland compared to the Waitemata Harbour.

The Waitemata and Manukau Harbours have elevated leaf total nitrogen contents compared to Mangawhai Harbour, which supports the δ¹⁵N observations that indicate higher anthropogenic nitrogen inputs at these two sites compared to Mangawhai. The mangrove leaf N data from the present study also correlate with Auckland water quality monitoring data. According to the 2013 Auckland Council Marine Water Quality Annual Report, calculated total inorganic nitrogen concentrations of Manukau Harbour surface water were significantly higher than in the Waitemata Harbour (0.15 and 0.05 mgL^-1, respectively; Walker and Vaughan, 2014).

The majority of studies in natural ecosystems have found that mangrove productivity is primarily limited by nitrogen and occasionally by phosphorus (Reef et al., 2010). A value for the level of nitrogen and phosphorus at which growth limitation of A. marina occurs has not been determined in natural ecosystems since many variables may impact growth (Reef et al., 2010; Alongi, 2011). However, Alongi (2011) studied the effect of nitrogen and phosphorus on the growth of A. marina and five other mangrove species under controlled tidal hydroponic conditions. Plants were grown in seawater with a range of nitrogen concentrations, and the growth rates and leaf nutrient contents were determined. The leaf nitrogen contents of A. marina increased with increasing concentrations of nitrogen in the seawater solutions, and ranged from a low of 1.1% at low nitrogen supplementation rates to a high of 3.4% total leaf nitrogen at very high supply rates. A. marina displayed an S-shaped nitrogen dependent growth curve, which plateaued at a nitrogen supply rate of 10 mmol m^-2d^-1, which resulted in a measured average leaf content of 2.1% nitrogen.

Thus, under otherwise optimal conditions, the growth of A. marina may be nitrogen limited up to a leaf content of slightly less than 2.1% N. In the present study it is noteworthy that the average A. marina leaf nitrogen contents from the two Auckland sites exceed this value, implying that the natural ecological nitrogen limitation on growth would be removed from these mangroves (under optimum salinity conditions). Alongi (2011) commented that the nitrogen solution concentration that gave this supply rate was likely to be the maximum that would be obtained under natural environmental conditions, but might be exceeded in polluted ecosystems, which gives further credence to the δ¹⁵N observations that indicate anthropogenic nitrogen inputs into the mangroves of the two sites close to Auckland.

Generally, nitrogen or phosphorus limitation in plants has been reported as an N/P ratio of less than 10 or greater than 20, respectively (Güsewell, 2004). Furthermore, an absolute value of less than 0.1% phosphorus in leaves is also generally indicative of phosphorus limitation (Güsewell, 2004). Based on these criteria none of the mangrove sites in the three contemporary harbor sites surveyed in the current study showed clear phosphorus limitation. However, while the mean N/P values for the three sites indicate that nitrogen limitation is unlikely to be present overall, several plots at all three sites may be nitrogen limited since 17 of the 69 plots surveyed had N/P ratios of less than 10.

None of the mangroves in the 69 locations sampled in the three main harbor sites had δ¹⁵N values in the -8 to +3‰ range, which is expected under pristine natural conditions (Fry, 2006). Costanzo et al. (2001) observed high δ¹⁵N values in mangrove leaves associated with sewage outfalls, but also observed low values (1.6–2.2‰) on an offshore island located approximately 20 km away from the main site. In contrast, in the present study the leaf δ¹⁵N and %N values from the off shore Great Barrier Island mangroves sites were relatively high, with means of 5.3‰ and 2.2%. Paradoxically, however, the %P values were relatively low with a mean of 0.15% P. This resulted in high N/P ratios, with a mean of 14.9. Indeed, this 14.9 N/P ratio approaches the values of 15–18 seen in old growth New Zealand pristine forests, which are typically phosphorus limited (Richardson et al., 2004; Parfitt et al., 2005). Thus, the phosphorus limitation results at this site suggest that the area might be pristine, while conversely the nitrogen results suggest the opposite.

General trends between herbarium samples and contemporary collections strongly suggest that there has been a decline in nitrogen stable isotope ratios and total nitrogen content in mangrove leaves over the past 100 years in Auckland’s estuaries. This finding is consistent with herbarium studies that have observed a decline in leaf %N and δ¹⁵N in a variety of plant species in other parts of the world over the last century (Peñuelas and Matamala, 1990; Peñuelas and Estiarte, 1997; Peñuelas and Filella, 2001; McLauchlan et al., 2010). The authors in these herbarium studies ascribed the decline in leaf %N and δ¹⁵N to either increased absorption of elevated atmospheric CO₂ related to anthropogenic activities over the last century, or to increased absorption of anthropogenically derived atmospheric nitrogen species depleted in ¹⁵N. However, in the present study, the decline is more likely a result of improvements in the Auckland sewage treatment system over the past 100 years (Fitzmaurice, 2009), potentially leading to reduced inputs of δ¹⁵N and %N.

In contrast to the present study, other herbarium studies used plant samples largely obtained from relatively pristine ecosystems where nitrogen was likely limiting. Under these conditions, elevated CO₂ levels have been shown to cause reductions in plant leaf nitrogen content (McGuire et al., 1995). However, under elevated CO₂ conditions and ample nitrogen availability, both reductions and increases in leaf and plant %N have been observed (McGuire et al., 1995). Likewise, in the herbarium study of Peñuelas and Filella (2001), low (-10 to 0‰) leaf δ¹⁵N was observed in samples from relatively pristine areas in the Mediterranean. In that study, precipitation derived anthropogenic nitrogen species depleted in ¹⁵N were likely to make a large contribution against a low natural background. Atmospheric nitrogen supply has been calculated to represent a high proportion (36–53%) of the N incorporated into biomass in Mediterranean systems due to low soil moisture, in contrast to more temperate ecosystems where nitrogen species, such as ammonium or nitrate, are derived from groundwater (Peñuelas and Filella, 2001). Therefore, in the present study, where consistently high δ¹⁵N values were observed, the major source of nitrogen is likely to be ammonium or nitrate from microbially processed urea in water derived from sewage or farm runoff, which is typically enriched in δ¹⁵N (rather than depleted in ¹⁵N as is atmospheric ammonium or nitrate). The observed historic decline in leaf δ¹⁵N and %N is hence more likely to be related to a reduction in sewage impact rather than to atmospherically derived nitrogen or to elevated CO₂ levels. However, impacts of elevated CO₂ levels over the last century cannot be excluded and the observed decline in leaf %N and δ¹⁵N may well be due to a combination of both elevated CO₂ levels and improvements in the Auckland sewage system.

Pitt et al. (2009) found that mangrove plants, unlike algae and crabs, had no detectable response in δ¹⁵N values to an improvement in sewage outfall quality in the Moreton Bay catchment, Australia, up to 2 years after an upgrade had been completed. They attributed the lack of change to the slower growth rate and turnover time of nitrogen in mangroves compared to algae. Mangroves source most of their N via their roots from sediments that can accumulate large quantities of N. In addition, mangroves are likely to store N for an extended period and unlike algae, mangroves recycle a proportion of their N internally since they resorb up to 64% of N from senescing leaves prior to abscission. Here, we see no or minimal changes occurring over short time periods (1–2 years), but significant changes occurring over longer time periods (decades) as indicated by the herbarium samples. Thus, algae may be more suitable for monitoring short term trends in eutrophication (Pitt et al., 2009) whereas mangroves are long lived and integrate nitrogen signatures across large spatiotemporal scales, making them more suitable for monitoring long term trends.