Supplementary materials to “Influence of legacy phosphorus, land use, and climate change on anthropogenic phosphorus inputs and riverine export dynamics”
Part A: Hydroclimate and land-use characteristics of the Yongan watershed
The Yongan watershed is located in one of the most developed regions of China, Taizhou area, Zhejiang province (Fig. S1). The Yongan River ultimately flows into the Taizhou Estuary and East China Sea. The downstream outlet examined in this study is 55 km upstream of the Taizhou Estuary. It has an average annual water depth of 5.42 m and discharge of 72.9 m3 s–1 at the downstream BZA sampling point. Six catchments of the Yongan watershed as defined by the location of the monitoring sites at HX, HB, BZA, ZK, HG and XZ, with areas of 547, 1650, 2474, 218, 35 and 357 km2, respectively. Catchments HB and BZA include all area up to the mainstream sampling sites HB and BZA, respectively. In other word, catchment BZA denotes the entire watershed. There is no river regulation, such as artificial dams/reservoirs and transboundary water withdrawal facilities within the Yongan watershed.
Fig. S1 Spatial distribution of land use in the Yongan River watershed (2009)
Agricultural land (i.e., the sum of paddy field, garden plot, and dry land) averaged ~12% of total watershed area, with developed land (i.e., the sum of rural and urban residential, roads, mining and industrial), woodland, and natural lands contributing ~3%, ~67%, and ~18%, respectively, Fig. S1). Rice, wheat, corn, vegetables, soybean, and potato are major crops cultivated in the agricultural lands. Due to increased development, the areas of woodland and natural lands were decreased, while developed land area was increased over the past 31 years, especially in the 2000s.
The climate of the Yongan watershed is subtropical monsoon with the six catchments having a long-term average annual precipitation of 1308–1463 mm during 1980–2010 (Fig. S2). The rainfall mainly occurs in May−September with the typhoon season usually occurring in July−September (Fig. S3). The Yongan watershed has experienced some of the most significant regional climate change in Zhejiang Province (The People’s Government of Zhejiang Province 2010). There has been a ~23 to ~28 day decrease in the number of rainy days, and a ~65% to ~76% increase in the number of storm events (>50 mm per 24 hr) over the 1980–2010 period. However, there were no significant trends in annual precipitation and average river discharge over the study period (1980−2010) (Fig. S2). The long-term trends in these hydroclimate variables were directly determined by regression analysis between each parameter and year number. Daily precipitation data for the study period were obtained from three weather monitoring stations within the Yongan watershed maintained by the Taizhou City Weather Bureau. Precipitation for each catchment was calculated using the Thiessen polygon method to spatially distribute the single point record from rain gauges, and the average precipitation was calculated by the following formula (Richard 1963):
(S1)
where, P is the mean rainfall of the region whose area is A, and P1, P2,…, Pn indicate rainfall samplings within those polygons having areas A1, A2, …，An.
Fig. S2 Changes of hydroclimatevariables in each of the six catchments within the Yongan watershed from 1980 to 2010: precipitation and river water discharge (a), annual number of rainy days and number of storm events (b)
Fig. S3 Average monthly precipitation in the Yongan watershed in 1980–2010. The error bars refer to standard deviations of measurements from the three weather monitoring sites within the watershed
Part B: Riverine TP flux estimate
River water samples were collected once every 4–8 weeks during the 1980–2010 period at six monitoring sites (Fig. S1). Water samples were collected between 8:00 and 9:00 and total phosphorus (TP) was analyzed in the certified laboratory of the Taizhou City Environmental Protection Bureau. Well mixed water samples (surface and bottom layers at three sites along the cross section) were collected and composited. Duplicate water samples were collected from the composite for analysis. The water samples were acidified with H2SO4 in the field (10 ml of concentrated H2SO4 per liter of sample) and analyzed within 4 hr after sampling. TP concentration (all dissolved and particle phosphorus) in the unfiltered water samples was determined by the spectrophotometric ammonium molybdate method (limit of detection: LOD = 0.01 mg P L-1), recommended by The Ministry of Environmental Protection of the People's Republic of China (State Environmental Protection Administration of China 2002). River discharge was measured once every 2 to 12 hours using the rotating-element current-meters method following the standard method recommended by The Ministry of Water Resources of the People's Republic of China (The Ministry of Water Resources of the People's Republic of China 1995).
To estimate annual TP flux using the discrete monitoring data from 1980–2010, the widely applied LOADEST model was utilized to predict daily TP flux (Runkel et al. 2004):
(S2)
where Ln is the natural logarithm function; Qt is daily average river discharge for a given P monitoring date (m3 s–1), Lt is the measured daily P flux (kg P d–1), which is estimated by multiplying measured P concentration by Qt; t is the decimal time for the corresponding monitoring date; tc is the center of decimal time for the study period (a constant); β0...β6 are the fitted parameters in the multiple regression; β1 and β2 describe the relation between flux and discharge; β3 and β4 describe the relationship between flux and time; and β5 and β6 describe seasonal variation in flux data.
In this study, the LOADEST model parameters shown in Eq. S2 (i.e., β0...β6) were calibrated by least squares fit using Matlab software (version 10.0, The MathWorks, Inc., US, 2010) for each of the six catchments. All calibrated parameter values were statistically significant (p0.05) with average relative errors of ±4%–±15% and high R2 between the modeled and measured TP concentration (Table S1). These results indicate that the established LOADEST models can be reasonably applied to estimate daily TP concentrations in the Yongan watershed (Fig. S4). Here, daily TP load was estimated by multiplying TP concentration and water discharge, and annual TP load was calculated from the sum of daily TP loads for a corresponding year. Annual TP flux was determined by dividing the annual TP load by the total drainage area upstream of each monitoring station.
Table S1 The calibrated LOADEST parameters for TP concentration for each of the six catchments in the Yongan watershed
Catchment / Parameter / β0 / β1 / β2 / β3 / β4 / β5 / β6 / R2HX / Mean / -8.838 / -0.119 / 0.015 / 0.282 / 0.179 / 4.501 / -0.161 / 0.76
(n=176)
95% CI / -12.842 / -0.273 / -0.068 / 0.026 / -0.062 / 2.449 / -0.235
-5.835 / 0.036 / 0.097 / 0.539 / 0.420 / 6.553 / -0.086
HB / Mean / -4.079 / 0.008 / 0.015 / -0.007 / 0.169 / 0.057 / 0.001 / 0.98
(n=174)
95% CI / -4.475 / -0.085 / -0.034 / -0.191 / -0.024 / 0.046 / -0.001
-3.683 / 0.100 / 0.064 / 0.178 / 0.362 / 0.067 / 0.003
BZA / Mean / -3.094 / -0.117 / -0.032 / -0.199 / -0.012 / 0.094 / -0.0004 / 0.78
(n=183)
95% CI / -4.165 / -0.171 / -0.059 / -0.293 / -0.108 / -0.104 / -0.009
-2.023 / -0.062 / -0.004 / -0.105 / 0.087 / 0.292 / 0.008
ZK / Mean / -5.458 / -0.098 / -0.016 / -0.057 / -0.0211 / 0.0100 / -0.002 / 0.92
(n=174)
95% CI / -8.275 / -0.205 / -0.069 / -0.198 / -0.160 / -0.348 / -0.019
-2.640 / 0.010 / 0.038 / 0.084 / 0.118 / 0.547 / 0.016
HG / Mean / -5.457 / -0.201 / -0.003 / 0.038 / 0.247 / 0.732 / -0.033 / 0.87
(n=170)
95% CI / -16.440 / -0.370 / -0.097 / -0.296 / -0.103 / -0.971 / -0.098
5.527 / -0.032 / 0.092 / 0.371 / 0.598 / 2.436 / 0.032
XZ / Mean / -4.912 / -0.221 / 0.019 / -0.513 / 0.239 / 0.227 / -0.005 / 0.78
(n=151)
95% CI / -9.848 / -0.447 / -0.096 / -0.855 / -0.088 / -0.633 / -0.041
0.024 / 0.005 / 0.135 / -0.171 / 0.566 / 1.086 / 0.031
Fig. S4 Measured daily riverine TP load versus the modeled values using the LOADEST model for each of the six catchment in the Yongan watershed
It is well known that nutrient inputs to rivers from point and nonpoint sources demonstrate significant hydrological differences. Point source nutrient input to rivers is relatively constant and hydrologically independent. In contrast, diffuse source nutrient inputs have a strong hydrologic dependence (Chen et al. 2013; Bowes et al. 2014). To address the role of point and nonpoint source P input to rivers, relationships between measured riverine TP concentration and water discharge in the six catchments were evaluated. Among the six catchments of the Yongan watershed, catchment ZK demonstrated a significant positive relationship between river TP concentrations and river discharge, while catchment HG presented a significant negative correlation (Fig. S5). These results indicate that P inputs from nonpoint (erosion and leaching) and point (domestic sewage) sources are the dominant cause of riverine TP in catchments ZK and HG, respectively (Chen et al. 2013; Bowes et al. 2014). No significant correlations were observed in the other four catchments, implying that P inputs from both point and nonpoint source produced comparable contribution to riverine TP flux (Fig. S5).
Fig. S5 The relationship between TP concentration and river discharge for each of the six catchments in the Yongan watershed
Part C: Soil characteristics and management
There are 5 soil groups, including 10 soil subgroups and 130 soil species, in the Yongan watershed. The red, yellow, and lithological soil groups were the dominant soil types and accounted for 64.6%, 15.4%, and 1.5% of total soil area in the watershed, respectively (Agricultural Bureau of Xianju County 2011). The dominant red and yellow soil groups correlate with Oxisols and Ultisols in USDA Soil Taxonomy. The alluvial and paddy soil groups were the dominant soil types in the plain area and accounted for 3.9% and 14.6% of total soil area, respectively. Extensive soil property measurements were conducted in 1984 and 2009 by the Agriculture Bureaus of Xianju County in Zhejiang Province, China (Soil Survey Office of Taizhou City 1987; Agricultural Bureau of Xianju County 2011; Chen and Lu 2013). In 1984, soil properties for the 0–20 cm, 20–40 cm, 40–100 cm soil layers of the different soil types and agricultural land types were hierarchically measured, which was supported by the Second National Soil Census of China. In 2009, soil samples were collected from the same locations as in 1984 and soil properties for the upper 20-cm layer of the different soil and farmland types were measured, which was supported by the National Soil Measurement and Formulated Fertilization Program of China. On average, one composite sample was collected per 15 ha for plain region soils and one sample per 25 ha for hilly region soils. Composite soil samples (20–30 subsamples from each 15–25 ha area) were collected from the upper 20-cm soil layer in October to December following crop harvest. Undisturbed soil sampleswere simultaneously collected in cylindrical cores for measuring soil bulk density. Soil samples were air-dried, milled, and passed through a 2-mm sieve for chemical analysis. Soil total P (TP) content and Olsen P were measured by the H2SO4-HClO4 degradation (all organic and inorganic P in soil sample were transformed into PO4-P in the solution)-molybdenum antimony colorimetric method and NaHCO3 extraction-molybdenum antimony colorimetric method (Agricultural Chemistry Specialty Committee of Soil Science Society Chinese 1983), respectively. Soil pH was measured potentiometrically using a pH meter in a 1:5, soil:distilled water suspension.To be comparable, data measured in the same catchment for upper 20-cm layer in both 1984 and 2009 (Table S2) were used to address the changes in soil properties over the 25 year period between sample collections.
Table S2 Changes of soil properties in the upper 20-cm soil layer of agricultural lands in the Yongan watershed between 1984 and 2009
Catchment / 1984 / 2009pH / Bulk density
(g cm3) / Olsen P
(mg P kg–1) / TP
(mg P kg–1) / pH / Bulk density
(g cm3) / Olsen P
(mg P kg–1) / TP
(mg P kg–1)
HX / Mean / 5.1a / 1.12a / 10b / 211b / 5.3a / 1.11b / 49a / 581a
97.5% / 6.1 / 1.34 / 100 / 611 / 6.4 / 1.35 / 341 / 1220
2.5% / 4.2 / 0.87 / 2.0 / 71 / 4.5 / 0.76 / 9 / 212
HB / Mean / 5.1a / 1.2a / 11.1b / 220b / 5.1a / 1.18a / 41a / 419a
97.5% / 6.2 / 1.54 / 185 / 586 / 6.3 / 1.46 / 312 / 843
2.5% / 4.1 / 0.92 / 1.2 / 44 / 4.2 / 0.86 / 5.1 / 98
BZA / Mean / 5.0a / 1.1a / 15.9b / 255b / 5.1a / 1.15a / 32a / 490a
97.5% / 6.4 / 1.3 / 117 / 870 / 6.3 / 1.24 / 453 / 1232
2.5% / 4.2 / 0.77 / 1.0 / 43 / 4.1 / 0.8 / 7.2 / 109
ZK / Mean / 4.8a / 1.21a / 8.8b / 144b / 4.9a / 1.18a / 56a / 629a
97.5% / 6.0 / 1.45 / 53 / 521 / 5.9 / 1.51 / 184 / 1132
2.5% / 4.0 / 0.82 / 2.0 / 46 / 4.1 / 0.76 / 13 / 123
HG / Mean / 5.2b / 1.21b / 15b / 157b / 5.4a / 1.15b / 49a / 443a
97.5% / 6.3 / 1.72 / 98 / 572 / 6.4 / 1.64 / 287 / 829
2.5% / 4.5 / 0.91 / 1.2 / 26 / 4.7 / 0.89 / 2.2 / 110
XZ / Mean / 5.5a / 1.31a / 24.2b / 241b / 5.1b / 1.21a / 57.2a / 503a
97.5% / 6.3 / 1.81 / 108 / 750 / 6.2 / 1.77 / 470 / 1321
2.5% / 4.8 / 0.86 / 1.3 / 50 / 4.9 / 0.71 / 12 / 128
Lower case letters denote significant differences in soil properties between 1984 and 2009 in each catchment (p<0.01).