Cryptosporidium spp. and Giardia duodenalis emissions from humans and animals in the Three Gorges Reservoir in Chongqing, China

Study area and model components

The TGR Area is located at 28°30′ to 31°44′N and 105°44′ to 111°39′E in the lower section of the upper reaches of the Yangtze River. It has a watershed of 46,118 km², reaches approximately 16.8 million residents, and its river system covers 38 Chongqing districts and counties (Fig. 1). Using the area’s population (Fig. 1A) and livestock density (Fig. 1B) (Table S1), we applied the GloWPa-TGR-Crypto model to estimate oocyst and cyst emissions in the Chongqing region of the TGR (Hofstra et al., 2013). We defined an emission as the annual total number of oocysts and cysts excreted by people and livestock found in surface water. We used the emission data from 2013 for our model since the records for human and livestock populations from that year were the most complete. The model (GloWPa-TGR-Crypto) consisted of two components: a human emission model and an animal emission model. Figure 2 shows a schematic sketch of the model’s components. We identified the two types of pollution sources for the total oocysts and cysts found in the TGR. Point sources were human emissions connected to sewage systems that indirectly reached the TGR after treatment, or directly before treatment. Nonpoint sources were emissions from rural residents or livestock resulting from manure being used as fertilizer and entering surface water via runoff. Our model was partly based on Hofstra et al. (2013) and other reviews suggesting improvements in manure treatment of livestock and human emissions (Hofstra et al., 2013; Vermeulen et al., 2015; Vermeulen et al., 2017). We ran the model at both district and county levels. Our model cannot differentiate Cryptosporidium species, as there was the paucity of the prevalence and excretion rates for different species in humans and livestock.

Calculating oocyst and cyst excretion in human feces (H)

Using population (P), sanitation availability (F), and excretion (O_p) data from 2013, the model first estimated human oocyst and cyst emissions. We divided the human populations into four emission categories: connected sources, direct sources, diffuse sources, and non-sources. The detailed descriptions of the emission categories are provided in Kiulia et al. (2015). The model not only calculated the human emissions connected to sewers in urban and rural areas, but it also calculated direct and diffuse emissions. Unlike Kiulia et al. (2015), our model did not differentiate across age categories. In rural areas of China, a portion of fecal waste is collected and used for fertilizer and irrigation. Therefore, we assumed that human feces runoff from septic tanks and pit latrines was a diffuse source of oocysts and cysts in surface water. The two protozoan parasites’ prevalence rate was 10% in developing countries (Human Development Index (HDI) <0.785; Hofstra et al., 2013), and the average excretion rate (O_p) was assumed to be 1.0 × 10⁸ and 1.58 × 10⁸ for oocysts and cysts, respectively (Ferguson et al., 2007; Hofstra et al., 2013). Our model used the secondary sewage treatment according to the Chinese discharge standard of pollutants for municipal wastewater treatment plants (GB18918-2002) and the Chinese technological policy for the treatment of municipal sewage and pollution control (http://www.mee.gov.cn/). The removal efficiencies were 10%, 50%, and 95%, for primary, secondary, and tertiary treatments, respectively (Hofstra et al., 2013). The results (H) were calculated using Eqs. (1) and (2):

$K_i\in \left\{K_1,K_2,K_3,K_4\right\},i\in \left\{1,2,3,4\right\}$, representing four state sets.

K₁—Urban connected emissions

K₂—Rural connected emissions

K₃—Urban direct emissions

K₄—Rural diffuse emissions.

Oocyst and cyst excretions (K_i) from each human emission (i) per district were calculated as follows: (1)${\displaystyle \mathrm{st}.\left\{\begin{array}{c}{K}_1={\mathrm{CE}}_u=P_u\times F_{cu}\times O_p\times \left(1-F_{\mathrm{rem}}\right)\hfill \\ K_2={\mathrm{CE}}_r=P_r\times F_{cr}\times O_p\times \left(1-F_{\mathrm{rem}}\right)\hfill \\ K_3={\mathrm{DE}}_u=P_u\times F_{du}\times O_p\hfill \\ K_4={\text{DifE}}_r=P_r\times F_{\text{difr}}\times O_p\hfill \end{array}\right.}$ (2)${\displaystyle H=\sum _{i=1}^4{K}_i}$

where H is the total oocyst and cyst excretion from different human emission categories in a district or county (oocysts and cycsts/year); P_u and P_r are total urban and rural populations in districts or counties, respectively; O_p is the average oocyst and cyst excretion rates per person per year (oocysts and cysts/year); F_cu and F_cr are the fractions of urban and rural populations connected to a sewer, respectively; F_du is the fraction of urban populations not connected to a sewer that is considered a direct source; F_difr is the fraction of rural populations not using sanitation that is considered a diffuse source; and F_rem is the fraction of oocysts and cysts removed by sewage treatment plants (STP). The values assumed for this study are summarized in Tables S1 and S2.

Calculating oocyst and cyst excretion in animal manure (A)

Using the number of livestock, breeding day, manure excretion, oocyst and cyst excretion rate, and prevalence rate, the model then estimated livestock oocyst and cyst emissions in 2013. We established six livestock categories: rabbits, pigs, cattle, poultry, sheep, and goats. Unlike Hofstra et al. (2013), our model used different livestock breeding day categories because each livestock species has a unique number of breeding days before slaughter and produces different amounts of manure and excretions each year. We also divided the animal populations into four emission categories: (1) connected emissions from livestock receiving manure treatment, (2) direct emissions from livestock directly discharged to surface water, (3) diffuse emissions resulting from using livestock manure as a fertilizer after storage, and (4) livestock manure that was not used for irrigation after storage or for any other use (e.g., burned for fuel) (Vermeulen et al., 2015; Vermeulen et al., 2017). We assumed that 10% of emissions would be connected emissions (Zhang et al., 2017). Illegal and undocumented direct emissions (e.g., dumped manure) were not included in our model due to a lack of data. The results (A) were calculated using Eqs. (3) and (4): ${\displaystyle X_j\in \left\{X_1,X_2,X_3,X_4,X_5\right\},j\in \left\{1,2,3,4,5\right\}}$

X₁—Rabbit emissions

X₂—Pig emissions

X₃—Cattle emissions

X₄—Sheep and goat emissions

X₅—Poultry emissions.

We calculated oocyst and cyst excretions (X_j) from each animal species (j) per district using the following equation:

(3)${\displaystyle X_j=N_{a_j}\times D_{a_j}\times M_{a_j}\times O_{a_j}\times P_{a_j}}$ (4)${\displaystyle A=\sum _{j=1}^5{X}_j}$

where X_j is the oocyst and cyst excretions from livestock species j in a district or county (oocysts and cysts/year), N_aj is the number of animals in a district or county, D_aj is the breeding day for different livestock species j (days), M_aj is the mean daily manure of livestock species j (kg day⁻¹), O_aj is the oocyst and cyst excretion rate per infected livestock species j in manure (¹⁰log (oo)cysts kg⁻¹ d⁻¹), and P_aj is the prevalence of cryptosporidiosis and giardiasis in livestock species j. The values assumed for this study are summarized in Tables S1, S3, S4, and S5.

Calculating oocyst and cyst emissions after manure storage (S)

In China, manure from human diffuse and livestock sources is collected and used for irrigation and fertilizer (Liu et al., 2019). The decay of oocysts and cysts in manure that has been stored before being applied as fertilizer in the TGR watershed is temperature-dependent during the storage period (Tang et al., 2011; Vermeulen et al., 2017). The number of oocysts and cysts in stored manure that has been loaded on land during irrigation was calculated using Eq. (5): (5)${\displaystyle S={\text{DifE}}_r\times F_{s,h}\times F_v+A\times F_{s,a}\times F_v}$

where S is the number of oocysts and cysts in manure that has been spread on land after storage in a district or county (oocysts and cysts/year); DifE_r and A are the number of oocysts and cysts in manure for rural residents (human diffuse sources) and livestock (oocysts and cysts/year), respectively; F_s,h and F_s,a are the proportions of stored manure applied as a fertilizer from rural residents and livestock, respectively (Table S2); and F_v is the proportion of average oocyst and cyst survival in the storage system.

The average oocyst and cyst survival rate (F_v) in the storage system depended on temperature (T) and storage time (t_s) (Vermeulen et al., 2017). The results were calculated using Eqs. (6), (7) and (8):

(6)${\displaystyle K_s=\frac{ln10}{-2.5586\times T+119.63}}$ (7)${\displaystyle V_s=e^{\left(-K_s\times t_s\right)}}$ (8)${\displaystyle F_v=\frac{{\int }_0^{t_s}{V}_{s}dt}{t_s}}$

where T is the average annual air temperature (°C) (Table S2), K_s is a constant based on air temperature, V_s is the survival rate of oocysts and cysts over time, and t_s is the manure storage time (days) (Table S2).

Calculating oocyst and cyst runoff (R) to the TGR

Oocysts and cysts in stored manure that have been applied to agricultural land as a fertilizer are transported from land to rivers largely via surface runoff (Velthof et al., 2009). Our model estimated oocyst and cyst runoff using the amounts of manure applied as fertilizer, maximal surface runoff, and a set of reduction factors (Velthof et al., 2009; Hofstra et al., 2013). The results (R) were calculated using Eq. (9): (9)${\displaystyle R=S\times F_{\mathrm{run},max}\times f_{lu}\times f_p\times f_{rc}\times f_s}$

where R is the number of oocysts and cysts in manure applied as a fertilizer that reached the TGR via surface runoff in a district or county (oocysts and cysts/year), S is the number of oocysts and cysts in manure that was spread on land after storage (oocyst and cysts/year), F_run,max is the fraction of maximum surface runoff across different slope classes, f_lu is the reduction factor for land use, f_p is the reduction factor for average annual precipitation, f_rc is the reduction factor for rock depth, and f_s is the reduction factor for soil type. The values assumed for this study are summarized in Table S2.

Calculating total emissions (E) and mean concentrations (C) of oocysts and cysts

Our model defined total oocyst and cyst emissions as the annual number of oocysts and cysts per district or county in the TGR. The results (E) were calculated using Eq. (10): (10)${\displaystyle E={\mathrm{CE}}_u+{\mathrm{CE}}_r+{\mathrm{DE}}_u+R}$

where E is the total oocyst and cyst emissions from humans and animals in a district or county (oocysts and cysts/year), CE_u is oocyst and cyst emissions in the TGR by urban populations connected to STP in a district or county, CE_r is oocyst and cyst emissions in the TGR by rural populations connected to STP in a district or county, DE_u is direct oocyst and cyst emissions in the TGR by urban populations in a district or county, and R is oocyst and cyst emissions in the TGR from human and livestock manure that has been applied as a fertilizer via runoff in a district or county.

According to total emissions from the two protozoa in Chongqing and the TGR’s hydrological information, we preliminarily calculated mean Cryptosporidium and Giardia concentrations in the TGR in 2013 using the GloWPa-Crypto C1 model (Vermeulen et al., 2019). Mean concentrations were calculated using Eq. (11). All equations and parameters used in this study were showed in Table S11. (11)${\displaystyle C=\frac{E_t\times e^{-\left(K_T+K_R+K_S\right)\times t}}{Q_s}}$

Where C is mean concentrations of oocysts and cysts (oocysts and cysts 10 L⁻¹), E_t is the sum of total oocyst and cyst emissions from humans and animals in all districts or counties in Chongqing (oocysts and cysts/year). K_T, K_R,  and K_S represent loss rate constants of temperature, solar radiation, and sedimentation, respectively (day⁻¹). t is residence time of oocysts and cysts in Chongqing section of the TGR, Q_s is the sum of the TGR’s annual inflow and storage capacity (m³ year⁻¹).

Sensitivity analysis

We tested our model’s sensitivity to change using input parameters in a nominal range sensitivity analysis (NRSA). Input parameter values were based on reasonable lower and upper ranges of a base model, and we tested each variable individually (Vermeulen et al., 2015; Vermeulen et al., 2017). We selected the NRSA because it provides quantitative insight into the individual impact of different parameters on the model’s outcome. Tables S6–S8 present the sensitivity analysis input variables.

Predicting total oocyst and cyst emissions for 2050

To explore the impact of future population, urbanization, and sanitation changes on human and animal Cryptosporidium and Giardia emissions in the TGR, we divided the emissions into urban resident, rural resident, and livestock categories to predict the total oocyst and cyst emissions for 2050 based on three scenarios. China’s projected population, urbanization, and livestock production data for 2050 can be found in the Shared Socioeconomic Pathways (SSPs) database (Zhao, 2018; Huang et al., 2019; Chen et al., 2020) (https://tntcat.iiasa.ac.at/SspDb). We based Scenario 1 on SSP1, which is entitled “Sustainability—Taking the green road” and emphasizes sustainability, well-being, and equity. In this scenario, there is moderate population change and well-planned urbanization (O’Neill et al., 2015; Jiang & O’Neill, 2017; Zhao, 2018; Huang et al., 2019; Chen et al., 2020). Scenario 2 was based on SSP3, which is entitled “Regional rivalry—A rocky road” and emphasizes regional progress. In this scenario, China’s population change is significant and urbanization is unplanned (O’Neill et al., 2015; Jiang & O’Neill, 2017; Zhao, 2018; Huang et al., 2019; Chen et al., 2020). To emphasize the importance of wastewater and manure treatment, we created Scenario 3 as a variation of Scenario 1 based on Hofstra & Vermeulen (2016). This scenario has the same population, urbanization, and sanitation changes as Scenario 1, but with the insufficient sewage and manure treatments from 2013. Since there were no available data for individual livestock species, we assumed that all livestock species will grow by the same percentage noted in the SSPs database. We also assumed that there will be changes only in population and livestock numbers, not in any other parameters (e.g., oocyst and cyst excretion rates and prevalence) (Iqbal, Islam & Hofstra, 2019). We based our sanitation, wastewater, and manure treatment predictions for these three scenarios on previous literature reviews (Hofstra & Vermeulen, 2016; Iqbal, Islam & Hofstra, 2019). Table S9 provides an overview of the scenarios.

Emissions and mean concentrations of oocysts and cysts in the TGR in 2013

Cryptosporidium oocyst and Giardia cyst emissions from humans, rabbits, pigs, cattle, sheep, goats, and poultry found in the TGR in 2013 are shown in Fig. 3. Chongqing had a total of 1.6 × 10¹⁵ oocysts/year and 2.1 × 10¹⁵ cysts/year of Cryptosporidium and Giardia emissions. Human Cryptosporidium and Giardia emissions contained a total of 1.2 × 10¹⁵ oocysts/year and 2.0 × 10¹⁵ cysts/year, and animal emissions had a total of 3.4 × 10¹⁴ oocysts/year and 1.5 × 10¹⁴ cysts/year. Humans and animals were responsible for 42% and 10% of total emissions in the TGR, respectively. Humans were responsible for 78% of oocyst emissions, followed by 14% from pigs, and 8% from poultry. Humans were responsible for 93% of cyst emissions, followed by 6% from pigs, and 0.5% from cattle. Ultimately, we found that humans were the dominant source of oocysts and cysts, followed by pigs, poultry, and cattle. The mean Cryptosporidium and Giardia concentrations in the TGR in 2013 were 22 oocysts/10 L and 28 cysts/10 L, respectively.

Oocyst and cyst emission sanitation types

We immediately observed the differences in sanitation types (connected emissions, direct emissions, diffuse emissions, and non-source) across the human, livestock, urban, and rural populations (Fig. 4). We found that 49% of the populations were connected to a sewer, 36% had diffuse sources, 13% had direct sources, and 2% were non-source. In livestock populations, only 10% were connected to a sewer (manure treatment), 80% produced diffuse emissions, and 10% were non-source. We divided the human and livestock emissions by region: urban areas (made up of urban residents) and rural areas (made up of rural residents and all livestock). In urban areas, the emissions connected to a sewer were prominent (78%), followed by direct sources (22%). In rural areas, diffuse sources produced approximately 85% of total emissions, followed by connected sources (9%), and non-source (6%).

Spatial distribution of oocyst and cyst emissions in the TGR in 2013

Our model produced a spatial distribution of Cryptosporidium and Giardia emissions in the TGR for each Chongqing district or county from 2013 (Fig. 5). The total human Cryptosporidium emissions ranged from 5.4 × 10¹² to 7.4 × 10¹³ oocysts/district (Fig. 5A) and the total human Giardia emissions ranged from 8.5 × 10¹² to 1.2 × 10¹⁴ cysts/district (Fig. 5B). Overall, the emission spatial differences depended on population density and urbanization rate. The largest emissions were from the densely-populated Yubei, Wanzhou, and Jiulongpo districts.

The total animal source emissions ranged from 0 to 1.8 × 10¹³ oocysts/district (Fig. 5C) and 0 to 7.2 × 10¹³ cysts/district (Fig. 5D) for Cryptosporidium and Giardia, respectively. We based our results on the number of animals, manure production, manure treatment and runoff, and invariant oocyst and cyst emissions from each animal category over one year. The lowest emissions were observed in areas with low animal populations, such as the downtown districts of Yuzhong, Dadukou, and Nanan.

The total Cryptosporidium and Giardia emissions ranged from 1.0 × 10¹³ to 8.1 × 10¹³ oocysts/district and 1.0 × 10¹³ to 1.2 × 10¹⁴ cysts/district, respectively (Figs. 5E and 5F). Total human emissions were approximately six-fold higher than animal emissions and played a decisive role in total emission distribution. We found slightly more cysts than oocysts in total emissions and human emissions. In contrast, there were slightly more oocysts than cysts in animal emissions. The highest total emissions were found in areas with large animal and human populations, such as the main districts of Wanzhou, Yubei, and Hechuan.

Human Cryptosporidium and Giardia emissions from urban and rural areas can be found in Fig. 6. In urban areas, Cryptosporidium and Giardia emissions ranged from 3.5 × 10¹² to 7.0 × 10¹³ oocysts/district and 5.5 × 10¹² to 1.1 × 10¹⁴ cysts/district, respectively (Figs. 6A and 6C). In rural areas, the emissions ranged from 0 to 9.8 × 10¹² oocysts/district and 0 to 1.6 × 10¹³ cysts/district (Figs. 6B and 6D). Rural emissions were spread over much larger areas than urban emissions. Human emissions in urban areas were six-fold higher than in rural areas and played a crucial role in total human emission distribution.

Sensitivity analysis

Since there were limited observational data on Cryptosporidium and Giardia, we performed a sensitivity analysis to verify the GloWPa-TGR-Crypto model’s performance. The sensitivity analysis (Tables S6–S8) showed that the model was the most sensitive to changes in excretion rate (shown for 1 log unit change in excretion rates), particularly the excretion rates of humans (factor 8.03), pigs (factor 2.22), and poultry (factor 1.74). The model was more sensitive to prevalence changes in humans, pigs, and poultry (factors 1.39, 1.14, and 1.08, respectively). The results confirmed that humans, pigs, and poultry were the dominant sources of oocyst and cyst emissions. Besides excretion rate and prevalence, the model was most sensitive to changes in the amount of runoff, STP oocyst and cyst removal efficiencies, the amount of connected emissions, human population, and manure storage time (factors 1.30, 1.23, 1.21, 1.16, and 1.11, respectively), as these parameters affected oocyst and cyst survival and emissions. The model was not very sensitive to changes in the amount of rural resident feces applied as fertilizer, rural wastewater treatment, and the excretion rates and prevalence of animal species that did not contribute much to the total oocyst and cyst emissions (e.g., cattle, rabbits, sheep, and goats).

Scenario analysis: the effect of population, urbanization, and sanitation changes in 2050

In Scenario 1, moderate population change, planned urbanization, and strong improvements in sanitation, wastewater, and manure treatments will decrease the total emissions in the TGR to 9.5 × 10¹⁴ oocysts/year and 1.2 × 10¹⁵ cysts/year by 2050 (Fig. 7). This would reduce approximately 40% of the Cryptosporidium emissions and 44% of the Giardia emissions measured in 2013. All emissions from all three sources would decrease, with a notable 61% decrease for rural residents (Table S10). Figures 8B, 8E, 9B and 9E show the decrease across all regions in Scenario 1. The largest decline would be found in the Yubei and Jiulongpo districts, where assumed urbanization rates would increase to 100% and 99%, respectively, and 99% of domestic sewers would obtain secondary or tertiary treatment. Scenario 1 also shows changes in the contributions to total emissions. Urban residents would be responsible for 64% and 81% of Cryptosporidium and Giardia emissions, respectively, which would be a 3% decrease and a 1% increase from 2013. Rural Cryptosporidium and Giardia emissions would decrease from 11% to 7% and 13% to 9%, respectively. Livestock Cryptosporidium and Giardia emissions would increase from 22% to 29% and 7% to 10%, respectively.

In Scenario 2, Cryptosporidium and Giardia emissions are expected to increase to 1.9 × 10¹⁵ oocysts/year and 2.4 × 10¹⁵ cysts/year by 2050 (Fig. 7), would be 19% and 12% growth, respectively, from 2013. Emissions from urban residents and livestock would increase (Table S10) due to strong population growth, unplanned urbanization, limited sanitation, and expanded livestock production practices where untreated manure used as fertilizer is emitted into surface water. Emissions from rural residents would decrease 8% because the rate of urbanization would increase while the same sanitation practices from 2013 are used by that smaller rural population. Figures 8C, 8F, 9C and 9F show that total emissions would increase in all regions (particularly the Wanzhou and Yubei districts) because of strong population growth and limited environmental regulation. In Scenario 2, urban residents would account for 63% and 79%, rural residents would account for 9% and 11%, and livestock would account for 29% and 10% of Cryptosporidium and Giardia emissions, respectively, by 2050.

Scenario 3 has the same population, urbanization, and sanitation changes as Scenario 1, but with limited wastewater and manure treatment facilities. Scenario 3 has the highest Cryptosporidium and Giardia emissions of all the scenarios. Total emissions would increase to 2.0 × 10¹⁵ oocysts/year and 2.7 × 10¹⁵ cysts/year, with 29% and 27% growth compared to 2013, respectively (Fig. 7). Livestock would see the most growth in emissions (an increase of 42%) (Table S10). Figures 8D, 8G, 9D and 9G show that an increase in emissions across all regions, except in regions with assumed urbanization rates of 100% and where 50% of emissions obtain secondary treatment (such as the Yuzhong and Shapingba districts). This result highlights the importance of wastewater and manure treatment. Connecting populations to sewers without appropriate sewage treatment introduces more waterborne pathogens to surface water, affecting water quality.