Int J Biometeorol DOI 10.1007/s00484-014-0873-x
ORIGINAL PAPER
Effects of climate change on the economic output of the Longjing-43 tea tree, 1972–2013 Weiping Lou & Shanlei Sun & Lihong Wu & Ke Sun
Received: 8 December 2013 / Revised: 8 July 2014 / Accepted: 9 July 2014 # ISB 2014
Abstract Based on phenological and economic output models established and meteorological data from 1972 to 2013, changes in the phenology, frost risk, and economic output of the Longjing-43 tea tree in the Yuezhou Longjing tea production area of China were evaluated. As the local climate has changed, the beginning dates of tea bud and leaf plucking of this cultivar in all five counties studied has advanced significantly by −1.28 to −0.88 days/decade, with no significant change in the risk of frost. The main tea-producing stages in the tea production cycle include the plucking periods for superfine, grade 1, and grade 2 buds and leaves. Among the five bud and leaf grades, the economic output of the plucking periods for superfine and grade 1 decreased significantly, that for grade 2 showed no significant change, and those for grades 3 and 4 increased significantly. The economic output of large-area tea plantations employing an average of 45 workers per hectare and producing superfine to grade 2 buds and leaves were significantly reduced by 6,745–8,829 yuan/decade/ha, depending on the county. Those tea farmers who planted tea trees on their own small land holdings and produced superfine to grade 4 tea buds and leaves themselves experienced no significant decline in economic output. Keywords Longjing-43 tea tree . Phenological model . Climate change . Economic output W. Lou (*) : K. Sun Xinchang Weather Bureau, Xinchang County 312500 Zhejiang Province, China e-mail: [email protected] S. Sun Applied Hydrometeorological Research Institute, Nanjing University of Information Science and Technology (NUIST), Nanjing 210044, China L. Wu Zhejiang Provincial Climate Center, Hangzhou 310017, China
Introduction Tea, coffee, and cocoa are the three major caffeinated drinks worldwide (Jamieson 2001). Increasing numbers of people are choosing green tea over other tea varieties because of its health benefits (Cooper et al. 2005; Cheng 2006; Zaveri 2006). Zhejiang Province is the most famous locality for green tea production in the world. In 2011, 1.8×105 ha of tea trees were cultivated in Zhejiang and accounted for 63.41 and 63.98 %, respectively, of the total volume and value of green tea exported from China (Zhejiang Tea Industry Association 2012). The economic output of tea trees depends on several factors, such as species, tea type, tea brand, and production dates. The benefits to farmers of planting tea trees cannot be measured in field yields alone but must be balanced with income. For example, while the output of spring green teas accounts for only 43.03 % of all tea production in Zhejiang, it accounts for 91.13 % of the total economic output (Zhejiang Tea Industry Association 2012), making this China’s famous tea the main income source for local tea farmers. The price of spring green tea fluctuates greatly with the bud and leaf plucking period. At the beginning dates of the tea bud and leaf plucking period (BDTP), farmers can charge higher prices, thus generating more income, while the price markedly declines after the plucking period, causing a decrease in income (Lou and Sun 2013). Late winter and early spring temperatures influence plant phenology, including budding, leafing, and flowering times (Kramer et al. 2000; Chmielewsky and Roetzer 2001; Wolfe et al. 2005; Avolio et al. 2012). A rise in air temperature would lead to early spring plant phenology (Ahas et al. 2002). In addition, a late spring frost can damage and even kill budding plants. Some studies have shown that climate warming may cause many trees to sprout earlier, increasing the risk of frost damage (Cannell and Smith 1986; Hänninen 1991). However, other studies have shown that climate warming increases the
Int J Biometeorol
temperature gradually and that the frost risk fluctuates depending on interactions between plant phenology and the local climate (Colombo 1998; Scheifinger et al. 2003). Spring frost can damage tea buds and decrease their economic value (Tomihama et al. 2009; Huang 1989); the resulting economic losses are related to the extent and severity of frost that occurs during the tea bud and leaf plucking period. Furthermore, the economic output of tea trees depends on the speed of shoot growth, which is closely linked to springtime temperature changes. Longjing-43 is the main tea cultivar grown in Zhejiang Province (Li et al. 2007). This study aimed (1) to establish a phenological model of the budding and leafing dates of this cultivar using a time series of temperatures and (2) to analyze the impacts of observed climate change on the tea’s economic output. Temperature is a main driver of many plant developmental processes. Higher temperatures have been frequently shown to accelerate plant development and the progression to the next ontogenetic stage (Ahas et al. 2000; Badeck et al. 2004; Linkosalo et al. 2006; Cleland et al. 2007). Effective accumulated temperature (EAT) and the accumulated portion of the daily mean temperature that exceeds the threshold temperature determine the duration of phenological phases (Jiang et al. 2011; Liang and Schwartz 2013). Changes in spring temperatures have resulted in the EAT method of predicting BDTP out-performing that of linear regression modeling (Huang 1989). The spring phenology of Longjing-43 tea trees was observed in 19 tea plantations and matched with recorded values from meteorological stations in the Yuezhou Longjing tea production area. Based on both sets of data, tea phenological and economic output models were developed. Using models and meteorological data from 1972 to 2013, the effect of climate change on the economic output of tea was analyzed.
laborers in four tea plantations (working from 07:00–11:30 and from 13:00–18:30) were also recorded from 2010 to 2013. Phenophases were based on the criteria of the GB 18650-2002 Product of Designations of Origin and Geographical Indications—Longjing tea (General Administration of Quality Supervision, Inspection and Quarantine of the People’s Republic of China, 2002) (Table 2). The tea producers could be classified into two groups: plantations and family farms. Plantations generally had more than 3.3 ha for planting Longjing-43 tea trees and employed workers to produce tea. To maximize their profits, they produced tea only from the high-grade tea buds and leaves, superfine to grade 2. The family farms generally had less than 0.2 ha for planting tea trees, with tea produced by family members. Because the costs for plucking and processing and the market values are less important to these farms, they produced tea from all grades of buds and leaves (superfine to grade 4).
Materials and methods
Fitting the BDTP model
Phenological data and survey sites
To establish a fitted model of BDTP, the observations (e.g., meteorological and phenological) from 2004 to 2010 and 2012 in 19 plantations were used. Based on regression analysis, a model was developed based on air temperatures from the period between tea bud initiation to BDTP (Huang 1989). Taking Xinchang County as an example, the BDTP equation for one plantation can be expressed as follows:
The Yuezhou Longjing tea production area is located in the center of Zhejiang Province, China (29.22–30.54° N, 119.88– 121.22° E), and includes Xinchang, Shengzhou, Shaoxing, Shangyu, and Zhuji counties (Fig. 1). This region belongs to a subtropical monsoon climate zone with an annual average temperature of 16.0–17.5 °C, annual precipitation of 900– 1,500 mm, and annual sunshine time of 1,100–2,200 h. As the largest Longjing tea production area in China, the Yuezhou area was named as origin appellation of the Longjing tea. Generally, most Longjing tea production in this area occurs from late February to April. The Longjing-43 tea tree phenology (i.e., BDTP and bud and leaf phenophases) was recorded in 19 plantations (Table 1). Moreover, the daily tea prices and amounts of tea buds and leaves plucked (yields) by four
Meteorological data In both Xinchang and Shangyu counties, the meteorological stations were built in 1971, while those in the other counties were built earlier. Thus, comparable data from all five counties were available from 1972 to 2013. Daily mean and minimum temperatures and daytime precipitation were collected from the Zhejiang Meteorological Information Network Center. Beginning in 2004, meteorological stations were established for collecting the meteorological observations (i.e., temperature and precipitation) in 19 tea plantations at altitudes ranging from 35 to 510 m (Table 1). The phenological duration and meteorological data acquired from 2010 to 2013 from four sites, along with 2013 data from 15 other sites, were used to determine EATs for each phenological duration.
BDTP ¼ 96:3−2:38T þ a
ð1Þ
where T is the average temperature of 30 days centered on Tday, where Tday is the first of six consecutive days prior to July with an average temperature of 10 °C or greater. The parameter a was introduced to account for differences in Tday among plantations. When Tday differed by less than 10 days, a was equal to zero. Otherwise, a was set at ten times the difference in Tday (Lou and Sun 2013).
Int J Biometeorol Fig. 1 Study area in the Yuezhou Longjing tea production area, Zhejiang Province, China (a, shaded area). Phenological survey locations (black dots; see Table 1 for names and details) and county meteorological stations (empty dots) are mapped (b)
Phenological duration Based on length and number of bud and leaf blades, tea buds and leaves could be divided into five categories, from highest quality to lowest: superfine and grades 1 through 4 (Table 2). The phenological duration of each stage could be computed as the period from the beginning to the end of plucking. Buds and leaves not meeting one of these standards cannot legally be processed into Longjing tea. To complete a developmental stage, plants require sufficient accumulated heat or summed temperature (also called growing degree days or TSUM) (van Vliet et al. 2002). Based on tea tree phenological and meteorological data, we calculated EATs for each grade tea bud and leaf: ∑T>0 °C, ∑T>5 °C, ∑T>6 °C, ∑T>7 °C, and ∑T>8 °C corresponding to the threshold temperature 0, 5, 6, 7, and 8 °C, respectively. Then, TSUM could
be regarded as an EAT with the minimum standard deviation (Wang et al. 1981). Detailed statistics are shown in Table 3. In the current study, the standard deviation of ∑T>5 °C was the smallest in all grades. For a temperature threshold of 5 °C, the EATs calculated for the phenological durations of superfine, grade 1, grade 2, grade 3, and grade 4 tea buds and leaves were 26.7, 35.3, 38.6, 47.6, and 87.5 °C, respectively. A thermal time approach was used to simulate bud and leaf growth in the spring plucking periods. The bud and leaf growth rate (DTj,t) was formulated as follows (Kropff et al. 1994): DT j;t ¼ Te=TSU M j ð j ¼ 1; 2; 3; 4; 5Þ
ð2Þ
where j is the phenological period corresponding to the five grades, and t is the day number from the first (t=1) to the last
Int J Biometeorol Table 1 Recording periods and survey approach at the Longjing tea phenological observation sites Site no.
Site name
Altitude (m)
Periods for which data was recorded Weather data
Beginning date of plucking period
Plucked period
Daily harvest and tea price
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Xuetou Shaxi Lidong Changzhao Pangdingshan Mingfu Xishan Pinghu Henbanqiao Liwang Shuangcai Hongzhang Yanko Jinlin
423 137 357 141 208 132 185 176 405 282 400 214 393 60
2005–2013 2007–2013 2010–2013 2007–2013 2010–2013 2009–2013 2005–2013 2010–2013 2007–2013 2007–2013 2007–2013 2010–2013 2010–2013 2008–2013
2005–2013 2007–2013 2010–2013 2007–2013 2010–2013 2009–2013 2005–2013 2010–2013 2007–2013 2007–2013 2007–2013 2010–2013 2010–2013 2008–2013
2013 2013 2013 2010–2013 2010–2013 2010–2013 2013 2013 2013 2013 2010–2013 2013 2013 2013
2013 2013 2013 2010–2013 2010–2013 2010–2013 2013 2013 2013 2013 2010–2013 2013 2013 2013
15 16 17
Juenong Shanjie Tongyuan
35 60 211
2010–2013 2005–2013 2005–2013
2010–2013 2005–2013
2013 2013
2013 2013
18 19
Zhongmei’ao Shiliping
107 122
2006–2013 2007–2013
2005–2013 2006–2013 2007–2013
2013 2013 2013
2013 2013 2013
Site numbers in table correspond to the station/plantation numbers in Fig. 1
plucking dates in jth phenological period. Te is the portion of the daily mean temperature that exceeded 5 °C on day t in the jth phenological period. TSUMj is the TSUM for the jth period. To determine the cumulative growth quantity for a period (Dj,i), DTj,t was accumulated:
Table 2 Criteria for Longjing tea phenological observation periods Period
Standard
BDTP
Ten to 15 buds per square meter of tea tree canopy surface meet the superfine tea plucking standard.
Superfine One leaf in the early opening stage. Small angle between bud and leaf. Bud is longer than leaf, and both are ≤2.5 cm. Grade 1 First leaf is open, second leaf is in early opening stage, and no more than 10 % of leaves in buds are open. Buds are longer than leaves, and both are ≤3 cm. Grade 2 Second leaf is open, and no more than 30 % of leaves in buds are open. Bud length equals leaf length, and both are ≤3.5 cm. Grade 3 Second leaf is open, third leaf is in early opening stage, and no more than 30 % of leaves in buds are open. Leaf length is longer than bud length, and both are ≤4 cm. Grade 4 Third leaf is open, and no more than 50 % of leaves in buds are open. Leaf length is longer than bud length, and both are ≤4.5 cm.
D j;i ¼
i X
ð3Þ
DT j;t
t¼1
When Dj,i equaled 1, the ith day could be considered the last day during the jth phenological period. From BDTP to the ith day of the jth phenological period, cumulative growth (ADj,i) was as follows: AD j;i ¼ j−1 þ D j;i ð j ¼ 1; 2; 3; 4; 5Þ
ð4Þ
Table 3 Effective accumulated temperature (EAT) for plucking periods of each grade of tea buds and leaves Grade
Threshold temperature (°C) (mean±SD) 0
Superfine 49.6±22.5 Grade 1 75.9±27.8 Grade 2 72.3±21.7 Grade 3 84.7±19.1 Grade 4 177.2±44.3
5
6
7
8
26.7±4.4 35.3±5.4 38.6±4.6 47.6±3.7 87.5±1.7
24.7±5.8 30.1±6.6 34.1±5.8 36.2±13.1 78.3±1.9
22.7±6.2 25.7±7.2 28.9±6.7 30.1±11.3 70.5±2.0
20.7±7.9 21.3±9.2 23.2±7.4 24.6±9.9 62.3±3.1
Phenological duration and meteorological data acquired between 2010 and 2013 from four sites, along with 2013 data from 15 other sites, were used to determine EATs for each phenological duration
Int J Biometeorol
Effect of rain on yield of tea buds and leaves At the beginning of plucking period, tea buds were small and only small amounts met the plucking standard, so harvest per worker was low. As the tea buds grew, more of them reached the plucking standard, and yield per worker increased. Based on data recorded at the Mingfu, Shuangcai, Changzhao, and Pangdingshan tea plantations, Fig. 2 depicts the changes in yield per worker per day along with ADj,i under no-rain conditions. The formula can be expressed by the following: Qq ¼ 1:06 þ 0:8928AD j; i −0:0536AD j;i 2
ð5Þ
where Qq (kg/person/day) is the yield of tea buds and leaves per worker per day under no-rain conditions. Rain in the daytime affected the workers’ ability to pluck, whereas rain at night did not affect harvesting. When daytime rainfall was less than 10 mm, the workers continued to pluck because of the high price of tea in spring. However, when daytime rainfall exceeded 10 mm, the workers could not harvest. Figure 3 shows the relationship between daytime rainfall and the ratio of rainy day yield to rainless day yield. The impacts of daytime rainfall on yield could be expressed as follows: 1−RR=10 RR < 10mm f ðRRÞ ¼ ð6Þ 0 RR ≥ 10mm where f(RR) is the influence coefficient of daytime rainfall on yield, and RR (mm) is the rainfall from 08:00 to 20:00. Without considering frost occurrences, a model to calculate yield per worker per day under rain condition was established based on Eqs. 5 and 6: TAD j; i ¼ Qq f ðRRÞ
ð7Þ
Fig. 2 Relationship between the amount of tea buds and leaves plucked by one worker and plucking time in rainless weather. ADj,i is the cumulative growth quantity from the beginning day of tea bud and leaf plucking to the ith day of the jth phenological period
Frost damage period Except for frost from late February to April, there were no other factors (e.g., pests and hail) to endanger the growth of tea buds and leaves in the Yuezhou Longjing tea production area. Based on the fact that tea buds and leaves damaged by frost were not suitable to produce Longjing tea, the impact of frost on tea yield is important. The frost damage period was the period during which the tea trees had no buds and leaves that met the standard of Longjing tea because of frost damage. Because frost damage affected tea yield differently before and after the BDTP, two methods were proposed to calculate the frost damage period. If the frost damage occurred during the harvest period, the period was determined by the following equation: 2:6−28:174T l −0:9043T 2l T l > −5:0 C FRT ¼ ð8Þ 120:9 T l ≤ −5:0 C
where TADj,i (kg/person/day) is the yield of tea buds and leaves per worker per day under rain and no-frost conditions on the ith day of the jth phenological period. Frost risk The frost end date can be defined as the last day of minimum temperatures ≤0 °C (Lou and Sun 2013; Cittadini et al. 2006). Differences between the frost end date and BDTP can reflect the risk that the tea trees suffered frost damage (Lou et al. 2013). When the frost end date is later than the BDTP, frost damage may occur during the Longjing-43 plucking period. However, when the frost end date is earlier than the BDTP, whether the tea trees suffer frost damage depends mainly on the minimum temperature and tea bud growth (Lou and Sun 2013). In general, the closer the frost end date to the BDTP, the higher the probability of frost damage.
Fig. 3 Relationship between the ratio of the amount of tea buds and leaves plucked during rain conditions to the amount plucked during rainless conditions (RATP) and daily precipitation from 8:00 to 20:00
Int J Biometeorol
TAD j;i Pl=4:3 0
where FRT is the accumulated portion of the daily mean temperature that exceeded 5 °C from the start to the end of the frost damage period, and Tl is the minimum temperature when frost damage occurred. When the minimum temperature was less than or equal to −5 °C, all the new tea buds were damaged. Because the lowest air temperature during one frost disaster was less than or equal to −5 °C, the value of FRT was fixed at 120.9 °C. Based on the minimum temperature, when frost damage occurred, we can determine the accumulated temperature required for tea buds to reach the plucking standard after frost damage. When frost damage occurred before the BDTP, the frost damage period was determined by the following equation: ( X X 0 FRT ≤ T X X k T¼ ð9Þ FRT − T k FRT > Tk
E j;i ¼
where ∑Tk is the accumulated portion of the daily mean temperature that exceeded 5 °C from the date of frost damage to the BDTP. ∑T is the accumulated portion of the daily mean temperature that exceeded 5 °C from the BDTP to the date that the buds and leaves met Longjing tea plucking standards. If FRT was less than or equal to ∑Tk, ∑T was set to 0, indicating that frost had not affected the tea yield. If FRT was greater than ∑Tk, frost affected the tea yield, and we calculated the accumulated portion of the daily mean temperature that exceeds 5 °C on every day; when EAT reached ∑T, the frost damage period ended.
AEO ¼ PEOðLengthÞ−EOðRRÞ−EOðFrost Þ
Economic output model for tea
No frost damage Frost damage period
ð11Þ
Integrating Eq. 11, the total economic output (E) of a worker in spring was X ð12Þ E¼ E j;i
In summary, the economic output of a specific plucking period was determined by three factors: the length of the plucking period, the daytime precipitation, and the frost damage. When the temperature was relatively low (but above 0 °C), tea production and economic output increases because of the extended plucking period. Considering the decreases in tea yield caused by daytime rainfall and frost damage, the actual economic output of a specific period (AEO) could be expressed as follows: ð13Þ
where PEO(Length) represents the potential economic output of this specific period without considering the impacts of daytime rainfall and frost damage. EO(RR) and EO(Frost) are the losses of economic output caused by daytime rainfall and frost damage, respectively. Model validation Using 2011 and 2013 data, model accuracy for the yield of tea buds and leaves was evaluated using the root mean square error (RMSE) between the predicted and observed values: vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi uX N 2 u u xpi −xoi t i¼1 RMSE ¼ ð14Þ N
Daily tea prices and amounts of tea buds and leaves plucked (yields) at four sites in 2010 and 2012 were collected to estimate economic output models for those years. The two models were evaluated using observations from 2011 and 2013. Figure 4 shows changes in tea price with ADj,i at Mingfu, Shuangcai, Changzhao, and Pangdingshan tea plantations in 2012. Thus, the relationship between tea price (Pl, yuan/kg) and ADj,i could be expressed as follows:
where xpi and xoi are the predicted and observed yields, respectively.
Pl ¼ 677:6692−256:0479AD j; i
Trend analysis
þ 44:6983AD j;i −3:0219AD j;i 2
3
ð10Þ
Based on the recorded data from these four tea plantations, making 1.0 kg of tea required 4.3 kg of tea buds and leaves. Therefore, the economic output (Ej,i) of a worker in ith day of the jth phenological period was as follows:
The longest period of meteorological data from the 19 observation sites was from 2004 to 2013. The shortest period of meteorological data from five counties was from 1972 to 2013. To reconstruct phenology over a long time, a model was based on data from the five county meteorological stations from 1972 to 2013 (Cleland et al. 2007). Then, the BDTP and phenological duration datasets from these five
Int J Biometeorol
Results Model reliability In 2011 and 2013, the differences between predicted and observed BDTPs ranged from −1 to +1 day for the 19 tea plantations. RMSEs of tea yield in the Mingfu, Shuangcai, Changzhao, and Pangdingshan tea plantations were 0.0392, 0.0351, 0.0451, and 0.0436 (kg/d), respectively. Generally, the model captured tea phenology relatively accurately.
Fig. 4 Change of tea price with ADj,i in 2012. ADj,i is the cumulative growth quantity from the beginning day of tea bud and leaf plucking to the ith day of the jth phenological period
stations were used to analyze their trends. Assuming that the tea prices from 1972 to 2013 were fixed at the 2012 level, the economic outputs in the five counties were calculated. We used linear regression analysis to examine trends in the phenological time series (Sparks and Menzel 2002; Roetzer et al. 2000; Stone et al. 1996). The Mann–Kendall test can reliably identify monotonic linear and non-linear trends in non-normal datasets with outliers; this method has been found to be an excellent tool for trend detection in a time series and has been used to assess the significance of trends in phenological series (Julien and Sobrino 2009; Menzel 2000). By comparing several statistical tests applied to a climate series, Gerstengarbe and Werner (1999) pointed out the value and strength of this method. In the present study, therefore, we used the Mann–Kendall test. Statistical significance and strong significance were inferred at P
ORIGINAL PAPER
Effects of climate change on the economic output of the Longjing-43 tea tree, 1972–2013 Weiping Lou & Shanlei Sun & Lihong Wu & Ke Sun
Received: 8 December 2013 / Revised: 8 July 2014 / Accepted: 9 July 2014 # ISB 2014
Abstract Based on phenological and economic output models established and meteorological data from 1972 to 2013, changes in the phenology, frost risk, and economic output of the Longjing-43 tea tree in the Yuezhou Longjing tea production area of China were evaluated. As the local climate has changed, the beginning dates of tea bud and leaf plucking of this cultivar in all five counties studied has advanced significantly by −1.28 to −0.88 days/decade, with no significant change in the risk of frost. The main tea-producing stages in the tea production cycle include the plucking periods for superfine, grade 1, and grade 2 buds and leaves. Among the five bud and leaf grades, the economic output of the plucking periods for superfine and grade 1 decreased significantly, that for grade 2 showed no significant change, and those for grades 3 and 4 increased significantly. The economic output of large-area tea plantations employing an average of 45 workers per hectare and producing superfine to grade 2 buds and leaves were significantly reduced by 6,745–8,829 yuan/decade/ha, depending on the county. Those tea farmers who planted tea trees on their own small land holdings and produced superfine to grade 4 tea buds and leaves themselves experienced no significant decline in economic output. Keywords Longjing-43 tea tree . Phenological model . Climate change . Economic output W. Lou (*) : K. Sun Xinchang Weather Bureau, Xinchang County 312500 Zhejiang Province, China e-mail: [email protected] S. Sun Applied Hydrometeorological Research Institute, Nanjing University of Information Science and Technology (NUIST), Nanjing 210044, China L. Wu Zhejiang Provincial Climate Center, Hangzhou 310017, China
Introduction Tea, coffee, and cocoa are the three major caffeinated drinks worldwide (Jamieson 2001). Increasing numbers of people are choosing green tea over other tea varieties because of its health benefits (Cooper et al. 2005; Cheng 2006; Zaveri 2006). Zhejiang Province is the most famous locality for green tea production in the world. In 2011, 1.8×105 ha of tea trees were cultivated in Zhejiang and accounted for 63.41 and 63.98 %, respectively, of the total volume and value of green tea exported from China (Zhejiang Tea Industry Association 2012). The economic output of tea trees depends on several factors, such as species, tea type, tea brand, and production dates. The benefits to farmers of planting tea trees cannot be measured in field yields alone but must be balanced with income. For example, while the output of spring green teas accounts for only 43.03 % of all tea production in Zhejiang, it accounts for 91.13 % of the total economic output (Zhejiang Tea Industry Association 2012), making this China’s famous tea the main income source for local tea farmers. The price of spring green tea fluctuates greatly with the bud and leaf plucking period. At the beginning dates of the tea bud and leaf plucking period (BDTP), farmers can charge higher prices, thus generating more income, while the price markedly declines after the plucking period, causing a decrease in income (Lou and Sun 2013). Late winter and early spring temperatures influence plant phenology, including budding, leafing, and flowering times (Kramer et al. 2000; Chmielewsky and Roetzer 2001; Wolfe et al. 2005; Avolio et al. 2012). A rise in air temperature would lead to early spring plant phenology (Ahas et al. 2002). In addition, a late spring frost can damage and even kill budding plants. Some studies have shown that climate warming may cause many trees to sprout earlier, increasing the risk of frost damage (Cannell and Smith 1986; Hänninen 1991). However, other studies have shown that climate warming increases the
Int J Biometeorol
temperature gradually and that the frost risk fluctuates depending on interactions between plant phenology and the local climate (Colombo 1998; Scheifinger et al. 2003). Spring frost can damage tea buds and decrease their economic value (Tomihama et al. 2009; Huang 1989); the resulting economic losses are related to the extent and severity of frost that occurs during the tea bud and leaf plucking period. Furthermore, the economic output of tea trees depends on the speed of shoot growth, which is closely linked to springtime temperature changes. Longjing-43 is the main tea cultivar grown in Zhejiang Province (Li et al. 2007). This study aimed (1) to establish a phenological model of the budding and leafing dates of this cultivar using a time series of temperatures and (2) to analyze the impacts of observed climate change on the tea’s economic output. Temperature is a main driver of many plant developmental processes. Higher temperatures have been frequently shown to accelerate plant development and the progression to the next ontogenetic stage (Ahas et al. 2000; Badeck et al. 2004; Linkosalo et al. 2006; Cleland et al. 2007). Effective accumulated temperature (EAT) and the accumulated portion of the daily mean temperature that exceeds the threshold temperature determine the duration of phenological phases (Jiang et al. 2011; Liang and Schwartz 2013). Changes in spring temperatures have resulted in the EAT method of predicting BDTP out-performing that of linear regression modeling (Huang 1989). The spring phenology of Longjing-43 tea trees was observed in 19 tea plantations and matched with recorded values from meteorological stations in the Yuezhou Longjing tea production area. Based on both sets of data, tea phenological and economic output models were developed. Using models and meteorological data from 1972 to 2013, the effect of climate change on the economic output of tea was analyzed.
laborers in four tea plantations (working from 07:00–11:30 and from 13:00–18:30) were also recorded from 2010 to 2013. Phenophases were based on the criteria of the GB 18650-2002 Product of Designations of Origin and Geographical Indications—Longjing tea (General Administration of Quality Supervision, Inspection and Quarantine of the People’s Republic of China, 2002) (Table 2). The tea producers could be classified into two groups: plantations and family farms. Plantations generally had more than 3.3 ha for planting Longjing-43 tea trees and employed workers to produce tea. To maximize their profits, they produced tea only from the high-grade tea buds and leaves, superfine to grade 2. The family farms generally had less than 0.2 ha for planting tea trees, with tea produced by family members. Because the costs for plucking and processing and the market values are less important to these farms, they produced tea from all grades of buds and leaves (superfine to grade 4).
Materials and methods
Fitting the BDTP model
Phenological data and survey sites
To establish a fitted model of BDTP, the observations (e.g., meteorological and phenological) from 2004 to 2010 and 2012 in 19 plantations were used. Based on regression analysis, a model was developed based on air temperatures from the period between tea bud initiation to BDTP (Huang 1989). Taking Xinchang County as an example, the BDTP equation for one plantation can be expressed as follows:
The Yuezhou Longjing tea production area is located in the center of Zhejiang Province, China (29.22–30.54° N, 119.88– 121.22° E), and includes Xinchang, Shengzhou, Shaoxing, Shangyu, and Zhuji counties (Fig. 1). This region belongs to a subtropical monsoon climate zone with an annual average temperature of 16.0–17.5 °C, annual precipitation of 900– 1,500 mm, and annual sunshine time of 1,100–2,200 h. As the largest Longjing tea production area in China, the Yuezhou area was named as origin appellation of the Longjing tea. Generally, most Longjing tea production in this area occurs from late February to April. The Longjing-43 tea tree phenology (i.e., BDTP and bud and leaf phenophases) was recorded in 19 plantations (Table 1). Moreover, the daily tea prices and amounts of tea buds and leaves plucked (yields) by four
Meteorological data In both Xinchang and Shangyu counties, the meteorological stations were built in 1971, while those in the other counties were built earlier. Thus, comparable data from all five counties were available from 1972 to 2013. Daily mean and minimum temperatures and daytime precipitation were collected from the Zhejiang Meteorological Information Network Center. Beginning in 2004, meteorological stations were established for collecting the meteorological observations (i.e., temperature and precipitation) in 19 tea plantations at altitudes ranging from 35 to 510 m (Table 1). The phenological duration and meteorological data acquired from 2010 to 2013 from four sites, along with 2013 data from 15 other sites, were used to determine EATs for each phenological duration.
BDTP ¼ 96:3−2:38T þ a
ð1Þ
where T is the average temperature of 30 days centered on Tday, where Tday is the first of six consecutive days prior to July with an average temperature of 10 °C or greater. The parameter a was introduced to account for differences in Tday among plantations. When Tday differed by less than 10 days, a was equal to zero. Otherwise, a was set at ten times the difference in Tday (Lou and Sun 2013).
Int J Biometeorol Fig. 1 Study area in the Yuezhou Longjing tea production area, Zhejiang Province, China (a, shaded area). Phenological survey locations (black dots; see Table 1 for names and details) and county meteorological stations (empty dots) are mapped (b)
Phenological duration Based on length and number of bud and leaf blades, tea buds and leaves could be divided into five categories, from highest quality to lowest: superfine and grades 1 through 4 (Table 2). The phenological duration of each stage could be computed as the period from the beginning to the end of plucking. Buds and leaves not meeting one of these standards cannot legally be processed into Longjing tea. To complete a developmental stage, plants require sufficient accumulated heat or summed temperature (also called growing degree days or TSUM) (van Vliet et al. 2002). Based on tea tree phenological and meteorological data, we calculated EATs for each grade tea bud and leaf: ∑T>0 °C, ∑T>5 °C, ∑T>6 °C, ∑T>7 °C, and ∑T>8 °C corresponding to the threshold temperature 0, 5, 6, 7, and 8 °C, respectively. Then, TSUM could
be regarded as an EAT with the minimum standard deviation (Wang et al. 1981). Detailed statistics are shown in Table 3. In the current study, the standard deviation of ∑T>5 °C was the smallest in all grades. For a temperature threshold of 5 °C, the EATs calculated for the phenological durations of superfine, grade 1, grade 2, grade 3, and grade 4 tea buds and leaves were 26.7, 35.3, 38.6, 47.6, and 87.5 °C, respectively. A thermal time approach was used to simulate bud and leaf growth in the spring plucking periods. The bud and leaf growth rate (DTj,t) was formulated as follows (Kropff et al. 1994): DT j;t ¼ Te=TSU M j ð j ¼ 1; 2; 3; 4; 5Þ
ð2Þ
where j is the phenological period corresponding to the five grades, and t is the day number from the first (t=1) to the last
Int J Biometeorol Table 1 Recording periods and survey approach at the Longjing tea phenological observation sites Site no.
Site name
Altitude (m)
Periods for which data was recorded Weather data
Beginning date of plucking period
Plucked period
Daily harvest and tea price
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Xuetou Shaxi Lidong Changzhao Pangdingshan Mingfu Xishan Pinghu Henbanqiao Liwang Shuangcai Hongzhang Yanko Jinlin
423 137 357 141 208 132 185 176 405 282 400 214 393 60
2005–2013 2007–2013 2010–2013 2007–2013 2010–2013 2009–2013 2005–2013 2010–2013 2007–2013 2007–2013 2007–2013 2010–2013 2010–2013 2008–2013
2005–2013 2007–2013 2010–2013 2007–2013 2010–2013 2009–2013 2005–2013 2010–2013 2007–2013 2007–2013 2007–2013 2010–2013 2010–2013 2008–2013
2013 2013 2013 2010–2013 2010–2013 2010–2013 2013 2013 2013 2013 2010–2013 2013 2013 2013
2013 2013 2013 2010–2013 2010–2013 2010–2013 2013 2013 2013 2013 2010–2013 2013 2013 2013
15 16 17
Juenong Shanjie Tongyuan
35 60 211
2010–2013 2005–2013 2005–2013
2010–2013 2005–2013
2013 2013
2013 2013
18 19
Zhongmei’ao Shiliping
107 122
2006–2013 2007–2013
2005–2013 2006–2013 2007–2013
2013 2013 2013
2013 2013 2013
Site numbers in table correspond to the station/plantation numbers in Fig. 1
plucking dates in jth phenological period. Te is the portion of the daily mean temperature that exceeded 5 °C on day t in the jth phenological period. TSUMj is the TSUM for the jth period. To determine the cumulative growth quantity for a period (Dj,i), DTj,t was accumulated:
Table 2 Criteria for Longjing tea phenological observation periods Period
Standard
BDTP
Ten to 15 buds per square meter of tea tree canopy surface meet the superfine tea plucking standard.
Superfine One leaf in the early opening stage. Small angle between bud and leaf. Bud is longer than leaf, and both are ≤2.5 cm. Grade 1 First leaf is open, second leaf is in early opening stage, and no more than 10 % of leaves in buds are open. Buds are longer than leaves, and both are ≤3 cm. Grade 2 Second leaf is open, and no more than 30 % of leaves in buds are open. Bud length equals leaf length, and both are ≤3.5 cm. Grade 3 Second leaf is open, third leaf is in early opening stage, and no more than 30 % of leaves in buds are open. Leaf length is longer than bud length, and both are ≤4 cm. Grade 4 Third leaf is open, and no more than 50 % of leaves in buds are open. Leaf length is longer than bud length, and both are ≤4.5 cm.
D j;i ¼
i X
ð3Þ
DT j;t
t¼1
When Dj,i equaled 1, the ith day could be considered the last day during the jth phenological period. From BDTP to the ith day of the jth phenological period, cumulative growth (ADj,i) was as follows: AD j;i ¼ j−1 þ D j;i ð j ¼ 1; 2; 3; 4; 5Þ
ð4Þ
Table 3 Effective accumulated temperature (EAT) for plucking periods of each grade of tea buds and leaves Grade
Threshold temperature (°C) (mean±SD) 0
Superfine 49.6±22.5 Grade 1 75.9±27.8 Grade 2 72.3±21.7 Grade 3 84.7±19.1 Grade 4 177.2±44.3
5
6
7
8
26.7±4.4 35.3±5.4 38.6±4.6 47.6±3.7 87.5±1.7
24.7±5.8 30.1±6.6 34.1±5.8 36.2±13.1 78.3±1.9
22.7±6.2 25.7±7.2 28.9±6.7 30.1±11.3 70.5±2.0
20.7±7.9 21.3±9.2 23.2±7.4 24.6±9.9 62.3±3.1
Phenological duration and meteorological data acquired between 2010 and 2013 from four sites, along with 2013 data from 15 other sites, were used to determine EATs for each phenological duration
Int J Biometeorol
Effect of rain on yield of tea buds and leaves At the beginning of plucking period, tea buds were small and only small amounts met the plucking standard, so harvest per worker was low. As the tea buds grew, more of them reached the plucking standard, and yield per worker increased. Based on data recorded at the Mingfu, Shuangcai, Changzhao, and Pangdingshan tea plantations, Fig. 2 depicts the changes in yield per worker per day along with ADj,i under no-rain conditions. The formula can be expressed by the following: Qq ¼ 1:06 þ 0:8928AD j; i −0:0536AD j;i 2
ð5Þ
where Qq (kg/person/day) is the yield of tea buds and leaves per worker per day under no-rain conditions. Rain in the daytime affected the workers’ ability to pluck, whereas rain at night did not affect harvesting. When daytime rainfall was less than 10 mm, the workers continued to pluck because of the high price of tea in spring. However, when daytime rainfall exceeded 10 mm, the workers could not harvest. Figure 3 shows the relationship between daytime rainfall and the ratio of rainy day yield to rainless day yield. The impacts of daytime rainfall on yield could be expressed as follows: 1−RR=10 RR < 10mm f ðRRÞ ¼ ð6Þ 0 RR ≥ 10mm where f(RR) is the influence coefficient of daytime rainfall on yield, and RR (mm) is the rainfall from 08:00 to 20:00. Without considering frost occurrences, a model to calculate yield per worker per day under rain condition was established based on Eqs. 5 and 6: TAD j; i ¼ Qq f ðRRÞ
ð7Þ
Fig. 2 Relationship between the amount of tea buds and leaves plucked by one worker and plucking time in rainless weather. ADj,i is the cumulative growth quantity from the beginning day of tea bud and leaf plucking to the ith day of the jth phenological period
Frost damage period Except for frost from late February to April, there were no other factors (e.g., pests and hail) to endanger the growth of tea buds and leaves in the Yuezhou Longjing tea production area. Based on the fact that tea buds and leaves damaged by frost were not suitable to produce Longjing tea, the impact of frost on tea yield is important. The frost damage period was the period during which the tea trees had no buds and leaves that met the standard of Longjing tea because of frost damage. Because frost damage affected tea yield differently before and after the BDTP, two methods were proposed to calculate the frost damage period. If the frost damage occurred during the harvest period, the period was determined by the following equation: 2:6−28:174T l −0:9043T 2l T l > −5:0 C FRT ¼ ð8Þ 120:9 T l ≤ −5:0 C
where TADj,i (kg/person/day) is the yield of tea buds and leaves per worker per day under rain and no-frost conditions on the ith day of the jth phenological period. Frost risk The frost end date can be defined as the last day of minimum temperatures ≤0 °C (Lou and Sun 2013; Cittadini et al. 2006). Differences between the frost end date and BDTP can reflect the risk that the tea trees suffered frost damage (Lou et al. 2013). When the frost end date is later than the BDTP, frost damage may occur during the Longjing-43 plucking period. However, when the frost end date is earlier than the BDTP, whether the tea trees suffer frost damage depends mainly on the minimum temperature and tea bud growth (Lou and Sun 2013). In general, the closer the frost end date to the BDTP, the higher the probability of frost damage.
Fig. 3 Relationship between the ratio of the amount of tea buds and leaves plucked during rain conditions to the amount plucked during rainless conditions (RATP) and daily precipitation from 8:00 to 20:00
Int J Biometeorol
TAD j;i Pl=4:3 0
where FRT is the accumulated portion of the daily mean temperature that exceeded 5 °C from the start to the end of the frost damage period, and Tl is the minimum temperature when frost damage occurred. When the minimum temperature was less than or equal to −5 °C, all the new tea buds were damaged. Because the lowest air temperature during one frost disaster was less than or equal to −5 °C, the value of FRT was fixed at 120.9 °C. Based on the minimum temperature, when frost damage occurred, we can determine the accumulated temperature required for tea buds to reach the plucking standard after frost damage. When frost damage occurred before the BDTP, the frost damage period was determined by the following equation: ( X X 0 FRT ≤ T X X k T¼ ð9Þ FRT − T k FRT > Tk
E j;i ¼
where ∑Tk is the accumulated portion of the daily mean temperature that exceeded 5 °C from the date of frost damage to the BDTP. ∑T is the accumulated portion of the daily mean temperature that exceeded 5 °C from the BDTP to the date that the buds and leaves met Longjing tea plucking standards. If FRT was less than or equal to ∑Tk, ∑T was set to 0, indicating that frost had not affected the tea yield. If FRT was greater than ∑Tk, frost affected the tea yield, and we calculated the accumulated portion of the daily mean temperature that exceeds 5 °C on every day; when EAT reached ∑T, the frost damage period ended.
AEO ¼ PEOðLengthÞ−EOðRRÞ−EOðFrost Þ
Economic output model for tea
No frost damage Frost damage period
ð11Þ
Integrating Eq. 11, the total economic output (E) of a worker in spring was X ð12Þ E¼ E j;i
In summary, the economic output of a specific plucking period was determined by three factors: the length of the plucking period, the daytime precipitation, and the frost damage. When the temperature was relatively low (but above 0 °C), tea production and economic output increases because of the extended plucking period. Considering the decreases in tea yield caused by daytime rainfall and frost damage, the actual economic output of a specific period (AEO) could be expressed as follows: ð13Þ
where PEO(Length) represents the potential economic output of this specific period without considering the impacts of daytime rainfall and frost damage. EO(RR) and EO(Frost) are the losses of economic output caused by daytime rainfall and frost damage, respectively. Model validation Using 2011 and 2013 data, model accuracy for the yield of tea buds and leaves was evaluated using the root mean square error (RMSE) between the predicted and observed values: vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi uX N 2 u u xpi −xoi t i¼1 RMSE ¼ ð14Þ N
Daily tea prices and amounts of tea buds and leaves plucked (yields) at four sites in 2010 and 2012 were collected to estimate economic output models for those years. The two models were evaluated using observations from 2011 and 2013. Figure 4 shows changes in tea price with ADj,i at Mingfu, Shuangcai, Changzhao, and Pangdingshan tea plantations in 2012. Thus, the relationship between tea price (Pl, yuan/kg) and ADj,i could be expressed as follows:
where xpi and xoi are the predicted and observed yields, respectively.
Pl ¼ 677:6692−256:0479AD j; i
Trend analysis
þ 44:6983AD j;i −3:0219AD j;i 2
3
ð10Þ
Based on the recorded data from these four tea plantations, making 1.0 kg of tea required 4.3 kg of tea buds and leaves. Therefore, the economic output (Ej,i) of a worker in ith day of the jth phenological period was as follows:
The longest period of meteorological data from the 19 observation sites was from 2004 to 2013. The shortest period of meteorological data from five counties was from 1972 to 2013. To reconstruct phenology over a long time, a model was based on data from the five county meteorological stations from 1972 to 2013 (Cleland et al. 2007). Then, the BDTP and phenological duration datasets from these five
Int J Biometeorol
Results Model reliability In 2011 and 2013, the differences between predicted and observed BDTPs ranged from −1 to +1 day for the 19 tea plantations. RMSEs of tea yield in the Mingfu, Shuangcai, Changzhao, and Pangdingshan tea plantations were 0.0392, 0.0351, 0.0451, and 0.0436 (kg/d), respectively. Generally, the model captured tea phenology relatively accurately.
Fig. 4 Change of tea price with ADj,i in 2012. ADj,i is the cumulative growth quantity from the beginning day of tea bud and leaf plucking to the ith day of the jth phenological period
stations were used to analyze their trends. Assuming that the tea prices from 1972 to 2013 were fixed at the 2012 level, the economic outputs in the five counties were calculated. We used linear regression analysis to examine trends in the phenological time series (Sparks and Menzel 2002; Roetzer et al. 2000; Stone et al. 1996). The Mann–Kendall test can reliably identify monotonic linear and non-linear trends in non-normal datasets with outliers; this method has been found to be an excellent tool for trend detection in a time series and has been used to assess the significance of trends in phenological series (Julien and Sobrino 2009; Menzel 2000). By comparing several statistical tests applied to a climate series, Gerstengarbe and Werner (1999) pointed out the value and strength of this method. In the present study, therefore, we used the Mann–Kendall test. Statistical significance and strong significance were inferred at P
