- Email: [email protected]
Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China
G Model
ARTICLE IN PRESS
AGWAT-4482; No. of Pages 9
Agricultural Water Management xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Agricultural Water Management journal homepage: www.elsevier.com/locate/agwat
Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China Tao Li a,b , Xinmei Hao a,∗ , Shaozhong Kang a a b
Center for Agricultural Water Research in China, China Agricultural University, Beijing 100083, PR China State Key Laboratory Base of Eco-Hydraulic in Arid Area, Xi’an University of Technology, Jinhua Road 5, Xi’an, Shaanxi,710048, PR China
a r t i c l e
i n f o
Article history: Received 9 February 2016 Received in revised form 5 May 2016 Accepted 5 May 2016 Available online xxx Keywords: Grapevine Soil water depletion Yield gap Boundary line analysis
a b s t r a c t Spatially variable soil properties, combined with the complex interrelationship between crop yield with environmental factors often lead to spatio-temporal variability of crop yield across the field. The objectives of this study were to investigate spatial variability of grape yield and estimate the potential yield at various soil water depletion levels within a vineyard of arid northwest China. Grape yield and soil water contents at different times were measured at 135 georeferenced points in 2012, and 147 points in 2013 within a 7.6-ha vineyard. Geostatistical approach was used to describe the spatial variation in grape yield, while boundary line analysis was used to estimate the potential yield at various soil water depletion levels. Some selected soil properties between the low and high yield gap groups, defined as the difference between the actual yield and the estimated potential yield, were compared. The spatial structure of yield distribution in the field was similar between the two years. The upper boundary line between grape yield with soil water depletion was best fitted by a quadratic function for both years. When soil water depletion was relatively high, soil sand content and saturated hydraulic conductivity of 0–20 cm soil were found to be significantly different between the high and low yield gap groups, while none of soil properties in the top 40 cm soil were found to be important in determining the yield gap when soil water supply was limited. Boundary line analysis method could be a valuable tool in developing better management practices to improve overall yield according to available soil water conditions in the field. © 2016 Elsevier B.V. All rights reserved.
1. Introduction Soil water content is an important variable of farmland hydrological cycle, and also an important factor affecting crop growth and yield formation. Due to inherent spatial variability of soil properties, soil water content is highly variable in space at a field scale. Better understanding of soil water depletion variations at field scale and its relationship with crop yield would help develop better management practices for improving crop yield. Soil water status within a vineyard has direct effects on vegetative growth, canopy microclimate, fruit growth, yield and quality, and the effect becomes particularly apparent at critical phonological stages of grapevine growth, such as new shoot and berry development stage (Serrano et al., 2012). Previous studies have assessed the interactive effect of water status on grapevine growth and yield under different irrigation treatments. Van Leeuwen et al.
∗ Corresponding author. E-mail address: [email protected] (X. Hao).
(2009) investigated vine shoot, berry weight and grape composition under different vine water status in a commercial vineyard in France, and found that water deficit stress anticipated shoot growth slackening and limited berry weight. Matthews and Anderson (1989) found that 30% to 40% yield increases can be obtained by increasing irrigation above the standard practice in a commercial hillside vineyard in Napa Valley near Saint Helena, California. The results of Medrano et al. (2003) indicated that moderate irrigation with about 30% potential evapotranspiration, compared with non-irrigation, could improve grape yield in two Spanish cultivars (Tempranillo and Manto Negro) of field-grown grapevine. Soil properties, through altering soil water and nutrient status, play an important role in determining grape yield and quality. Previous studies have explored the relationship between soil properties (soil texture, organic matter, and etc.) and grape yield at field scale in attempts to develop better management practices for higher yield. Van Leeuwen et al. (2004) found that the 32% of total yield variation could be explained by the soil type, and yield on sandy soil was 32 and 62% higher than that on clayey and gravelly soil, respectively, from 1996 to 2000 in three Saint-Emilion vine-
http://dx.doi.org/10.1016/j.agwat.2016.05.006 0378-3774/© 2016 Elsevier B.V. All rights reserved.
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
G Model AGWAT-4482; No. of Pages 9 2
ARTICLE IN PRESS T. Li et al. / Agricultural Water Management xxx (2016) xxx–xxx
Fig. 1. The location of the experimental site in the country, along with layout of the sampling location and elevation map of the field.
yards in France. Tardaguila et al. (2011) found that yield per plant was strongly affected by a soil index that was a linear combination of soil thickness, organic matter content, clay and cation exchange capacity in a five hectare commercial vineyard in Spain. Other soil properties have been shown to have some effect on the yield of the vines, such as soil electrical resistivity, bulk density, penetration resistance, water holding capacity (Bellvert et al., 2012; Mugnai et al., 2012; Quezada et al., 2014; Reynolds et al., 2007; Rossi et al., 2013). Given that grape yield is determined by the combined effects of many factors, it is difficult, by using traditional correlation and regression methods, to pinpoint the relationship of yield with one single factor without the confounding effects from other factors. The principle of boundary line approach was first introduced by Webb (1972), which attempts to separate yield responses to a single causal factor from responses to other independent variables that might affect the yield (Shatar and McBratney, 2004). The approach is based on the assumption that for a sufficiently large dataset, there are the maximum potential values for the response variable at different levels of a predictor variable of interest, and any points lower than the maximum values are limited by other predictor variables (Shatar and McBratney, 2004). To do the analysis, the response data are often subdivided into groups corresponding to the quantitative categories of the potential limiting factor of interest and a subset of the highest values was isolated from the response data within each group (Huang et al., 2008). Then the subset dataset was fitted as the boundary line using some statistical techniques. The boundary line represents the relationship between the response variable and the predictor variable without interferences of other independent variables. Points on the boundary line was considered as the maximum attainable values at the corresponding range of the predictor variable, which has important implications when applied into crop yield responding to a number of factors at as the yield potential at the site (Shatar and McBratney, 2004). The approach has been successfully used to assess response of crop growth or yield to a single soil property with the response could be limited by other factors. Shatar and Mcbratney (2004) identified yield responses to changes in soil properties (soil organic carbon content, K, pH and Fe) using boundary line analysis method in northern New South Wales. Huang et al. (2008) obtained the relationship between wheat yield data and wetness index by fitting the boundary line using the spline regression method in southwest Michigan in 1998, 2001, and 2004. Kitchen et al. (2003) reported that on the boundary line, yield decreased with increasing soil electrical conductivity at a Kansas field.
Boundary lines based on various independent variables could be used to estimate the maximum yield potential, and the difference between the estimated yield potential and the actual yield, i.e. yield gap, could be used to evaluate the yield status of the field (Fermont et al., 2009; Grassini et al., 2009; Hochman et al., 2009; Wairegi et al., 2010). The larger yield gap implies there is more room for improvement, and yield potential and yield gap maps could help identify low yield zones and yield-limiting factors of those areas. In this study, a 2-year experiment was carried out in a typical mature vineyard in arid northwest China. The main objectives of this study were to investigate spatial variability of grape yield, and estimate the maximum attainable yield at different crop water consumption levels at the vineyard using the boundary line analysis method. Soil properties between the estimated high and low yield gap groups were also compared to explore the possible contributing factors to the lower yield over the field.
2. Materials and methods 2.1. Field description and data collection The study was carried out at a 7.6 ha vineyard of the Shiyanghe Experimental Station for Improving Water Use Efficiency in Agriculture, Ministry of Agriculture (MOA), located in Huangtai, Wuwei, Gansu Province, China (N 37◦ 52 20 , E 102◦ 50 50 , and altitude 1581 m). Fig. 1 shows the location of an experimental site in the country. This region is limited in water resources with an average groundwater table of deeper than 25 m, mean annual precipitation of 164 mm and mean pan evaporation of about 2000 mm. The experiments were conducted during the period of April to October in 2012 and 2013. The study area was about 275 m long and 275 m wide. The 2-year data were collected mostly from regular spaced points as shown in Fig. 1, with a 25-m apart in both east-west and north-south directions for each grid. Number of samples was set as large as possible within the labor and budget constraints, and sampling locations were spread evenly over the entire experimental vineyard to be representative of the field. In 2013, 12 additional sampling points with the distance to the respective grid points less than 25 m were added to have better estimates of varirogram model parameters (Fig. 1). As a result, there was a total of 135 and 147 points in 2012 and 2013, respectively. Elevation of each sampling point was obtained through a global positioning system (GPS, Trimble Recon, USA) and the digital elevation model (DEM) for the sampling area
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
G Model AGWAT-4482; No. of Pages 9
ARTICLE IN PRESS T. Li et al. / Agricultural Water Management xxx (2016) xxx–xxx
3
Table 1 Start and end dates of major phenological stages for grapevines, and precipitation during each stage for 2012 and 2013 growing seasons. Phenological stage
Date
Duration (d)
Precipitation (mm)
2012 New shoot (NS) Flowering and berry development (FB) Veraison (V) Berry mature (BM)
5 May–1 June 2 Jun–21 July
28 50
9.8 56.6
22 Jul–17 August 18 Aug–30 September
27 43
34.0 29.0
2013 New shoot (NS) Flowering and berry development (FB) Veraison (V) Berry mature (BM)
5 May–30 May 31 May–21 July
25 52
8.2 23.0
22 Jul–15 August 16 Aug– 28 September
25 44
24.8 21.6
2012 whole 2013 whole
5 May–30 September 5 May–28 September
148 146
124.4 77.6
was created using Spatial Analyst of ArcGIS (ArcGIS10.0, ESRI, Redlands, CA) (Fig. 1). The grapevines (Vitis vinifera L. cv Merlot Noir) were established in 1999 at a spacing of 270 cm between rows oriented east–west and 100 cm between vines. Vines were trained to a vertical plane by pillar spaced at 9 m with three wires supported by a 1.5-m-high trellis (Li et al., 2014). Each vine was spur pruned to 2 or 3 nodes per spur several days earlier before being buried, usually at the end of leaf fall period. About one month after bud burst, every new shoot was removed top and there were one or two flower buds and about 10 leaves per shoot remaining after the removal. In this study the growing season of grapevine was divided into 4 phenological stages: New shoot, flowering and berry development, veraison and berry mature, and Table 1 shows the start and end date of the four phenological stages. The vineyard was furrow-irrigated four times in 2012 on 4 May, 27 May, 1 July, and 26 August, and in 2013, irrigation was applied 6 times on 16 April, 29 April, 20 May, 29 June, 30 July and 24 August (Fig. 2). The water amount for each irrigation event was about 70 mm with a salinity of 0.61 g L−1 (Zhang et al., 2010). Before each irrigation event, fertilizers usually including urea, zinc compound fertilizer, boric acid compound fertilizer and phosphate diamine were applied into holes between grapevines. Grapes of three grapevine was collected by manual harvesting at the end of berry mature stage on 26 September in 2012 and 2013 and was weighed by a scale with accuracy of 0.01 kg. Yield data of each sampling point in the study was represented by the mean values of those three grapevines. For yield data, there was a total of 129 and 143 data points in 2012 and 2013 respectively due to missing data in some points. Three soil cores surrounding each sampling locations were collected and mixed during May of 2011. Soil properties of total nitrogen (TN), total phosphorus (TP), total potassium (TK), organic carbon (OC), Ca2+ , Mg2+ , Clay, Slit and Sand at 135 grid points at 0–20 cm and 20–40 cm depths, bulk density (BD) at 135 grid points, and soil saturated hydraulic conductivity (Ks) at 147 points at 0–20 cm soil depth were measured. TN was determined using the Kjeldahl method, TP was measured by molybdenum-antimony anti-spectrophotometric method, TK was measured by alkali fusion and flame photometer method, Mg2+ and Ca2+ were measured using atomic absorption spectrophotometry, and BD was determined gravimetrically using oven drying method (Bao, 2000). Organic Carbon was measured using the potassium dichromate method (National Soil Survey Office, 1995). Soil texture analysis (clay, slit, sand) was performed using a MaterSizer2000 laser particle size analyzer (Malvern Instruments Ltd., U.K). Ks was mea-
Fig. 2. Irrigation and precipitation at the vineyard in (a) 2012 and (b) 2013.
sured using the falling head method (SL237-1999). Soil volumetric water content was measured at each sampling location at different depths: 0–20, 20–40, 40–60 cm in 2012 and 0–20, 20–40, 40–60, 60–80 and 80–100 cm in 2013, using a portable device (Diviner 2000, Sentek Pty Ltd.) approximately every 10 d, and additional measurements were taken before and after an irrigation event. Soil moisture content then was measured 17 and 19 times in 2011 and 2013 respectively. The data was calibrated by the gravimetric method at intervals during the growing season. 2.2. Data analysis The spatial structure of yield for 2012 and 2013 was evaluated using geostatistical approach. Variogram of yield in 2012 and 2013 was calculated as 1 2 [Z (xi + h) − Z (xi )] 2Nh Nh
␥ (h) =
(1)
i=1
where ␥ (h) is the semivariogram of certain variable Z, grape yield in this study, xi and xi + h are sampling locations separated by a distance h,Z (xi ) and Z (xi + h) are measured values of the variable Z at the corresponding locations, and Nh is the total number of sample pairs for the distance h. The calculated sample variograms for yield were fitted with the three commonly empirical variogram models: spherical, exponential and Gaussian model, there are three parameters of models: nugget (C0 ), sill(C0 + C1 ) and range, as shown by Burgess and Webster (1980). According to Cambardella et al. (1994), the degree of spatial dependence (DSD), defined as the ratio of nugget to sill values in a variogram model, reflects the extent of spatial autocorrelation; if DSD ≤ 25%, the variable is considered to be strongly spatially dependent, 26% < DSD < 75%, moderately spatially dependent, DSD > 75%, weakly spatially dependent.
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
G Model
ARTICLE IN PRESS
AGWAT-4482; No. of Pages 9
T. Li et al. / Agricultural Water Management xxx (2016) xxx–xxx
4
Table 2 Descriptive statistics of soil water storage change (mm) at different phenological stages. Year
Phenological stage
N
Min
Max
2012
NS FB V BM
134 133 135 134
12.36 63.38 36.98 35.90
32.92 152.17 91.50 124.29
2013
NS FB V BM
137 131 139 147
17.35 53.18 42.92 35.84
38.26 197.21 140.16 139.12
Std
CV (%)
NID* (d)
Mean Daily W
15.61 90.08 56.68 70.09
8.24 28.90 14.12 19.76
52.79 32.08 24.91 28.18
14 43 27 36
1.11 2.09 2.10 1.95
18.04 126.45 65.34 83.16
7.50 32.34 22.22 20.65
41.57 25.58 34.01 24.83
11 45 18 36
1.64 2.81 3.63 2.31
Mean
* NID: Non-irrigation days; W: soil water storage change; N: sampling number. Skw: Skewness. NS: New shoot; FB: Flowering and berry development; V: Veraison; BM: Berry mature.
The change of soil water storage (W) was calculated based on the difference of soil water content between two different times within depth from 0 to 60 cm in 2012 and from 0 to 100 cm in 2013. However, at the vineyard it usually took 4 days to complete an irrigation event and another 3-day for the field to dry up enough for soil moisture measurement. Consequently, for stages during which irrigation was applied, soil water storage change reported in this study was the sum of soil water depletion before and after irrigation events without accounting for soil water change and irrigation water input during the 7-day period for each irrigation event. For stages during which there was no irrigation, reported soil water storage change was the difference in soil water content measured close to the start and end dates of the stage. Precipitation was assumed uniformly distributed across the vineyard due to relatively small researched area. In this way, variation in irrigation amount over the vineyard, which was difficult to quantify, was not included in the water balance calculation directly, but was reflected in the measured soil water content after an irrigation event. Assuming soil water exchange between the top (60 and 100 cm for 2012 and 2013 respectively) with the deeper soil layer was negligible, soil water change of any two different times without irrigation in between was considered as evapotranspiration during the corresponding period and calculated as follows:
W = z ((t2 ) − ((t1 ) × 1000 + P
(2)
where W is the change of soil water storage between two different times (t2, t1), mm; (t1 ) and (t2 ) are soil volumetric water content at time t1 and t2 respectively, %; z is the calculated depth, m, z = 0.6 and 1.0 m in 2012 and 2013, respectively; P is precipitation, mm; In order to make the comparison more meaningful between the two years, the W of each year was standardized for the analysis. The boundary line analysis was used for studying the relationship between yield and W, and to do that, W was divided into 5 equal groups and in each group the top two yield value points were selected as upper boundary yield. The selected boundary points and the corresponding W values then were fitted with quadratic lines through simple linear regressions. This line represents the maximum yield under according W levels, while the difference between the maximum yield and the actual yield at the same w level was considered as the yield gap, which might be reducible through management practices improvements. After sorting yield gaps in ascending order, the low and high yield gap groups were classified, i.e. low group when yield gap was less than its 30th percentile, and high group when larger than the 70th percentile. In order to evaluate the potential soil factors contributing to the yield gap, differences in soil properties between the low and high yield gap groups were compared using an ANOVA model. ANOVA analyses were conducted using SAS software (Version 9.1.3, SAS Institute). Descriptive statistics were carried out by SPSS20.0 for windows software (SPSS Inc., Chicago, IL, USA, 2009).
Table 3 Descriptive statistics of grape yield (kg vine−1 ) in 2012 and 2013. Year
N
Min
Max
Mean
Std
Skewness
Kurtosis
CV (%)
2012 2013
129 143
0.04 0.06
3.18 3.10
1.32 1.28
0.70 0.70
0.34 0.53
−0.66 −0.44
53.47 54.69
Std: standard deviation; CV: coefficient of variation.
3. Results and discussion 3.1. Spatial variability of grape yield Table 2 shows descriptive statistics of W at different growing stages excluding the change during the period affected by irrigation events as described earlier. Apparently, mean daily W of different phenological stages in 2013 was higher than that in 2012, over 70% higher at veraison stage in particular, probably due to more irrigation in 2013. In both years, W showed moderate spatial variation across the field with CV ranging from 24.8 to 52.8% at different phenological stages, while the spatial variation was found larger in 2013 at three of the four stages, except for verasion stage. Among different stages, daily mean W was the largest at veraison stages in both years, which was consistent with the results reported by Zhang et al. (2010). Descriptive statistics of grape yield in 2012 and 2013 are presented in Table 3. The grape yield of 2012 and 2013 was similar with mean of 1.32 and 1.28 kg per vine respectively. According to the classification proposed by Warrick and Nielsen (1980), the yield data showed moderate heterogeneity with CV of 53.47 and 54.69% in 2012 and 2013 respectively. Variogram of grape yield was similar for 2012 and 2013 as Fig. 3 has shown. The fitted total sill value was quite similar for the two years, which was consistent with the similar standard deviation of grape yield in both years. The fitted nugget value was 0.058 for 2013, larger than 0.016 for 2012, and the difference might be due to different sampling layout in the two years. In 2013, additional 12 points with shorter distance were sampled. The relative smaller value of DSD value in both years, 3.33 and 12.08% for 2012 and 2013, respectively, indicated that grape yield was strongly spatially structured and the majority of the variation over the vineyard was spatially-dependent. The result was consistent with the findings reported by Baluja et al. (2012), who found that DSD of yield in a commercial vineyard was 13.9%. Grape yield map over the vineyard created by ordinary kriging with the yield at the sampling points showed that yield spatial distribution was similar for the two years (Fig. 3). In both years, yield was relatively higher at the southwestern part of the field. This result is in accordance with previously reported results that pattern of grape yield spatial distribution at a vineyard tended to be stable over time (Bramley and Hamilton, 2004; Bramley et al., 2011). Therefore, yield variation over the vineyard to some extent could be attributed to inherent invariant factors such as soil texture, water and nutrient levels,
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
G Model AGWAT-4482; No. of Pages 9
ARTICLE IN PRESS T. Li et al. / Agricultural Water Management xxx (2016) xxx–xxx
5
Fig. 3. Sample and fitted variogram for grape yield of (a) 2012 and c) 2013, along with yield map created through kriging based on the variogram for (b) 2012 and (c) 2013.
Fig. 4. Scatter plot, along with the upper boundary line of relative yield and standardized soil water storage change (W) in 2012 and 2013.
etc. The relative stable spatial structure of grape yield distribution across the field would be beneficial for developing and implementing some precision agricultural management practices (Bramley and Hamilton, 2004). 3.2. Boundary line analysis Fig. 4 shows scatter plot of relative yield and standardized W in 2012 and 2013, along with the respective upper boundary lines. Apparently, the spread of yield with the change of W showed a similar pattern in the two years, while the upper boundary line in the two years was both best fitted by a parabolic function (Eqs. (3)–(4)). For both years, grape yield first increased at decreasing
rate with increasing W until to the maximum yield at certain W value, then decreased at increasing rate with continue increase of W. The fitted quadratic equation was similar in shape between the two years with x-coordinate of the vertex of both parabolas close to zero, indicating that the maximum yield was achieved with the average water supply applied at the field in both years. Y2012 = −0.144x2 + 0.048x + 0.932R2 = 0.918
(3)
Y2013 = −0.145x2 − 0.082x + 0.948R2 = 0.899
(4)
where Y2012 and Y2013 are relative yield in 2012 and 2013 respectively, x is the standardized W. The maximum attainable yield in both years was similar, which could be considered as the poten-
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
G Model AGWAT-4482; No. of Pages 9 6
ARTICLE IN PRESS T. Li et al. / Agricultural Water Management xxx (2016) xxx–xxx
Fig. 5. Relationships between the relative yield with standardized soil water storage change (W) and fitted quadratic boundary lines of different phenological stages. (a) 2012 new shoot; (b) 2012 flowering and berry development; (c) 2012 veraison; (d) 2012 berry mature; (e) 2013 new shoot; (f) 2013 flowering and berry development; (g) 2013 veraison; h) 2013 berry mature.
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
G Model
ARTICLE IN PRESS
AGWAT-4482; No. of Pages 9
T. Li et al. / Agricultural Water Management xxx (2016) xxx–xxx 275
275
(a) 2012
250
High
High
7
(b) 2013
250
Low 225
225
200
200 175 North-south(m)
High
150 125
Low
High
North-south(m)
175
100
Low
High
Low
50
25
50
75
100
125
150
175
200
225
Low Low
25
High 0
Low
125
75
Low
25 0
High
150
100
75 50
Low High
Low
250
275
0
0
25
50
High
75
100
125
150
175
200
225
250
275
East-west(m)
East-west(m)
Fig. 6. The spatial distribution of the high and low yield gaps group. (a) 2012; (b) 2013.
tial yield attainable in this field with the optimum management practices. In order to understand the effect of W at different phenological stages on yield, boundary line analyses were also conducted on the relative yield and W at each stage for 2012 and 2013 (Fig. 5). Large R2 from the upper boundary points regressions at flowering and berry development stages in both years, relative to the other stages, suggested that the influence of soil water status on grape yield was more pronounced at berry development stage, as this stage generally requires larger evapotranspiration (McCarthy, 1997; Zhang et al., 2010). The standardized W value at which the maximum yield was obtained was close to the middle of whole scatter points for most of phenological stages in the two years, except that at veraison of 2012 and new shoot stage of 2013, the standardized W achieving the maximum yield was severely on the left side of the scatter, indicating soil water content at 2012 veraison stage and 2013 new shoot stage that led to the maximum attainable yield was lower than the average soil water content over the field at the respective stage. Precipitation in veraison stage of 2012 was relatively higher while at new shoot stage of 2013, an additional irrigation event was applied (Fig. 2). Therefore, the boundary analysis for different growth stages could be a useful tool in evaluating soil water or nutrient status in order to develop better management practices to meet the crop demand at different growth periods (Table 4).
3.3. Spatial distribution of yield gap zones Fig. 6 shows the spatial distribution of small and large yield gap groups in 2012 and 2013, and the larger gaps indicated the greater difference between the actual yield and the potential yield with the same amount of water consumption obtained through boundary line analysis method. The spatial distribution of yield gap zones showed some degree of similarities between the two years. The northern part of the vineyard tended to be large yield gap zone while the southwest part of the vineyard had smaller yield gap. On the other hand, the apparent discrepancy in yield gap maps between the two years suggested the inter-annual variation in the
Table 4 Fitted parameters of variogram model for grape yield (kg vine−1 ) in 2012 and 2013. Year
Model
2012 Exponential 2013 Exponential
C0 0.016 0.058
R2
C0 +C Range (m) DSD (%) RSS 0.48 0.48
53.7 30.0
3.33 12.08
−3
1.578 × 10 1.266 × 10−3
0.801 0.931
C0 : nugget; C0 + C: sill; DSD: degree of spatial dependence = C0 /(C0 + C) × 100%; R2 : the determination coefficient; RSS: the residual sums of squares.
complex interrelationship between soil properties including but not limited to soil water status and crop yield. To further investigate the possible environmental factors contributing to the relatively lower actual yield with similar amount of soil water depletion, selected soil properties at 0–20 cm and 20–40 cm soil depths were compared between the low and high yield gap groups at various W levels for 2012 and 2013 (Table 5). In this section, standardized W was divided into 3 groups: [−2 −1], [−1 0] and [0 1], which represented low, average and high crop water consumption levels, respectively. The results showed that when W was at the average or high level, Ks and sand content of 0–20 cm soil was significantly higher at the low yield gap group in both years, indicating that potential yield was more likely to be achieved at soils with high sand content and large Ks values when water supply was ample. Therefore, measures to facilitate surface soil water movement, such as periodically shallow soil plowing during the growing period, would be helpful in achieving the potential yield when there is enough water supply. When W level was high or average, TP, and OC of 20–40 cm tended to be higher for the low yield gap group in both years. These results indicated that target fertilization, such as improving OC and TP at this particular field, would be important in reducing yield gap and improving overall yield of the field when there was no water stress. Meanwhile, there was no significant difference in any of the selected soil properties of both soil depths when W level was low, suggesting that factors other than those soil properties, e.g. inherent individual grapevine difference, were likely more important contributing to the lower yield when water supply is limited.
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
G Model
ARTICLE IN PRESS
AGWAT-4482; No. of Pages 9
T. Li et al. / Agricultural Water Management xxx (2016) xxx–xxx
8
Table 5 Comparisons of selected soil properties at 0–20 cm and 20–40 cm soil depths between the low and high yield gap groups at different soil water storage change levels in 2012 and 2013. Soil depth (cm) Year
2012
W
[0 1]
Yield gap N
High 13
2013 (−2 −1]
(−1 0]
[0 1]
(−2 −1]
(−1 0]
Low 10
High 11
Low 12
High 5
Low 8
High 11
Low 9
High 16
Low 4
High 4
Low 5
0–20
TN (%) 0.05 TP (%) 0.14 TK (%) 1.77 0.99 OC (%) Ca (mg kg−1 ) 3300.17 −1 440.03 Mg (mg kg ) 17.02 Clay (%) 69.53 Slit (%) Sand (%) 13.45 BD (g cm−3 ) 1.57 1.20 Ks (cm d−1 )
0.05 0.16 1.76 0.96 3096.39 356.78 13.54 62.17 24.29* 1.60 2.34*
0.05 0.15 1.71 0.94 3641.35 415.73* 15.77 67.82 15.78 1.63 0.85
0.07 0.14 1.69 0.96 3132.22 373.63 14.16 66.57 19.27 1.58 1.96
0.05 0.15 1.67 0.78 3661.45 315.18 15.69 67.65 16.66 1.57 1.44
0.05 0.16 1.67 0.80 3220.39 368.09 14.76 65.52 19.73 1.58 1.63
0.05 0.12 1.66 0.89 3795.71 377.49 16.67 68.51 14.82 1.62 0.93
0.05 0.12 1.79 0.93 3131.16 388.75 13.93 64.32 21.75* 1.6 1.43**
0.05 0.12 1.65 0.8 3237.04 350.38** 16.75 66.82 16.44 1.64 1.54
0.07 0.13 1.68 0.86 3150.44 260.24 15.65 67.04 17.31 1.61 1.8***
0.05 0.12 1.71 0.79 3496.4 407.95 16.51 67.87 15.62 1.58 0.66
0.04 0.11 1.68 0.82 3442.81 401.48 16.07 67.32 16.61 1.57 1.37
20–40
TN (%) 0.05 TP (%) 0.07 1.81 TK (%) 0.74 OC (%) Ca (mg kg−1 ) 3734.11 370.67 Mg (mg kg−1 ) Clay (%) 17.7 68.71 Slit (%) Sand (%) 13.58
0.05 0.12* 1.8 0.83* 3640.72 347.46 15.3 66.23 18.48*
0.05 0.07 1.81 0.79 3670.3 401.38 17.16 70.36 12.48
0.05 0.09 1.81 0.92** 3581.86 373.44 16.88 69.67 13.46
0.04 0.06 1.72 0.65 4235.16 412.87 16.57 71.21 12.23
0.05 0.1 1.8 0.72 3324.77 322.99 16.57 70.69 12.74
0.05 0.06 1.74 0.67 3584.44 342.65 15.53 66.57 17.91
0.05 0.12* 1.86* 0.71 3557.64 330.43 15.49 69.02 15.49
0.05 0.07 1.7 0.75 3679.45 346.53 18.47 69.83 11.7
0.05 0.09 1.71 0.85* 3341.29 311.57 15.56 70.01 14.43
0.04 0.06 1.81 0.79 4173.36 411.99 18.73 71.55 9.72
0.04 0.1 1.79 0.86 4373.49 394.88 16.12 69.95 13.92
N: sampling number; TN: total nitrogen; TP: total phosphorus; TK: total potassium; OC: organic carbon; BD: soil bulk density; Ks: soil saturated hydraulic conductivity. *, ** and *** denote significance at the 0.05, 0.0l and 0.005 level.
4. Conclusions In this study, spatial variability of grape yield, and its association with soil water depletion were investigated at a 7.8 ha vineyard in 2012 and 2013 at the arid region of northwest China. For both years, the majority of yield variation at the field was found to be spatially structured, and the spatial structure was similar for the two years. The upper boundary line of grape yield with the change of soil water depletion could be best fitted by a quadratic function in both years and the fitted function was similar between the two years. In both years, the maximum yield was achieved at locations where soil water depletion was close to the spatial-averaged soil water depletion over the field. Based on the difference between the estimated maximum attainable yield with the actual yield, yield gap maps were created separately for the two years. Yield gap map demonstrated some degree of similarities for the two years. Comparisons of soil properties between high and low yield gap groups indicated that soil sand content and saturated conductivity were important affecting grape yield when soil water depletion level was high, while soil properties seem to have no significant impact on the yield gap when water depletion level was relatively lower. This study suggested that boundary line analysis could help identify those factors other than soil water content contributing to the low yield across the field, thus help develop precise management practices to improve yield when water supply is limited. Acknowledgements The project was supported by the National Natural Science Foundation of China (No. 51209205 and No. 51321001), and the Programme of Introducing Talents of Discipline to Universities (No. B14002). References Baluja, J., Diago, M., Goovaerts, P., Tardaguila, J., 2012. Assessment of the spatial variability of anthocyanins in grapes using a fluorescence sensor: relationships with vine vigour and yield. Precis. Agric. 13 (4), 457–472.
Bao, S.D., 2000. Soil Agrochemical Analysis. China Agricultural Press, Beijing (in Chinese). Bellvert, J., Marsal, J., Mata, M., Girona, J., 2012. Identifying irrigation zones across a 7.5-ha ‘Pinot noir’ vineyard based on the variability of vine water status and multispectral images. Irrig. Sci. 30 (6), 499–509. Bramley, R., Hamilton, R., 2004. Understanding variability in winegrape production systems: 1. Within vineyard variation in yield over several vintages. Aust. J. Grape Wine Res. 10 (1), 32–45. Bramley, R.G.V., Ouzman, J., Boss, P.K., 2011. Variation in vine vigour, grape yield and vineyard soils and topography as indicators of variation in the chemical composition of grapes, wine and wine sensory attributes. Aust. J. Grape Wine Res. 17 (2), 217–229. Burgess, T.M., Webster, R., 1980. Optimal interpolation and isarithmic mapping of soil properties: I. The semi-variogram and punctual kriging. J. Soil Sci. 31 (2), 315–331. Cambardella, C.A., Moorman, T.B., Parkin, T.B., Karlen, D.L., Novak, J.M., Turco, R.F., Konopka, A.E., 1994. Field-scale variability of soil properties in central Iowa soils. Soil Sci. Soc. Am. J. 58, 1501–1511. Fermont, V.A., Van Asten, P.J.A., Tittonell, P., Van Wijk, M.T., Giller, K.E., 2009. Closing the cassava yield gap: an analysis from smallholder farms in East Africa. Field Crop Res. 112 (1), 24–36. Grassini, P., Hall, A.J., Mercau, J.L., 2009. Benchmarking sunflower water productivity in semiarid environments. Field Crop Res. 110 (3), 251–262. Hochman, Z., Holzworth, D., Hunt, J.R., 2009. Potential to improve on-farm wheat yield and WUE in Australia. Crop Pasture Sci. 60 (8), 708–716. Huang, X., Wang, L., Yang, L., Kravchenko, A.N., 2008. Management effects on relationships of crop yields with topography represented by wetness index and precipitation. Agron. J. 100 (5), 1463–1471. Kitchen, N.R., Drummond, S.T., Lund, E.D., Sudduth, K.A., Buchleiter, G.W., 2003. Soil electrical conductivity and topography related to yield for three contrasting soil-crop systems. Agron. J. 95 (3), 483–495. Li, T., Hao, X., Kang, S., 2014. Spatiotemporal variability of soil moisture as affected by soil properties during irrigation cycles. Soil Sci. Soc. Am. J. 78, 598–608. Matthews, M.A., Anderson, M.M., 1989. Reproductive development in grape (Vitis vinifera L.): responses to seasonal water deficits. Am. J. Enol. Vitic. 40 (1), 52–60. McCarthy, M.G., 1997. The effect of transient water deficit on berry development of cv. Shiraz (Vitis vinifera L.). Aust. J. Grape Wine Res. 3 (3), 2–8. Medrano, H., Escalona, J.M., Cifre, J., Bota, J., Flexas, J., 2003. A ten-year study on the physiology of two Spanish grapevine cultivars under field conditions: effects of water availability from leaf photosynthesis to grape yield and quality. Funct. Plant Biol. 30 (6), 607–619. Mugnai, S., Masi, E., Azzarello, E., Mancuso, S., 2012. Influence of long-term application of green waste compost on soil characteristics and growth, yield and quality of grape (Vitis vinifera L.). Compost Sci. Util. 20 (1), 29–33. National Soil Survey Office (NSS), 1995. Chinese Soil Genus Records, 1-6. China Agricultural Press, Beijing (In Chinese). Quezada, C., Soriano, M.A., Díaz, J., Merino, R., Chandía, A., Campos, J., Sandoval, M., 2014. Influence of soil physical properties on grapevine yield and maturity components in an ultic palexeralf soils, Central-Southern, Chile. OJSS 4 (4), 127.
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
G Model AGWAT-4482; No. of Pages 9
ARTICLE IN PRESS T. Li et al. / Agricultural Water Management xxx (2016) xxx–xxx
Reynolds, A.G., Senchuk, I.V., Van Der Reest, C., De Savigny, C., 2007. Use of GPS and GIS for elucidation of the basis for terroir: spatial variation in an Ontario Riesling vineyard. Am. J. Enol. Vitic. 58 (2), 145–162. Rossi, R., Pollice, A., Diago, M.P., Oliveira, M., Millan, B., Bitella, G., Amato, M., Tardaguila, J., 2013. Using an automatic resistivity profiler soil sensor on-The-Go in precision viticulture. Sensors 13 (1), 1121–1136. Serrano, L., González-Flor, C., Gorchs, G., 2012. Assessment of grape yield and composition using the reflectance based Water Index in Mediterranean rainfed vineyards. Remote Sens. Environ. 118, 249–258. Shatar, T., McBratney, A., 2004. Boundary-line analysis of field-scale yield response to soil properties. J. Agric. Sci. 142 (05), 553–560. Tardaguila, J., Baluja, J., Arpon, L., Balda, P., Oliveira, M., 2011. Variations of soil properties affect the vegetative growth and yield components of Tempranillo grapevines. Precis. Agric. 12 (5), 762–773. Van Leeuwen, C., Friant, P., Choné, X., Tregoat, O., Koundouras, S., Dubourdieu, D., 2004. Influence of climate, soil, and cultivar on terroir. Am. J. Enol. Vitic. 55 (3), 207–217.
9
Van Leeuwen, C., Tregoat, O., Choné, X., Bois, B., Pernet, D., Gaudillère, J.P., 2009. Vine water status is a key factor in grape ripening and vintage quality for red Bordeaux wine. How can it be assessed for vineyard management purposes. J. Int. Sci. Vigne Vin 43 (3), 121–134. Wairegi, L.W., Van Asten, P.J., Tenywa, M.M., Bekunda, M.A., 2010. Abiotic constraints override biotic constraints in East African highland banana systems. Field Crop Res. 117 (1), 146–153. Warrick, A.W., Nielsen, D.R., 1980. Spatial variability of soil physical properties in the field. In: Hillel, D. (Ed.), Applications of Soil Physics. Academic Press, New York, pp. 319–344. Webb, R.A., 1972. Use of the boundary line in the analysis of biological data. J. Hortic. Sci. 47, 309–319. Zhang, B., Kang, S., Li, F., Tong, L., Du, T., 2010. Variation in vineyard evapotranspiration in an arid region of northwest China. Agric. Water Manage. 97 (11), 1898–1904.
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
ARTICLE IN PRESS
AGWAT-4482; No. of Pages 9
Agricultural Water Management xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Agricultural Water Management journal homepage: www.elsevier.com/locate/agwat
Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China Tao Li a,b , Xinmei Hao a,∗ , Shaozhong Kang a a b
Center for Agricultural Water Research in China, China Agricultural University, Beijing 100083, PR China State Key Laboratory Base of Eco-Hydraulic in Arid Area, Xi’an University of Technology, Jinhua Road 5, Xi’an, Shaanxi,710048, PR China
a r t i c l e
i n f o
Article history: Received 9 February 2016 Received in revised form 5 May 2016 Accepted 5 May 2016 Available online xxx Keywords: Grapevine Soil water depletion Yield gap Boundary line analysis
a b s t r a c t Spatially variable soil properties, combined with the complex interrelationship between crop yield with environmental factors often lead to spatio-temporal variability of crop yield across the field. The objectives of this study were to investigate spatial variability of grape yield and estimate the potential yield at various soil water depletion levels within a vineyard of arid northwest China. Grape yield and soil water contents at different times were measured at 135 georeferenced points in 2012, and 147 points in 2013 within a 7.6-ha vineyard. Geostatistical approach was used to describe the spatial variation in grape yield, while boundary line analysis was used to estimate the potential yield at various soil water depletion levels. Some selected soil properties between the low and high yield gap groups, defined as the difference between the actual yield and the estimated potential yield, were compared. The spatial structure of yield distribution in the field was similar between the two years. The upper boundary line between grape yield with soil water depletion was best fitted by a quadratic function for both years. When soil water depletion was relatively high, soil sand content and saturated hydraulic conductivity of 0–20 cm soil were found to be significantly different between the high and low yield gap groups, while none of soil properties in the top 40 cm soil were found to be important in determining the yield gap when soil water supply was limited. Boundary line analysis method could be a valuable tool in developing better management practices to improve overall yield according to available soil water conditions in the field. © 2016 Elsevier B.V. All rights reserved.
1. Introduction Soil water content is an important variable of farmland hydrological cycle, and also an important factor affecting crop growth and yield formation. Due to inherent spatial variability of soil properties, soil water content is highly variable in space at a field scale. Better understanding of soil water depletion variations at field scale and its relationship with crop yield would help develop better management practices for improving crop yield. Soil water status within a vineyard has direct effects on vegetative growth, canopy microclimate, fruit growth, yield and quality, and the effect becomes particularly apparent at critical phonological stages of grapevine growth, such as new shoot and berry development stage (Serrano et al., 2012). Previous studies have assessed the interactive effect of water status on grapevine growth and yield under different irrigation treatments. Van Leeuwen et al.
∗ Corresponding author. E-mail address: [email protected] (X. Hao).
(2009) investigated vine shoot, berry weight and grape composition under different vine water status in a commercial vineyard in France, and found that water deficit stress anticipated shoot growth slackening and limited berry weight. Matthews and Anderson (1989) found that 30% to 40% yield increases can be obtained by increasing irrigation above the standard practice in a commercial hillside vineyard in Napa Valley near Saint Helena, California. The results of Medrano et al. (2003) indicated that moderate irrigation with about 30% potential evapotranspiration, compared with non-irrigation, could improve grape yield in two Spanish cultivars (Tempranillo and Manto Negro) of field-grown grapevine. Soil properties, through altering soil water and nutrient status, play an important role in determining grape yield and quality. Previous studies have explored the relationship between soil properties (soil texture, organic matter, and etc.) and grape yield at field scale in attempts to develop better management practices for higher yield. Van Leeuwen et al. (2004) found that the 32% of total yield variation could be explained by the soil type, and yield on sandy soil was 32 and 62% higher than that on clayey and gravelly soil, respectively, from 1996 to 2000 in three Saint-Emilion vine-
http://dx.doi.org/10.1016/j.agwat.2016.05.006 0378-3774/© 2016 Elsevier B.V. All rights reserved.
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
G Model AGWAT-4482; No. of Pages 9 2
ARTICLE IN PRESS T. Li et al. / Agricultural Water Management xxx (2016) xxx–xxx
Fig. 1. The location of the experimental site in the country, along with layout of the sampling location and elevation map of the field.
yards in France. Tardaguila et al. (2011) found that yield per plant was strongly affected by a soil index that was a linear combination of soil thickness, organic matter content, clay and cation exchange capacity in a five hectare commercial vineyard in Spain. Other soil properties have been shown to have some effect on the yield of the vines, such as soil electrical resistivity, bulk density, penetration resistance, water holding capacity (Bellvert et al., 2012; Mugnai et al., 2012; Quezada et al., 2014; Reynolds et al., 2007; Rossi et al., 2013). Given that grape yield is determined by the combined effects of many factors, it is difficult, by using traditional correlation and regression methods, to pinpoint the relationship of yield with one single factor without the confounding effects from other factors. The principle of boundary line approach was first introduced by Webb (1972), which attempts to separate yield responses to a single causal factor from responses to other independent variables that might affect the yield (Shatar and McBratney, 2004). The approach is based on the assumption that for a sufficiently large dataset, there are the maximum potential values for the response variable at different levels of a predictor variable of interest, and any points lower than the maximum values are limited by other predictor variables (Shatar and McBratney, 2004). To do the analysis, the response data are often subdivided into groups corresponding to the quantitative categories of the potential limiting factor of interest and a subset of the highest values was isolated from the response data within each group (Huang et al., 2008). Then the subset dataset was fitted as the boundary line using some statistical techniques. The boundary line represents the relationship between the response variable and the predictor variable without interferences of other independent variables. Points on the boundary line was considered as the maximum attainable values at the corresponding range of the predictor variable, which has important implications when applied into crop yield responding to a number of factors at as the yield potential at the site (Shatar and McBratney, 2004). The approach has been successfully used to assess response of crop growth or yield to a single soil property with the response could be limited by other factors. Shatar and Mcbratney (2004) identified yield responses to changes in soil properties (soil organic carbon content, K, pH and Fe) using boundary line analysis method in northern New South Wales. Huang et al. (2008) obtained the relationship between wheat yield data and wetness index by fitting the boundary line using the spline regression method in southwest Michigan in 1998, 2001, and 2004. Kitchen et al. (2003) reported that on the boundary line, yield decreased with increasing soil electrical conductivity at a Kansas field.
Boundary lines based on various independent variables could be used to estimate the maximum yield potential, and the difference between the estimated yield potential and the actual yield, i.e. yield gap, could be used to evaluate the yield status of the field (Fermont et al., 2009; Grassini et al., 2009; Hochman et al., 2009; Wairegi et al., 2010). The larger yield gap implies there is more room for improvement, and yield potential and yield gap maps could help identify low yield zones and yield-limiting factors of those areas. In this study, a 2-year experiment was carried out in a typical mature vineyard in arid northwest China. The main objectives of this study were to investigate spatial variability of grape yield, and estimate the maximum attainable yield at different crop water consumption levels at the vineyard using the boundary line analysis method. Soil properties between the estimated high and low yield gap groups were also compared to explore the possible contributing factors to the lower yield over the field.
2. Materials and methods 2.1. Field description and data collection The study was carried out at a 7.6 ha vineyard of the Shiyanghe Experimental Station for Improving Water Use Efficiency in Agriculture, Ministry of Agriculture (MOA), located in Huangtai, Wuwei, Gansu Province, China (N 37◦ 52 20 , E 102◦ 50 50 , and altitude 1581 m). Fig. 1 shows the location of an experimental site in the country. This region is limited in water resources with an average groundwater table of deeper than 25 m, mean annual precipitation of 164 mm and mean pan evaporation of about 2000 mm. The experiments were conducted during the period of April to October in 2012 and 2013. The study area was about 275 m long and 275 m wide. The 2-year data were collected mostly from regular spaced points as shown in Fig. 1, with a 25-m apart in both east-west and north-south directions for each grid. Number of samples was set as large as possible within the labor and budget constraints, and sampling locations were spread evenly over the entire experimental vineyard to be representative of the field. In 2013, 12 additional sampling points with the distance to the respective grid points less than 25 m were added to have better estimates of varirogram model parameters (Fig. 1). As a result, there was a total of 135 and 147 points in 2012 and 2013, respectively. Elevation of each sampling point was obtained through a global positioning system (GPS, Trimble Recon, USA) and the digital elevation model (DEM) for the sampling area
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
G Model AGWAT-4482; No. of Pages 9
ARTICLE IN PRESS T. Li et al. / Agricultural Water Management xxx (2016) xxx–xxx
3
Table 1 Start and end dates of major phenological stages for grapevines, and precipitation during each stage for 2012 and 2013 growing seasons. Phenological stage
Date
Duration (d)
Precipitation (mm)
2012 New shoot (NS) Flowering and berry development (FB) Veraison (V) Berry mature (BM)
5 May–1 June 2 Jun–21 July
28 50
9.8 56.6
22 Jul–17 August 18 Aug–30 September
27 43
34.0 29.0
2013 New shoot (NS) Flowering and berry development (FB) Veraison (V) Berry mature (BM)
5 May–30 May 31 May–21 July
25 52
8.2 23.0
22 Jul–15 August 16 Aug– 28 September
25 44
24.8 21.6
2012 whole 2013 whole
5 May–30 September 5 May–28 September
148 146
124.4 77.6
was created using Spatial Analyst of ArcGIS (ArcGIS10.0, ESRI, Redlands, CA) (Fig. 1). The grapevines (Vitis vinifera L. cv Merlot Noir) were established in 1999 at a spacing of 270 cm between rows oriented east–west and 100 cm between vines. Vines were trained to a vertical plane by pillar spaced at 9 m with three wires supported by a 1.5-m-high trellis (Li et al., 2014). Each vine was spur pruned to 2 or 3 nodes per spur several days earlier before being buried, usually at the end of leaf fall period. About one month after bud burst, every new shoot was removed top and there were one or two flower buds and about 10 leaves per shoot remaining after the removal. In this study the growing season of grapevine was divided into 4 phenological stages: New shoot, flowering and berry development, veraison and berry mature, and Table 1 shows the start and end date of the four phenological stages. The vineyard was furrow-irrigated four times in 2012 on 4 May, 27 May, 1 July, and 26 August, and in 2013, irrigation was applied 6 times on 16 April, 29 April, 20 May, 29 June, 30 July and 24 August (Fig. 2). The water amount for each irrigation event was about 70 mm with a salinity of 0.61 g L−1 (Zhang et al., 2010). Before each irrigation event, fertilizers usually including urea, zinc compound fertilizer, boric acid compound fertilizer and phosphate diamine were applied into holes between grapevines. Grapes of three grapevine was collected by manual harvesting at the end of berry mature stage on 26 September in 2012 and 2013 and was weighed by a scale with accuracy of 0.01 kg. Yield data of each sampling point in the study was represented by the mean values of those three grapevines. For yield data, there was a total of 129 and 143 data points in 2012 and 2013 respectively due to missing data in some points. Three soil cores surrounding each sampling locations were collected and mixed during May of 2011. Soil properties of total nitrogen (TN), total phosphorus (TP), total potassium (TK), organic carbon (OC), Ca2+ , Mg2+ , Clay, Slit and Sand at 135 grid points at 0–20 cm and 20–40 cm depths, bulk density (BD) at 135 grid points, and soil saturated hydraulic conductivity (Ks) at 147 points at 0–20 cm soil depth were measured. TN was determined using the Kjeldahl method, TP was measured by molybdenum-antimony anti-spectrophotometric method, TK was measured by alkali fusion and flame photometer method, Mg2+ and Ca2+ were measured using atomic absorption spectrophotometry, and BD was determined gravimetrically using oven drying method (Bao, 2000). Organic Carbon was measured using the potassium dichromate method (National Soil Survey Office, 1995). Soil texture analysis (clay, slit, sand) was performed using a MaterSizer2000 laser particle size analyzer (Malvern Instruments Ltd., U.K). Ks was mea-
Fig. 2. Irrigation and precipitation at the vineyard in (a) 2012 and (b) 2013.
sured using the falling head method (SL237-1999). Soil volumetric water content was measured at each sampling location at different depths: 0–20, 20–40, 40–60 cm in 2012 and 0–20, 20–40, 40–60, 60–80 and 80–100 cm in 2013, using a portable device (Diviner 2000, Sentek Pty Ltd.) approximately every 10 d, and additional measurements were taken before and after an irrigation event. Soil moisture content then was measured 17 and 19 times in 2011 and 2013 respectively. The data was calibrated by the gravimetric method at intervals during the growing season. 2.2. Data analysis The spatial structure of yield for 2012 and 2013 was evaluated using geostatistical approach. Variogram of yield in 2012 and 2013 was calculated as 1 2 [Z (xi + h) − Z (xi )] 2Nh Nh
␥ (h) =
(1)
i=1
where ␥ (h) is the semivariogram of certain variable Z, grape yield in this study, xi and xi + h are sampling locations separated by a distance h,Z (xi ) and Z (xi + h) are measured values of the variable Z at the corresponding locations, and Nh is the total number of sample pairs for the distance h. The calculated sample variograms for yield were fitted with the three commonly empirical variogram models: spherical, exponential and Gaussian model, there are three parameters of models: nugget (C0 ), sill(C0 + C1 ) and range, as shown by Burgess and Webster (1980). According to Cambardella et al. (1994), the degree of spatial dependence (DSD), defined as the ratio of nugget to sill values in a variogram model, reflects the extent of spatial autocorrelation; if DSD ≤ 25%, the variable is considered to be strongly spatially dependent, 26% < DSD < 75%, moderately spatially dependent, DSD > 75%, weakly spatially dependent.
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
G Model
ARTICLE IN PRESS
AGWAT-4482; No. of Pages 9
T. Li et al. / Agricultural Water Management xxx (2016) xxx–xxx
4
Table 2 Descriptive statistics of soil water storage change (mm) at different phenological stages. Year
Phenological stage
N
Min
Max
2012
NS FB V BM
134 133 135 134
12.36 63.38 36.98 35.90
32.92 152.17 91.50 124.29
2013
NS FB V BM
137 131 139 147
17.35 53.18 42.92 35.84
38.26 197.21 140.16 139.12
Std
CV (%)
NID* (d)
Mean Daily W
15.61 90.08 56.68 70.09
8.24 28.90 14.12 19.76
52.79 32.08 24.91 28.18
14 43 27 36
1.11 2.09 2.10 1.95
18.04 126.45 65.34 83.16
7.50 32.34 22.22 20.65
41.57 25.58 34.01 24.83
11 45 18 36
1.64 2.81 3.63 2.31
Mean
* NID: Non-irrigation days; W: soil water storage change; N: sampling number. Skw: Skewness. NS: New shoot; FB: Flowering and berry development; V: Veraison; BM: Berry mature.
The change of soil water storage (W) was calculated based on the difference of soil water content between two different times within depth from 0 to 60 cm in 2012 and from 0 to 100 cm in 2013. However, at the vineyard it usually took 4 days to complete an irrigation event and another 3-day for the field to dry up enough for soil moisture measurement. Consequently, for stages during which irrigation was applied, soil water storage change reported in this study was the sum of soil water depletion before and after irrigation events without accounting for soil water change and irrigation water input during the 7-day period for each irrigation event. For stages during which there was no irrigation, reported soil water storage change was the difference in soil water content measured close to the start and end dates of the stage. Precipitation was assumed uniformly distributed across the vineyard due to relatively small researched area. In this way, variation in irrigation amount over the vineyard, which was difficult to quantify, was not included in the water balance calculation directly, but was reflected in the measured soil water content after an irrigation event. Assuming soil water exchange between the top (60 and 100 cm for 2012 and 2013 respectively) with the deeper soil layer was negligible, soil water change of any two different times without irrigation in between was considered as evapotranspiration during the corresponding period and calculated as follows:
W = z ((t2 ) − ((t1 ) × 1000 + P
(2)
where W is the change of soil water storage between two different times (t2, t1), mm; (t1 ) and (t2 ) are soil volumetric water content at time t1 and t2 respectively, %; z is the calculated depth, m, z = 0.6 and 1.0 m in 2012 and 2013, respectively; P is precipitation, mm; In order to make the comparison more meaningful between the two years, the W of each year was standardized for the analysis. The boundary line analysis was used for studying the relationship between yield and W, and to do that, W was divided into 5 equal groups and in each group the top two yield value points were selected as upper boundary yield. The selected boundary points and the corresponding W values then were fitted with quadratic lines through simple linear regressions. This line represents the maximum yield under according W levels, while the difference between the maximum yield and the actual yield at the same w level was considered as the yield gap, which might be reducible through management practices improvements. After sorting yield gaps in ascending order, the low and high yield gap groups were classified, i.e. low group when yield gap was less than its 30th percentile, and high group when larger than the 70th percentile. In order to evaluate the potential soil factors contributing to the yield gap, differences in soil properties between the low and high yield gap groups were compared using an ANOVA model. ANOVA analyses were conducted using SAS software (Version 9.1.3, SAS Institute). Descriptive statistics were carried out by SPSS20.0 for windows software (SPSS Inc., Chicago, IL, USA, 2009).
Table 3 Descriptive statistics of grape yield (kg vine−1 ) in 2012 and 2013. Year
N
Min
Max
Mean
Std
Skewness
Kurtosis
CV (%)
2012 2013
129 143
0.04 0.06
3.18 3.10
1.32 1.28
0.70 0.70
0.34 0.53
−0.66 −0.44
53.47 54.69
Std: standard deviation; CV: coefficient of variation.
3. Results and discussion 3.1. Spatial variability of grape yield Table 2 shows descriptive statistics of W at different growing stages excluding the change during the period affected by irrigation events as described earlier. Apparently, mean daily W of different phenological stages in 2013 was higher than that in 2012, over 70% higher at veraison stage in particular, probably due to more irrigation in 2013. In both years, W showed moderate spatial variation across the field with CV ranging from 24.8 to 52.8% at different phenological stages, while the spatial variation was found larger in 2013 at three of the four stages, except for verasion stage. Among different stages, daily mean W was the largest at veraison stages in both years, which was consistent with the results reported by Zhang et al. (2010). Descriptive statistics of grape yield in 2012 and 2013 are presented in Table 3. The grape yield of 2012 and 2013 was similar with mean of 1.32 and 1.28 kg per vine respectively. According to the classification proposed by Warrick and Nielsen (1980), the yield data showed moderate heterogeneity with CV of 53.47 and 54.69% in 2012 and 2013 respectively. Variogram of grape yield was similar for 2012 and 2013 as Fig. 3 has shown. The fitted total sill value was quite similar for the two years, which was consistent with the similar standard deviation of grape yield in both years. The fitted nugget value was 0.058 for 2013, larger than 0.016 for 2012, and the difference might be due to different sampling layout in the two years. In 2013, additional 12 points with shorter distance were sampled. The relative smaller value of DSD value in both years, 3.33 and 12.08% for 2012 and 2013, respectively, indicated that grape yield was strongly spatially structured and the majority of the variation over the vineyard was spatially-dependent. The result was consistent with the findings reported by Baluja et al. (2012), who found that DSD of yield in a commercial vineyard was 13.9%. Grape yield map over the vineyard created by ordinary kriging with the yield at the sampling points showed that yield spatial distribution was similar for the two years (Fig. 3). In both years, yield was relatively higher at the southwestern part of the field. This result is in accordance with previously reported results that pattern of grape yield spatial distribution at a vineyard tended to be stable over time (Bramley and Hamilton, 2004; Bramley et al., 2011). Therefore, yield variation over the vineyard to some extent could be attributed to inherent invariant factors such as soil texture, water and nutrient levels,
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
G Model AGWAT-4482; No. of Pages 9
ARTICLE IN PRESS T. Li et al. / Agricultural Water Management xxx (2016) xxx–xxx
5
Fig. 3. Sample and fitted variogram for grape yield of (a) 2012 and c) 2013, along with yield map created through kriging based on the variogram for (b) 2012 and (c) 2013.
Fig. 4. Scatter plot, along with the upper boundary line of relative yield and standardized soil water storage change (W) in 2012 and 2013.
etc. The relative stable spatial structure of grape yield distribution across the field would be beneficial for developing and implementing some precision agricultural management practices (Bramley and Hamilton, 2004). 3.2. Boundary line analysis Fig. 4 shows scatter plot of relative yield and standardized W in 2012 and 2013, along with the respective upper boundary lines. Apparently, the spread of yield with the change of W showed a similar pattern in the two years, while the upper boundary line in the two years was both best fitted by a parabolic function (Eqs. (3)–(4)). For both years, grape yield first increased at decreasing
rate with increasing W until to the maximum yield at certain W value, then decreased at increasing rate with continue increase of W. The fitted quadratic equation was similar in shape between the two years with x-coordinate of the vertex of both parabolas close to zero, indicating that the maximum yield was achieved with the average water supply applied at the field in both years. Y2012 = −0.144x2 + 0.048x + 0.932R2 = 0.918
(3)
Y2013 = −0.145x2 − 0.082x + 0.948R2 = 0.899
(4)
where Y2012 and Y2013 are relative yield in 2012 and 2013 respectively, x is the standardized W. The maximum attainable yield in both years was similar, which could be considered as the poten-
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
G Model AGWAT-4482; No. of Pages 9 6
ARTICLE IN PRESS T. Li et al. / Agricultural Water Management xxx (2016) xxx–xxx
Fig. 5. Relationships between the relative yield with standardized soil water storage change (W) and fitted quadratic boundary lines of different phenological stages. (a) 2012 new shoot; (b) 2012 flowering and berry development; (c) 2012 veraison; (d) 2012 berry mature; (e) 2013 new shoot; (f) 2013 flowering and berry development; (g) 2013 veraison; h) 2013 berry mature.
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
G Model
ARTICLE IN PRESS
AGWAT-4482; No. of Pages 9
T. Li et al. / Agricultural Water Management xxx (2016) xxx–xxx 275
275
(a) 2012
250
High
High
7
(b) 2013
250
Low 225
225
200
200 175 North-south(m)
High
150 125
Low
High
North-south(m)
175
100
Low
High
Low
50
25
50
75
100
125
150
175
200
225
Low Low
25
High 0
Low
125
75
Low
25 0
High
150
100
75 50
Low High
Low
250
275
0
0
25
50
High
75
100
125
150
175
200
225
250
275
East-west(m)
East-west(m)
Fig. 6. The spatial distribution of the high and low yield gaps group. (a) 2012; (b) 2013.
tial yield attainable in this field with the optimum management practices. In order to understand the effect of W at different phenological stages on yield, boundary line analyses were also conducted on the relative yield and W at each stage for 2012 and 2013 (Fig. 5). Large R2 from the upper boundary points regressions at flowering and berry development stages in both years, relative to the other stages, suggested that the influence of soil water status on grape yield was more pronounced at berry development stage, as this stage generally requires larger evapotranspiration (McCarthy, 1997; Zhang et al., 2010). The standardized W value at which the maximum yield was obtained was close to the middle of whole scatter points for most of phenological stages in the two years, except that at veraison of 2012 and new shoot stage of 2013, the standardized W achieving the maximum yield was severely on the left side of the scatter, indicating soil water content at 2012 veraison stage and 2013 new shoot stage that led to the maximum attainable yield was lower than the average soil water content over the field at the respective stage. Precipitation in veraison stage of 2012 was relatively higher while at new shoot stage of 2013, an additional irrigation event was applied (Fig. 2). Therefore, the boundary analysis for different growth stages could be a useful tool in evaluating soil water or nutrient status in order to develop better management practices to meet the crop demand at different growth periods (Table 4).
3.3. Spatial distribution of yield gap zones Fig. 6 shows the spatial distribution of small and large yield gap groups in 2012 and 2013, and the larger gaps indicated the greater difference between the actual yield and the potential yield with the same amount of water consumption obtained through boundary line analysis method. The spatial distribution of yield gap zones showed some degree of similarities between the two years. The northern part of the vineyard tended to be large yield gap zone while the southwest part of the vineyard had smaller yield gap. On the other hand, the apparent discrepancy in yield gap maps between the two years suggested the inter-annual variation in the
Table 4 Fitted parameters of variogram model for grape yield (kg vine−1 ) in 2012 and 2013. Year
Model
2012 Exponential 2013 Exponential
C0 0.016 0.058
R2
C0 +C Range (m) DSD (%) RSS 0.48 0.48
53.7 30.0
3.33 12.08
−3
1.578 × 10 1.266 × 10−3
0.801 0.931
C0 : nugget; C0 + C: sill; DSD: degree of spatial dependence = C0 /(C0 + C) × 100%; R2 : the determination coefficient; RSS: the residual sums of squares.
complex interrelationship between soil properties including but not limited to soil water status and crop yield. To further investigate the possible environmental factors contributing to the relatively lower actual yield with similar amount of soil water depletion, selected soil properties at 0–20 cm and 20–40 cm soil depths were compared between the low and high yield gap groups at various W levels for 2012 and 2013 (Table 5). In this section, standardized W was divided into 3 groups: [−2 −1], [−1 0] and [0 1], which represented low, average and high crop water consumption levels, respectively. The results showed that when W was at the average or high level, Ks and sand content of 0–20 cm soil was significantly higher at the low yield gap group in both years, indicating that potential yield was more likely to be achieved at soils with high sand content and large Ks values when water supply was ample. Therefore, measures to facilitate surface soil water movement, such as periodically shallow soil plowing during the growing period, would be helpful in achieving the potential yield when there is enough water supply. When W level was high or average, TP, and OC of 20–40 cm tended to be higher for the low yield gap group in both years. These results indicated that target fertilization, such as improving OC and TP at this particular field, would be important in reducing yield gap and improving overall yield of the field when there was no water stress. Meanwhile, there was no significant difference in any of the selected soil properties of both soil depths when W level was low, suggesting that factors other than those soil properties, e.g. inherent individual grapevine difference, were likely more important contributing to the lower yield when water supply is limited.
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
G Model
ARTICLE IN PRESS
AGWAT-4482; No. of Pages 9
T. Li et al. / Agricultural Water Management xxx (2016) xxx–xxx
8
Table 5 Comparisons of selected soil properties at 0–20 cm and 20–40 cm soil depths between the low and high yield gap groups at different soil water storage change levels in 2012 and 2013. Soil depth (cm) Year
2012
W
[0 1]
Yield gap N
High 13
2013 (−2 −1]
(−1 0]
[0 1]
(−2 −1]
(−1 0]
Low 10
High 11
Low 12
High 5
Low 8
High 11
Low 9
High 16
Low 4
High 4
Low 5
0–20
TN (%) 0.05 TP (%) 0.14 TK (%) 1.77 0.99 OC (%) Ca (mg kg−1 ) 3300.17 −1 440.03 Mg (mg kg ) 17.02 Clay (%) 69.53 Slit (%) Sand (%) 13.45 BD (g cm−3 ) 1.57 1.20 Ks (cm d−1 )
0.05 0.16 1.76 0.96 3096.39 356.78 13.54 62.17 24.29* 1.60 2.34*
0.05 0.15 1.71 0.94 3641.35 415.73* 15.77 67.82 15.78 1.63 0.85
0.07 0.14 1.69 0.96 3132.22 373.63 14.16 66.57 19.27 1.58 1.96
0.05 0.15 1.67 0.78 3661.45 315.18 15.69 67.65 16.66 1.57 1.44
0.05 0.16 1.67 0.80 3220.39 368.09 14.76 65.52 19.73 1.58 1.63
0.05 0.12 1.66 0.89 3795.71 377.49 16.67 68.51 14.82 1.62 0.93
0.05 0.12 1.79 0.93 3131.16 388.75 13.93 64.32 21.75* 1.6 1.43**
0.05 0.12 1.65 0.8 3237.04 350.38** 16.75 66.82 16.44 1.64 1.54
0.07 0.13 1.68 0.86 3150.44 260.24 15.65 67.04 17.31 1.61 1.8***
0.05 0.12 1.71 0.79 3496.4 407.95 16.51 67.87 15.62 1.58 0.66
0.04 0.11 1.68 0.82 3442.81 401.48 16.07 67.32 16.61 1.57 1.37
20–40
TN (%) 0.05 TP (%) 0.07 1.81 TK (%) 0.74 OC (%) Ca (mg kg−1 ) 3734.11 370.67 Mg (mg kg−1 ) Clay (%) 17.7 68.71 Slit (%) Sand (%) 13.58
0.05 0.12* 1.8 0.83* 3640.72 347.46 15.3 66.23 18.48*
0.05 0.07 1.81 0.79 3670.3 401.38 17.16 70.36 12.48
0.05 0.09 1.81 0.92** 3581.86 373.44 16.88 69.67 13.46
0.04 0.06 1.72 0.65 4235.16 412.87 16.57 71.21 12.23
0.05 0.1 1.8 0.72 3324.77 322.99 16.57 70.69 12.74
0.05 0.06 1.74 0.67 3584.44 342.65 15.53 66.57 17.91
0.05 0.12* 1.86* 0.71 3557.64 330.43 15.49 69.02 15.49
0.05 0.07 1.7 0.75 3679.45 346.53 18.47 69.83 11.7
0.05 0.09 1.71 0.85* 3341.29 311.57 15.56 70.01 14.43
0.04 0.06 1.81 0.79 4173.36 411.99 18.73 71.55 9.72
0.04 0.1 1.79 0.86 4373.49 394.88 16.12 69.95 13.92
N: sampling number; TN: total nitrogen; TP: total phosphorus; TK: total potassium; OC: organic carbon; BD: soil bulk density; Ks: soil saturated hydraulic conductivity. *, ** and *** denote significance at the 0.05, 0.0l and 0.005 level.
4. Conclusions In this study, spatial variability of grape yield, and its association with soil water depletion were investigated at a 7.8 ha vineyard in 2012 and 2013 at the arid region of northwest China. For both years, the majority of yield variation at the field was found to be spatially structured, and the spatial structure was similar for the two years. The upper boundary line of grape yield with the change of soil water depletion could be best fitted by a quadratic function in both years and the fitted function was similar between the two years. In both years, the maximum yield was achieved at locations where soil water depletion was close to the spatial-averaged soil water depletion over the field. Based on the difference between the estimated maximum attainable yield with the actual yield, yield gap maps were created separately for the two years. Yield gap map demonstrated some degree of similarities for the two years. Comparisons of soil properties between high and low yield gap groups indicated that soil sand content and saturated conductivity were important affecting grape yield when soil water depletion level was high, while soil properties seem to have no significant impact on the yield gap when water depletion level was relatively lower. This study suggested that boundary line analysis could help identify those factors other than soil water content contributing to the low yield across the field, thus help develop precise management practices to improve yield when water supply is limited. Acknowledgements The project was supported by the National Natural Science Foundation of China (No. 51209205 and No. 51321001), and the Programme of Introducing Talents of Discipline to Universities (No. B14002). References Baluja, J., Diago, M., Goovaerts, P., Tardaguila, J., 2012. Assessment of the spatial variability of anthocyanins in grapes using a fluorescence sensor: relationships with vine vigour and yield. Precis. Agric. 13 (4), 457–472.
Bao, S.D., 2000. Soil Agrochemical Analysis. China Agricultural Press, Beijing (in Chinese). Bellvert, J., Marsal, J., Mata, M., Girona, J., 2012. Identifying irrigation zones across a 7.5-ha ‘Pinot noir’ vineyard based on the variability of vine water status and multispectral images. Irrig. Sci. 30 (6), 499–509. Bramley, R., Hamilton, R., 2004. Understanding variability in winegrape production systems: 1. Within vineyard variation in yield over several vintages. Aust. J. Grape Wine Res. 10 (1), 32–45. Bramley, R.G.V., Ouzman, J., Boss, P.K., 2011. Variation in vine vigour, grape yield and vineyard soils and topography as indicators of variation in the chemical composition of grapes, wine and wine sensory attributes. Aust. J. Grape Wine Res. 17 (2), 217–229. Burgess, T.M., Webster, R., 1980. Optimal interpolation and isarithmic mapping of soil properties: I. The semi-variogram and punctual kriging. J. Soil Sci. 31 (2), 315–331. Cambardella, C.A., Moorman, T.B., Parkin, T.B., Karlen, D.L., Novak, J.M., Turco, R.F., Konopka, A.E., 1994. Field-scale variability of soil properties in central Iowa soils. Soil Sci. Soc. Am. J. 58, 1501–1511. Fermont, V.A., Van Asten, P.J.A., Tittonell, P., Van Wijk, M.T., Giller, K.E., 2009. Closing the cassava yield gap: an analysis from smallholder farms in East Africa. Field Crop Res. 112 (1), 24–36. Grassini, P., Hall, A.J., Mercau, J.L., 2009. Benchmarking sunflower water productivity in semiarid environments. Field Crop Res. 110 (3), 251–262. Hochman, Z., Holzworth, D., Hunt, J.R., 2009. Potential to improve on-farm wheat yield and WUE in Australia. Crop Pasture Sci. 60 (8), 708–716. Huang, X., Wang, L., Yang, L., Kravchenko, A.N., 2008. Management effects on relationships of crop yields with topography represented by wetness index and precipitation. Agron. J. 100 (5), 1463–1471. Kitchen, N.R., Drummond, S.T., Lund, E.D., Sudduth, K.A., Buchleiter, G.W., 2003. Soil electrical conductivity and topography related to yield for three contrasting soil-crop systems. Agron. J. 95 (3), 483–495. Li, T., Hao, X., Kang, S., 2014. Spatiotemporal variability of soil moisture as affected by soil properties during irrigation cycles. Soil Sci. Soc. Am. J. 78, 598–608. Matthews, M.A., Anderson, M.M., 1989. Reproductive development in grape (Vitis vinifera L.): responses to seasonal water deficits. Am. J. Enol. Vitic. 40 (1), 52–60. McCarthy, M.G., 1997. The effect of transient water deficit on berry development of cv. Shiraz (Vitis vinifera L.). Aust. J. Grape Wine Res. 3 (3), 2–8. Medrano, H., Escalona, J.M., Cifre, J., Bota, J., Flexas, J., 2003. A ten-year study on the physiology of two Spanish grapevine cultivars under field conditions: effects of water availability from leaf photosynthesis to grape yield and quality. Funct. Plant Biol. 30 (6), 607–619. Mugnai, S., Masi, E., Azzarello, E., Mancuso, S., 2012. Influence of long-term application of green waste compost on soil characteristics and growth, yield and quality of grape (Vitis vinifera L.). Compost Sci. Util. 20 (1), 29–33. National Soil Survey Office (NSS), 1995. Chinese Soil Genus Records, 1-6. China Agricultural Press, Beijing (In Chinese). Quezada, C., Soriano, M.A., Díaz, J., Merino, R., Chandía, A., Campos, J., Sandoval, M., 2014. Influence of soil physical properties on grapevine yield and maturity components in an ultic palexeralf soils, Central-Southern, Chile. OJSS 4 (4), 127.
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006
G Model AGWAT-4482; No. of Pages 9
ARTICLE IN PRESS T. Li et al. / Agricultural Water Management xxx (2016) xxx–xxx
Reynolds, A.G., Senchuk, I.V., Van Der Reest, C., De Savigny, C., 2007. Use of GPS and GIS for elucidation of the basis for terroir: spatial variation in an Ontario Riesling vineyard. Am. J. Enol. Vitic. 58 (2), 145–162. Rossi, R., Pollice, A., Diago, M.P., Oliveira, M., Millan, B., Bitella, G., Amato, M., Tardaguila, J., 2013. Using an automatic resistivity profiler soil sensor on-The-Go in precision viticulture. Sensors 13 (1), 1121–1136. Serrano, L., González-Flor, C., Gorchs, G., 2012. Assessment of grape yield and composition using the reflectance based Water Index in Mediterranean rainfed vineyards. Remote Sens. Environ. 118, 249–258. Shatar, T., McBratney, A., 2004. Boundary-line analysis of field-scale yield response to soil properties. J. Agric. Sci. 142 (05), 553–560. Tardaguila, J., Baluja, J., Arpon, L., Balda, P., Oliveira, M., 2011. Variations of soil properties affect the vegetative growth and yield components of Tempranillo grapevines. Precis. Agric. 12 (5), 762–773. Van Leeuwen, C., Friant, P., Choné, X., Tregoat, O., Koundouras, S., Dubourdieu, D., 2004. Influence of climate, soil, and cultivar on terroir. Am. J. Enol. Vitic. 55 (3), 207–217.
9
Van Leeuwen, C., Tregoat, O., Choné, X., Bois, B., Pernet, D., Gaudillère, J.P., 2009. Vine water status is a key factor in grape ripening and vintage quality for red Bordeaux wine. How can it be assessed for vineyard management purposes. J. Int. Sci. Vigne Vin 43 (3), 121–134. Wairegi, L.W., Van Asten, P.J., Tenywa, M.M., Bekunda, M.A., 2010. Abiotic constraints override biotic constraints in East African highland banana systems. Field Crop Res. 117 (1), 146–153. Warrick, A.W., Nielsen, D.R., 1980. Spatial variability of soil physical properties in the field. In: Hillel, D. (Ed.), Applications of Soil Physics. Academic Press, New York, pp. 319–344. Webb, R.A., 1972. Use of the boundary line in the analysis of biological data. J. Hortic. Sci. 47, 309–319. Zhang, B., Kang, S., Li, F., Tong, L., Du, T., 2010. Variation in vineyard evapotranspiration in an arid region of northwest China. Agric. Water Manage. 97 (11), 1898–1904.
Please cite this article in press as: Li, T., et al., Spatial variability of grape yield and its association with soil water depletion within a vineyard of arid northwest China. Agric. Water Manage. (2016), http://dx.doi.org/10.1016/j.agwat.2016.05.006