Soil water retention is important for the study of water availability to germinating weed seeds. Six soil water retention models (Campbell, Brooks-Corey, four- and five-parameter van Genuchten, Tani, and Russo) with residual soil water parameter derivations were evaluated to describe water retention for weed seed germination at minimum threshold soil water potential for three hillslope positions. The Campbell, Brooks-Corey, and four-parameter van Genuchten model with modified or estimated forms of the residual parameter had superior but similar data fit. The Campbell model underestimated water retention at a potential less than −0.5 MPa for the upper hillslope that could result in underestimating seed germination. The Tani and Russo models overestimated water retention at a potential less than −0.1 MPa for all hillslope positions. Model selection and residual parameter specification are important for weed seed germination by representing water retention at the level of minimum threshold water potential for germination. Weed seed germination models driven by the hydrothermal soil environment rely on the best-fitting soil water retention model to produce dynamic predictions of seed germination.
The soil water retention characteristic (SWRC) is a basic hydrophysical property of the soil that relates the water content of soil water to its energy state [
The timing of seed germination is a function of soil water potential [
One of the greatest challenges in characterizing the SWRC for the shallow depth of the seedling recruitment zone (soil layer from which seeds germinate and emerge) across field topography is obtaining the parameters of the soil hydrological property. Determining the SWRC by direct measurement is time-consuming to obtain sufficient representation of a field due to spatial variability of soil properties [
Diverse SWRC often exist in soils along a hillslope as a result of variability in texture and pore-size distribution [
The SWRC is expressed as a nonlinear function in which water content decreases from saturation to dryness in a sigmoidal manner with decreasing (more negative) water potential [
The SWRC in the shallow seedling recruitment zone was evaluated in three hillslope positions (summit, backslope, and toeslope) on two hillslopes having opposing aspects in an annually cropped agricultural field at Graysville, MB, Canada. Each hillslope contained six replications that were arranged perpendicular to the hillslope gradient to maximize homogeneous soil conditions at each hillslope position.
Soil samples were extracted from the 25–50 mm soil depth in three replicates by pressing 50 mm diameter by 25 mm deep rings into the soil. The 25–50 mm soil depth is representative of average weed seedling recruitment depth in a conventionally tilled field [
Subsamples of soil weighing 12 g were placed in shallow Plexiglas cylinders and saturated for 24 h. Volumetric water content at saturation was determined by placing the soil samples on saturated porous plates in covered ceramic suction cups to equilibrate saturation water content over 48 h. Volumetric water content was determined on desaturation at pressures of 0.01, 0.03, 0.05, and 0.1 MPa with a model 1600 0.5-MPa pressure plate extractor with a 0.1-MPa porous ceramic pressure plate and determined at pressures of 0.5 and 1.5 MPa with a model 1500 1.5-MPa pressure plate extractor with 0.5-MPa and 1.5-MPa porous ceramic pressure plates (Soil Moisture Equipment Corp., Santa Barbara, CA, USA) according to methods by [
Soil particle size was analysed by the hydrometer method [
Commonly used SWRC equations having a low number of parameters were evaluated for their ability to fit the data. The Brooks-Corey (BC) model [
The fitted SWRC for the 25–50 mm soil depth of the hillslope positions was determined by nonlinear analysis with the likelihood-based NLMIXED procedure using iterative optimization to compute the parameter estimates [
The Akaike information criterion (AIC) [
Model comparisons were facilitated using delta AIC
Model inference was based on evidence ratios that evaluate the relative likelihood of model pairs. Evidence ratios are calculated as the ratio of Akaike weights
The best-fitting curve according to Akaike weights from the
Mean water contents for the best-fitting
The hillslope positions (summit, backslope, and toeslope) represented a categorical range in soil physical properties along the hillslope. Surface soil texture ranged from loamy fine sand to silty clay (Table
Average measured soil physical properties for the shallow seedling recruitment zone of the hillslope positions.
Hillslope position | Textural class | Clay | Silt | Sand | Organic matter | Bulk density |
(g/kg) | (g/cm3) | |||||
SW summit | Silt loam | 56.7d | 746.7a | 196.7c | 35.3d | 1.12a |
SW backslope | Silt loam | 96.7c | 616.7c | 286.7b | 38.7cd | 1.04a |
SW toeslope | Clay loam | 286.7a | 323.3e | 390.0a | 60.6a | 0.86b |
NE summit | Silt loam | 83.3cd | 690.0b | 226.7c | 44.0bc | 1.09a |
NE backslope | Silt loam | 93.3c | 640.0bc | 266.7b | 34.7d | 1.13a |
NE toeslope | Loam | 156.7b | 463.3d | 380.0a | 46.0b | 1.05a |
Soil property means within a column, followed by different letters are significantly different at
The respective SWRC models (excluding CA) all exhibited better fit with modified
Average fitted parameter and AIC values for the soil water retention models across hillslopes.
Hillslope position | Average parameter values (SE) | AICg | ||||||
| Model | |||||||
(cm3/cm3) | (MPa) | (dimensionless) | ||||||
Summit | 0.379 | |||||||
Absent | CA | 0.000 (—) | −0.0022 (0.0006) | — | 3.29 (0.34) | — | −27.6 | |
Measured | BC | 0.065 (—) | −0.0042 (0.0006) | — | 0.60 (0.05) | — | −28.5 | |
VG5 | 0.065 (—) | — | −0.0042 (0.0006) | 11.14 (0.04) | 0.054 (0.005) | −26.5 | ||
VG4 | 0.065 (—) | — | −0.0053 (0.0010) | 1.65 (0.07) | — | −28.1 | ||
TA | 0.065 (—) | — | −0.0115 (0.0035) | — | — | −11.1 | ||
RU | 0.065 (—) | −0.0011 (<.0001) | — | 16.66 (5.07) | — | −13.5 | ||
Modified | BC | 0.064 (0.011) | −0.0042 (0.0008) | — | 0.69 (0.16) | — | −30.1 | |
VG5 | 0.064 (0.008) | — | −0.0042 (<.0001) | 11.83 (<.01) | 0.059 (0.006) | −28.1 | ||
VG4 | 0.066 (0.011) | — | −0.0053 (0.0012) | 1.79 (0.20) | — | −29.4 | ||
TA | 0.088 (0.005) | — | −0.0055 (0.0007) | — | — | −24.1 | ||
RU | 0.087 (0.005) | −0.0011 (<.0001) | — | 5.43 (1.40) | — | −24.5 | ||
Estimated | BC | 0.051 (0.015) | −0.0037 (0.0007) | — | 0.51 (0.10) | — | −29.4 | |
VG5 | 0.051 (0.011) | — | −0.0037 (<.0001) | 11.05 (<.01) | 0.046 (0.004) | −27.4 | ||
VG4 | 0.055 (0.015) | — | −0.0048 (0.0012) | 1.58 (0.13) | — | −28.5 | ||
TA | 0.106 (0.014) | — | −0.0070 (0.0019) | — | — | −15.6 | ||
RU | 0.098 (0.015) | −0.0011 (<.0001) | — | 9.93 (4.11) | — | −16.8 | ||
Backslope | 0.391 | |||||||
Absent | CA | 0.000 (—) | −0.0023 (0.0004) | — | 3.61 (0.23) | — | −32.1 | |
Measured | BC | 0.073 (—) | −0.0040 (0.0006) | — | 0.54 (0.04) | — | −28.8 | |
VG5 | 0.073 (—) | — | −0.0040 (0.0006) | 21.21 (0.05) | 0.025 (0.002) | −26.8 | ||
VG4 | 0.073 (—) | — | −0.0053 (0.0009) | 1.59 (0.06) | — | −29.4 | ||
TA | 0.073 (—) | — | −0.0140 (0.0008) | — | — | −10.2 | ||
RU | 0.073 (—) | −0.0011 (<.0001) | — | 20.87 (6.05) | — | −12.8 | ||
Modified | BC | 0.061 (0.014) | −0.0034 (0.0006) | — | 0.51 (0.10) | — | −33.6 | |
VG5 | 0.061 (0.010) | — | −0.0034 (<.0001) | 9.68 (<.01) | 0.052 (0.005) | −31.6 | ||
VG4 | 0.066 (0.009) | — | −0.0044 (<.0001) | 1.58 (0.06) | — | −32.9 | ||
TA | 0.102 (0.006) | — | −0.0055 (0.0009) | — | — | −21.9 | ||
RU | 0.100 (0.007) | −0.0011 (<.0001) | — | 5.57 (1.94) | — | −22.5 | ||
Estimated | BC | 0.042 (0.015) | −0.0032 (0.0005) | — | 0.39 (0.06) | — | −33.3 | |
VG5 | 0.042 (0.011) | — | −0.0032 (<.0001) | 9.38 (<.01) | 0.042 (0.003) | −31.3 | ||
VG4 | 0.049 (0.011) | — | −0.0042 (<.0001) | 1.45 (0.03) | — | −32.3 | ||
TA | 0.120 (0.018) | — | −0.0078 (0.0031) | — | — | −13.6 | ||
RU | 0.108 (0.019) | −0.0011 (<.0001) | — | 13.00 (5.66) | — | −15.4 | ||
Toeslope | 0.417 | |||||||
Absent | CA | 0.000 (—) | −0.0014 (0.0003) | — | 5.63 (0.30) | — | −34.5 | |
Measured | BC | 0.120 (—) | −0.0030 (0.0009) | — | 0.41 (0.06) | — | −23.3 | |
VG5 | 0.120 (—) | — | −0.0030 (0.0009) | 19.63 (0.33) | 0.021 (0.003) | −21.3 | ||
VG4 | 0.120 (—) | — | −0.0039 (0.0013) | 1.44 (0.07) | — | −24.0 | ||
TA | 0.120 (—) | — | −0.0177 (0.0057) | — | — | −8.1 | ||
RU | 0.120 (—) | −0.0011 (<.0001) | — | 29.02 (10.26) | — | −10.5 | ||
Modified | BC | 0.000 (—) | −0.0014 (0.0003) | — | 0.18 (0.01) | — | −32.5 | |
VG5 | 0.000 (—) | — | −0.0014 (<.0001) | 14.34 (0.24) | 0.012 (<.001) | −30.5 | ||
VG4 | 0.000 (—) | — | −0.0016 (<.0001) | 1.18 (<.01) | — | −32.8 | ||
TA | 0.149 (0.005) | — | −0.0039 (0.0008) | — | — | −23.4 | ||
RU | 0.149 (0.005) | −0.0011 (<.0001) | — | 2.35 (1.23) | — | −23.4 | ||
Estimated | BC | 0.000 (—) | −0.0014 (0.0003) | — | 0.18 (0.01) | — | −32.5 | |
VG5 | 0.000 (—) | — | −0.0014 (<.0001) | 14.45 (1.64) | 0.012 (0.001) | −30.5 | ||
VG4 | 0.000 (—) | — | −0.0016 (<.0001) | 1.18 (<.01) | — | −32.8 | ||
TA | 0.182 (0.020) | — | −0.0065 (0.0019) | — | — | −11.3 | ||
RU | 0.162 (0.028) | −0.0011 (<.0001) | — | 15.39 (10.95) | — | −12.4 |
a
b
c
d
e
f
gA lower value of Akaike Information Criterion (AIC) indicates a better model fit.
Fitted SWRC for soil water contents
Summit
Backslope
Toeslope
Fitted SWRC for soil water contents
Summit
Backslope
Toeslope
Fitted SWRC for soil water contents
Summit
Backslope
Toeslope
The evidence ratio provided a discrete comparison of the water retention models in a pairwise manner. The water retention models were compared to the BC model with
Akaike weights and evidence ratios for the soil water retention models.
Akaike weighta | Evidence ratiob | Rank | |
---|---|---|---|
Absent | |||
CA | 0.143 | 1.4 | 4 |
Measured | |||
BC | 0.015 | 13.2 | 9 |
VG5 | 0.006 | 35.9 | 10 |
VG4 | 0.018 | 11.4 | 8 |
TA | 0.000 | 67, 045.1 | 16 |
RU | 0.000 | 19, 531.6 | 15 |
Modified | |||
BC | 0.203 | 1.0 | 1 |
VG5 | 0.075 | 2.7 | 6 |
VG4 | 0.169 | 1.2 | 3 |
TA | 0.002 | 85.3 | 12 |
RU | 0.003 | 72.2 | 11 |
Estimated | |||
BC | 0.172 | 1.2 | 2 |
VG5 | 0.063 | 3.1 | 7 |
VG4 | 0.132 | 1.5 | 5 |
TA | 0.000 | 10, 542.0 | 14 |
RU | 0.000 | 5, 323.0 | 13 |
bEvidence ratio
The BC and VG5 models represented the SWRC similarly except at the wetter end of the curve, where
The BC and VG4 models with
The VG5 model with modified and estimated forms of
The CA and BC models may provide unrealistic description of water retention at the wet end of the SWRC where
Minimum threshold water potential for germination of weed species grouped by intervals of germination sensitivity to water potential.
Sensitivity interval (MPa) | Common name | Scientific name | Reference |
---|---|---|---|
B (−0.1 to −0.5) | |||
−0.10 | Common cocklebur | [ | |
−0.20 | Redroot pigweed | [ | |
C (−0.5 to −1.0) | |||
−0.50 | Common groundsel | [ | |
−0.50 | Large crabgrass | [ | |
−0.60 | Wild oat | [ | |
−0.64 | Common lambsquarters | [ | |
−0.64 | Velvetleaf | [ | |
−0.69 | Yellow foxtail | [ | |
−0.70 | Green foxtail | [ | |
−0.80 | Goosegrass | [ | |
−0.80 | Common ragweed | [ | |
−0.83 | Large crabgrass | [ | |
D (−1.0 to −1.5) | |||
−1.13 | Common chickweed | [ | |
−1.15 | Rigid ryegrass | [ | |
−1.20 | Wild oat | [ | |
−1.21 | Goosegrass | [ | |
−1.50 | Wild oat | [ | |
−1.50 | Perennial ryegrass | [ | |
−1.53 | Blackgrass | [ |
The VG4 model was superior to the VG5 model no matter which derivation of
The SWRC for the coarse soils in the summit and backslope positions were best described by the BC model, based on
The measured water potential interval (−0.01 to −1.5 MPa) was divided into four subintervals to represent sensitivity groups of minimum threshold water potential for germination of weed species (Table
The best-fitting SWRC for each model from the various
The mean water content was averaged across hillslopes because water retention was not different between hillslopes (Table
Mean water content for water potential intervals based on weed seed germination sensitivity to water potential.
Mean water content | |||||
Water potential interval (MPa) | |||||
Main effect | Full | A | B | C | D |
−0.01 to −1.5 | −0.01 to −0.1 | −0.1 to −0.5 | −0.5 to −1.0 | −1.0 to −1.5 | |
(cm3/cm3) | |||||
Hillslope | |||||
SW | 0.158 | 0.202 | 0.132 | 0.108 | 0.098 |
NE | 0.150 | 0.190 | 0.126 | 0.102 | 0.092 |
Hillslope position | |||||
Summit | 0.124b | 0.164b | 0.099b | 0.080b | 0.072b |
Backslope | 0.141b | 0.185b | 0.114b | 0.091b | 0.082b |
Toeslope | 0.196a | 0.239a | 0.175a | 0.145a | 0.132a |
Model | |||||
CA | 0.153c | 0.202a | 0.125d | 0.095c | 0.083c |
BC | 0.153bc | 0.198b | 0.126c | 0.100b | 0.090b |
VG5 | 0.153bc | 0.199b | 0.126c | 0.100b | 0.089b |
VG4 | 0.153b | 0.198b | 0.126c | 0.101b | 0.091b |
TA | 0.156a | 0.189d | 0.137a | 0.118a | 0.109a |
RU | 0.156a | 0.191c | 0.135b | 0.116a | 0.108a |
Hillslope ( | 0.216 | 0.189 | 0.313 | 0.243 | 0.222 |
Position ( | <.001 | <.001 | <.001 | <.001 | <.001 |
| 0.360 | 0.414 | 0.161 | 0.095 | 0.078 |
Model ( | <.001 | <.001 | <.001 | <.001 | <.001 |
| 0.001 | <.001 | <.001 | 0.003 | 0.013 |
| <.001 | 0.010 | 0.429 | <.001 | <.001 |
| <.001 | 0.431 | 0.682 | 0.411 | 0.309 |
Means within a column and main effect, followed by different letters are significantly different at
The interactions for hillslope position by SWRC model are shown in Figure
Fitted SWRC for soil water contents
Summit
Backslope
Toeslope
The CA model underestimated water retention at a potential less than −0.5 MPa for the summit and backslope positions (Figures
The mean water content of the TA and RU models differed from that of the BC model across all water potential intervals for both upper and lower hillslope positions (Figure
Models of the SWRC with derivations of
Evaluation of analytical models to describe the SWRC of the shallow seedling recruitment zone indicated that the BC, VG4, and CA models with modified and estimated forms of
Weed seed germination is generally predicted by models using hydrothermal properties of the soil environment, of which water potential is a component. Because the best-fitting SWRC is needed for a dynamic model to predict weed germination within the range of minimum threshold water potentials for germination, water retention models that specifically deal with the SWRC at the dry end of the curve need further investigation. Future efforts could include analysis of the SWRC below −1.5 MPa to include species with very low minimum threshold water potentials for germination. As well, future research could evaluate the SWRC for a range of soil textures beyond that examined in the current study.
The evaluated
This paper was funded by the Manitoba Rural Adaptation Council, the Natural Sciences and Engineering Research Council of Canada, and the Canadian Wheat Board.