Table 1: Linear Mixed-Effects Model Summary for Soil Water Storage (SWS)

1. Linear mixed-effects model fit by REML

AIC 1422.68

BIC 1451.88

logLik -701.3402

2. Random effects:

Formula: ∼ 1 ∣ Replications
0.006221491
(Intercept) 0.006221
54.35841
Residual 54.35841

3. Correlation Structure: AR(1)

Formula: ~ TimeIndex | Replications Parameter (Phi) lications 0.757596

4. Fixed effects: SWS ∼ Treatment + Precipitation + Temperature + Year

Predictor Value Std. Error DF t-value p-value

120.5172 10.4588
Intercept 1260.4683 135 ¿ 0.0001
7 2

Treatment2 43.8520 5.83551 135 7.51468 ¿ 0.0001

Treatment3 63.1195 6.83426 135 9.23575 ¿ 0.0001

11.6025
Treatment4 72.4312 6.24269 135 ¿ 0.0001
6

Precipitatio
-0.1170 0.08387 135 -1.39481 0.1654
n

Temperature -47.0302 10.91637 135 -4.30823 0.0000

Year -8.1375 3.21599 135 -2.53032 0.0125

Trtmnt Trtmnt
(Intr.) Trtmnt3 Precip Temp
2 4

Treatment2 -0.059

Treatment3 -0.092 0.585

Treatment4 -0.139 0.361 0.605

Precipitation -0.761 0.007 0.014 0.028

Temperature -0.941 0.023 0.042 0.077 0.566

Year -0.504 0.101 0.170 0.284 0.396 0.292

Note. Model was fit by REML with a random intercept for Replications and an AR(1)
correlation structure based on TimeIndex.
Table 2: Linear Mixed-Effects Model Summary for Soil Desiccation (SD)

1. Linear mixed-effects model fit by REML

AIC 986.0763

BIC 1015.276

logLik -483.0382

2. Random effects:

Formula: ∼ 1 ∣ Replications

(Intercept) 0.001139812

Residual 10.43577

3. Correlation Structure: AR(1)

Formula: ∼ TimeIndex ∣ Replications

4. Fixed effects: SD ∼ Treatment + Precipitation + Temperature + Year

Predictor Value Std. Error DF t-value p-value

Intercept 136.54075 24.440758 135 5.586601 < 0.0001

Treatment2 8.79487 1.204705 135 7.300437 < 0.0001

Treatment3 11.28886 1.405151 135 8.033916 < 0.0001

Treatment4 13.18584 1.275245 135 10.339854 < 0.0001

Precipitation -0.01601 0.016965 135 -0.943815 0.3470

Year -1.05033 0.601710 135 -1.745567 0.0832

5. Correlation:

Trtmnt Trtmnt
(Intr.) Trtmnt3 Precip Temp
2 4

Treatment2 -0.056

Treatment3 -0.086 0.582

Treatment4 -0.130 0.358 0.599

Precipitation -0.770 0.007 0.015 0.030

Temperature -0.946 0.022 0.042 0.078 0.580

Year -0.532 0.086 0.146 0.246 0.436 0.336

Note. Model was fit by REML with a random intercept for Replications and an AR(1)
correlation structure based on TimeIndex.
Table 3: Linear Mixed-Effects Model Summary for Water Consumption (WC)

1. Linear mixed-effects model fit by REML

AIC 1587.979

BIC 1617.179

logLik -783.9894

2. Random effects:

Formula: ∼ 1 ∣ Replications

Residual 70.36616

3. Correlation Structure: AR(1)

Formula: ~ TimeIndex | Replications

4. Fixed effects: WC ~ Treatment + Precipitation + Temperature + Year

Predictor Value Std. Error DF t-value p-value

Intercept 123.65512 198.19851 135 0.623895 0.5337

Treatment2 -68.43743 12.64373 135 -5.412757 < 0.0001

Treatment3 -69.66944 14.22003 135 -4.899386 < 0.0001

Treatment4 -76.62666 12.89257 135 -5.943473 < 0.0001

Precipitation 0.16076 0.13186 135 1.219209 0.2249

Temperature 27.59966 17.41874 135 1.584481 0.1154

5. Correlation:

Trtmnt Trtmnt
(Intr.) Trtmnt3 Precip Temp
2 4

Treatment2 -0.042

Treatment3 -0.055 0.561

Treatment4 -0.079 0.380 0.569

Precipitation -0.840 0.003 0.009 0.024

Temperature -0.968 0.009 0.019 0.044 0.700

Year -0.727 0.023 0.038 0.074 0.665 0.611

Note. Model was fit by REML with a random intercept for Replications and an AR(1)
correlation structure based on TimeIndex.
Table 4: Linear Mixed-Effects Model Summary for Fodder Yield (FY)

5. Correlation:

Trtmnt Trtmnt
(Intr.) Trtmnt3 Precip Temp
2 4

Treatment2 -0.057
 Treatment3 -0.089 0.584

Treatment4 -0.135 0.359 0.602

Precipitation -0.766 0.007 0.014 0.029

Temperature -0.944 0.022 0.042 0.078 0.573

Year -0.518 0.093 0.157 0.264 0.417 0.314

Note. Model was fit by REML with a random intercept for Replications and an AR(1)
correlation structure based on TimeIndex.

Table 5: Linear Mixed-Effects Model Summary for Water Use Efficiency (WUE)

1. Linear mixed-effects model fit by REML

AIC 1052.679

BIC 1081.879

logLik -516.3394

2. Random effects:

Formula: ∼ 1 ∣ Replications

(Intercept) 0.001117985

Residual 12.76811

3. Correlation Structure: AR(1)

Formula: ~ TimeIndex | Replications Parameter (Phi) 0.6941402

4. Fixed effects: WUE ~ Treatment + Precipitation + Temperature + Year

Predictor Value Std. Error DF t-value p-value

Intercept -86.47517 31.073715 135 -2.782904 0.0062

Treatment2 8.37886 1.556951 135 5.381581 < 0.0001

Treatment4 13.88351 1.637017 135 8.480983 < 0.0001

Precipitation 0.07855 0.021506 135 3.652552 0.0004

Temperature 7.52807 2.799632 135 2.688951 0.0081

Year 2.16202 0.722268 135 2.993379 0.0033

5. Correlation:

Trtmnt Trtmnt
(Intr.) Trtmnt3 Precip Temp
2 4

Treatment2 -0.053

Treatment3 -0.082 0.581

Treatment4 -0.124 0.358 0.596

Precipitation -0.778 0.007 0.015 0.031

Temperature -0.950 0.021 0.041 0.077 0.591

Year -0.556 0.076 0.129 0.220 0.467 0.371

Note. Model was fit by REML with a random intercept for Replications and an AR(1)
correlation structure based on TimeIndex.

Table 6: Combined ANOVA Summary for All Dependent Variables

Effect SWS SD WC FY WUE

F ¿-value ¿ F ¿ -value ¿ F ¿ -value ¿ F ¿ -value ¿ F ¿ -val)

Intercept 5169.44 (¿ 0.0001) 31.23(¿ 0.0001) 0.39( 0.5337) 28.87(¿ 0.0001) 7.74 ¿

Treatment 61.709(¿ 0.0001) 74.28(¿ 0.0001) 29.94 (¿ 0.0001) 24.92(¿ 0.0001) 46.23 ¿ .
Precipitatio
2.998(0.0856) 0.891(0.3470) 1.49(0.2249) 38.94 (¿ 0.0001) 13.34 ¿
n

Temperatur
13.929(0.0003) 13.49(0.0003) 2.51(0.1154 ) 42.97 (¿ 0.0001) 7.22 ¿
e

Year 6.403(0.0125) 3.04 (0.0832) 5.16(0.0247) 9.06 (0.0031) 8.96 ¿

Note. For each effect, the F-values and p-values were computed from the corresponding
linear mixedmodels for each dependent variable. SWS ¿ Soil Water Storage; SD=¿ Soil
Desiccation; WC=¿ Consumption; FY = Fodder Yield; WUE = Water Use Efficiency.

Interpretation of all results:

Soil Water Storage (SWS):

The mixed–effects model for soil water storage (SWS) demonstrated an acceptable fit (AIC =
1422.68, BIC = 1451.88, logLik = –701.34). A minimal random intercept variance (0.00622)
relative to a residual standard deviation of 54.36 mm, along with an AR(1) parameter of 0.758,
confirmed that temporal autocorrelation was appropriately modeled. Fixed effects indicated that
the baseline SWS (intercept = 1260.47) was highly significant (t = 10.46, p < .0001). In
comparison with the reference biochar application (Flat Planting), Treatment2, Treatment3, and
Treatment4 were associated with significant increases in SWS of 43.85 mm (t = 7.51, p < .0001),
63.12 mm (t = 9.24, p < .0001), and 72.43 mm (t = 11.60, p < .0001), respectively. Although
precipitation was not a significant predictor (β = –0.1170, t = –1.39, p = 0.1654), higher
temperatures significantly reduced SWS (β = –47.03, t = –4.31, p < .0001), and later years were
associated with a modest decline (β = –8.14, t = –2.53, p = 0.0125). These results underscore that
biochar application robustly enhances soil water storage despite adverse effects of increasing
temperature and time.

Soil Desiccation (SD):

For soil desiccation (SD), the model fit was strong (AIC = 986.08, BIC = 1015.28, logLik = –
483.04), with a low random intercept variance (0.00114) and a residual standard deviation of
10.44. The AR(1) parameter (Φ = 0.723) indicated that repeated measurements were adequately
correlated over time. The intercept for SD was 136.54 t= 5.59, p < .0001). All biochar treatments
significantly increased soil desiccation relative to the reference, with Treatment2 (β = 8.79, t =
7.30, p < .0001), Treatment3 (β = 11.29, t = 8.03, p < .0001), and Treatment4 (β = 13.19, t =
10.34, p < .0001) showing robust effects. Precipitation (β = –0.0160, t = –0.94, p = 0.3470) did
not significantly affect SD, while temperature significantly reduced it (β = –8.11, t = –3.67, p =
0.0003). The effect of Year was marginal (β = –1.05, t = –1.75, p = 0.0832). Thus, while biochar
consistently increased soil desiccation, the roles of temperature and time were less pronounced
for this variable.
Water Consumption (WC):

The mixed–effects model for water consumption (WC) indicated moderate model fit (AIC =
1587.98, BIC = 1617.18, logLik = –783.99) and a lower AR(1) parameter (Φ = 0.396), suggesting
less temporal autocorrelation. The intercept of 123.66 was not statistically significant (t(135) =
0.62, p = 0.5337). However, all biochar treatments significantly reduced water consumption
relative to the reference: Treatment2 decreased WC by 68.44 mm (t = –5.41, p < .0001),
Treatment3 by 69.67 mm (t = –4.90, p < .0001), and Treatment4 by 76.63 mm (t = –5.94, p
< .0001). Although precipitation (β = 0.1608, t = 1.22, p = 0.2249) and temperature (β = 27.60, t
= 1.58, p = 0.1154) did not reach significance, Year was a significant negative predictor (β = –
7.83, t = –2.27, p = 0.0247). These findings support the conclusion that biochar application
effectively reduces water consumption, reinforcing its potential role in improving water-use
efficiency.

Fodder Yield (FY):

The model for fodder yield (FY) exhibited an acceptable fit (AIC = 2401.32, BIC = 2430.52,
logLik = –1190.66) despite high residual variability (residual SD = 1877.14; random intercept SD
= 0.288). With an AR(1) parameter of 0.740, the model appropriately addressed repeated
measures. The intercept was significantly negative (–23020.95, t(135) = –5.37, p < .0001).
Notably, while Treatment2 was not significantly different from the reference (β = 27.33, t = 0.13,
p = 0.8963), both Treatment3 (β = 1021.52, t = 4.18, p = 0.0001) and Treatment4 (β = 1684.45, t
= 7.57, p < .0001) significantly increased fodder yield. In addition, precipitation (β = 18.59, t =
6.24, p < .0001), temperature (β = 2539.81, t = 6.55, p < .0001), and Year (β = 329.88, t = 3.01, p
= 0.0031) were significant positive predictors. These results indicate that higher rates of biochar
application, together with favorable weather conditions and advancing stand age, substantially
enhance fodder yield.

Water Use Efficiency (WUE):

For water use efficiency (WUE), the model yielded favorable fit statistics (AIC = 1052.68, BIC =
1081.88, logLik = –516.34) and an AR(1) parameter of 0.694, indicating that the correlation of
repeated measures was well-captured. The intercept was significantly negative (–86.48, t(135) = –
2.78, p = 0.0062). All biochar treatments significantly improved WUE, with increases of 8.38
(Treatment2; t = 5.38, p < .0001), 11.13 (Treatment3; t = 6.15, p < .0001), and 13.88
(Treatment4; t = 8.48, p < .0001) units compared to the reference. In addition, precipitation (β =
0.0786, t = 3.65, p = 0.0004), temperature (β = 7.53, t = 2.69, p = 0.0081), and Year (β = 2.16, t =
2.99, p = 0.0033) were all significant positive predictors of WUE. These outcomes imply that
biochar not only enhances soil water storage and reduces water consumption but also improves
the efficiency with which water is utilized by the crop.

Combined ANOVA Summary:

A combined ANOVA table across all dependent variables reveals that, for soil water storage
(SWS), the intercept effect was extremely robust (F(1,135) = 5169.44, p < .0001) and treatment
effects were also highly significant (F(3,135) = 61.71, p < .0001), while precipitation approached
significance (F(1,135) = 2.998, p = 0.0856) and both temperature (F(1,135) = 13.93, p = 0.0003)
and Year (F(1,135) = 6.40, p = 0.0125) significantly predicted SWS. Similar patterns were
observed for soil desiccation (SD), water consumption (WC), fodder yield (FY), and water use
efficiency (WUE), where treatment effects were consistently significant (all p < .0001) and
weather variables as well as Year showed varying degrees of significance. These ANOVA results
corroborate the mixed–effects model findings by demonstrating that biochar application—
especially at higher levels—is a key driver of improved soil moisture conditions and enhanced
crop performance, while also reflecting the significant impacts of climatic factors and temporal
progression.