A New Method for Computing Reliable Sample Size for Prescribing Soil Test Based Nutrient Management Interventions

ABSTRACT

Conventional sampling schemes for soil test guided nutrient management do not duly consider spatial variability. Fisher’s least significant difference (LSD) classical technique is sometimes manipulated for computing minimum sample size. However, it does not consider spatial dependence and relies on sample variance. Here, we present a new LSD-based robust method that uses semivariogram sill as a variance surrogate and then explore through sensitivity analysis novel alternative measurement units to reduce sample size rendered large by spatial variability. For differentiating crop response based categories, 273–22,320 samples were required for primary nutrients. Required sample size for detecting desired critical shifts in micronutrient status varied from 16–28,854. Changing to millimole units for potassium (K) and iron (Fe) further reduced sample size significantly. Thus, LSD-based technique can be made robust by using geostatistical techniques. Conventional measuring units in highly variable plant nutrients can be replaced with more practicable and economical units.

KEYWORDS:

• Spatial variability sensitivity analysis
• crop response
• geostatistics
• South Asia
• soil testing
• soil sampling
• soil fertility

Acknowledgments

Punjab Agricultural University and Indian Council of Agricultural Research provided infrastructural support. Staff at Krishi Vigyan Kendra, Budh Singh Wala, Punjab, India provided logistic support.