Using spatial structure analysis of reflectance data to assess spider mite stress in maize plants

The basic concept of spectral-based analyses in agricultural applications is to associate reflectance values in narrow spectral bands with a given response variable, for instance, plant biomass or fertilizer requirements of a given plant. Regarding crop stress detection, the general assumption is that a biotic or abiotic stressor reduces the plant’s ability to utilize sunlight as efficiently as healthy plants. Consequently, sunlight reflection off a stressed plant is typically comparatively higher than from a healthy plant. Although spectral-based detection of crop stress is widespread, one important challenge is to separate effects of stressors, when crops are subjected to multiple stress levels (i.e. combinations of biotic and abiotic stressors). Another common challenge is low repeatability of spectral data across time and space, which hampers use of absolute reflectance values and indices as direct indicators of stress. An alternative analytical approach is to examine the spatial structure of spectral data. In this study, we examined spectral data from two maize hybrids subjected to one of three drought regimes (representing 0, 12, and 18% stress) and low (0-100 mites per plant) or medium (500-3,000 mites per plant) spider mite densities. Semi-variogram parameters (nugget, sill and range) were used as potential indicators of biotic stress, and we found that: 1) the sill increased in response to mite stress when drought stress was low, and 2) nugget values increased in response to drought stress but were consistently lower in mite infested maize plants compared to non-infested plants.