Study on Hyperspectral Reflectance of Brown Planthopper (Nilaparvata lugens) Infestation in Rice

Rice (Oryza sativa) is a major cereal crop with a rich source of carbohydrates and starch. Different species of rice occupy different parts of the world (Latif et al., 2011). Insects pose a colossal threat to cereal crops and significantly reduce crop production and productivity (Jena et al., 2006). Among various insect pests, the BPH (Nilaparvata lugens) is currently thought to be Asia’s most destructive rice pest. The hotspots of BPH were prone to higher relative humidity of >80% and maximum temperature of >30°C at peak hours and heavy application of nitrogen fertilizer (Wang et al., 2008). The BPH affected plants showed diminished chlorophyll and photosynthetic rate and Hopper burn symptoms such as yellowing and wilting of plants (Maisarah and Habibuddin, 2018). High usage of chemicals to control BPH leads to the developing resistance BPH biotypes and expanding environmental hazards over a large area (Zhang et al., 2016). Improved crop monitoring techniques, such as hyper and multispectral sensors, are being developed and promoted to increase the sustainability of pest management in modern agriculture Iost (href="#iost_2020">Filho et al., 2020). From the perspective of pest management, it is crucial to quantify early symptoms. Biophysical remote sensing is used to identify plant stress brought on by disease or insect activity remotely (Jensen, 1983). The effective use of spectral reflectance to detect of insect infestation depends on determining the specific wavelength highly correlated to a specific infestation. The spectroradiometers use various ranges of light to identify chlorophyll, anthocyanin and carotenoids, and the water content in leaves (Mahlein et al., 2013). Using hyper spectral remote sensing, the better forewarning system helps solving farmers pest problems, makes farming more profitable and efficient (Saraf et al., 2019). For managing and monitoring natural resources, such as crops, identifying changes in vegetation patterns are crucial. The benefits of vegetation indices in agriculture are the availability of high level analysis over a vast region, and access to different image sources in one place. Some unique vegetation indices are helps to predict crop yield detection, plant vigour and stress (Panda et al., 2010). The vegetation index uses linear or nonlinear combinations of reflectance in different spectral bands to produce associated spectral signals (Lei et al., 2021). The need of the study is to utilize hyperspectral sensor for early detection of BPH infestation so as to reduce the cost of cultivation in rice growing areas.  

The Infestation of N. lugens was heavily damaged in Rice along farmers field at Palayagudalur village, Kuttalam Block, Nagapattinam district Tamil Nadu (11.032932°N and 79.549255°E) and it is measured with the instrument-Spectroradiometer to know spectral reflectance of the crop. The growing season was Rabi (2022) and the age of seedlings was 120 days old transplanted seedlings which is the long duration variety (ADT 51). Alluvial clay was the dominant soil type over the block. The fertiliser used was 150:60:60 NPK/ha. Irrigation was done on alternate days. The average fortnight maximum and minimum temperature occurred in the study area was 30°C and 20°C with a relative humidity of  more than 93%. During observations, we found only BPH damage at the field. Because BPH damage was more likely to occur in that particular block during rabi season by every year than any other pest damage. We observed the appearance of BPH Symptoms and the number of BPH as <25 BPH/hill for Class 1 symptoms, 25-50 BPH/hill for Class 3 symptoms, 50-100 BPH/hill for Class 5 symptoms and >100 BPH for Class 7 symptoms and Class 8 symptoms were not discussed in this paper since there is no need of forewarning. Moreover in our study area, the complete wilting and dead plants not available before that pesticides application was done to control over it. We followed the standard evaluation for the BPH damage given by INGER (IRRI, Cuttack) (Table 1).
 

       
This BPH damage was captured by Spectroradiometer GER 1500, (S/N 1500-002128, Year-2009) a portable device covering UV, Visible and Infra-red region from 350 nm to 1050 nm. The observation of spectral reflectance was taken between 1200 and 1400 hours during the daytime. The device should be kept 1 m above the height of the plant, and the sighting laser should have a 4° field of view and be oriented closer to the plant leaves. Initially, a white reference board was used for reference prior to data collection to calibrate for the lighting conditions. After the captured reflectance, Certain vegetation indices were evaluated to know sensitive spectral bands to give better forewarning (Table 2).
 

 
Statistical analysis
 
A general linear model was generated with univariate ANOVA with Classes as the main effects at various wavebands. A post hoc analysis with Duncan’s homogenous test was employed to identify similar mean reflectance values.  Superscripted alphabets that are different are statistically significantly different (Table 3).

Spatial signature of N. lugens
 
In the present study, we observed a significant variance between the spectral reflectance of BPH infected rice plants and the symptoms of different classes (p<0.001). Yang and Cheng (2001) evaluated the spectral properties of rice plants at various levels of N. lugens infestation and he stated that across infestations, there were noticeable changes in reflectance at 755 and 890 nm wavelengths. In the current study also, the highest significant difference of reflectance was observed in the NIR (740 nm-925 nm) and Red region (640 nm-739 nm) than visible and UV region (Table 3) (Fig 1). This is due to high absorption of radiation by visible region and maximum reflection and transmission in the NIR region Lei et al., (2021). The wavelength range (640-740 nm) in the red region is regarded as a reflection sign of insect damage Liu and Sun (2016). Here, the spectral signature of BPH damaged plants showed a decreased trend with slight increase in the visible region (350 nm-639 nm) and a decrease with a steep rise in the red region (640 nm-739 nm). However, it follows steady and stable up to the NIR region (740 nm-925 nm) (Fig 1). Our Study results were highly correlated with the BPH study of Prasannakumar et al., (2013) and he reported that BPH damage was greater at longer wavelengths (740 nm-925 nm) and lower at shorter wavelengths (350 nm-730 nm). The same trend followed in our study; the healthy plants showed higher reflectance radiation in NIR region than the BPH-infected plants. The Previous study of Lu et al., (2022) also revealed similar spectral signature for leaf mite infestation in the crop jujube. A similar trend was pursued in the earlier study of Riedell and Blackmer (1999) observed in wheat seedlings with aphids (Diuraphis noxia) and greenbugs (Schizaphis graminum). Prabhakar et al., (2013) also revealed the minimal reflectance was observed in BPH infected plants in Rice and Mealy bug infected plants in cotton. He stated that the reduced reflectance in the advanced stage of infestation (class 3 and above) is due to the reduction in leaf pigments and damages in internal leaf structure as a result of loss of xylem and phloem tissue.
 

 
Spatial indices of N.lugens
 
As the result of this study, we estimated five vegetation indices with respect to different classes of BPH infestation for forewarning. The selection of these indices were made, because they significantly affect the amount of leaf chlorophyll in the plant canopy. BPH sucks plant sap, which results in decreased chlorophyll content in the plants. In the indices, NDVI (R²=0.62, P<0.001, F<0.05), CIred edge (R²=0.76, P<0.001, F<0.05), GNDVI (R²=0.15, P<0.001, F<0.10) follows slight variation upto class 3 symptoms then rapidly decrease with rise upto class 7 symptoms (Fig 2). Huang et al., (2015) manifested that the estimation of BPH infestation through the indices NDVI, CIred edge, GNDVI was unsatisfactory due to poor detection at early infestation. Hence these indices are very difficult to predict and interpret over large area. The index NPCI (R²=0.75, P<0.1, F<0.05), it showed an inverse relationship with respect to class 0 to 7 symptoms which is very controversial and complex to interpret. A Similar outcome was suggested by Prabhakar et al., (2013) in BPH infestation in Rice and he stated that the values of NPCI were meandering with low to high infestation where it doesn’t follow any trend. However, in our present study, the indices MCARI (R²=0.93, P<0.001, F<0.05) revealed high significance in all statistical analyses and it proved to be worth prediction in all classes of BPH infestation. Hence it decreased linearly from lower to higher infestation (Fig 2). Luo et al., (2013) suggested in the same way that MCARI was the best indices to predict Aphid (Sitobion avenae) infestation in Wheat under field conditions based on the high correlation and significant difference of p<0.0001. He also stated that among various indices such as NPCI, PRI, NDWI, NBNDVI and PSRI, the indices MCARI was performed better.  It is proved by the results of Lu et al., (2022) an leaf mite (Tetranychus truncatus Ehara) infestation in the tree jujube. He also suggested that the indices MCARI can be used to predict from mild infestation to severe infestation. The results of our study were confirmed by the previous studies of Huang et al., (2015) in rice BPH infestation using the indices MCARI and he reported that MCARI spectral index was good to predict the BPH damage at slight infestation level across different rates of nitrogenous fertilizer in the field conditions.
 

In future climate change scenarios, due to adverse weather conditions pest outbreaks and survival will become more. In an effort to control, farmers use enormous pesticides which could lead to loss of biodiversity and environmental risk. In this situation better forewarning only helps to avoid the detrimental impact of pests and, economic loss to the farmers and stakeholders. In the upcoming days, the decision-making policymakers should utilize hyperspectral remote sensing, vegetation indices, satellite data and crewless aerial vehicle for the accurate forewarning of the early infestation of any pest and to minimize the production expenditure to the farmers over a large scale. As a result, the above mentioned practice was utilised to identify BPH infestation in rice by large-scale mapping, and it is very beneficial for small-scale farmers for pest management.

We thank Tamil Nadu Rice Research Institute (TRRI), Aduthurai for providing information about BPH occurred area around Thanjavur and Nagapattinam district of Tamil Nadu. This information helped us reach the afflicted area.

None.