Occurrence and short-distance migration of Nilaparvata lugens (Hemiptera: Delphacidae) in the Vietnamese Mekong Delta

Abstract

In migration analysis of the brown planthopper, Nilaparvata lugens (Stål), in a tropical region, it is difficult to know the arrival timing of the immigrants because rice plants in different growth stages are simultaneously cultivated, and local insects may already occur in destination areas. Due to this analytical difficulty, additional information is necessary to discriminate between immigrants and locally originated individuals in the light trap catch. This study applied rice phenology maps retrieved from satellite images to analysis of N. lugens’s migration in the Vietnamese Mekong Delta. The analysis also employed occurrence data by dense light traps and trajectory analysis. The result showed that the occurrence of N. lugens peaked periodically at an interval of 27.2–30 days, changing in association with the rice cropping area in the harvesting stage. For an N. lugens occurrence in late July 2009 in the delta, the rice phenology suggested catch increases in light traps in a destination area could likely be attributed to immigrants from outside the area. Therefore, the migration analysis suggested that N. lugens that emigrated from paddy fields in An Giang and Kien Giang provinces possibly moved to east-neighboring provinces. The migration distance was estimated to be about 100 km at a maximum.

Appendix

A crop phenology detection method based on time-series MODIS data used in this study consisted of three steps (method WFCP in Sakamoto et al. 2005): (1) the prescription of multi-temporal MODIS/Terra data; (2) filtering of time-series Enhanced Vegetation Index (EVI) data by time–frequency analysis to reduce short-term noise components mostly caused by cloud coverage; (3) specifying the rice heading stages by detecting the local maximum point from the smoothed EVI time profile. The computation algorithm used in this study was compliant with a wavelet-based filter for evaluating the spatial distribution of the cropping system (WFCS method in Sakamoto et al. 2006), which was an improved version of the WFCP for revealing the spatiotemporal distribution of rice phenology and cropping systems in the delta. Sakamoto et al. (2009) provided a brief overview of the wavelet-based algorithms and new findings regarding yearly changes in land use and flood-affected areas.

This study used an 8-day time series of 250- and 500-m MODIS surface reflectance data (MOD09A1 and MOD09Q1, Collection 5, tile: h28v07 and h28v08) obtained by the Terra satellite (EOS AM). The data were freely distributed through an FTP server of the Land Processes Distributed Active Archive Center (LP DAAC), US Geological Survey (LP DAAC 2013). The 8-day composite MODIS surface reflectance (MOD09 product) contains the best surface spectral reflectance data for each 8-day period on the basis of the CM-MVC technique (Huete et al. 2002), and these data were also systematically corrected for the atmospheric effects of gaseous absorption and aerosol scattering (Vermote and Vermeulen 1999).

The EVI has a higher sensitivity to the leaf area index (LAI) than does the NDVI, especially in high biomass areas (Huete et al. 2002). The EVI was calculated from three optical bands by the following equation:

$$ {\text{EVI }} = \, 2.5 \, (\rho_{\text{NIR}} {-} \, \rho_{\text{red)}} )/{ (}\rho_{\text{NIR}} + \, 6 \, \rho_{\text{red}} {-} \, 7.5 \, \rho_{\text{blue}} + \, 1 ) $$

where ρ _red and ρ _NIR are the red (band 1, 621–670 nm) and near infrared (band 2, 841–875 nm) reflectance of the MOD09Q1 product, respectively. ρ _blue is the blue reflectance (band 3, 459–479 nm) of the MOD09A1 product, which was resampled to 250-m resolution using the nearest neighbor method. A map of heading dates for every 3 months (Fig. 2), whose pixel value is the EVI peak value, was made for the cropping years from 2007 to 2009 from multiple resampled images.