Hyperspectral imaging (also known as spectrum imaging) requires software for extracting the signatures present in every spectrum. However, commercial software available for spectrum analysis remains expensive, complicated, and often not transparent regarding the internal workings and approximations made. For user facilities, educational institutes, and other settings where multiple users on a single tool can be expected, the limited availability of software becomes the bottleneck to data analysis, user training, and throughput. The Cornell Spectrum Imager (CSI) was developed as a universal data analysis tool to be freely distributed, to run on all computers, and to minimize training. This is accomplished by using one simple interface for imaging, cathodoluminescence, Raman, Electron Energy Loss Spectroscopic (EELS), and EDX data analysis. This article demonstrates the CSI plugins for ImageJ by guiding you through the basic workflow for processing EELS maps.

The high beam current and subangstrom resolution of aberration-corrected scanning transmission electron microscopes has enabled electron energy loss spectroscopy (EELS) mapping with atomic resolution. These spectral maps are often dose limited and spatially oversampled, leading to low counts/channel and are thus highly sensitive to errors in background estimation. However, by taking advantage of redundancy in the dataset map, one can improve background estimation and increase chemical sensitivity. We consider two such approaches—linear combination of power laws and local background averaging—that reduce background error and improve signal extraction. Principal component analysis (PCA) can also be used to analyze spectrum images, but the poor peak-to-background ratio in EELS can lead to serious artifacts if raw EELS data are PCA filtered. We identify common artifacts and discuss alternative approaches. These algorithms are implemented within the Cornell Spectrum Imager, an open source software package for spectroscopic analysis.

There is much discussion of the probability of transgene flow from transgenic crop varieties to landraces and wild relatives in centers of origin or diversity, and its genetic, ecological, and social consequences. Without costly research on the variables determining gene flow, research on transgene frequencies in landrace (or wild relative) populations can be valuable for understanding transgene flow and its effects. Minimal research requirements include (1) understanding how farmer practices and seed systems affect landrace populations, (2) sampling to optimize Ne/n (effective /census population size), (3) minimizing variance at all levels sampled, and (4) using Ne to calculate binomial probabilities for transgene frequencies. A key case is maize in Mexico. Two peer-reviewed papers, based on landrace samples from the Sierra Juárez region of Oaxaca, Mexico, reached seemingly conflicting conclusions: transgenes are present (Quist and Chapela, 2001, Nature414: 541–543; 2002, Nature416: 602) or “detectable transgenes” are absent (Ortiz-García et al., 2005, Proc. Natl. Acad. Sci. USA102: 12338–12343 and 18242). We analyzed these papers using information on Oaxacan maize seed systems and estimates of Ne. We conclude that if Quist and Chapela’s results showing presence are accepted, Ortiz-García et al.’s conclusions of no evidence of transgenes at detectable levels or for their introgression into maize landraces in the Sierra de Juárez of Oaxaca are not scientifically justified. This is because their samples are not representative, and their statistical analysis is inconclusive due to using n instead of Ne. Using estimates of Ne based on Ortiz-García et al.’s n, we estimate that transgenes could be present in maize landraces in the Sierra Juárez region at frequencies of ~1–4%, and are more likely to be present in the 90% of Oaxacan landrace area that is not mountainous. Thus, we have no scientific evidence of maize transgene presence or absence in recent years in Mexico, Oaxaca State, or the Sierra Juárez region.