The abundance effect on network nestedness is stronger for parasitic than herbivory interactions

It has been suggested that the importance of network architecture to species diversity and stability should be based on preference networks(comprised of niche differentiations),rather than observational networks,because species abundance may significantly affect interaction frequencies.Considering that resource abundance is usually greater for herbivores than parasites,we hypothesize that the abundance effect is stronger for parasitic than herbivory interactions.To test this hypothesis,we collected 80 quantitative observational networks including 34 herbivorous and 46 parasitic networks from the published literature,and derived preference networks by removing the effects of species abundance.We then determined the network nestedness using both weighted NODF and spectral radius.We also determined species degree distribution,interaction evenness,weighted connectance and robustness for both observational and preference networks.The observational networks(including both herbivory and parasitic networks)were more nested judged by weighted NODF than spectral radius.Preference networks were less nested for parasitic than herbivory networks in terms of both weighted NODF and spectral radius,possibly because removing the abundance effect increased interaction evenness.These trends indicate that the abundance effect on network nestedness is stronger for parasitic than herbivory networks.Weighted connectance and robustness were greater in most preference networks than observational networks,indicating that preference networks may have high network stability and community persistence compared with observational ones.The data indicate that future network analyses should not only address the structural difference between mutualistic and antagonistic interactions,but also between herbivory and parasitic interactions.

A Novel De-Embedding Technique of Packaged GaN Transistors

This paper presents a novel de-embedding technique of packaged high-powertransistors. With the proposed technique, the packaged model of the power amplifier (PA)tube can be divided into the frequency independent de-embedded intrinsic device (DID)and the frequency dependent internal parasitic network (IPN), which is of great help in reducingthe design complexity of a broadband PA. Different from the conventional techniqueof parasitic extraction, the proposed technique only requires external measurements.The frequency independent characteristic of DID is verified and the IPN is modeledand calibrated for a 50 W gallium-nitride (GaN) transistor. At last, a broadbandDoherty PA is fabricated with the de-embedding technique. According to the measured results,the PA exhibits satisfactory power and efficiency performance.