It depends on the plant and herbivore species, and more importantly on the mode of feeding by the herbivore, and by extension, the type of damage normally caused. Generally, biomass of plant material eaten would be the simplest (for foliage feeders), but other parameters could include level of penetration (for insect borers), extent of wilting (for sap suckers), extent of plant recovery after herbivore attack (with yield data, if applicable). Comparing profiles of volatile and non-volatile metabolites of stressed plants with those of non-stressed plants is another parameter used in the case of insect and other invertebrate herbivores, if information is available on the biochemical cues used by the herbivore to recognise and find its host plant under normal conditions.

Depends on the feeding guild of the herbivore. But its important that you analyse the plant tissue to get the chemical profiles of plants in different treatments. Depending on the capacity of your laboratory, do as much profiling as possible, if capacity is limiting, at least get the Tissue N, P, K. Take feeding symptom severity, population dynamics of the herbivores, and their biology. Herbivore biology is very important in host plant resistance studies. So capture information on the length of the periods between life stages/instars, weight of immatures and adults..