Selective pesticides and or selective rates can work over the short term, assuming your herbivores have greater tolerance/resistance. If your goal is to determine the potential impact of herbivores on crop development, you can also seed plots with different densities of herbivores and natural enemies. Both approaches were used in the following paper to create a wide range of pest and natural enemy densities, resultant pest population rates of increase, and damage impacts.

Wilson, L. T., P. J. Trichilo, and D. Gonzalez. 1991. Spider mite (Acari: Tetranychidae) infestation rate and initiation: Impact on cotton yield. Journal Economic Entomology 84: 593-600.

Bioquip sells larger cages, even up to 6 feet wide by 8 feet long by 6 feet tall. The screen is fine enough mesh to exclude most predators (maybe not fairy flies, small mites, or small thrips --- in part this will depend on the behavior of the insect). That said I have not had any problems in the 5 years that I have been using these cages to exclude all arthropods. They have held up well in the Florida sun, though it helps to apply a lubricant to the zippered door.

You might also explore BugDorm.com.http://bugdorm.megaview.com.tw/bugdorm-2-insect-tents-c-22.html. I have several of these cages and have been happy with the product. The screen is a bit finer mesh, but the cages are a bit smaller. While the cages are designed to be completely enclosed, you could remove the bottom and use it as a tent. Since these people make these cages, I suspect they may also take custom orders.

I hope that helps.

Apart from size-exclusion already pointed out by Timothy, I cannot see. But if your "real guys" fit the adequate size-distribution (i.e. size (predators) > sise (herbivores), then go for it.

Take a look at the following ref as an example of field level net-based size exclusion (and here lu!ckily the herbivore is small :-)

Han P, Niu CY, Desneux N. PLoS One. 2014 Aug 29;9(8):e102980. doi: 10.1371/journal.pone.0102980. eCollection 2014. Identification of top-down forces regulating cotton aphid population growth in transgenic Bt cotton in central China.

Yvan

Selective pesticides and or selective rates can work over the short term, assuming your herbivores have greater tolerance/resistance. If your goal is to determine the potential impact of herbivores on crop development, you can also seed plots with different densities of herbivores and natural enemies. Both approaches were used in the following paper to create a wide range of pest and natural enemy densities, resultant pest population rates of increase, and damage impacts.

Wilson, L. T., P. J. Trichilo, and D. Gonzalez. 1991. Spider mite (Acari: Tetranychidae) infestation rate and initiation: Impact on cotton yield. Journal Economic Entomology 84: 593-600.

It really depends on the predator. I used exclusion bands around the base of trees to keep ants out of the canopy.

Pesticides might work, but there is a great risk that there will be sublethal effects on herbivore behavior and reproduction. The pesticide will also have greater effect on early instars, so you might kill off a whole cohort.

Depends on the herbivore and predator species, the cropping system, and the particular pesticides. For a poorly studied system, this would certainly be a concern. 

Also depends on what you are measuring. I assume that the contrast of interest is a treatment with herbivores but not predators versus a treatment with both. The problem is that one can argue all sorts of outcomes.

1) Sublethal treatment reduces competition and makes more food available to later instars. These mature sooner than they would otherwise and thereby reproduce sooner. There is also less crowding. The sublethal effect increases herbivore populations and therefore there is a greater difference between "herbivore+predator" and "herbivore alone" treatments than one would expect had the insecticide treatment not influenced the outcome.

2) Sublethal treatment reduces the apparent effect of the treatment. The larvae don't eat as much and pupate sooner due to long term sublethal effects. The smaller adults don't produce as many eggs, and the populations don't achieve the levels they would have if the herbivore populations had developed unaltered. The end result is that there is less difference between the treatments.

Of course you could apply similar logic to the idea of cages. The cage blocks some sunlight, so the plants don't grow as quickly. Cages change the humidity and temperature. So the herbivores may have more impact on plant development than if there was full sunlight and the plant could grow faster to compensate for herbivory. There may also be disease problems that affect the plant or the herbivore. The cage will also prevent colonization by herbivores, so you will either have to introduce the herbivores to the cage, or hope that the plants in the cage have already been colonized. Finally, the cage may be relatively ineffective against predators like ground beetles that can burrow or move through cracks in the soil.

The bottom line is that you can't win. You must make a decision and know that your methodology will influence your results. You will not be able to define how your methods influenced your results. If that is a problem, then you could have several predator exclusion techniques and compare them. Of course this is now a huge experiment.

I guess it also depends on the definition of "all herbivores". Some caterpillars are herbivores in some conditions but carnivores under other conditions. They may be opportunistic or indiscriminate. I'll eat this leaf, and if there are a few aphids present so much the better. The herbivores will interact, so that will increase variability.

There is no assurance that this is even a cropping system we are discussing. We don't know scale: is this a dandelion or an oak tree, a whole plant or several square meters, days or years? We don't know the end point: harvest, total biomass, number of herbivores, or herbivore biomass. I am guessing outdoors, but it could be a greenhouse study. Are the plants being studied or are the insects the focus?

If you have plants with a well defined trunk, and the primary predators is ants, then Paulson's answer is great unless the ants are keeping other predators/parasitoids/herbivores away. It also assumes that your herbivores don't migrate up and down the trunk every evening.

The correct answer depends on your research question, the nature of the system you are studying, and the problems that you are willing to accept given your constraints of time and funding.

Good response! I've tended to minimize the use of greenhouse and field cages due to the impact each has on the light/temperature/vapor pressure environment. An exception is when I have conducted host preference studies where I prefer to only confine the plants in cages for a few days to minimize cage effects.

As you correctly convey, pesticides can also impact the host. Depending on the herbivore species, resource competition also may or may not be important. For example, spider mites can reach very high densities, while most noctuid only rarely reach levels that result in severe resource competition. The strengths and limitations of each experimental method is very much system and species dependent.