Field ecology experiments are fickle. Even with best laid plans in place, they can fail…Nature doesn’t follow sampling protocols.
When this happens, should you publish the results? Most people would say no, and I would generally agree. Failed experiments are different to negative results. The latter are important additions to the scientific literature, but the former have very limited use. The results of failed experiments will have limited value, depending on why the experiment failed and how many data points were left intact. But they can have some use as ‘what not to do’ baselines for other researchers.
I recently wrote up a failed experiment and submitted it as a ‘Short Communication’ to a journal that published relevant topics. It was rejected, unsurprisingly. As the reviewers very astutely recognised, I had written “what amounts to a failed experiment”.
One of the recurring comments from reviewers was that instead of trying to publish these results, I should design a better study, collect another year of data, and then publish it. Yes, that’s the ideal way to deal with a failed experiment and it works for established scientists who have access to the same system every year. But early career researchers on fixed-term contracts with limited operating funds don’t always have that luxury.
My failed experiment happened in the last year of my previous postdoc. I’ve since moved to a new region to start a new project, so I probably won’t have the chance to run the experiment again any time soon. And I definitely won’t be able to run the experiment again in the same system.
In this case, I thought that publishing the failed experience for someone else to build on was better than letting it gather dust in the file drawer. I assumed that ‘Short Communications’ categories were the ideal forum for failed experiments, pilot studies and negative results. But the very few journals that publish short communications seem to prefer complete experiments (as a contribution to the literature, I’m not sure how this is different to standard research papers).
Lucky for blogs! Academic blogs are a great forum for sharing and discussing unpublished results or failed experiments.
In my previous postdoc I worked on insect-related ecosystem services in apple orchards in south-eastern Australia. During field work, I noticed that one of our organic orchards had a high number of European, or common, earwigs (Forficula auricularia) on apple trees, compared to the other orchards. Most people in Australia think of common earwigs as pests, and they are notorious for damaging vegetables and plants in home gardens. However, studies from the northern hemisphere have shown they can also be great biocontrol agents. In Australia, earwigs are known to be effective biocontrol agents in apple orchards, but there is limited knowledge of their effects in other systems.
This orchard with lots of earwigs was interesting to me because it grew multiple fruit varieties. The owners told us that the earwigs were annoying pests on their stonefruit trees, where they damaged a lot of fruit, but were ‘goodies’ on their apple trees, where they controlled pests like woolly aphid. Because all of our other orchards mostly grew pome fruit (apples and pears), I thought this would be a great opportunity to run a little pilot study testing the tradeoffs between the ‘pest’ and ‘beneficial’ activity of the earwigs in a single system. Doing this in a single system would limit the confounding effects of management across multiple orchards.
Cylinders of rolled corrugated cardboard are a standard method of trapping earwigs. Earwigs seek shelter during the day, and the dark cosiness of cardboard corrugations are the perfect hiding spot. A lot of published studies have used this method to measure the dynamics of earwig populations in orchards. However, most of these have only trapped for a short period of time, or have regularly disturbed the traps by opening and counting earwigs inside. Also, many of these studies set traps out in spring, after earwigs have already made nests and laid eggs.
I was interested to know how this trap method could work as a way of enhancing biological control in orchards, as well as how the costs and benefits of earwig activity trade off against each other within a single orchard.
I made up 40 earwig traps and set them up across the orchard in 10 trees each of four fruit varieties: apple, pear, apricot and plum. I used cable ties to attach the traps securely to a fork near the bottom of the trunk of each tree. I set the traps out in May, just before winter, to give the earwigs time to locate the traps and set up camp ready for spring flowering and fruiting. Then I left the traps there all season until December (mid-summer). In December, I opened each trap carefully and counted the earwigs inside. I then measured the proportion of damaged fruit out of 10 fruits on each tree. I also measured damage on a paired control tree closest to, but not touching, each study tree.
Ideally, I would have left the traps there until the end of summer and measured fruit damage around harvest time in February-March. But my contract ended at the end of December, so I had to finalise all my field work before then…
Out of 40 traps, a grand total of 12 remained in place when I went back in December. All the rest had disappeared. It looked like a few had been attached to branches that had been pruned off, while the remainder had been removed or fallen out of their holding place. Cable ties were still intact on the trees and I found some of the cardboard traps shredded in grass nearby. I assume birds, or maybe curious mammals, had pulled the traps out of trees to get the earwigs inside, and then the discarded traps had been mowed over.
Unfortunately, the trap thieves didn’t leave me enough traps in each of the four fruit varieties to answer my original question of how damage compared between fruit varieties. And with 12 traps, there wasn’t really a lot of data points to test much at all, even for a pilot study.
For what it’s worth, there appeared to be more fruit damage on control trees (without traps) than there was on trees that supported enhanced earwig populations. There also looked like there might be a difference in damage levels between the varieties, but my data aren’t solid enough to confirm this…more studies needed!
Table 1: Fruit damage and number of earwigs per trap for each pair of study trees.
|Fruit Variety||% damage trap||% damage control||∆ damage||Earwig counts|
trap = trees holding earwig traps; control = paired control trees without earwig traps; ∆ damage = difference in damage between trap and control trees; earwig counts = number of earwigs inside the shelter trap.
I also found predatory spiders sheltering inside a couple of traps…were they feasting on earwig smorgasbord, or living communally?
Any animal species can be a ‘goodie’ and a ‘baddie’ on farms, but we tend to label things as one or the other because we think it makes it easier to manage them. Our previous research shows that looking at trade-offs between these costs and benefits is more informative than measuring either costs or benefits as isolated outcomes. Earwigs are a great example! Identifying how earwig costs (fruit damage) trade off against their benefits (predators of other fruit pests) can help orchardists manage to optimise their benefits.
© Manu Saunders 2017