How to write your ESA abstract even though you haven’t analyzed your data yet

Note from Jeremy: this post is by Meg and originally ran in 2015 under the title “I have data, ESA, I promise!” I’m re-upping it because it’s timely.

Last week, as I was working on my ESA abstract, I realized that I was including things that I wouldn’t normally, just to make sure I showed I have data in hand. The ESA Abstract Guidelines include this requirement:

The abstract must report specific results. The results may be preliminary but they may not be vague. Abstracts without explicitly stated results will be rejected. It is understandable that abstracts describing non-traditional work may lack quantitative data; however, it is still expected that the abstract will address some question and have a “take-home message” describing specific findings.

The abstract I submitted this year combines what will end up being two different publications. We’re working on one of those publications now, and just have a few loose ends to tie up before it will be ready for submission. That project redescribes a parasite that attacks developing embryos of Daphnia, and characterizes its phylogeny, virulence, and ecology. But I’m guessing the phylogeny part won’t be as exciting to an ESA audience, and the virulence stuff can be summarized quickly, so I decided to combine portions of this first manuscript with a second, less fully developed analysis. That second project deals with both the parasite we’re redescribing and another one that is similar in some ways (in that both sterilize their hosts but do not affect lifespan) but differs in other key ways (one is an obligate killer, the other is not). The main data component of this second project is two years of field data on these parasites in 15 lakes. We have lots of data, and I’ve done some analyses on them, but haven’t fully analyzed them.

So, to summarize, I was writing an ESA abstract for a talk where we’ve done a ton of work, but I haven’t finished analyzing everything. The abstract guidelines are clear that that’s okay (and I’m sure it’s common). I, like many other people, want to talk about new work at ESA, and also use it as a deadline to motivate me to finish up some analyses. But, since I don’t know what the threshold is for enough specific results in an abstract and I want to be sure I’m above it, I suspect I tend to put more in than really would be needed. (*see footnote below; I am NOT criticizing this requirement!)

When I wrote my abstract last week, the point where it really struck me that I was tailoring my abstract because of this requirement was when I wrote:

Infections of both parasites were observed in all six host species and all 15 lakes. However, there was substantial variation between lakes in the prevalence of infection, with infections rare in some lakes but common in others.

After writing that, I thought, “Hmmm, is that specific enough? That could sound kind of vague and like I haven’t really analyzed the data on this.” So, I added in these sentences right after:

In 2014, maximum infection prevalences of the brood parasite reached 4.9-8.7% of the entire population and 9.1-20% of the asexual adult female population. Maximum prevalences of the bacterium ranged from 0.2-54.5% of the population.

That’s probably excessive detail for an abstract, but at least it makes it clear that we really do have data and it’s certainly specific!**

How much do you tailor your ESA abstract to address that part of the guidelines? Have you had an abstract rejected because it didn’t contain enough specific results?

*I want to emphasize that I am NOT criticizing ESA for this policy. It makes sense to me that they want to be sure there’s a reasonable chance the person can give a talk that will be interesting to others (and a talk with no data is less likely to be interesting). And I can’t even imagine the amount of work that goes into sorting through all the abstracts and making those decisions. I’m glad I do not have to do that! I’m simply describing how I think about that guideline (maybe more than is necessary) while writing my abstract, to try to make sure I’m above whatever bar there is for specific results.

**My abstract ends by talking about ongoing analyses that we are doing, so it makes it clear that we haven’t fully analyzed the data yet.

 

2 thoughts on “How to write your ESA abstract even though you haven’t analyzed your data yet

  1. Back in ESA 2014, I had an abstract come back to me with a “revise” decision, because they claimed it wasn’t obvious I had results. I think this is partially due to the reviewer not being a theoretician (and that in the first abstract, I was not absolutely crystal clear that this study was a theory paper), but in addition, I did not have any specific numbers in my first abstract. Below I include the abstract they were on the verge of rejecting and the one they accepted, in hopes the comparison may be useful for others. Parts with significant additions/changes are in { }. Adding the “25%” figure may have been key. I think based on the two abstracts below, it is possible that Meg’s initial sentence could lead to skepticism by a reviewer. I believe, to this day, I hold the distinct honor of being the only student in my advisor’s lab who was ever required to revise an ESA abstract. One point to make: If you get back a “revise” decision, this occurs a couple of months after you submit, and you presumably have clearer, more specific results by then. As was the case in my situation

    Nearly Rejected abstract:
    *Optimal detection of invasive species when eradication is and is not possible*
    While there is a rich literature describing the optimal deployment of pesticides and natural enemies to control invasive species, very little is known about how managers should sample to detect them. Nearly all past studies rely on the assumption that local populations of the invader can be completely eradicated soon after they are detected, and no other control options are available. However, eradication is often slow or impossible, and when it is possible there may be alternative controls to choose from. We consider a simple model and describe how different control strategies: (1) local eradication, (2) reduced growth and (3) reduced spread, affect optimal sampling protocols.
    We find the control action can greatly affect the optimal sampling strategy. For example, in many scenarios, if the invasive is immediately eradicated following detection, a manager should sample less than in the case where detected patches are quarantined and prevented from growing into local outbreaks. However, when the spread of the invasive is much slower than it’s local growth, this pattern can be reversed. Our results suggest that models based on the eradication assumption may not provide good sampling guidelines for systems where managers use other control methods.

    Accepted:
    While there is a rich literature describing the optimal deployment of pesticides and natural enemies to control invasive species, little is known about how managers should sample to detect them. { Past studies focus on equilibrium sampling protocols that are constant through time, despite the importance of transient dynamics in invasion biology } . In addition, these models rely on the assumption that local populations of the invader are immediately eradicated after detection, even though eradication is often slow or impossible, and when it is possible there may be alternative controls to choose from. We consider a simple model { that relaxes these assumptions, and use optimal control theory to } describe how different control strategies: (1) eradicating local populations, (2) preventing outbreaks and (3) reducing spread, affect the sampling protocols which minimize the total cost of management and damages.
    We find the control action can greatly affect the optimal sampling strategy. In many scenarios, if the invasive is immediately eradicated following detection, a manager should sample less than in the case where detected patches are prevented from growing into local outbreaks. However, when the spread of the invasive is much slower than it’s local growth, this pattern can be reversed. { In addition, for invasive species that spread quickly, the best constant sampling strategy is up to 25% more costly than the optimal sampling strategy that varies through time. On the other hand, if the spread is slow, the best constant strategy can perform just as well as the optimal time varying one. Our results suggest that sampling guidelines from past studies may be effective for slowly spreading invaders, but potentially under predict sampling effort for invaders that spread quickly. }

    In case the abstracts piqued your interest, you can read the polished paper here http://onlinelibrary.wiley.com/doi/10.1111/1365-2664.12617/pdf

Leave a Comment

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s