"LPS Eating Flatworms"- do we really know nothing scientifically about them?

[DavidA]

Is this Claude summary accurate? It essentially says that we have no established scientific knowledge about "euphylllia eating flatworms" including taxonomy, life cycle, mechanism of damage if any and dependence on euphyllia for survival.

My mixed reef is 14 months old with a preponderance of softies, hammers and torches (and a handful of montis). My parameters run Ph 8.1, dKH 8, nitrate 10, phosphate .2, calcium 440. The hammers and torches have thrived but recently I lost 2 hammer heads over 2 separate nights. The many other hammer heads and the torches looked great and were fully open. Because of the losses I inspected the LPS and found many reddish brown flatworms (see photos) which appeared to be around 5mm. I dipped all LPS in Coral RX (with turkey basting) and the corals shed hundreds of worms. I dipped again the next day and many worms were once again shed.The LPS continue to fully open and have no evidence of recession.

I used a Google search and found inconsistent information. In general instead of asking questions of Claude I ask it to write a scientific paper. This guarantees a predictable structure, citations and and a review based on established science.

The lack of scientific information surprised me and makes me wonder if euphyllia eating flatworms really exist or we hobbyists make up causation when we encounter an LPS problem and as a result examine the corals and find incidental flatworms. It is hard for me to believe that the hundreds of flatworms that were shed are harmful if the corals look good with no evidence of recession. That said, I plan on assuming that they may be harmful and will do 6 weekly dips

 

[DavidA]

PDF of Euphyllia eating AI scientific paper

 

[thamnasteroid]

I wouldn't recommend using AI for information, especially for niche and under-studied topics like this.

Euphyllia-eating flatworms are much larger than what you have shown. Generally the flatworms we get on Euphyllia/Fimbriaphyllia and other LPS are Waminoa (I think), but those don't actually eat corals themselves, but rather feeding on their mucus.

 

[EnterName]

To be sure about Order, Family, and Genus, you usually need a full breakdown to the microscopic level. I did something similar (on a very basic level) for a Convolutriloba species in my tank:

First sighting: https://www.reef2reef.com/threads/microscopy-thread.1138845/post-14464850
More details: https://www.reef2reef.com/threads/microscopy-thread.1138845/post-14498224

So if someone has access to one of those EEFW and a microscope I would be more than happy to see some recordings.

 

[Useful_Idiot]

They do exist and they are much larger. I set up a QT after getting them years ago. Big enough to scarf down a tentical and their body looks similar to the flesh line as you can see

 

[Decamicron Labs Verified Sponsor]

I find AI to be very unreliable for anything biological like this. There isn't a lot of reliable training data online, and I haven't heard of any of the companies starting to recruit ecologists to train their models. I doubt the ecologists would know much about something specialized like this either. You'd be better off asking hobbyists who've dealt with this.

When you read a paper, you are reading a finished product with lots and lots of thought and effort that has gone into it. I've first-authored and coauthored two species descriptions of new sea slugs before. For the first author paper, it took us years of data collection before we were ready for the first submission, and it took maybe three quarters of a year with two revisions from peer reviewers. Sure, there was a learning curve for me, and AI could automate and help with a lot of the minutiae, but there are some things like data collection and visual examinations that AI simply cannot do and/or are terrible at right now.

 

[DavidA]

I find AI to be very unreliable for anything biological like this. There isn't a lot of reliable training data online, and I haven't heard of any of the companies starting to recruit ecologists to train their models. I doubt the ecologists would know much about something specialized like this either. You'd be better off asking hobbyists who've dealt with this.

When you read a paper, you are reading a finished product with lots and lots of thought and effort that has gone into it. I've first-authored and coauthored two species descriptions of new sea slugs before. For the first author paper, it took us years of data collection before we were ready for the first submission, and it took maybe three quarters of a year with two revisions from peer reviewers. Sure, there was a learning curve for me, and AI could automate and help with a lot of the minutiae, but there are some things like data collection and visual examinations that AI simply cannot do and/or are terrible at right now.

I agree that AI is not useful for a number of things. But when I approach a reefing question, my first question is always "what do we know from peer reviewed journals?" I find AI to be helpful in this first phase of lit review as long as I instruct it to use the scientific paper template and search the peer reviewed lit. AI is not actually writing a novel scientific paper- it is simply using the template of scientific papers..

In this case AI was helpful. It indicated that there is nothing in the peer reviewed literature about LPS eating flatworms.

 

[BeanAnimal]

. In general instead of asking questions of Claude I ask it to write a scientific paper. This guarantees a predictable structure, citations and and a review based on established science.

PDF of Euphyllia eating AI scientific paper…AI is not actually writing a novel scientific paper- it is simply using the template of scientific papers..

It’s not a scientific paper, that is not how an LLM works. It may read peer reviewed sources, but it is still an LLM and connects dots via language occurrence, not logic. It really should be removed from this site.

 

[Cornelia_Reefer]

I find AI to be very unreliable for anything biological like this. There isn't a lot of reliable training data online, and I haven't heard of any of the companies starting to recruit ecologists to train their models. I doubt the ecologists would know much about something specialized like this either. You'd be better off asking hobbyists who've dealt with this.

When you read a paper, you are reading a finished product with lots and lots of thought and effort that has gone into it. I've first-authored and coauthored two species descriptions of new sea slugs before. For the first author paper, it took us years of data collection before we were ready for the first submission, and it took maybe three quarters of a year with two revisions from peer reviewers. Sure, there was a learning curve for me, and AI could automate and help with a lot of the minutiae, but there are some things like data collection and visual examinations that AI simply cannot do and/or are terrible at right now.

I agree that AI is not useful for a number of things. But when I approach a reefing question, my first question is always "what do we know from peer reviewed journals?" I find AI to be helpful in this first phase of lit review as long as I instruct it to use the scientific paper template and search the peer reviewed lit. AI is not actually writing a novel scientific paper- it is simply using the template of scientific papers..

In this case AI was helpful. It indicated that there is nothing in the peer reviewed literature about LPS eating flatworms.

I use LLMs all day professionally. No format instructions or prompt engineering will prevent the LLM from occasionally hallucinating information including inventing references that don’t exist or telling you no references exist when they actually do. It’s the nature of the technology.

In some occasions it has given me references as evidence and reading the actual material shows the exact opposite conclusion is in the actual content.

LLMs are useful tools but it’s risky to trust any text they generate without verification.

 

[DavidA]

. In general instead of asking questions of Claude I ask it to write a scientific paper. This guarantees a predictable structure, citations and and a review based on established science.

PDF of Euphyllia eating AI scientific paper…AI is not actually writing a novel scientific paper- it is simply using the template of scientific papers..

It’s not a scientific paper, that is not how an LLM works. It may read peer reviewed sources, but it is still an LLM and connects dots via language occurrence, not logic. It really should be removed from this site.

I should have said scientific paper template rather than scientific paper. My apologies. I don’t see an obvious way to remove it. How do I remove?