How fast can sea level go up with Polar Ice Cap melt?

I can't recommend this video. For the most part it is right, but it goes way overboard at the end. At about 4:45 Peter's voice-over announces "Eric Rignot, writing in the Guardian, pointed out that while at the current rate of melt, the area might take 200 years to flow into the sea, evidence indicates the rate will continue to accelerate".

Contrast this with the transcript of the NASA press conference,

http://www.npr.org/2014/05/12/311896973/past-the-point-of-no-return-an-…

wherein Ian Joughin, author of another prominent study recently released, says "It's not like a building collapse that, you know, would occur over seconds. It's a collapse that's going to occur over centuries. ... Our worst case scenario had the rapid onset of the collapse occurring in just over a couple hundred years." [ The best case took over 900 years. ]

So is Rignot disagreeing with his colleague? Well, look at what he actually said in the Guradian article: "At the current rate, a large fraction of the basin will be gone in 200 years, but recent modelling studies indicate that the retreat rate will increase in the future."

By putting this material in juxtaposition with talk of the five-meter-per-century sea level rise of 14,000 years ago, and speaking of "acceleration" the video encourages the viewer to interpret this as saying that the ice sheet itself will be gone in 200 years at current rates and far less given the acceleration discussed here.

A simple extrapolation of sea level rise shows this can't be the case, so that has to be a misinterpretation.

Indeed, Rignot in the same Guardian piece that Peter paraphrases says " It remains difficult to put a timescale on [the fast retreat of West Antarctica] because the computer models are not good enough yet, but it could be within a couple of centuries".

So the answer to your question is "very fast compared to what you might expect", but not "very fast compared to very fast compared to what you might expect"!

While sea level rise rates approaching or even exceeding 5 meters per century may yet occur, there is no indication that this is the century in which it will happen, and the expert opinion is that the peak collapse rate remains some few centuries in the future.

Some people also claim that predictions of the imminence of sea level rise keep getting worse and worse, but they weren't talking to actual ice sheet glaciologists. By sheer happenstance, I was. A few years ago, a collapse within this century was considered to be in the plausible range. It isn't now.

It is no joke that we are committing our descendants, barring heroic measures, to losing vast coastal areas. But there's no new evidence that the brunt of it will begin in our lifetimes.

Michael, the key point seems to be that there is no means of placing an upper limit on the speed of the process. I can't see how extrapolating SLR does that. Models necessarily do, but confidence that they correctly capture all the processes affecting ice loss can't be high.

Re "our lifetimes," if by that you mean the next 20 years or so, maybe not. But it's a mistake to use that yardstick.

Here's the take-home from the paper:

Since the 1970’s, the ASE ice discharge into the Southern Ocean has been increasing, especially in 2003-2008 when Pine Island and Smith/Kohler glaciers un-grounded from their ice plains, as documented by the abrupt change in speed during that time period (Mouginot, et al. 2014). Ice flow changes are detected hundreds of km inland, to the flanks of the topographic divides, demonstrating that coastal perturbations are felt far inland and propagate rapidly. As shown here, the glacier grounding lines retreat rapidly, at km/yr, over the entire sector. On Smith/Kohler, the retreat rate of 1.8 km/yr is even greater than its rate of horizontal motion of 1.1 km/yr. The retreats proceed, as expected, along deeper subglacial channels, and accelerate where the surface is at the onset of floatation. The grounding lines retreat dominantly on a retrograde bed, except for the west branch of Kohler, hence are inherently unstable, with no clear feature of stabilization upstream. These observations of change in velocity and grounding line retreat therefore concur with recent ice sheet model simulations to indicate that this sector of West Antarctica has developed a marine instability. (Emphasis added.)

The bolded bit, stating that the drainage effects observed at the grounding lines have propagated very quickly all the way to the top of the basins and done so very recently, is an indicator of a very fast process indeed. Note that this is before there's been a chance for the abrupt collapse process described by Joughin et al. to kick in.

"So is Rignot disagreeing with his colleague?"
Michael Tobis

"Eric Rignot, a climate scientist at the University of California, Irvine, and the lead author of the GRL radar mapping study, is skeptical of Joughin’s timeline because the computer model used estimates of future melting rates instead of calculations based on physical processes such as changing sea temperatures. “These simulations ought to go to the next stage and include realistic ocean forcing,” he says. If they do, he says, they might predict an even more rapid retreat."

http://news.sciencemag.org/climate/2014/05/west-antarctic-ice-sheet-col…

I am not qualified to comment further on this subject, but there does seem to be some difference between Rignot and Joughin regarding the rate of melting. Personally, I find a difference of this kind between scientists quite natural. I would find it far more disturbing if scientists (using different methods) always reached the same results and evaluated their results in the same exact way.

A lot of interesting comments.

Michael, I basically agree with you. First, there is no doubt that the current level of CO2 seems to be associated with massive sea level rise in the past, in some cases much more than any of the numbers discussed in the video or Rignot's guardian piece, though there is a lot of variation. Second, as we continue to pump CO2 into the air, that only gets worse.

Obviously the key thing here is this: Increasing CO2 raises surface temperatures slowly, and temperature increases melt ice over time, not instantly. So it is possible to say that at current CO2 levels/temps, or at some future reasonable CO2 level/temp the potential sea level rise is X+/-Y meters.

What we are totally blind on, and I think the models are very immature though improving quickly, is the rate of sea level rise. So we might reach 5 m potential and then see that realized in centuries hence. Or at some other time scale.

I agree that Peter's video, near the end, is worded incorrectly and perhaps he should change that. The implication many will walk away with, putting together the comments throughout the video and Peter's bit at the end, is that in less than 200 years, or sooner, that entire basin (or most of it) will melt. That is not what the current science is saying

I don't quit get what you are saying about glaciology decreasing estimates. That may be something over the long term. But in recent years, looking at what has gotten in the IPCC reports and what has come out recently, too late to include in the present IPCC report, we can see two things: 1) A general upward trend in estimated total SLR, and in some cases, maybe many cases, rate; and 2) a general tendency for models and/or current observational estimates to predict something that looks very little like the paleo record in terms of magnitude or rate.

When I show a graph of this in talks I tend to use a "?" as the time scale and make sure to emphasize that.

Also, while I think sea level rise rates during the late P and early H are important and informative, that ice melt rate may be pretty much useless. Imagine knowing a lot about Greenland, and then being shown a rough map of Antarctica. It would be very difficult to use the understanding of Greenland to predict changes in a mainly unknown Antarctic. How do we use an increasing but inadequate understanding of Greenland and Antarctica to understand the Laurentian Ice Sheet?

In the end I have two opinions formed from reading the literature and knowing as a paleo person who's dealt with sea level rise before it was popular: 1) the total melting potential is almost always understated for reasons unclear to me (there has never, or very rarely, been a time in the past with the present CO2/Temperature configuration with this low a sea level) and 2) the rate over which this may happen is best characterized as unknown. So far.

I think this second point will be addressed empirically before it is addressed adequately with models, frankly. The reason climate models work so well is because detailed observations can be used to develop them or challenge them. There are no detailed observations of polar ice sheet degradation. With luck (both very bad and good, depending on one's goal) we'll see more activity that will inform the science over the next several years.

In the novel I'm working on, which is in part an outlet for thinking on issues of race, society, climate change, and such, I put the total melting of the ice caps out some distance, so by the 26th century it is done and no one really knows when it happened or how long it took because such knowledge is lost. That may be too soon for some models, but I'm banking on factors that degrade ice sheets being discovered at a higher rate than factors that preserve or grow them in a warming world, including the development of surface rivers to drain seasonal melt efficiently, strong effects of SLR on grounding line stabilities, and isostatic rebound causing tectonic effects that encourage collapse and movement. If I finish my novel at around the same time the models prove (more or less) convincingly that that is not enough, I'll invoke a comet. No problem.

But yes, Peter would do well to adjust the wording in this otherwise excellent video.

This video seems more realistic to me:

https://www.youtube.com/watch?v=t8SPrR4K-tM

It lacks the polemical bite, but it seems closer to representing the immediate scale of the issue (in the most extreme worst case my sense is that Amundsen Bay phenomena add maybe a half meter in the next century).

Again, I agree with Greg that in the long run, barring a huge project for capture and sequestration of ambient CO2, the sea level rise is enormous. And indeed, the West Antarctic kicks in massively and relatively soon. As someone who does think of very long times in the past and future as nevertheless real actual events, I think this matters a great deal.

Apparently neither politics nor economics agree, which may be why there is so much focus on the present day impacts.

But like it or not, present-day and near-term impacts are tiny compared to those of the long run.

I still don't see the problem with the video. There's a claim of a risk of something like MWP 1A, but that seems supportable (although the scale of the risk is unknown). The ASE basin itself contains only a couple meters of SLR-equivalent, but note that it was Rignot who raised the issue of the vulnerability of other ice streams to the same rapid melt process.

I also want to point to the choice of time scale in the very last sentence of the paper:

"We conclude that this sector of West Antarctica is undergoing a marine ice sheet instability that will significantly contribute to sea level rise in decades to come."

Not centuries, decades.

We should be clear that both the ice sheet models and the GCMs have major problems in terms of projecting anything along these lines, the former because they're still missing processes and the latter because they can't manage expansion of the tropics (=> poleward movement of the westerlies => warm intermediate water getting pushed up against the ice) at the observed scale.

Maybe the most important thing to bear in mind is that, while at first glance an MWP 1A-like event would seem less likely because the candidate ice is so much closer to the poles, today's forcing is qualitatively different as GHGs operate all the time, including at night and during polar winter. Expansion of the tropics is being driven primarily from the tropics, where orbital forcing changes are weak, so It seems unlikely that the westerlies would have been affected so strongly during the deglacials. Plus let's not forget that present GHG forcing is far from static, rather it's increasing rapidly. Current effects are thus lagged, and we can expect them to grow much larger in the near future.

Finally, recent results have identified the ice-elevation feedback to be the key process whereby rapid retreat of land-based ice occurred during deglacials. The poleward location and high elevation of the Antarctic ice may mean that it's not very vulnerable to this effect since surface melt seems to be a requirement , but that situation could change rapidly if the westerlies continue to move south and strengthen, and if inland retreat of ice streams such as the Totten create the ice saddles that seem to be most vulnerable.

I mentioned this in the thread at P3, but I'll repeat that what really made me stand up and pay attention to this issue wasn't MWP 1A, that ice being long gone, but rather the ANDRILL results for the Pliocene that came out in 2007, showing repeated high-volume obliquity-paced Antarctic ice sheet loss. That stuff can move fast when it's pushed hard.

Joughin et al. discuss the extensive limitations of their results:

Our simulations are not coupled to a global climate model to provide forcing nor do they include an ice-shelf cavity-circulation model to derive melt rates. Few if any such fully coupled models presently exist. As such, our simulations do not constitute a projection of future sea level in response to projected climate forcing. The results, however, indicate the type of behavior that is likely to occur. In particular, all the simulations show the grounding line stepping back in stages with concurrent increases in discharge, consistent with other models and observations of paleo-ice-stream retreat. The intensity of melt in our simulations regulates the time scale over which this pattern of retreat occurs. Thus, while cavity-circulation models driven by regional ocean-circulation models coupled to global climate models might yield differing spatio-temporal variation in melt, they should produce patterns of retreat similar to those we have simulated but with tighter constraints on the timing.

An import (sic) feature of our numerical simulations is that they reveal a strong sensitivity to mechanical and/or rheological weakening of the margins, which can accelerate the rate of collapse by decades to centuries. Thus, future models will require careful treatment of shear margins to accurately project sea-level rise. Our simulations also assume that there is no retreat of the ice-shelf front. Full or partial ice-shelf collapse should produce more rapid retreat than we have simulated. In addition, we have not modeled ocean-driven melt that extends immediately up-stream of the grounding line, which could also accelerate retreat.

As far as I know GCMs still don't do ice sheet dynamics at all. If anybody does, that would have to be considered an ESM (broader than a GCM) in my opinion.

We do need to note that Joughin and Rignot are discussing different glacier basins, the former a part of the East Antarctic sheet, and the later a piece of the WAIS. These are still relatively small chunks of the entire ice sheet. They are not in disagreement, although an apples to oranges comparison might make it seem so. There is also a third paper out, with radar results showing deep channels leading from the coast into interior greenland, so there is also the potential for the GIS to collapse/decay on timescales of centuries instead of millenia.

We really don't yet have a handle on this, this is just an indication that the long timescales for the large ice sheets to reach equilibrium, which had been assumed to be millenia are probably wrong.

"Few if any" is strange phrasing. Maybe what they mean is that people are working on adding this to GCMs? In any case it sounds like a key lack, although as they note it's far from the only one.

Later they refer to "cavity-circulation models driven by regional ocean-circulation models coupled to global climate models," where presumably an ice sheet model would be the first link in the chain. This makes it pretty clear that to start with GCMs need to be able to get the westerlies right, which so far they can't.

But for sure these are interesting times for modelers!

"significantly" doesn't necessarily mean "consequentially". It's easy to see that claim as spin.

I am finding Rignot's imprecision tendentious and possibly self-serving.

That said, I agree that the recent Greenland topography result is very concerning - the now demonstrated existence of sub-sea-level channels on Greenland was exactly what we hoped would not turn out to be the case. But that doesn't enter into the video.

I think we should await what the ice sheet models say about the time scale of the Greenland failure, but it looks likely that the GIS is going to fail too. But if the science says hundreds of years until the big pulse, we should still not try to confuse people about that.

I have always advocated decision making is rationally dominated by plausible worst cases. But plausibility is pretty much the key to making that work. It's tricky business right there at the plausibility margin, and going beyond a realistic worst case has huge negative consequences too.

OC, both papers discuss the ASE, although Joughin et al. focus on the Thwaites. Neither studied the EAIS or indeed the remainder of the WAIS, although Rignot has stated that the marine-based ice sheets throughout the continent are vulnerable to the same processes. Note that the ASE ice streams were the first to be impacted by westerly-driven encroaching warm water (pulled up from a considerable depth). That water has not significantly affected any other ice streams, but the expectation seems to be that this is only a matter of time as the westerlies continue to move south and strengthen.

"But if the science says hundreds of years until the big pulse"

Michael, please point to where it says so. The science says that's plausible, but doesn't exclude something much quicker. Pretty clearly the models aren't in a position to do better at the moment.

Re the GIS, it's much farther south on average and is experiencing extensive surface melt and rainfall, so I think it's fair to say that unlike the EAIS or WAIS it has the prerequisites for near-future vulnerability to the ice-elevation feedback. The only things it has going for it are a remaining degree of topographic restraint (although nothing like what Pfeffer et al. postulated a few years ago -- I recall glaciologists very recently still referring to that paper as pretty much settled science) and the lack of (extensive anyway) retrograde slopes underlying the ice streams.

I'm such a naysayer today...

I'm not a big fan of the ESM approach. Let me put it this way - if you don't believe an ice sheet model standing alone, why would you believe it fully coupled into a climate system?

Science doesn't "exclude" unknown phenomenology, ever, but that's a lame excuse that deniers dredge up in various forms all the time.

The question is whether we have enough to explain what we see. Known phenomenology now seems to explain what we see pretty well, and known phenomenology does seem to counterindicate a WAIS collapse in less than two centuries as I currently understand ice sheet dynamics. What I've heard from Joughin fits in nicely with that, and what I've heard from Rignot seems like spin, rather than counterargument.

"I am finding Rignot’s imprecision tendentious and possibly self-serving."

This seems to come down to differences in physical intuition. I seem to share Rignot's (and Jim Hansen's). I don't know about Rignot, but I won't claim for a moment that mine started out science-based, rather it's a result of very close observation of small-scale snow and ice melt, although subsequently reinforced by the ANDRILL results I mentioned above.

It's interesting to look again at this RC discussion of Pfeffer et al. from 2008. Jim Hansen appears once again to be likely to have been prematurely correct. To be clear, his 5 meters was stated as a maximum and we may well get less, but if any hubris is evident in retrospect it's on the part of some reticent glaciologists.

What "known phenomenology"?

Pfeffer et al. relied on some known phenomenology when it came to bedrock depth. Not so much, it turns out.

Also, note that there have been IIRC on the order of half a dozen papers in the last couple years adding significant new pieces to the ice dynamics puzzle.

Greg, there have been a number of ice sheets that have been observed to collapse. For example the Larsen B.

http://earthobservatory.nasa.gov/Features/WorldOfChange/larsenb.php

All ice sheets that have collapsed have done so catastrophically. The reason for this is because liquid water is denser than solid water, so when the top of an ice sheet starts to melt, the water flows down (because the pressure at the bottom of a column of water is higher than the equivalent depth of ice). As the water flows deeper, the pressure differential gets higher. Until either the water hits bedrock, or until the water hits ice cold enough to freeze it.

When it freezes, it deposits its heat of fusion and the surrounding ice increases in temperature until it hits the melting point.

The strength of ice becomes zero at the melting point.

When the strength of the ice sheet goes to zero, then it disintegrates.

The only reason the ice at the bottom of the ice sheet remains below the freezing point is because of heat conducted up through the ice sheet and radiated away into space during the winter. When the ice sheet becomes isothermal, there is no gradient for the heat to flow down, and geothermal heat starts melting ice at the bottom.

Once the ice sheet starts to collapse, gravitational energy from the collapsing ice sheet adds heat to the ice already at the point of failure. Heating the ice that is already failing, decreases its strength even more and accelerates its failure.

daedalus2u, Larsen B was an ice shelf (floating) rather than a marine-based ice sheet (grounded below sea level), but many of your points are applicable to both. More than anything, the Larsen B collapse was a big surprise, I suspect not the last such.

daedalus2u, we are using the term differently. I'm talking about the continental glacier level ice sheet. Larson is an ice shelf, a much smaller thing, part of an ice sheet.

So, for example, much of that ice sheet was far to the south and not in a polar circle, while all the existing ice sheet are within the polar circles, except for little bits. The Laurentide ice sheet would have been subjected to tropical storms, very hot summers, very warm sea currents. etc.

EG at 11K when southern parts of the ice sheet were still in New England, we believe sea currents off the coast of Boston had a temperature similar to the Carolinas, based on clams found in the sediments from the glacial outwash that live in subtropical waters. (That was probably some short term quirk of the gulf stream, but lasted long enough for clams to migrate!). The southerly margins of that ice sheet would have been subjected to different levels of stress than the existing ice sheets.

But yes to all that other stuff.

11 kya was toward the end of the deglacial process, when all of the ice sheets were retreating or losing ice quickly, and was a disruptive time for ocean and atmosphere circulation, so that warm New England current sounds more like a deglaciation feedback than a driver. These are comparatively much earlier days for the impact of increased CO2 on the ice sheets.

But even orbital forcing (with feedbacks) can do a lot to to ice sheets located near the poles, as witnessed by the record of fast obliquity-paced changes I mentioned above. Per Lake E drill core pollen and Ellesmere Island fossils (featuring e.g. camels), the Arctic at least, and perhaps much of the Antarctic periphery (harder to find proxies there), got pretty toasty ~3.3 mya when CO2 levels were similar to present.

At ~600 ppm CO2, which we seem very likely to reach, given a little time the ice sheets will disappear entirely. So for processes sufficient to melt the ice sheets, even near the poles, it's a matter of when rather than if.

Once the ice-elevation feedback kicks in, I see no reason to expect it to go any more slowly than it did during the deglacial for less poleward ice sheets. Insolation will be much lower, but a lot of heat will be getting transported to polar regions due to continuing expansion of the tropics, and as I also mentioned above CO2 levels much higher than those of the deglacial will act to keep it warmer at night and during the winter.

I'm curious. I'm told there's no chance of going Venus -- at least not yet -- but I wonder what the threshold CO2 level is for that.

Anyway as it is we're talking about hitting 660+ ppm and the sea levels going up to levels akin to the Cretaceous. There's a lot of forested land and plant life in those low-lying areas -- take, for example, the stretch in the lower Amazon River, or much of Connecticut and Rhode Island and Louisiana. Underwater all those trees are going to be dead, and releasing CO2 as they decay. So the CO2 levels get jacked even more.

And I see no reason whatever to assume that sea level rise won't be catastrophic. In fact to assume that we're going to get away in the next 50 years without New York City being underwater seems folly, since in just about every case I can think of environmental disasters have been far, far worse than anyone ever predicted. Really, can anyone come up with a single case where we could say, "wow, it wasn't so terrible?"

Or to put it another way: I am rather pessimistic that we will have great-grandchildren to worry about. Do people who run fossil fuel companies have no children? That's the only conclusion I can arrive at. Maye this whole evolving intelligence idea wasn't such a good one.

Mr Wizard, yay!

https://www.youtube.com/watch?v=iKiq5EwkzDg

This proves that Mr. Wizard is a charlatan and a liar!

The "third theory" is correct (and wrt climate change applies to SEA ICE not GLACIERS). Fine. The problem here is that the water plus ice, with the water height buoyancy corrected (by itself), fills the glass But that is tap water that has been further chilled by ice. It is very cold, down near freezing, but above freezing. Room temperature is much warmer, and the temperature of a studio is perhaps even warmer. The water would have expanded and overflowed the glass.

This is why Mr. Wizard got rid of the kid ... so he could wipe up the overflow and pretend like he was right!

Jesse, you may be right about the carbon release from forests. On the other hand, if we ruin the chemistry of the ocean and with mass extinction it becomes a carbon sink (as has happened in the past) that will leave more room for the next generation (as in aliens that come to a post-human earth) supply of hydrocarbon fuels!

And, again, the thing people often forget and is not mentioned to any great extent here: If sea level goes up just a foot that can be disastrous. Remember that the sea does not go up. It goes across.

As to Mr Wizard - the physics isn't that hard. The ice takes up more volume than water of the same mass. So if I have 100g of ice, it takes up ~110 cm^3, but once it melts that goes down to 100. So if I stick 100g of ice and add volume volume of water, once it melts you will have whatever the volume of water is plus 100 cm^3, even though the ice would jut out from the top of the glass.

So that's great as long as the volume of water and ice are relatively constant, and it applies to ice sheets on water. But it doesn't work if you add more ice from the land. And it ignores the fact that when you freeze water it stops flowing -- that's why sea levels drop during ice ages, the water is stuck at the poles. Conservation of mass still applies. There's a relatively fixed mass of water in the oceans and ice combined. It has to go someplace. We haven't even gotten to the effects of thermal expansion yet.

The Yale forum ones and earlier (and still going) Climate Denial Crock of the Week series are among my very favourite and most useful series on youtube - the eponymous blog is superb too - which had a major impact on changing my views on this issue a number of years ago. A series everyone should watch and learn from I reckon.

> if we ruin the chemistry of the ocean and with mass
> extinction it becomes a carbon sink

What? Citation for that? I'd thought primary productivity was expected to collapse along with the ecosystems?

We've only known about prochlorococcus since 1989, and it's the reason we're breathing -- are you assuming a mass extinction wouldn't affect photosynthesis in the oceans?
http://www.ncbi.nlm.nih.gov/pubmed/10066832

and
http://www.esajournals.org/doi/abs/10.1890/13-0421.1
"Ecosystem-wide primary productivity generally increases with primary producer diversity, emphasizing the importance of diversity for ecosystem function...."
Nicole L. Goebel, Christopher A. Edwards, Michael J. Follows, and Jonathan P. Zehr 2014. Modeled diversity effects on microbial ecosystem functions of primary production, nutrient uptake, and remineralization. Ecology 153. http://dx.doi.org/10.1890/13-0421.1

If an ocean basin becomes anoxic, surface productivity continues (though we don't know at what rate) but a large part of the output of that productivity which would normally cycle in the global biosphere and short term carbon cycle drifts to the bottom of the sea and become future hydrocarbon reserves.

A good amount of the oil and gas used over the last century, for example, is the product of an anoxic Tethys ocean which produced what later came to nearer the surface in places like the Arabian Peninsula and the oil and gas in the mountain ranges and valleys east of that region in, for example, Pakistan.

A great recent work that covers this and much more ... the long geological view of the quarter of the globe of concern, is here: http://scienceblogs.com/gregladen/2012/12/17/the-amazing-and-compelling…

ah (that was fast !) -- ok, I understand that; I had thought the past great extinctions involved the photosynthesizers as well. Perhaps that happened but later, when hydrogen sulfide reached the surface as Peter Ward wrote about a while back? Thanks for the pointer, I'll try to catch up.

This isn't necessarily one of the great extinctions. Just the occasional multi million year period of an anoxic se hear or there.

> drifts to the bottom

Or maybe we need those deep sea fish after all:

http://qz.com/216826/fish-can-slow-down-global-warming-but-not-if-we-ke…
"... Until now, scientists have chalked it up mainly to gravity—to the sinking of phytoplankton that have either died or been eaten and excreted by fish. But new research reveals that we have deep-sea fish to thank for transferring a lot of that carbon into the depths—and that sinking alone wouldn’t do the trick.

In fact, bottom-dwellers transfer more than a million tonnes of CO2 a year from surface waters of the UK and Ireland, ... says a new study (paywall) by a University of Southampton team. Killing too many of those fishes, as well as the ones they feed on, risks damaging the ocean’s ability to store carbon, leaving more CO2 in the atmosphere...."

http://rspb.royalsocietypublishing.org/content/281/1787/20140669.abstra…
"... mid-water and bentho-pelagic-feeding demersal fishes play an important role in the ocean carbon cycle, bypassing the detrital particle flux and transferring carbon to deep long-term storage. "