Friday, 18 June 2010

How screwed is the polar bear?
Even deniers like Akasofu wouldn't argue that the planet isn't warming. He just casts the blame elsewhwere, in a probably incorrect manner. It's the naïve who say that global warming is a hoax because "it snowed a lot last winter" (and so are those who believe in global warming because it's hot today, but for different reasons), that or the ideologues who view this through a partisan lens. Let there be no mistake, creatures like polar bears are adapted for conditions that won't occur again until the next ice age, and things are only going to change more as time goes on.

That's not to say we know everything about polar bears. The problems are not trivial - logistics are massive, the creatures elusive, and weather is ever your foe. It's difficult to get a reasonable trend on the species, in whole or in part, because counting is easier said than done. That's where morphemetrics supposedly can come in. The idea is that we can measure things, take masses, count the cubs, and figure out from there how well everyone is doing. In some respects, this has many advantages over census, where populations can be high, but unwell and ready for a massive decline.

And so, Dr. Rode, Dr. Amstrup and Dr. Regehr, all three Forest Service or USGS here in the state, have published a paper collating 28 years of morphometric data from polar bears in the southern Beaufort sea, which spans from midway between Point Hope and Wainwright, to somewhere just west of Nunavut's western most continental extreme. Thankfully, the spell out their study aims very nicely:
Specifically, we addressed the following four questions:
(1) Is reproductive output, quantified as litter mass, associated with maternal condition? If so, what measures of female stature/condition (condition indices, body mass, skull size) are most closely related to reproductive output?
(2) Did body mass, skull size, or condition relate to interannual variation in available ice habitat?
(3) Did body mass, skull size, or condition of polar bears exhibit a trend between 1982 and 2006?
(4) Did reproductive output (litter mass and cubs per-female) exhibit a trend between 1982 and 2006? Was it related to interannual variation in available ice habitat?
The study is deceptively simple in terms of field methods - they flew out in a helicopter oppertunistically tracking polar bears in May when polar bears are aggregated on the continental shelf, or in a few years in the Autumn when they're more on the shorefast ice or land. They would dart the bear, jab any cubs, and then take some basic measurements including skull width, body length, mass, they would age the bear (which can be done from the teeth), and quantify the cubs. In all, they captured between 5 and 152 bears a year

For the ice, things became more... mmm... abstract, we'll say. Instead of taking satellite measurements of ice, they use collar data to generate a resource selection function - which is to say, where the bears prefer to be. They called the top 20%, where 70% of the bears dwell, optimum habitat, or "Ice." From there, they dive into what I can only call model-mania, which is typical of a paper of this nature. They compete different models to see which are the best, and generate many tedious tables that all but the most interested individuals find themselves nodding off during. Don't get me wrong, the information of these models can be highly informative, as can their weights, but they're incredibly dry. I've sometimes caught individuals trying to use this fact to their advantage in sneaking through poor data and hoping people nod off after a hearty meal of model stew. I'll spare you the details of the growth curves, but suffice to say they broke the bears up into 2 age classes for each gender, and 1 for cubs which included all but bi-yearlings.

The answer to question (1) appears to be an unqualified yes, with high probability. Sow mass influences cub mass, and not inconsiderably. This is true of cubs of the year in both autumn and spring, and yearlings in just the autumn. This is unsurprising, but it's good to see that their variables are measuring what they think they're measuring, for purposes of our future analyses. The answer to (2) is also yes, but only for some age groups. Curiously, a trend over time is only evident if you look at age classes  (7-11yr olds), but there appears to be a simple explanation for this - lighter weight bears tend to die more. This leaves only the older bears, for whom there is no trend over time. They would be left at a reduced number, if this was the case. The best trends appear to come from skull width for females and "Ice", and Skull width and body length for both sexes early in life. This is curious, since neither variable acts fast. Bone is a slow growing (though dynamic) tissue, and you would expect it to react slowly to yearly variation of ice. This is to contrast with mass, which can change quite rapidly as conditions worsen. An explanation why it's a major factor for fast growing polar bears is that this is a key stage in their development, and environmental stress is getting 'recorded' in their body. An older bear just needs to pile on the pounds, and care for cubs, whereas a younger bear needs to do this as well as continue to grow larger and larger. As anyone who has been around teenagers know, this takes a lot of food.

Question (3) is a yes - there were trends over time in female skull width, body length, and some fewer variables in males. I'm less interested in this fact than I am in the trend in reproduction, where litter mass is trending down over time (while being positively associated with "Ice") at the same time that yearlings are trending down over time, with the same positive association with "Ice." This suggests there is a "Double Whammy" - not only are there fewer cubs, but lighter cubs. Recall that mass is good for survival, so these lighter cubs will be more likely to die earlier, and sire or whelp fewer cubs of their own. This is a real matter of concern, and should highlight how important it is to keep a close watch on the SB sea polar bears in the future.

They discuss the variance and trend in their "Ice" variable. As "Ice" doesn't truly measure ice, I'm hesitant to draw too much from this. What the graph I've put to the right really shows is that in some years, bears are more clustered than others, and over time the clustering is increasing. There are many  factors that could lead to this pattern, including decrease in true ice coverage, changing patterns of seal availability, and increased human related disturbance. The arctic is a changing place, and it is becoming a busy place, and I saw no attempts to disentangle their "Ice" from this. We don't actually know that true ice declines lead to any future trends we see in this paper, we only know that an increase in bear density (or a decrease in habitat selected) does. This is this paper's strongest weakness, and makes all further "Ice" models slightly weaker, in my mind.

Can be density increasing? If the population is declining, then a decrease in habitat selected could maintain the same density of bears. However, there is no statistically robust trend of bear population for the SB sea. The problem is that measurements in the 80s had a large deal of imprecision associated with them. I generally suspect bear census is declining, but as a properly skeptical scientist, I must point out this has not been well demonstrated. Discarding that hypothesis, density would increase, leading to many of the downstream
 findings that I've previously discussed. We know that density is bad for survival (males kill cubs), increases stress, and can decease resource availability to any given bear. Additionally, it can lead to an increased transmission of disease. The authors rightly discount selective hunting for large bears driving the trends (polar bears are shot opportunistically), and argue that contaminants are just as unlikely to be a source for a decline. But given the poor support for their "Ice' variable, we can equally argue that what we're seeing is density dependant effects from some other factor that might not stem from ice per-se. I am a fan of keeping my datasets simple. If I wish to measure ice's association with polar bears, I would prefer to measure the ice.

Rode KD, Amstrup SC, & Regehr EV (2010). Reduced body size and cub recruitment in polar bears associated with sea ice decline. Ecological applications : a publication of the Ecological Society of America, 20 (3), 768-82 PMID: 20437962


Hoover said...

"It's the naïve who say that global warming is a hoax because "it snowed a lot last winter""

This is a straw man. I have never heard or read anyone say that global warming is a hoax because it snowed a lot last winter.

TwoYaks said...

If only it was a strawman. This particularly inane meme has a fair bit of play. Here it is in the popular press,

... and I've heard it repeated by the public. If you haven't heard or read it, it must be because you keep very fine company.