The [email protected]
*...work done w/Erik...*First, some facts about the little-known
pika. Pikas are a small relative of the rabbit, about as big as
your fist. One important trait they share with rabbits is that they
do not hibernate during winter. Well come back to this point. Other
than that, they dont behave much like rabbits. For one thing,
theyre highly vocal, making a wide array of calls that can be
interpreted somewhat even by humans. For example, at my main study
site, I know whether to look up for an eagle or down for a weasel,
depending on the sound of a pikas alarm call. One of the most
important characteristics of the American pika is that its a
habitat specialist. With almost no exceptions, you wont find it
away from piles of broken rock, such as taluses or rockslides. Like
rabbits, theyre vegetarians. Theyre also generalists, eating a wide
variety of vegetation, from grasses to flowering plants; even
lichens and bark.The American pika is also highly territorial, with
each individual defending its own territory on the talus. This
territoriality probably stems from the need to cache food for the
winter. Because they dont hibernate, they make hay while the sun
shines, and stack it in a big haypile for the winter. Theyll fight
vigorously to defend this haypile, and the territory around it.
This female in the lower corner is calling to defend her haypile,
which is developing in the background. This male in the upper
corner is checking a scent marker used to mark his territory.The
American pika has a broad distribution that stretches quite a ways
into Canada, spanning a wide variety of climatesfrom inland to
coastal, from montane to lowlandbut always in rocky habitats. In
Asia, about half of the pika species live like prairie dogs in
colonies on grasslands. But here, they stick to rocks, and our
challenge is to figure out why theyre disappearing from many of
their historical talus habitats.**The literature is full of studies
showing that pikas can overheat (MacArthur & Wang 1973, Smith
1974a). They have a very high resting metabolic rate, relative to
other animals (Li et al. 2001). Its believed that pikas are
operating so close to their thermal maximum in the summertime that
they need constant access to cooler microclimates like the shadow
being used by this little juvenile. But what do pikas do when its
too cold? Heres a cold, puffed-out pika waiting for this sparse
snow cover in the background to be replaced by a thick, insulating
blanket. There have been a couple of studies suggesting that snow
cover insulates pikas during cold winters, because survival has
been shown to decline in years with less snow cover (Tapper 1973,
Smith 1978; and see Morrison and Hik 2007 for collared pikas).
Recent work that Ive done with Erik Beever, Nifer Wilkening and
others provides evidence for both effects of heat-stress and
cold-stress. Weve shown that local extinctions of the pika in the
Great Basin can be explained by warmer summer temperatures and more
frequent periods of extreme cold. Pikas have disappeared from sites
with warmer average summer temperatures and especially from sites
with more days when the temperature drops below -10 degrees C.But
none of these studies has related the temperatures actually
experienced by pikas to individual mortality or morbidity. We know
that pikas spend more time under the talus when its hot out, and we
know that pikas disappear from warmer sites and from sites with
more extremely cold days. We even have data from the collared pika
suggesting that the timing and quality of snow cover may be the
cause of localized mortality. But we dont know anything for sure
about the mechanism or mechanisms through which the pikas
microclimate may affect its survival or recruitment. Here Ill
discuss a comparative study of pika survival and stress in Colorado
and Montana that is designed to illustrate these mechanisms.*Most
of the evidence for both prehistoric and very recent losses comes
from the Great Basin. This map shows whats happened to the American
pika in the Basin over the past 20,000 years. Everything in blue
here is pika habitat and was probably occupied by pikas right after
the last glacial maximum. Much of the blue area is still occupied
today. Sites of pre-historic pika extinctions are indicated in
yellow. Red and blue symbols indicate 25 local populations that
persisted at least into the 1900. Well zoom in to see these
better.*The 25 historical populations were recorded between 1898
and 1990. The average date of these records is 1933. Erik Beever
began resurveying these sites in the 1990s, and found that 6 of
these populations were extinct by 1999. Weve continued surveying
them every couple of years, and fully 9 were extinct by 2007. There
have been no recolonizations. So, well over one-third of this
sample has gone extinct within the past century, and the rate of
local extinction has risen during the past decade. In order to
forecast future extinctions in this region, were modeling
extinction as a function of likely predictor variables, and were
able to test and update our models as more extinctions occur.
*Lets look more closely at how the range of elevations occupied
by pikas has been changing within this sample. Youll recognize that
this mountain is a little bigger than anything in the Great Basin,
but just bear with me. The red bracket might represent the range of
elevations occupied by pikas on a more typical mountain. In our
historical sample, the average elevation at which pikas were
recorded was about 2300 meters. This isnt a proper minimum
elevation, because previous researchers werent focused on the
elevation question. This minimum is biased high. Nevertheless, by
1999, Erik had determined true minimum elevations for these
populations, and the average of these minimum elevations was 2500
meters. By 2008, it was even higher.The average rate of uphill
movement during the last century was more than 3 meters per year,
and during the past decade its been 4 meters per year. We need to
put some confidence intervals on these rates, but the implications
are disturbing. In North America, we dont have any 6800-meter peaks
like Ama Dablam here. The highest peak occupied by pikas in the
Great Basin isnt much more than 1000 meters above this lower bound
recorded in 2005. At the current rate, pikas would pop off the top
of most of the peaks in the Basin within a couple hundred years.
Other researchers are suggesting similar trends for this species in
the central Sierra Nevada, although weve documented pikas thriving
in some low-elevation habitats in northern California, and Connie
Millar has documented widespread populations on the eastern slope
of the southern Sierras.
*This slide and the next summarize work Ive completed recently
with Erik Beever, Nifer Wilkening and others. Weve modeled local
extinctions observed in the Great Basin according to hypothesized
impacts. The usual suspects in the case of extinction are listed on
the left. In order to model extinction, we need some data on these
potential predictors. We take what we can get. To address these
generalized drivers, weve characterized each historical pika site
in the Great Basin according to the factors on the right: the
amount of habitat available in the mountain range, accessibility by
road, presence or absence of long-term livestock grazing, climatic
variables that should lead to acute heat stress, chronic heat
stress or acute cold stress, and a habitat factor that might
ameliorate effects of thermal stresswhether or not a high-elevation
refuge is present near the historical site. This refuge might
harbor a persistent population that could recolonize the site if
pikas there succumbed to other factors. We constructed logistic
regression models of local extinction based on these factors,
including multiple regression models allowing for the combined
effects of up to four factors. On the next slide, I summarize the
relative support for each of these factors as a predictor of local
extinction, taking into account each factors effect across single-
and multiple-regression models.*Heres how we define local
extinction. This cartoon represents a mountain range, and the blue
outline represents the current distribution of pikas. The blue dot
represents the location of an historical record of pika presence.
If this historical site lies within the current distribution, we
say that the population is persisting. If it lies outside the
current distribution, and pikas cant be found within 3km of the
historical record, we say the population is extinct. If it lies
within 3km of the current distribution, but at least 200 vertical
meters lower than pikas are currently found, then the population is
likely making the transition between persistence and extinction. We
call this transitional, but persisting. To determine absence, each
site was surveyed several times and by different teams over the
course of two or more years, with each survey conducted for 8 hours
during the cooler portions of the day, in late summer, when haying
activity should be apparent.Of the 25 historical populations, only
11 are persisting in this sense, and another 4 have retracted their
range uphill by at least 200 vertical meters.*Ill introduce each
predictor using this same cartoon of historical and current
population records in a mountain range. The first predictor is
Habitat, a categorical variable describing the total area of talus
within the mountain rangecategorical because theres still no way to
really quantify this type of habitat easily. The second predictor
might be called a climate surrogate. Erik reasoned that the climate
at the top of the mountain was probably better for this
cold-adapted species, so if there was some talus at the top of the
mountain, pikas could survive there and recolonize lower habitats.
So, for each historical site, Erik recorded the highest talus
occurring within 3 km. Think of this as a climatic Refuge for the
population. This one gets slightly complicated, because mountains
in the Basin get higher as you go south*So, weve covered two
predictors related to habitat size and structure. Weve also
considered predictors related to direct or indirect human
impacts.We characterized each historical site in terms of its
accessibilty by road, and whether or not it was grazed long-term by
livestock. Finally, we also characterized each site by its
temperature profile...*Heres what the thermal regime looks like
inside a pikas haypile at a couple of my study sites. These points
represent temperatures recorded every 2 hours from Sep 2004 to Aug
2005. The red squares represent temperatures in a single haypile in
southern Montana; blue circles, a single haypile right up above us
on Niwot Ridge.Relative to Niwot Ridge, the Montana site is further
north, but its also a bit lower in elevation. So I guess its not
too surprising that these factors balance out so that these sites
exhibit similar temperatures in the summer. But look at the other
data! Niwot Ridge is much colder in the winter, and more variable
in the spring. These differences arent just due to latitude. The
Montana site is down in a box canyon. Niwot Ridge is much more
exposedsnow cover always blows offso it can get a lot colder in the
haypile during the winter. Niwot Ridge also gets more sun than a
box canyon, so spring snows can melt here and then freeze into
thinner sheets of ice, which dont have the same insulating
properties as snow. Ive been gathering this sort of data at my
study site in Montana for several years now, and Ive begun to link
individual survival data to temperatures around the haypile, where
a pika spends a lot of its time. *We split our hindcast for each
site into two periods: 1945 to 1975, and 1976 to 2006. This allowed
us to characterize what we call the prevailing or long-term
climate, as opposed to climate change between these two periods.
From here on, Ill use omega to symbolize metrics of stress based on
the prevailing climate, and delta to symbolize metrics of stress
related to climate change.By metrics of stress, we mean climate
statistics that capture the sorts of cold stress and heat stress
that pikas are supposed to suffer, given results in the literature.
For example, survival declines in winters with lower snow cover,
and snow cover tends to keep ground temperatures from falling far
below zero, so we guessed that temperatures in the talus that were
below zero or negative 5 degrees Celsius might be stressful to
pikas. For each of these thresholds, we estimated both the
cumulative number of days below threshold, and the change in number
of days below threshold.For heat stress, we hypothesized effects of
chronic stress, represented by the mean temperature in June, July
and August, as well as effects of acute heat stress, represented by
days in which the temperature rose above 28 degrees Celsius. *We
were trying to predict the pattern of local pika extinction in the
Great Basin, including 10 sites where populations no longer persist
(red and blue arrows) among 25 historical population records (red
and blue circles and stars). (The yellow squares represent
prehistoric extirpations.)The relative size of the pikas on the
right represents the relative support for each of several candidate
predictors of local extinction. Some predictors were negatively
related with persistence (upside-down pikas), and some had positive
effects (right-side-up pikas). The best predictor of local
extinction, among the candidates selected, was the number of
extremely cold days at the site over the past 60 years. Pikas
disappeared from sites where temperatures under the talus more
often dropped below -10 deg C. For temperatures in the talus to get
this cold, there has to be a cold snap during a period of low snow
cover.The second best predictor was the relative elevation of an
upslope refuge in the form of talus at a relatively high elevation
and within the maximum dispersal distance of a pika. Pikas were
more likely to persist in locations where this local refugium was
at a relatively high elevation, suggesting that lower populations
are often recolonized or rescued by populations at higher
elevations. In turn, this suggests an effect of climate on
population demographic rates.Other predictors werent so well
supported, but habitat areaat smaller and larger scalesreceived
moderate support as a positive covariate of persistence. Average
summer temperature also received moderate support as a negative
covariate of persistence, with pikas being lost from sites that
were warmer during summer months in recent yearsspecifically, duing
2005-2006, when we had temperature sensors in place at each
site.There was relatively low support for the presence of long-term
grazing at a site as a predictor of extinction, and the other
candidate predictors that I mentioned received so little support
that you wouldnt even be able to see them on this relative
scale.So, the best predictor of local extinction in the Great Basin
is extreme cold, which also appears to predict individual survival
in Rocky Mountain pikas. We certainly need more data before
concluding that we understand any mechanisms through which climate
may affect pika extinction, but Ill be putting a lot of my effort
into examining mechanisms related to cold stress. *There are many
ways that climate might affect pikas. Although this is just a
simplified cartoon of possible relationships, its complex enough to
keep me busy for a long time.**Ive been studying pikas at this site
in Montana since 1989. This is a high-elevation site, at over 3000
m or 9500 ft. Here, I mark individuals and follow their survival
across years. I also place temperature data loggers in the taluses
where pikas live, looking for evidence that survival is related to
microclimate.*Over the past few years, pikas that died appeared to
experience temperatures similar to those that survived. But notice
the slightly higher proportion of extremely low temperatures
measured at sites where pikas died.*In some years, such as
2008-2009, this pattern was quite striking. Well come back to this
in a couple of slides when we compare survival in CO and MT during
this year.*Here we see another interesting pattern. The proportion
of adults in the population declined significantly between 1998 and
1999, and has never fully rebounded from this decline. This
suggests that recruitment (survival) of juveniles into the adult
population has been depressed since the late 1990s.*Juveniles fail
to recruit into the adult population if they dont survive their
first winter. One reason for this may be lack of sufficient food
stores. Here we see the mean size of pika haypiles on August 31
each year, for both adults and juveniles. Since 2000 (the first
year I had reliable data on this variable), adult haypiles have
been about a 1 on this arbitrary scaleno change in haypile size
over time. But juvenile haypiles have declined significantly in
size over this period. This may be related to the continued
depression of the adult-to-juvenile ratio in this population.**Now
lets see whats happening with pika survival and stress in Colorado.
In 2008 we tagged 39 pikas at 4 study sites around Niwot Ridge: 2
north-facing sites and 2 south-facing sites. Of those 39 pikas,
only 5 survived through 2009, and all 5 surviving pikas were in
south-facing sites.*In contrast, the survival rate was normal
(about 60%) during the same year in Montana. The survival rate in
Colorado was off-the-charts low, relative to the pika literature
and previous results from research on Niwot Ridge. Overall, it
ranged 13-22%, and it was 0-11% at north-facing sites (the range is
due to the possibility of missing survivors during our surveysa
probably that has been estimated as quite low in several
studies).*We also measured two metrics of stress in pikas from CO
and MT during the summer of 2008. We had low sample sizes for our
stress studies, due to problems with our pilot study. But one
metric exhibited significant differences between MT and CO pikas.
Plasma glucosemobilized as part of the fight-or-flight responsewas
much higher in CO pikas during the summer of 2008. (95% confidence
intervals are shown.)*Differences in temperature appear to explain
these differences in survival and stress. Pikas in CO experienced
warmer summer temperatures and colder winter temperaturesnote that
the frequencies of both very warm and very cold temperatures
recorded were significantly higher in CO (dotted lines appear above
the range of temperatures that differed significantly between CO
and MT by bootstrap analysis).*Pikas on north-facing slopes in CO
experienced more cold days, which may explain the lower survival
observed on these slopes. *So far, data from many studies suggest
effects of climate on vegetation, and data from our studies at two
very different scales (individual survival/stress in the Rockies
and population loss in the Great Basin) suggest direct effects of
climate on pikas. But recall that there are other possible
explanations for the relationship between pika loss and
climate.*This is the current hypothesis-to-beat regarding
pika/population loss.*An intriguing part of this hypothesis
involves the pikas food cache, which may be affected not only by
changes in foraging behavior but also by climatic influences on
available vegetation/quality.*The potential importance of the
haypile, and of summer temperatures that may impede haypile
development, is pretty obvious at some sites. Haypiles this size
are common at my long-term study site. And this isnt the full
cache. There are other haypiles within this pikas territory, and at
least some of the hay is hidden under the rocks. But, as the hay
piles up, a lot of it is visible by late summer. And that makes it
easy to analyze what pikas are caching. Its also easy to observe
pikas as they forage, so several studies have quantified foraging
behavior. And these studies suggest that pikas can be selective,
and selective behavior suggests that the type of available forage
may be very important. In a study of 15 sites throughout the
southern Rockies, Justine Smith found that pikas selected
especially for higher water and nitrogen content, and selectivity
increased as water and nitrogen content declined in the available
vegetation.**Heres the yellow-bellied marmot. The distribution of
the American pika in the lower 48 states is generally nested within
the distribution of this marmot, and the two species share similar
habitats. Heres one of my tagged pikas communing with a your
marmot. In fact, I often find pika haypiles right on top of marmot
dens, as shown in this photo by my colleague Shana Weber. I also
very often see pikas collecting marmot scat and placing it in their
haypiles, suggesting this is an important resource for pikas.
Marmots and pikas also share a similar set of predators, so they
might benefit from each others alarm calls. So far, my research
indicates this is more of a one-way street, with pikas using
marmots more than the other way around.*One of the fun things to do
is to set up automated cameras to see whos using who. I set these
up at pika haypiles and at marmot dens, and I find much more
evidence of pikas entering marmot dens than marmots visiting
haypiles. Heres one series taken at the entrance to a marmot den.
Hungry marmot emerges in the early light and soon returns with some
forage. Later in full sunlight a pika enters the den. This pika
entered this den many times during the week. Heres another shot of
her rump disappearing into the den in the afternoon.What they do in
there, I dont know, but they do it, which suggests its important.If
climate change alters the frequency of interaction between these
species, there may be important consequences for the pika. There
are already signs that climate change is altering the timing of
marmot hibernation.**The only predators that can follow pikas into
the rocks are weasels and martens. Both of these guys are emerging
empty-handed from pika haypiles.This pika wasnt so lucky, but this
is the only time Ive witnessed a weasel successful in capturing a
nearly adult pika. Generally, weasels take juveniles, and you can
really tell thats the case just by observing the behavior of adults
and juveniles when a weasel comes through. Adults tend to jump up
on a high rock to get the view, and just stay there giving alarm
calls. The calls let the weasel know where the pika is, but they
also let the weasel know that its been seen, and no longer has the
element of surprise. Juveniles tend to dive under the rocks in
response to the adult alarm calls. Weasels tend to keep searching
the rocks, ignoring the calling adults that are in plain view. When
a weasel does chase an adult into the rocks, the adult usually gets
away and pops up somewhere else, probably because it is so familiar
with its own territory. Juveniles are clearly not familiar with the
territorywhen theyre in a hurry to run they often bump into rocks
and run into dead-ends.So, if global warming ends up increasing the
frequency of pika encounters with these guys, recruitment may
decline. These guys also eat a variety of small mammals, so their
distributions may shift in response to the distribution ofsayvoles.
So it gets pretty complex.To my knowledge, theres very little
information on the density and distribution of these predators.
Personally, Ive noticed that in Montana the decline of the pine
martin over the past several decades appears to be turning around.
I didnt see a single pine martin for the first 15 years of my
study, but Ive seen them every few days for the past 3 years. So
something appears to be changing.*****
