Nils Chr. Stenseth, Centre for Ecological and Evolutionary Synthesis (CEES), Department of Biology, University of Oslo, PO Box 1066 Blindern, N-0316 Oslo, Norway.
1Animals are affected by local weather variables such as temperature, rainfall and snow. However, large-scale climate indices such as the North Atlantic Oscillation (NAO) often outperform local weather variables when it comes to explain climate-related variation in life history traits or animal numbers.
2In a recent paper, Hallet et al. (2004, Nature, 430, 71–75) document convincingly why this may happen. In this perspective, we identify from the literature three mechanisms why this is so: (1) the time window; (2) the spatial window; and (3) the weather composition component of climate.
3Such an understanding may be used to derive even better ‘weather packages’ than the NAO.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Any sensible biologist, especially those getting their feet wet and cold while working in the field, would intuitively use covariates such as temperature, precipitation and snow depth whenever the aim is to deduce the climate-related variation in life history traits or animal numbers. No doubt, these are the climatic factors giving animals a good or hard time. Nevertheless, in a recent Nature paper, Hallett et al. 2004) demonstrate convincingly why global scale climate indices such as the North Atlantic Oscillation (NAO) index may better explain animal performances than the local weather variables biologists often select.
In 1997 (Post et al. 1997), the winter index for the NAO (Hurrell et al. 2003) – the sea surface pressure difference between the Azores and Iceland for months December through March – was, for the first time, used to explain variation in performance of red deer along the west coast of Norway. At first, the position that a large-scale climate proxy such as the NAO should be a better predictor of ecological phenomena in terrestrial ecosystems than local weather variables, such as precipitation and temperature was met by much scepticism. A follow-up study including some 16 populations of seven species of large herbivores showed that the NAO predicted variation in life history traits and population dynamics better than local climate in about half of the cases (Post & Stenseth 1999). It became difficult to question the predictive power of the NAO. The NAO subsequently became quite popular in ecological studies focusing on how ecology might be affected by climate variation and possibly climate change (Ottersen et al. 2001; Stenseth et al. 2002; Mysterud et al. 2003).
The time window component
What Hallett et al. (2004) have done is to convincingly demonstrate why the NAO index sometimes better explain animal performance than local weather variables – using the Soay sheep Ovis aries from Hirta in the St Kilda archipelago as an example. At high population density, the Soay sheep die in large numbers during periods of severe rainfall in winter due to starvation, caused, in part, by energetic constraints imposed by harsh weather (Coulson et al. 2001). Hallett and coworkers demonstrate that as previous studies used monthly values of rainfall (Catchpole et al. 2000), which is the most common temporal scale used (Weladji et al. 2002), they fail to capture that such incidences of severe rainfall may happen during the entire winter. In contrast, as the NAO is indexed for the entire winter (December–March), the NAO index is a fairly good predictor of whether such a severe spell is coming this winter or not (Shabbar & Bonsal 2003). This is the time window component of climate (Hallett et al. 2004; Fig. 1). Indeed, the temporal scale provides an important feature of the link of climate and ecology, and as such explains why indices like the NAO may work better than local variables.
The spatial window component
Earlier we have reported another feature explaining why the NAO give a better predictive power than single site climate variables (Fig. 2). As the NAO is correlated with both increased precipitation and temperature along the west coast of Norway, the effect of the NAO on snow depth will vary according to altitude in heterogeneous landscapes, as altitude is a main determinant of whether precipitation comes as rain or snow (Mysterud et al. 2000). High phases of the NAO are thus correlated with less snow at low altitude, and more snow at high altitude. The same principle also applies to increasing latitude. This large-scale geographical feature of the NAO and similar indices explain why performance of migratory deer increase during high NAO values due to less snow in winter areas at low elevation, while more snow in high altitude summer ranges deer are offering deer better quality summer ranges due to an extended period of snow melt (Mysterud et al. 2001). We suggest to call this the spatial window component of climate (Mysterud et al. 2000, 2001).
The weather composition component
Another feature of the NAO and similar indices is that they incorporate the variation in several climate variables – temperature, precipitation, snow depth, wind speed, etc. It may not be that an animal is responding only to (for example) snow depth or cold temperature. Already in 1924, Elton (1924) noted that ‘A good climate, like a good dinner, is more than the mere sum of its separate parts’. This weather-feature integrating aspect relates to what we suggest to call the weather composition component of climate (Elton 1924; Stenseth et al. 2003; Mysterud et al. 2003; Hallett et al. 2004). This highlights that it typically is not only one component of the weather that is an important determinant of the ecological processes of interest, but rather an entire package (including their intrinsic interactions) that is relevant to animal performance. In the case of the Soay sheep, increased mortality might be due to high rainfall, high winds or low temperatures or a combination of these (which all correlate with the NAO on St Kilda), as they all put a severe cost on the energy budget of the sheep (Fig. 3; Hallett et al. 2004). Hallett et al. (2004) is also the first convincing empirical demonstration of this principle. Typically, extreme cold or severe rainfall combined with wind is the more difficult for the animals to cope with (Moen 1973).
Discussion and conclusions
It is an empirical fact that indices such as the NAO often explains more ecological variation than single weather components. We suggest this is due to the fact that the NAO combines – what turns out to be – appropriate temporal and spatial features of several weather components (all of which being appropriately integrated). We are convinced that these are the features making the NAO (and similar large-scale climate indices) an appropriate ‘weather package’ applicable for several ecological studies (Mysterud et al. 2003; Stenseth et al. 2003). The study by Hallett et al. (2004) provide one important step towards an understanding of why this is so.
In our further efforts to understand how climate affect ecological processes, it is important that we take account of the fact that such climate-driven ecological effects typically are quite complex: organisms move over space (making the spatial window component of climate important); organisms are generally affected over a period of time (making the time window component of climate important); and organisms are typically affected by more than one feature of climate (making the weather composition component of climate important). This is why no single or a simple combination of a few meteorological features often convey less information than a package of weather such as the NAO, an index that empirical observations now demonstrate that integrate these three components (or dimensions) such as to convey ecological interpretable signals. Other packages of weather might certainly outperform the NAO and similar indices. However, to derive such new and better weather packages, much detailed understanding of the system in question is needed. For this reason more mechanistic approaches to understanding variation in demographic rates and how these are influenced by climate is highly recommended.