Scientific foundations for boreal conservation

我提出了加拿大北方森林保护规划和可持续管理的三个挑战：预测未来的火灾状况，预测森林的未来状态，开发计算机化工具或模拟模型，以在一到两个世纪的大范围和时间跨度内进行这些预测。******保护区网络的设计考虑了我们对火情的理解，即对特定地区典型火灾的数量、大小和其他特征的统计描述。预测气候和森林结构可能与今天不同的未来是困难的，因为我们的历史信息是不完整的。在加拿大的许多地方，直到最近，大多数小火灾都没有被发现。这种“大小偏差抽样”可能导致气候和植被对结果影响的模型严重错误。同样，灭火效应也可能会影响我们的模型，同时在重要方面改变森林。统计方法可以同时考虑所有这些因素，从而得出可以做出可靠预测的无偏模型。******森林动态模型经常使用分类卫星数据来描述森林的初始状态，如年老或年轻、落叶或针叶林。显然，相同的位置不会永远保持相同的状态。随着时间的推移，森林会通过“演替”和再生而发生变化，例如在火灾之后。这些过程可以概括为可以用计算机模拟的“状态转换规则”；不幸的是，我们对这些规则并没有非常清楚的了解，迄今为止，我们更多地依赖于专家意见，而不是定量分析来为计算机模型设定参数。现在有可能做得更好，因为我们现在可以为整个加拿大建立相当长的时间序列卫星图像，例如1990年，2000年和2010年。这是一个足够长的时期，因此应该可以通过仔细比较三幅图来估计和模拟过渡概率。随着管理和火灾继续在北方地区展开，这将使人们能够更可靠地预测未来的森林状况。这样做是本建议的第二步。******该提案的第三部分包括将前两部分的结果与大量外部努力的结果综合起来，这些努力包括模拟森林鸣禽、水禽和北美驯鹿的丰度，以及保护区规划。这种综合将通过继续发展一些已建立的建模工具来实现；目标是预测气候、森林管理和保护活动将如何影响22世纪北方森林及其野生动物的未来状态。这样的工具将使我们能够着眼未来进行管理，并尽可能实现加拿大人所希望的结果。

I address three challenges for conservation planning and sustainable management of Canada's boreal forests: predicting future fire regimes, predicting the future state of the forest, and developing computerized tools, or simulation models, to make these predictions over large areas and time spans of one or two centuries. ******Protected areas network designs take account of our understanding of fire regime, that is, of statistical descriptions of the number, size, and other characteristics of the fires typical of a given area. Forecasts into a future where climate and forest structure may be different than they are today is difficult, because our historical information in incomplete. In many parts of Canada, up until very recently, most small fires were not detected. This "size-biased sampling" can lead to seriously incorrect models of the effects of climate and vegetation on the outcomes of interest. Similarly, fire suppression effects may also bias our models, while at the same time changing the forest in important ways. Statistical methods can account for all these factors simultaneously, leading to unbiased models that can make reliable forecasts. ******Models of forest dynamics often use classified satellite data to describe the initial state of the forest, as being old or young, deciduous or coniferous. Obviously, the same location does not stay in the same state forever. Forests change over time through "succession" and regeneration, for example after a fire. These processes can be summarised as "state transition rules" that can be simulated by computer; unfortunately, we do not have a very clear understanding of what the rules are, and to date have relied more on expert opinion than on quantitative analysis to set parameters for computer models. It is now possible to do better, because we can now build reasonably long time series of satellite imagery for all of Canada, e.g. for the years 1990, 2000, and 2010. That is a long enough period so that it should be possible to estimate and model transition probabilities by careful comparison of the three maps. This would enable more reliable projections of future forest conditions as management and fire continue to unfold across the boreal. Doing so forms the second leg of this proposal.******The third part of this proposal involves synthesising the results of the previous two parts with the results of substantial external efforts in modelling forest songbirds, waterfowl and caribou abundances, and in protected areas planning. This synthesis will take place by continuing the development of some established modelling tools; the objective is to project how climate, forest management and conservation activities will affect the future state of the boreal forest and its wildlife into the 22nd century. Tools like this will enable us to manage with the future in mind and to achieve, as far as possible, the outcomes preferred by Canadians.