Range of Natural Variability

Range of Natural Variability (RNV) can be computed for most ecological variables. The purpose of computing RNV is to provide stakeholders with an empirical estimation of the inter-annual variation in indicator performance occurring in a world without industrial activity but driven by natural disturbance regimes (fire, climate, insect outbreaks). The “driving” variables that can be allowed to vary stochastically throughout RNV simulations include:

• average annual temperature;
• average annual precipitation; and
• area burned by fire

The “natural” drivers of fire, insect outbreaks and climate can influence a suite of landscape metrics (landscape fragmentation, forest age and core area, water flow dynamics, sediment movement, for example) being tracked in ALCES®. These landscape metrics can also be affected by anthropogenic drivers (forestry, energy, agriculture, settlements, etc.). In turn, these landscape metrics are key input variables to most ecological indicators being tracked in ALCES®.

When only stochastic natural disturbance regimes are operating on a landscape, ALCES generates the RNV for each ecological indicator. Displaying the RNV band when simulating an indicator into the future allows the user to assess the level of departure of performance of an indicator relative to its “pre-industrial” reference point. It is not implicit in this approach that ecological indicators should remain within RNV, although it can be logically argued that risk to maintenance of a indicator increases as its status moves further away (above or below) a defined RNV envelope.

It is common in ALCES projects for stakeholders to define levels of performance, relative to RNV, that suggest different levels of risk to indicators. For example, indicator performance within 25% of RNV may be considered to be acceptable and assigned a low risk factor of “green”. Indicator values between 25-75% below RNV could be assigned a moderate risk factor of “yellow”, and values ranging between 75-100% below RNV could be assigned the highest risk factor of “red”. It is the role of scientists using ALCES to inform stakeholders about systems dynamics and relative change in risk, but is generally the responsibility of stakeholders, or policy makers, to assign thresholds depicting tolerance to risk for given indicators.

Ecosystem attributes vary naturally in response to fluctuations in natural disturbance and climate. The dominant natural disturbance agent in Canada’s boreal forest is fire, and variable fire rates over time and space create a dynamic patchwork of forest types that provide habitat for a diverse assemblage of species. Variation in precipitation is influential in aquatic ecosystems, causing fluctuations in water and nutrient levels that, in turn, control habitat for aquatic species.

The range of natural variability can be estimated for environmental indicators to provide a baseline from which to interpret projected future conditions associated with land use scenarios. Following the approach recommended by Schieck (2009), range of natural variability can be estimated by conducting 100 independent runs in ALCES® that incorporated known ranges of variation in wildfires, precipitation and precipitation. Fire, precipitation and temperature can be simulated as stochastic processes such that fire rate, rainfall and temperature vary across years. Fire rate (percent of forest landscape burned) in each year can be simulated as a random draw from a lognormal distribution with a specified mean. Annual precipitation can be simulated as a random draw from a normal distribution with a specified mean and standard deviation.

An indicator’s range of natural variation equals the interval defined by the minimum and maximum value of the indicator during the last 100 years of a 200-year simulation. The first 100 years should not be considered so that the legacy of historical land use on forest age-class did not influence the range of natural variation .  Upper and lower bounds of the estimated ranges of natural variation can subsequently be used to scale land use simulation results to percent departure from natural.