An important feature of our simulations with the HadAM3 model is that we use a Biome model (BIOME 4) to translate climate model output from the Pliocene control experiment, (specifically the seasonal cycle of temperature, precipitation, and solar radiation) into biome distributions. Such an approach has two specific advantages. The first is that it condenses multivariate climate data into a small number of biomes producing an effective and informative summary of the middle Pliocene simulations produced by the HadAM3 model. Secondly, it greatly improves our ability to compare model-simulated climate with estimations of palaeoclimate derived from fossil pollen. In this study we use the BIOME 4 model, which is the latest in a series of mechanistically based models that were developed from physiological considerations that place constraints on the growth and regeneration of different plant functional types. These constraints were calculated via the use of limiting factors for plant growth. These factors include the mean temperature of the coldest and warmest months, the number of growing degree days (GDD) above 0º and 5º C, and via the calculation of a coefficient (Priestley – Taylor coefficient) for the extent to which soil moisture supply satisfies atmospheric moisture demand. GDD are calculated by linear interpolation between mid-months and by a one-layer soil moisture balance model (Prentice et al. 1993) independent of the HadAM3 model hydrology.

The BIOME 4 model is representative of a gradual evolution of biome models (see Prentice et al. 1992; Prentice et al. 1993; Prentice et al. 1998) and differs from earlier BIOME models by the addition of additional plant functional types for tundra climates and by factoring in the importance that differing atmospheric CO₂ concentrations have on plant growth (Kaplan 2001). To minimise any model bias,the BIOME 4 model, like its parent models, has been validated and calibrated to vegetation distribution, productivity and other biogeochemical data for the present-day and the recent past. Known biases in the BIOME 4 model relate to the precise location of the forest-grassland boundary in temperate and subtropical latitudes (Kaplan 2001).

As an equilibrium vegetation model, BIOME 4 is an ideal tool for assessing vegetation distribution at a particular time in the past. However, plant assemblages heavily influenced by short-term dynamic processes (e.g., fire, recurrent extreme weather events) may be incorrectly simulated in model. In this application, we have run BIOME 4 using a 2.5º by 3.75º grid corresponding to the same resolution as the HadAM3 GCM. Absolute values for monthly mean surface temperature, precipitation, and cloud cover derived from the ten-year climatological means from the HadAM3 Pliocene control run were used to provide the climatic information necessary for the BIOME 4 model.