Processes influencing soil carbon storage following afforestation of pasture with Pinus radiata at different stocking densities in New Zealand

Abstract

Since 1992, afforestation with Pinus radiata D. Don in New Zealand has led to the establishment of over 600 000 ha of new plantation forests, about 85% of which are on fertile pastures used previously for grazing sheep and cattle. While this leads to rapid accumulation of carbon (C) in vegetation, the effects of afforestation on soil C are poorly understood. We examined key soil C cycling processes at the (former) Tikitere agroforestry experimental site near Rotorua, New Zealand. In 1973, replicated stands of P. radiata (100 and 400 stems/ha) were established on pastures, while replicated pasture plots were maintained throughout the first 26-year rotation. In 1996, soil C and microbial biomass C in 0–0.10 m depth soil, in situ soil respiration and net N mineralisation, and soil temperature were lower in the forest than in the pasture, and tended to decline with increasing tree-stocking density. In the 400 stems/ha stands, mineral soil C (0–0.50 m depth) was lower than in the pasture (104 and 126 Mg C/ha, respectively; P < 0.01). Carbon accumulation in the forest floor during the first rotation of these forest stands was 12 Mg C/ha. Using the Rothamsted soil C model (Roth-C), we examined how changes in plant C inputs following afforestation might lead to changes in soil C content to 0.30 m depth. Steady-state pasture inputs of 9.0 Mg C/ha.year were estimated using Roth-C; these C inputs were assumed to decrease linearly during the first 12 years following tree establishment (until canopy closure). Below-ground C inputs in the forest were estimated using steady-state relationships between litterfall and soil respiration; these inputs were assumed to increase linearly between years 1 and 12, after which they remained constant at 1.53 Mg C/ha.year until harvest. Measured changes in soil C (0−0.30 m) during the first rotation, in conjunction with the below-ground inputs, were used to estimate above-ground inputs (as a proportion of total litterfall [3.81 Mg C/ha.year]) to the soil. Our results suggest 10% of litterfall C over one rotation actually entered the mineral soil. Using these results and estimates of additional C inputs to the soil from harvest slash and weeds following harvest, we found mineral-soil C stocks would continue to decline during second and third rotations of P. radiata; the magnitude of this decline depended in part on how much slash enters the mineral soil matrix. We confirmed our modelling approach by simulating soil C changes to within 8% over 19 years following afforestation of pasture at another previously studied site, Purukohukohu. Whether afforestation leads to an increase or decrease in mineral-soil C may depend on previous pasture management; in highly productive pastures, high C inputs to the soil may maintain soil C at levels that cannot be sustained when trees are planted onto these grasslands.