In a paper (Marriott C A, Fisher J M, Hood K and Pakeman R J (2010) 'Impacts of extensive grazing and abandonment on grassland soils and productivity' Agriculture, Ecosystems & Environment Volume 139, Issue 4, 476–482) the abstract states: "Two long-term (16 year) experiments on intensively managed pastures compared extensive grazing, abandonment and continued intensive grazing and were assessed for impacts on soil parameters, plant nutrient content and ecological indicator values. There was a reduction in soil carbon and nitrogen in the abandoned treatment compared to the intensively managed treatment at the wetter site. "At the drier site, extensive grazing resulted in a build up of soil carbon. There was a build up of dead organic matter and a reduction in the nutritive value of the vegetation as grazing was reduced. Indicator values confirmed the reduced soil nutrients and a fall in site pH. There was also a rise in the dominance of plants preferring moist conditions, especially at the wetter site. "As biodiversity gains are small, the management of these systems could be seen as a trade-off between managing for production and for soil organic carbon. At the drier site this trade-off is apparent, whereas at the wetter site managing for production also maximises soil carbon content." A few extra paragraphs copied add clarification: "The overall aim of this analysis was to understand the long term (16 year) impacts of the removal of fertiliser inputs and reduced grazing/abandonment on site fertility and nutrient stocks through

1. monitoring of major soil chemical indices,
2. monitoring of biomass and biomass nutrient content and
3. monitoring of changes in plant indicator values related to soil properties.

"An experiment comparing sward responses in established ryegrass/white clover pastures under productive management and a range of more extensive management options was set up in 1990 at two upland sites in Scotland (Hartwood, a wetter site and Sourhope, a drier site). There were four treatments, each replicated twice, at each site:

1. productive management, i.e. fertilised annually in March and August with 50 kgNha−1 as ammonium nitrate and in May with a compound fertiliser supplying 40, 8.7, 16.6 kg N, P and k ha−1, and maintained at a sward surface height of 4 cm from March until mid-November (4F);
2. an unfertilised, grazed treatment maintained at 4 cm over the same period (4U);
3. an unfertilised, grazed treatment maintained at 8 cm over the same period (8U);
4. no fertiliser applications and no grazing (UN).

"Grazing was with Scottish Blackface ewes and single lambs (until weaning in mid-August). Since there were site differences in plant growth, the number of livestock required to maintain the sward height treatments varied. "The 4F treatment here reflects the usual management of ryegrass clover pastures, though more uniform in its application. This management ensures high productivity: at Hartwood the mean annual stocking rate was 1.84 LU ha−1 yr−1 and at Sourhope 1.60 LU ha−1 yr−1. It, however, required continual high inputs: 140 kg ha−1 yr−1 of N, 8.7 kg ha−1 yr−1 of P and 16.6 kg ha−1 yr−1 of K. "The result of this high input and high output management was little plant biodiversity: averaged over the period 2000–2005 only 8.8 species were recorded on the 4F treatment at Hartwood and 11.8 at Sourhope. If inputs are reduced then there is a reduction in production. Thus stocking rates on the 4U and 8U treatments at Hartwood were 1.25 and 0.76 LU ha−1 yr−1 and at Sourhope 1.04 and 0.61 LU ha−1 yr−1 respectively. "At Hartwood, extensification resulted in a drop in soil carbon stocks though no change in soil C:N ratio, though this is partly offset by higher storage of C in live and dead parts of the vegetation. This possibly indicates that soil carbon at this relatively wet site is dependent on short-term inputs from production. "At Sourhope, on the other hand, there is more carbon in the vegetation and under the extensively grazed treatments and a higher C:N ratio in the soil of the ungrazed treatment. At this drier site, the changes may reflect slower turnover under less intensive management as plant traits related to slower decomposition predominate, and a build up of litter in the ungrazed treatment that is slow to be incorporated into the soil. "Biodiversity gains in the extensively managed treatments were not large. Over the latter part of the experiment at Hartwood (2000–2005), there were 45% more species present on the 4U treatment, 48% on the 8U and 51% on the UN, though this started from a low base. At Sourhope the gains were smaller, 8%, 21% and 31% respectively, but they started from a higher base. However, there were shifts in species towards those more characteristic of nutrient poor conditions which can be seen as an increase in nature conservation value (Fig. 1c and g). "This pattern counters the continued increase in the nitrogen indicator value in national monitoring of vegetation over the same period.  For the type of grassland studied here it suggests that the biodiversity gains that result from extensive management are relatively modest. This suggests that unless there is a significant demand for restoring species-rich grassland, then management decisions may focus on the trade-off between productivity and soil carbon. However, within the two sites studied it appears that there is no need to trade-off at the wetter Hartwood site, as productivity and soil carbon are maximised in the conventional management treatment (4F). At the drier Sourhope site, it does appear that there is a potential trade-off: the most productive treatment has the lowest soil carbon and lowest soil C:N ratio. "Below ground C was positively correlated to grazing intensity at Hartwood but peaked in the extensively grazed treatments at Sourhope. Soil C might be expected to increase as input into the soil increases; i.e. with less biomass consumed by livestock. However, this effect appears to be slight and obscured by the impact of fertiliser inputs increasing productivity, and hence inputs, at Hartwood whereas at Sourhope, the build up of litter in the ungrazed treatment at this dry site may have prevented this treatment demonstrating higher soil C levels. "Above-ground C increased in all the grazing reduction treatments in proportion to the amount of stock removal. These patterns partly confirm previous findings that a suppression of N fertiliser supply and a strong reduction in grazing pressure appear not to offer any considerable increase in greenhouse gas (GHG) sinks strength in the short-term. Extensive grazing at Sourhope did increase soil C, but it did not at Hartwood. "Abrupt reductions in fertiliser and stocking levels appear not to be a universal mitigation strategy as productivity declines and the evidence suggests that soil carbon stocks and recalcitrance are little changed under the more extensive treatments. Soil N increased in all treatments at both sites and combined with the changes in soil C meant that there was no difference between treatments in soil C:N at Hartwood and a negative correlation between grazing intensity and C:N ratio at Sourhope." The full paper may be accessed here. (This is a pay service from ScienceDirect.)