This work is based on an excellent PhD thesis by Matthew Carroll (2011) who conducted his PhD work at York (Biology department) as part of a CASE PhD with the RSPB (supervisors were Prof Chris Thomas at York and Dr James Pierce-Higgins at the RSPB).

During his PhD the MILLENNIA model (Heinemeyer et al., 2010) was updated to allow a monthly time step (Carroll et al., 2015); water table outputs form the model were compared to three UK blanket bog sites and used to predict soil moisture across large catchments for those four upland areas (Wales; Peak District; North York Moors) and the model was also validated against water table data for Moor House (NNR; UK). The model performed well for all sites (compared to site water table and peat depth measurements; see slide (1)) but required to consider peat development period (i.e. time since peat initiation). This varied between sites and peat initiation times were partly based on data by Tallis (1991). The model code also allows to consider topographic effects (i.e. aspect and slope) on temperature (and thus evaporation and decomposition processes) as well as erosion through runoff.

The predicted soil moisture predictions were based on a relationship between observed soil moisture and predicted water tables (see slide (2)) and were then used to predict crane fly emergence in the following year based on field relationships of cranefly emergence with soil moisture (see slide (3)). Based on those contemporary relationships we extrapolated cranefly abundance across the landscape and into the future (slide (4)) using climate change scenarios for the UK (UKCIP). We kindly acknowledge the use of the UK's ECN data for Moor House (i.e. water table depths and climate data to enable the hydrological model validation).

The above images were taken from a talk given by A. Heinemeyer at the BES conference in Bangor 2012: "Modelling past, present and future UK upland peatland carbon dynamics and implications for restoration projects" ANDREAS HEINEMEYER, M.J. CARROLL, C.D. THOMAS, A.R.M. HANLON; also see the actual paper (Carroll et al., 2015). In this paper the impact of peatland restoration via blocking peat drainage channels (so called grips) was investigated, linking water tables to soil moisture and cranefly emergence and thus bird populations (breeding success is linked to chicks having access to plenty of cranefly larvae, which thrive in wet but die in dry summer conditions). The project considered MILLENNIA model scenarios until 2080 for three UK upland areas: Wales (wet), Peak District (intermediate) and North York Moors (dry).

In the project we also modelled cranefly abundance and subsequent impacts on bird populations. For this we considered WTD affecting peat soil moisture, which then regulates cranefly emergence and thus abundance as a food source for certain upland birds (i.e. dunlin, golden plover and red grouse). For this we used actual field measurements of soil moisture versus water table depth per management and our moisture relationship versus cranefly abundance, which was very similar to that observed by Carroll et al., 2015. We then used Carroll et al. (2011, 2015) equations to relate cranefly numbers to bird populations. The MILLENNIA model (as applied in Carroll et al., 2015) allowed predicting future water table depth and this soil moisture under different management scenarios, linking measurements to model predictions. A major finding was that soil moisture was higher under mowing with leaving brush and lowest on burnt plots. As cranefly numbers show a positive relationship to soil moisture, this resulted in more resilient bird numbers under future drier climate, particularly at the driest site, Nidderdale.

Above from left to right:

Water table depth: for uncut management scenario at the three sites: Annual mean water table depth (WTD) predictions for the 5 x 5 km squares for each site for (left) the current baseline (1961-1990) and (right) future climate (2051-2080) for the cranefly scenarios. Whitendale areas under 200 m a.s.l. were not included in the model as they do not support observable blanket bog peatland habitat.

Peat soil moisture versus water table depth per management: Relationships between soil moisture (SM, %) and water table depth (cm) across all sites, and years for each (plot) management. Days with significant amounts of rainfall over the previous 3 days were removed. Regression equations were:

• Mown (brash left): SM = 1.00 – 0.000758(WTD) – 0.000197(WTD2), R2 = 0.557, P< 0.001;
• Mown brash removed (Brash): SM = 1.00 – 0.000903(WTD) – 0.000275(WTD2), R2 = 0.550, P< 0.001;
• Uncut (Do Nothing): SM = 1.00 – 0.00256(WTD) – 0.000165(WTD2), R2 = 0.411, P< 0.001;
• Burnt: SM = 1.00 – 0.00342(WTD) – 0.000205(WTD2), R2 = 0.400, P< 0.001.

Cranefly number predictions: Projected average total annual cranefly abundance (individuals per m²) in the 5 x 5 km square surrounding the Mossdale field site under (left) baseline (1961-1990) and (right) future (2051-2080) climate scenarios with different methods of habitat management (mowing with brash left, mowing with brash removed, no management (uncut) and burning).

Final bird model predictions: Projected average annual golden plover abundance (individuals per km²), and the overall mean, within the 5 x 5 km square surrounding the three sites under baseline (1961-1990) and future (2051-2080) climate scenarios with different methods of habitat management (mowing with brash left, mowing with brash removed, no management/uncut and burning). White areas are those under 250 m a.s.l. that would not support the moorland habitat that is the focus of this study. Predictions are based on the cranefly abundance estimates and relationship between cranefly abundance and bird abundance described by Carroll et al. (2015).