Search

Understanding long-term woodland dynamics is regarded as essential to investigating the legacy of past human actions and adequately conserving woodland habitats of higher ecological and cultural value. In Scotland, direct evidence of woodland extent and management is usually scarce and incomplete before the mid-nineteenth century First Edition Ordnance Survey maps (OS). For this PhD thesis, historical plans of private estates, underutilised in the UK, are proposed to fill this gap. Although each of these plans covers only a small area, they were drawn at large scale, permitting a detailed and accurate depiction of woodland prior to the First Edition OS. This research collates 352 Scottish estate plans, dated c.1740 to c.1835, from both private and public collections. Hundreds of these plans are privately-owned and had been previously unavailable to researchers. A GIS-based method is implemented to integrate the spatial information of a large variety of historical plans into a homogenous database. The consistency in the depiction, accuracy and reliability of estate plans enables spatially explicit reconstructions of the woodland cover extent for two time series, namely T1 (1740-1799) and T2 (1801-1833). These reconstructions cover a total of 107,700 ha in Nithsdale and Annandale (Dumfries and Galloway) and can be compared with the First Edition OS maps (i.e. T3, c.1860) and modern data (i.e. T4, 2014). The uncertainties resulting from the challenges of working on a large variety of plans drawn by different mapmakers are also assessed. This assessment uses a conceptual framework that explores in a chronological manner the uncertainty arisen from the estate plans production to their practical use for research on woodland cover changes. Quantitative analysis based on the reconstructions shows a marked and consistent growth in woodland cover during the eighteenth and nineteenth centuries. While the woodland covered about 3% of the study area in T1, it increased to 4.5% in T2, and to between 6.5 and 8.5% in T3. The lowest coverage in the study area occurred, at the latest, some time in the mid-eighteenth century with an upper estimate of 2.5% of woodland coverage. Although it is not possible to determine to what extent the trends observed in the study area apply to the rest of Scotland, this research casts doubt on the methods and assumptions that may have led previous research to overestimate the amount of past woodland cover for the whole Scotland. Change detection analysis allowed mapping and quantification of where the woodland was new, lost or extant between two time series. Trajectory analysis enabled the tracking, mapping and categorising of the various historical trajectories of present-day broadleaved woodland since the eighteenth century. Along with the study of various woodland metrics, these spatial analyses underline how the present-day woodland cover has been progressively shaped by plantations and clearance at different spatio-temporal rates. In addition, a modelling approach using binary logistic regression allows a better understanding of past woodland distribution and changes in relation to landscape physical contexts such as slope steepness, elevation, soils and distance to the nearest streams. Hence, these models highlight changes in woodland plantation practices over time and space. Other historical sources, including the first and second Statistical Accounts of Scotland (1791-1845), offer a complementary insight into tree composition of past woodland cover, management practices and the reasons that encouraged planting woodland. These sources indicate how the temporal resolution of the mapping reconstruction may still underestimate the magnitude of past woodland changes. Estate plans can also provide a rare insight into the history and ecology of the so-called 'ancient woodland sites'. Defined in Scotland as continuously wooded since c.1750, these sites have been compiled in the Scottish Ancient Woodland Inventory (AWI) using evidence from official historical mapping, namely the Roy map (c.1750) and First Edition OS mapping (c.1860). At present, ancient woodlands are recognised as areas of greater ecological and cultural value that deserve priority for UK woodland conservation. A comparison between the Scottish AWI and the woodland reconstructions in T1 and T2 indicates that the former is largely inaccurate. The amount of 'pseudo-ancient woodland' - of more recent origin than expected in the AWI - is estimated to reach at least 40% of the woodland compiled in the AWI. This discrepancy is discussed in the light of a new assessment of spatial accuracy of the Roy map and First Edition OS maps. In addition, the logistic regression models of past woodland cover can help distinguishing the pseudo-ancient woodland from the probable ancient woodland in the AWI. Using evidence from estate plans, 41 native woodland sites of different continuity classes were selected for botanical survey. Although ancient woodland sites are likely to exhibit more species supposed to be 'indicators' of ancient woodland, several recent woodland sites are very similar in plant assemblage to those of ancient woodlands. The occurrence of many indicator species of ancient woodland in non-ancient woodland habitats indicates how rapidly these species can establish themselves in recent woodlands, helping to blur the distinction between woodland habitats of different continuities. While critically assessing the current criteria for defining ancient woodland and the applicability of those criteria to woodland conservation, this study suggests that, depending on the connectivity to ancient woodland, environmental conditions and history of woodland, relatively recent plantations (i.e. plantations made after c.1750) could deserve the same recognition as ancient woodland for conservation.