Land transformation for the period 1944–1999 was quantified according to a set of land categories (Table 1) based on a refinement of a physiognomic–structural vegetation classification system incorporating growth form, stratification (height), projected canopy cover and crown-to-gap ratio (Edwards 1983:705–712). Hard-copy panchromatic aerial photographs of the catchment in 1944 (first time flown) and 1996 (most recent at time of study), as well as the corresponding orthophotographs compiled from aerial photographs taken in 1976 and 1981, were obtained. Areas occupied by the land categories (Table 1) that were present in 1944 and 1996 were delineated on tracing film overlaid on the orthophotographs of 1976 and 1981. Areas mapped using aerial photographs from 1996 were updated to 1999 by field verification, justified owing to the lack of change over that time. The tracings were then digitised and land area statistics were extracted using the GIS package ArcView 3.0. The poor quality of the 1944 aerial photographs precluded the use of geo-referencing, orthocorrecting and on-screen digitising.

GIS databases of property ownership as in 1944 and 1999 were created for the area based on title deed searches at the Deeds Office, Surveyor General, Pietermaritzburg.

We sought to understand the implications of land transformation on biodiversity integrity and hydrological functioning, as relevant to aquatic biodiversity, based on the approach of O’Connor and Kuyler (2009:387). Their study determined the relative impact of different land uses on biodiversity integrity and hydrological functioning for the moist Grassland Biome based on the analytic hierarchy process (Saaty 1990:9–26). Experts were interviewed to score 37 indicators across a predefined set of land uses with regard to the effect on ecosystem composition, structure and functioning. Hydrological functioning, a component of ecosystem functioning, was scored as an integrated effect of a land use on the amount, quality and seasonality of water flow.

The scores from each reviewer for a specific indicator are integrated into a single weight, which quantifies the effect of each land use per unit area for the indicator. From these, an integrated weight across all indicators is derived to express the effect of land use on biodiversity integrity per unit area. The overall effect of a particular land use is then determined as the product of this weight and the proportion of the catchment area covered by that land use. The weights for land use provided by O’Connor and Kuyler (2009:390) are directly applicable to this study, as the area was part of their broader study area. Specifically, a natural asset was accorded a weight of zero as its value per unit area should not change. Plantations and alien plants (infestations, habitation) were accorded the value for plantation forestry, commercial cropping was accorded the mean of dryland and irrigated cropping, subsistence cropping was accorded the value for rural settlement, and fairly irreversible transformations were accorded the value for an urban environment (see Results).

Values represent a change in area of each land category between 1944 and 1999 as a percentage of the area occupied in 1944. The most notable land transformation was the loss of grassland (54% decrease) to commercial timber plantations (427% increase) and commercial agricultural cropping (311% increase) (Table 1; Figures 2 and 3). In contrast, subsistence cultivation had almost disappeared (98% decrease). Aquatic systems were transformed through an increase in the number and size of farm dams (2503% increase). Forest and woodland, although a minor part of the catchment, increased (4% and 22%, respectively).

The number of registered properties increased by 61%, with mean property size decreasing by 41% (Table 2).

In 1944, land was owned predominantly by private individuals, with private trusts or trustees and state-owned land accounting for the remainder (Table 3; Figure 4). By 1999, however, there were 10 additional categories of ownership (Figure 5). Although private individual ownership remained the largest category, more than half the catchment area had become the property of private businesses (31%) and commercial timber plantation companies (26%).

The loss of grassland to commercial timber plantations and agricultural cropping resulted in the biodiversity integrity and hydrological functioning index deteriorating by 326% and 166%, respectively (Table 4). The relative change in area of other land covers, whether positive (such as an increase in types of natural asset) or negative (such as an increase in the area affected by alien plants), was too small to influence the overall trend. However, the potential future threat of alien plants should be recognised on the basis of their rapid rate of establishment and consequent increase (Table 1). The increase in commercial timber plantations and agricultural cropping has had a pronounced effect on all three of the main components of biodiversity integrity. Landscape composition is changed extensively as a result of the extent of land transformation. Loss of habitat and the proliferating invasion of alien invasive plants, which affect indigenous species negatively, are the result of such transformation. Landscape structure is substantially impaired by transformation, due mostly to commercial timber operations, with an associated negative effect imposed on the extent, porosity, connectivity and geometry of fragments.

Landscape functioning is impaired through altered regimes of fire and grazing, accelerated soil erosion associated with perturbed bio-geochemical processes, including reduced carbon storage, and altered patterns of hydrological functioning. The latter is impaired through impacts on the amount of water and its seasonality of flow, in particular the reduction of base flows during the dry season, which are critical for aquatic biodiversity. In the case of commercial agricultural cropping, water quality (independent of soil erosion) is also severely adversely affected.

The most notable land transformation in the Karkloof catchment between 1944 and 1999 is grassland (54%) being lost to commercial timber plantations and agricultural cropping. The area of grassland within the catchment prior to colonial settlement was 334 km² (Weyer 2000:57, 59). It is estimated that 63% of this had been lost by 1999. Mucina and Rutherford (2006:423) and Jewitt (2011:11) estimate similar losses (50% and 76%, respectively) for the Midlands Mistbelt Grassland, which extends beyond the Karkloof catchment. These estimates are based on the assumed potential for indigenous vegetation if human influence on vegetation was removed and on outlines of remnant vegetation determined for the period 2004–2006 (Mucina & Rutherford 2006:15) and 2008 (Jewitt 2011:11). Estimates for transformation of the Grassland Biome overall are 26% between 1988 and 2000 (Fairbanks et al. 2000:77) and 47% for 1994–1995 (Neke & Du Plessis 2004:472). The loss of grassland in the Karkloof catchment therefore parallels the general trend in the moist eastern portion of the Grassland Biome of South Africa. The remaining grasslands are particularly suitable for commercial timber plantations and agricultural cropping owing to a favourable climate and fertile soils and are therefore highly threatened.

In the 55-year period reviewed in this study, the combined area of forest and woodland did not decrease as it did between 1880 and 1942 (Rycroft 1944:20), but rather increased by 3.02 km². This increase is ascribed to a number of influences. In 1944, subsistence cultivation was practised in areas within and adjacent to the indigenous forest. When these areas were abandoned, they were re-colonised by indigenous woody vegetation. Expansion of forest and woodland in the eastern portion of South Africa is constrained by fire (Bond, Midgley & Woodward 2003:80). Discussions with land owners revealed that grasslands are burnt less frequently than they were in 1944. A reduction in fire frequency may account for the increase in indigenous woody cover (O’Connor, Puttick & Hoffman 2014:75). In contrast to our findings, Lawes, Macfarlane and Eeley (2004:620) reported a decrease of 5.7% in the area covered by the Karkloof–Balgowan forest archipelago in the midlands of KwaZulu-Natal between 1944 and 1996. This difference may be due to differences in mapping methodologies and category interpretations and a larger study area being used in their study.

The total area of alien invasive plants increased by 7.71 km² (397%). Alien invasive vegetation poses a major threat to indigenous grassland and forest of eastern South Africa and their associated biodiversity and uses more water than indigenous vegetation (Driver et al. 2012:135). Commercial timber plantations similarly may cause an estimated mean annual stream flow reduction of 3% and a low-flow (dry-season flow) reduction of 8% (Scott, Le Maitre & Fairbanks 1998:193). By 1999, 36% of the total catchment area was under commercial timber plantations (Table 1; Figure 3).

The number of properties increased and the mean size of a property decreased between 1944 and 1999, as similarly recorded by Scotney (1970:185) and Rivers-Moore (1997:35) for adjacent regions. This may be attributed to population growth in South Africa, which increased fourfold in the reviewed period, from 11 415 000 people in 1946 to 43 054 000 in 1999 (Statistics South Africa 2002:8, 10). This may have driven the subdivision of large properties. Furthermore, in 1944 no legislation existed to regulate farm size and it was only with the promulgation of the Subdivision of Agricultural Land Act (Act 70 of 1970) that the indiscriminate subdivision of land was halted. (This act is still in force, but will be repealed once the Draft Preservation and Development of Agricultural Land Framework Bill, as gazetted on the 13 March 2015, is enacted.) Smaller property sizes have implications for reduced biodiversity and hydrological functioning, as farmers are forced to intensify agricultural production to remain economically viable. The aerial photographs of 1996 show clear examples of commercial timber plantations abutting directly onto indigenous forests and commercial cultivation encroaching into wetland areas.

There is a distinct difference between registered farm sizes and managed operational farm sizes. Although mean property size in the catchment generally decreased, the shift from private ownership towards corporate forestry and business entities (agricultural cropping farms and some corporate forestry) suggests that, by 1999, a large proportion of the small properties were farmed as large, consolidated land holdings, with land use and management being influenced accordingly. The economic difficulties of farming small rangeland cattle and agricultural cropping farms that existed prior to 1970 may have precipitated the move towards corporate business ownership of farms. Other economic factors in the late 1970s and early 1980s also had an influence. Worldwide demand for soluble pulp peaked in the 1980s and, in response, two large corporate timber companies began purchasing large tracts of land from as early as 1950, intensifying from 1986 to 1990 (Cairns 2000:7). This is supported by title deed searches, which show ownership transfers taking place in the 1980s. In the 1980s, beef prices were low and many cattle farmers in the catchment sold their land, seeing it as an opportunity for economic survival. The farms sold comprised mostly grassland and at that time no environmental legislation existed to control the potential environmental impacts of development changes and the conversion of natural land cover. Although an environmental clause is included in the Bill of Rights of the Constitution of South Africa, the first specific protective environmental legislation was the Environment Conservation Act 73 of 1989. This has been followed by the National Environmental Management Act (Act 107 of 1998) and the most recent Environmental Impact Assessment regulations of 2014, which afford current protection.

The increase in the number of farm dams in the catchment (Table 1) is directly related to the increase in the number of properties. The increase in dam size can be attributed to the need to optimise or intensify land use on the smaller property sizes. In the adjacent Midmar catchment, reservoirs upstream of the Midmar Dam could reduce median annual streamflow into the dam by 6% (Tarboton & Schulze 1991:229). Increases in the number and area of farm dams in the Karkloof catchment may have a similar effect on water availability in the uMngeni catchment.

Loss of grassland has affected a number of grassland types and parallels similar losses noted elsewhere (e.g. Rivers-Moore 1997:35). Such losses have been sufficient to render some grassland types of conservation concern in South Africa (Mucina & Rutherford 2006:362). Midlands Mistbelt Grassland (53% transformed) is endangered, whereas Ngongoni Veld (39% transformed), Mooi River Highland Grassland (24% transformed) and KwaZulu-Natal Hinterland Thornveld (22% transformed) are classified as vulnerable. In contrast, Drakensberg Foothill Moist Grassland (18% transformed) is classified as least threatened (Mucina & Rutherford 2006:423, 511, 422, 510, 424). Under the National Forest Act (Act 84 of 1998 as amended) forests are afforded protection in South Africa and, consequently, Southern Mistbelt Forest (5% transformed) is classified as least threatened, concordant with the slight increase observed in the Karkloof catchment (Table 1).

The Karkloof catchment is home to a number of red data species (Appendix 1), including 10 plant, 21 bird, two amphibian, one reptile and 11 mammal species (excluding bats). The majority of these are associated with grassland, for which loss of habitat due to land transformation is one of the main threats to their persistence. Six of the ten plant species are grassland species. Of these, five occur in the remaining areas of Midlands Mistbelt Grassland, with three being classified as vulnerable. In contrast, harvesting presents the primary threat to two forest red data plant species, as their habitat has remained secure (Table 1). The vulnerable aquatic species Hydrostachys polymorpha (rivers) and Nerine pancratioides (wetland) are indirectly threatened by loss of grassland to plantation forestry and cropping owing to their sensitivity to altered hydrological functioning.

Although only 12 of the 21 bird species are grassland species, with another two using grassland and forest (Appendix 1), all the bird species use all grassland types in the catchment. The social units of these species use a fairly large home range (Hockey, Dean & Ryan 2005), which for some is greater than 100 km². Seven of the mammal species rely on grassland (Appendix 1). Oribi antelope, serval and weasel (Friedman & Daly 2004) require large tracts of grassland and are therefore expected to have been adversely affected by the degree of fragmentation in the catchment (compare Figure 2 and 3). The blue swallow has become locally extirpated in this catchment and is threatened with local extinction in South Africa (Wakelin & Hill 2007:252, 254). In contrast, the two amphibian species, as well as the other mammal, reptile and insect species, show fine-grained use of the habitat owing to their smaller body sizes (Appendix 1). These species will therefore likely be less affected by fragmentation, provided the remaining fragments contain the specific habitat features they require. Bird and mammal species whose habitat is forest are unlikely to have been negatively impacted by land transformation, as forests have increased marginally in size and have been spared harvesting in recent decades; however, increases in human density may have escalated indirect pressures.

The decline in biodiversity integrity due to a loss in grassland is probably confounded by deterioration in the botanical composition of grassland for livestock production and a decline in plant species diversity. Grassland dominated by the palatable Themeda triandra has been widely replaced by the unpalatable Aristida junciformis (Camp 1997:16; Scotney 1970:154), ostensibly as a consequence of poor grassland management practices earlier, including overstocking, frequent burning, burning early in winter followed immediately by grazing, and continuous selective overgrazing (Acocks 1988:7; Camp 1997:16; Tainton 1999:281). This compositional change has been associated with a loss of forb species (Scott-Shaw & Morris 2015:24).