Fragment quality rather than matrix habitat shapes forest regeneration in a South African mosaic-forest landscape

Land-use change and agricultural intensification are responsible for a global decline of forest cover entailing the fragmentation of forests. Landscapes are increasingly shaped by a mosaic of forest fragments within variable matrix habitat. It is thus essential to understand how these habitat alterations influence function and stability of forest ecosystems. However, the complex relationships between anthropogenic disturbance of forests, concurrent species loss and ecosystem functions are not fully clarified yet. In the present thesis, I investigated the impact of structural forest fragment quality and variable matrix habitat on biodiversity and ecological processes of forest regeneration. For this purpose, I conducted three studies in a fragmented mosaic-forest landscape in KwaZulu-Natal, South Africa. In particular, I assessed (1) seed predation by rodents, (2) establishment of woody seedlings and saplings, and (3) leaf damage on woody seedlings and saplings by insects and pathogens in 24 study plots in scarp forest fragments. These fragments were embedded in four variable matrices: two natural, heterogenous (forest, grassland) and two modified, homogenous matrices (eucalypt plantations, sugarcane fields). In the first field study I assessed rodent diversity and conducted seed predation experiments. As a measure for forest fragment quality important for rodents, I estimated herbal ground vegetation cover. For the second study I identified all trees on 500 m², seedlings on 10 m² and saplings on 50 m² per study plot. I categorized trees, seedlings and saplings as either early- or late-successional species. Moreover, I differentiated between seedlings and saplings of external and local origin depending on the presence of conspecific adult tress within fragments to be able to identify potential seed influx by seed dispersers. Additionally, I measured canopy cover, light intensity and vegetation complexity as parameters of forest fragment quality relevant to seedling establishment. In the third study I used beating samples to collect the arthropod community. Furthermore, I estimated proportions of leaf damage by insect herbivory and leaf pathogens on woody seedlings and saplings. I determined tree diversity, canopy cover and vegetation complexity in terms of forest fragment quality. My investigations showed an increase of rodents and seed predation in small forest fragments, potentially caused by enhanced ground vegetation cover. Especially in fragments with sugarcane matrix rodent abundance and seed predation were significantly higher than in fragments with forest matrix. Further, I found reduced seedling and sapling establishment in forest fragments with modified homogenous surroundings. In particular, these fragments consisted of less late-successional species. This could predominantly be ascribed to reduced canopy cover and increased light intensity. However, seed influx occurred in all forest fragments indicating high matrix permeability for seed dispersers. Moreover, forest fragment quality amplified arthropod predator abundance and reduced arthropod herbivore abundance. Fragment quality effects on herbivory were variable. This mismatch might be related to weak trophic interactions. Matrix habitat had merely marginal effects on the investigated factors. Overall, my results illustrate that forest fragment quality as well as matrix habitat have the potential to alter biodiversity and ecological processes of forest regeneration, but effects on the observed processes differed in strength. Yet, I generally found that forest fragment quality appears to be of high relevance for regeneration, indicating its potential for conservation management of the remaining fragments. Further, the rather weak overall effects of matrix habitat indicate a comparable and generally high permeability for the investigated groups of species. This emphasizes the significant value of forest fragments for the connectivity of remaining forests and the conservation of biodiversity and ecological processes at a landscape scale. Nevertheless, conclusions have to be treated with care. Due to the historical natural fragmentation of scarp forests in KwaZulu-Natal, which might have caused a higher robustness of species towards human-induced fragmentation, my findings might not be transferable to other regions. The scarp forest fragments are characterized by high habitat quality for the different species groups. However, potential shifts in the plant community from late- to early-successional species in fragments enclosed by modified matrices might entail unforeseen cascading effects and negative feedback loops within the ecosystem that still need to be examined. Thus, the value of natural forest as a source of propagules and as habitat for forest specialists remains indisputable