Seed dispersal in agricultural habitats and the restoration of species-rich meadows = Dispersie van zaden in cultuurlandschappen en het herstel van soortenrijke graslanden

The restoration of species-rich meadows on former agricultural land in the Netherlands has a high priority, because these ecosystems have been disappearing rapidly due to eutrophication and acidification and falling water tables. In order to be able to restore such ecosystems on wet nutrient-poor soils, the <em>suitability</em> and <em>accessibility</em> of target sites have to be improved.The starting point for the restoration of species-rich meadows is frequently a soil that has been drained and enriched with fertilizers and polluted with pesticides for decades. Increasing the suitability of a site then involves ameliorating the habitat quality for the selected plant species by rewetting the soil, by reducing the availability of nutrients to plants through cutting and grazing and by removing the enriched topsoil, so that new individuals can establish. It is generally assumed that seeds are still available or will soon become available, This is not necessarily always the case. There are two alternative strategies by which plants may (re)colonize new sites; either through the germination of seeds buried in the soil or through the dispersal of seeds.In situations where soil seed banks have been depleted, dispersal of seeds from neighbouring sources via water, wind, animals and humans is the only natural option to restock a site with seeds. Since most grassland species have a limited dispersal capacity, the distances between seed sources and target sites become crucial. Ecological corridors could facilitate the dispersal of species in agricultural habitats if they satisfy the habitat requirements of the selected species.This thesis deals with a few aspects of the regeneration ecology of several meadow plant species (Chapters 1-5). Two questions are raised in particular: is it possible to restore species-rich meadows on previously farmed fields? and do ditch banks function as ecological corridors for species that are absent from a site undergoing restoration? To provide a contrast with typical grassland species, fleshy-fruited plant species and their specific dispersal characteristics have also been studied (Chapter 6).<strong>Chapter 1</strong> describes the first phase of the restoration of species-rich meadows on former agricultural land in an intensively farmed landscape in the centre of the Netherlands. A comparison between the species pools of the former (pre1950) species-rich meadows, a set-aside area undergoing restoration and the ditch banks in the surrounding formed landscape, revealed that 106 out of the 145 meadow species of the former species-rich meadows were still present in refugia on ditch banks. Eighty-five out of these 145 meadow species have survived or already recolonized the set-aside area, but another 60 species has not yet recolonized the site due to insufficient seed dispersal, depleted soil seed banks and/or too few appropriate microsites in the vegetation for germination and establishment.An increase in the number of meadow species would be attainable if the accessibility and suitability of the site can be maximized. The effectiveness of the dispersal vectors water, humans and animals is extremely limited in the study area, leaving wind as the principal dispersal vector. Although some of the missing (extinct) species with long-range dispersal or permanent seed banks will reach the site without help, most species will not re-establish without being introduced deliberately.<strong>Chapter 2</strong> analyses the dynamic distribution of ten perennial plant species typical of species-rich meadows in a Dutch agricultural landscape (220 ha). Mapping in 1990, 1991 and 1992 showed that ditch banks in the study area (comprising a set- aside area and the surrounding farmed landscape) form an important refugium for the selected species. Ditch banks were managed by regular mowing plus removal of the harvested biomass in the set-aside area and by grazing, mulching and dredging in the farmed landscape. In the setaside area, seven species were more frequent, whereas two species were less frequent. One species did not differ in frequency between the two areas.The distribution of the selected species varied greatly between years, suggesting frequent extinction and colonization events. This type of variation was expressed as E/C, i.e. the mean ratio of the number of extinctions and colonizations, The overall E/C index for all species and all years was 0.99. Four species appeared to be decreasing in distribution in the study area (E/C>1.0), six species appeared to be stable (E/C=1.0) or even increasing (E/C<1.0). A related index used was the proportion of cells that was occupied permanently (P/(T+P)).For the the selected plant species, the variation in the indices E/C and (P/(T+P)). was related to five life-history attributes (seed weight, dispersal mechanism, dispersal distance, ability to vegetatively spread and seed bank type) and proved to be not significantly associated. It is concluded that different combinations of life-history attributes (i.e. regeneration strategies) lead to species stability in this type of agricultural landscape.<strong>Chapter 3</strong> explores the effects of five wind speeds (variable V: 2-13.5 m/s) and five release heights (variable H: 0.2-0.6 m) on the dispersal distances of seeds of six barochorous grassland perennials in a wind tunnel. The variation in dispersal distances within a seed population and between species with different aerodynamic attributes was expressed as 1-percentile, mode and 99-percentile values. Regression analyses showed that a model with three terms (V, V*H and V ²) best explained the variance in the dispersal distances across all species. According to the regression models, the dispersal distances of seeds in the tall of a frequency distribution (99-percentile values) increased exponentially with wind speed. At wind speeds of 14 m/s, predicted maximum distances were 10 to 15 m for small and relatively heavy spherical seeds and 20 to 30 m for large and relatively light cylindrical or disk-like seeds.A review of meteorological data showed that wind gusts>10 m/s at plant height occur at least annually. The long life-spans of plants of the selected species (up to several decades) suggests a large potential for long-range dispersal during their life-time. Individual populations appeared to be less isolated from other populations than can be inferred from distribution patterns of seed sources.<strong>Chapter 4</strong> reports on the success of establishment after adding seeds of ten selected perennial plant species to a grassland undergoing restoration. The recolonization of former agricultural grasslands by perennial grassland species is assumed to be delayed or even prevented by a lack of seeds, by a lack of microsites offering opportunities for germination and establishment, or by both.Sampling the seed rain with sticky traps recorded the seeds of resident species and ubiquitous wind-dispersed species of the genera <em>Betula, Cirsium</em> and <em>Epilobium.</em> Given the spatial distribution of seed sources in the surrounding agricultural landscape and the limited dispersal capacity of the selected species, the fields of the restoration site are largely inaccessible. Lack of seeds was a major cause of their absence.Seeds of ten plant species were also added to a sward that was mown, clipped or from which the sod had been stripped. Established plants were allowed to grow for two years and then harvested. The establishment success of the selected species on sod-stripped plots was significantly higher than on mown or clipped plots. Differences between these treatments can be explained by the low density and short duration of gaps in the intact (mown and clipped) vegetation. Differences between species were related to seed weight; species with large heavy seeds had a significantly higher establishment success than species with small and light seeds. The lack of appropriate microsites, especially for species with small seeds, was another cause of their absence.Recruitment from old buried seeds is another recolonization route. Burial of seeds for two years revealed very low mortality rates in species with small, spherical and hard-coated seeds, and moderate mortality rates in species with seeds of high area/content ratios and direct germination. Species of the first group are expected to be frequently recruited from seeds buried in the soil when sod stripping has been applied.<strong>Chapter 5</strong> explores the importance of linear landscape elements as ecological corridors. A cellular automaton model was built in order to determine the relative importance of the principal factors which determine the rate of migration of plants through corridors: the width and habitat quality of patches within a corridor (expressed as the population growth rate λ) and the dispersal capacity of plants (expressed as the slope αof the relationship between seed number and log-distance).Simulations with the model using different levels of the principal factors indicated highly significant and positive main effects of dispersal capacity, habitat quality and width of corridors on the rate of migration. Significant interactions existed for dispersal capacity x width and dispersal capacity x habitat quality, indicating that the effects of width and habitat quality depended on the dispersal capacity. In narrow corridors most of the dispersed seeds were deposited outside the corridor, which significantly reduced migration rates, especially for species with long-range dispersal of seeds. In wide corridors (up to 20 m), seed losses were much smaller and migration rates approximated those of continuous habitats. The contribution of the few longrange dispersed seeds to the rate of migration was significant when the quality of habitat patches was high. In all simulations, migration rates were <5 m/yr.Linear landscape elements are not effective corridors for plants with shortrange dispersal of seeds, because migration rates are low (<5 m/yr), high quality patches are few, and refugia and suitable habitat patches are frequently several kilometres apart, making a cohesive infrastructure for plants elusive.<strong>Chapter 6</strong> deals with the dispersal interactions between fruit-eating birds and fleshy-fruited plants that grow on wooded banks in an agricultural landscape in Twenthe, Overijssel province, the Netherlands. Wooded banks are a characteristic feature of the landscape, but their density in the landscape is changing, i.e. in some areas wooded banks have been removed whereas new banks are being planted elsewhere. On average, there were 7 fleshy-fruited plant species per 100-m transect (the range was 2 to 14 species). The number of fleshy-fruited species of the transects did not correlate with the density of wooded banks or of woodland.Eight fruit-eating passerine bird species were regarded as the major avian seed dispersers. They were divided into <em>longitudinal</em> dispersers which carry the seeds of the majority of fleshy-fruited species over short (<0.1 km) distances only and <em>transverse</em> dispersers which carry the seeds of species with conspicuous fruit crops over larger distances (>0.1 km). Seed dispersal by longitudinal dispersers is limited to the network of wooded landscape elements whereas transverse dispersers frequently disperse seeds to and from dissimilar landscape elements.The bird-mediated seed rain on wooded banks was sampled in twelve transects by using 120 seed collectors and by systematically collecting bird droppings. The seed rain was dominated by <em>Rubus fruticosus, Sorbus aucuparia, Rhamnus frangula, Lonicera periclymenum</em> and <em>Sambucus nigra.</em> The density of the seed rain was 215 ± 68 seeds m ^-2. Most seeds (87,6%) were deposited during the fruiting period of the plant species involved; the rest (12.4%) was deposited when ripe fruits were no longer available or was deposited on transects where fruiting adults were absent and should be regarded as immigrants.In order to clarify the complex interactions between birds, plants and landscapes, the ecological differences between longitudinal and transverse dispersers deserve more attention. It will help to better predict the consequences of changing the density of wooded landscape elements on the species richness and distribution of fleshy-fruited plants in fragmented landscapes.