Soil Seed Bank and Aboveground Vegetation in Grazing Lands of Southern Marmara, Turkey

The composition and conservation of plant communities is greatly influenced by the soil seed bank. Information on the soil seed banks and the remaining vegetation in these ecosystems is crucial for guiding the restoration efforts. This study examines the size, species richness, diversity, uniformity, and similarity of soil seed banks and aboveground vegetation in 6 different grazing lands including coastal pasture, reseeded pasture, artificial pasture, lowland shrubland, ungrazed pasture, and hillside shrubland. Forty-eight soil samples were taken by cores with a diameter and depth of 10 cm from each of grazing lands in August of 2007. A vegetation survey was conducted using a 0.5 x 0.5-m quadrant in both the spring and fall. Eighty species were observed in soil seed banks and aboveground vegetation. The largest seed bank was observed in reseeded pasture (7,715 seed/m2), while the smallest seed bank was found in coastal pasture (2,755 seed/m2). Coastal pasture also possessed the least amount of aboveground vegetation (131 plants/m2). The most aboveground vegetation was found in ungrazed pasture (155 plants/m2). The most common species in seed banks were annual and perennial grasses in reseeded pasture, annual forbs in artificial pasture and hillside shrubland, and perennial forbs in low shrubland and ungrazed pasture. Species richness, diversity, and uniformity in seed banks were highest in lowland shrubland and lowest in artificial pasture. The seed bank and aboveground vegetation were similar in ungrazed pasture, coastal pasture, reseeded pasture, low shrubland, hillside shrubland and artificial pasture. Shrublands play an important role in species richness and the number of germinated seeds from seed banks of grazing lands in southern Marmara. The results showed that reseeding or a decrease in grazing pressure may improve the condition of grazing lands.


Introduction
Southern Marmara has a Mediterranean climate; thus, shrublands (maquis) are common in the region (Yılmaz, 1993).Shrublands are composed of evergreen and deciduous shrubs and annual or perennial plants.Annuals generally grow in heavily grazed lands on dry, low yielding soils and are the most common vegetation in the Mediterranean (Yılmaz, 1996;Marcelo et al., 2003).Mediterranean shrubs provide habitats for wildlife (Papachristou et al., 2003) and are an important food source for domestic animals.Throughout the year, shrublands are grazed by sheep and goats.The combination of shrubs and herbaceous plants produce feed in the spring and fall.Goats that graze heavily on shrubs prefer green herbaceous varieties (Papachristou, 1994), and many grazing lands in the Mediterranean are composed entirely of herbaceous plants.The loss of shrubs from grazing land may be due to conversion to farmland, poor soil quality or exploitation.For instance, coastal land is composed of sandy soil with a low water holding capacity that suffers from drought in the summer.Drought resistant therophytes are dominant in these areas (Raunkiaer, 1934).Studies to improve the landscape have been conducted on abandoned lands due to the loss of natural and quality plants from these areas.Reseeding is the best method for improving abandoned areas (Altın et al., 2005), leading to an increase in yield and quality and a decrease in species richness (Bakker, 1989;Tallowin, 1996).
Soil seed banks play an important role in the composition and conservation of plant communities (Grubb, 1977;Leck et al., 1989;Wisheu and Keddy, 1991).The composition of the seed bank depends on the composition and production of current and previous plant communities, as well as the longevity of seeds under local conditions.Seeds that have been brought in from surrounding areas also contribute to the seed bank.
Seed production is dependent on land use and environmental factors: production can be diminished by a reduction in leaf area due to livestock grazing, the allocation of photosynthate to reproductive organs, or by the direct removal of flowers and seeds (Sternberg et al., 2003).Ungrazed sites typically possess a greater number of species and higher species density in the seed bank compared to grazed sites (O'Cannor and Pickett, 1992;Jutila, 1998;Mayor et al., 2003).Navie et al. (1996) found that Subsequently, the land was reseeded with Lolium perenne, Bromus inermis, Medicago sativa, and Lotus corniculatus in 2005 and fertilized with N and P. Controlled grazing practice are applied in this site.The soils of reseeded pasture are loamy, pH=7.27,EC= 0.29 dSm, CaCO 3 =4.29%,organic matter= 5.16%, available phosphorus and potassium 62.80, 469 mg kg respectively.
Artificial pasture: Artificial pasture was tilled after which wheat was grown.An artificial pasture is composed of a mixture of 20% Medicago sativa, 20% Lotus corniculatus, 30% Lolium perenne, and 30% Bromus inermis.These species are the most common in grazing lands, although other types of herbaceous plants have also been found.Artificial pasture is grazed under controlled conditions.The soil is loamy sand, pH=7.38,EC= 0.38 dSm -1 , CaCO 3 =11.90%,organic matter= 3.98%, available phosphorus and potassium 29.74, 394 mg kg -1 respectively.

Seed bank sampling
The seed bank was sampled in August 2007 before fall rains and the onset of germination.Four plots were chosen from each of the six types of grazing lands.Twelve randomly sampled soil cores were obtained from each plot.Cores had a diameter and depth of 10 cm.Soil cores samples were stored in a dark, cold storage area (1°C) for 6 months.The samples were sorted to eliminate plant fragments and stones.Soil from each plot was combined and the total seed density in the soil seed bank of Queensland decreased with high levels of grazing and increased with intermediate grazing levels due to an increase in the number of annuals.
On the other hand, soil features (humidity, texture, etc.) determine the composition of vegetation and amount of seeds produced by plants.Due to the prevalence of drought in sandy soils, annual species are common, and the seed number in the seed bank is generally low (Staaf et al., 1987).Alternatively, water resources are not limited in low areas during dry seasons.Perennials are able to grow and produce enough seeds to perpetuate themselves, even under heavy grazing.
Due to deep root systems, shrubs in the Mediterranean region are able to adapt and can provide invaluable food to grazing animals and sustain soil productivity (Bochet et al., 1999).Because shrubs endure under grazing conditions, they also enable the survival of many herbaceous species: perennial herbaceous species can mature under the canopy of shrubs.
There are many studies on the botanical composition, yield (Ozaslan Parlak et al., 2011), and animal performance (Tölü et al., 2007) of grazing lands in southern Marmara.However, studies on seed banks in the region have not been conducted.The objectives of this study were to determine the size, composition, species richness and diversity of the seed bank and aboveground vegetation of different grazing lands representing southern Marmara region and to describe the relationship between the seed bank and existing vegetation within grazing lands.

Materials and methods
The study was conducted on Çanakkale grazing lands, located in northwest part of Turkey.Çanakkale has a typical Mediterranean climate (Türkeş et al., 2002): the mean annual precipitation is 615.5 mm and falls mainly in the winter and spring.The rainy season typically begins in October and ends in May, while summers are dry and hot.The mean annual temperature is 15 o C.
Six different grazing lands located in southern Marmara were chosen as study sites, including coastal pasture, reseeded pasture, artificial pasture, low shrubland, ungrazed pasture, and hillside shrubland.
Reseeded pasture: Originally, reseeded pastures possessed a large amount of Paliurus spina-christi; however, these plants were removed because they restricted grazing.spread onto a plastic tray (25 x 50 x 7 cm) containing 2 cm of sterile potting soil in February of 2008.
Trays were placed in a greenhouse with natural light and temperature (between 14 and 25°C) and watered frequently.Trays were rotated monthly to a different area of the greenhouse to reduce any effect of placement.Four control trays filled with sterilized potting soil were also randomly placed in the greenhouse to test for contamination.Germination was not observed in control trays.
Seedlings were identified immediately after germination and were counted and removed, or, if they were not immediately identified, transplanted into pots for further growth.After two months, soil samples were carefully turned over to facilitate the emergence of new seedlings (Roberts, 1981).Germination trials lasted for approximately 8 months.

Vegetation survey
Vegetation was sampled twice by recording the species present in a 0.5 x 0.5-m quadrant randomly placed 10 times on the surface of each type of pasture.The first survey was conducted on May 4, 2007, when Mediterranean plants were in the flowering stage.The second survey was conducted on November 4, 2007, when plants were in the reproductive stage.Only herbaceous plants germinated in the experiment; thus, the composition of these plants was used exclusively in the vegetation survey.

Data analyses
To determine species diversity, the Shannon-Wiener index, Simpson index, and Hurlbert's probability of intraspecific encounter were determined.The Pielou index was used to indicate species uniformity (Shannon and Weaver, 1949;Pielou, 1975).
Simpson diversity index: Where S = species richness, N i = number of the ith species, N = number of all species, and P i = proportion of the ith species.
Soil seed bank density, aboveground vegetation density, and soil properties of grazing lands were compared by a one-way analysis of variance (ANOVA) and Duncan's multiple-range test with p = 0.05.All analyses were conducted with MINITAB Release 14.13 (Minitab INC., 1972-2004).Before conducting ANOVA, all data were calculated as the mean ± standard error.
Seed bank and aboveground vegetation were assessed for similarity using Jaccard's similarity coefficient based on the presence or absence of species ( Jaccard, 1912).This coefficient was calculated for all types of vegetation.
where a = the number of species in sample A and B, b = the number of species present in A and absent in B, c = the number of species present in B and absent in A. Jaccard's coefficient ranges from 0, indicating a lack of mutual species, to 1, denoting complete concurrence.

Overall richness and density of the soil seed bank and vegetation
A total of 12,000 seedlings (1,039 in coastal pasture, 2,911 in reseeded pasture, 1,640 in artificial pasture, 2,382 in low shrubland, 1,646 in ungrazed pasture, and 2,382 in hillside shrubland) and 80 species (29 in coastal pasture, 33 in reseeded pasture, 23 in artificial pasture, 34 in low shrubland, 33 in ungrazed pasture, and 32 in hillside shrubland) were observed.The largest seed bank was found in reseeded pasture (7,715 seeds/m 2 ), followed by hillside shrubland, low shrubland (6,314 seeds/m 2 ), ungrazed pasture (4,363 seeds m -2 ), and artificial pasture (4,347 seeds/ m 2 ).The smallest seed bank was observed in coastal pasture with 2,755 seeds/m 2 .Annual and perennial grasses were predominant in reseeded pasture.However, grasses were rare on artificial pasture and annual forbs were the most common plant (3,105 plant/m 2 ).Annual forbs were also widespread in hillside shrubland (2,444 plant/ m 2 ) but were lowest in reseeded pasture (581 plant/m 2 ).Low shrubland and ungrazed pasture contained the highest level of perennial forbs.The occurrence of legumes on different types of land was not statistically significant (Tab.1).
A total of 80 species of aboveground vegetation were observed.Of these species, 36 were found in coastal pasture, 29 in reseeded pasture, 25 in artificial pasture, 29   (Tab. 3,4).

Properties of the soil seed bank and aboveground vegetation
The seed banks and aboveground vegetation of artificial pasture had the lowest species richness.The highest number of species was found in low shrubland (34), reseeded pasture (33), ungrazed pasture (33) and hillside shrubland (32).Coastal pasture had the highest number of species (36) in aboveground vegetation; however, there were less species in seed banks of coastal pasture and artificial pasture compared to aboveground vegetation.Alternatively, other types of grazing lands contained more species in seed banks than in vegetation.The highest seed bank diversity (D = 0.958 and PIE = 0.957) was observed in lowland shrubland, while diversity was the lowest in the seed bank and aboveground vegetation of artificial pasture (D = 0.842, 0.854 and PIE = 0.838, 0.839, respectively).The highest diversity (D and PİE) in aboveground vegetation was found in coastal pasture.Artificial pasture displayed the lowest uniformity in both the seed bank and vegetation ( J = 0.881, 0.890).The highest value of uniformity was found in the seed bank of lowland shrubland ( J = 0.987) and in vegetation of coastal pasture ( J = 0.961) (Tab.5).

Similarity and relationship between soil seed bank and aboveground vegetation
Similarity between soil seed bank and aboveground vegetation was strong.The values of Jaccord's coefficient of community ranged from 0.411 to 0.676 (with a mean of 0.518) (Tab.6).Fluctuations in similarity were observed between different grazing lands in vegetation quadrants and seed banks.The similarity between different types of aboveground vegetation varied between 0.100 and 0.410 in the number of seeds in the seed bank.Reseeded pasture has productive soils with the least amount of erosion due to a low slope.The species diversity of germinative seeds increased with increasing soil fertility (Staff et al., 1987).Shrublands were rich in germinating seeds; specifically, low shrubland displayed the highest species richness, diversity and uniformity.When compared to other types of shrublands, lowland shrubland experienced the least water stress and little to no erosion due to a lack of slopes.Animals were not able to access herbaceous vegetation between shrubs; thus, plants produced seeds even under heavy grazing.Seeds in low shrublands disperse by many mechanisms: small seeds dispersed by wind, animals and overflow (Harper, 1977;Grant, 1983).
In ungrazed pasture, less germination occurred compared to other types of shrubland.Similarly, Bakoğlu et al. (2009) reported more germination in grazed land com- (Tab. 6: v).Between different types of seed banks, similarity ranged from 0.173 to 0.459 (Tab.6: s).The degree of similarity in seed banks was moderately stronger than in vegetation.The weakest similarity was found in the vegetation of hillside shrubland and coastal pasture.The strongest similarity was observed in the seed bank and vegetation of reseeded and artificial pasture.In seed banks, the weakest similarity was observed between ungrazed and coastal pasture.

Discussion
Reseeded pasture was rich with germinated seeds.Seeds of herbaceous species were produced in between Paliurus spina-christi plants, which had been growing for a long period of time and prevented grazing.After removing shrubs, reseeding with legumes and grass led to an increase The number of seeds that germinated in artificial pasture was low.These lands also displayed the lowest species richness, diversity and uniformity.Artificial pasture pared to ungrazed land.Grazing allows seeds to become buried in the soil (Altın et al., 2005) and grazing animals cause a decrease in the number of seed predators (birds, rodents, etc.) (Gökkuş and Koç, 2001).Despite a higher seed production in ungrazed land, a lower germination Coastal pasture had the lowest number of germinated seeds and vegetation on these lands (Tab.3).This may be due to the texture of coastal soils: buried seeds were able to resurface from the sandy soil and seeds left on the soil surface were eaten by insects such as ants and beetles and small mammals (Thompson, 1987).Similarly, Staff et al. (1987) found that less seeds were germinated from sandy loams.In coastal pasture, species richness was higher in vegetation than in the seed bank.This result indirectly suggests that asexual reproduction was the primary method of reproduction in coastal grassland.Similar results have been obtained in steppe ecosystems (Zhan et al., 2007).Thus, the preservation of perennial vegetation may play a more significant role in grassland recovery than soil seed banks.
The most germinated species in all grazing lands were annuals.Research has shown that annuals are a major component of Mediterranean grasslands (Marcelo et al., 2003).The grazing of Mediterranean grasslands can have a large impact on vegetation (Montalvo et al., 1993;Seligman, 1996;Lavorel et al., 1999;Sternberg et al., 2000); however, the soil seed bank may buffer the impact of differences in inter-annual grazing (Russi et al., 1992;Ortega et al., 1997).It is generally assumed that Mediterranean annuals show adaptive responses to unpredictable environments by producing dormant seeds that persist during periods of limited resources (Shmida and Ellner, 1984;Thompson, 1992;Lavorel et al., 1993;Perevolotsky and Seligman, 1998).Thus, a persistent soil seed bank acts as a long-term refuge for annual plant populations in variable environments typical of the Mediterranean region (Kigel, 1995).
With the exclusion of coastal pasture, seed banks of grazing land contained more species than aboveground vegetation.This is expected in heavily grazed pastures be- Many studies have been conducted on the similarity of seed banks and aboveground vegetation.Low similarities were observed between the seed bank and aboveground vegetation in most studies involving grasslands dominated by perennial grasses (Rabinowitz, 1981;Archibold, 1981;Schenkeveld and Verkaor, 1984;Thomson, 1986;Fischer, 1987;Millberg, 1995;Bakkler et al., 1996).These discrepancies were explained by the minor contribution of dominant perennial grassland species to seed banks.In general, perennial grasses have low seed production because they alternate between sexual reproduction and vegetative forms and their seeds persist in the soil for short periods of time (Touzard et al., 2002).
The strongest similarity between vegetation and seed bank was observed in coastal and ungrazed pastures.Annual grasses and forbs commonly grow in coastal pasture and water stress often occurs due to the presence of sandy soils.This leads to rapid seed production by annuals and results in increased similarity (Peco et al., 1998).Plants in ungrazed pasture produce seeds in vast quantity.Since there is no grazing pressure, plants do not disappear from the vegetation.Due to the fact that seeds on the aboveground vegetation are buried in the soil, similarity between the soil seed bank and vegetation increases (Grime, 1979).The weakest similarity between vegetation was observed in hillside shrubland and coastal pasture.Sandy soil in coastal pasture results in unique plant species and the weakest similarity.
The strongest similarity was observed in seed banks and vegetation of reseeded and artificial pasture.These pastures are located close to each other and possess similar soils.Although the use of reseeded and artificial pastures is different, their characteristics are quite similar.This result indicates that soil properties have an effect on vegetation type.

Conclusions
The grazing lands of Southern Marmara are under heavy grazing pressure and shrublands play a crucial role in species richness and the number of seeds that germinate.