Temporal turnover of the flora of lake islands: The island of Lake Pamvotis (Epirus, Greece)

Abstract Lake Pamvotis is one of the Balkan "ancient" lakes, a Quaternary refugium of great environmental importance and ecological value, that is under various anthropogenic pressures. It belongs to a Natura 2000 Special Area for Conservation and Special Protection Area. Almost in the middle of the lake, there is an inhabited island - one of the two lake islands in Greece – that also attracts touristic interest. Τhe main objectives of the present study are to provide a floristic inventory of the protected island, combining data of two different sampling periods, within a 25 year interval, in order to estimate temporal beta diversity and species turnover of the island’s plant diversity. The value of the absolute and relative turnover rates of the floristic diversity of the island studied are 4.24 and 1.72, respectively and are amongst the higher rates reported for plants. The absolute difference between extinct (E) and immigrant (l) taxa is to a great extent accounted, concerning life forms, by therophytes (1.86), hemicryptophytes (1.56) and geophytes (1.04) and, for habitats, by taxa preferring agricultural and ruderal forms (2.52).


Introduction
Lakes are considered as islands in many biogeographical and ecological respects and more specifically, they are "negative islands", that is, they are more or less isolated freshwater areas surrounded by a hostile land matrix (Whittaker and Fernandez-Palacios 2007). Lake Pamvotis is one of the Balkan "ancient" lakes that are restricted within a radius of 300 km around Lakes Ohrid and Prespa (see Touka et al. 2018, Figure 1) and the term "ancient" refers to long-living modern or palaeo-lakes Frogley and Preece 2004). Lake Pamvotis has been in existence throughout the Plio-Pleistocene period, as shown by the identification of several endemic mollusc taxa which are known to be 500,000 years old and it has also been characterised as a Quaternary refugium (Frogley et al. 2001;Tzedakis et al. 2002).
The lake is located at the south-eastern foot of the mountain Mitsikeli and covers 22 km of the basin. It is a shallow lake with an average depth of 4.5 m (Sarika-Hatzinikolaou 1999;Sarika-Hatzinikolaou et al. 2003). It was formed by water concentration and is fed by the mountain Mitsikeli springs, while it has no physical surface outlet. The outflow of the water is carried through Lapsista's ditch and is diverted to the river Kalamas. The lake belongs to the Natura 2000 Special Protection Area and Site of Community Importance GR2130005 and it is under the responsibility of the Lake Pamvotis Management Body. It has attracted research interests as a sedimentary archive on long term environmental and climate history and as a hotspot for European biodiversity (Touka et al. 2018). The lake could be characterised as an island (sensu Whittaker and Fernandez-Palacios 2007) and almost in the middle of the lake's vast extent, opposite the town of Ioannina, a true island is formed, named "Nisi" (Fig. 1). Chelli et al. (2019), demonstrated the urgent need for monitoring species diversity over time in protected areas by approaches that are scientifically sound and comparable over time and they underlined that the assessment of species diversity is crucial and that the analysis of temporal beta diversity -the variation of species composition over timereflects species turnover. MacDonald et al. (2018) investigated relationships between fragmentation and species diversity concerning lake islands but there are very few previous studies on species turnover dealing with lake islands (e.g. Nilsson and Nilsson 1982).
In this context, the aquatic macrophyte flora of the Lake is well studied (Sarika-Hatzinikolaou 1999;Sarika-Hatzinikolaou et al. 2003;Stephanides and Papastergiadou 2007), but there is no previous literature information concerning temporal turnover of the plant species diversity of the island studied. Therefore, the changes that may have occurred over time in the floristic composition and plant species diversity on the island 2 "Nisi" of Lake Pamvotis, which is heavily influenced by human interventions, have not yet been investigated. Τhe main objectives of the present study are to provide a floristic inventory of the protected island, by combining data of two different sampling periods in order to estimate the temporal beta diversity and species turnover of the island's plant diversity within a 25-year period (1993-2019).

Survey area
The small island "Nisi" of Lake Pamvotis is a tectonic prominence of the alpine background of the area which is attributed to the same formation mechanism of the lake (Karakitsios 2005;Karakitsios 1995). The shape of the island is elongated, its relief is relatively mild, its highest elevation reaches 529 m above sea level or 59 m above lake level and its surface area is about 20 hectares. The island consists of limestone of the Upper Senonian (Upper Cretaceous). It is made up of compact strata of limestone with Rudist fragments (drifted from Gavrovo Zone) and layers of the pelagic limestone with Globotruncanidae. Their thickness vacillates between 200 to 400 m (Karakitsios 1995;Karakitsios 2005).
The area belongs to the wet bioclimatic level of the Emberger-Sauvage climatic diagram, with cool winters and the adverse xerothermic period for the plants is relatively short and not intense, lasting from the beginning of June until the middle of August.
The island has been inhabited since the 13 century AD and its flora and vegetation have been strongly influenced by human activities. It is also an area of cultural interest attracting thousands of tourists every year. th Figure 1.

Map of the area studied.
Temporal turnover of the flora of lake islands: The island of Lake Pamvotis ...

Data Collection, Database and Floristic Analysis
The present study is based on plant diversity data collected during two different sampling periods within a 25 year interval. During the first sampling period (sampling period A), Th. Chitos (TC) collected a large number of plant specimens by regularly visiting the island from autumn 1993 until autumn 1995. Α few taxa were also collected or observed by A. Yannitsaros, on a one-day visit to the island in October 1997(Yannitsaros et al. 2005. During the second sampling period (sampling period Β) from May 2018 until April 2019, sample collections were carried out by Α. Papanikolaou and Μ. Panitsa. The sampling scheme, used for both sampling periods, included fieldwork (plant specimen collections and observations) all over the island during all seasons of each year.
Along with the terrestrial plants, aquatic plants growing close to island shores were also collected. Plant specimens are deposited in the personal herbaria of collectors TC and AY and in UPA Herbarium. The indication "obs." means that the record is based on field observation.
For the determination of the plant material, Tutin et al. (1964), Tutin et al. (1981, Tutin et al. (1972), Tutin et al. (1976), Tutin et al. 1980, Strid and Tan (1997 and Strid and Tan (2002) were used. Taxonomy of the family of Poaceae follows Valdés and Scholz (2006). The families, genera, species and subspecies are listed in alphabetical order, within the major taxonomic groups. Nomenclature, life form, chorology, status categories and habitat preferences of plant taxa follow Dimopoulos et al. (2013). The main life form categories are summarised as follows: phanerophytes, chamaephytes, hemicryptophytes, geophytes, therophytes and aquatics. Chorological categories have been grouped into 5 main chorological elements: Widespread, Mediterranean, Balkan, Greek endemics and Alien.
In the framework of the authors' research, a list of all information recorded has been created including the plant taxa, their biological and chorological types and their habitat preferences, according to data provided by the "Vascular Plants of Greece" (Dimopoulos et al. 2013; see also "Flora of Greece web" 2017+, http://portal.cybertaxonomy.org/floragreece-intro).
Absolute (S2; Herwitz et al. 1996) and relative (Rt;Diamond 1969, Morrison 2003 temporal turnover rates have been quantified using the formulas: where t is the time period between censuses, E stands for species present only in sampling period A (extinct), l for species present only in sampling period B (immigrants), SA for all species recorded in sampling period A and SB for all species in sampling period B.

Results
Field investigations on the lake island "Nisi" during two sampling periods, sampling period A (1993)(1994)(1995)1997) and sampling period B (2018-2019) revealed a total number of 350 plant taxa recorded, of which 5 are Pteridophyta, 3 Gymnospermae and 342 Angiospermae, belonging to 246 genera and 84 families, as is shown in Table 1. The total numbers of families and genera in sampling period A were 79 and 204, respectively, while in sampling period B, 71 and 163, respectively. A total of 263 taxa were present during sampling period A (SA) and 231 in sampling period B (SB). The total number of taxa present only in sampling period A was 119 (l) consisting 34% of the total flora, while 87 taxa (24.85%) were present only in sampling period B (E); 144 taxa (41.15%) were common to both sampling periods (C).  [1993][1994][1995]1997, Sampling period Β = Collections and observations from fieldwork period 2018-2019, X = alien, xenophyte, non-native plant taxa including cultigens, permanently established in at least one floristic region of Greece and r = range-restricted plant taxon characterised by a restricted distribution, by populations occurring along a linear distance not exceeding 500 km.
Temporal turnover of the flora of lake islands: The island of Lake Pamvotis ...  (14 taxa) are the richest in taxa families in the flora of the island. Taxa belonging to these families represent 46.2% of the total island's flora. The richest in taxa genera are: Trifolium (8 taxa), Veronica (7 taxa), Medicago (5 taxa), Bromus (5 taxa), Crepis (4 taxa) and Geranium (4 taxa). The genus Erigeron is represented by 3 alien taxa. Most of the genera are represented by less than 3 taxa.

Plant taxa A B Status
Regarding the chorology, widespread taxa dominate (78.7%), followed by Mediterranean elements (33.1%) and alien taxa (6%). Fig. 2 presents proportions of different chorological elements recorded during the two sampling periods. The presence of Balkan elements is quite significant since 15 taxa (4.2%) have been registered, of which 6 are range-restricted (Table 1), while Heliotropium halacsyi Riedl is the only Greek endemic and range-restricted taxon registered. Interesting is the presence of widespread taxa that are rather rare in Greece like Cicuta virosa, a wetland toxic plant, which was found in a small population (not more than 1000 individuals) near the island shore, adapted to the lake's ecological conditions but at risk because of anthropogenic interference. It is classed as Vulnerable in Greece (Sarika and Yannitsaros 2009).
On the life form spectrum, Therophytes dominate (38.5%) followed by Hemicryptophytes (34.8%) and Geophytes (10.7%). Life forms of taxa present only in sampling period A, only in sampling period B and common to both sampling periods are presented in Fig. 3 and proportions of different life forms recorded during the two sampling periods in Fig. 4. Phanerophytes present a rather low proportion and, amongst them, there are tree species like Cupressus sempervirens L., as well as cultivated stands of Pinus brutia which form artificial small woods in the study area.    The evaluation on the habitat preferences of plant taxa reveals that most common are plants of agricultural and ruderal habitats (36.7%), followed by plants of grasslands (15.7%), of shrub formations (13.8%), of woodland and scrub (11.4%) and of freshwater habitats (11.1%) (Fig. 5).
Absolute turnover (S2) and relative turnover (Rt) rates for the total flora, for different life forms and for taxa with different preferences on habitats, are presented in Table 2. For the total flora, S2 is 4.24 species per year and Rt is 1.72 species per year (for t = 25 this gives 43% species change). The highest relative turnover rates amongst taxa of different life forms have been revealed for therophytes (2.05) and geophytes (2.16), while the highest absolute turnover rate concerning habitat preferences was calculated for taxa preferring agricultural and ruderal habitats (2.52). Relative turnover seems to be high for taxa preferring xeric phryganic formations because the percentage of these taxa in total is rather low (4.8%).

S2 Rt
Total

Discussion
Lake Pamvotis is the second oldest lake at European level after Lake Ohrid and it is a sedimentary archive on long term environmental and climate history (Touka et al. 2018). Lake Pamvotis has also been characterised as a Quaternary refugium, that is an ecologically stable area, critical not only for the long-term survival of existing species, but also for the emergence of new ones (Tzedakis et al. 2002). Lake Pamvotis island is one of the two lake islands occurring in Greece -the second one is the island of Agios Achilleios Table 2.
Absolute turnover (S2) and relative turnover (Rt) rates for the grand total, for different life forms and for taxa with different preferences on habitats.
in Prespa Lake -both belonging to protected areas. There has been a gap so far concerning plant species richness and temporal turnover of the diversity of the island studied, which is included in the protected area of Lake Pamvotis, GR2130005.
During our field investigations on the island in two sampling periods with a 25 year interval, 351 plant taxa were registered in total belonging to 84 families and 246 genera. Of the taxa recorded in total, 42.4% belong to the families Asteraceae, Poaceae, Lamiaceae, Fabaceae and Brassicaceae, which are amongst the best-adapted families to the ecological conditions of the Mediterranean area, as is confirmed by many floristic studies of Greek insular areas (amongst others Panitsa and Tzanoudakis 2001; Panitsa and Tzanoudakis 2010; Kougioumoutzis et al. 2012;Iliadou et al. 2014a;Iliadou et al. 2014b). This is also the case when floristic composition of the two sampling periods (SA and SB) is examined separately (  Dimopoulos et al. 2013), but in a small area.  recognised i) the human population density as a significant predictor of the spatial distribution of alien species in Greek Natura 2000 sites, assuming a positive relationship between human presence and alien plant species richness and ii) a possible preference of invasive plants for northernwestern Greek Natura 2000 sites that could be explained by the cross-border trade that may facilitate the spread of invasive species.
The Pamvotis island plant diversity lost 10.5% of plant taxa during the studied interval of 25 years. The patterns of occurrence have changed for 60.4% of the plant taxa with 43% having being subjected to maximum turnover. It has to be noted that the absolute difference between extinct (E) and immigrant (l) taxa is mainly accounted by therophytes, hemicryptophytes and geophytes.
The value of turnover rate on the studied island is amongst the largest reported for plants (amongst others Cody 2006;Panitsa et al. 2008) and two factors should be taken into account, the small surface area (20 ha) and the continental character of the island. Cody (2006) showed that plant species turnover is a highly variable phenomenon, depending on the functional traits of each species, the local geomorphology and the ecological conditions of the islands. Morrison (2016) and Chelli et al. 2019) mentioned that temporal analyses could be affected by the "pseudoturnover" effect, the imperfect detection of species because of, for example, their presence during the sampling period only in the soil seed or bulb banks, but this is, more or less, the case for all plant species turnover studies. In order to avoid this "pseudoturnover" effect, the adequate time framework for the study of temporal plant species turnover on islands must include fieldwork distributed over all the different seasons per year and exhaustive field collections and observations all over the islands. This is what has been followed for the studied island.
Human activities are amongst the factors enhancing turnover (Panitsa et al. 2008). It is noteworthy that the absolute turnover rate also reveals the high contribution of ruderal plant taxa. Ruderal taxa are typically occurring and prevailing in disturbed areas, in agricultural and ruderal habitats and especially in sites with pronounced direct or indirect human activity, rural and urban sites, roadsides, excessively grazed and trampled sites, as well as naturally nutrient-rich and frequently disturbed pioneer habitats (Dimopoulos et al. 2013). Pyšek et al. (2017) noticed that ruderal flora and vegetation generally represent one of the most dynamic floristic-vegetation complexes that have a huge impact on the homogenisation of biodiversity and vegetation. Dornelas et al. (2014) and McGill et al. (2015) mentioned that changes in community composition (temporal beta diversity) and spatial structure (spatial beta diversity) need much more work to be quantified, but our focus should also include the quality and the composition of biodiversity.
Protected areas are the most widely applied policy tool for biodiversity conservation and there is now an increasing need to incorporate social impacts in decision-making processes by providing accurate estimations and developing ways to forecast their change in the future (Jones et al. 2015, Jones et al. 2017, Jones et al. 2018). Lazoglou and Vagiona (2018) assessed the willingness of inhabitants of the nearby city of Ioannina to pay for environmental actions and projects for the management of the protected area of Lake Pamvotis. The lake, despite its environmental importance and ecological value, is under various anthropogenic pressures (Latinopoulos et al. 2018) and Lazoglou and Vagiona (2018) suggested that economic valuation could provide an effective tool for planning processes and decision-making policies for the protection of the environment. Chelli et al. (2019) demonstrated the usefulness of long-term plant diversity programmes for the study of species richness and patterns of species assemblages over time in a protected area. Social impacts of protected areas should not be seen as static concepts, but should be seen as dynamic and long-term factors which need to be incorporated in decision-making processes (Jones et al. 2017). The knowledge of the plant species diversity, the high turnover rate and the continuing human interference on the protected lake island studied should be considered for biodiversity conservation actions which should also concern the whole protected area of Lake Pamvotis.