Facoltà di Scienze Matematiche Fisiche e Naturali
Corso di Laurea in Conservazione ed Evoluzione
Master Thesis
Effects of honeybees (Apis mellifera Linneaus,
1758) on the reproductive success of the
psammophyte Helichrysum stoechas (L.) Moench
in a coastal protected area in Tuscany
Gemma Giannetti
Supervisors:
General Index
1 Introduction...3
1.1 Sand dunes...4
1.1.1 Geomorphology, Structure and Dynamics...4
1.1.2 Conservation status of the dune habitats...7
1.1.3 Vegetational framework...9
1.2 Pollination...15
2 Materials and methods...17
2.1 Study species...17 2.1.1 Phytophagous Insects...20 2.2 Study area...21 2.3 Data collection...25 2.3.1 Insect observations...29 2.4 Data organization...31 2.5 Statistical analyses...34
3 Results...36
3.1 Helichrysum stoechas...36 3.2 Insects observations...454 Discussion...46
4.1 Effects acting on the reproductive success of H. stoechas...46
4.2 Fruit set...51
4.3 Phytophagous insects...57
5 Conclusions...58
1 Introduction
Sandy beaches and dunes occupy an important role in coastal ecosystems. They occur in places where waves and wind move sand grains towards the land. They provide coastal protection, buffering tides and waves, which may be particularly important where relative sea level is rising, and during storms. They support a fauna and flora especially adapted to the habitat. They are found almost in all latitudes, covering ecological habitats which range from polar to tropical latitudes, and from deserts to tropical rain forests (Fig. 1) Despite this, they are among the most vulnerable and threatened habitats as a result of an excessive exploitation of natural resources, chaotic demographic expansion, and industrial growth. The present work gives a contribution to the knowledge and to the protection of the environmental heritage of the Migliarino - San Rossore - Massaciuccoli Regional Park, a protected area hosting one of the largest Italian sand dunes system. It is part of a wider project aiming to evaluate the impact of managed honeybees have on vegetation and entomofauna. This work focuses on the effects that Apis mellifera have on a plant that is well represented in sandy habitats, Helichrysum stoechas.
1.1 Sand dunes
1.1.1 Geomorphology, Structure and Dynamics
Dunes are littoral eolian landforms characterized by moderate elevations (from 1.50 m to 12 m above sea level) made up of loose, sand-sized sediment (grains with a diameter lower than 2 mm) (Audisio, 2002; Martínez et al. 2004). They are part of unique ecosystems which are at the spatial transition between continental/terrestrial and marine/aqueous environments. They differ from continental mobile dunes for the presence of vegetation blocking the progress of sediments inland mediated by the wind (Audisio, 2002). In fact, coastal sandy dunes are the result of the combination of the presence of vegetation, sediments and the action of the wind. Vegetation has a crucial role in the formation and stabilization of the coastal sand dunes. Perennial psammophilous grasses accumulate the sand transported by the wind around their leaves. This permits dunes to grow upward through accreting sand layers (Maun, 1998; Perrow & Davy, 2002; Acosta et al., 2015) (Fig.
2) .
Fig. 2. Schematic representation of the phases occurring during dune formation (Acosta et al., 2015).
They also allow to slow the erosive action of the wind and this in turn increases the sand accretion in the lee side of the dune (Ranwell, 1972; chapman, 1976; Perrow & Davy, 2002).
The whole littoral dunal system consists of:
• a beach (highly dynamic and lacking vegetation), • active foredunes and
• a resting zone (stable dune) (Psuty, 2004; Acosta et al., 2015) (Fig.3).
Sediment is supplied:
• by longshore drift from eroding headlands, • cliffs and other dune systems
• by rivers and
• by the (deeper) sea bottom (Van der Meulen & Salman, 1996).
In Italy the entire coastline is about 7500 Km, 3000 of which are represented by sandy beach-dune systems (Fig. 4). Unfortunately, as a consequence of human exploitation, few of them are well conserved and present the classic zonation previously described (Audisio, 2002).
Fig 3. Schematic zonation of sandy coasts. Are also represented some chemical and physical gradients acting on dunal ecosystem (Acosta et al., 2015).
The growth of human populations and economies constitute the ultimate pressure acting on coastal environment (Schlacher et al., 2014). In Italy, among the most important causes damaging this ecosystem there are coastal erosion, soil exploitation, pollution and tourism (WWF Italia, 2014). The most critical situations are the Sicilian and Sardinian coasts, heavily compromised due to uncontrolled building, and all the Adriatic coast, in which only the 30% of the coastline is free from an urbanization process (WWF Italia, 2014).
Fig 4. Distribution of beach-dune systems in Italy (Audisio, 2002).
1.1.2 Conservation status of the dune habitats
Dunes are of great importance to society as prime sites for housing and recreation, buffers against storms, and providers of fisheries and mineral resources (Schlacher et al., 2014). Contrary to common perception, beaches and dunes contain a diverse and unique set of species, many of them found nowhere else. In addition a lot of wildlife species (e.g. birds, turtles, fishes) are dependent on dunes and beaches for nesting and feeding (Schlacher et al., 2014). Nevertheless sandy dunes ecosystems are extremely threatened all over the world. Threats to beaches arise from a range of stressors which span a spectrum of impact scales from localised effects (e.g. trampling) to a truly global level (e.g. sea-level rise), but the major one comes from the interaction of human coastal development and climate changing (Defeo et al., 2009; Schlacher et al., 2014). Urbanization, industrialization, touristic activities, farming practices, erosion and pollution are the main factors originating such an alteration (Cori, 1999; Brown & McLachlan, 2002; Van der Maarel, 2003; Coombes et al., 2008; Carboni et al., 2009; Gornish & Miller, 2010; Miller et al., 2010; Ciccarelli et al., 2012).From a report on the state of conservation of EU species and habitats under the European Directive 92/43/CEE, dune habitats resulted to be the most endangered. They present the highest assessments marked as inadequate” and “Unfavourable-bad” (European Commission, 2015). The status “Unfavourable-inadequate” indicates a necessity of changing in the conservation management actions even if there is not an imminent risk of extinction and “Unfavourable-bad” describes the presence of a serious danger of extinction (Acosta & Ercole, 2015). Also at a national scale dunes habitats seem to be the most threatened (40% are “Unfavourable-inadequate” and 46.7% are “Unfavourable-bad”) (Biondi & Zivkovic, 2014). The major sources of risk are erosion, urbanization and transport infrastructure (Biondi & Zivkovic, 2014). Pressures come mainly from activity linked to seaside tourism, invasion from alien species, erosion and other modifications of natural balance (Acosta & Ercole, 2015).
Nevertheless, in Tuscan coastline there are many portions with a high level of biodiversity that are preserved, notably Maremma Regional Park (25 Km of protected coastline) and
Migliarino-San Rossore-Massaciuccoli Regional Park, with 35 Km of protected coastline having different degrees of protection (Vagge & Biondi, 1996).
1.1.3 Vegetational framework
The coastal environment is characterized by a restricted availability of space and particular abiotic factors causing a condition of high stress for living beings hosted in this ecosystem. These are the causes making pressure for the selection of highly specialized plants and animals, very often present just in the coastal habitat and not in the continental environment (Audisio et al. 2002; Ciccarelli, 2014).
From a vegetation point of view in Migliarino-San Rossore-Massaciuccoli Park the typical vegetative chain succession of sandy dunes can be found. It is determined by a gradient of ecological factors, such as wind, salinity and dryness and mobility of the soil (Biondi, 2007) (Fig. 3). On this basis, it is possible to divide the coastal sand dunes ecosystem into three main habitats that host dissimilar plant communities: the foredune, the interdune and the backdune (Psuty, 2004) (Fig. 3).
The foredune is defined as the portion of the beach-dune profile that extends from the mean tide line to the top of the frontal dune, including the upper beach, embryonic dunes and mobile dunes (Barbour, 1992). The main plant communities that occupy the foredune in San Rossore Park are (Arrigoni, 1990; Vagge & Biondi, 1999):
• Salsolo kali-Cakiletum maritimae Costa & Manz. 1981 corr. Rivas Martínez, Costa & Loidi 1992 constituted by annual halonitrophilous plants, such as Cakile maritima Scop. (Fig.5),
Salsola kali L., Chamaesyce peplis (L.) Prokh.
and Polygonum maritimum L. in the upper beach (Arrigoni, 1990; Vagge & Biondi, 1999). This is the harshest part of the beach in which plants are found. Salt spray, wind, incoherent soil and other stressing factors are the strongest; • Echinophoro spinosae-Elymetum farcti Géhu
1988, in which the main species are Elymus farctus (Viv.) Runemark, Echinophora
spinosa L., Achillea maritima (L.) Ehrend. Y.P. Guo, Medicago marina L., Calystegia soldanella (L.) Roem. & Schult. and Eryngium maritimum L. (Fig. 6)
Fig 5. Cakile maritima (picture by Gemma Giannetti).
(Arrigoni, 1990; Vagge & Biondi, 1999). These plants live in the embryonic dunes (Arrigoni, 1990; Vagge & Biondi, 1999);
• in taller mobile dunes, the so-called “white dunes”, are mostly found the
Echinophoro spinosae-Ammophiletum arundinaceae Géhu, Biondi, Géhu-Franck,
Taffetani 1987 association, represented by Ammophila arenaria (L.) Link, Eryngium
maritimum L. (Fig. 6), Euphorbia paralias L. (Fig. 7) and Pancratium maritimum L.
(Fig. 8) (Arrigoni, 1990; Vagge & Biondi 1999). A. arenaria is the main species that contributes to construction of dunes.
The middle portion of the coastal dunes is the interdune, in which we can find stabilized dunes, protected from the action of the wind by mobile dunes built by A. arenaria. Interdune is characterized by a chamaephytic garrigue, in particular, by the Pycnocomo rutifolii-Seseletum tortuosi Géhu et al. 1987 association. This type of habitat is dominated by H. stoechas, Lomelosia rutifolia (Vahl) Avino & P. Caputo, Seseli tortuosum L., Pancratium maritimum (Fig. 8).
Fig. 6. Eryngium maritimum (picture by Gemma Giannetti).
Fig. 7. Euphorbia paralias (picture by Gemma Giannetti).
In the more humid area of Macchia di Migliarino – Tenuta di San Rossore, is present also the tuscan coastal endemism Solidago litoralis Savi (Vagge and Biondi,1999). In the interdunal belt is also quite common the Vulpietum fasciculatae-Silenetum coloratae
(Pignatti 1953) Géhu & Scoppola 1984 community (Fig. 9), made up of Malcolmia
ramosissima (Desf.) Thell., Silene canescens Ten., Vulpia fasciculata (Forssk.), Ononis variegata L., Pseudorlaya pumila (L.) Grande, Cutandia maritima (L.) Barbey, Medicago littoralis Loisel., Lagurus ovatus L (Vagge & Biondi, 1999).
Relatively far from the sea, there is the backdune zone characterized by reduced wind and fixed dunes with a layer of humus. Here, shrubs can grow, such as Juniperus oxycedrus L. subsp. macrocarpa Sibth. & Sm., Pistacia lentiscus L., Phillyrea angustifolia L., Phillyrea
latifolia L., that all together form the Asparago acutifolii-Juniperetum macrocarpae Géhu
& Biondi 1994 (Vagge & Biondi 1999) (Fig. 9). This is a transition belt from a strict sandy beach environment to a more continental one (Arrigoni, 1990). Finally, the backdune environment is followed by a forest of Pinus pinaster Aiton and Pinus pinea L. that are residues of a reafforestation done since the latter half of the 18th century to protect
agricultural crops. In the part of the backdune, between juniper scrubland and pine forest, hollows permanently or seasonally filled with water can be present. Here, hygrophilous or
aquatic plants typical of marshes or plain forests live, e.g. Carex spp, , Potamogeton spp.,
Typha angustifolia L., Phragmites australis (Cav.) Trin. Ex Steud., Scirpus spp, Juncus spp.
and Utricularia vulgaris L. (Arrigoni, 1990).
Fig. 9. Interdune of marina di Torre del Lago. In the foreground there is there plant assosiation Vulpietum fasciculatae-Silenetum coloratae; in the backdune some Juniperus oxycedrus subsp. macrocarpa shrubs are visible (picture by Gemma Giannetti).
In Marina di Torre del Lago the vegetation series previously described is well defined (Fig. 9). This does not occur in San Rossore Estate (Fig. 10) because of the action of erosion. Here there is not enough space for the interdune belt, because the sea enters the beach up to the backdune line. In this way the landscape does not follow the strict psammophytes vegetation series, but becomes a mixing of coastal dune vegetation.
Psammophytes, i.e. plants adapted to live in dune systems, are subjected to particular stresses that let them to evolve some particular traits. The most important environmental stresses are:
• salt spray, produced by breaking waves;
• low nutrient and water availability due to the fact that sand soil is incoherent and it does not retain the water;
• the high temperatures reached by the sand;
• the substrate instability and sand burial (Hesp, 1991; Biondi, 2007).
In the foredunes these ecological stresses are particularly strong and decline markedly away from the beach.
Fig. 10. San Rossere estate dunes. The different plant associations previously described, here do not follow the classical succession. For example E. paralias and A. arenaria is close to J. macrocarpa (in the right). Also P. lentiscus is unusually found next to the see (foreground in the left) (picture by Gemma Giannetti).
The most peculiar adaptions of psammophytes that permit them to cope with the severe stresses of this environment are:
• succulent tissues to store water (for example Cakile maritima);
• pubescent epidermis (e.g. H. stoechas) with the aim of preventing excessive transpiration and plant warming;
• thick cuticle in order to reduce loss of water by evapotranspiration; • deep roots to withstand sand burial;
• annual habit to avoid the hot and arid summer period (Hesp, 1991).
In general, as a result of the selective factors typical of coastal environment, plants adapted to the coastal environment are rarely found in continental habitats. The coastal flora is not characterized by a great level of endemism, nevertheless it has a high conservational value precisely because it is very specialized (Audisio et al. 2002). Despite this, along the Tuscany coasts three edemic specieis are present: S. litoralis, Centaurea aplolepa Moretti subsp.
1.2 Pollination
The pollination of flowering plants by animals represents a critical ecosystem service of great value to humanity, both monetary and otherwise. Honeybees (Apis mellifera Linneaus, 1758) are critically important for crop pollination worldwide (Klein et al. 2007), and the yields of some fruit, seed and nut crops can decrease by more than 90% without these pollinators (Southwick and Southwick 1992). Less investigated is the role that managed honeybees have on the pollination of wild
plants. Few studies have focused on the effects of managed honeybees on the pollination of spontaneous plants in parts of the world in which honeybees are not native, reporting either positive effects (e.g., Gross (2001) in Australian woodlands; Chamberlain and Schlising (2008) in Californian savannahs), no effects (e.g., Dupont and others (2004) in sub-alpine deserts of the Canary Islands, Spain), or even negative effects (e.g. Kato et al., 1999) when compared to the performance of wild pollinators.
Nowadays A. mellifera is spread worldwide, but before European settlers brought them in the New World continents, they were found
in Europe, Africa and Near and Middle East (Garnery et al., 1992). Morphological, paleogeographical (Ruttner et al., 1978) and mitochondrial DNA analyses (Garnery et al., 1992) agree that A. mellifera originated in Asia and evolved according to three different lines: the African subspecies (Fig.11 branch A), west Mediterranean subspecies (Fig.11 branch M) and north Mediterranean subspecies (Fig.11 branch C).
Fig. 11. Hypothesis for the origin and evolution of
current Apis mellifera populations (Garnery et al., 1992).
The simplest hypothesis is that the branch M results from an initial westward progression from Asia north to some obstacles represented by the Alpine arc, the Black Sea, the Caucasus and the Caspian Sea whilst the other two branches progressed more southward (along the coast of the Persian Gulf). This leads to the location of the initial centre of dispersion of the proto-mellifera species in the Middle East (Fig. 11). Another separation should have occurred when the species came to the Mediterranean Sea, northern populations skirting it by the north (branch C) and southern populations invading Africa. Once separated by distance and natural obstacles, populations diverged progressively (Garnery et al., 1992). Honeybee colonies are declining in many parts of the world (Delaplane and Mayer 2000). This is mainly attributable to:
1. the spread of pests such as parasitic mites (Varroa jacobsoni Oudeman, V. destructor Anderson & Trueman) and Acarapis woodi Hirst; the small hive beetle (Aethina
tumida Murray) and the microsporidian parasite Nosema ceranae n. sp.;
2. improper pesticide and herbicide use;
3. ageing of the beekepers’ population, especially in Europe and North America;
4. low market prices for their products and services. The number of beekeepers has declined, and so the number of colonies being kept over most of Europe and North America (Cayuela et al. 2011).
5. the Colony Collapse Disorder. Over the winters of the 2006 and the 2007, there have been large-scale, unexplained losses (30-90%) of Apis mellifera colonies in the United States (Evans et al., 2009; Ellis et al. 2010). In the absence of a known cause, this syndrome was named Colony Collapse Disorder because the main trait was a rapid loss of adult worker bees. Causes of this syndrome remain unresolved, but it may involves an interaction between pathogens and other stress factors (Evans et al., 2009).
2 Materials and methods
2.1 Study species
Helichrysum stoechas L. Moench belongs to the wide family of Asteraceae, more in detail
to the tribe of Gnaphalieae (Anderberg, 1991). The term “helichrysum” comes from the Greek “helios” and “chrysos” that mean respectively “sun” and “gold” and refers to the form and colour of the characteristic capitula of many species belonging to this genus (Ansaldi, 1994). The genus Helichrysum Mill. is one of the most numerous genera of Asteraceae, it covers 600 different species (Anderberg, 1991; Bayer et al. 2007) and it is distributed throughout Africa, Madagascar, the Mediterranean basin, Macaronesia, western and central Asia and India (Anderberg, 1991). In the Mediterranean region and western central Asia 41 taxa of Helichrysum, including subspecies are present (Galbany-Casals et al., 2004). They have a southern African origin and the whole Mediterranean-Asiatic group derives from a unique ancestor that had reached the Mediterranean region (Galbany-Casals et al., 2009) with the general African migration of flora of the Pliocene (5,300-2.5 Ma) and Pleistocene (2.5-0,012 Ma) (Quézel, 1978). In Italy there are six endemic species of
Helichrysum, found in particular in Sicily: Helichrysum errerae Tineo, Helichrysum hyblaeum Brullo, Helichrysum nebrodense Heldr. ex Guss. Helichrysum litoreum Guss. Helichrysum panormitanum Tineo ex Guss, Helichrysum pendulum (C.Presl) C.Presl; and
three endemic subspecies (Helichrysum italicum (Roth) G.Don subsp. pseudolitoreum (Fiori) Bacch., Brullo & Mossa, Helichrysum saxatile Moris subsp. morisianum Bacch., Brullo & Mossa, Helichrysum saxatile Moris subsp. Saxatile (Peruzzi et al., 2014).
H. stoechas (fig.12-13) is a chamaephyte with
tomentose leaves, growing from 15 to 30 cm in height. It has a strong aromatic odor. Woody stems, covered by white hairs, form a dense basal cushion. Leaves too are covered by a silver indumentum and they have revolute margins. The extremities of the erect flower-bearing stems bear sets of small golden yellow capitula (5-10) grouped in corymbs (Fig.13). Each capitulum is surrounded by yellow-brown scales and composed by yellow tubulous flowers (Pignatti, 1997).
It is a steno-Mediterranean plant distributed in the Mediterranean basin up to the Atlantic coast of France (Guinochet & Vilmorin, 1973). It is found in a variety of habitats, such as coastal dunes, mediterranean scrubland, garrigues, road embankments or abandoned fields (Pignatti et al., 2001).
Fig 12. H. stoechas plants in Torre del Lago that have faded florets (picture by Simone Anzà).
Fig. 13 H. stoechas corymb with florets
Many species belonging to the genus Helichrysum are employed in folk medicine as expectorants, antipyretic, antimicrobial, antiviral, diuretic, and for snake bites and sciatica (Font Quer, 1981; Rios et al., 1991; Meriçli et al., 1992). These properties are presumably conferred by secondary products contained in their essential oils (A.C.U. Lourens et al., 2008). In fact, club-shaped glandular trichomes that produce these important oils are scattered throughout the plant (Fig. 14): they can be found in the abaxial surface of leaves, on the involucral bracts of the capitula, on the corollas of the flowers and on the ovary surface (Ascensão et al. 2001, Rodrigues et al. 2015). In particular, there are two different types of glandular trichomes: the type I made up of 10-14 cells arranged into two series that exhibit a stalk of variable length and a head with three or four pairs of secretory cells. The apical pair of cells produces a subcuticular space, where secretion is temporarily accumulated, being released by cuticle rupture. The other kind of glandular trichomes, the type II, is formed of a stalk and some glandular cells that are smaller and with a narrower cuticular space compered to type I (Ascensão et al., 2001; Rodrigues et al., 2015).
Fig. 14. SEM micrographs of H. stoechas leaf. 1) Cross section of a young leaf showing a dense mass of
non glandular trichomes on both epidermal surfaces. Glandular trichomes are scattered among this non glandular indumentum just in the abaxial surface (arrows). 2) A glandular trichome after the cuticle rupture (arrows) (picture by Ascensão et al., 2001).
2.1.1 Phytophagous Insects
Among the most important abundant components of the coastal dune fauna there are insects (McLachlan, 1991). Phytophagous insects can be present in the coastal environment. Louda (1982) describes as plant consumers insects belonging to the orders of Diptera, Lepidoptera and Thysanoptera. Rivosecchi (2009) reported the presence of phytophagous Diptera on sandy coastal dunes. Also Lepidoptera feeding on psammophilous plants, such as
Pancratium maritimum are present in the Italian Ionic coast (Arpaia, 2008). Many are the
studies occurring in Italian coastlines revealing the presence of phytophagous beetles (Contarini, 1992; Fattorini & Carpaneto, 2001; Audisio et al., 2002; Carpaneto et al. 2006). Carpaneto et al. (2006) reported in a coastal protected area of Lazio, the presence of phytophagous and antophagous species of Coleoptera. Among them there are three species that are very common in the south-central Italy: Pentodon bidens Pallas, 1771 (Dynastidae),
Oxythyrea funesta Poda, 1761 and Netocia morio Fabricius, 1781 (Cetoniidae). They are
not exclusive of the coastal environment. Cetoniidae mostly feed on Asteraceae in this protected area.
I wanted to check if some phytophagous insects are present in my study area, at which species they belong and if they can affect the reproductive success of H. stoechas.
2.2 Study area
The study area is within the Regional Natural Park “Migliarino, San Rossore, Massaciuccoli”, a protected area located in the provinces of Pisa and Lucca (Tuscany, Italy). It faces the southern part of the Ligurian Sea. The River Arno, the longest river of Tuscany, and the River Serchio are the most important rivers flowing across the park. It is characterized by a Mediterranean sub-humid climate, according to Rapetti (2003): its mean annual temperature is around 15° C and the mean rainfall is 800-900 mm.
The littoral drift runs northwards on the right side of the River Arno delta and southwards
on the left side (Pranzini, 2001; Bertoni et al., 2012). The River Arno contributes the most to a coastline of about 30 Km characterized by dunes formed during period of intense progradation (Ruocco et al., 2014). However, those dunes have experienced a strong coastal erosion for about 150 years. The northern and the southern sectors of San Rossore are accreting (about 2 m per years), whereas the central sector is still subjected to a retreat of
about 4 m per year (Cipriani et al. 2004). The coastal dunes of San Rossore belongs to the Natura 2000 network with two Sites of Community Importance (SCI), i.e. “Selva Pisana” (IT5160002) and “Dune litoranee di Torre del Lago” (IT5170001) (http://www.parcosanrossore.org/pagina.php?id=92).
The areas of survey of Migliarino - San Rossore Park are two: Marina di Torre del Lago (hereby, TDL) and San Rossore Estate dunes (hereby, SRO).
TDL is a locality made up of a wide accreting sandy beach. In this part of the park, people can easily access to the beach through some boardwalks limiting the access to dunes. Despite this, it is subjected to intense human disturbance activity, such as trampling. On the other hand, SRO is a strict protected area in which the entrance of people is not allowed. It is characterized by a strong erosion process and the continuous degradation of dunes is due to a decrease in sediment supply by the Arno river (Anfuso et al. 2011). Various mechanisms have been hypothesized as causes of the decrease of the beach, such as extraction of sediments from the riverbed, stabilization of the river banks, reduction of agricultural activities and reforestation (Billi and Rinaldi, 1997).
The park grants, under an agreement, the right to some agricultural holdings to raise honeybees. For this reason, many are the apiaries present in the territory that provides honey of different type. For our study we selected two of these apiaries because they are close to dunes and to the study species. They belong to the agricultural holding “Sapori Mediterranei” of Mrs. Donatella Baldi. They produce the so-called “Beach honey”, a very particular honey with a distinctive taste given by the essential oils of H. stoechas. One apiary is present in TDL and the other is located in SRO.
Fig. 16. Marina di Torre del Lago (TDL) (picture by Gemma Giannetti).
Fig. 18: Migliarino, San Rossore, Massaciuccoli Regional Park map
2.3 Data collection
In each fieldwork locality, I set up one transect on the foredune and on the interdune, perpendicular to the coastline, at the nearest point to the apiaries (Tdl0). I then set up further transects at 500 m intervals, as follows (Fig. 19):
• 3 transects in TDL (Tdl0, Tdl500, Tdl1000),
• 6 transects in SRO (SRo0, SRo500, SRo1000, SRo1500, SRo2000, SRo2500), • 1 in a site 5000 m far from the beehives in SRO, corresponding to the control site
(Sro5000).
with a total of 10 transects. Each transect represents a sampling site.
The control site is located at 5000 m distant from the apiary because it was expected to have absence of honeybees or at least a very reduced foraging activity due to the fact that the maximum bees foraging distance estimated is of ca. 2000 m (Couvillon, 2014; Garbuzov, 2014). Despite this, it has been recorded that honeybees under extraordinary conditions, such as low availability of resources or a patch particularly abundant flowering plants, can cover a foraging distance up to 12 km (Ratnieks, 2000).
The length of transects is about 100 m in TDL and 40 m in SRO, depending on the depth of dune system that is approximately 50 m in SRO area, due to the erosion process, and 200-220 m in TDL (Ciccarelli, 2014).
Unfortunately, it was not possible to have the same number of transects in both sampling localities for many reasons. Firstly TDL is smaller (about 1.4 Km in width) than the SRO (about 5 Km in width), as a consequence of the fact that beach establishments interrupt the dune system. Furthermore, transects were settled southward from the apiaries due to the fact that the mouth of the Serchio River causes a change in the erosion pattern of the coast that did not allow locating any site northward the apiary present in SRO. In order to keep the
Fig. 19. Overview of the fieldwork area. Green points are transects, red points represent hives. Pictures a and
experimental design homogeneous between the two localities, I decided not to select any site also northward from TDL apiary.
Sampling period ranged from 27-06-2015 to 08-07-2015. In every transect I collected, with a random stratified sampling method and right before the seed dispersal, 80 capitula potentially bearing fruits. To verify if there is a plant size effect on the production of fruits, sampling was conducted considering two different categories: small plants (from 10 to 40 cm in diameter of the cushion of the plant) and large plants (40 to 70 cm in diameter of the cushion of the plant) (Fig. 20). So, I sampled a total of 80 capitula, 40 belonging to the category small plants and 40 to the category large plants. Each capitulum was put in a 2 mL Eppendorf tube, and the tubes were put in pre-labelled plastic bags. During collection in the fieldwork plastic bags were preserved in a refrigerated bag.
Labels included: Sample type: “Fruits”;
Site and transect code: TDL/SRO and 0/500/1000...; Plant size: large/small;
Date of collection.
Bags were then left in laboratory for a few days to let all fruits reach full ripening stage, and finally stored at -20° C in a freezer (Fig. 21)
2.3.1 Insect observations
I conducted surveys focused on insects found on the target species during the flowering season. On 29/04/2016 and on 03/05/2016 I set out a transect 200 m long parallel to the sea in the interdunal belt in which the H. stoechas plants are the most present (Fig. 22). Along the transects the number of coleoptera found on H. stoechas plants was recorded. This sampling was provided both in TDL and SRO. In addition, I collected approximatively 40 Coleoptera individuals in both TDL and SRO for identification. These samples were collected in a tube with an alcohol swab and then stored at -20° C in a freezer. Finally they were sent to the Department of Agriculture, Food and Environment of the University of Pisa in order to be identified.
2.4 Data organization
After the sample collection, a subsequent laboratory step was performed. I counted the number of ovaries and fruits present in every capitulum. I conducted the tally under a stereomicroscope (Swift Stereo 80) in order to safely distinguish ovaries and fruits.
In particular, ovaries and fruits present in each capitulum were separated into three different categories: unfertilised ovaries, aborted cypselae and well-formed cypselae (Fig. 23). For this study, ovaries were considered as “unfertilised” (Fig. 23.1) when they appeared thin and withered. In this case, there was no trace of an eventual embryo development process. I identified as “aborted” (Fig. 23.2), cypselae being withered but not thin, in which there is evidence that the embryo development likely occurred but stopped for unknown reasons at a certain time. “Well-formed cypselae” (Fig. 23.3) refers to cypselae appearing round swollen and firm. These characteristics suggest that the seed development most probably occurred. Well-formed cypselae were subjected to cut test to check the presence in the seed of embryo (Baskin and Baskin, 2014).
Data regarding the count of ovaries were organized as follow:
Fig. 23: 1) Unfertilised ovaries. They are undeveloped, withered, empty and ochre and brown. 2) Aborted
achenes. Some developmental processes seem to have started but they are empty They are dark-brown. 3) Well-formed achenes. They are brown tending to red well developed, round and turgid.
During the tally, many larvae were found among the fruits of the H. stoechas infructescence. In most of the cases in which larvae were present, a portion or the total of fruits of the flower head were damaged. In particular they lacked the inner tissues of the cypselae, suggesting that presumably larvae feed on H. stoechas fruits. I was not able to ascribe to any category damaged fruits, so I decided to not count them.
The success of a plant species in nature largely depends on its ability to reproduce, that is to form ripe seeds (Went, 1950). A method to estimate the reproductive success of plants is the
fruit set (Shivanna & Tandon, 2014). It indicates the proportion of ovaries that set fruits. In
general, it is calculated by:
Fruit set = total number of mature fruits/ total number of ovaries
In many cases the number of ovaries is represented by the number of flowers and in many species it is possible to follow the development of a specific flower into fruit, in order to exactly know which flower develop into fruit. This is not possible for H. stoechas due to the reduced dimension of flowers that should force to sever flower heads to count the number of flowers. So in this study, the total number of ovaries is represented by the sum of “unfertilised ovaries”, “aborted cypselae” and “well-formed cypselae” found in each flower head. Moreover, I took into account not only the number of mature fruits, but also the amount of aborted cypselae, because together they represent the portion of ovaries in which a process of fertilization occurred, even if some of them do not became mature fruits.
So, in this study, I refer to fruit set as:
Fruit set = (number of aborted cypselae + number of mature cypselae)/total number of ovaries per flower head
2.5 Statistical analyses
I performed all the calculation with R Core Team (2016), language and environment for statistical computing.
I tested the effect of damaged achenes on the fruit set by fitting GLM.
In all my statistical computations the variable distance of sites from the apiary was considered as a proxy for presence of honeybees.
To test differences among sites I performed an Analysis of Variance (ANOVA) test, that compares means of distributions among different sites. I also run the ANOVA to check if there is a difference in the means between large plants and small plants. I applied the test taking into account as dependent variables:
• the total number of ovaries per capitulum; • the fruit set for each capitulum.
I used the Tukey's post-hoc pairwise comparison test to show which distributions differ from the other. TDL data and SRO data were analysed separately. For TDL area related to the variable distance there are 3 different levels (TDL0, TDL500, TDL1000), whereas for SRO area there are 7 sites (SRO0, SRO500, SRO1000, SRO1500, SRO2000, SRO2500, SRO5000).
I also conducted an explorative analysis trough linear regression to describe an eventual relationship between the production of ovaries per capitulum and the production of fruits per capitulum. Linear regression was performed between the variables y representing aborted cypselae and mature cypselae per capitulum and tot (the total number of ovaries per capitulum) (Fg. 25) at different distances. Record of site TDL0, TDL500, TDL1000 and SRO0, SRO500, SRO1000 were grouped together.
My hypothesis is that, if there is an effect of honeybees on the fruit set, there should be a negative relationship between fruit set and distance from the apiary. I tested this hypothesis by fitting Generalized Linear Models (GLM) with a binomial distribution and a logit link function. With GLM I also analysed the effect of size. GLM were performed on the overall data collected. Subsequently, TDL area and SRO area datasets were analysed independently.
I considered as response variables aborted cypselae + mature cypselae (y) and the total number of ovaries per flower head minus y (tot - y). Variable y represents the portion of ovaries that have been fertilized. Explanatory variables are represented by distance from the apiary (distance) and plant size (size) (Fig. 25). I fitted GLM:
1. considering distance and size both together ;
2. and separately, in order to avoid an eventual influence of one variable on the other.
GLM were fitted comparing the AIC (Akaike Information Criterion) of each model. The AIC is a criterion that numerically express the quality of a statistical model in respect of other models used to describe the same data set (Yamaoka et al., 1977). I decided to show just models with the lowest AIC value, so models that the best explain the relationship between y and explanatory variables.
Fig. 25. Example of dataset used for linear and
3 Results
3.1 Helichrysum stoechas
The total number of counted ovaries is 17,760. The majority of them are unferfertilised (11,840, = 66,7%), 4437 are aborted cypselae (25%) and the smallest fraction is represented by mature cypselae (1,483) (Fig. 26).
The percentage of damaged capitula gets to 19.7%. Anyway, consumers do not seems to have an effect on the fruit set (Table 1). The proportion of damaged flower head for each site can be seen in Fig 27. SRO sites are more attacked compared to TDL sites.
Fig. 26. Proportion of unfertilised ovaries, aborted
Presence of damaged capitula is not statistically significant in explaining fruit set values. Differences in the fruit set are better explained by place and apiary distance (Table 1).
Table 1.Results of linear models performed considering as response variable fruit set in each site (Adjusted
R-squared = 0.8871, p-value = 0.0009071).
Estimate Standard Error t value Pr(>|t|)
Intercept 2.561e-01 8.090e-02 3.166 0.01942 *
Proportion of damaged capitula
6.070e-02 2.951e-01 0.206 0.84384
Place (TDL) 3.094e-01 6.270e-02 4.934 0.00262 **
Apiary distance -4.089e-05 1.532e-05 -2.670 0.03705 *
The total number of ovaries per flower head is shown in table 2.
Table 2. Average number counted ovaries per capitulum and the three different categories of ovaries and
fruits.
General TDL SRO
Total Mean Total Mean Total Mean
Total 17760 23.648 ± 0.343 6653 24.281 ± 0.393 11107 23.285 ± 0.455
Unfertilised 11840 15.766 ± 0.374 2966 10.825 ± 0.538 8874 18.604 ± 0.453
Aborted 4437 5.908 ± 0.241 2493 9.099 ± 0.465 1944 4.075 ± 0.231
Well-formed 1483 1.975 ± 0.177 1194 4.358 ± 0.393 289 0.606 ± 0.127
The value of the fruit set is 0.341. In TDL it is higher than SRO (Table 3). This can be observed also taking in account single sites (Fig. 28).
Table 3. General fruit value, for TDL and SRO.
General TDL SRO
Fruit set 0.341 0.616 0.207
Concerning with distance from the apiary, in SRO all the ANOVA tests are statistically significant, so for every response variable (total ovaries, fruit set) at least one mean is different from the others. In TDL, means are significantly different just when is considered the total number of ovaries (Table 4).
Moreover, results showed that the effect of size in itself was not significant in all cases (Table 4).
Table 4. Results of ANOVA regarding the total number of ovaries and fruit set, both for TDL and SRO in
relation to distance.
Df SumSq MeanSq F value Pr(>F)
Distance TDL Total ovaries 2 266 132.88 4.33 0.0144 * Fruit set 2 0.037 0.018 237 0.789 SRO Total ovaries 6 732 122.00 3.71 0.00137 **
Fruit set 6 3.194 0.532 11.98 3e-12 ***
Size
TDL Total ovaries 1 8 8.070 0.254 0.615
Seed set 1 0.016 0.016 0.206 0.650
SRO Total ovaries 1 50 50.19 1.461 0.228
Fig. 29. Boxplots illustrating distributions of the total number of ovaries and fruit set considering sites, both
Linear regression models at different distances from the apiary can be seen in Fig. 31. It is possible to observe that the higher is the number of ovaries per capitulum the higher is the production of fruits. The effect of size is not homogeneous at all distances. Despite this there is not a particular difference in regression lines for small plants and large plants, except at distances 500 and 1000 (Fig. 31). Unfortunately, is not possible to appreciate the difference at distances 1500 and 5000 because. Low values of total number of ovaries correspond to capitula that have been subjected to parasitism, but they do not have en effect on the regression line slope.
Fig 30. Boxplots illustrating distributions of the total number of ovaries and fruit set considering the
size of plants, both for TDL (left side) and for SRO (right side). Letters represent the results of Tukey test.
Fig. 31. Plots illustrating linear regression models at different distances from the apiary. Red dots represent
large plants. Blue lines describe regression model considering just large plants. Red dots represent small plants. Red lines describe regression model considering just small plants. Black lines are regression lines of
Table 5. Generalized linear model (GLM) results for the effect of plant size and distance from the apiary on
the production of fruits, both mature or aborted, of the entire dataset, TDL and SRO.
Explanatory
variables Coefficient Standard Error z value Pr(>|z|)
General
Intercept 4.805e-02 2.872e-02 1.673 0.0943 .
Apiary distance -6.457e-04 2.008e-05 -32.154 < 2e-16 ***
Size-small -2.232e-01 3.305e-02 -6.754 1.44e-11 ***
AIC: 9515.5
TDL
Intercept 4.341e-01 4.686e-02 9.263 < 2e-16 ***
Apiary distance -8.818e-05 6.156e-05 -1.432 0.152
Size-small -3.511e-01 4.953e-02 -7.089 1.35e-12 ***
AIC: 4083.7
SRO
Intercept -6.823e-01 4.606e-02 -14.814 <2e-16 ***
Apiary distance -6.218e-04 3.159e-05 -19.680 <2e-16 ***
Size-small -3.943e-02 5.254e-02 -0.751 0.453
AIC: 3434
GLM results show that in the general case, distance from the apiary seems to have an effect on the production of fruits. This effect is negative, so closer to the apiary there is a higher production of fruits. For TDL data are not sufficient to guarantee a significant result for the distance from the apiary. In SRO distance have a negative significant effect. Plant size have a negative effect in all cases but it is not significant in SRO. So large plants produce more fruits than small plants.
In Table 6 I have analysed the effect the distance from the apiary and the size separately to avoid any eventual effect of one variable on the other. There is not any substantial difference in results between Table 6 and Table 5.
Table 6. GLM results considering distance and size analysed separately. Explanatory
variables Coefficient Standard Error z value Pr(>|z|)
Distance General
Intercept -6.261e-02 2.368e-02 -2.644 0.0082 **
Apiary distance -6.461e-04 2.028e-05 -31.865 <2e-16 ***
AIC: 9559.2
TDL
Intercept 2.631e-01 3.986e-02 6.601 4.09e-11 ***
Apiary distance -8.928e-05 6.132e-05 -1.456 0.145
AIC: 4132.1
SRO
Intercept -7.008e-01 3.895e-02 -17.99 <2e-16 ***
Apiary distance -6.221e-04 3.159e-05 -19.70 <2e-16 ***
AIC: 3432.5 Size General Intercept -0.62738 0.02204 -28.466 < 2e-16 *** Size-small -0.13601 0.03188 -4.266 1.99e-05 *** AIC: 10904 TDL Intercept 0.38923 0.03478 11.192 < 2e-16 *** Size-small -0.35133 0.04952 -7.094 1.3e-12 *** AIC: 4083.8 SRO Intercept -1.32794 0.03539 -37.522 <2e-16 *** Size-small -0.05835 0.05135 -1.136 0.256 AIC: 3858.3
3.2 Insects observations
I observed 523 individuals of insects (Table 7) on H. stoechas flowering summits.
The identification revealed that all individuals belong to genus Stilbus Seidlitz, 1872 of the family Phalacridae Leach 1815, order of Coleoptera Linneaus 1758, they are also known as shining flower beetles (Gimmel, 2013).
I found that in SRO there was a higher presence of Stilbus consumers than TDL (Table 7).
Table 7. Number of Coleoptera found in each location.
TDL SRO
4 Discussion
4.1 Effects acting on the reproductive success of H. stoechas
Asteraceae are a very huge and globally distributed family and three kinds of pollination processes contribute to the reproduction of these plants: anemophily, entomophily, and orni-thophily (Cerana, 2004). The entomophily constitutes the vast majority of presumed or documented pollination types. Among the most important pollinators there are insects belonging to the orders of Hymenoptera, Lepidoptera, Diptera, Coleoptera and Hemiptera (Esten & Thorp, 1975; Cerana, 2004; Grombone-Guaranti et al., 2004; Janšta et al., 2015). A singular species can be pollinated by all or some of this kind of pollinators; it is the case of
Jurinea cyanoides (L.) Rchb. (Janšta et al., 2015), or Mikania spp. (Cerana, 2004), or Pyrrhopappus carolinianu (Walter) DC. (Esten & Thorp, 1975). Also on H. stoechas I
observed many species belonging to all the orders previously cited. A portion of them could presumably act as pollinators of H. stoechas, but insects can also be just visitors of a flower or a floret. In Helichrysum bracteatum (Vent.) Andr. and Helichrysum viscosum Sieber ex Spreng, two endemic Helichrysum species of Australia, a particular phenomenon occurs: ants defend their capitula from seed predators to forage on extrafloral nectaries (O'Dowd & Catchpole, 1983). Anyway, the ant behaviour does not prevent damage of flowerheads by seed predators or increase the reproductive output of H. bracteatum and H. viscosum (O'Dowd & Catchpole, 1983). In any case, there are some studies reporting that A. mellifera forages on Helichrysum species (Golding et al. 2001; Merti, 2003).
Statistical analyses demonstrate that in general honeybee activity increases production of fruits in H. stoechas in my study areas, although with differences between SRO and TDL (Table 5). The effect of distance from the apiary on the reproduction is low (-6.457e-04) and it is weaker than the effect of size (-2.232e-01). The effect of size is proven in some cases also by linear regression models. For SRO, the difference among distributions according to fruit set revealed by ANOVA reflects the output given by GLM.
These results are in accordance with the study of Cayuela et al. (2011) that investigated the effect of honeybees on wildcherry (Prunus avium L.), hawthorn (Crataegus monogyna Jacq.), and bilberry (Vaccinium myrtillus L.) present in a Spanish traditional agroecosystem. Pollination mediated by honeybees enhances the reproductive success of two of the studied species, even if the effect of distance is visible just for P. avium.
Many other studies on the impact of honeybees on native plants and pollinators are focused on areas in which A. mellifera is not native. In fact, in California for example, managed honeybees play an important role in the pollination of the native Triteleia laxa Bentham (Chamberlain & Shiling, 2008). In Dillwynia juniperina Lodd., a native legume of Australia,
native bees could on their own adequately service flowers in some years at some sites while at other times introduced honeybees may be necessary to augment pollination services (Gross, 2001). In other cases, the presence of honeybees does not influence the pollination of native plant species, even if they stay longer and visit more flowers on the same inflorescence than native bees, thus potentially promoting self-pollination of the plants (Dupont et al., 2004). On the other hand, honeybees reduced fitness in Melastoma affine D. Don in montane tropical-rainforest systems in Australia (Gross and Mackay, 1998).
Benelli et al. (2017) conducted a study in a coastal environment of Tuscany on Anthyllis
barba-jovis L., a Mediterranean spontaneous shrub. They highlighted the pivotal role that
honeybees have for this plant, not only because they are the most abundant pollinator but also because plant prevented from entomophilous pollination show inbreeding depression. Other studies investigate the effects of A. mellifera on native pollinators in terms of overlap in resource use between honey bees and native bees; the change in visitation rates of native bees; and the change in the levels of resource harvested by native bees when honeybees are present (Goulson 2003; Paini, 2004). Honeybees may have the potential to impact negatively on native bees, but this is not always true. In the Bonin Islands, a group of Pacific Japanese islands rich in endemic plants, honeybees result to be really detrimental because they compete for floral resources with native bees and enhance the spread of weedy species (Kato et al., 1999).
In Europe and in the Mediterranean region, in which honeybees are native they have a weak or a negligible effect on wild pollinators (Pechhacker & Zeillinger, 1994; Paini, 2004; Forup & Memmot, 2005; Goras et al., 2016). Nonetheless, Shavit et al. (2009) found that in some
cases in two protected areas in Israel, honeybees had a negative effect on visitation rates of the other bees, while in other cases they did not find such an effect. Even so, they recommend to avoid the introduction of honeybees in Israeli natural reserves as a precaution aimed to protect the native flora and bee fauna. Also Elbgami et al. (2013) reported that
proximity to managed honeybee hives was associated with significantly reduced fitness of
bumblebee colonies. Another study supports the hypothesis that high honeybees densities may have a negative impact on other pollinators via competition for flower resources (Torné-Noguera, 2014). Anyway the impact of honeybees on wild pollinators is complex and it depends on many different factors (i.e. floral resources, foraging behaviour, population density, fecundity) (Goras et al., 2016).
The efficiency of honeybee in pollination of crop plants compared to wild pollinators has been investigated in many studies. For example, Pisanty et al. (2013) demonstrates that confection sunflower crops in Israel rely entirely on the presence of honeybees for seed production, native insects have a very scarce contribution. In Parthenium argentatum Gray, (Asteraceae), a rubber-producing plant, individuals exposed to a high abundance of honeybees produce 150% more seeds than plots without honeybees, indicating that honeybees have a strong positive effect on the pollination of this plant (Mamood et al., 1990). On the other hand, the development of strawberry depends both on wild pollinators and honeybees, as a consequence of their behaviour. The former visit mainly the basal region and the area next to the stamens, the latter the apical region of the receptacle (Chagnon et al., 1993). In Wisconsin wild bees play a significant and unique role in apple pollination and they cannot therefore be replaced by managed bees (Mellinger & Gratton, 2015).
ANOVA results show different a situation between TDL and SRO study areas concerning the fruit set distributions; the former is more homogeneous, the latter presents differences among means. It may reflect the physical dynamical processes of the coastline. The severe erosion (Ciccarelli et al., 2012) in SRO could strongly affect the good condition and resource availability necessary to H. stoechas to grow and produce flowers and fruits. A study of Ciccarelli et al. (2012) conducted on the dunal systems of Migliarino- San Rossore-Massaciuccoli Regional Park stresses on the fact that erosion causes an instability and a
strong disturbance of plant communities that results in an unstable equilibrium. In TDL this problem does not occur because it is an accreting zone. Nevertheless, in TDL data are not enough to understand if there is an effect of the distance from the apiary. Moreover, in SRO it is possible to see a gradient of erosion from the southern part of the mouth of Serchio River, corresponding to the site SRO0, southward until to the northern part of the Morto Nuovo River (Fig. 33). Then south to it there is again an accreting phenomenon, corresponding to SRO5000. The erosion gradient correspond to a fruit set gradient: in less eroded sites (SRO0, SRO500) there is a higher fruit set values. Nevertheless, also SRO5000 is not subjected to erosion process but fruit set is really low. This corroborates the hypothesis that honeybees give an important contribution to the pollination of H. stoechas. Anyway, more investigations and statistical analyses should be taken for future research.
Fig. 32. Erosion process in SRO. Plant associations are disrupted and some plants of H. stoechas are falling from the top of the dune.
Not only distance from bee hive, but also plant size has an effect on the reproductive success of H. stoechas: large plants produce more seeds than small plants do. Results like these were observed also in the perennial Ipomopsis aggregata (Pursh) V. Grant: there is a positive correlation between seeds per expanded fruit and size (Jong et al., 1992). We could suppose that larger plants can be more attractive to pollinators, but this is not the case of I.
aggregata (Campbell et al. 1991). In Cynoglossum officinale L., for example, this
phenomenon depends on the light condition in which plants grow: in shaded populations plant size effect on pollinator visitor is strong, instead in unshaded condition it is less pronounced (Klinkhamer et al., 1989).
4.2 Fruit set
Considering the total fruit set, it resulted to be 0.341. It includes not only mature fruits, but also aborted achenes. Compared to other Asteraceae (Burd, 1994; Byers, 1995; Wagenius, 2004) it is a very low value. It is lower also compared with a very similar species,
Helichrysum italicum subsp. picardii (Boiss. & Reut.) Franco; in fact its fruit set value is
about 0.680 in the coastal environment of the Doñana National Park (Herrera J.,1987). Measuring the seed set of H. stoechas as the average number of seeds per inflorescence per capitulum the result is 0.835. In this case, it is higher compered to the seed set of A.
barba-jovis, a mediterranean plant living in a similar environment in Tuscany, because it produces
0.250 seeds per inflorescence (Benelli et al., 2017). But, fruit production percentages of species belonging to different families in the coastal mediterranean environment can differ a lot (Herrera J.,1987).
Among factors causing the failure of flowers developing into fruits there are pollination failure, abortion of young fruits due to genetic defects, microbial attack of fruits and seeds, and resource limitation (Shivanna & Tandon, 2014). Reduced reproductive success may be explained by a combination of them.
Pollination is the process of transport and deposition of pollen from the male part of a flower to a stigma (Wilcock & Neiland, 2002). Pollination failure resulted to be very frequent among plants. It can occur at various levels: during the process of release, transport and deposition of pollen on stigma. The process of failure has been attributed to pollen limitation, pollination limitation, and pollinator limitation. Pollen limitation occurs when pollen does not reach the stigma, whereas in pollination limitation pollen is deposited on the stigma but it is of inadequate quantity or quality for fruit or optimum seed set.
Moreover, pollination failure can occurs in three different steps of pollination:
• pre-dispersal. It concerns all mechanisms operating before the arrival of pollen on the stigma. For example, pollen can be lost by flower visitors at the anthers before transport or during transport. Pollination can also fail before pollen reaches the
stigma as a consequence of failure of microsporogenesis derived from chromosomal imbalance or environmental influences at the time of the development of the pollen; • dispersal
• Post-dispersal. Even if pollen reaches the stigma it can be not sufficient in quantity and in quality. Small amounts of pollen may not stimulate flower to set seed. Also the presence of heterospecific pollen leads to a failure in fertilization by conspecific pollen. Pollen belonging to other species may anyway initiate the fecundation process and induce chemical or physical inhibition of the stigma, this have the consequence to inhibit also the fertilization by pollen of the same species . Also the arrival of self pollen can halt fertilization by xenogenous pollen (Wilcock & Neiland, 2002).
Other more general factors influence pollination, some example are the lack of pollinators and the reduction of pollinators activity as a consequence of habitat disruption or fragmentation. The breeding system of plants themselves influences pollination success, failure is less intense in self-compatible species than self-incompatible. Floral traits, such as the reduction of petal size or the absence of floral rewards for pollinators, can induce a failure in pollination. Nectarless species are less competitive in terms of attractiveness for pollinators than nectariferous plants (Neiland & Wilcock, 1998), they often rely on other species of the community for the transport of pollen (Pellmyr, 1986). H. stoechas is a non-nectariferous plant (Petanidou et al., 2006), therefore its reproductive success can suffer for this characteristic. Nevertheless, during the sampling periods, I found many insects on this plant, more in detail Coleoptera, Diptera and Hymenoptera. It is also well known that population density and size have an impact on the pollen flow, in fact pollination success is reduced in more sparse populations (Kunin, 1997). Small populations appear less attractive to pollinators, resulting in a general decline in pollinators visitation according to size. This findings may explain the difference in seed set between TDL and SRO, respectively 0.554 and 0.201. Erosion in in SRO may affect the density of H. stoechas plants, causing a diminished reproductive success of H. stoechas in SRO.
Among factor influencing the production of seeds in plants that do not strictly concern pollination there is the presence of parasites or consumers of flowers, fruits or seeds (Janzen, 1971, Shivanna & Tandon, 2014). As a general rule, the number of ovaries in a plant should reflect the number of flowers (Janzen, 1971), but in my case they differ considerably. Some capitula lacked entirely or just in part ovaries. The lack of achenes often corresponded with the presence of a larva of an insect. In other cases there were no larvae but many ovaries were damaged as well. According to this, also the mean number of flowers (28.7) differs from the mean number of ovaries per flower head (23.6). Ignoring damaged achenes, the mean number of achenes per capitulum is 26.6. The percentage of damaged ovaries gets to 19.7%. Insects and more in general animals can influence the development of a flower into fruit by predation and consumption: they can eat directly flowers or they can consume undeveloped fruits (Janzen, 1971). The identification of Coleoptera found on H.
stoechas summits, revealed that these Coleoptera belongs to a genus that feeds on flowers
and pollen of Asteraceae (more details can be found below). Despite these findings, statistical analysis revealed that damages provoked by consumers do not negatively influence seed set. The production of fruits seems to be affected more by the distance from the apiary and by location, rather than the presence of insects.
However, hermaphroditic angiosperms produce more flowers than fruits and seeds, so fruit developmental process can be arrested also by mechanisms internal to plants (Janzen, 1971, Johnston,1991; Burd, 1994; Byers, 1995; Wilcock & Nieland, 2002). For example, plants can produce more flowers that do not develop into fruit just to attract pollinators, to selectively abort “poor quality” fruits or they may function just as pollen donors despite having a morphologically complete gynoeceum (Janzen, 1971; Guitian, 1993). An excess of flowers can be of potential use in the case of extraordinary resource availability or may provide a reserve supply in case of high mortality during the flowering period (Guitian, 1993).
Moreover, in many plants bearing flowers grouped in inflorescences not all ovaries get to maturation because the production of seeds or fruits depends on the position of flowers within the inflorescence (Medrano et al. 2000). Such within-inflorescence variation of the ovary reproductive success can be attributable to: a competition of ovaries among an inflorescence for resources, a non uniform receipt of pollen or for an intrinsic or
“architectural” reason simply linked to the position of flowers in the inflorescence (Guitian, 1994; Diggle, 1995; Medrano et al. 2000;). Furthermore, pollinators have the tendency to visit multiple flowers within an inflorescence creating an opportunity for self- pollination among flowers (geitonogamy). Self-pollination among flowers can reduce the pollen available for export to other plants (pollen discounting) and can increase the incidence of inbreeding depression for embryos and offspring. In this way, geitonogamy should cause a plant that displays a large fraction of its flowers simultaneously to have lower reproductive output than a plant that displays fewer flowers at once ( Harder et al., 2004).
The low seed set found for H. stoechas may be due in part to the fact that it is a perennial plant. In fact, it has been observed that perennial plants in general have lower seed set values compared with annual plants (Weins, 1984; Burd, 1994). There is a strong link between perennation, outcrossing and low seed set. On the other hand, annual plants have reproductive strategy that allow self-compatibility and show high seed set. Outbreeding is a genetic reproductive system that tends to produce high levels of heterozigosity, that is more likely found by long-living organisms having multiple reproductive opportunities during their life span. Presumably, the genetic load associated with such genetic systems greatly reduces the success of single reproductive efforts (Wiens, 1984). Conversely, annual plants need to invest higher reproductive efforts for survival of seed because they have a single reproductive episode before death. For this reason they present a reproductive strategy based on self-compatibility tending to minimize the risk of reproductive failure (Ehrlén, 2002). Another explanation for the higher seed set among self-compatible plants is that they can have a pollen supply by transfer of pollen within a single flower, by geitonogamy and from other plants. A self-incompatible plant strictly relies on external provision of pollen (Burd, 1994). Anyway the difference between perennials and annual plants is not so clear and not all long-living plants are strictly outbreeders (Ehrlén, 2002). We do not know whether H.
stoechas is self-compatible or self-incompatible, but we know that it is a perennial plant,
therefore is possible that this characteristic have a role in the low seed set found.
Many species experience fluctuating levels of pollen limitation and fruit set values within a season (Ashman & Baker, 1992; Burd, 1994; Medrano et al. 2000). For example, in
Pancratium maritimum L. the earliest opening flowers have a higher probability of setting
observed in Silene laxipruinosa Mayol & Rosselló by Guitián & Navarro (1996). This phenomenon, has been interpreted as an indication that the mobilisation of resources for fruit production occurs largely after pollination (Stephenson, 1981). According to this hypothesis, the first flowers to be pollinated receive the most resources for fruit production and therefore make a greater contribution to female reproductive success (Guitián & Navarro, 1996). On these basis, is possible to reinterpret the low value of seed set of H.
stoechas considering also the effect of the season. I sampled capitula containing fruits quite
close to the end of the dispersal period. In fact, in some sites I hardly found capitula having fruits not yet dispersed. Furthermore, both processes of flowering and fruit maturation have presumably been accelerated by temperatures occurred during the sampling period (Fig. 34)
Finally, another important factor provoking the discrepancy between the number of flowers and fruits produced by angiosperms is resource availability (Burd, 1994). The coastal dune environment is quite scarce in terms of nutrients and organic matter and this can negatively
Fig. 34. Minimum (blue) and maximum (red) daily temperatures in 2015 from S.Rossore meteo-idrological
influence the productivity of psammophytes. In addition, comparing seed set of the two fieldwork, SRO plants have a lower probability to set seed than TDL ones. This is likely due to the fact that SRO is affected by erosion, that results into a low availability of resources.
4.3 Phytophagous insects
Switching to Coleoptera found on H. stoechas corymbs, the identification revealed that they belong to the family of Phalacridae, also known as shining flower beetles (Gimmel, 2013), and genus Stilbus (Fig.35). Studies on this family show that Phalacridae feed mostly on fungi, a significant numbers are palynophagous on angiosperm flowers and at least one species feed on cycad pollen (Mifsud & Svec, 2001; Gimmel, 2013). In many cases adults are found on Asteraceae flowers and larvae develop in Asteraceae capitula or in aerial parts of other herbaceous plants (Grandi, 1951; Crowson, 1981).
The Phalacridae are morphologically a well-described taxon, but among the most poorly known taxonomically (Gimmel, 2013). They include almost 600 species divided into 52 genera (Gimmel, 2013). Species belonging to this family occur nearly worldwide in terrestrial environment,
with the exception of New Zealand and Chile. In Italy, Svec & Angelini (1996) described 25 species.
Anyway, I just observed the presence of this consumers on plants. In addition, statistical analysis demonstrate that parasites do not influence the production of achenes in this plant.