Plant Ecol (2012) 213:581–590 DOI 10.1007/s11258-012-0023-7

Seed production and pre-dispersal reproductive losses in the narrow endemic Euphorbia pedroi (Euphorbiaceae)

Ma´rio Boieiro • Carla Rego • Artur R. M. Serrano • Xavier Espadaler

Received: 18 January 2011 / Accepted: 12 January 2012 / Published online: 24 January 2012 Ó Springer Science+Business Media B.V. 2012

Abstract Euphorbia pedroi is a narrow endemic variables indicative of plant size and fecundity, and species with three known populations located in showing no consistency at the individual level. coastal areas of western Portugal. This study focused Specialist seed-wasps inflicted the highest losses to on the reproductive biology of this species from E. pedroi and their impact was intimately associated flowering to dispersal, aiming to identify the factors with the magnitude of yearly variation in seed causing decrease in seed production potential and to production. This finding highlights the role of the assess the spatio-temporal patterns of seed production inter-annual variation in seed production as a key at the individual and population levels. The abortion of feature in this plant-seed predator system. The effect reproductive structures, particularly seeds, repre- of the two groups of seed predators on the reproductive sented a major fraction of losses in the potential seed output of E. pedroi was additive and those do production of E. pedroi. larvae destroyed a not seem to exert an important selective pressure on variable proportion of cyathia in a large number of the traits studied. The proportion of intact seeds plants from the two populations regardless of their produced by E. pedroi differed between locations, but degree of isolation. Furthermore, generalist and spe- not between individuals within each population, cialist pre-dispersal seed predators were responsible highlighting the major contribution of larger plants for temporally variable seed losses unrelated with to the seed pool.

Keywords Andromonoecy Á Arra´bida Á Fruit and Electronic supplementary material The online version of seed abortion Á seed predation Á Reproductive this article (doi:10.1007/s11258-012-0023-7) contains success supplementary material, which is available to authorized users.

M. Boieiro (&) Á A. R. M. Serrano Centro de Biologia Ambiental, Faculdade de Cieˆncias da Universidade de Lisboa, 1749-016 Lisboa, Portugal e-mail: [email protected] Introduction

C. Rego The conservation management of a particular plant Azorean Biodiversity Group CITA-A, Universidade dos species can be much facilitated with a better under- Ac¸ores, 9700-851 Angra do Heroı´smo, Ac¸ores, Portugal standing of its reproductive biology. Reproduction is a critical stage in the life-cycle of plants, particularly for X. Espadaler CREAF and Unitat d’Ecologı´a, Universitat Auto`noma de those species where regeneration occurs exclusively Barcelona, 08193 Bellaterra, Barcelona, Spain through seeds. During plant reproduction, a variety of 123 582 Plant Ecol (2012) 213:581–590 abiotic and biotic factors interfere with the processes the reproductive output of E. pedroi and the consis- of fruit and seed production, leading to a reduction in tency of their attack at the individual level. the reproductive potential of individual plants. Amongst the abiotic factors, harsh atmospheric con- ditions have been frequently identified as a cause of Materials and methods reproductive losses in plants (Stephenson 1981). For example, coastal plants are subjected to strong winds, Study plant regular frosts and salt spray which usually damage a number of flowers and fruits and may even preclude Euphorbia pedroi Molero & Rovira is a narrow insect pollination (A˚ gren 1988; Rovira et al. 2004). endemic species of western Portugal restricted to three The destruction of flowers, fruits and seeds by populations along the coastline (Fig. 1). There, this is also a common cause of plant reproductive losses, species is restricted to the slopes of south-facing rocky which in some cases may be very severe and even cliffs being subjected to regular mist, strong winds and influence the dynamics and survival of plant popula- to high levels of insolation throughout the year. tions (Crawley 2000; Kolb et al. 2007). Seed predators Euphorbia pedroi is a caducifolious sub-succulent belong to a variety of groups, but many of them xerophyte that can attain about 2 m in height and more are inconspicuous insects that usually feed on one or a than 2 m width (Fig. 2a). This perennial andromo- few closely related host plant species and have their noecious species is included in the E. lamarckii life-cycle synchronized with the phenology of their complex showing affinities with Macaronesian den- hosts (Hulme 2002). The discrimination of the impact droid spurges. Annually, each adult plant produces inflicted by each seed predator is paramount since it cyathia, which are arranged in clusters at the top of allows the identification of patterns of interaction loose pleiochasial synflorescences. Each cyathium between predators and their host plants and may even comprises a single female flower with a three-lobed enable some predictability on subsequent reproductive ovary surrounded by some male flowers, but some losses. cyathia may contain solely staminate flowers (male In narrow endemics, all those particularities and cyathia) (Fig. 2b). During the flowering period a circumstances may be exacerbated by the isolation variety of insects visit the inflorescences in search of and/or fragmentation of populations (George et al. nectar or pollen and may also be engaged in pollen 2009). They are more vulnerable to environmental transfer between cyathia. Other insects, however, have hazards (Garcı´a 2008) and are at higher extinction risk a negative impact on the reproductive success of through loss of genetic variability resulting from E. pedroi. The larvae of Acroclita subsequana reduction or variation in pollination services (Srimu- (Herrich-Scha¨ffer) (), a moth species spe- ang et al. 2010), death of individuals by anthropogenic cific to the genus Euphorbia, develop inside the buds disturbance (Lo´pez-Pujol et al. 2005; Rucin´ska and and, as they grow, they join the adjacent cyathia Puchalski 2011) or ineffective recruitment due to together with the surrounding leaves, usually leading limited seed dispersal, rarity of safe-sites or seedling to the destruction of the whole cluster of cyathia. herbivory (Navarro and Guitia´n 2003). Therefore, life- During fruit maturation, seed-wasps—Eurytoma history parameters are a sound base to understand and fumipennis Walker (Eurytomidae)—are regularly promote the conservation of narrow endemics. In this observed ovipositing in full-sized fruits. This seed- study we investigate seed production and the causes of wasp species is specific to a few Euphorbia species pre-dispersal reproductive losses in the narrow and depends upon their seeds to complete its life cycle. endemic Euphorbia pedroi. More specifically, we Seed-wasp larvae grow and develop inside the seeds identify the factors responsible for reproductive losses feeding on their contents and there remain throughout from anthesis to seed dispersal and study the spatio- the winter until spring, when the adult emerges temporal variation in their magnitude. We also assess (Fig. 2c). Three species of hemipterans—Cydnus how the variation in the effect of each mortality factor aterrimus (Forster) (Cydnidae), Spilostethus pandurus relates with variables indicative of plant size and (Scopoli) (Lygaeidae) and Dolycoris baccarum (Lin- fecundity. Furthermore, we examine the differential naeus) (Pentatomidae)—have also been found punc- impact of specialist and generalist seed predators in turing the mature fruits of E. pedroi (personal observ.). 123 Plant Ecol (2012) 213:581–590 583

These bugs are generalist species that feed upon Plant fecundity and the production of fruits different structures in a variety of plant species, but usually prefer seeds. They puncture the fruits and In 2002, at the beginning of the flowering period, we inject saliva into the seed leaving it irreversibly tagged 25 randomly chosen individual plants of damaged. Seed dispersal in E. pedroi involves two Euphorbia pedroi in each population. During the different mechanisms with myrmecochory following flowering and fruiting periods (i.e. from March to explosive fruit dehiscence. July) we visited regularly the study areas to record the total number of cyathia produced, the number of Study sites cyathia that lacked ovary, the number of ovaries and fruits aborted and the number of cyathia damaged by This study took place in two of the three known moth larvae in all marked plants. This procedure was populations of this species: Azo´ia and Ares (Fig. 1). In repeated on the same plants in the two following years. Azo´ia, the number of E. pedroi individuals (around Male cyathia were easily distinguished from the 800–1,000) is much lower than that found in Ares, hermaphrodite ones. In male cyathia the pistillate where more than 3,000 plants are believed to occur. flower was generally absent, although a vestigial non- The study areas are located about 6 km apart and the functional pistillate flower was also occasionally habitats present similar characteristics. At both sites found. The data on male cyathia levels were analyzed we found cliffs and steep rockwalls, where E. pedroi as a form of reduction of the potential seed production, plants grow on rocky crevices, and less inclined slopes but taking in consideration that these results translate or even balconies, where soil pockets harbour a much the reproductive allocation strategy of plants we larger diversity of plant species. The vegetation is discuss these findings separately. Individual seed dominated by Pistacia lentiscus L., Quercus coccifera production was estimated by multiplying the number L., Rhamnus oleoides (L.) Jahand. & Maire and of intact fruits by three, the number of seeds enclosed Juniperus turbinata Guss. The habitat-type where our in each trilocular capsule (Berg 1990; Traveset 1995). study plant occurs is discontinuous along a short For each individual plant, we recorded plant height, course of the coastline, it is extremely localised and crown width and stem diameter (at 10 cm from the presents unique characteristics from the biological and trunk base) as attributes indicative of plant size, geological point of view (ICN 1997). These peculiar together with the distance to the nearest conspecific, in habitats are also equivalent to ecological islands to order to evaluate if there was an influence of these some other rare or narrow endemic plant species and, variables on seed production and pre-dispersal seed for this reason, were recently included in the predation levels. Portuguese Reserve System. Seed production and losses to seed predators

Each year, mature fruit collection from the selected individuals was regularly performed during the fruit- ing season with an increased sampling effort during the peak of fructification (in May–June). The fruits were brought to the lab and seeds were obtained by fruit dissection. A random sample of 100 seeds/ individual was then dissected under a stereomicro- scope to determine which were preyed, aborted or intact. Aborted seeds were readily recognized due to their whitish or yellowish colour, light weight and wizened appearance whilst the distinction between preyed and intact seeds was only possible after inspecting seed content. Seeds preyed by generalist Fig. 1 Map of the study area showing the location of E. pedroi hemipterans were left empty as a result of feeding populations (grey circles). 1 Espichel; 2 Azo´ia; 3 Ares whilst the presence of a larva inside the seed was 123 584 Plant Ecol (2012) 213:581–590

Fig. 2 An individual of Euphorbia pedroi (a) with a detail of the synflorescence (the arrow points to a male cyathium) (b). A recently emerged adult seed-wasp from a seed of E. pedroi (c) (photo by Israel Silva)

indicative of seed-wasp predation. In intact seeds the (only considering non-aborted seeds) and the average locule was filled with an unharmed embryo sur- seed predation levels by the two insect groups. rounded by white endosperm and the seed coat showed no signs of puncturing by seed predators. Results Data analysis Spatiotemporal variation in cyathia and fruit The effects of population, year and their interaction on production cyathia and fruit production, fruit-set, abortion levels and losses due to insects were analyzed using There was temporal variation in cyathia (F2,96 = repeated-measures ANOVA, with year as the within 53.81; P \ 0.0001) and fruit (F2,96 = 30.50; P \ subject (repeated) factor. Before performing the 0.0001) production within the two study populations, analyses the data were tested for normality and with the average number of cyathia produced in Ares homocedasticity. In order to meet the assumptions of showing higher fluctuations during the study period than ANOVA some variables were transformed (see sup- it was recorded at the same time in Azo´ia (Table 1). plementary material). Since multiple comparisons Nevertheless, no significant differences were found in were performed between variables indicative of cyathia and fruit production between the two study fecundity and plant size, a caution to control for false populations. In both study populations, the individuals discovery rates has been applied following the proce- with a larger crown diameter produced a high number of dure in Waite and Campbell (2006). When signifi- cyathia and a high number of fruits (all r [ 0.72; cance was rejected this is noted at the appropriate test P \ 0.001). The other variables indicative of plant size with ‘H0 accepted; FDR’. were less consistently related with cyathia and fruit We tested the association between variables indica- production. Plant size differed amongst the two popu- tive of plant size (height, crown width and stem lations with the individuals from Ares being taller diameter), plant fecundity (cyathia and fruit produc- (mean ± SD: 137.2 ± 37.1 vs. 98.2 ± 22.7 cm; tion), plant isolation and insect predation levels by t = 4.86, P \ 0.0001) but not having a larger crown Spearman Rank correlation analysis. The association diameter (mean ± SD: 162.2 ± 38.4 vs. 133.6 ± analyses were performed separately for each population 33.4 cm; t = 2.35, P \ 0.05; H0 accepted; FDR) and study year. Interpopulation comparisons on traits than those in Azo´ia. There was a strong consistency in indicative of plant size were made by means of Student’s cyathia production at the individual level (all t tests. The temporal dynamics of seed predation by r [ 0.84; P \ 0.0001), meaning that, in general, the generalist and specialist seed predators was examined most and less productive plants were the same by plotting together the average annual seed production throughout the years.

123 Plant Ecol (2012) 213:581–590 585

Table 1 Cyathia and fruit production per plant and proportion of losses due to the different factors that reduced fruit-set in the study populations of Euphorbia pedroi from 2002 to 2004 Location Variable 2002 2003 2004

Ares N of cyathia 628.1 ± 408.9 266.8 ± 142.0 447.8 ± 226.6 % Male cyathia 11.4 ± 2.7 17.1 ± 2.4 13.5 ± 2.9 % Ovary and fruit abortion 5.9 ± 2.7 6.8 ± 2.6 9.5 ± 4.9 % Cyathia infested by larvae 9.6 ± 6.1 10.9 ± 6.6 7.3 ± 7.0 N of fruits 471.6 ± 337.0 179.3 ± 103.2 319.8 ± 173.9 % Fruit set 73.0 ± 7.2 65.0 ± 8.0 69.8 ± 10.3 Azo´ia N of cyathia 542.4 ± 304.9 459.9 ± 272.2 335.3 ± 201.3 % Male cyathia 14.5 ± 2.4 13.9 ± 2.4 16.3 ± 2.8 % Ovary and fruit abortion 17.0 ± 9.4 11.6 ± 7.6 12.8 ± 7.2 % Cyathia infested by larvae 14.1 ± 8.9 10.3 ± 10.4 17.0 ± 10.4 N of fruits 321.6 ± 219.5 312.9 ± 218.3 202.0 ± 149.0 % Fruit set 54.4 ± 16.1 64.1 ± 15.7 53.9 ± 16.2 The number of individuals sampled was 25. Results are presented as mean ± SD

The proportion of cyathia that set fruit was higher in A large proportion of plants in the two populations

Ares than at Azo´ia (F1,48 = 18.16; P \ 0.0001), but was attacked by moth larvae (Ares, mean ± SD: no significant variation was observed within each 88.0 ± 4.0%; Azo´ia, mean ± SD: 96.0 ± 6.9%) and population. In general, fruit-set was not correlated the average proportion of cyathia destroyed per with any of the variables indicative of plant size individual was moderate (Table 1). The losses due to (|r| \ 30.0; P [ 0.05) and there was no consistency of moth larvae showed significant differences between fruit-set at the individual level. the study areas (F1,48 = 8.15; P \ 0.01), but not between years (F2,96 = 0.53; P [ 0.05). The relative Magnitude and variability in losses due number of cyathia damaged by moth larvae was not to the factors that affect fruit-set correlated with any of the variables indicative of plant size (|r| \ 0.24; P [ 0.05), not even with the distance The factors responsible for reduction in fruit-set were between individual plants (all |r| \ 0.35; P [ 0.05). the lack of a functional pistillate flower in the cyathia, Cyathia production also did not influence infestation inflorescence destruction by moth larvae and the levels by moth larvae, with a single exception abortion of ovaries and fruits (Table 1). Together, recorded at Azo´ia in 2004 (r =-0.59; P \ 0.01). these factors accounted for considerable reproductive No consistency was found in the proportion of cyathia losses and their combined effect showed high varia- damaged by moth larvae at the individual level in the tion between individuals (Ares, mean ± SD: 30.7 ± two study populations (all r \ 0.30; P [ 0.05). 9.6%; range = 13.6–67.2%; Azo´ia, mean ± SD: The proportion of aborted ovaries and fruits

42.5 ± 16.8%; range = 18.9–88.4%). differed amongst populations (F1,48 = 25.71; P \ Male cyathia levels were similar between the two 0.0001), being lower in Ares and higher at Azo´ia. study areas (F1,48 = 1.79; P [ 0.05), but showed Amongst-year variation in the proportion of aborted significant temporal variations throughout the study ovaries and fruits was not significant (F2,96 = 2.34; (F2,96 = 34.03; P \ 0.0001). The absolute number of P [ 0.05). Occasionally, ovary and fruit abortion male cyathia was positively correlated with the levels were found to be negatively correlated with the number of cyathia produced (all r [ 0.92; P \ number of cyathia produced and with plant size (both 0.0001) and with the variables indicative of plant plant height and crown diameter), but this pattern size, particularly crown diameter (where all r [ 0.71; failed to be consistent throughout the study. Annual P \ 0.001). variations in losses due to ovary and fruit abortion

123 586 Plant Ecol (2012) 213:581–590 were also not consistent at the individual level (all losses due to hemipterans. The damage inflicted by r \ 0.32; P [ 0.05). hemipterans was not related with plant traits related to size or fecundity and only once their effect was Seed production, abortion and the impact negatively correlated with the losses inflicted by seed- of specialist and generalist insect predators wasps (in Azo´ia during 2004, r =-0.54; P \ 0.01). Seed-wasps inflicted heavy damage to E. pedroi plants The proportion of viable, aborted and preyed seeds is in both populations. Seed predation levels by seed- shown in Table 2. Heavy losses due to seed abortion wasps showed a wide temporal variation (F2,96 = were recorded in the two study populations and 34.82; P \ 0.0001), particularly in Ares (Fig. 3). The there were also significant spatial (F1,48 = 259.52; highest losses due to seed-wasps were recorded during P \ 0.0001) and temporal variations (F2,96 = 11.38; years of low seed production whilst the lowest losses P \ 0.0001) in seed abortion levels. No associations occurred in high productive years. Thus, the impact of were found between seed abortion levels and any of the specialist seed-wasps seems to be intimately asso- the variables indicative of plant size or fecundity. ciated with the seed production patterns of E. pedroi The two groups of insect seed predators had a (Fig. 3). The losses inflicted by seed-wasps were not different impact on the reproductive output of E. pedroi related with variables indicative of plant size, fecundity and they also showed distinct temporal patterns of or isolation. We found no consistency of seed predation interaction with this host species. On average, hemipt- levels on individual plants throughout the study for both erans attacked a lower proportion of individuals than did specialist and generalist seed predators. The average seed-wasps (mean ± SD: 76.0 ± 18.4 vs. 99.3 ± number (mean ± SD) of intact seeds produced by 1.6%) and they also inflicted minor losses to the individual during the study was 399.0 ± 455.5 at Ares attacked plants (Fig. 3). There were spatial (F1,48 = and 228.3 ± 251.9 at Azo´ia. Intact seed production 89.37; P \ 0.0001) but not temporal (F2,96 = levels showed wide temporal fluctuations throughout 4.68; P \ 0.05; H0 accepted; FDR) variations in the the study (F2,96 = 39.03; P \ 0.0001) and were

Table 2 Proportion of intact, aborted and preyed seeds in the study populations of Euphorbia pedroi from 2002 to 2004 Location Variable 2002 2003 2004

Ares % Seed predation 18.2 ± 12.0 49.2 ± 10.0 33.2 ± 11.3 % Seed abortion 31.7 ± 4.6 34.8 ± 5.4 29.9 ± 4.5 % Viable seeds 50.2 ± 13.4 16.0 ± 8.3 36.9 ± 11.0 Azo´ia % Seed predation 28.3 ± 11.4 22.9 ± 10.7 36.6 ± 9.1 % Seed abortion 41.8 ± 4.5 47.7 ± 5.1 51.0 ± 7.0 % Viable seeds 30.0 ± 12.9 29.4 ± 11.4 12.4 ± 10.2 The number of plants sampled was 25. Results are presented as mean ± SD Proportion of seeds lost ab1200 75 1200 75 Proportion of seeds lost Ares Azóia

900 900 50 50

600 600

25 25 300 300 Seed production Seed production

0 0 0 0 2002 2003 2004 2002 2003 2004

Fig. 3 Dynamics of seed production and pre-dispersal seed Annual average (?SE) seed production per plant (bars)is predation by seed-wasps (solid line) and hemipterans (dashed plotted together with the mean proportion of losses due to the line) in the study populations of E. pedroi from 2002 to 2004. two different kinds of seed predators. a Ares; b Azo´ia 123 Plant Ecol (2012) 213:581–590 587 unrelated with the traits measured. No consistency in pollination during the period of stigma receptivity. intact seed production levels was detected amongst the Flying and non-flying insects are important visitors of study individuals. E. pedroi and the activity of both groups is strongly influenced by weather conditions. Other studies deal- ing with the reproductive biology of coastal plants Discussion have shown a decrease in pollen transfer by insects as a consequence of unfavorable weather (Kevan and Cyathia production in E. pedroi was variable through- Baker 1983; Rovira et al. 2004). A number of aborted out the 3-year period and showed no significant ovaries also corresponded to late flowering cyathia differences between Ares and Azo´ia, despite the which have failed to develop probably due to resource differences on traits indicative of plant size recorded limitation. By the time of the development of these amongst the two populations. However, within each ovaries, most of the fruits were maturing and a few population, cyathia production was positively related others had already dehisced. So, maternal resources with crown diameter and showed a strong consistency are probably very limited at this stage and are mostly at the individual level, with the most and least channelled to maturing fruits. The temporal decline in productive plants being, in general, the same during fruit-set has been documented in a large number of the study period. Many other studies have also found a species and experimental work has provided evidence high positive correlation between flower production that this is mostly due to competition of flowers and and traits indicative of plant size (e.g. Traveset 1995; fruits for limited maternal resources (Stephenson McIntosh 2002) leading several authors to consider 1981; Lee 1988 and references therein). that this association appears to hold universally The two major forms of reproductive losses in both amongst plants (Herrera 1991 and references therein). populations of E. pedroi were seed abortion and losses Male cyathia levels in E. pedroi were relatively due to insects. Seed abortion levels were quite high in constant with annual averages usually around 15%. the two populations, averaging from a third to half the These values differ from the ones reported to other potential seed production following fruit-set. High Iberian perennial Euphorbia (Traveset 1995; Narbona levels of seed abortion have been frequently reported et al. 2002; Boieiro et al. 2010a) probably due to in narrow endemic and rare species and often reflect differences in phylogeny and morphology, particu- the negative effects of inbreeding on seed set (Keller larly inflorescence architecture. The prevalence of and Waller 2002; Shi et al. 2005; Ledig et al. 2006). In male cyathia in perennial Euphorbia has been inter- small and isolated populations, as is characteristic of preted as a mechanism to promote outcrossing in long- many narrow endemics, gene flow between popula- lived plants by enhancing pollen receipt of pistillate tions is usually low leading to a decrease in genetic flowers (Narbona et al. 2002, 2005). In these species, diversity and the accumulation of deleterious muta- protogyny (together with pollen transfer probabilities) tions. As a consequence, those populations become may have favored variation in sex allocation leading to more prone to negative genetic effects on fitness traits a higher presence of male cyathia in early-blooming and to decreased potential for adaptation (Keller and flowers (Brunet and Charlesworth 1995; Narbona et al. Waller 2002; Willi et al. 2006). Thus, in recent years, 2005), which occur on the lowest levels of the genetic rescue has been a conservation tool increas- inflorescence. ingly considered to ensure the survival of many The harsh environmental conditions found in the narrow endemics (Holmes et al. 2008; Finger et al. coastal populations where the study took place were 2011). We cannot exclude the possibility that other responsible for some of the reproductive losses that causes of seed abortion (e.g. pollen or resource resulted in ovary and fruit abortion. The strong winds, limitation) may also be responsible for a fraction of frosts and regular salt spray during the flowering and the losses recorded. At least, there is some evidence fruiting periods damaged both ovaries and early fruits that the higher abortion levels in Azo´ia comparatively in some individuals leading to the abortion of these to Ares may be due to a lower availability of resources. reproductive structures. Furthermore, the unfavour- In Azo´ia, the individuals of E. pedroi are shorter than able atmospheric conditions may have also played an their conspecifics at Ares and grow predominantly on indirect role in ovary abortion by precluding insect rocky crevices or inclined coastal slopes, where the 123 588 Plant Ecol (2012) 213:581–590 availability of nutrients is usually low. A similar (De Steven 1983; Solbreck and Sille´n-Tulberg 1986) pattern of spatial variation in seed abortion levels was the inter-annual variation in seed production is consid- also detected in a co-occurring congener (Boieiro et al. ered a key feature in these plant-seed predator systems 2010a). These findings reinforce the idea that specific since it allows the regulation of seed predators’ experimental work is thus needed to assess the causes populations providing cyclical opportunities of seed of high seed abortion in the two populations of this escape from predation. Some other studies have narrow endemic. reported similar patterns of interaction between spe- Insects had a significant negative impact on the cialist seed predators and their host plants (Sperens reproductive output of E. pedroi. The destruction of 1997;Poncetetal.2009; Boieiro et al. 2010b). A cyathia by moth larvae affected most of the individuals number of studies have shown that narrow endemics in both populations but losses were moderate. On the seem to be particularly vulnerable to seed predator other hand, pre-dispersal seed predators besides attack (e.g. Menges et al. 1986;Kaye1999;Combs attacking virtually all individuals of E. pedroi were et al. 2011) and occasionally, as it happened with also responsible for high seed losses. At this stage, it is Ebenus armitagei (Hegazy and Eesa 1991), population however important to discriminate the reproductive survival may also be under threat due to extraordinary losses due to generalist and specialist seed predators. levels of seed predation. Furthermore, some compara- Losses due to generalist seed predators were mainly tive studies on the reproductive ecology of narrow due to Cydnus aterrimus, the most common of the endemics and their more widespread congeners have three hemipteran species detected in the two study shown that the former seem to be more susceptible to populations. Traveset (1995) clearly misinterpreted losses due to seed predation (e.g. Lavergne et al. 2005; the feeding biology of C. aterrimus by considering that Young et al. 2007) and even in one occasion where seed this species punctured the fruits of E. dendroides to predator intensity was similar between two congeners, feed exclusively upon the seed-wasp larvae develop- it was shown that seed predation was the key factor ing inside infested seeds. Cydnus aterrimus is a true causing low population growth rate in the rare species insect seed predator that feeds upon the seeds of a (Mu¨nzbergova´ 2005). Nevertheless, amongst others, a variety of plant species, but also on other plant comprehensive study on the biology of 20 congeneric material (Schorr 1957; W. Rabitsch, pers. comm.). pairs of rare and widespread Mediterranean species Seed losses inflicted by hemipterans were low to challenged this view by failing to found significant moderate, with significantly higher predation levels differences in seed predation levels between the two recorded at Azo´ia where these insects were more species groups (Lavergne et al. 2004). abundant. Seed losses due to specialist seed-wasps In spite of the high levels of seed predation detected were high and highly variable in time in both in our study species predispersal seed predators per se populations of E. pedroi. The damage caused by do not seem to pose a serious threat to the survival of seed-wasps to some individuals was extremely high this narrow endemic. Euphorbia pedroi is a long-lived (above 85%), but no consistency of seed predation perennial species that may benefit from the oscilla- levels amongst individuals was found during the tions in seed production patterns amongst seasons to study. Interestingly, the temporal variation in seed provide cyclical opportunities of seed escape from predation levels due to seed-wasps was intimately specialist predators. Furthermore, the number of seeds associated with the magnitude of yearly variation in per plant that escape predispersal seed predation is seed production in E. pedroi. Seed-wasps are specialist relatively high when compared to those from other seed predators that depend upon the availability of narrow endemics (Kaye 1999;Mu¨nzbergova´ 2005; intact seeds to complete their life-cycle and conse- Young et al. 2007; Combs et al. 2011) and similar to quently their population dynamics is mostly deter- the ones reported from a widespread congener (Tra- mined by the availability of resources for oviposition veset 1995). and larvae development during the previous season. By the time xerochastic dehiscence occurs, the For this reason, seed-wasps show a delay in tracking proportion of intact seeds dispersed is just a fraction of the variation in their food resources leading to the the potential seed production in E. pedroi (24% at Ares observed mismatch between seed predation levels and and less than 14% at Azo´ia). Intact seed production in seed production patterns. According to several authors this species is crucial not just only for providing 123 Plant Ecol (2012) 213:581–590 589 propagules to the replacement of individuals, but also Acknowledgments We dedicate this article to the botanist for the colonization of new sites since some of the Jose´ Gomes Pedro for the discovery of interesting plant taxa, including Euphorbia pedroi, and for his commitment to the areas occupied by E. pedroi are prone to rock slides study and safeguard of the flora of Arra´bida. We thank J. which usually lead to individual mortality. These Boieiro, I. Silva, M. Zerova, W. Rabitsch, E. Marabuto and M.J. events seem important for the population dynamics of Pinto for their help. Support was provided by Fundac¸a˜o para a E. pedroi since they provide an opportunity for the Cieˆncia e a Tecnologia through grants PRAXIS XXI/BD/21407/ 99 to MB and SFRH/BPD/66934/2009 to CR. establishment of individuals in areas free of compet- itors and with available microsites.

Concluding remarks References

Information on plant reproductive biology has proved A˚ gren J (1988) Between-year variation in flowering and fruit set to be extremely valuable for the design of more in frost-prone and frost-sheltered populations of dioecious effective conservation measures (Colas et al. 2001; Rubus chamaemorus. Oecologia 76:175–183 ´ Berg RY (1990) Seed dispersal relative to population structure, Navarro and Guitian 2002) and this has led several reproductive capacity, seed predation, and distribution in authors to defend that those studies should be ‘an Euphorbia balsamifera (Euphorbiaceae), with a note on integral feature of all conservation projects’ (Moza sclerendochory. Sommerfeltia 11:35–63 and Bhatnagar 2007). Even so, insufficient under- Boieiro M, Rego C, Serrano ARM, Espadaler X (2010a) The impact of specialist and generalist pre-dispersal seed pre- standing of species biology is still considered a serious dators on the reproductive output of a common and a rare handicap for the adoption of more effective conser- Euphorbia species. Acta Oecol 36:227–233 vation efforts. By providing information on the Boieiro M, Serrano ARM, Rego C, Espadaler X (2010b) Plant predispersal reproductive losses of the narrow ende- fecundity and pre-dispersal reproductive losses in a com- Euphorbia E. pedroi mon and a rare species (Euphorbiaceae). Ecol mic we intend to contribute to a better Res 25:447–456 management of its populations. As intact seeds are a Brunet J, Charlesworth D (1995) Floral sex allocation in relatively small fraction of potential seed production, sequentially blooming plants. Evolution 49:70–79 following their fate is needed to establish if further Colas B, Olivieri I, Riba M (2001) Spatio-temporal variation of reproductive success and conservation of the narrow- losses may threaten population persistence. Further- endemic Centaurea corymbosa (Asteraceae). Biol Conserv more, studies on the genetic diversity within and 99:375–386 amongst populations should be carried to ascertain the Combs J, Reichard S, Groom M, Wilderman D, Camp P (2011) genetic status of E. pedroi and to evaluate if the high Invasive competitor and native seed predators contribute to rarity of the narrow endemic Astragalus sinuatus Ecol abortion levels detected in our study have a genetic Appl 21:2498–2509 basis. All this should be done in a plant population Crawley MJ (2000) Seed predators and plant population monitoring framework, where specific demo- dynamics. In: Fenner M (ed) Seeds, the ecology of regen- graphic, ecological and genetic parameters should be eration in plant communities. CABI Publishing, Oxford, pp 167–182 evaluated. De Steven D (1983) Reproductive consequences of insect seed The recent inclusion of all three populations of predation in Hamamelis virginiana. Ecology 64:89–98 E. pedroi in a natural reserve was an important step for Finger A, Kettle CJ, Kaiser-Bunbury CN, Valentin T, Doudee the conservation of this narrow endemic. However, D, Matatiken D, Ghazoul J (2011) Back from the brink: potential for genetic rescue in a critically endangered tree. due to the reduced number of populations and high Mol Ecol 20:3773–3784 habitat specialization, this species remains extremely Garcı´a MB (2008) Life history and population size variability in vulnerable to demographic and environmental sto- a relict plant. Different routes towards long-term persis- chasticity. Ex-situ conservation should thus be a tence. Divers Distrib 14:106–113 George S, Sharma J, Yadon VL (2009) Genetic diversity of the complementary strategy for the conservation of endangered and narrow endemic Piperia yadonii (Or- E. pedroi. In the short term, efforts should be carried chidaceae) assessed with ISSR polymorphisms. Am J Bot out to cultivate some individuals in botanical gardens 96:2022–2030 and to collect and store seeds from the three known Hegazy AK, Eesa NM (1991) On the ecology, insect seed-pre- dation, and conservation of a rare and endemic plant spe- populations to ensure genetic variability amongst cies: Ebenus armitagei. Conserv Biol 5:317–324 propagules for hypothetical reintroductions or Herrera C (1991) Dissecting factors responsible for individual reinforcements. variation in plant fecundity. Ecology 72:1436–1448 123 590 Plant Ecol (2012) 213:581–590

Holmes GD, James EA, Hoffmann AA (2008) Limitations to Navarro L, Guitia´n J (2002) The role of floral biology and reproductive output and genetic rescue in populations of breeding system on the reproductive success of the narrow the rare shrub Grevillea repens (Proteaceae). Ann Bot endemic Petrocoptis viscosa rothm. (Caryophyllaceae). 102:1031–1041 Biol Conserv 103:125–132 Hulme PE (2002) Seed-eaters: seed dispersal, destruction and Navarro L, Guitia´n J (2003) Seed germination and seedling demography. In: Levey D, Silva WR, Galetti M (eds) Seed survival of two threatened endemic species of the north- dispersal and frugivory: ecology, evolution and conserva- west Iberian peninsula. Biol Conserv 109:313–320 tion. CABI Publishing, Wallingford, pp 257–273 Poncet BN, Garat P, Manel S, Bru N, Sachet JM, Roques A, ICN (1997) Sı´tio Arra´bida/Espichel. Plano sectorial da Rede Despres L (2009) The effect of climate on masting in the Natura 2000. Instituto da Conservac¸a˜o da Natureza, Lisboa European larch and on its specific seed predators. Oeco- Kaye TN (1999) From flowering to dispersal: reproductive logia 159:527–537 ecology of an endemic plant, Astragalus australis var. Rovira ML, Bosch M, Molero J, Blanche´ C (2004) Pollination olympicus (Fabaceace). Am J Bot 86:1248–1256 ecology and breeding system of the very narrow coastal Keller LF, Waller DM (2002) Inbreeding effects in wild popu- endemic Seseli farrenyi (Apiaceae). Effects of population lations. Tree 17:230–241 fragmentation. Nord J Bot 22:727–740 Kevan PG, Baker HG (1983) Insects as flower visitors and Rucin´ska A, Puchalski J (2011) Comparative molecular studies pollinators. Annu Rev Entomol 28:407–453 on the genetic diversity of an ex situ garden collection and Kolb A, Ehrle´n J, Eriksson O (2007) Ecological and evolu- its source population of the critically endangered polish tionary consequences of spatial and temporal variation in endemic plant Cochlearia polonica E. Frohlich. Biodivers pre-dispersal seed predation. Perspect Plant Ecol Evol Syst Conserv 20:401–413 9:79–100 Schorr H (1957) Zur verhaltensbiologie und symbiose von Lavergne S, Thompson JD, Garnier E, Debussche M (2004) The Brachypelta aterrima Fo¨rst. (Cydnidae, Heteroptera). biology and ecology of narrow endemic and widespread Z Morph u O¨ kol Tiere 45:561–602 plants: a comparative study of trait variation in 20 conge- Shi XJ, Michaels HJ, Mitchell RJ (2005) Effects of self-polli- neric pairs. Oikos 107:505–518 nation and maternal resources on reproduction and off- Lavergne S, Debussche M, Thompson JD (2005) Limitations on spring performance in the wild lupine, Lupinus perennis reproductive success in endemic Aquilegia viscosa (Ran- (Fabaceae). Sex Plant Reprod 18:55–64 unculaceae) relative to its widespread congener Aquilegia Solbreck C, Sille´n-Tulberg B (1986) Seed production and seed vulgaris: the interplay of herbivory and pollination. Oec- predation in a patchy and time-varying environment. ologia 142:212–220 Dynamics of a milkweed-tephritid fly system. Oecologia Ledig FT, Hodgskiss PD, Johnson DR (2006) Genetic diversity 71:51–58 and seed production in Santa Lucia fir (Abies bracteata), a Sperens U (1997) Fruit production in Sorbus aucuparia L. relict of the Miocene Broadleaved Evergreen Forest. (Rosaceae) and pre-dispersal seed predation by the apple Conserv Genet 7:383–398 fruit moth (Argyresthia conjugella Zell.). Oecologia Lee TD (1988) Patterns of fruit and seed production. In: Lovett 110:368–373 Doust J, Lovett Doust L (eds) Plant reproductive ecology. Srimuang KO, Watthana S, Pedersen HAE, Rangsayatorn N Oxford University Press, New York, pp 179–202 (2010) Aspects of biosubsistence in Sirindhornia (Or- Lo´pez-Pujol J, Zhang FM, Ge S (2005) Population genetics and chidaceae): are the narrow endemics more reproductively conservation of the critically endangered Clematis aceri- restricted than their widespread relative? Ann Bot Fenn folia (Ranunculaceae). Can J Bot 83:1248–1256 47:449–459 McIntosh ME (2002) Plant size, breeding system, and limits to Stephenson AG (1981) Flower and fruit abortion: proximate causes reproductive success in two sister species of Ferocactus and ultimate functions. Annu Rev Ecol Syst 12:253–279 (Cactaceae). Plant Ecol 162:273–288 Traveset A (1995) Spatio-temporal variation in pre-dispersal Menges ES, Waller DM, Gawler SC (1986) Seed set and seed reproductive losses of a Mediterranean shrub, Euphorbia predation in Pedicularis furbishiae, a rare endemic of the dendroides L. Oecologia 103:118–126 St. John River, Maine. Am J Bot 73:1168–1177 Waite TA, Campbell LG (2006) Controlling the false discovery Moza MK, Bhatnagar AK (2007) Plant reproductive biology rate and increasing statistical power in ecological studies. studies crucial for conservation. Curr Sci 92:1207 Ecoscience 13:439–442 Mu¨nzbergova´ Z (2005) Determinants of species rarity: popu- Willi Y, Buskirk JV, Hoffmann AA (2006) Limits to the lation growth rates of species sharing the same habitat. Am adaptive potential of small populations. Annu Rev Ecol J Bot 92:1987–1994 Evol Syst 37:433–458 Narbona E, Ortiz PL, Arista M (2002) Functional andromono- Young AS, Chang S-M, Sharitz RR (2007) Reproductive ecol- ecy in Euphorbia (Euphorbiaceae). Ann Bot 89:571–577 ogy of a federally endangered legume, Baptisia arachnif- Narbona E, Ortiz PL, Arista M (2005) Dichogamy and sexual era, and its more widespread congener, B. lanceolata dimorphism in floral traits in the andromonoecious (Fabaceae). Am J Bot 94:228–236 Euphorbia boetica. Ann Bot 95:779–787

123 Supplementary material

Table S – Results of the repeated measures ANOVA for the effects of population, year and their interaction on the studied variables. The ANOVA were conducted on original or transformed data. Data on cyathia and fruit production, and fruit set were log transformed while data on andromonoecy levels, proportions of aborted ovules and fruits, aborted seeds and losses to moth larvae, hemipterans and seed-wasps were arcsine square root transformed. FDR corrected (=H0 accepted) is highlighted in yellow.

Cyathia production Fruit production Fruit set Andromonoecy levels Source of variation df MS F Df MS F df MS F df MS F

Population 1 104 0.00 ns 1 75085 0.64 ns 1 0.522 18.16*** 1 0.003 1.79 ns Year 2 729732 53.81*** 2 343856 30.50*** 2 0.010 0.89 ns 2 0.009 34.03*** Population x Year 2 357956 26.40*** 2 301428 26.74*** 2 0.114 10.19*** 2 0.016 62.29*** Error 96 13560 96 11274 96 0.011 96 0.000

Prop. of losses to moth Prop. of aborted ovules Prop. of losses to Prop. of losses to seed- larvae and fruits hemipterans wasps Source of variation df MS F Df MS F df MS F df MS F

Population 1 0.076 8.15** 1 0.152 25.71*** 1 0.465 89.37*** 1 0.086 2.02 ns Year 2 0.003 0.53 ns 2 0.007 2.34 ns 2 0.034 4.68* 2 0.795 34.82*** Population x Year 2 0.033 5.24** 2 0.021 6.83** 2 0.017 2.41 ns 2 1.074 47.03*** Error 96 0.006 96 0.003 96 0.007 96 0.023

Prop. of aborted seeds Prop. of intact seeds Source of variation df MS F df MS F

Population 1 0,807 259.52*** 1 0.406 26.33*** Year 2 0.029 11.38*** 2 0.452 39.03*** Population x Year 2 0.041 15.97*** 2 0.540 46.71*** Error 96 0.003 96 0.012 ns nonsignificant, *P<0.05, **P<0.01, ***P<0.001