Ecological response to human trampling and conservation status of Helianthemum caput-felis (Cistaceae) at the eastern periphery of its range

Abstract

Border and isolated plant populations represent an interesting target for ecological and conservation issues. We analysed the ecological constraints and the conservation status of the eastern population of Helianthemum caput-felis Boiss. (Cistaceae), located in Sardinia. The distribution of H. caput-felis was verified via field surveys; ecological data, morphological and reproductive traits, were recorded in 40 permanent plots randomly established; the human trampling effects on plant density, plant size and plant performance were analysed. Plant density was higher in bedrock and lowland areas, in garrigue and maquis habitats; however, the differences among plants growing in different ecological conditions were not statistically significant; only human trampling intensity significantly affected plant density and lowest values were observed in areas with intense trampling pressure. All ecological variables analysed had a statistically significant effect on plant size and on the number of fruits per plant. In particular, larger plants were found in areas with the following ecological features: presence of structured soil, on the slopes, in the maquis habitat, and in areas with intensive human trampling. Conversely, plants displayed a higher fruits output per plant in deep and structured soil, in lowland areas, and in the garrigue and maquis habitats; the mean fruits output per plant increased as human trampling intensified. Human-induced threats are the main hazards threatening the remaining Sardinian population. In particular, the major threats are linked to tourism and other outdoor activities (i.e. human trampling), followed by the expansion of agricultural activities; all of these threats result in the disappearance of small localities and in reduced population size due to habitat loss and fragmentation. Our study indicates that H. caput-felis should be considered as Critically Endangered (CR) at the regional level. Urgent measures should be undertaken to protect the remaining H. caput-felis population in Sardinia and a possible integrated strategy for the conservation and management consists of a combination of in situ and ex situ measures. In particular, greater emphasis should be given to minimizing the negative impacts of unsustainable tourism and recreation use, in order to exclude human trampling and to facilitate the plant recruitment process and population renewal. In addition, an ex situ conservation strategy must be implemented and the seeds collected could be used for future translocations in suitable areas. Moreover, considering the threats observed, a long-term monitoring programme must be developed to reveal changes in the species conservation status.

Keywords:

• border plant populations
• Cistaceae
• conservation status
• Habitat Directive
• IUCN assessment
• Mediterranean coastal plant
• plant density and size
• reproductive traits
• Sardinia
• threatened plant population

Introduction

Several plant species are characterized by a disjunctive distribution in which peripheral populations can be isolated from the main home range. Plants that display a peculiar distribution and, in particular, species with border and peripheral populations, represent interesting targets in ecology, evolutionary biology and genetics (Eckert, Samis, and Lougheed 2008; Sexton et al. 2009; Pouget et al. 2013). In addition, they provide insight into critical phenomena, such as speciation, adaptive radiation and natural selection (Grant and Antonovics 1978; Holt and Keitt 2005).

Independent of the central/marginal model debate, border populations as well as peripheral isolated plant populations are particularly important from both ecological and genetic points of view (Lesica and Allendorf 1995; Conradt 2001; Holt and Keitt 2005) and require more attention from conservation biologists (Abeli et al. 2009). Border populations are usually considered more vulnerable and are more prone to local extinction because of their isolation and restriction to marginal habitats (e.g. Gargano et al. 2007; Del Vecchio et al. 2012; Villellas et al. 2013; Villellas, Morris, and García 2013). As suggested by international organizations (e.g. IUCN, European Council) and according to The European Strategy for Plant Conservation (Planta Europa 2008), border and isolated populations should be considered an important resource for biodiversity and should therefore be included in conservation programmes. Consequently, several plant species in the Mediterranean Basin that show outlying populations isolated ecologically and geographically from the rest of their distribution range have been investigated in recent years (e.g. Gargano et al. 2007; Del Vecchio et al. 2012; Fois et al. 2015).

Cistaceae is a medium-size family with eight genera and approximately 180 taxa distributed in temperate and subtropical regions of the northern hemisphere, and it displays the highest genus and species diversity in the Mediterranean floristic region (Guzmán and Vargas 2009). Within this family, Helianthemum Miller is the most diverse genus, with approximately 100 taxa that grow from sea level up to approximately 3000 m in a diverse array of substrates (limestone, dolomite, marl, gypsum, saline and sand-soils) that are concentrated in the western Mediterranean area (Próctor and Heywood 1968). Within this genus, Helianthemum caput-felis Boiss. deserves particular attention because it is considered the only extant representative of an ancient lineage (Arrigoni 1971; López-González 1992). Helianthemum caput-felis is a coastal plant distributed throughout the western Mediterranean Basin (southeastern Iberian Peninsula, Mallorca, Sardinia and northwest Africa) in several fragmented populations; the widest distribution and the largest populations are located in Spain (Agulló et al. 2011; López-González 1992), whereas the presence of this species in Sardinia and northwest Africa is restricted to small areas in unique or reduced places (Arrigoni 1971; Fenu and Bacchetta 2008; Quézel and Santa 1963).

In Sardinia, H. caput-felis grows in two main localities (Capo Mannu and Su Tingiosu, central-west part of the island; Figure 1) that are approximately 3 km apart; small patches are also found in the coastal cliff along the Sinis Peninsula (Arrigoni 1971; Fenu and Bacchetta 2008). This plant is found in the discontinuities of the Juniperus micro-forest and into the maquis, but it mainly grows in the coastal garrigues, where cushion chamaephytes are dominant. Helianthemum caput-felis is a member of a rupicolous coastal plant community that is rich in narrow endemics, such as Limonium lausianum Pignatti and Polygala sinisica Arrigoni, as well as other western Mediterranean plants, such as Viola arborescens L., Coris monspeliensis L. and Erica multiflora L. (Fenu et al. 2012).

Fig. 1. Study area in the Sinis Peninsula (CW-Sardinia); the circles indicate the main localities of Capo Mannu and Su Tingiosu where Helianthemum caput-felis grows.

Helianthemum caput-felis is an emblematic plant from a conservation point of view; it is included in the Berne Convention and Appendix II of the Habitats Directive (European Community Council Directive 92/43/EEC), and it is classified as endangered on the European red list of vascular plants (Bilz et al. 2011) according to the IUCN protocol (IUCN 2001). However, to date, no exhaustive reproductive and ecological studies have been conducted on this plant, and in particular on the Sardinian population, and no conservation actions have been implemented for this species.

Mediterranean coastal habitats have been altered by human activity for several thousand years, and coastal environments are strongly affected by tourism and related infrastructures (Davenport and Davenport 2006). In particular, touristic and recreational activities in coastal areas appear to be common threats to a wide variety of European threatened plants (Bilz et al. 2011; Ballantyne and Pickering 2013; Fenu et al. 2013; Rossi et al. 2015). However, few studies have focused on the effects of human trampling on Mediterranean coastal ecosystems (Comor et al. 2008; Kutiel, Eden, and Zhevelev 2000; Kerbiriou et al. 2008). Although for threatened plants the impact of tourism is particularly severe (Pickering and Hill 2007), to our knowledge, the effect of human trampling on threatened plants growing on coastal areas has yet to be accurately assessed in Mediterranean coastal habitats. In general, tolerance of species to human trampling varies, sometimes markedly. In Mediterranean coastal ecosystems some plant species (including threatened plants) are very sensitive to trampling, while others seem to be tolerant or even to benefit from trampling (Kerbiriou et al. 2008; Yu, Bell, and Kutiel 2009; Fenu et al. 2013). Human trampling has been demonstrated to be generally an important threat for threatened plant species in Sardinia (Quilichini and Debussche 2000; Fenu, Mattana, and Bacchetta 2011; Fenu et al. 2013; Rossi et al. 2015) and we hypothesize that also the Sardinian population of H. caput-felis is particularly affected by human disturbance, as previously reported for the Spanish populations (Dominguez Lozano et al. 1996).

The main aims of this study were to verify the distribution range and population size of H. caput-felis in Sardinia (eastern periphery of its distribution range), to analyse plant size and reproductive traits, to analyse the effect of human disturbance, to identify its principal threats and to assess its conservation status at the regional level following the IUCN protocol.

Methods

Study species

Helianthemum caput-felis is a perennial small shrub that grows to a height of 35(50) cm. Its flowers, which are arranged in inflorescences at the tip of new branches, are yellow and hermaphroditic, open at dawn and close at dusk, and have a short lifespan (3–4 days; Rodríguez-Pérez 2005). Based on studies carried out in Spain, the flowering period is from late February to late May, and the fruiting season runs from late April to July–August (Rodríguez-Pérez 2005). Tébar, Gil, and Llorens (1997) and Rodríguez-Pérez (2005) reported the allogamous character of this species, being a generalist entomogamous plant. Fruits are capsules that detach at maturity, and seed germination occurs in autumn, at the onset of the rainy season (Rodríguez-Pérez 2005).

From an ecological point of view, H. caput-felis is a thermophilous plant that typically grows in coastal environments under the direct influence of the sea, mostly on calcareous rocky cliffs (0–200 m above sea level) with garrigues or scrublands (Arrigoni 1971; Agulló et al. 2011; Fenu et al. 2012); some populations also grow on different habitats, such as sand dunes (Mallorca and Melilla), rocky slopes bordering inland ravines (Melilla; Agulló et al. 2011) or, rarely, in open wooded areas (Raynaud 1999).

Data sampling

The distribution of H. caput-felis was verified by field surveys conducted in the localities for which herbarium specimens and/or published data (Arrigoni 1971; Fenu and Bacchetta 2008) were available. The geographical limits of the main localities of Capo Mannu and Su Tingiosu (CM and ST, hereafter) were mapped using a global positioning system, and the areas where the plants were found were estimated using ArcView v. 3.2 (ESRI, Redlands, CA, USA). Localities containing only scattered plants were excluded a priori in this study (Figure 1).

Forty permanent plots of 2 × 1 m (20 per locality) were randomly established (in the area where the plant was found) to estimate the plant densities and extrapolate the population size. Over 1 year, plants were monitored bimonthly, approximately on the 10^th and 20^th days of each month. Within the plots, all plants (including seedlings) were counted, marked and measured from March to August, in order to analyse the whole reproductive season of the plant species. During each monitoring, survival, morphological (height, maximum and minimum diameter) and reproductive traits (number of flowers and fruits per plant) were recorded for every plant. Each plant was assigned to one of the following habitats: garrigue, maquis and micro-forest. Similarly, each plant was assigned to a geomorphology and substrate (lowland versus slope areas and soil versus bedrock, respectively).

Human trampling intensity was visually estimated for each plot and four levels of intensity were considered: absent, low (≤ 30% of the plot surface), moderate (31–60%), and intense (≥ 61%); 10 plots for each category were considered in this study.

Phenological and reproductive traits were monitored in 378 marked plants. Phenological phases were considered present when > 5% of the plants showed the same pattern and each plant was classified as being reproductive based on the presence of flowers/fruits (censuses were carried out in May). The average number of fruits per plant was determined at the peak of fruiting season (May) from a ratio of the total number of fruits/total number of plants monitored. Seed-set (considered as number of seeds per fruit) was calculated in May by collecting randomly 400 mature fruits from 40 randomly selected plants. Seeds were extracted and counted, and the average number of seeds per fruit was multiplied by the average number of fruits per reproductive plant to predict the mean reproductive capacity of each plant.

IUCN assessment at regional level

Globally, the IUCN Red List procedure is the most widely used protocol for species risk assessment because it facilitates objective, repeatable and flexible assessments (e.g. Rodrigues et al. 2006; Grammont De and Cuarón 2006; Rossi et al. 2015). For IUCN assessment, we evaluated all of the localities where H. caput-felis grows (Fenu and Bacchetta 2008). IUCN categories and criteria version 3.1 (IUCN 2001) were applied according to the procedure for regional assessment (IUCN 2003). According to the New Italian Red List project (Rossi et al. 2013), a grid of 2 × 2 km generated by the ESRI^® ArcGis™ 9.2 package and superimposed onto a map of Italy was used to assess the area of occupancy (AOO, defined as the area within the extent of occurrence, EOO, that is occupied by a taxon, where EOO is defined as the area contained within the shortest continuous imaginary boundary that can be drawn to encompass all the known sites of occurrence of a taxon, excluding cases of vagrancy, sensu IUCN 2001, 2014a). All of the parameters required by the IUCN protocol (i.e. EOO, locations, decline rate) and conservation status were assessed following the latest guidelines of the IUCN (2014a).

The major threats to H. caput-felis were determined through field observations and categorized following the version 3.2 of IUCN threats classification scheme (IUCN 2014b).

Statistical analysis

Considering that the peak of the reproductive season occurred in May, only the data of this month have been used in the analyses. Morphometric traits collected in the field were used to calculate the volume of each plant (plant size, hereafter). We calculated plant size (V, expressed in dm³) using individual height (h, expressed in dm) and the maximum and minimum diameter (d_M and d_m, respectively), according to the following formula:

The Pearson correlation value was calculated to correlate plant size with reproductive parameter. To evaluate the effect of locality, ecological parameters (geomorphology, substrate and habitat type) and disturbance on plant density, plant size and fruit production, three independent generalized linear models were fitted using a normal function for continuous variables, Poisson error distribution and log as a link function for count data. GLM analyses were performed by a stepwise procedure using JMP 7.0 (SAS, SAS Institute, Cary, NC, USA).

The non-parametric Mann–Whitney U-test was performed to verify differences between localities in the mean percentage of seeds per fruit; these analyses were performed using Statistica 8.0 software (Statsoft, Tulsa, OK, USA).

Results

Surface-area and plant density

The overall surface-area occupied by the H. caput-felis population is 30.61 ha (17.87 ha for CM and 12.74 ha for ST); these locations represent almost the entire Sardinian population (99.99%), although a few isolated plants grow in some southern coastal sites along the Sinis Peninsula (data not shown).

The estimated mean density was 4.73 ± 2.31 plants/m², and it varied from 4.63 ± 2.25 (CM) to 4.83 ± 2.42 (ST) plants/m²; the minimum and maximum density values ranged from 2 to 9.5 plants/m² for CM and from 1 to 8.5 plants/m² for ST (Table 1).

Table 1. Helianthemum caput-felis localities and their population traits: area, mean density, estimated population range, mean plant size, percentage of reproductive plants, mean number of flowers and fruits per plant, percentage of empty fruits and mean number of seeds per fruit.

Locality	Area (ha)	Plant density ± SD (plants/m^2)	Estimated mean no. plants	Estimated population range (min–max)	Plant size ± SD (dm^3)	Reproductive plants (%)	Flowers per reproductive plant (mean ± SD)	Fruits per reproductive plant (mean ± SD)	Empty fruits (%)	Seeds per fruit (mean ± SD)
Capo Mannu (CM)	17.87	4.63 ± 2.25	862,230	1,295,400–429,100	11.21 ± 48.68	66.49	30.42 ± 48.00	36.69 ± 49.43	17.5	4.32 ± 1.40
Su Tingiosu (ST)	12.74	4.83 ± 2.42	589,225	876,250–302,200	6.11 ± 13.72	89.64	39.01 ± 44.52	78.49 ± 107.15	22.5	4.27 ± 1.37
Overall population	30.61	4.73 ± 2.31	1,446,320	2,154,000–738,600	8.61 ± 35.48	78.31	33.23 ± 46.98	58.03 ± 86.47	20.0	4.29 ± 1.39

Plant density was higher in bedrock and lowland areas, and the highest density value was observed in garrigue, followed by maquis (Table 2); however, the differences among plants growing in different ecological conditions are not statistically significant (p > 0.05; Table 3). Our results revealed that only human trampling intensity significantly affected plant density (< 0.05; Table 3): in fact, the lowest plant density was observed in the plot with intense trampling pressure (6.67 ± 4.27 plants/m²) compared with undisturbed areas (trampling low and absent, 10.38 ± 4.61 and 10.48 ± 4.96 plants/m², respectively; Figure 2).

Table 2. Helianthemum caput-felis traits at population and plant levels in relation to substrate, geomorphology and habitat in Sardinia.

		Plant density (plant m^2)			Plant size (plant volume - dm^3)			No. Fruits per reproductive plant
		Overall	CM	ST	Overall	CM	ST	Overall	CM	ST
Substrate
Bedrock	Mean ± SD	4.89 ± 2.41	5.17 ± 2.84	4.82 ± 2.43	3.64 ± 8.10	3.44 ± 6.23	3.69 ± 8.59	68.27 ± 80.26	60.61 ± 73.94	69.75 ± 81.72
	Min	1.5	2	1.5	0.0002	0.0012	0.0002	3	4	3
	Max	8.5	7.5	8.5	61.28	24.46	61.28	500	325	500
Soil	Mean ± SD	4.63 ± 2.3	4.53 ± 2.23	4.83 ± 2.56	11.43 ± 43.79	12.78 ± 53.17	9.05 ± 17.73	80.21 ± 98.14	55.85 ± 46.81	112.48 ± 133.40
	Min	1	2	1	0.0003	0.0003	0.009	2	2	2
	Max	9.5	9.5	8.5	628.32	628.32	117.76	600	218	600
Geomorphology
Plain	Mean ± SD	5.23 ± 2.52	5.56 ± 2.26	5.03 ± 2.72	6.56 ± 15.47	5.72± 15.87	7.12 ± 15.22	79.87 ± 102.61	44.08 ± 36.43	96.43 ± 118.13
	Min	1	2	1	0.0003	0.0003	0.002	2	2	2
	Max	9.5	9.5	8.5	117.76	107.35	117.76	600	146	600
Slope	Mean ± SD	3.97 ± 1.77	3.86 ± 2.04	4.20 ± 1.15	12.65 ± 57.16	17.67 ± 69.39	2.49 ± 4.21	66.83 ± 62.98	69.91 ± 61.22	61.86 ± 66.30
	Min	2	2	3	0.0002	0.0003	0.0002	4	4	4
	Max	8.5	8.5	5.5	628.32	628.32	24.93	325	325	300
Habitat
Garrigues	Mean ± SD	4.59 ± 2.18	4.35 ± 1.92	4.72 ± 2.36	5.11 ± 12.26	3.08 ± 5.66	6.15 ± 14.43	82.22 ± 103.07	59.12 ± 55.46	91.06 ± 115.19
	Min	1	2	1	0.0002	0.0003	0.0002	2	2	2
	Max	8.5	7.5	8.5	117.76	31.73	117.76	600	325	600
Maquis	Mean ± SD	4.50 ± 2.50	3.67 ± 1.53	5.75 ± 3.89	27.64 ± 94.19	50.45 ± 132.19	5.82 ± 6.59	76.00 ± 62.95	77.00 ± 66.96	75.24 ± 61.39
	Min	2	2	3	0.014	0.088	0.014	8	8	10
	Max	8.5	5	8.5	628.32	628.32	133.99	217	217	207
Micro-forest	Mean ± SD	5.43 ± 2.91	5.43 ± 2.91	–	9.17 ± 18.73	9.17 ± 18.73	–	46.51 ± 37.10	46.51 ± 37.10	–
	Min	2	2	–	0.0006	0.0006	–	2	2	–
	Max	9.5	9.5	–	107.35	107.35	–	152	152	–

Table 3. Generalized linear models results: effects of locality and ecological parameters on plant density.

Source	χ^2	DF	p
Locality	0.42	1	0.515
Site geomorphology	1.40	1	0.237
Substrate	0.20	1	0.653
Habitat type	4.99	2	0.082
Human trampling	14.44	3	0.002
Site geomorphology × substrate	0.01	1	0.984
Site geomorphology × habitat type	4.09	2	0.129

Observations = 40. Model: –LogLikelihood = 13.80; χ² = 27.60; DF = 11; p = 0.004.

Fig. 2. Effect of human trampling on plant density, plant size and fitness.

Plant size

The mean plant size was 8.61 ± 35.48 dm³, ranging from a minimum of 0.20 × 10⁻³ dm³ to a maximum of 628.32 dm³. Plants growing in CM (mean value 11.21 ± 48.68 dm³) were larger than those growing in ST (mean value 6.11 ± 13.72 dm³); indeed, a higher percentage of large plants (plant volume > 25.00 dm³) was observed in CM compared with ST (10.3% of the plants measured for CM and 3.1% for ST).

All ecological variables analysed in our study had a statistically significant effect on plant size, as well as the interactions between geomorphology and substrates and between geomorphology and habitat significantly affected plant size (p < 0.001; Table 4); in particular, larger plants were found in areas with the following ecological features: presence of structured soil, on the slopes, in the maquis habitat (Table 2) and in areas with intensive human trampling (Figure 2). In contrast, smaller plants were observed in areas without human trampling (5.55 ± 13.08 dm³).

Table 4. Generalized linear models results: effects of locality and ecological parameters on plant size (plant volume).

Source	χ^2	DF	p
Locality	105,189.91	1	< 0.001
Site geomorphology	122,056.28	1	< 0.001
Substrate	62,790.64	1	< 0.001
Habitat type	186,002.73	2	< 0.001
Human trampling	432,806.60	3	< 0.001
Site geomorphology × substrate	3533.98	1	< 0.001
Site geomorphology × habitat type	1,030,061.80	2	< 0.001

Observations = 378. Model: –LogLikelihood = 1,665,332.51; χ² = 3,330,665; DF = 11; p < 0.001.

Reproductive traits

Flowering season occurs from March to late May, but isolated flowers were present until mid-June. In CM, the flowering season ranged from March to mid-May, and peak of flowering was recorded in mid-April. The flowering season in ST ranged from early April to mid-May, and peak flowering was recorded in early May. Generally, the fruiting season lasts 1 month; here, it lasted from April to July, and the peak occurred from mid-May to mid-June in both localities.

The mean percentage of reproductive plants was 76.72%, ranging from 64.86% (CM) to 88.08% (ST). The mean number of flowers per plant was 33.23 ± 46.98, ranging from 1 to 327. Notably, the number of ST flowers per reproductive plant (39.01 ± 44.52) was higher than CM (30.42 ± 48.00).

The mean number of fruits per reproductive plant was 58.03 ± 86.47 and it was higher in ST than in CM (78.49 ± 107.15 and 36.69 ± 49.43, respectively). All ecological variables analysed had a statistically significant effect on the number of fruits per plant, as well as the interactions between geomorphology and habitat significantly affected plant size (p < 0.001; Table 5); plants displayed a higher mean number of fruits in deep and structured soil and in lowland areas, preferably in the garrigue and maquis habitats (Table 2). The mean number of fruits per plant increased as human trampling intensified, with mean values ranging from 50.33 ± 71.17 (absent human trampling) to 113.18 ± 190.62 (intense human trampling; Figure 2).

Table 5. Generalized linear models results: effects of locality and ecological parameters on fruit output per plant.

Source	χ^2	DF	p
Locality	1125.31	1	< 0.001
Site geomorphology	68.21	1	< 0.001
Substrate	296.44	1	< 0.001
Habitat type	394.20	2	< 0.001
Human trampling	1005.00	3	< 0.001
Site geomorphology × substrate	0.34	1	0.558
Site geomorphology × habitat type	201.26	2	< 0.001

Observations = 378. Model: –LogLikelihood = 2696.70; χ² = 5393.41; DF = 11; p < 0.001.

A positive correlation between number of fruits and plant size was observed (r = 0.76, r² = 0.57 and p < 0.001; Figure 3).

Fig. 3. Relationships between plant size (plant volume) and reproductive capacity (number of fruits per plant) in Helianthemum caput-felis. The following equation was used: No. of fruits = – 0.320 + 0.517 × plant volume; r² = 0.57; r = 0.76; p < 0.001.

Overall, approximately 20% of the fruits were empty, ranging from 17.5% for CM to 22.5% for ST. The mean number of seeds per fruit was 4.29 ± 1.39, ranging from 1 to 7 seeds per fruit; the fruits collected in CM had a higher mean number of seeds per fruit (4.31 ± 1.40) compared with those in ST (4.27 ± 1.37), but this difference was not statistically significant (Mann–Whitney U-test, p > 0.05).

IUCN assessment at regional level

The same threats were detected in all localities where H. caput-felis was found, including areas with scattered plants growing in the southern part of the Sinis Peninsula. The major threat is tourism and other outdoor activities (such as human trampling), followed by the expansion of agricultural activities (such as agriculture and wood plantations) and the invasion of exotic plants. All of these pressures result in reduced population size due to habitat loss and fragmentation. Stochastic environmental events (e.g. landslides) pose a significant potential threat. All of these threats could result in the disappearance of small localities with a consequent reduction in EOO and AOO values, as well as in the number of localities. Additionally, continuous, human-induced habitat degradation will continue to occur in a predictable manner in the future.

Based on the EOO (3.99 km²), AOO (16 km², i.e. four cells of 2 × 2 km), decline rate and number of locations (one, sensu IUCN 2014a), we propose that H. caput-felis should be included on the Red List categorization of Critically Endangered at the regional level, based on the following formula: B1ab(i,ii,iii,v) + 2ab(i,ii,iii,v).

Discussion

Our study highlighted important information regarding the distribution pattern, reproductive traits and ecological requirements of H. caput-felis, which are relevant issues for developing future conservation measures for this species in Sardinia. To the best of our knowledge, no exhaustive studies have been performed on the central populations of H. caput-felis, and the present study is the first investigation to analyse the population traits of this threatened species under natural conditions.

In the Sardinian population, density did not vary between localities and ecological features (geomorphology, substrate and habitat type), suggesting the absence of strong differences in ecological stress among different local conditions. The actual distribution of H. caput-felis in Sardinia is restricted in the two main localities, which constitute “ecological islands” (acting as local refuges) in coastal areas strongly modified by human activity. Hence, H. caput-felis could represent a “refuge-model” plant in Sardinia, with a range that simply occupies a reduced fraction of a wider habitat from which it is excluded by intensive human-induced habitat degradation, similar to several endemic species in the Mediterranean basin (Thompson 2005). In this restricted ecological context, H. caput-felis represents one of the principal species among coastal vegetation types.

Our data detected a significant effect of topography, landform and habitat type on plant fitness. Helianthemum caput-felis prefers lowland areas with deep, structured soil due to the amount of water and resources available and morphological stability. Moreover, as highlighted by the mean number of fruits per plant, H. caput-felis was ecologically optimum in garrigues, whereas it produced few fruits in micro-forests.

Considering previous studies (Tébar, Gil, and Llorens 1997; Rodriguez-Pérez 2005), the phenological pattern of H. caput-felis is similar in the Sardinian and Balearic populations. Information on the reproductive biology of endangered plants is crucial for predicting their survival capacity and establishing the appropriate measures for their conservation (e.g. Schemske et al. 1994; Cogoni, Fenu and Bacchetta 2015). However, despite the ecological importance of the Cistaceae in the Mediterranean Basin, few studies have been carried out on this topic for this family (Herrera 1992; Talavera, Gibbs, and Herrera 1993; Rodríguez-Pérez 2005; Guzmán, Narbona, and Vargas 2011). Local studies have been carried out on Helianthemum species (Tébar, Gil, and Llorens 1997; Rodriguez-Perez 2005; Aragón and Escudero 2008), but it remains one of the genera for which reproductive biology remains less documented.

The H. caput-felis population in Sardinia is mainly composed of reproductive plants (78.31%), while the percentage of saplings and seedlings is relatively low; this result, together with the high number of seeds produced each year, suggests that seedling establishment represent the main critical stage for this plant. However, further studies are needed to understand whether this is related to marginalization of the population or the lack of suitable micro-sites for plant establishment. In fact, in the case of locally restricted and threatened species, many seeds may end up in unsuitable areas depending on their peculiar ecological requirements (e.g. Cogoni et al. 2012). The lowest percentage of reproductive plants was observed in CM (two-thirds of the total), suggesting a greater rate of recruitment, whereas in ST, four-fifths of the plants were reproductive. The similar number of empty fruits per plant and number of seeds per fruit between the two localities (lower values in ST) could be related to the impact of agricultural activities that determine habitat fragmentation. However, at the population level, multiplying the number of viable seeds per fruit per the estimated number of reproductive plants per locality by the mean number of fruit per plant gives approximately 75 million viable seeds for CM and 137 million for ST; hence, seed production is not a limiting factor for this plant.

Reproductive limitations were not detected for this species (i.e. fruit and seed set, pollination service and seedling survival on natural populations) in previous studies (e.g. Rodríguez-Pérez 2005); considering our findings on the high seed output, the increasing rarity of this species is probably a direct result of the destruction of its natural habitat.

Habitat fragmentation or destruction caused by human disturbance is currently considered one of the main factors responsible for reducing population viability and increasing the extinction risk of rare plants and/or of marginal and small populations (Schemske et al. 1994; Holsinger 2000; Schleuning and Matthies 2009). As a consequence, many threatened or rare plants are confined to naturally fragmented habitats or ecological islands that might be separated by large inhospitable areas, as noted in this study.

Among other factors, human trampling is an important threat for threatened endemic species in Sardinia (Quilichini and Debussche 2000; Fenu, Mattana, and Bacchetta 2011; Fenu et al. 2013; Rossi et al. 2015). A consistent reduction in reproductive traits was observed in plant populations subjected to human trampling (e.g. Rossi et al. 2006; Fenu et al. 2013), resulting in a serious threat to the persistence of the population. Accordingly, in the present study, a negative effect of human trampling was observed on plant density. Surprisingly, human trampling enhanced the plant size and the rate of fruit production, suggesting that the reproductive plant of H. caput-felis is tolerant to direct damage and probably benefits from the reduction of inter-specific and intra-specific competition. However, considering the critical limitation in seedling recruitment, as suggested by the negative effect on plant density, human trampling should be considered a significant threat to the persistence of the Sardinian population.

Red lists highlight the most pressing issues in biodiversity conservation matters, representing taxa that are closer to extinction (Rossi et al. 2013, 2015). In this context, Red lists may be policy-relevant for promoting conservation efforts, but they cannot be considered policy-prescriptive (Bilz et al. 2011; IUCN 2014a). Very little attention has been given to border populations in the application of the IUCN criteria at the regional level (IUCN 2003; Miller et al. 2007) and only recent research assigned some Italian marginal populations of widespread species to high-risk categories (e.g. Gargano et al. 2007; Del Vecchio et al. 2012; Rossi et al. 2015). This study confirms this trend for H. caput-felis, and it allows us to raise the risk category at the regional level for this species from Lower Risk (LR; Conti, Manzi, and Pedrotti 1997) to Critically Endangered (CR). Although criterion B is biased by restricted range and could overestimate the extinction risk, it is strongly supported by the population decline observed in Sardinia. However, to confirm or reject the assumed worse performance and higher vulnerability of border populations, an extensive and integrative approach that compares all population over a wide temporal context is needed; in addition, the effect of population attributes (e.g. population size or structure) or the particular conditions where populations occur (e.g. human trampling intensity analysed in this study) should be taken into account to separate the role of local and positional factors that drive populations (Garcia, Goñi, and Guzmán 2009). This approach is essential for all plants at the margin of their distribution range to plan appropriate conservation measures to reduce the extinction risk of these populations.

Implication for conservation

According to the European Strategy for Plant Conservation (Planta Europa 2008), populations at the border of their distribution area, such as H. caput-felis in Sardinia, should be considered of high interest, with a need for urgent conservation measures. In situ and ex situ (e.g. seed conservation in seed banks, cultivation in botanical gardens) conservation efforts should be improved. In particular, greater emphasis should be given to minimizing the range of negative impacts, including unsustainable tourism and recreation use (Fenu, Mattana, and Bacchetta 2011; Ballantyne and Pickering 2013; Fenu et al. 2013; Rossi et al. 2015). Therefore, touristic and recreational activities should be regulated in known localities, and no new pathways should be opened. A management strategy should exclude trampling in different portions of the population to facilitate the plant recruitment process and population renewal.

In addition, an ex situ conservation strategy must be implemented and the seeds collected could be used for future reinforcement or reintroduction of this species in suitable areas, as realized in Spanish regions or following low-cost programmes carried out in Sardinia (Cogoni et al. 2013). These actions may be extremely important for conservation in a changing climate (Sala et al. 2000; Godefroid et al. 2011). Moreover, because many threats will affect plant species over the next few decades (e.g. climate change, biological invasions), long-term monitoring programmes must be developed to reveal changes in the species conservation status (Balmford, Green, and Jenkins 2003; Fenu et al. 2015), as well as monitoring both the vegetation and the human threat dynamics.

Acknowledgements

The authors thank Sergio Sulis for his help with the fieldwork. We thank Elsevier Language Editing Services for linguistic revision (Project nr: 47013).