SEARCH

SEARCH BY CITATION

Keywords:

  • species introduction;
  • population dynamics;
  • matrix population model;
  • population restoration;
  • long-term monitoring;
  • species management strategies;
  • vital rates;
  • plant survival;
  • plant fecundity;
  • cliff-dwelling species

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • 1
    In a context of increasing human impact on ecosystems and species distributions, population restoration (introductions, reintroductions, reinforcements) is an essential management tool, especially for plant species with limited colonization ability. However, detailed demographic surveys following restoration and comparisons of demographic rates between restored and natural populations, although essential for identifying the key factors of restoration success, are lacking.
  • 2
    We compared the demography over 10 years of six natural and two experimentally introduced populations of the narrowly endemic, cliff-dwelling, self-incompatible plant species Centaurea corymbosa. We analysed the fate of two cohorts of individuals that emerged simultaneously from seed introduction and natural germination. We then built a matrix model of population dynamics (using 6 years of data) and compared the demographic rates and asymptotic growth rate between the natural and introduced populations.
  • 3
    Overall, survival rates were higher in the introduced than in the natural populations, either due to better habitat conditions at the cliff scale or to better conditions in microsites selected for seed introduction compared to those reached by chance following natural seed dispersal.
  • 4
    In contrast, introduced populations exhibited lower fecundity than natural populations, probably due to the introduction protocol which led, in combination with self-incompatibility, to severely reduced mate availability.
  • 5
    Despite clear differences in population dynamics between introduced and natural populations, no significant difference in the asymptotic growth rates could be detected, because the higher survival compensated for the lower fecundity in introduced populations.
  • 6
    Synthesis and applications. Creating new populations of C. corymbosa in suitable unoccupied sites seems straightforward, provided that the introduction protocol allows sufficiently high fecundity. This key parameter for restoration success can be optimized by sowing seeds from several sources at high density and in several consecutive years, which should increase mate availability for self-incompatible flowering individuals. We suggest that population introduction might be successful for many (endemic) plant species whose geographical range is mainly limited by low colonization ability, especially in Mediterranean landscapes. We show that the simultaneous monitoring of restored and natural populations enables identification of the key parameters to be targeted for management optimization of restored populations.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

The introduction or reintroduction of populations can be a useful conservation tool (Falk, Millar & Olwell 1996). All other things being equal, increasing the number of populations must increase the probability of species’ persistence over a given time period. This remains true whatever the relative importance of demographic, environmental and genetic stochasticities on the variation of population sizes (reviews in Menges 2000 and Oostermeijer 2003). The success of (re-)introductions relies on the establishment of viable populations (Griffith et al. 1989). This implies that introduced individuals not only survive, but also produce enough offspring to grow and reproduce in their turn, so that the population demographic growth rate allows population persistence over generations without recurrent human intervention (Sarrazin & Barbault 1996). However, the vast majority of published studies have focused on the technical aspects of introductions, such as habitat choice, habitat management, or planting method (reviews in Bottin et al. 2007 and Menges 2008). Although these studies provide valuable information about how to establish individuals in natural habitats, they do not give any information on population persistence. Detailed demographic monitoring following introduction is essential to the assessment of (re-)introduction success. With some exceptions (Cully 1996; Bell, Bowles & McEachern 2003; Liu et al. 2004; Holl & Hayes 2006; Bottin et al. 2007), demographic analysis of (re-)introduced populations has generally been neglected.

In order to understand the reasons for a possible failure of population establishment after (re-)introduction, it is crucial to compare the population dynamics between introduced and natural populations surveyed simultaneously. It would be speculative to ascribe the possible failure of an introduction to a particular demographic rate (fecundity or survival of a given stage) without knowing the relative value of this demographic rate compared to the values in natural populations. To date, few studies have compared the demography of (re-)introduced and natural populations. Previous studies have focused on part of the life cycle (reproductive success in Morgan 2000; survival in Maschinski, Baggs & Sacchi 2004) or used demographic data collected over the whole life cycle (Pavlik & Espeland 1998). Recently, Bell, Bowles & McEachern (2003) used demographic modelling to compare growth rate, elasticities, and minimum viable population size between one restored and one natural population of Cirsium pitcheri, but no statistical test was conducted to assess the significance of the observed demographic differences.

In this study, we compare the population dynamics over 10 years between two experimentally introduced populations and all six known natural populations of the narrowly endemic, cliff-dwelling, monocarpic-perennial Centaurea corymbosa Pourret (Asteraceae). Although many nearby unoccupied cliffs appear suitable, the species distribution area (3 km2) seems limited by low colonization ability (details in Colas, Olivieri & Riba 1997). This very low colonization ability means that persistence of each of the six small extant populations is critical for the persistence of the species. However, no impending threat on any population was detected at the beginning of this study in 1994, and seed introductions to unoccupied cliffs were initiated mainly as an experimental goal, rather than to decrease the species extinction risk.

We conducted two sets of demographic analyses. First, we compared the fate of introduced and natural cohorts from emergence to death. Secondly, we integrated demographic data into a stage-structured matrix population model and compared the asymptotic growth rates (λ), the demographic vital rates, and their contributions to the difference in λ between natural and introduced populations. In doing so, we tested whether introduced individuals and their offspring performed as well as those from natural populations. We examined whether the observed differences between introduced and natural populations could be accounted for either by habitat conditions or by population parameters resulting from the introduction scheme. Our results show that population introduction is a promising conservation tool for species whose geographical range appears mainly limited by dispersal ability. The comparative approach of this study may be used in climate change studies where the introduction of populations outside the current range may be necessary to ensure species persistence.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

study species

Centaurea corymbosa is endemic to a 3-km2 area representing about 6% of the limestone ‘Massif de la Clape’ in Southern France (Colas, Riba & Molina 1996). It occurs in six populations ranging from 18 to 200 flowering plants on average. Seedling emergence occurs mostly in the autumn following summer dispersal of seeds which do not form any persistent seed bank (details in Supplementary Material Appendix S1). Plants then grow as vegetative rosettes during 3–12 years (mean 5·5 years) in rock clefts on the top of cliffs and in nearby rocky areas where competition with other species is very low. At flowering (late spring), self-incompatible reproducing individuals produce 1–200 insect-pollinated capitula (Fréville 2001; Kirchner et al. 2005). The frequency of monocarpic individuals (which die after flowering) is about 95%. Seed dispersal by wind and ants is very limited (Colas, Olivieri & Riba 1997; Riba et al. 2005; Imbert 2006), and pollen flow between populations is low (Hardy et al. 2004a). Consequently, populations are genetically highly differentiated for both allozyme (Colas, Olivieri & Riba 1997) and microsatellite markers (Fréville, Justy & Olivieri 2001), although they are only separated by 0·3–2·3 km.

experimental introductions and data collection

Two populations were created in 1994 on the top of unoccupied cliffs where C. corymbosa had never been recorded, near the natural populations (see map in Supplementary Material Appendix S2). Sites of about 2 m2 were selected for seed introduction on the basis of their apparent ecological similarity with the species habitat in natural populations (i.e. medium-sized clefts on stable rocks with almost no soil and vegetation). In November 1994, we introduced seeds from three natural populations into rock clefts from eight and 13 sites, respectively, in the two newly created populations. In October 1995, we introduced seeds from four natural populations into 18 new sites in one introduced population (details in Supplementary Material Appendix S1). Each introduction site contained 50 seeds. We then monitored emergence, survival and size (rosette diameter) every 3 months in the first week of September, December, March and June, at every introduction site. Similar demographic data were collected from 41 permanent quadrats in the six natural populations (3–10 quadrats per population; details in Supplementary Material Appendix S1).

cohort analyses

We compared the fate of emerging seedlings from the 1994 and 1995 cohorts between introduced and natural populations. After pooling data over populations for both population types (introduced and natural, hereafter called PopType), we analysed the pre-reproductive mortality from June in year y to June in year y + 1 using a known-fate Kaplan–Meier model as implemented in the program mark (White & Burnham 1999). This procedure allowed us to model survival as a product of binomial estimates for a group of individuals whose fate is known. The proportion of flowering individuals among all living individuals from these cohorts in a given year was analysed as a binomial variable using a generalized linear model with a logit link specifying a binomial error using r (r Development Core Team 2007). For both pre-reproductive mortality and proportion of flowering individuals, the explanatory variables were year, the age of individuals, cohort and PopType. We applied a model selection approach using the Akaike's Information Criterion (AIC; Burnham & Anderson 2002) and Akaike weights (ω) to identify the most relevant models (details in Supplementary Material Appendix S1).

matrix model construction

Matrix models of population dynamics were built as in Fréville et al. (2004), using demographic data from 1998 to 2004, since flowering in the introduced populations first occurred in 1998. The projection matrix model was based on a pre-breeding census in early June with a 1-year time step, and a three-stage life cycle (Fig. 1). All vital rates were first calculated for each population and each pair of years over the period 1998–2004. However, because samples were too small to obtain accurate estimates of all vital rates for each population in each pair of years, we built one matrix per PopType (introduced and natural) and per pair of years, pooling the data over populations within each PopType, and one matrix per PopType pooling the data over years and populations. The asymptotic growth rate λ was computed as the dominant eigenvalue of each matrix.

image

Figure 1. Life-cycle graph and corresponding transition matrix for analysis of population dynamics in Centaurea corymbosa. The life cycle has a 1-year time step and a census time in June, at the peak of flowering. Stage-1 individuals are young rosettes (< 1 year old), stage-2 individuals are old rosettes (> 1 year old), and stage-3 individuals are flowering plants. The fecundity term f is the number of seedlings (recorded every 3 months) between time t − 1 and time t over the number of flowering plants in June at time t − 1; s0 is the seedling survival probability until June; s1, s2 and s3 are the survival probabilities of young rosettes, old rosettes and flowering plants, respectively, from t − 1 to t; α1, α2 and α3 are the flowering probabilities at t of young rosettes, old rosettes and flowering plants at t − 1, respectively, that survived until t.

Download figure to PowerPoint

comparison of life-cycle parameters

To investigate variation in the life-cycle parameters s0, s1, s2, and α2 (Fig. 1) according to year and PopType, we used a generalized linear model with a logit link specifying a binomial error using r. We then followed the model selection approach using the AIC to assess the relevance of including PopType in addition to year in the models. We also carried out Fisher's exact tests on s0, s1, s2, and α2 between both PopTypes for each year. The 24 two-tailed P values (four variables times 6 years) were compared to a 5% level of significance corrected for multiple tests using the method of Simes (1986), which is less conservative and more powerful than the classical Bonferroni procedure (Sarkar & Chang 1997). Because individual fecundity values were not available from our demographic monitoring, we simply performed a paired t test to compare the means of f (Fig. 1), obtained by averaging the annual values over populations, between PopTypes (log-transformed data). Finally, we did not examine the patterns of variation of α1, s3, and α3, because they either represented rare events (α1 and s3) or were based on a very small number of individuals (α3), and had a negligible contribution to the variation of λ[Fréville et al. 2004, and results of the life-table response experiment (LTRE) analysis below].

comparison of λs and contribution of vital rates to the difference in λ

We used a non-parametric randomization procedure based on random permutations of individuals (Caswell 2001; Fréville et al. 2004) to test for the difference in λ between PopTypes, using data pooled over years. We thereby generated the distribution of the differences in λ between PopTypes under the null hypothesis that the fate of an individual was independent of the PopType to which it belonged (i.e. no difference between natural and introduced populations). The observed difference in λ between PopTypes was then tested against the null distribution using a two-tailed test. Finally, we performed an LTRE, Caswell 2001) to investigate how the difference in each vital rate contributed to the observed difference in λ between PopTypes (details in Supplementary Material Appendix S1).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

cohort analyses

The 1994 cohort comprised 356 and 194 seedlings from introduced and natural populations, respectively, whereas the 1995 cohort comprised 272 and 206 individuals, respectively. In June 2004, all individuals from the 1994 cohort had died, but 19 individuals from the 1995 cohort were still alive (16 and 3 in the introduced and natural populations, respectively). The model including the effects of year, age, and PopType within the 1995 cohort only had the highest probability (ω = 0·644) of being the best model among all candidate models for pre-reproductive mortality (Table 1). Aside from the expected effects of year and age, this result was explained by a higher pre-reproductive mortality in the natural than in the introduced populations for the 1995 cohort, whereas the 1994 cohort suffered from a much higher early mortality in both PopTypes (Fig. 2a). The models which included a cohort effect in addition to the effects cited above, or a cohort x PopType interaction, also had some support (ω = 0·237 and 0·088, respectively). Evidence ratios (ratios of Akaike weights between models) showed that even the ‘worst’ model of the three ‘best’ models was more than 500 000 times more likely than any model which did not include PopType as a predictor variable (year + age or year + age + cohort, see Supplementary Material Appendix S3).

Table 1.  Model selection results for pre-reproductive mortality and proportion of flowering individuals among all living individuals from two cohorts in Centaurea corymbosa. k is the number of parameters in the model. ΔAIC is the difference in values of the Akaike Information Criterion (AIC) between each model and the ‘best’ model having the lowest AIC, and ω is the Akaike weight of each model. The notation ‘PopType (cohort 95)’ denotes an effect of the population type (introduced vs. natural) within cohort 1995 only. Complete results including other candidate models that failed to explain the observed data (ΔAIC ≥ 10) can be found in Supplementary Material Appendix S3
ModelDeviancekΔAICω
Pre-reproductive mortality
 year + age + PopType (cohort 95)20·9200·00·644
 year + age + cohort + PopType (cohort 95)20·9212·00·237
 year + age + cohort × PopType 20·9224·00·088
 age + year × PopType + cohort × PopType 5·5316·60·024
 year + age × PopType + cohort × PopType 8·2319·30·006
Flowering probability
 age + cohort + PopType (cohort 95)19·2110·00·527
 age + cohort × PopType18·2121·00·322
 age + PopType (cohort 95)25·0103·80·078
 age + PopType26·1104·90·046
 age + cohort + PopType26·0116·80·017
 Age × cohort + cohort × PopType10·9209·70·004
image

Figure 2. Cumulated percentage of individuals (a) which died before flowering, and (b) which reached flowering among all seedlings from the 1994 and 1995 cohorts.

Download figure to PowerPoint

The proportion of flowering individuals was, on average over years, higher in the introduced than in the natural populations for the 1994 cohort (0·132 and 0·056, respectively). For the 1995 cohort, it was the reverse (0·109 and 0·206, respectively). The best model to explain variation in the proportion of flowering individuals included the effects of age, cohort, and PopType in the 1995 cohort only (ω = 0·527, Table 1). This was probably accounted for by the lower survival of the 1994 cohort, so that this cohort was weighted less than the 1995 cohort in the analyses. However, the model including age and the cohort × PopType interaction had substantial support as well (ω = 0·322).

Over the whole 1994–2004 decade, the number of individuals that reached flowering were 2 (out of 194) and 10 (out of 356) from the 1994 cohort in the natural and introduced populations, respectively, and 24 (out of 206) and 38 (out of 272), respectively, from the 1995 cohort (Fig. 2b). After pooling both cohorts, the proportion of individuals that reached flowering among all individuals that emerged was slightly, although not significantly, higher in the introduced (8·2%) than in the natural populations (6·7%, two-tailed P value from Fisher's exact test: 0·40, Fig. 2b).

matrix model analyses

The mean demographic rates in each PopType are shown in Table 2. Overall, survival rates (s0, s1 and s2) were higher in the introduced populations and fecundity (f) was higher in the natural populations. The observed variation in seedling survival (s0) and in survival of young rosettes (s1) was best explained by the model including an interaction between year and PopType (Table 3), due to the fact that the natural populations performed better than the introduced ones in some years (Fig. 3). The variation in survival of old rosettes (s2) was best explained by the model including year and PopType effects without interaction, although the model with interaction had some support (Table 3). The best model explaining the variation of flowering probability of old rosettes (α2) only included a year effect, although the model which included both year and PopType (without interaction) had also some support (Table 3).

Table 2.  Mean demographic vital rates in the natural and introduced populations of Centaurea corymbosa, and sensitivity of λ to each vital rate. Sensitivities were calculated from the arithmetic mean matrix of ANat and AIntro, the corresponding matrices from data pooled over years and populations in natural and introduced populations, respectively
Vital rateNatural populationsIntroduced populationsSensitivity
f14·0915·1150·018
s0 0·5840·7020·263
s1 0·3520·4150·422
s2 0·6900·7940·778
s3 0·0400·0660·156
α1 0·0020·0000·305
α2 0·1240·1150·889
α3 0·7500·7500·006
Table 3.  Results of model selection procedures for the analysis of demographic rates in Centaurea corymbosa. PopType is for population type (introduced vs. natural). ΔAIC and ω are as in Table 1
Life-cycle parameterBest modelAlternative model (ΔAIC < 4)
s0year × PopType (ω = 0·998)
s1year × PopType (ω = 0·927)
s2year + PopType (ω = 0·767)year × PopType (ΔAIC = 3·03, ω = 0·168)
α2year (ω = 0·660)year + PopType (ΔAIC = 1·96, ω = 0·248)
image

Figure 3. Demographic rates over 1998–2004 in the natural (black circles) and introduced (white triangles) populations. (a) s0, (b) s1, (c) s2, and (d) α2. Significant differences between natural and introduced populations detected from Fisher's exact tests are indicated: *, P < 0·05; **, P < 0·01. Parentheses indicate the differences that were no longer significant after correcting for multiple tests.

Download figure to PowerPoint

Among the 24 Fisher exact tests between PopTypes performed per year on the survival rates and α2, six were significant at the 5% level (Fig. 3). Three remained significant after correcting for multiple tests: higher values in the introduced populations for s0 in 2 years and s1 in 1 year. The difference in f between PopTypes (Table 2 and Fig. 4) was highly significant (two-tailed P value from paired t test: 0·0099).

image

Figure 4. Fecundity (f) over 1998–2004 in the natural (black circles) and introduced (white triangles) populations.

Download figure to PowerPoint

After pooling data over years and over populations within each PopType, λ equalled 0·887 and 0·914 in introduced and natural populations, respectively. This difference was not significant (randomization procedure, P = 0·511).

The fecundity f and the survival of old rosettes s2 explained most of the difference in λ between PopTypes (Fig. 5). The high contribution of f was due to its large difference between PopTypes, in favour of natural populations, as the sensitivity of λ to f was very low (Fig. 4, Table 2). The high contribution of s2 was explained by the combined effect of the high sensitivity of λ to this vital rate and a relatively large difference in this vital rate between PopTypes. On the contrary, because the flowering probability of old rosettes (α2) differed weakly between PopTypes, its contribution to the difference in λ was small, despite the fact that the sensitivity of λ to this vital rate was the highest (Table 2). Finally, survival probabilities s0 and s1 contributed slightly to the difference in λ between PopTypes (in favour of the introduced populations), and s3, α1 and α3 had a negligible contribution (Fig. 5).

image

Figure 5. Contribution of demographic rates to the difference in λ between the natural and introduced populations in Centaurea corymbosa. Positive contributions indicate lower values of demographic rates in introduced populations, while negative contributions reflect lower values in the natural populations.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

main demographic differences between natural and introduced populations

Over 10 years, there was a clear trend towards higher survival in the introduced compared to the natural populations of C. corymbosa. Whereas mortality of the 1994 cohort was very high in all populations, the lower pre-reproductive mortality in the introduced than in the natural populations was clear for the 1995 cohort during the 1995–1997 period (Fig. 2a and Table 1). This trend was confirmed by the analyses of survival rates based on matrix models, built over the last 6 years (1998–2004). The survival of old rosettes (s2) proved significantly higher in the introduced than in the natural populations, and the other survival rates (s0 and s1) also tended to be higher in the introduced populations, although the interaction between PopType and year complicated the picture (Tables 2 and 3, and Fig. 3).

Flowering probability did not show any clear trend among the natural and introduced populations, either in the cohort analyses or in the analysis of α2 from the demographic matrices. On the contrary, the threefold higher mean f value in the natural populations was highly significant and undoubtedly indicates that the introduced populations suffered from low fecundity compared to natural populations.

Thus, the natural and introduced populations clearly did not behave similarly from a demographic point of view. The LTRE analysis showed that the observed differences in fecundity (f) and in the survival of rosettes (s2) were the most important in generating a difference between the asymptotic growth rates of natural and introduced populations. However, they acted in opposite ways such that no overall significant difference in λ could be detected between both types of populations.

differences in fecundity between the natural and introduced populations

The difference in fecundity between the introduced and the natural populations, although already significant, was probably underestimated. Indeed, f in all populations was estimated as the annual number of emerging seedlings (recorded every 3 months) over the number of flowering plants in the previous June. Because we did not demarcate the monitoring area of introduction sites, every seedling that was detected in the vicinity was included in the data, so that the estimates of f in introduced populations are unbiased. On the contrary, data from natural populations were collected exclusively in permanent quadrats within which the initial density was much higher than outside. The seeds dispersing out of the quadrats outnumbered those entering within, which led to an underestimation of f in natural populations. As a likely consequence of this mechanism, unpublished results show a significant negative temporal trend in the proportion of flowering plants occurring within quadrats among all flowering plants in the natural populations. Thus, f as well as λ was underestimated in natural populations, so that the real difference in these variables between introduced and natural populations is probably greater than what has been observed.

The large difference in f between the natural and the introduced populations could be due either to a lower number of seeds produced per flowering plant or a lower emergence rate in the introduced populations. We could roughly estimate emergence rates in natural populations by combining demographic and seed set data in 1995 and 1996. In both years, the emergence rate of the naturally dispersed seeds was approximately 1·5%, while it was higher than 30% from the seeds sown in the clefts of the introduction sites in 1994 and 1995 (Colas 1997), suggesting that there was no barrier to emergence at these sites. On the contrary, the number of seeds produced per flowering plant was probably smaller in the introduced populations than in the natural ones. By counting seeds and unexpanded ovules in capitula collected from the first five flowering plants in introduced populations in 1998, estimates of fertilization rate appeared very low (0–11%, Fréville 2001) compared to usual values in natural populations (mean = 58% over 1995 and 2002, Kirchner et al. 2005).

Four source populations were used to form the pool of introduced seeds. Mixing seed sources may result in low fecundity due to outbreeding depression (Montalvo & Ellstrand 2001). However, in a crossing experiment using plants from the four natural populations from which introduced seeds originated, inter-population crosses led to equal or higher fertilization rates and equal abortion rates compared to intra-population crosses (Fréville 2001). Thus, we can exclude the possibility that the low reproductive success in the introduced populations was due to outbreeding depression.

Important determinants of fertilization success in cross-pollinated plant species are population size and density (Kunin 1997; Brys et al. 2007). In the introduced populations of C. corymbosa, only a few flowering plants were observed every year, many of which were isolated by several tens or even hundreds of metres from the closest conspecific flowering individuals. In 2002, for instance, the number of conspecific flowering individuals within 10 m of each flowering individual was on average 6·5 in the natural populations (Kirchner et al. 2005), but only 0·7 in the introduced populations. Fertilization rate in the natural populations of C. corymbosa is known to be positively related to the density of conspecific flowering individuals occurring within a 10-m radius (Colas, Olivieri & Riba 2001; Kirchner et al. 2005). This relationship is for the most part due to a reduction in the availability of compatible mates at low density rather than to a reduced visitation rate (Hardy et al. 2004b). In the introduced populations, the proportion of compatible fathers among all flowering plants might be high due to the mixing of seed sources. However, the number and the density of flowering plants (and hence potential fathers) occurring in the introduced populations was always lower than the lowest values observed in the natural populations. Thus, it is likely that the generalist pollinators of C. corymbosa were less attracted by the small and low-density introduced populations than by the natural ones, and/or that they carried a much higher fraction of non-specific pollen to C. corymbosa stigmas in the introduced populations. In another restored-population study, focusing on population structure rather than vital rates, Hegland, van Leeuwen & Oostermeijer (2001) also found such an Allee effect, (i.e. lower-than expected numbers of juveniles as a result of small population sizes).

differences in survival between the natural and introduced populations

Survival rates have generally been considered as too low in restored populations to ensure population establishment (e.g. Pavlik & Espeland 1998; Drayton & Primack 2000; Holl & Hayes 2006). In C. corymbosa, plant survival was higher overall in the introduced than in the natural populations and all significant survival differences that could be detected were in favour of the introduced populations. This is, to our knowledge, the second example of a significantly higher survival in (re)introduced populations compared to natural populations (after Helenurm 1998).

Enhanced intraspecific competition due to higher density is unlikely to be responsible for the lower survival in natural populations (unpublished results even show significant positive density dependence on s0). Two other non-exclusive hypotheses can be advanced. First, it is possible that the cliffs selected for the introductions provided on average better habitat conditions for survival than cliffs where natural populations occur, due for instance, to lower competition with other plant species. Secondly, even given similar habitat quality at the cliff scale, it could be that the microsites selected for seed introduction (clefts about 1-cm wide) provided more suitable conditions than microsites reached by chance after seed dispersal in natural populations. This could explain the higher survival of the introduced 1995 cohort in the cohort analysis and the higher survival of old rosettes (s2) in the introduced populations. However, the survival rates s0 and s1 from the 1998–2004 data used in the matrix model were calculated from naturally dispersed seeds only, and those rates also tended to be higher in the introduced populations (although less clearly for s1 than for s0), which suggests an environmental effect at the cliff scale. Nevertheless, the microsite hypothesis cannot be rule out at this stage. Survival rates are likely to depend upon complex interactions between year, population and microsite which would necessitate very large data sets to be analysed in a single integrative model.

implications for conservation and management

Our study showed that suitable sites occur in the Massif de la Clape for the creation of new C. corymbosa populations. Indeed, the populations that we experimentally introduced exhibited on average higher plant survival than the natural populations. Their lower fecundity was most likely due to our introduction scheme (low density of sown seeds) rather than to habitat features (see also Morgan 2000). A simulation model of population dynamics, integrating C. corymbosa demographic parameters with intraspecific competition, positive density dependence on fecundity, and self-incompatibility, confirmed this point (Kirchner, Robert & Colas 2006). The extinction probability of an introduced population after 100 years was very sensitive to the density of sown seeds because it determined the subsequent density of compatible mates (see Hardy et al. 2004a for the pollen dispersal curve) and fecundity following introduction.

Generally, population restoration experiments require intensive management interventions such as the ex-situ growth of restoration propagules (DeMauro 1994; Cully 1996; Maschinski & Duquesnel 2006; Bottin et al. 2007), soil preparation (Sinclair & Catling 2003; Maschinski, Baggs & Sacchi 2004), grass cover reduction (Pavlik, Nickrent & Howald 1993), caging protection (Maschinski, Baggs & Sacchi 2004), or supplemental watering (DeMauro 1994). Compared to these studies, creating new populations of C. corymbosa was straightforward. First, many cliffs appear suitable all over the massif of which only 3 km2 out of 50 km2 harbour natural populations. Secondly, sowing seeds into clefts on the top of cliffs did not present any technical problem. Thirdly, no expensive or time-consuming habitat management was required following the introduction. However, for future introductions, sowing seeds several times, eventually at yearly intervals, might be necessary to avoid an Allee effect due to the low density of flowering plants until introduced populations reach a minimum size and density. This strategy may also be applied to reinforce the two introduced and six natural extant populations, although the low λ value calculated in the present study for natural populations is probably underestimated, as discussed above.

We intensively monitored plant emergence, growth and reproduction in both introduced and natural populations. This comparative approach was necessary to assess the success of introduction strategies and to identify the possible causes of introduction failures (Sarrazin & Barbault 1996; Menges 2008). The 10-year monitoring period allowed us to observe the fate of almost all introduced individuals from emergence to death and to analyse vital rates from the whole life cycle, including fecundity, over several years. For species with shorter generation times than in C. corymbosa, similar results might be obtained from shorter monitoring times.

Finally, this study confirms that colonization ability can limit plant distribution, even at the local scale (Primack & Miao 1992). This may be especially true in many narrowly endemic plants in the Mediterranean, which tend to occur on steep rocky slopes and cliffs (Lavergne et al. 2004). In such stable habitats where community succession and habitat closure are prevented (Thompson 2005), and which are generally surrounded by unsuitable patches such as farmland and forest, colonization ability has probably been selected against (Colas, Olivieri & Riba 1997), while traits promoting local persistence have been selected for (Thompson 2005). Hence, the occurrence of unoccupied suitable habitats near to (but not adjacent to) natural populations may be common in Mediterranean endemic species. This makes population introduction a very promising management strategy for such species, because it is easier to artificially disperse seeds to suitable unoccupied sites than, for example, to restore a degraded habitat. More generally, our study illustrates how comparative demographic studies can provide conservation biologists and field managers with valuable knowledge on the key factors determining the viability of natural and restored populations.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

This work was funded by the Conservatoire Botanique National Méditerranéen de Porquerolles, the Muséum National d’Histoire Naturelle (BQR 2005–2007 to B.C.), the French Ministry of Research (ACI ‘Ecologie Quantitative’ to I.O. and D. Jolly, University of Montpellier, 2002–2005; ANR ‘ABIME’ to J. Thompson, CEFE-CNRS Montpellier, 2006–2008; PhD grants to F.K. and C.B.), and by the French and Spanish Ministries of Foreign Affairs (PICASSO program to M.R. and I.O.). Many people helped in the field, in particular Sandrine Maurice, Maria Mayol, Sheila Luijten, Thomas Bataillon, Christophe Petit and Ophélie Ronce. The manuscript benefited from helpful discussions with Alexandre Robert and useful comments from Eric Menges, an anonymous reviewer and the editorial board of JAE. This is publication no. 2008–50 of the Institut des Sciences de l’Evolution of Montpellier.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Bell, T.J., Bowles, M.L. & McEachern, A.K. (2003) Projecting the success of plant population restoration with viability analysis. Population Viability in Plants: Conservation, Management, and Modeling of Rare Plants (eds C.A.Brigham & M.W.Schwartz), pp. 313348. Springer-Verlag, Berlin.
  • Bottin, L., Le Cadre, S., Quilichini, A., Bardin, P., Moret, J. & Machon, N. (2007) Re-establishment trials in endangered plants: a review and the example of Arenaria grandiflora, a species on the brink of extinction in the Parisian region (France). Ecoscience, 14, 410419.
  • Brys, R., Jacquemyn, H., De Bruyn, L. & Hermy, M. (2007) Pollination success and reproductive output in experimental populations of the self-incompatible Primula vulgaris. International Journal of Plant Sciences, 168, 571578.
  • Burnham, K.P. & Anderson, D.R. (2002) Model Selection and Inference: A Practical Information–Theoretic Approach, 2nd edn. Springer-Verlag, Berlin.
  • Caswell, H. (2001) Matrix Population Models: Construction, Analysis, and Interpretation, 2nd edn. Sinauer, Sunderland, MA, USA.
  • Colas, B., Riba, M. & Molina, J. (1996) Statut démographique de Centaurea corymbosa Pourret (Asteraceae), Hormatophylla pyrenaica (Lapeyr.) Cullen & Dudley (Brassicaceae), et Marsilea strigosa Willd. (Marsileaceae-Pteridophyta), trois plantes rares dans le sud de la France. Acta botanica Gallica, 143, 191198.
  • Colas, B. (1997) Centaurea corymbosa: chronique d’une extinction annoncée. PhD Thesis, University of Tours, Tours, France.
  • Colas, B., Olivieri, I. & Riba, M. (1997) Centaurea corymbosa, a cliff-dwelling species tottering on the brink of extinction: a demographic and genetic study. Proceedings of the National Academy of Sciences of the United States of America, 94, 34713476.
  • Colas, B., Olivieri, I. & Riba, M. (2001) Spatio-temporal variation of reproductive success and conservation of the narrow-endemic Centaurea corymbosa (Asteraceae). Biological Conservation, 99, 375386.
  • Cully, A. (1996) Knowlton's cactus (Pediocactus knowltonii) reintroduction. Restoring Diversity: Strategies for Reintroduction of Endangered Plants (eds D.A.Falk, C.I.Millar & M.Olwell), pp. 403410. Island Press, New York.
  • DeMauro, M.M. (1994) Development and implementation of a recovery program for the federal threatened lakeside daisy (Hymenoxys acaulis var. glabra). Restoration of Endangered Species: Conceptual Issues, Planning and Implementation (eds M.L.Bowles & C.J.Whelan), pp. 298321. Cambridge University Press, Cambridge.
  • Drayton, B. & Primack, R.B. (2000) Rates of success in the reintroduction by four methods of several perennial plant species in eastern Massachusetts. Rhodora, 102, 299331.
  • Falk, D.A., Millar, C.I. & Olwell, M. (1996) Restoring Diversity. Island Press, New York.
  • Fréville, H. (2001) La centaurée de la Clape: biologie d’une espèce rare et réflexions méthodologiques. PhD Thesis, University of Montpellier, Montpellier, France.
  • Fréville, H., Colas, B., Riba, M., Caswell, H., Mignot, A., Imbert, E. & Olivieri, I. (2004) Spatial and temporal demographic variability in the endemic plant species Centaurea corymbosa (Asteraceae). Ecology, 85, 694703.
  • Fréville, H., Justy, F. & Olivieri, I. (2001) Comparative allozyme and microsatellite population structure in a narrow endemic plant species, Centaurea corymbosa Pourret (Asteraceae). Molecular Ecology, 10, 879889.
  • Griffith, B., Scott, J.M., Carpenter, J.W. & Reed, C. (1989) Translocation as a species conservation tool: status and strategy. Science, 245, 477480.
  • Hardy, O.J., González-Martínez, S.C., Fréville, H., Boquien, G., Mignot, A., Colas, B. & Olivieri, I. (2004a) Fine-scale genetic structure and gene dispersal in Centaurea corymbosa (Asteraceae) I. Pattern of pollen dispersal. Journal of Evolutionary Biology, 17, 795806.
  • Hardy, O.J., González-Martínez, S.C., Colas, B., Fréville, H., Mignot, A. & Olivieri, I. (2004b) Fine-scale genetic structure and gene dispersal in Centaurea corymbosa (Asteraceae) II. Correlated paternity within and among sibships. Genetics, 168, 16011614.
  • Hegland, S.J., Van Leeuwen, M. & Oostermeijer, J.G.B. (2001) Population structure of Salvia pratensis in relation to vegetation and management of Dutch dry floodplain grasslands. Journal of Applied Ecology, 38, 12771289.
  • Helenurm, K. (1998) Outplanting and differential source population success in Lupinus guadalupensis. Conservation Biology, 12, 118127.
  • Holl, K.D. & Hayes, G.F. (2006) Challenges to introducing and managing disturbance regimes for Holocarpa macradenia, an endangered annual grassland forb. Conservation Biology, 20, 11211131.
  • Imbert, E. (2006) Dispersal by ants in Centaurea corymbosa (Asteraceae): what is the elaiosome for? Plant Species Biology, 21, 109117.
  • Kirchner, F., Robert, A. & Colas, B. (2006) Modelling the dynamics of introduced populations in the narrow-endemic Centaurea corymbosa: a demo-genetic integration. Journal of Applied Ecology, 43, 10111021.
  • Kirchner, F., Luijten, S.H., Imbert, E, Riba, M., Mayol, M., González-Martínez, S.C., Mignot, A. & Colas, B. (2005) Effects of local density on insect visitation and fertilization success in the narrow-endemic Centaurea corymbosa (Asteraceae). Oikos, 111, 130142.
  • Kunin, W.E. (1997) Population size and density effects in pollination: pollinator foraging and plant reproductive success in experimental arrays of Brassica kaber. Journal of Ecology, 85, 225234.
  • Lavergne, S., Thompson, J.D., Garnier, E. & Debussche, M. (2004) The biology and ecology of narrow endemic and widespread plants: a comparative study of trait variation in 20 congeneric pairs. Oikos, 107, 505518.
  • Liu, G., Zhou, J., Huang, D. & Li, W. (2004) Spatial and temporal dynamics of a restored population of Oryza rufipogon in Huli Marsh, South China. Restoration Ecology 12, 456463.
  • Maschinski, J. & Duquesnel, J. (2006) Successful reintroductions of the endangered long-lived Sargent's cherry palm, Pseudophoenix sargentii, in the Florida Keys. Biological Conservation, 134, 122129.
  • Maschinski, J., Baggs, J.E. & Sacchi, C.F. (2004) Seedling recruitment and survival of an endangered limestone endemic in its natural habitat and experimental reintroduction sites. American Journal of Botany, 91, 689698.
  • Menges, E.S. (2000) Population viability analyses in plants: challenges and opportunities. Trends in Ecology & Evolution, 15, 5156.
  • Menges, E.S. (2008) Restoration demography and genetics of plants: when is a translocation successful? Australian Journal of Botany, 56, 187196.
  • Montalvo, A.M. & Ellstrand, N.C. (2001) Nonlocal transplantation and outbreeding depression in the subshrub Lotus scoparius (Fabaceae). American Journal of Botany, 88, 258269.
  • Morgan, J.W. (2000) Reproductive success in reestablished versus natural populations of a threatened grassland daisy (Rutidosis leptorrhynchoides). Conservation Biology, 14, 780785.
  • Oostermeijer, J.G.B. (2003) Threats to rare plant persistence. Population Viability in Plants: Conservation, Management, and Modeling of Rare Plants (eds C.A.Brigham & M.W.Schwartz), pp. 1758. Springer-Verlag, Berlin.
  • Pavlik, B.M. & Espeland, E.K. (1998) Demography of natural and reintroduced populations of Acanthomintha duttonii an endangered serpentinite annual in northern California. Madroño 45:3139.
  • Pavlik, B.M., Nickrent, D.L. & Howald, A.M. (1993) The recovery of an endangered plant. I. Creating a new population of Amsinckia grandiflora. Conservation Biology, 7, 510526.
  • Primack, R.B. & Miao, S.L. (1992) Dispersal can limit local plant distribution. Conservation Biology, 6, 513519.
  • r Development Core Team (2007) r: A Language and Environment for Statistical Computing. r Foundation for Statistical Computing, Vienna, Austria.
  • Riba, M., Mignot, A., Fréville, H., Colas, B., Imbert, E., Vile, D., Virevayre, M. & Olivieri, I. (2005) Variation in dispersal traits in a narrow-endemic plant species, Centaurea corymbosa Pourret (Asteraceae). Evolutionary Ecology, 19, 241254.
  • Sarkar, S.K. & Chang, C.-K. (1997) The Simes method for multiple hypothesis testing with positively dependent test statistics. Journal of the American Statistical Association, 92, 16011608.
  • Sarrazin, F. & Barbault, R. (1996) Reintroduction: challenges and lessons for basic ecology. Trends in Ecology & Evolution, 11, 474478.
  • Simes, R.J. (1986) An improved Bonferroni procedure for multiple tests of significance. Biometrika, 73, 751754
  • Sinclair, A. & Catling, P.M. (2003) Restoration of Hydrastis canadensis by transplanting with disturbance simulation: results of one growing season. Restoration Ecology, 11, 217222.
  • Thompson, J.D. (2005) Plant Evolution in the Mediterranean. Oxford University Press, Oxford, UK.
  • White, G.C. & Burnham, K.P. (1999) Program mark: survival estimation from populations of marked animals. Bird Study, 46, 120139.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Appendix S1. Detailed materials and methods

Appendix S2. Map of Centaurea corymbosa populations

Appendix S3. Detailed results from model selection procedures applied to pre-reproductive mortality and flowering probability of rosettes from two cohorts of emerging seedlings in natural and introduced populations of Centaurea corymbosa

Please note: Blackwell Publishing are not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
JPE_1536_sm_Appendix S1-S3.doc1079KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.