Potential impact of harvesting on the population dynamics of two epiphytic bromeliads

Acta Oecologica 59 (2014) 52e61

Contents lists available at ScienceDirect

Acta Oecologica journal homepage: www.elsevier.com/locate/actoec

Original article

Potential impact of harvesting on the population dynamics of two epiphytic bromeliads
ndez-Apolinar b, Teresa Valverde b Tarin Toledo-Aceves a, *, Mariana Herna Red de Ecología Funcional, Instituto de Ecología, A.C., Carretera antigua a Coatepec no. 351, El Haya, CP 91070, Xalapa, Veracruz, Mexico
noma de M
Departamento de Ecología y Recursos Naturales, Facultad de Ciencias, Universidad Nacional Auto exico, Circuito Exterior S/N Ciudad Universitaria, 04510 M
exico, D.F., Mexico a

b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 13 February 2014 Accepted 19 May 2014 Available online

Large numbers of epiphytes are extracted from cloud forests for ornamental use and illegal trade in Latin America. We examined the potential effects of different harvesting regimes on the population dynamics of the epiphytic bromeliads Tillandsia multicaulis and Tillandsia punctulata. The population dynamics of these species were studied over a 2-year period in a tropical montane cloud forest in Veracruz, Mexico. Prospective and retrospective analyses were used to identify which demographic processes and life-cycle stages make the largest relative contribution to variation in population growth rate (l). The effect of simulated harvesting levels on population growth rates was analysed for both species. l of both populations was highly influenced by survival (stasis), to a lesser extent by growth, and only slightly by fecundity. Vegetative growth played a central role in the population dynamics of these organisms. The l value of the studied populations did not differ significantly from unity: T. multicaulis l (95% confidence interval) ¼ 0.982 (0.897e1.060) and T. punctulata l ¼ 0.967 (0.815e1.051), suggesting population stability. However, numerical simulation of different levels of extraction showed that l would drop substantially even under very low (2%) harvesting levels. Matrix analysis revealed that T. multicaulis and T. punctulata populations are likely to decline and therefore commercial harvesting would be unsustainable. Based on these findings, management recommendations are outlined. © 2014 Elsevier Masson SAS. All rights reserved.

Keywords: Demography Elasticity Life table response experiments (LTRE) Non-timber forest products (NTFP) Population matrix models Retrospective analysis Tillandsia multicaulis Tillandsia punctulata

1. Introduction Epiphytes are extremely vulnerable to forest disturbance, fragmentation (Holbrook, 1991; Zhu et al., 2004; Wolf, 2005) and climate change, due to total dependence on established vegetation in order to complete their life cycle and a high sensitivity to microclimate (Benzing, 1998; Nadkarni and Solano, 2002). These plants are important elements of the diversity and physiognomy of tropical montane cloud forest (TMCF), one of the most threatened ecosystems worldwide (Scatena et al., 2010). Epiphyte populations in small forest remnants face high extinction risks since they are particularly vulnerable to environmental, demographic and genetic stochasticity (Menges, 1998; Young et al., 1996; Turner et al., 1994; Sodhi et al., 2008). In addition to the detrimental effect of forest disturbance and deforestation on epiphyte populations, over-

* Corresponding author. Tel.: þ52 228 8421800x4217. E-mail addresses: [email protected], [email protected] (T. Toledo
ndez-Apolinar), Aceves), [email protected] (M. Herna [email protected] (T. Valverde). http://dx.doi.org/10.1016/j.actao.2014.05.009 1146-609X/© 2014 Elsevier Masson SAS. All rights reserved.

exploitation is also an important factor affecting remnant epiphyte populations (Turner et al., 1994; Wolf, 2005; Haeckel,
n et al., in press). 2009; Mondrago The use of epiphytic bromeliads for ceremonial and ornamental purposes is widespread in Mexico and Guatemala (Hietz et al., 2002; Flores-Palacios and Valencia-Díaz, 2007). In the construction of just three ceremonial floral arches, 2438 inflorescences of Tillandsia multicaulis and 109 of Tillandsia punctulata were recorded (FloresPalacios and Valencia-Díaz, 2007). In just four municipalities of the Mexican state of Veracruz, more than 70 floral arches are constructed every year to celebrate religious festivities (Haeckel, 2008). Moreover, the illegal collection of epiphytes for commercial purposes is commonplace in this region: each collector is expected to gather between 2000 and 6000 rosettes of Tillandsia kirchhoffiana per day (Toledo-Aceves, unpublished data). During this process, targeted trees are stripped bare of epiphytes, including orchids, ferns and mosses. This type of harvesting technique is extremely destructive and completely disregards the recovery capacity of the resources involved. Demand for epiphytic bromeliads has increased in recent years in southern Mexico (Toledo-Aceves and Wolf, 2008; Haeckel,

T. Toledo-Aceves et al. / Acta Oecologica 59 (2014) 52e61

2008), adding to the already significant loss of TMCF in this region (Toledo-Aceves et al., 2011). As a result, various bromeliad species have now become locally rare and are listed as threatened (FloresPalacios and Valencia-Díaz, 2007; Toledo-Aceves and Wolf, 2008; Haeckel, 2009). Implementation of carefully designed extraction systems in areas with robust populations has been proposed as a viable alternative to contribute to epiphyte conservation (Verhoeven and Beckers, 1999; Wolf and Konings, 2001). However, in order to develop appropriate management guidelines for sustainable harvesting systems, prior demographic analysis of the target population is required. The canopy is a very restrictive environment from an ecological perspective (Benzing, 1990) and therefore many epiphytes present low individual growth rates, reach maturity relatively late in life and suffer high mortality rates (Hietz et al., 2002; Schmidt and Zotz,
n et al., 2014). Previous studies evaluating the 2002; Mondrago population dynamics of epiphytic bromeliads report that many populations display l values below unity, indicating that they are likely in decline. Such is the case with Tillandsia brachycaulos
n et al., 2004), Catopsis sessiflora, T. deppeana, (Mondrago T. multicaulis, T. punctulata (Winkler et al., 2007), T flexulosa (Wester
nand Zotz, 2010), T. violacea and T. macdougalii (Mondrago Chaparro and Ticktin, 2011). Conversely, Zotz (2005) report average l values above unity for Werauhia sanguinolenta and Valverde and Bernal (2010) for T. recurvata. The study of population dynamics is a key tool with which to evaluate the likelihood of population growth or decline (Olmsted
and Alvarez-Buylla, 1995; Ticktin et al., 2002). However, various demographic processes and life cycle routes can be responsible for such numerical trends in each population (Mandujano et al., 2007). Prospective analyses (i.e. traditional sensitivity and elasticity matrices) allow us to estimate how potential changes in matrix entries may affect l. On the other hand, retrospective analyses (also called life table response experiments) allow us to examine the relative contribution of the observed variation in each vital rate to the recorded variation in l, either overtime or between populations
et al., 2005). The combined use of (Caswell, 1989; Martínez-Balleste these two analyses has proved to be a powerful tool for assessing harvest sustainability in non-timber forest products (Zuidema et al., 2007; Schmidt et al., 2011). In the present study, we carried out a traditional demographic analysis of undisturbed populations of the epiphytic bromeliads T. multicaulis and T. punctulata over a 2year period in a TMCF in the Mexican state of Veracruz. We then used both prospective and retrospective analysis to identify which vital rates, life-cycle stages and demographic processes make the largest contributions to the population dynamics of the species. To contribute to the design of appropriate management strategies, we used the demographic data to carry out numerical simulations assessing the potential impact of different harvesting regimes (including a non-harvesting scenario) on the population dynamics of these species. To support our management recommendations, we also used the approach proposed by Zuidema et al. (2007) to identify the vital rates that could be crucial in determining longterm population trends (i.e. those with high elasticities and high LTRE contributions). 2. Methods 2.1. Study area The study was carried out in the tropical montane cloud forest of
s Tlalnelhuayocan, located in the the municipality of San Andre centre of the Mexican state of Veracruz (19 3100300 N, 97 000 2500 W; elevation: 1660 m; mean annual temperature: 18  C). Total annual precipitation reported for the region in 2010 and 2011 was 1567

53

and 1757 mm, respectively (National Water Commission; http:// www.conagua.gob.mx/). Three main seasons occur annually in this area: a cool and relatively dry season from November to March; a warm, dry season from April to May; and a warm, rainy season from June to October (Williams-Linera, 1997). The forest in the study area is highly fragmented, and features fragments of ca. 1e100 ha immersed within a matrix of pasturelands, crops (mainly maize) and abandoned pastures occupied by early successional forest; the fragments predominantly consist of secondary forests. Mean tree diameter at 1.3 m height (dbh) ¼ 27.4 cm (only trees  10 cm dbh were included; Toledo-Aceves et al., 2014). The dominant tree species in this forest are Quercus delgadoana S. Valencia, Nixon & M.L. Kelly, Liquidambar styraciflua L. and Q. lancifolia Schltdl. & Cham. (Toledo-Aceves et al., 2014). 2.2. Study species The two studied epiphytic bromeliad species share several biological features. Both produce rosettes that reach 40 cm in height (Espejo-Serna et al., 2005). Individuals (genets) are sympodial, composed of a succession of rosettes, shoots, or modules that show determinate growth. The leading rosette dies after fruiting, but module production continues, thus the genets are polycarpic. This vegetative growth (i.e., production of new rosettes) can be observed even in relatively small plants (infants and juveniles). A single genet therefore simultaneously present several reproductive shoots that die after fruiting, and the genet regress to a smaller size category from one year to the next. Individual rosettes also decrease in size due to herbivory, leaf senescence, and/or desiccation. Vegetatively produced rosettes were never observed to detach from parent rosettes. Both studied species feature a C3 photosynthetic metabolism and are distributed mainly in TMCF in Mexico and Central America. T. multicaulis is generally found at altitudes of 1150e1900 m, and T. punctulata at altitudes between 1400 and 1800 m. Both inhabit TMCF but can occupy relatively more temperate forests, and T. punctulata colonize more tropical ecosystems, such as tropical rain forest, and tropical dry forest (Espejo-Serna et al., 2005; Toledo-Aceves et al., 2012a). Seeds of both species are plumose, ca. 3 mm long with a ca. 2-cm long appendix. Fruits mature in about 11 mo and seeds are wind-dispersed during the dry season (October to May; Toledo-Aceves et al., 2012b). Rosettes of T. multicaulis have tank morphology. They produce one to five red inflorescences per rosette. A fertile rosette produces approximately 1129 ± 102 seeds (mean ± s.e, N ¼ 7; Toledo-Aceves et al., 2012b) and a genet comprise up to 14 rosettes; a single rosette produces as many as eight new rosettes per year. The morphology of T. punctulata is intermediate between tank and atmospheric. Each fertile rosette bears only one red inflorescence that produce approximately 571 ± 89 seeds (N ¼ 11; ToledoAceves et al., 2012b). A genet comprises up to 36 rosettes; a single rosette produces as many as 14 new rosettes per year. Single rosette abundance per tree in the forest is: 342.6 ± 58.4 in T. multicaulis and 299.6 ± 101.7 in T. punctulata (Toledo-Aceves et al., 2014). In this study, we analysed the population dynamics of T. multicaulis and T. punctulata based on the identification and monitoring of genets. Previous studies on epiphytic bromeliads have used stage classes defined by the size of single rosettes (Hietz et al., 2002; Zotz et al., 2005; Winkler et al., 2007). Others have considered the size of the whole plant (genet), or its number of shoots, in order to subdivide the population into size categories
n-Chaparro and Ticktin, (Valverde and Bernal, 2010; Mondrago 2011). The latter categorizations based on genet size recognize that genets of differing sizes show differential growth, survival and
n-Chaparro reproductive rates (Clark-Tapia et al., 2005;Mondrago

54

T. Toledo-Aceves et al. / Acta Oecologica 59 (2014) 52e61

and Ticktin, 2011). Plant harvesting in our study system is mainly carried out by illegal harvesters, who remove whole genets from host trees. We therefore used a plant categorization method based on genet size (rather than rosette or ramet size) to analyse the population dynamics of the studied species. The studied populations were not subjected to anthropogenic disturbance during our study period; however, the forest owners reported the occurrence of illegal harvesting in the region in previous years although there are no reports of the intensity of these activities (ToledoAceves unpublished data). 2.3. Field methods This study was carried out from 2009 to 2011. An initial sample of 105 genets of T. multicaulis (making a total of 197 rosettes) and 62 genets of T. punctulata (for a total of 159 rosettes) were tagged and monitored for two years. These were established on the branches and trunk (T. multicaulis: main trunk ¼ 83%, branches ¼ 17%; T. punctulata: main trunk ¼ 39%, branches ¼ 61%) of four L. styraciflua and three Q. lancifolia trees. Single rope techniques were used to climb the trees and monitor the epiphytes. Plants were grouped into five categories based on total plant size and reproductive status (Appendix A). To obtain total plant size, we measured and summed the length of all the rosettes that comprised a genet (Clark-Tapia et al., 2005; Esparza-Olguín et al., 2005) (Appendix A). Rosette size was measured with a calliper as the length from the base to the tip of the longest leaf. The seedling category consisted of plants 1 cm in height. For both species, the limit between infants and juveniles was established as the midinterval between the upper size threshold of seedlings and the lower size threshold of non-reproductive adults, following Winkler et al. (2007). The presence of inflorescences was used to define the adult category; the lowest size at which plants were fertile was 31 cm of rosette height in T. multicaulis, and 21 cm in T. punctulata. Based on cumulative height, adult plants were divided in two categories (Appendix A). Seed germination and seedling establishment experiments were carried out in the field to obtain estimates of survival and transition probabilities for these life-cycle stages (Toledo-Aceves et al., 2012b). These had to be obtained experimentally due to the difficulty of distinguishing between species at the seedling stage. The resulting probability of seed germination immediately after dispersal and subsequent seedling survival probability were incorporated into the matrices as part of the fecundity values. Transition probabilities from the seedling to the infant category were calculated based on the results of field experiments on seedling survival and growth (Toledo-Aceves et al., 2012b), assuming that seedlings would present a constant growth rate until they reach 2 cm in height. 2.4. Demographic analysis We constructed two 5  5 population projection matrices (periods 2009e2010 and 2010e2011) for each studied species in order to characterize their population dynamics and project the potential impact of harvesting. These Lefkovitch matrices summarize the survival, growth and fecundity rates of the different life-cycle stages into which populations are structured (Appendix A). Sizespecific survival and growth were calculated from the proportion of individuals in each size class that died, remained in the same size class or grew to the following size class from one year to the next. Since no deaths were observed in some adult size-classes and years (A2 in T. multicaulis for 2010e2011; and A2 in T. punctulata for 2009e2010), the probability of dying in these categories was calculated based on average annual growth rate, according to

Enright and Ogden (1979). Following this procedure, we estimated the time needed for a plant to move from one category to the next, or to reach the maximum observed size, at which point it is assumed that death followed. Since the average plant expansion rates were measured for the periods 2009e2010 and 2010e2011, we used the data of the corresponding year to calculate the transition probabilities of the corresponding matrix. In addition to growth and survival, the contribution of adult individuals to the first size category i.e. fecundity rate, was estimated in terms of seedlings. Seeds of the two species studied remain viable for only a few months (Toledo-Aceves, pers. obs.); thus, reproduction of adult plants over a one-year period is expressed in the form of seedlings emerging the following year. Fecundity entries are therefore given as the average number of seedlings produced by an individual of each adult category (A1 and
n et al., 2004) as: A2) (Mondrago

Fj ¼ cj *s*d*g




nj

where Fj is the fecundity of an average individual in the j-th category; cj is the total no. of capsules produced by all individuals in the j-th category; s ¼ mean no. of seeds per capsule (T. multicaulis ¼ 242, T. punctulata ¼ 179; n ¼ 20);; d ¼ probability of successful dispersal (here we used a value of 0.037, which was reported for T. deppeana in TMCF; García-Franco and Rico-Grey, 1988); g ¼ germination probability based on the results of field experiments (T. multicaulis ¼ 0.137, T. punctulata ¼ 0.318) and nj is the number of individuals in the j-th category. Due to a small sample size in some categories of both species (Appendix A), we applied a bootstrapping method to our data set to estimate mean vital rates (survival, growth, and fecundity), and mean population growth rate with confidence intervals, as well as stable size distribution, reproductive values and elasticities for the different matrix entries (Caswell, 2001; Efron, 1982; Levin et al., 1996). We performed separate bootstrap resampling procedures for each year and each species, using the percentile method with a bootstrap sample size of 1000 (Caswell, 2001). 2.5. Perturbation analysis Prospective and retrospective analyses were used to evaluate the relative importance of different matrix elements to the population growth rate (l). Elasticity (prospective) analysis evaluates the relative contribution of different matrix elements to the population growth rate (l) by providing information on the potential impact on l of relatively small changes in matrix elements (de Kroon et al., 1986, 2000). Retrospective analysis evaluates the contribution of the observed variation in matrix entries to the observed variation in l by providing information on the extent of the variation in l as a result of variation in particular matrix entries (Horvitz et al., 1997; Caswell, 2001). The elements of the elasticity matrices were calculated as:



eij ¼ vi wj < w; v > * aij l where w and v represent the vectors corresponding to the stable size-class distribution and size-specific reproductive values, respectively. Since entries in an elasticity matrix sum to unity (de Kroon et al., 1986), they may be used to calculate the relative contribution to l made by the different size-classes and the demographic process, by simply summing the relevant elasticity entries (Silvertown et al., 1993). Life table response experiments (LTRE) are based on the construction of contribution matrices derived from the observed variation in matrix entries, combined with their sensitivities

T. Toledo-Aceves et al. / Acta Oecologica 59 (2014) 52e61

(Heppell et al., 2000; Caswell, 2001). In this study, we carried out two one-way fixed LTREs: (1) an analysis of the temporal variation in matrix entries (within each species) and their contribution to temporal variation in l by comparing each annual l to the l of an average matrix (obtained by averaging the two yearly bootstrapped matrices per species; M. Mandujano pers. comm.); and (2) an analysis of the variation between species, evaluating the contribution of matrix entries to an increase or decrease in the l of the average matrix per species in relation to the l of an overall mean matrix obtained by averaging the two average matrices (one per species). We then followed the procedure introduced by Zuidema et al. (2007) in which elasticity and LTRE contribution values are simultaneously evaluated, by plotting the LTRE contribution of each matrix entry against its elasticity value in the two years of study. Although this approach was originally proposed as a tool to evaluate the sustainability of harvesting through a contrast between managed and undisturbed populations (Zuidema et al., 2007;
n-Chaparro and Ticktin, 2011), we use it in this study Mondrago only to identify the matrix entries that could be crucial in determining population trends (i.e. those with high elasticities and high LTRE contributions). 2.6. Numerical simulations: effects of harvesting regimes on population dynamics
ndez-Apolinar et al. (2006), we used the Following Herna average matrix per species to carry out numerical simulations and evaluate the sustainability of different extractive regimes. For this purpose, we used the mean matrix obtained from the bootstrapped analysis. The main assumption behind the use of average matrices is that the overall demographic trend of the population is determined by an average effect of the demographic response of the population to annual variations in the environment (Mandujano et al., 2001). The simulation series included the extraction of different percentages of genets of T. multicaulis and T. punctulata. Following the suggestion of Wolf and Konings (2001) regarding the harvest of plants from the lower forest stratum only (phorophyte main trunk), extraction regimes were selected based on plant distribution within phorophytes. In 32 trees registered in the studied forest, 32% of T. multicaulis genets and 27% of T. punctulata were located on the main trunk (Toledo-Aceves, unpubl. data). Thus, the extraction of all the plants from the trunk is equivalent to 32% of the total number of genets on the whole tree for T. multicaulis and 27% for T. punctulata. The slopes of the regression of the level of extraction vs. the resulting l values were compared between species using a modified t-test (Zar, 1996).

55

2.8. Expected time to extinction An estimate of expected time to population extinction was obtained by applying the numerical approach used by Valverde (1995), in which time to extinction (te) is estimated as:

te ¼ ðln 0:05Þ=lnl where l is the finite population growth rate obtained from each theoretical simulation carried out for the T. multicaulis and T. punctulata populations. According to this approach, a population is considered virtually extinct when it reaches 5% of its original population size (Valverde and Silvertown, 1998). 3. Results 3.1. Matrix analysis 3.1.1. Tillandsia multicaulis The population growth rate (l) for T. multicaulis calculated from the bootstrapped projection analysis varied annually (Fig. 1). In 2009e2010, l was significantly below unity, suggesting a declining population. However, in 2010e2011, l did not differ from unity, suggesting an annual population increase of ~6% (Fig. 1). Population growth rate for the average matrix was below the unity, but not significantly so given its 95% confidence intervals, suggesting on average a numerically stable population (Table 1). Fecundity was low in both years as a result of low seed germination and dispersal probabilities, as well as a low proportion of reproductive adults (Table 1a). Mortality tended to decrease towards the larger size categories; in fact, none of the plants in category A2 died during the 2010e2011 study period. The matrix entries representing stasis showed high values, indicating that most plants remain in the same size category from one year to the next. Some juvenile and adult plants (A1 and A2) decreased in size due to rosette loss. Some plants in category A2 lost as many as three rosettes from one year to the next, which was the maximum decrease recorded and was reflected in positive retrogression matrix entries (Table 1a). The bootstrapped stable size-class distribution of the average T. multicaulis matrix projected that, at equilibrium, ~70% of the population would correspond to seedlings, infants, and juveniles (vector w in Table 1). This projected population structure differed significantly (G ¼ 50.57, d.f. ¼ 4, P < 0.05) from that observed in the field (row Nx in Table 1a). The main differences were associated

2.7. Potential effect of increased recruitment on population growth We recognize that there is a level of uncertainty in our fecundity and recruitment estimates, since we have used experimental data on seed germination and seedling survival, and the dispersal probability values from a different species and forest, and from an experimental study. We therefore conducted further numerical simulations by directly manipulating the fecundity and seedling establishment matrix entries, to both estimate the impact of these uncertainties on our results, and to evaluate the potential success of particular conservation practices involving seedling establishment (Contreras and Valverde, 2002). Different ecological scenarios were simulated by independently increasing fecundity (i.e. entries A1-S and A2-S) and seedling transitions (i.e. entry SeI) until a l value equal or higher than unity was reached.

Fig. 1. l values with lower and upper 95% confidence intervals obtained from the annual matrices for Tillandsia multicaulis (closed symbols) and T. punctulata (open symbols).

56

T. Toledo-Aceves et al. / Acta Oecologica 59 (2014) 52e61

below unity, according to the 95% confidence intervals (Fig. 1). In addition, the l value of the average matrix was below unity, although not significantly so, suggesting a numerically stable population (Table 1b). The matrix entries with the highest values were those related to the persistence of individuals in their same category (stasis), followed by growth, which most frequently occurred into the following size category. Category A2 was the only one in which retrogression was observed as a result of rosette loss (Table 1b). Some plants in category A2 lost as many as three rosettes from one year to the next. As with T. multicaulis, fecundity was low given that only a small proportion of the plants in our sample reproduced during the study period, which added to the low seed germination and dispersal probabilities. With the exception of A1, mortality tended to decrease towards large size categories and no mortality was observed in category A2. The stable size-class distribution of the two annual matrices projected that, at equilibrium, ~70% of the population would correspond to seedlings, infants, and juveniles (vector w in Table 1b). The stable size-class distribution obtained from the average matrix differed significantly from the observed population structure (row Nx in Table 1b) (G ¼ 3776, d.f. ¼ 4, P < 0.05). Statistical differences were associated with category S, which was more abundant in the observed than in the projected population structure; and also with categories I, J and A1, in which the opposite was true. Size-specific reproductive values were low for plants smaller than 10 cm (i.e. seedlings and infants) but increased considerably in juveniles and adults (vector v in Table 1), with the latter contributing ~70% of the total reproductive value. Elasticity matrices for T. punctulata showed that the matrix entries with the higher elasticity values were persistence of individuals in A1 and A2 (22% and 39%of total elasticity, respectively), followed by category J (16.2%). As in T. multicaulis, stasis was the demographic process with the highest contribution to l, accounting for ~85% of total elasticity (Table 2). Fecundity entries added up to 3.2% of total elasticity, while growth elements accounted for 10.8% and retrogression 0.5% of the l value (Table 2).

Table 1 Average population projection matrices for (a) Tillandsia multicaulis and (b) T. punctulata. Bootstrapping methods were applied to estimate annual vital rates per year per species. The l values (±95% confidence intervals) are shown above each matrix. Entries in the main diagonal are italicized for clarity; Categories are: S ¼ seedling, I ¼ infant, J ¼ juvenile, A1 ¼ adult 1, A2 ¼ adult 2. w ¼ stable size category distribution, v ¼ size-specific reproductive values, qx ¼ size-specific mortality. Category (ntþ1)

Category (nt) S

A1

A2

w

v

(a) T. multicaulis, l ¼ 0.982 (0.897e1.060) S 0 0 0 I 0.170 0.569 0 J 0 0.230 0.563 A1 0 0 0.278 A2 0 0 0 0.830 0.201 0.159 qx Nx 165 29.5 30

I

J

1.067 0 0.036 0.705 0.153 0.106 16

1.732 0 0 0.100 0.806 0.094 20.5

0.379 0.210 0.117 0.149 0.144

0.018 0.105 0.192 0.301 0.384

(b) T. punctulata, l ¼ 0.967 (0.815e1.051) S 0 0 0 I 0.355 0.643 0 J 0 0.113 0.732 A1 0 0 0.183 A2 0 0 0 qx 0.645 0.245 0.085 Nx 382 20 11.5

0.735 0 0 0.549 0.069 0.382 10

2.931 0 0 0.126 0.785 0.089 16

0.239 0.293 0.208 0.177 0.083

0.016 0.045 0.117 0.162 0.660

with category S, which was relatively more abundant in the observed than in the projected population structure; and also with categories I and A1 and A2, where the opposite was true. Sizespecific reproductive values were low for plants smaller than 10 cm (i.e. seedlings and infants), but increased considerably in juvenile and adult plants (vector v in Table 1). Elasticity matrices showed a clear pattern: the matrix entries that most contributed to population growth rate were the persistence of individuals in categories A1 and A2 (22.1 and 31.6% of total elasticity, respectively) followed by category I (10.3%) (Table 2). Thus, stasis was the demographic process that made the highest contribution to l (73.6%). Fecundity entries contributed only 4.4% to the l value, while growth elements contributed 18.9% and retrogression accounted for 3.1% of the total elasticity.

3.2. Life table response experiments (LTRE) The population growth rate of T. multicaulis was highly influenced by temporal variation. Stasis in T. multicaulis contributed negatively to the population growth rate (Fig. 2) in the first year and the opposite pattern was observed during the second year (2010e2011); the positive contribution of growth and stasis marked an increase in the l value. In general, the growth of

3.1.2. T. punctulata The finite rate of population growth (l) estimated from the bootstrapped projection matrix analysis also varied over time in this species (Fig. 1). In 2009e2010, l did not differ from unity, suggesting a growing population with an annual increase of ~3%. However, the l value of the 2010e2011 matrix was significantly

Table 2 Elasticity matrices for T. multicaulis and T. punctulata, corresponding to the average matrices reported in Table 1. Average elasticity values (95% confidence values) obtained from a bootstrap sample size of 1000. The three highest elasticity values are highlighted in bold in each matrix. Only positive entries are presented. Category ntþ1

T. multicaulis S I J A1 A2 T. punctulata S I J A1 A2

Category (nt) S

I

0.044 (0.01e0.07)

0.103 (0.01e0.30) 0.044 (0.01e0.07)

0.032 (0.01e0.06)

0.088 (0.03e0.14) 0.032 (0.01e0.06)

J

0.096 (0.02e0.25) 0.050 (0.02e0.09)

0.162 (003e0.59) 0.032 (0.01e0.06)

A1

A2

0.019 (0e0.05)

0.025 (4E03e0.04)

0.006 (0.05e0.02) 0.221 (0.05e0.49) 0.051 (0.02e0.09)

0.025 (0e0.07) 0.316 (0.07e0.67)

0.024 (0e0.04)

0.008 (4E04e0.04)

0.220 (0.12e0.42) 0.013 (8E-04e0.08)

0.005 (0e0.05) 0.386 (2E03e0.60)

T. Toledo-Aceves et al. / Acta Oecologica 59 (2014) 52e61

57

individuals of categories I and J, as well as stasis in categories A1 and A2, contributed negatively to l for T. multicaulis in 2009e2010and vice versa in 2010e2011. The population growth rate of T. punctulata was also strongly affected by temporal variation. During the first year (2009e2010), fecundity made considerable high positive contributions to the l value of T. punctulata. In contrast, the lower l value of T. punctulata in 2010e2011 was associated mainly with the negative contributions of stasis (Fig. 2). The positive contribution of fecundity in categories A2 and of stasis in category A1 were associated with the l value above unity for T. punctulata in 2009e2010, while the negative contribution of stasis in category A1 meant a lower l value in 2010e2011. The results of the LTRE comparing the population dynamics of the two species showed that the slightly lower l value of T. punctulata compared to T. multicaulis may be explained by the negative contribution of stasis and growth in the former; the positive contributions of fecundity to the average l of this species were insufficient to outweigh such negative contributions (data not shown). At the same time, the positive contributions of growth in T. multicaulis account for the higher average l value in this species. The matrix entries responsible for the differences between the two species were the persistence of individuals in the adult categories and the growth of all categories, with the exception of the Seedling category (data not shown).

3.3. Combining prospective and retrospective analyses The combined analyses using elasticities and LTRE provided contrasting results for the two species: In T. multicaulis, persistence of individuals in theA1 and A2 category made a high contribution to l (negative in 2009e2010 and positive in 2010e2011) and also had a relatively high elasticity value in both years (0.0125 and 0.107 in 2009e2010 and 0.197 and 0.074 in 2010e2011, respectively) (Fig. 3a). A high positive contribution to variation in l in 2009e2010 was also associated with transition of I to J and J to A1, entries with relative low elasticity (0.10, and 0.08). For T. punctulata, a different picture emerged where the A1

Fig. 3. Relationship between LTRE contributions and elasticity values of the different matrix entries for T. multicaulis and T. punctulata. Black symbols ¼ 2009e2010; open symbols ¼ 2010e2011. Matrix entries representing different demographic processes are shown with different symbols: fecundity ¼ squares, growth ¼ triangles, stasis ¼ circles, and retrogression ¼ diamonds.

matrix entry with a high negative contribution had simultaneously low elasticity values; or where entries with high elasticities (stasis of A2 individuals) made low contributions to variation in l (Fig. 3b).

Fig. 2. Results of the life table response experiments (LTRE) conducted to evaluate the contribution to l of (a) the four main demographic processes in 2009e2010 (white) and 2010e2011 (black), and (b) the different matrix entries for T. multicaulis (left) and T. punctulata (right). In (b), the elements in the y axis refer to the subindices of matrix entries (aij) where j refers to the source category (at time t), and i is the fate category (at time t þ 1).

58

T. Toledo-Aceves et al. / Acta Oecologica 59 (2014) 52e61

3.4. Numerical simulations: effects of harvesting regimes on population dynamics The simulation of different levels of extraction showed that population growth rate (l) would drop significantly below unity, even if only a small amount of the plants were harvested from the populations: 2% in T. punctulata and 8% in T. multicaulis (Fig. 4). Although the l values decreased to a lower level in T. punctulata than in T. multicaulis, the slopes of the regression equations did not differ significantly between species (P > 0.05). Regarding the potential effect of increased recruitment on population growth, in T. multicaulis, a 0.5-fold increase in fecundity caused the l value to increase from 0.982 to 1.0005, while a 4-fold increase in the original fecundity values of T. punctulata resulted in l increasing from 0.967 to 1.018. In the same way, lower increases in seedling transition probabilities were sufficient to obtain a l above unity in T. multicaulis compared to T. punctulata; in the former, a 0.5fold increase in this matrix entry was necessary in order to obtain a l value of 1.005, while in T. punctulata, this entry had to be increased 3-fold for l to reach a value of 1.008. When both fecundity and seedling establishment were modified simultaneously in the same proportion, a positive l value was obtained with a 0.25fold increase in T. multicaulis, and a 2-fold increase in T. punctulata.

4. Discussion

3.5. Estimated time to extinction

4.1. Population dynamics

A projected population is considered extinct when the number of individuals fell to 5% of the initial size (Valverde and Silvertown, 1998). Time to extinction was calculated as the time required for each population to reach this critical level, assuming a decline rate given by l (obtained after simulating the different harvesting intensities) (Fig. 5). Given the below unity l value of both average matrices (not considering 95% CI), the projections suggest that both populations would go extinct even with no harvesting. However, T. punctulata would reach the extinction threshold in half the time of T. multicaulis. The simulated removal of a small proportion of individuals (2%) had a more dramatic effect on T. multicaulis than T. punctulata (i.e., the slope is steeper in the former; Fig. 5). According to the simulation, regular harvest of all the plants from the trunk of the phorophytes (a harvesting level equivalent to 32 and 27% of all plants for T. multicaulis and T. punctulata, respectively) would imply the extinction of both populations in approximately 8 years.

The matrix analyses revealed that, over the study period, the populations of T. multicaulis and T. punctulata at the study site presented an overall trend of decline (i.e. l < 1, although not significantly). The l values of these populations concur with previous demographic studies of epiphytic bromeliads, in which a l
n value below unity was reported in 9 out of 12 species (Mondrago et al., 2004, 2014; Zotz et al., 2005; Martínez-García, 2006; Winkler
net al., 2007; Haeckel, 2009; Valverde and Bernal, 2010; Mondrago Chaparro and Ticktin, 2011). However, our matrix results should be treated with some caution. The natural probability of seeds landing on a safe site, as well as seed germination and seedling survival rates in the field are required for a full account of population dynamics. Estimates of germination and seedling survival rates from experimental data may differ from those obtained under natural conditions. However, as has been reported by Winkler et al. (2007) for various Tillandsia species in the same region, seedlings and infants make a small contribution to population growth rate, and therefore the main outcome from the present analysis should not be severely affected by uncertainty in seedling recruitment, as shown by the effect on l of the simulated increase in fertility and seedling establishment. Position in the canopy can also affect
n epiphyte population performance (Winkler et al., 2007; Mondrago et al., in press.). For example, Hietz (1997) reported that species growing on thinner branches experience higher mortality because of the relative insecurity of their support. However, when branchfall was excluded, Tillandsia displayed lower mortality in less humid and shaded zones of the canopy. Hietz et al. (2002) also found that growth and mortality of established bromeliads differ depending on their position in the canopy. Site-specific mortality has also been reported for experimentally sown seedlings of epiphytic bromeliads (Zotz and Vollrath, 2002; Winkler et al., 2005; Toledo-Aceves et al., 2012b). In this study, due to the difficulty of reaching the thinner and outer branches in the canopy, all sampled plants were located on the main trunk and on the branches of the inner canopy. Even though evaluation of the position effect on plants was not the aim of this study, the plants in our sample may perform differently to those in the outer canopy, which may in turn affect the overall population growth rate. While our study samples are not necessarily representative of whole populations and longer-term

Fig. 4. Effect of different simulated harvesting levels on the population growth rates (l) of T. multicaulis and T. punctulata.

Fig. 5. Time to extinction projected under different simulated harvesting intensities. A population is considered extinct when the number of individuals falls to 5% of its initial size.

T. Toledo-Aceves et al. / Acta Oecologica 59 (2014) 52e61

observations are necessary to evaluate population status with higher confidence, this assessment suggests that, under current conditions, the populations may be in decline. We observed temporal variation in the demographic behaviour of both populations, but particularly in T. multicaulis, in which the l value fluctuated from 0.91 in the first year to 1.06 in the second. Similar temporal demographic variation has been observed in other epiphytes, such as T. brachycaulos, T. juncea, T. recurvata, and Wer
n et al., 2004; Zotz, 2005; Winkler auhia sanguinolenta (Mondrago et al., 2007; Valverde and Bernal, 2010). Since yearly variation in vital rates may cause long-term numerical fluctuations in popula
n et al., 2004), it is undoubtedly necessary tion numbers (Mondrago to adopt longer observation periods in order to obtain a clearer picture of the population dynamics of these species. In both species, population growth rate was determined most strongly by adult survival. As the average l values were below unity, the high dependence of l on survival implies that populations decrease in numbers at the rate at which plants die, and that seedling establishment does not make an important contribution to l. The categories with the highest elasticity were A1 and A2 in both species. Retrogression occurred frequently in both species as a result of the loss of one or more rosettes, mostly in reproductive adults that had lost shoots after fruiting. This demographic process had low elasticity values in the two average matrices of both T. multicaulis and T. punctulata, and the same pattern was observed in the LTRE comparison between species. However, the latter results suggest contrasting population dynamics between the two species. Rosette loss in T. punctulata, mainly in adult plants, had a negative impact on l value, since it frequently implied the loss of more than one reproductive rosette at once. In the case of juveniles, biomass must be gained before reaching the reproductive stage: if death occurs, juveniles do not normally have other rosettes with which to replace themselves. Conversely, rosette loss mainly in adult T. multicaulis plants contributed positively to population growth rate, as it was associated with a positive contribution of growth. Due to transfer of resources from the leading shoot, growth rates in rosettes of T. macdougalii were reported to be higher than those of seed-germinated plants, and these even became repro
n-Chaparro and Ticktin, ductive within the first year (Mondrago 2011); it is possible that a similar process may occur in T. multicaulis. In other epiphytic bromeliads, it has also been shown that l is influenced mostly by survival (stasis), to a lesser extent by growth,
n et al., 2004; Zotz, 2005; and only slightly by fecundity (Mondrago
n-Chaparro and Ticktin, 2011; Winkler et al., 2007; Mondrago
n et al., in press.). Vegetative growth certainly plays a Mondrago fundamental role in the life cycle of these organisms. In the studied species, vegetative growth occurred through successive rosette production and replacement, giving rise to a complex and dynamic process, the importance of which is somehow obscured by the fact that it is implicit within other demographic processes according to the categorization used (i.e. transition probabilities to larger sizeclasses, or even persistence in the same category). However, there is no doubt that adult plant survival is fundamental to population maintenance, a fact that has important implications for management, especially since harvesting is targeted towards adult flowering plants. As we expected given its lower l, T. punctulata reached the extinction threshold in a shorter time period than was the case with T. multicaulis, and the simulated harvesting had also a more dramatic effect on T. punctulata than on T. multicaulis. This could be the result of the higher contribution of growth to l in T. multicaulis compared to T. punctulata. While matrix projections shed some light on potential population patterns, they are based on a series of assumptions that generally do not reflect reality. Thus, evaluation of actual harvesting impact on populations should always accompany matrix results in order to offer a more conclusive

59

diagnostic. Based on the present evaluation, we concur with
n-Chaparro and Ticktin (2011) in that the harvesting of Mondrago individuals cannot be buffered by vegetative growth and that epiphytic bromeliads with a high investment in vegetative growth may be even more vulnerable to harvest than their nonvegetatively reproducing relatives. 4.2. Management implications Our demographic assessment suggests that the evaluated populations are likely to decline, even in the absence of harvesting. Conclusions drawn from the potential impact of harvesting on l should be treated with caution since direct knowledge of population dynamics under natural conditions cannot be directly replaced by numerical simulations. Nevertheless, various studies have reported results that support our own: Hietz (1997), Winkler et al. (2007), and Haeckel (2009) all studied different epiphytic bromeliad species in the same region and, with the exception of T. juncea, they all show l values of below unity. Thus, it would seem realistic to suggest that, under current conditions, even very low levels of harvesting would be unsustainable. A high proportion of the TMCF in the region has been transformed to other land uses (40% in the last two decades; WilliamsLinera et al., 2002) and the remnant fragments have been subjected to selective logging for more than 80 years (Fuentes-Pangtay, 2013). The latter is targeted towards the largest Quercus spp. trees,
z et al., which are used to produce charcoal and fire wood (Gere 2012). Clearly, the elimination of these important support trees and the illegal stripping of epiphytes in the region contribute to the decline in epiphyte populations. In addition, given their strong dependence on microclimate, global climate change threatens the persistence of epiphytes, particularly in cloud forest ecosystems (Benzing, 1998; Nadkarni and Solano, 2002). However, forest loss is the most severe and immediate threat to their persistence. Given the level of threat to this terrestrial ecosystem worldwide, and the fact that many human communities depend on the valuable resources offered by these ecosystems, appropriate management plans and sustainable extraction of non-timber forest products has been suggested as an alternative to encourage landowners to preserve their forest areas (Peters, 1994; Kaushal and Melkani, 2005). Active management strategies that promote the recovery of epiphyte populations are required in the area, as it appears that they may be declining even in relatively well-conserved TMCF fragments. Even though seed production and early establishment make a relatively small contribution to the population growth rate of the studied populations, and seed germination and transplant survival rates have been reported to be low in other epiphytic bromeliads (Toledo-Aceves and Wolf, 2008; Winkler et al., 2007; ToledoAceves et al., 2012b), our numerical simulations suggest that the reintroduction of seeds and seedlings would have a positive impact on l. Other conservation strategies that could be suggested for these species should consider increasing the persistence of adult individuals, since the combined analysis of elasticities and LTRE showed that this process merits further attention. Plant reintroduction may be impractical due to the challenges involved in working in the canopy and, as epiphytic bromeliads tend to be slow-growing (Hietz et al., 2002; Schmidt and Zotz, 2002;
n et al., in press.), it would be expensive and time Mondrago consuming to cultivate plants from seed. Thus, in the short term, an alternative strategy that may produce some income while facilitating epiphyte conservation is the collection and utilization of
n-Chaparro and Ticktin, 2011; naturally fallen plants (Mondrago Toledo-Aceves et al., 2013). One of the main causes of natural death in epiphytes is detachment from the phorophyte (Hietz,

60

T. Toledo-Aceves et al. / Acta Oecologica 59 (2014) 52e61


n-Chaparro and Ticktin, 2011); 1997; Zotz et al., 2005; Mondrago thus, collecting fallen plants would not imply an additional mortality to the population. The recovery of fallen plants may become an effective conservation strategy, by supplying the existing market for epiphytic bromeliads and thus reducing the pressure of illegal collectors. The collection and marketing of fallen epiphytic bromeliads could serve as an example of management diversification for the TMCF by increasing the benefits derived from non-timber forest products. 5. Conclusions Matrix analysis revealed that T. multicaulis and T. punctulata populations are likely to be in decline at the study site and therefore the natural populations of these plants should not be subjected to harvesting. The l value of both populations is influenced mostly by survival (stasis), to a lesser extent by growth, and only slightly by fecundity. The combination of elasticity analyses and LTRE showed that, for both species, the persistence of individuals in the adult categories (principally larger adults) is the most important demographic process for population maintenance, which supports our suggestion regarding the importance of preventing harvest. One of the main causes of natural death in epiphytes is detachment from phorophytes, yet these detached plants actually represent a potential source of epiphyte supply for commercial use that would have no impact on population viability. Acknowledgements This project was funded by CONABIO (project HQ001), and by via the Kleinhans fellowship to T. Toledo-Aceves. INECOL A.C. provided facilities to carry out the fieldwork. We thank J. García-Franco and SENDAS A.C. for their support throughout the project. We thank H.
ndez, M.L. Leo
n and S. Landero for help with data collection Herna
ndez and R. Acosta for permission to work on their land. and R. Herna ~ iga-Vega for his assistance with data analysis. We also thank J. J. Zún We thank K. MacMillan and two anonymous reviewers for their helpful comments on a previous version of the manuscript. Appendix A Categories defined to analyse the population dynamics of Tillandsia multicaulis and Tillandsia punctulata, according to individual size (cumulative height, see text). The average number of rosettes per category (±s.e.) was obtained from data measured over two years. Categories are: S ¼ seedling, I ¼ infant, J ¼ juvenile, A1 ¼ adult 1, and A2 ¼ adult 2.

Species

Category

Cumulative height (cm)

Average number of individuals per category

Average number of rosettes per genet

T. multicaulis

S I J A1 A2

<1 2e10 11e30 31e50 >50

165 29.5 30 16 20.5

1 1.77 1.48 2.54 3.94

± ± ± ± ±

0 0.16 0.14 0.33 0.44

T. punctulata

S I J A1 A2

<1 2e11 12e20 21e50 >50

382 20 11.5 10 16

1 1.03 1.10 2.78 9.13

± ± ± ± ±

0 0.02 0.07 0.34 0.88

References Benzing, D.H., 1998. Vulnerabilities of tropical forests to climate change: the significance of resident epiphytes. Clim. Chang. 39, 519e540. Benzing, D.H., 1990. Vascular Epiphytes: General Biology and Related Biota. Cambridge University Press, Cambridge. Caswell, H., 1989. Analysis of life table response experiments I. Decomposition of effects on population growth rate. Ecol. Model 46, 221e237. Caswell, H., 2001. Matrix Population Models: Construction, Analysis and Interpretation. Sinauer Associates, Sunderland, Massachusetts. Clark-Tapia, R., Mandujano, M.C., Valverde, T., Mendoza, A., Molina-Freaner, F., 2005. How important is clonal recruitment for population maintenance in rare plant species?: the case of the narrow endemic cactus, Stenocereus eruca, in Baja California, Mexico. Biol. Conserv. 124, 123e132. Contreras, C., Valverde, T., 2002. Evaluation of the conservation status of a rare cactus (Mammilaria crucigera) through the analysis of its population dynamics. J. Arid. Environ. 51, 89e102. de Kroon, H., Plaisier, A., van Groenendael, J., Caswell, H., 1986. Elasticity: the relative contribution of demographic parameters to population growth rate. Ecology 67, 1427e1431.
n, J., 2000. Elasticities: a review of methods deKroon, H., van Groenendael, J., Ehrle and model limitations. Ecology 81, 617e618. Efron, B., 1982. The jackknife, the bootstrap and other resampling plans. Society for Industrial and Applied Mathematics, Philadelphia, Pennsylvania. Enright, N.J., Ogden, J., 1979. Applications of transition matrix models in forest dynamics: Araucaria in New Guinea, and Nothofagus in New Zaeland. Aust. J. Ecol. 4, 3e23. Esparza-Olguín, L., Valverde, T., Mandujano, M., 2005. Comparative demographic analysis of three Neobuxbaumia species (Cactaceae) with differing degree of rarity. Pop. Ecol. 47, 229e245.
pez-Ferrari, A.R., Ramírez-Morillo, I., 2005. Bromeliaceae. Espejo-Serna, A., Lo Fascicle 136. In: Sosa, V. (Ed.), Flora of Veracruz. Instituto de Ecología/University of California, Mexico/Riverside, CA, p. 307. Flores-Palacios, A., Valencia-Díaz, S., 2007. Local illegal trade reveals unknown diversity and involves a high species richness of wild vascular epiphytes. Biol. Conserv. 136, 372e387.
filo de Fuentes-Pangtay, T., 2013. Usos tradicionales de la madera del bosque meso ~ a en la subcuenca del río Pixquiac. MSc Thesis. Universidad Veracrumontan zana, Mexico. García-Franco, J., Rico-Grey, V., 1988. Experiments on seed dispersal and deposition patterns of epiphytes: the case of Tillandsia depeanna Steudel (Bromeliaceae). Phytologia 65, 73e78.
z, P., Fuentes, T., Vidriales-Chan, G., Toledo-Aceves, T., Pe
rez, K., 2012. CarGere
tica de la subcuenca. In: Pare, L., Gerez, P. (Eds.), acterísticas sociales y problema
n de la subcuenca del Pixquiac, Veracruz. INE, Juan Al Filo del Agua: cogestio Pablos Ed., Mexico, pp. 135e189. Haeckel, I.B., 2008. The “Arco Floral”: ethnobotany of tillandsia and Dasylirion spp. in a mexican religious Adornment. Econ. Bot. 62, 90e95. Haeckel, I.B., 2009. Ethnobotany, Ecology, and Harvest Impacts of Tillandsia imperialis (Bromeliaceae) in Veracruz, Mexico. M.Sc thesis. University of Texas. Heppell, S., Caswell, H., Crowder, L.B., 2000. Life histories and elasticity patterns: perturbation analysis or species with minimal demographic data. Ecology 8, 654e665.
ndez-Apolinar, M., Valverde, T., Purata, S., 2006. Demography of Bursera Herna glabrifolia, a tropical tree used for folk wood crafting in Southern Mexico: an evaluation of its management plan. For. Ecol. Man. 223, 139e151. Hietz, P., 1997. Population dynamics of epiphytes in a Mexican humid montane forest. J. Ecol. 85, 767e775. Hietz, P., Ausserer, J., Schindler, G., 2002. Growth, maturation and survival of epiphytic bromeliads in a Mexican humid montane forest. J. Trop. Ecol. 18, 177e191. Holbrook, N.M., 1991. Small plants in high places: the conservation and biology of epiphytes. Tree 6, 314e315. Horvitz, C.C., Schemske, D.W., Caswell, H., 1997. The “importance” of life history stages to population growth: prospective and retrospective analyses. In: Tuljapurkar, S., Caswell, H. (Eds.), Structured Population Models in Marine, Terrestrial, and Freshwater Systems. Chapman and Hall, New York, pp. 247e272. Kaushal, K.K., Melkani, V.K., 2005. India: achieving the millennium development goals through non-timber forest products. Int. For. Rev. 7, 128e134. Levin, L., Caswell, H., Bridges, T., DiBacco, C., Cabrera, D., Plaia, G., 1996. Demographic responses of estuarine polychaetes to pollutants: life table response experiments. Ecol. Appl. 6, 1295e1313. ~ a, C., Franco, M., Golubov, J., Flores, A., 2001. Integration of Mandujano, M.C., Montan demographic annual variability in clonal desert cactus. Ecology 82, 344e359. Mandujano, M.C., Gobulov, J., Huenneke, L.F., 2007. Effect of reproductive modes and environmental heterogeneity in the population dynamics of a geographically widespread clonal desert cactus. Pop. Ecol. 49, 141e153.
, A., Martorell, C., Martínez-Ramos, M., Caballero, J., 2005. Martínez-Balleste Applying retrospective demographic models to assess sustainable use: the Maya management of Xa'an palms. Ecol. Soc. 10, 17. Martínez-García, E., 2006. Din
amica poblacional de Tillandsia makoyana Baker (Bromeliaceae) en la selva baja caducifolia de la reserva de la biosfera Sierra de Huatla, Morelos (Population dynamics of Tillandsia makoyana Baker

T. Toledo-Aceves et al. / Acta Oecologica 59 (2014) 52e61 (Bromeliaceae) in the tropical drydeciduous forests of the Sierra de Huatla
noma de Biosphere Reserve, Morelos). BS thesis. Universidad Nacional Auto
xico, Mexico. Me Menges, E.S., 1998. Evaluating extinction risks in plants. In: Fielder, P.L., Kareiva, P.M. (Eds.), Conservation Biology The Coming Decade. Chapman and Hall, New York, pp. 49e65.
n, D., Dura
n, R., Ramírez, I., Valverde, T., 2004. Temporal variation in the Mondrago demography of the clonal epiphyte Tillandsia brachycaulos (Bromeliaceae) in the Yucat
an Peninsula, Mexico. J. Trop. Ecol. 20, 189e200.
n-Chaparro, D., Ticktin, T., 2011. Demographic effects of harvesting Mondrago epiphytic bromeliads and an alternative approach to collection. Conserv. Biol. 25, 797e807.
n, D., Valverde, T., Hern
Mondrago andez-Apolinar, M., Population ecology of epiphytic angiosperms: a review. Trop. Ecol. in press. Nadkarni, N.M., Solano, R., 2002. Potential effects of climate change on canopy communities in a tropical cloud forest: an experimental approach. Oecologia 131, 580e586.
Olmsted, I., Alvarez-Buylla, E., 1995. Sustainable harvesting of tropical trees: demography and matrix models of two palm species in Mexico. Ecol. Appl. 5, 484e500. Peters, C.M., 1994. Sustainable Harvest of Non-timber Plant Resources in Tropical Moist Forest: An Ecological Primer. Biodiversity Support Program, Washington, DC. Scatena, F.N., Bruijnzeel, L.A., Bubb, P., Das, S., 2010. Setting the stage. In: Bruijnzeel, L.A., Scatena, F.N., Hamilton, L.S. (Eds.), Tropical Montane Cloud Forests Science for Conservation and Management. Cambridge University Press, UK, pp. 38e63. Schmidt, I., Mandle, L., Ticktin, T., Gaoue, O., 2011. What do matrix population models reveal about sustainability of harvesting non-timber forest products (NTFP)? J. Appl. Ecol. 48, 815e826. Schmidt, G., Zotz, G., 2002. Inherently slow growth in two Caribbean epiphytic species: a demographic approach. J. Veg. Sci. 13, 527e534. Silvertown, J.W., Franco, M., Pisanty, I., Mendoza, A., 1993. Comparative plant demography-relative importance of life cycle components to the finite rate of increase in woody and herbaceous perennials. J. Ecol. 81, 465e476. Sodhi, N.S., Koh, L.P., Peh, K.S.-H., Tan, H.T.W., Chazdon, R.L., Corlett, R.T., Lee, T.M., Colwell, R.K., Brook, B.W., Sekercioglu, C.H., Bradshaw, C.J.A., 2008. Correlates of extinction proneness in tropical angiosperms. Divers. Distrib. 14, 1e10. Ticktin, T., Nantel, P., Ramírez, F., Johns, T., 2002. Effects of variation on harvest limits for non-timber forest species in Mexico. Conserv. Biol. 16, 691e705. Toledo-Aceves, T., Wolf, J.H.D., 2008. Germination and establishment of Tillandsia eizii (Bromeliaceae) in the canopy of an oak forest in Chiapas, Mexico. Biotropica 40, 246e250. Toledo-Aceves, T., Meave, J.A., Gonz
alez-Espinoza, M., Ramírez-Marcial, N., 2011. Tropical montane cloud forests: current threats and opportunities for their conservation and sustainable management in Mexico. J. Environ. Manag. 92, 974e981. Toledo-Aceves, T., García-Franco, J., Hern
andez-Rojas, A., MacMillan, K., 2012a. Recolonization of vascular epiphytes in a shaded coffee agroecosystem. Appl. Veg. Sci. 15, 99e107.
n-Mateos, M.L., Toledo-Aceves, T., García-Franco, J., Landero-Lozada, S., Leo MacMillan, K., 2012b. Germination and seedling survivorship of epiphytic bromeliads in the cloud forest canopy. J. Trop. Ecol. 28, 423e426.

61


ndez-Rojas, A., Sosa, V., Toledo-Aceves, T., Mehltreter, K., García-Franco, J., Herna 2013. Benefits and costs of epiphyte management in shaded coffee plantations. Agr. Ecosyst. Environ. 181, 149e156. Toledo-Aceves, T., García-Franco, J., Williams-Linera, G., MacMillan, K., Gallardo
ndez, C., 2014. Significance of remnant cloud forest fragments as resHerna ervoirs of tree and epiphytic bromeliad diversity. Trop. Conserv. Sci. 7, 250e262. Turner, I.M., Tal, H.T.W., Ibrahim, A.B., Chew, P.T., Corlett, R.T., 1994. A study of plant species extinction in Singapore: lessons for the conservation of tropical biodiversity. Conserv. Biol. 8, 705e712. Valverde, T., 1995. Metapopulation Dynamics of Primula vulgaris. PhD thesis. The Open University, UK.
fica entre poblaciones locales Valverde, T., Bernal, R., 2010. Hay asincronía demogra de Tillandsia recurvata? Evidencia de su funcionamiento metapoblacional. Bol.
x. 86, 23e36. Soc. Bot. Me Valverde, T., Silvertown, J., 1998. Variation in the demography of a woodland understory herb (Primula vulgaris) along the forest regeneration cycle: projection matrix analysis. J. Ecol. 86, 545e562. Verhoeven, K.J.F., Beckers, G.J.L., 1999. Canopy farming: an innovative strategy for the sustainable use of rain forests. Selbyana 20, 191e193. Wester, S., Zotz, G., 2010. Growth and survival of Tillandsia flexuosa on electricity cables in Panama. J. Trop. Ecol. 26, 123e212. Williams-Linera, G., 1997. Phenology of deciduous and broadleaved-evergreen tree species in a Mexican tropical lower montane forest. Glob. Ecol. Biogeogr. Lett. 6, 115e127.
n delbosWilliams-Linera, G., Manson, R.H., Isunza-Vera, E., 2002. La fragmentacio
filo de Montana y patrones de uso del suelo en la regio
n oeste de que meso
xico. Madera Bosques 8, 73e89. Xalapa,Veracruz, Me Winkler, M., Hülber, K., Hietz, P., 2005. Effect of canopy position on germination and seedling survival of epiphytic bromeliads in a Mexican humid montane forest. Ann. Bot 95, 1039e1047. Winkler, M., Hülber, K., Hietz, P., 2007. Population dynamics of epiphytic bromeliads: life strategies and the role of host branches. Basic Appl. Ecol. 8, 183e196. Wolf, J.H.D., Konings, C.J.F., 2001. Toward the sustainable harvesting of epiphytic bromeliads: a pilot study from the highlands of Chiapas, Mexico. Biol. Conserv. 101, 23e31. Wolf, J.H.D., 2005. The response of epiphytes to anthropogenic disturbance of pine
xico. For. Ecol. Man. 212, 376e393. oak forests in the highlands of Chiapas, Me Young, A., Boyle, T., Brown, T., 1996. The population genetic consequences of habitat fragmentation for plants. Tree 11, 413e418. Zar, J.H., 1996. Bioestatistical Analysis. Prentice Hall, New Jersey, US. Zhu, H., Xu, Z.F., Wang, H., Li, B.G., 2004. Tropical rain forest fragmentation and its ecological and species diversity changes in southern Yunnan. Biodivers. Conserv. 13, 1355e1372. Zotz, G., 2005. Differences in vital demographic rates in three populations of the epiphytic bromeliad, Werauhia sanguinolenta. Acta Oecol. 28, 306e312. Zotz, G., Laube, S., Schmidt, G., 2005. Long-term population dynamics of the epiphytic bromeliad Werauhia sanguinolenta. Ecography 28, 806e814. Zotz, G., Vollrath, B., 2002. Substrate preferences of epiphytic bromeliads: an experimental approach. Acta Oecol. 23, 99e102. Zuidema, P., de Kroon, H., Marinus, J., Werger, A., 2007. Testing sustainability by prospective and retrospective demographic analyses: evaluation for palm leaf harvest. Ecol. Appl. 17, 118e128.