Fire response of the endangered Pinus culminicola stands after 18 years in Cerro El Potosí, northeast Mexico

Aim of the study: To analyze the impact of the 1998 wildfire on dwarf pine (Pinus culminicola) population stands 18 years after the fire. Area of study: Cerro el Potosí, Nuevo León (México). Material and methods: We measured regeneration, biovolume of individuals, diversity and species composition in unburned and burned plots randomly selected in the area. Main results: Recovery was very slow, with limited dense natural regeneration and very slow growth. Significant differences between control and burned plots were found with respect to regeneration, species composition, evenness and soil nutrients, suggesting that fire could promote natural regeneration and biodiversity in these communities. Research highlights: Management efforts are necessary to maintain a natural fire regime and ensure grazing exclusion in order to favor the conservation of the community dominated by this rare species P. culminicola. Additional keywords: fire ecology; DCA; Permanova; richness. Abbreviations used: CIC (Cationic Interchange Complex); DCA (Detrended Correspondence Analysis); EC (Exchangeable Cation); OM (Organic Matter). Authors ́ contributions: Conceived, designed and performed the experiments: JRA, EE and HGR. Analyzed the data: JRA, JAV, JRE and YM. Contributed reagents/materials/analysis tools: JAV, JRE, YM, IC and JU. Wrote the paper: JRA, JAE and EE. All authors read and approved the final manuscript. Citation: Arévalo, J. R.; Estrada, E.; Encina, J. A.; Villarreal, J. A.; Escobedo, J. R.; Morales, Y.; Cantú, I.; González-Rodríguez, H.; Uvalle, J. (2017). Fire response of the endangered Pinus culminicola stands after 18 years in Cerro El Potosí, northeast Mexico. Forest Systems, Volume 26, Issue 3, e015. https://doi.org/10.5424/fs/2017263-11566 Received: 20 Apr 2017. Accepted: 21 Nov 2017. Copyright © 2017 INIA. This is an open access article distributed under the terms of the Creative Commons Attribution (CC-by) Spain 3.0 License. Funding: Program of cooperation Universidad de La Laguna-Universidad Autónoma de Nuevo León, included in the official frame agreement of these institutions; Facultad de Ciencias Forestales of the Universidad Autónoma de Nuevo León; University of La Laguna (Research Program) covered the travel expenses of JRA. Competing interests: The authors have declared that no competing interests exist. Correspondence should be addressed to José R. Arévalo: jarevalo@ull.edu.es Forest Systems 26 (3), e015, 9 pages (2017) eISSN: 2171-9845 https://doi.org/10.5424/fs/2017263-11566 Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria O. A., M. P. (INIA)


Introduction
Habitat loss and invasive species are the main threats to the survival of most endangered species (D 'Antonio & Kark, 2002), especially for species with small populations and slow regeneration.This is common at high altitudes due to the limited productivity of these sites and the short growing season (Mark & Baylis, 1963;Whittaker & Fernández-Palacios, 2007).If these species have also been reduced to just small remnant populations, natural processes such as fire can mean a dangerous threat of extinction.In addition, there are many other elements of change that can endanger these populations (Walker & Steffen, 1997) in the long term, such as global warming, which can be intensified in these small and environmentally limited populations (Farjon, 2013).
Many high elevation forest ecosystems are adapted to stand-replacing fires (Baker, 2006), and fire can be a key disturbance process influencing tree survival and regeneration (DeBano et al., 1998).The recovery of a stand will depend on seed production and dispersion (Donato et al., 2009), and climate conditions will govern the success of seedling and sapling establishment (Arévalo & Fernández-Palacios, 2008;Allen et al., 2012), with post-fire conditions playing an important role.
Dwarf pine (Pinus culminicola) was described by Andresen & Beaman (1961) as a new species from the highest ridges of Cerro El Potosí, Mexico.In 1966, the species occupied a total area of 106 ha (Beaman & Andresen, 1966), and a few years later, it was found in Sierra La Marta and Sierra La Viga in the state of Coahuila, and Cerro Peña Nevada in Nuevo León, with spatial patterns and altitudinal distribution ranges similar to those determined for Cerro El Potosí (Riskind & Patterson, 1975).Fire is a key component of the dynamics of these communities, and in 1978, wildfires on Cerro El Potosí burned 34% of the area occupied by dwarf pine (García, 1989).Other anthropogenic impacts have also reduced the area formerly covered by dwarf pine, including road building, cattle grazing, mining and land use changes for the installation of telecommunication and aerial navigation infrastructure (Jiménez et al., 2005).
In 1998, several wildfires drastically reduced the area covered by dwarf pine on Cerro El Potosí.Currently, only 30 ha of severely fragmented dwarf pine area exist in Cerro El Potosí, including adults with low seed production that are subject to the impact of grazing and trampling by cattle and horses (Jiménez et al., 1999).There are no studies of P. culminicola populations on other mountains where it occurs, and unfortunately most of these areas are not legally protected and are also subjected to grazing and human-induced fires.
This is a species with very small distribution range, highly sensitive to fire, whose population has been reduced and is endangered.Furthermore, it grows in an area where environmental conditions, including drought and very low temperatures, lead to a short growing season.Therefore, changes in the natural disturbances patterns could result in the extinction of the species.
The main objective of this study was to study the regeneration, almost 20 years after the last fire affected this small stand.We tested the hypothesis that P. culminicola follows a pattern of stand replacement after fire and that its regeneration requires fire in order to renew individuals in the area.

Study area
The study area is located at the top of the Cerro El Potosí.This is a legally protected natural area, established by the government of the state of Nuevo León under the category of ecological conservation (PO, 2000).This mountain is located on ejidal and private lands in the south of the State of Nuevo León, Mexico (lat.24°52´N, long.100°13´W) within the Sierra Madre Oriental range, between the Gulf Coastal Plain and the Mexican High Plateau (Fig. 1) and dates in some parts from the inferior cretaceous with a marine sedimentary origin (García & González, 1991).Soils are classified as Orthens and also some Rendolls with poor discrimination of horizons, as well as Litosoils (González, 1999).The maximum elevation at Cerro El Potosí reaches 3,721 m.
The climatic data indicate that it is an E(T) H(e) type climate following the Köppen classification modified by García (1988).Mean annual precipitation is low, less than 600 mm, falling mainly in the month of September and supplemented by fog condensation.Average annual temperature is 12 o C, with an average minimum in January of -1.5 o C. The mountain top is exposed to high winds for most of the year.Dense fogs often occur throughout the year.Snow may fall several times during winter, persisting for a few days on northern and northeastern slopes (García, 1996).
In this area, the population of P. culminicola occurs mainly over 3,600 m a.s.l. with isolated individuals below that altitude.Density is highly variable ranging from 200 to 25 ind/ha.Above 3,600 m, it is the dominant shrub, forming shrublands associated with Juniperus zanonii interspersed with occasional gnarled and stunted Pinus hartwegii (Estrada et al., 2014).This community was described by Rzedowski (2006) as Pinus shrubland and occupies a narrow altitudinal belt, growing between the Pinus hartwegii forest and the alpine meadow.

The species
Pinus culminicola Andresen & Beaman is a dwarf shrub 0.5-2.5 m tall and with more than 1.0 m of canopy cover.It is multi-stemmed with low, dense and rounded crowns.The bark is grey-brown, thin and scaly at the base of the trunk.The leaves ('needles') are in fascicles of five, slender, 3-5.5 cm long, and deep green to blue-green, with stomata confined to a bright white band on the inner surfaces.The cones are globose, 3-4 cm long and broad when closed, green at first, ripening yellow-brown when 16-18 months old, with only a small number of thin, fragile scales, typically 6-14 fertile scales.The cones open to a breadth of 4-6 cm when mature (nonserotinous), approximately in October, holding the seeds on the scales after opening.The seeds are 9-12 mm long, with a thick shell, a white endosperm, and a vestigial 1-2 mm wing (Estrada et al., 2014).They are dispersed by the Clark's nutcracker, which pluck the seeds out of the open cones.These jays, which use the seeds as a major food resource, store many of the seeds for later use; however, some of these stored seeds are not used and are able to grow into new plants.The species is endemic to a few mountain tops in the states of Coahuila and Nuevo León in northeast Mexico (Villaseñor, 2016).The conservation status of the species according to SEMARNAT ( 2010) is in danger of extinction.Following the IUCN Red List of Threatened Species, this species is assessed as endangered under criterion B1+2 (restricted distribution to few populations), and grazing and fire are threatening these populations (Farjon, 2013).P. culminicola has a very restricted range, and while not subject to intensive cutting for fuel wood, it is, during long dry periods, very susceptible to fire (Perry, 1991).

Sampling design
During October-November 2015, we sampled 40 plots in the Cerro El Potosí, in the area with the densest existing population, affected by fire in 1998 and 1978, both of which were considered intense and severe fires during summer periods (García & Gonzalez, 1991).In the burned area (hereafter burned), we randomly located 21 plots, while in the unburned area (hereafter unburned) we located 19 plots (Fig. 1).Each plot was 15 × 15 m (225 m 2 ).
In each plot, we collected one composite soil sample of about 1 kg from 0-30 cm depth.Each sample was mixed, dried, and sieved through a 2 mm sieve, and debris and stones were removed from the soil.We analyzed pH, EC (electrical conductivity of extract, dS/m), Olsen P (ppm), N (ppm), % organic matter (OM), available cations in ppm (Na, K, Ca, Mg) and CIC (Cationic Interchange Complex, meq/100 g).Percentages of clay, sand and silt were also analyzed.We followed standard methods of analyses (AOAC, 1990;MAPA, 1986).Altitude, aspect and slope were measured in each plot.
The available information on fires indicated that in 1978, 30% of the pine population was burned, while in 1998, after several fire events, the population was reduced to 30 ha of fragmented patches.The fire intensity was high enough to eliminate all the individuals' affected (Jiménez et al., 2005).In the burned plots, all regeneration woody plants (trees and shrubs) after 1998 were recorded together with the height and two crown diameters (the larger one and its perpendicular), which allowed for the calculation of the total biovolume of the species (Arévalo et al., 2015).Dead trees were also counted and measured.In the unburned plots, due to the dense cover of the pines, only the total cover of pines and other woody species on each plot were recorded, and regeneration was noted (individuals under 130 cm tall).
For species names, we followed the checklist of flora for the summit of Cerro El Potosí (García & González, 1991).

Statistical analyses
We performed a one-way distance-based permutational t-test (Anderson et al., 2008) for comparison of species richness and Smith & Wilson evenness values of unburned plots vs. burned plots.The analyses were based on Brian-Curtis similarity of the raw data, with p-values (p < 0.05) obtained with 9999 permutations and a Monte Carlo correction when necessary.Primer 6 and Permanova+ (PRIMER-E Ltd, Plymouth, UK) were used to perform all PERMANOVA statistical procedures.
We used DCA (Detrended Correspondence Analysis; Hill & Gauch, 1980) to analyse the species composition (based on species cover) of burned vs. unburned plots.The coordinates on the first DCA axis for each plot were correlated with the environmental and soil nutrient variables using the Pearson correlation coefficient and tested for significance (p< 0.05) and the multiple test procedure of Holm for independent tests (Legendre & Legendre, 1998) was also applied.
For the significantly correlated variables with the coordinates of the plots, we performed a one-way distance-based permutational t-test as before.To determine if DCA axes discriminated the species composition based on burned vs. unburned, the coordinates of transects from the DCA for axes I and II were analyzed using logistic regression and the Wald statistic to determine its significance (for p<0.05).We performed all multivariate analysis with the CANOCO package (ter Braak & Ŝmilauer, 1998).The statistical methods used followed Zar (1984) and were implemented using Primer 6 and Permanova+ (PRIMER-E Ltd, Plymouth, UK).

Results
Significant differences were found between unburned and burned plots, for almost all the variables included in this study.A total richness of 22 species was found for the study area, herbs were the most common (Table 1).We did not find pine regeneration in unburned plots, while in burned plots pine regeneration was present in all the plots.The density ranged between 0 and 46 individuals.Average number of pines for plots with regeneration was 11.1 (±12.8)individuals and biovolume 4.7 (±6.6) m 3 .More than 40% of the individuals presented mature pine cones.In addition to P. culminicola, we also found another woody species, Juniperus zanonii, regenerating in 5 of the burned plots and in one unburned plot.Density of this species ranged between 13 and 3 in burned plots and one individual in an unburned plot.Biovolume ranged between 0.03 and 0.01 m 3 in burned plots and 0.08 m 3 in unburned ones.No evidence of dead individuals after 1998 was found in burned or unburned plots.Richness did not differ significantly between unburned vs. burned plots, however, Smith & Wilson evenness index was significantly higher (pseudo F 38 =2.95, p<0.05) in burned plots (Table 2).
The environmental variables and soil nutrient variables average per plot are indicated in Table 3.Only 5 of these variables are correlated with the DCA axis I site scores: Litter, rock, pH, Ca and Mg (Pearson correlation coefficient -0.81, 0.49, 0.64, 0.45 and -0.53; all of them significant for a p<0.01 after Holm correction).We found all of these variables were significantly different when comparing burned vs. unburned plots.
The DCA analysis (eigenvalues of axes I and II were 3.41 and 3.29, respectively, and the cumulative percentage of variance explained by both axes was 33.3 %) discriminated significantly unburned and burned plots along axis I using the logistic regression analysis of the plot scores taking as dependent variable burned vs. unburned Wald = 19.119,p<0.01).Based on cover, P. culminicola, Salvia microphylla and Tauschia madrensis were the most representative of the species composition in unburned plots (solid line polygon, Fig. 2), while the shrub Ribes ciliatum and forbs Hymenoxys insignis or Villadia aristata were more representative for the species composition of burned plots (dashed line polygon, Fig. 2).The rest of the species have similar importance in plots, such as the shrub Juniperus zanonii or well-distributed species like Achillea millefolium, Penstemon leonensis or Senecio loratifolius (Fig. 2).

Discussion
Studies into forest fire effects have mainly concentrated on the analysis of the canopy structure, but there have been fewer on the effects on the understory Table 1.List of the species recorded on the plots and Raunkaier classification.

Unburned Burned
Regeneration density/plot P. culminicola 0 11.1 (12.8) species composition (Arévalo et al., 2001).In our study, multivariate analysis revealed some statistically significant patterns of species composition, and the bivariate analyses revealed low differences in species richness.Also we found differences in the structure of the pine stand and soil nutrient composition.
Although we did not find significant differences in species richness, evenness was significantly higher in the burned plots.A reduction of the dominance of P. culminicola explains these differences.It is more common to find an increase in species richness together with evenness after natural fires (Shafi & Yarranton, 1973;Chaturvedi & Rafhubanshi, 2014;Markham & Essery, 2015).In any case, in spite of the higher evenness of the burned plots, the values were very low and far from maximum values, indicating the dominance of a few species such as P. culminicola, Senecio coahuilensis and T. madrensis in unburned plots and Ceanothus fendleri and Euphorbia beamanii in burned plots.
Regeneration of woody species has been considered to be susceptible to climate conditions and particularly to extreme events, responding faster than adults to environmental changes or extreme climatic conditions (Lloret et al., 2012).In our case, regeneration was present only in burned plots and not in unburned ones (only adult trees were present).Nevertheless, a low growth rate was apparent and recovery was still very low from the situation prior to the fire.Though there was a source of seeds close enough and almost all the burned plots were colonized by P. culminicola.However, in this environment, due to the short growing period and low temperatures, some constraints (Körner & Peláez, 1989;Häsler et al., 1999) are expected on root or shoot growth.This alpine ecosystem also requires higher rates of respiration of assimilated CO 2 , which can explain in part the low growth rate of species (Atkin et al., 1996) together with large pools of nonstructural carbon in field-growth alpine leaves (Mooney & Billings, 1961).As a result of the low recruitment and growth in alpine environments, P. culminicola, like other species, revealed a high survivorship rate (Forbis & Doak, 2004), with no dead individuals being found in the burned plots since the burning.
Species composition was significantly different between unburned vs. burned plots.Dominance of P. culminicola in burned plots did not reduce richness, but significantly reduced evenness (Fig. 2).Species as S. microphylla and T. madrensis were dominant in control plots, while the shrub R. ciliatum and the forbs H. insignis or V. aristata were dominant in burned plots.Evident environmental differences among plots explain these differences in species composition, such as the litter or rock cover.But nutrient level (higher Ca in the burned plots and Mg in unburned plots) is also related to these differences (Certini, 2005).So, more time is still required for the recovery of the area.Table 3. Environmental and nutrient content variables measured in unburned plots vs. burned plots (mean and SD).The variables significantly correlated with the axis I of DCA analysis are indicated in bold (for these variables, significant differences between unburned vs. burned are indicated as p<0.05 as "*" and p<0.01 as "**").Fire is an intrinsic characteristic of this and similar ecosystems, and fires may be needed to maintain such systems (Arno & Allison-Bunnell, 2002), though it is also true that some studies have found negatives effects of fire on alpine plant communities (Kirkpatrick & Dickinson, 1984).Previous studies (García & González, 1991) did not report seedlings or saplings in burned areas of Cerro El Potosí during the field trips 13 years after the 1978 fire.In the case of the P. culminicola stand there is no information available about fire rates, but based on these results it is expected to take over 50 years for the plant community to recover (as a conservative estimate).Fires appear to favor natural regeneration (which is absent in unburned plots, with only adult trees present), and their impact on recurrence in vegetation recovery is well known (Tessler et al., 2016), especially in these areas with such extreme environmental conditions.It is also clear that mountains areas are especially sensitive to human disturbances or changes in the natural disturbance rate (Guisan & Theurillat, 2000).Grazing is another factor that can limit recovery of plant cover in the area (Jiménez et al., 2005) and is determinant in some ecosystems of the plant community (Arévalo et al., 2007).Although we did not record of any grazing in the studied area, probably due to the steep slope of the burned and unburned area, there is some evidence around.In this case, grazing of cattle and horse should be prevented as it has been demonstrated that they are potential seedling grazers of this species (Jiménez et al., 2005).
In conclusion, continuous monitoring of the affected area will be required to supervise the level of recovery and prevent external disturbances (as commented) to favor the establishment of natural regeneration of P. culminicola.However, we do not share the negative view about fire, as we consider it an intrinsic element of plant community dynamics.Nevertheless, conservation efforts are urgent to maintain a natural fire regime (recurrence, intensity, severity) and ensure the exclusion of grazing in order to favor the natural succession of the stand dominated by this unique species.

Figure 1 .
Figure 1.Location and map of the studied area in Cerro El Potosí, Mexico.Burned and unburned plots are indicated.