Secondary Succession under invasive species (Pteridium aquilinum) conditions in a seasonal dry tropical forest in southeastern Mexico

Study site

The study was carried out in the ejido (communally managed land) Laguna Om (−89.15W 18.70N, −88.87W 18.40N WGS 84) in the south of the state of Quintana Roo, Mexico with an area of 84,998 ha (Fig. 1). The ejido is located in the geological formation called Petén, which belongs to the Paleocene-Eocene and is characterized by massive, compact macro and microcrystalline limestones, which have yellow to white coloring, with brown parts stained by iron oxides (Wright, 1967; INEGI, 1993). The terrain is flat with an undulated microrelief, and with wide depressions that present small plains. Its altitude above sea level varies between 100 and 150 m (García, 1997; INEGI, 1993; IUSS Working Group WRB, 2015). The climate is Aw (x’) i, warm subhumid, with precipitation in summer and some in the winter (García, 1997). The climatological station located in the center-south part of the ejido reports 1,290 mm/year of precipitation and an average annual temperature of 26 °C. Soils are Leptosols, Vertisols and Gleysols (INEGI, 1993). The dominant vegetation in the study area is seasonal dry tropical forest, and the terrain is low seasonal flooded forest and savanna (Miranda & Hernández, 1963).

Sampling design

Sampling units were selected based on several conditions. Each of the four sampling units has eight plots (SU) of 10 ×10 m (100 m²), which were randomly selected. This gives a total of 32 plots. To evaluate the importance of the soil properties on the richness and abundance of species, soil samples were taken in all the sampling units. 12 soil samples per plot were mixed, giving one composite sample of 2 kg for each sampling unit. Soil analyses were based on the Official Mexican Standard (NOM) 021 RECNAT-2000. The soil samples were dried at room temperature (22–25 °C), macerated and sieved with a 10 mm mesh screen and the following characteristics were analyzed: Total nitrogen (Nt, Micro Kjeldahl method), Phosphorus (Olsen method), Potassium (K, AS-12 with ammonium acetate), pH (Walkley and Black method), texture (Bouyoucos, method), Ca, Na, and soil organic matter (SOM).

Plots characteristics

Three sampling units were established in areas invaded by P. aquilinum for approximately 50 years (from 1960 till present). The age and the frequency of fires of the sample units were defined according to the information provided by local farmers. Sampling unit 1 (SU1) consisted of areas with regular periods of burning every two or three years during the last fifty years. The last recorded fire occurred in 2015; vegetation sampling was carried out one year after the last fire (2016). Sampling unit 2 (SU2) was established in areas invaded by ferns during the last 50 years, but the last burning occurred 12 years ago (2007). In SU1 and SU2, fires were natural and the presence of abundant biomass of P. aquilinum and together with higher temperatures during the dry season increased fire probability. Contrary, sampling unit 3 (SU3) was established in areas with a similar land use history, but since 2010 fire was controlled and the removal method applied by Macario (2010) was applied. This method consists of weekly cuttings (locally known as chapeos) during two months, followed by monthly cuttings throughout the year. Later quarterly and semi-annual cuttings are applied depending on the density of the fern. A fourth sampling unit (SU4) was established in areas that were burned for the last time in 2007 for the establishment of the milpa system (slash-and-burn agriculture), and was left for recovery since 2009. This area never presented invasion by ferns. All the sampling units presented similar conditions in relation to climate, stoniness, slope and altitude.

Sampling

The sampling was carried out between May and June 2016. In each of the plots the following variables were recorded: diameter at breast height (DBH) of all tree species >0.5 cm and the taxonomic identity of each species (Wyatt-Smith & Foenander, 1962). All registered individuals were identified from dichotomous botanical keys (Standley & Steyermark, 1958) and existing floristic listings (Sousa & Cabrera, 1983; Pennington & Sarukhán, 2005). The community of plant species was characterized in each of the four SUs establishing richness and abundance. The Importance Value Index (IVI) was determined in each of the treatments. ${\displaystyle IVI=\text{relative density}+\text{relative basal area}+\text{relative frequency}}$

Where: relative density = (number of individuals of species/total number of individuals)*100; relative basal area = (basal area of a species/basal area of all species)*100; relative frequency = (frequency of a species/ frequency of all species)*100 (Kent, 2011).

Data analysis

To compare the taxonomic diversity of the treatments, a rarefaction analysis (interpolation) and extrapolation (prediction) of the Hill numbers were performed, based on sample size and coverage, which represents a unified criterion to contrast the diversity of multiple assemblages (Hsieh, Ma & Chao, 2016). The analysis was carried out based on the order q (richness of species), and richness estimators were determined with the iNEXT software package R (Chambers, 2008; Hsieh, Ma & Chao, 2016). The variation in the composition of species was analyzed through a non-metric multidimensional scaling (NMDS). To reduce the effect of very abundant species, and to level the effect of rare species, the fourth root transformation in the PRIMER-E 6.1.12 software was applied. Statistically significant differences in the composition of tree species between the different plots were calculated with a dissimilarity analysis (ANOSIM) in PRIMER-E 6.1.12 (Clarke & Gorley, 2006). Finally, a generalized linear model (GLM) was carried out between the species richness and abundance and the soil variables with the package glm2 (Marschner, 2011) in the R software (R Development Core Team, 2012). Prior to the analysis, multicollinearity of variables was established to avoid redundancy (Zuur, Hilbe & Ieno, 2013).

Composition and floristic richness

There were 2,162 individuals belonging to 33 families, 56 genera and 63 species, recorded in the 32 sampled plots (SU1 to SU4). Of the total recorded individuals, 1884 (87%) were trees and 278 (13%) shrubs. The sampling unit with the highest diversity was SU4 with 32 families, 49 genera, 54 species and 1091 individuals, of which 882 were trees and 209 shrubs. On the other hand, SU1 showed the lowest species richness with 11 species, belonging to nine families, and a total of 193 trees without the presence of shrubs (Table 1). The families with a higher richness of recorded species were Fabaceae with 11 species, followed by Asteraceae, Meliaceae, Rubiaceae, Sapindaceae and Verbenaceae with three species each, contributing 41% of the total of the species registered in the plots. The genera that presented the highest number of species were Lonchocarpus with five, and Thevetia, Eugenia and Coccoloba with two species respectively. The largest number of individuals was distributed among the following families: Rubiaceae with 572, Araliaceae (237), Fabaceae (233), Polygonaceae (159), Cecropiaceae (134), Asteraceae (115), Ulmaceae (110) and Sapindaceae with 104. Species with higher densities were Guettarda combsii Urb. (522), Dendropanax arboreus (L.) Decne. & Planch. (237), Coccoloba spicata Lundell (158), Cecropia peltata L. (134) and Trema micrantha (L.) Blume (108), and together formed 54% of the inventoried species.

NMDS (Fig. 2) showed clear differences between sampling units. The stress obtained in the test (0.18%) is probably influenced by the variability in the composition of species of SU4. The dissimilarity of species is a good measure to identify the distribution of the ecological niche between the species. The dissimilarity analysis (ANOSIM) determined significant differences in the dissimilarity between the different plots, indicating that beta diversity varied between sampling units. In this case, the variation percentage ranged from 60% to 91% between the plots. Our results showed that the plots SU1 and SU3 are the plots with a higher degree of differentiation in the composition of species, and SU1 and SU2 are those with a lower percentage of differentiation (Table 2).

Statistically significant differences were observed for species richness among the evaluated sampling units. It was observed that species richness decreased from SU4 to SU1. Likewise, abundance showed the same pattern, except for SU2 which presented the lowest abundance. It was also observed that the sampling effort performed is representative for the richness and composition of the biota of the region (between 98 and 99% for SU1, SU3 and SU4 and 88% for SU2; Table 3).

Relative Importance value index (IVI)

The IVI was obtained in each of the four SU and IVI data for each of the species are presented (Fig. 3). In SU1, the species with the highest IVI are Coccoloba spicata, Lonchocarpus rugosus Benth., Croton sp1, Hippocratea excelsa Kunth. and Diospyros verae-crucis Standley. Regarding SU2, the species that characterize this community considering the highest IVI were: Lysiloma latisiliquum (L.) Benth., Bursera simaruba (L) Sarg., Malvaviscus arboreus Cav., Trema micrantha, Metopium brownei (Jacq.) Urb., Simarouba glauca DC and Guettarda combsii. The characteristic species of SU3 based on IVI were Guettarda combsii, Hampea trilobata Stand., Zuelania guidonia (Sw.) Britton & Millsp., Coccoloba spicata, Lonchocarpus rugosus Benth., Lysiloma latisiliquum, Bursera simaruba, Swietenia macrophylla King, Trema micrantha and Croton sp2. Finally, the species with greater relative importance in SU4 were Cecropia peltata, Dendropanax arboreus, Trema micrantha, Bursera simaruba, Eupatorium albicaule Sch. Bip. ex Klatt, Piscidia piscipula (L.) Sarg., Hamelia patens Jacq., Allophylus cominia (L.) Sw., Lonchocarpus guatemalensis Benth., Nectandra salicifolia Kunth., Coccoloba spicata Lundell, Aegiphila monstrosa Moldenke, Bourreria oxyphylla Stand., Verbesina gigantea Jacq. and Guettarda combsii.

Diversity and abundance

The species richness and conditions of the sampling units considering the disturbance frequency resulted in significant differences. SU3 showed almost twice the number of species found in SU1 and SU2. As expected, SU4 was the site with the highest species richness (Table 3). Table 4 summarizes soil characteristics. Soil organic matter and N were excluded from the analysis, as those values indicated collinearity >10. GLM showed a strong correlation between the species richness of tree species. Variables, which explained best species richness, were site conditions, followed by Ca and soil organic matter. Also, GLM showed that abundance resulted in a strong correlation with site conditions and soil characteristics. Species abundance was best explained by site conditions, P, Ca and organic C (Table 5).