Introduction

The soil seed bank plays an essential role in plant dynamics; it constitutes an important stage of regeneration, as it stores the plant species that will potentially become established in the soil (^{De Souza et al., 2006}). Seed banks are dynamic in space and time (^{Fenner and Thompson, 2005}); in space, fruits and seeds are dispersed when they fall to the ground or by several vectors such as wind, water or animals (^{Pijl, 1969}). Over time, seeds may remain dormant in the soil, that is, they delay germination due to physical or physiological characteristics in their structures, or simply because the right environmental conditions for germination are not present (^{Baskin and Baskin, 2004}). The seeds reach the soil through seed dispersal and rainfall, and they are eliminated from the soil through germination, predation, and death by pathogens (Fenner and Thompson, 2005).

Soil seed banks may contain species present in the vegetation or that were present before a disturbance and in nearby areas (^{Bossuyt and Honnay, 2008}), and therefore are an important resource for the restoration and resilience of a damaged plant community (^{Shiferaw et al., 2018}). Traditionally, in disturbed communities, degradation is assessed only by the status of the standing vegetation, often forgetting that the analysis of seeds stored in the soil provides crucial information for understanding plant dynamics, interpreting the conservation status, and determining the potential for plant recovery (^{Ma et al., 2021}).

Research on the temporal and spatial variations of the seed bank is approached by studying its basic characteristics such as density, diversity, or richness. An important aspect that influences the spatial distribution of seeds is their dispersal mechanism, since it is through this process that plant populations grow, exchange genes, and survive environmental changes (^{Levin and Muller-Landau, 2000}). For example, seeds dispersed by animals are more frequently deposited near trees with similar means of dispersal than those with different dispersal mechanisms (^{Saatkamp et al., 2013}). This is due to the fact that plants with fruits attractive to animals tend to be visited by these for feeding, protection, nesting, or perching (in the case of birds), and while performing these activities, they defecate and thereby deposit the seeds under the tree canopy (^{DeMars et al., 2010}).

Isolated trees that remain standing after a disturbance or are intentionally left in deforested areas are widely studied and utilized for their influence on the dynamics of regeneration and ecological restoration (^{Guevara et al., 2005}). In addition, if these species attract frugivorous animals, they tend to act as seed reservoirs, increasing the interactions with other plants and animals and thereby maintaining high biodiversity (^{Camargo et al., 2020}). This is likely to occur more frequently in forests and jungles because there are more zoochorous species there than in other ecosystems (^{Gentry, 1982}; ^{Willson et al., 1990}), however, in disturbed areas of arid zones, zoochorous dispersal and seed accumulation under isolated trees also occurs (^{Filazzola et al., 2019}), and its study depends on whether it is a useful tool for the recovery of vegetation and the interactions that are lost with the continuous change of land use.

The Tamaulipan Thorny Scrub is located in a semi-arid zone in northeastern Mexico. It is a very fragmented vegetation type, composed of a matrix of induced vegetation (for livestock or agriculture) with small remnant patches of native vegetation, as well as isolated trees of different species (^{Molina-Guerra et al., 2013}). In the thorny scrub, 37 species are recognized as having zoochorous dispersal, which represents a higher percentage of taxa with this type of dispersal compared to other semiarid zones (^{Jurado et al., 2001a}). However, the relationships between plants and dispersers, as well as the influence of the type of dispersal on regeneration, have been scarcely studied in this ecosystem.

Different methods have been used for seed bank research in the Tamaulipan Thorny Scrub (^{Pando-Moreno et al., 2010}; ^{Martínez-Adriano et al., 2021}; ^{Valdes-Alameda et al., 2021}); seed banks are known to occur within small undisturbed vegetation areas and at their edges, but there is no information on seed banks in cattle ranching sites, the most common land use in the region. On the other hand, due to the fragmentation of the scrubland, the role of isolated trees has been studied in the context of livestock production or agroforestry (^{Sarmiento-Muñoz et al., 2019}), while in the ecological context, seed removal and fruit production were studied in order to compare them with those produced in the remaining continuous vegetation (^{Jurado et al., 2006}; ^{Cuéllar-Rodríguez and Jurado, 2016}).

The characteristics of the seed bank were determined in terms of richness and diversity, under isolated individuals of tree species with different dispersal syndromes (classified by the characteristics of their fruits). The soil seed bank was compared under the native species Diospyros texana Scheele and Neltuma laevigata (Humb. & Bonpl. ex Willd.) Britton & Rose (previously Prosopis laevigata (Humb. & Bonpl. ex Willd.) M. C. Johnst.), Yucca filifera Chabaud (of zoochorous dispersion), Parkinsonia aculeata L., and Vachellia farnesiana (L.) Wight & Arn. (previously Acacia farnesiana (L.) Willd.) (without apparent dispersion syndrome). The temporal variation of the seed bank studied was also explored. The hypotheses were: (1) Isolated individuals of zoochorously dispersed species have a greater richness and density of seeds under their canopies than species with other types of dispersal, due to a greater seed deposition generated by the visit of dispersers that are attracted to the zoochorous fruits, and (2) The composition of the banks is variable between seasons because the plants in the community do not all produce seeds in the same season.

Materials and Methods

Results

A total of 57 species belonging to 29 families were identified in the seed bank (Table 1): 21 were arboreal or shrubby, and 36 were herbaceous (Figure 1). Of the total number of species, 21 were classified as having unassisted dispersal, 19 as zoochorous, 14 as anemochorous, three as having explosive dispersal or autochorous, and one species as hydrochorous or water-dispersed (^{Jurado et al., 2001a}). Most taxa (45) were considered native or probably native, and five were exotic; the most abundant taxa during the two years of study were Malvastrum coromandelianum (L.) Garcke (754 seedlings total) and Nama jamaicensis L. (616 seedlings total), both herbaceous. The most abundant tree species were Neltuma laevigata (54 seedlings total) and Yucca filifera (49 seedlings total), mostly under the parent plant, and the most abundant seedlings found under a different species were Opuntia Mill. sp. (44 seedlings) and Zanthoxylum fagara (L.) Sarg. (39 seedlings). The best represented families were Asteraceae (9) and Euphorbiaceae (7), followed by Fabaceae (6).

Figure 1 Habit of the species and their dispersal type. 

Based on the analysis of the soil seed bank under each isolated tree species, it was determined that seedling richness and density were similar between the two study years (Richness N. laevigata: U=186, Z=-0.38, P=0.71; V. farnesiana: U=194, Z=-0.79, P=0.88; D. texana: U=134, Z=-1.79, P=0.76; Y. filifera: U=135, Z=-1.78, P=0.81; P. aculeata: U=192, Z=0.22, P=0.84; Density N. laevigata: U=197, Z=-0.08, P=0.97; V. farnesiana: U=180, Z=-0.58, P=0.60; D. texana: U=147, Z=-1.43, P=0.15; Y. filifera: U=184, Z=-0.42, P=0.67, P. aculeata: U=147, Z=-1.43, P=0.15).

The comparison between collecting seasons resulted in a higher richness and density of seedlings in the autumn bank under all species (Richness N. laevigata: U=49, Z=-4.12, P<0.01; V. farnesiana: U=86, Z=-3.10, P=0.002; D. texana: U=80, Z=-3.27, P=0.001; Y. filifera: U=60, Z=-3.82, P<0.01; P. aculeata: U=64, Z=-3.77, P<0.01; Density N. laevigata: U=40, Z=-4.13, P<0.01; V. farnesiana: U=80, Z=-3.24, P=0.001; D. texana: U=132, Z=-1.82, P=0.06; Y. filifera: U=74, Z=-3.42, P<0.01; P. aculeata: U=63, Z=-3.71, P<0.01).

The zoochorous dispersal taxa (N. laevigata and D. texana), had a higher average canopy richness than the rest of the species (H=21.14; d. f.=4; P<0.001). In addition, the seedling density was higher under N. laevigata than under V. farnesiana and P. aculeata; the number of species with unassisted seeds was even larger under N. laevigata with respect to Y. filifera, which is also zoochorous (H=23.44; d. f.=4; P<0.001.) (Figure 2).

Density data are rounded to the nearest whole number. The error bars represent the standard deviation.

Figure 2 Average seedling density under each species. 

In general, more richness and density of herbaceous than of arboreal species was obtained under all species (Richness U=3 765, Z=-3.11; P=0.002; Density U=4 075, Z=-2.31, P=0.02). More tree seedlings emerged on average under D. texana and Y. filifera (two species, H=29.93, d. f.=4, P<0.001), and more herbaceous species under N. laevigata (five species, H=13.96, d. f.=4, P=0.007). The highest density of arboreal species seedlings was found to occur under N. laevigata, D. texana, and Y. filifera, while the density of herbaceous plants was highest under N. laevigata (H=22.47, d. f.=4, P<0.001; H=16.97, d. f.=4, P=0.002, respectively) (Figure 3).

Density data are rounded to the nearest whole number. The error bars represent the standard deviation.

Figure 3 Average density of tree and herbaceous seedlings under each species. 

Discussion

The results of this study showed a higher species richness in the seed bank than in any other study conducted in the Tamaulipan Thorny Scrub; 27 more species were identified than those recorded by ^{Martínez-Adriano et al. (2021)}, who cite 30 species in seed banks present in remnant fragments of the Tamaulipan Thorny Scrub. Likewise, the richness was higher than that estimated in an area of dense scrub (33 species) (^{Pando-Moreno et al., 2010}). On the other hand, the average densities of seedlings m² under N. laevigata and V. farnesiana are between those recorded in the dense area (1 900 seeds m²) and in an eroded area (62 seeds m²), both with Tamaulipan Thorny Scrub (^{Pando-Moreno et al., 2010}).

There was no annual variability in seed bank density and richness, but there was seasonal variation, as there were more species and density in the seed bank collected in autumn. In other semi-arid regions, variations in species richness, abundance, and seed density in the soil are recorded in different years, due to variations caused by extreme climatic conditions (^{Quevedo-Robledo et al., 2010}; ^{Dreber and Esler, 2011}; ^{da Silva et al., 2013}). Fruit production can change depending on environmental factors and on the resources that are variable over time, consequently, temporal changes are observed in seed banks (^{Quevedo-Robledo et al., 2010}); although some plants are very constant and maintain their phenological patterns or respond to the variability of resources with a shift of a few days in their phenological stages (^{Gordo and Sanz, 2009}). Slight variation in fruit production may not be sufficient to be reflected in the seed bank composition in short-term studies.

On the other hand, in zoochorous plants, the abundance of fruit consumers may increase or reduce the density of deposited seeds (^{García and Martínez, 2012}), however, there is little or no information on these interactions in the region. Furthermore, the various species in a community usually have different germination strategies, and with the method used in this research, these strategies directly influence the results, so that the recorded seasonal variation may depend on whether the necessary germination conditions are met or not. For the plants of the thorny scrub, ^{Jurado et al. (2001b)} indicated that some of the studied species germinated more in autumn, consistently with the results documented herein. Seed bank research using other methods such as sieving, may help to confirm whether there are differences in seed numbers or whether the seasonal variations detected are caused by the germination adaptations of the community. The analysis of the seed bank must take into account this seasonal variation, since sampling in a single season would provide incomplete information for the proper evaluation of disturbed areas or the determination of the potential plant richness.

The results show that zoochorous plants had more species and seed density under their canopies, consistently with the stated hypothesis. Similar results have been described for other arid, semi-arid, and dry forests (^{Warnock et al., 2007}; ^{Hadinezhad et al., 2021}), as well as for tropical (^{Camargo et al., 2020}), Mediterranean (^{González‐Varo et al., 2017}), and temperate ecosystems (^{Rodríguez‐Pérez et al., 2014}).

Most of the seeds of the identified tree species have zoochorous dispersal characteristics and are found under a zoochorous tree species (D. texana); thus, the dispersing fauna, among other factors, may shape the tree community as is the case in other ecosystems (^{Warnock et al., 2007}; ^{Hadinezhad et al., 2021}). Moreover, seed removal is slower under isolated trees than in the dense thornscrub (^{Jurado et al., 2006}), which may also influence the number of species identified. According to ^{Jurado et al. (2006)}, the removal was slower, especially in isolated N. laevigata trees, this is one of the species with the highest richness and density of seeds in the studied Tamaulipan Thorny Scrub. The same removal process may determine the seed composition of isolated chapote individuals, which was similar to that of mesquite. The above seems to indicate that in this semi-arid region, the choice of tree species to leave standing or for reforesting influences the ability of the seeds to reach disturbed sites and remain in the seed bank.

The largest number of herbaceous species was obtained in the seed bank under N. laevigata, with a high density and anemochorous or unassisted dispersal. Likewise, the herbaceous species outnumbered the tree species under all species. This is related to the fact that herbaceous plants tend to produce many small seeds, whereas trees produce fewer but larger seeds (^{Moles et al., 2004}; ^{Moles et al., 2005}), also, they are selected more often by predators (^{Dylewski et al., 2020}), a fact which may have influenced the recorded lower density. Another factor is that in general, more herbaceous than woody species are dormant in ecosystems with extreme conditions such as those with frost or drought, as is the case of the Tamaulipan Thorny Scrub. The ratio of dormant to non-dormant species does not differ significantly (^{Jurado and Flores, 2005}).

All this information is important because it represents the potential richness that would be obtained in a natural regeneration trajectory. With the evaluation of seed banks and standing vegetation, it is possible to interpret whether areas that appear degraded will recover the pre-disturbance vegetation structure and diversity or have an alternative status (^{Ma et al., 2021}). In the present study, a high density and richness of native species was determined, in addition, it is inferred that there is an influx of seeds from nearby areas, which suggests that the studied area is in a suitable condition to recover its vegetation through natural regeneration, once disturbances cease to occur.

Conclusions

Isolated trees of the native species Diospyros texana, Neltuma laevigata, Yucca filifera, Parkinsonia aculeata and Vachellia farnesiana of the Tamaulipan Thorny Scrub, are reservoirs of a high richness and diversity of seeds. Seed banks composed under species with zoochorous dispersal maintain a higher seed richness and density than those consisting under taxa with other dispersal syndromes. The number of species and the density of germinable seeds undergo a seasonal variation, as the assessed parameters are higher in autumn than in spring.