2. Method
2.1. Study area
The study area is limited to the upper area of ​​the Papaloapan River in the state of Oaxaca, southern Mexico. In this area the rivers drain from a maximum altitude of 3,216 m a.s.l. towards a minimum altitude of 1,833 m a.s.l. Strahler order of rivers ranges from 1 at sites in the smallest rivers (headwaters) to 5 at sites that are located on the largest rivers (downstream). The flow ranges from 0.01 l/s to 2.5 l/s and has a negative relationship with altitude (Flores-Galicia et al., 2021). The climate in the upper part of the watershed corresponds to humid temperate with summer rains and in the lower and middle part it corresponds mainly to sub-humid temperate with summer rains. The upland forests are Abies hickelii and Pinus hartwegii forests in the upper areas and mixed forests of A. hickelli -Pinus patula and Pinus pseudostrobus var.apulcensis -Quercus crassifolia towards the middle and low areas of the watershed (Piña & Trejo, 2014). The vegetation of the riparian zone changes along the streams, in the headwaters arboreal elements of species such as A . hickelli , Pinusspp., Quercus spp., towards the intermediate parts Alnus acuminata , Oreopanax xalapensis and Meliosma dentata , while in the lower areas A . acuminata and Salixhumboldtiana are observed (Flores-Galicia et al., 2021).
To evaluate the changes in the communities of propagules, nine sampling sites were established in two tributary rivers (Yavesía’s river and Lachatao’s river) that are part of the same drainage network in the basin, the location of the sites was determined based on a progressive criterion of distance with respect to the origin of the rivers (Fig. 1).
2.2. Sampling propagules
To characterize the communities of propagules dispersed by hydrochory and other types of dispersal in the riparian zone, generative propagules were collected, that is, those that are the product of the sexual reproduction of plants such as seeds or fruits (Pijl, 1969) in two types of traps that were placed in sediment deposition zones adjacent to the streams: (i) Leyer boxes to collect propagules dispersed by hydrochory and (ii) mesh traps to collect propagules dispersed by other types of dispersal. Leyer boxes (sensu Leyer, 2006) were boxes open at the bottom, covered by a 0.15 mm mesh size at the top and fastened by wires. The construction ensured the free flow of water, sediments, and propagules into the box, inside an artificial turf mat were placed (0.5x0.5 m) to retain the propagules. The boxes were tied to 75 cm long metal rods to prevent them from being dragged by the waterflood (Araujo et al., 2015). On the other hand, the mesh traps to collect propagules dispersed by other types of dispersal consisted of extended squares (1.5 × 1.5 m) of shade cloth (1.5 mm mesh) that were placed near the Leyer boxes. Leyer boxes were placed at the lowest line of perennial vegetation (greenline) (Fig. 1c).
In each of the nine sampling sites, two Leyer boxes and three mesh traps were placed to collect propagules dispersed by hydrochory and by other types of dispersal (Fig. 1b). The traps were visited every two or three months to collect the propagules trapped in them between January 2017 and January 2018 (they were visited five times: March, June, August, and November 2017 and January 2018). At each visit the traps were removed and new traps were placed, so that a total of 90 Leyer boxes and 135 mesh traps were placed throughout the sampling period. Due to river flooding and other events, 36% of the Leyer boxes (n=32) and 15% of the mesh traps (n=20) were lost. During this period, propagules in the dry season (March 2017 and January 2018) and in the rainy season (June, August, and November 2017) were collected.
The propagules were identified (to species level or otherwise assigned to a morphospecies), counted and removed manually. The propagules were compared with specimens deposited in the Collection of Fruits and Seeds of the National Herbarium (MEXU), of the Institute of Biology of the UNAM.
2.3. Data analysis
2.3.1. Temporal variations
To detect differences between seasons, a database was prepared differentiating the propagules collected in the dry season (propagules collected in March 2017 and January 2018) and the rainy season (propagules collected in June, August and November 2017), with which performed the following analyses in R ver. 4.0.3 (R Core Team, 2020).
Since differences in the number of propagules between seasons, a standardization of the sampling effort was first carried out by randomly selecting 52 traps for hydrochory (dry= 26 traps, rainy=26 traps) and 92 traps for other types of dispersal (dry= 46 traps, rainy= 46 traps). Subsequently, and because the data were not normal, we performed a Mann-Whitney test to determine differences in the number of propagules between seasons.
To contrast species richness, a rarefaction analysis was performed using the iNEXT function (Hsieh et al., 2016) from the iNEXT package. We calculated the estimated species richness (q0 ) and its 95% confidence intervals for the dry and rainy seasons considering the lower coverage that we recorded in the traps (traps for hydrochory in the rainy season=0.995) (Hsieh et al., 2016). Finally, to determine changes in species composition, a permutational multivariate analysis of variance (PERMANOVA) was performed, applying a Hellinger transformation to the abundance data using the adonis function from the vegan package in R (Oksanen et al., 2013). For the PERMANOVA, a distance matrix was used that was obtained with the Bray Curtis coefficient. To graphically evaluate the associations in the composition of the propagules dispersed by hydrochory and by other types of dispersal in the dry and rainy season, an NMDS was performed with the metaMDS function of the vegan package (Oksanen et al., 2013).
2.3.2. Longitudinal variations along the streams
To evaluate changes in the abundance and richness of species with the distance to the origin of the river, a Generalized Linear Mixed Effects Model (GLMM) with a negative binomial link function was performed using the glmer.nb function of the lme4 package (Bates, 2010). The number of propagules/m2 and the number of species estimated at each site were used as a fixed effect and the stream in which the sites were established (Yavesia river or Lachatao river) as a random effect (Fig. 1).
To determine changes in the turnover and nesting of propagule communities with increasing distance to the origin of the river, values were obtained for the Simpson dissimilarity index (βSim ) and nesting between sites near the river origin (site one: 0.5 km and site two: 0.8 km) and the downstream sites, in this analysis the sites that were located in the Lachatao stream (sites six, seven and nine) were excluded because they are located in a different stream than the rest of the sites. The turnover and nesting values were obtained using the functions vegdist andnestednodf of vegan package (Oksanen et al., 2013). Subsequently, a Pearson correlation analysis was carried out to determine if there was a relationship between the distance to the origin of the river and the turnover and nesting values.