Aim
The expansion of agriculture is promoting the loss of natural environments and their biotic homogenization. We aimed at understanding whether the replacement of forests and grasslands by tree plantations leads to biotic homogenization of mammal assemblages of two contrasting Neotropical ecoregions or if dispersal or environmental limitations keep their original assemblages clearly differentiated.

Location
Argentina, Upper Paraná Atlantic Forest, Southern Cone Mesopotamian Savannas and Iberá marshes

Taxon
Mammals

Methods
We conducted two camera-trap surveys, deploying 184 camera-trap stations in continuous forest, fragmented forest, and pine plantations in the Atlantic Forest, and 234 in grassland, fragmented forest, and pine plantations in the Mesopotamian Savannas. We evaluated the similarity of the assemblages among the environments and regions, generating all possible pairwise comparisons using three similarity indices: Sørensen (species identity), Horn (common species), and Morisita-Horn (dominant species). Using variation partitioning diagrams and redundancy analysis, we evaluated the spatial structure of mammal assemblages and the influence of environmental variables.

Results
There was a greater similarity in species identity between different environments within each region than between similar environments in different regions. Common and dominant mammal assemblages of tree plantations tended to be similar between regions and were different from assemblages of the natural environments within the same region. Fragmented forest assemblages were very similar among regions. Assemblages were spatially structured but most of the variation between regions was explained by the environmental variables.  

Main conclusions
Each region has a distinct pool of species, which is partially explained by environmental factors, such as the differential representation of native environments in the landscape. However, an expansion of tree plantations and forest fragmentation in the Atlantic Forest could lead to biotic homogenization between regions due to an increase in the abundance of generalist species.

We conducted camera-trap (Reconyx HC500) surveys in two contrasting ecoregions ("Region"): in the Upper Paraná Atlantic Forest ecoregion ("Atlantic Forest" in the dataset), and in the Southern Cone Mesopotamian Savannas and the Iberá marshes ("Mesopotamian Savannas" in the dataset).  

In the Atlantic Forest, we randomly distributed 184 camera-trap stations (with a minimum distance of 2 km between stations) in three main environments: continuous forest, fragmented forest, and tree plantations (Iezzi et al., 2018). Continuous forest stations (53 camera-trap stations) were located in the largest continuous native forest block (~370,000 ha), surrounded in a 2-km radius by more than 75% of forest cover. Fragmented forest stations (69 stations) were in forest remnants with varying degrees of anthropogenic disturbance, outside the continuous forest (Iezzi et al., 2019). Stations in tree plantations (62 stations) were located in 4- to 14-year-old pine plantation stands. The cameras were active between May 2013 and December 2014 for an average of 49.8 days (range: 12 - 123 days), totaling 9,171.8 camera days of effort.

In the Mesopotamian Savannas, we deployed 234 camera-trap stations, at least 2 km apart, in three environments: native grassland (89 camera-trap stations), fragmented forest (54 stations), and tree plantations (91 stations; Iezzi et al., 2020). Stations in tree plantations were immersed in 1- to 30-year-old pine plantation stands. Approximately half of the stations within each environment were located in fields with cattle (N=107; Di Bitetti et al., 2020). Cameras were active between May 2016 and March 2017, with a  mean effort of 44.9 days (range: 21 - 67 days) per station, totaling 10,493.9 camera-trap days.

Stations were not baited and were located off-road, attached to a tree trunk or stake at 25-50 cm above ground level. More than 1 h had to pass for successive pictures of a species to be considered independent records. Records of small (<200 g) sigmodontine rodents were categorized as “sigmodontines” since it was difficult to identify them at species level. We excluded the records of domestic exotic mammals and 328 records that were impossible to identify at species level. Those sites with no records (9 stations) were excluded from the dataset.

For each species, we report the relative frequency of records at each station (# of independent records / # of camera-trap days).

Independent variables

We estimated the percentage of each natural and anthropogenic land cover (“% of forest”, “% of grassland” and “% of wetland”, "% anthropogenic", "% of tree plantation") around the camera-trap stations using a Geographic Information System (GIS) and a land-use raster layer (pixels of 30x30 m) created for 2013-2014 by Zuleta et al. (2015). "% anthropogenic" was estimated as the percentage of all anthropogenic land uses (urban areas, crops, shrub and tree plantations) and "% of tree plantation" as the percentage of tree plantation only. The heterogeneity of land uses ("heterogeneity") was estimated using the Shannon-Wiener index (Shannon & Weaver, 1949) generated with the Fragstat software (ver. 4.2), using the number of pixels of each type of land use, including anthropogenic uses. These variables were estimated at five different radii centered on the camera-trap locations (200, 500, 1000, 2000, and 5000 m). Variable "cost of access" is an indirect measure of human impact and hunting pressure and was estimated as the time required for a person to reach the station from the nearest town or city following De Angelo, Paviolo & Di Bitetti (2011) and Iezzi et al. (2018; 2020). We estimated the livestock density (“livestock”) at local scale as the number of independent cattle records/sampling effort (in days) at each camera-trap station. Climate data were obtained for each station from WorldClim (Fick & Hijmans, 2017), a set of global climate layers with a spatial resolution of about 1 km²: BIO5 corresponds to Max Temperature of Warmest Month; BIO6 to Min Temperature of Coldest Month; BIO14 to Precipitation of Driest Month; and BIO15 to Precipitation Seasonality (Coefficient of Variation).

GIS variables were estimated using ArcGIS 10.6 with the support of the ESRI Conservation Program and the Society for Conservation GIS (SCGIS).

References

De Angelo, C., Paviolo, A., & Di Bitetti, M.S. (2011). Differential impact of landscape transformation on pumas (Puma concolor) and jaguars (Panthera onca) in the Upper Paraná Atlantic Forest. Diversity and Distributions, 17(3), 422–436.

Di Bitetti, M.S., Iezzi, M. E., Cruz, P., Varela, D., & De Angelo, C. (2020). Effects of cattle on habitat use and diel activity of large native herbivores in a South American rangeland. Journal for Nature Conservation, 58(September), 125900.

Fick, S.E., & Hijmans, R.J. (2017). WorldClim 2: new 1-km spatial resolution climate surfaces for global land areas. International Journal of Climatology, 37(12), 4302–4315.

Iezzi, M.E., Cruz, P., Varela, D., De Angelo, C., & Di Bitetti, M.S. (2018). Tree monocultures in a biodiversity hotspot: Impact of pine plantations on mammal and bird assemblages in the Atlantic Forest. Forest Ecology and Management, 424, 216–227.

Iezzi, M.E., Cruz, P., Varela, D., Di Bitetti, M.S., & De Angelo, C. (2019). Fragment configuration or environmental quality? Understanding what really matters for native mammals conservation in the Atlantic Forest of Argentina. Journal for Nature Conservation, 52(June), 125751.

Iezzi, M.E., De Angelo, C., & Di Bitetti, M.S. (2020). Tree plantations replacing natural grasslands in high biodiversity areas: how do they affect the mammal assemblage? Forest Ecology and Management, 473(June), 118303.

Shannon, C., & Weaver, W. (1949). The mathematical theory of communication. Urbana, USA. University of Illinois Press.

Zuleta, G. A., Gauto, O. A., Varela, D. M., Angelo, C. De, Johnson, B. G., Lorán, D., … Zurita., A. A. (2015). Evaluaciones Ambientales Estratégicas y Programa de Monitoreo de la Biodiversidad en las Regiones de Mesopotamia y Delta del Paraná. Proyecto de Conservación de la Biodiversidad en Paisajes Productivos Forestales (GEF TF 090118). Technical report. Buenos Aires, Argentina.

More information about the camera-trap dataset in:

da Rosa, C. A. et al. (2020) NEOTROPICAL ALIEN MAMMALS: a data set of occurrence and abundance of alien mammals in the Neotropics. Ecology. Accepted Author Manuscript. doi:10.1002/ecy.3115

Lima, F., et al. (2017) ATLANTIC-CAMTRAPS: a dataset of medium and large terrestrial mammal communities in the Atlantic Forest of South America. Ecology, 98:2979. doi:10.1002/ecy.1998.

Marques Santos P., et al. (2019) NEOTROPICAL XENARTHRANS: a data set of occurrence of xenarthran species in the Neotropics. Ecology, e02663.

Nagy-Reis, M., et al. (2020) NEOTROPICAL CARNIVORES: a data set on carnivore distribution in the Neotropics. Ecology. Accepted Author Manuscript. https://doi.org/10.1002/ecy.3128