By Barberis Davide, Pittarello Marco, Lombardi Giampiero and Lonati Michele
From Ente di gestione delle Aree Protette delle Alpi Marittime and University of Torino
“In European mountain regions, it is very common to use hay transfer from healthy donor sites to restore grasslands. A new R package and shiny app can help practitioners match the suitability of a donor site to a recipient site, based on elevation, slope and aspect.”
The rising demand for restoration requires specific and subjective methods for the planning and operational parts of restoration programs. In particular, the sourcing of native seeds is essential for a good restoration outcome. Understanding the suitability of a seed mixture from a given donor site to a restoration site may be challenging. The lack of available research and modeling regarding species and habitat distribution along vegetation gradients implies that the success of restoration is nowadays monitored mostly ex-post rather than at the planning stage, and mainly through expert-based approaches.
Native seeds must be used in homogeneous areas at biogeographical, and thus macro-climatic, level. Consequently, differences among seed provenances should be found in micro-climate, which is mainly influenced by topography, in particular elevation, but also slope and aspect. They don’t have a direct impact on plant physiology, but directly affect some variables which do. Elevation is a good proxy for temperature and rainfall, while aspect and slope are mainly proportional to the solar radiation reaching plants and soil (Austin, 2013; Boehm et al., 2021). For these reasons, the three topographic variables mentioned above correlate well with the vegetation composition. Thus, topography could indirectly affect the outcome of a restoration process and it could be useful for local predictions.
ResNatSeed (Barberis et al., 2023) is a tool built in R language (R Core Team, 2019) that computes the suitability of a certain seed mixture with the conditions of the site where the seeds would be used for restoration purposes. The computation process uses the composition of the seed mixture, or the vegetation composition of the donor grassland where the mixture is harvested, and the topographic features of the restoration site. Based on statistical models, the expected abundance in the restoration site is calculated for each species, and then an index of suitability of the seed mixture (Suitability Index, SI) is predicted, as a function of elevation, slope, and aspect of the restoration site. The abundance of the species potentially occurring in the mixture is determined from a large training database of vegetation surveys. Such a database can be either the default vegetation database, which contains 4081 vegetation surveys in the Piedmont Region (NW Italy), or provided by the user to allow the usage of ResNatSeed for any geographical area. ResNatSeed is available both as a package running under R and as a Shiny app available offline and on the web either.
The training database for ResNatSeed originates from a set of located vegetation surveys and their corresponding topographic variables. For reliable modeling, the customized databases should include a number of vegetation surveys sufficient to take the main environmental gradients into account. Nowadays, a lot of large vegetation databases already exist for different geographical areas, e.g. VegBank for North America (http://vegbank.org/vegbank/index.jsp), European Vegetation Archive for Europe (http://euroveg.org/eva-database) and sPlot for the whole world (https://www.idiv.de/?id=176&L=0). These huge databases could potentially be used as input of ResNatSeed for specific areas of interest. If topographic variables are not available from databases, Digital Elevation Models (DEMs) are often made accessible for free by local administrations at a fine spatial grain and, if not, 30-m grid DEMs covering the whole world are available too (e.g. Space Shuttle Radar Topography Mission, https://earthexplorer.usgs.gov and ASTER Global Digital Elevation Model, https://asterweb.jpl.nasa.gov/gdem.asp).
This method necessarily simplifies complex germination processes under the following assumptions. With regards to the abundance of seeds in the batch, the amount of seeds per species is a good approximation of the germination potential of the different species, not considering any kind of dormancy. This approximation is possible only for areas with a low rate of dormancy, like temperate European areas, where dormancy is usually below 50% (e.g. Scotton, 2018) and is mostly broken after the first cold season. In drylands, where dormancy rate is well above 50% (Kildisheva et al., 2020), this method could be highly unreliable. An additional approximation concerns the use of the donor grassland composition to estimate the seed mixture composition. The differences in the phenology of the various grassland species mean that when harvesting seeds, only the species with mature propagules will be included in the seed mixture, which will likely result in an overestimation of the number of species present in the batch. However, if harvested when most of the dominant species are ripe, the abundance of the species in the grassland can represent most of the species and can be used as a proxy to calculate the potential adaptability of a seed mixture to a restoration site.
Figure 1. Workflow of the algorithm.
ResNatSeed could help reduce the subjectivity during the assessment of ecological compatibility between donor grasslands and restoration sites. While this application cannot calculate the germinability of the mixture, it can improve the choice of the best-suiting species or donor grasslands according to their ecological compatibility with the restoration site. ResNatSeed uses only three main topographic variables easily extractable from a DEM, but it could be further improved with virtually every topographic climatic or litho-pedological variable available for different areas or purposes, thanks to the open access coding. The implementation of ResNatSeed on an easy-to-use Shiny web app further extends the applicability of the tool, allowing professionals and non-experts of the software R to use this modeling method for restoration purposes.
Software and data availability
R package
Source code of Resnatseed R package: https://github.com/MarcoPittarello/ResNatSeed
Package website and vignettes: marcopittarello.github.io/ResNatSeed/
Shiny app
Source code of Resnatseed Shiny App: https://github.com/MarcoPittarello/ResNatSeed_ShinyApp
Shiny app website: https://marco-pittarello.shinyapps.io/ResNatSeed_ShinyApp/
In addition to the ResNatSeed application, the website includes a series of tutorials for the correct use of the tool and a glossary of terms used.
Research article
https://doi.org/10.1016/j.envsoft.2023.105813
References
Austin, M.P., 2013. Vegetation and Environment: Discontinuities and Continuities. In: van der Maarel, E., Franklin, J., Vegetation Ecology, Second Edition. Wiley-Blackwell, 71-106 https://doi.org/10.1002/9781118452592.ch3
Barberis, D., Pittarello, M., Lombardi, G., Lonati, M., 2023. ResNatSeed: An R package and shiny web app to predict the REStoration potential of NATive SEEDs using topographic factors. Environmental Modelling and Software, 169: 105813. https://doi.org/10.1016/j.envsoft.2023.105813
Boehm, A.R., Hardegree, S.P., Glenn, N.F., Reeves, P.A., Moffet, C.A., Flerchinger, G.N., 2021. Slope and Aspect Effects on See db e d Microclimate and Germination Timing of Fall-Planted Seeds. Rangeland Ecology & Management, 75: 58–67. https://doi.org/10.1016/j.rama.2020.12.003
Kildisheva, O.A., Dixon, K.W., Silveira, F.A.O., Chapman, T., Di Sacco, A., Mondoni, A., Turner, S.R., Cross, A.T., 2020. Dormancy and germination: making every seed count in restoration. Restoration Ecology, 28 (S3): S256-S265. https://doi.org/10.1111/rec.13140
R Core Team, 2019. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/
Scotton, M., 2018. Seed production in grassland species: Morpho-biological determinants in a species-rich semi-natural grassland. Grass and Forage Science, 73 (3): 764-776. https://doi.org/ 10.1111/gfs.12359