Research Article |
Corresponding author: Andrea Mojzes ( mojzes.andrea@ecolres.hu ) Academic editor: Milan Chytrý
© 2024 György Kröel-Dulay, Attila Rigó, Eszter Tanács, Katalin Szitár, Gábor Ónodi, Eszter Aradi, Zsolt Bakró-Nagy, Marianna Biró, Zoltán Botta-Dukát, Tibor Kalapos, András Kelemen, Annamária Laborczi, László Pásztor, Quinter Akinyi Rabuogi, Andrea Mojzes.
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation:
Kröel-Dulay G, Rigó A, Tanács E, Szitár K, Ónodi G, Aradi E, Bakró-Nagy Z, Biró M, Botta-Dukát Z, Kalapos T, Kelemen A, Laborczi A, Pásztor L, Rabuogi QA, Mojzes A (2024) Explosive spread of sand dropseed (Sporobolus cryptandrus), a C4 perennial bunchgrass, threatens unique grasslands in Hungary (Central Europe). NeoBiota 95: 59-75. https://doi.org/10.3897/neobiota.95.124667
|
Sporobolus cryptandrus is a C4 perennial bunchgrass native to extensive areas of North America. As a non-native species, it has been reported from several continents, and it has been described as a transformer species in sand steppes of Central and Eastern Europe. However, its spreading ability across the landscape and within habitats has not been quantified, and factors determining its success have not yet been assessed.
In this study, we focused on the largest stronghold of S. cryptandrus invasion in Hungary, where the species was first recorded in 2016, and investigated its present distribution in the landscape by mapping along dirt roads. In a separate local study in a heavily infested sand dune site of 2 km2, we assessed the infestation level and factors affecting the species’ establishment.
Our landscape-scale mapping found that in April 2023, the distribution of S. cryptandrus encompassed a largely contiguous 600 km2, with documented presence from 282 1-km2 mapping units. The species occurred more than 5 m away from roads in 71 mapping units, mostly in the centre of its distribution area. Sporobolus cryptandrus presence was negatively related to soil organic matter content and positively related to sand content. At the local scale, we found the species in 39% of vegetation plots in a sand dune site originally covered by Pannonic sand steppes, a priority habitat in the EU Habitats Directive. Sporobolus cryptandrus presence at this site was negatively related to the total cover of resident grassland but, surprisingly, was unrelated to the distance from roads. Collectively, these results suggest that primary spread occurs mostly along roads; these roadside populations likely serve as sources for establishment in neighbouring grasslands, but subsequent mass invasion becomes independent from roads.
Together with the species’ broad macroclimatic tolerance in North America and reported mass invasion events in Ukraine and Russia, our results suggest that S. cryptandrus likely poses a broad-scale threat to Eurasian dry grasslands, in particular on coarse-textured sandy soils with low vegetation cover.
Dirt roads, distribution, invasive grass, mapping, open habitats, sand steppe, sandy soil, unproductive soil
Invasive grasses pose serious threats to natural communities by substantially altering the biodiversity and functioning of ecosystems (
Sand dropseed (Sporobolus cryptandrus (Torr.) A. Gray) is a C4 NAD-ME perennial bunchgrass (
Outside of its native range, casual or naturalised populations of S. cryptandrus have been reported from numerous locations in Eurasia (from Spain to Japan), Australia and New Zealand (Suppl. material
The overall objective of our study was to investigate the current distribution of S. cryptandrus at the species’ largest naturalised population in Hungary and identify factors associated with its successful spread at the landscape and local scales. At the landscape scale, we mapped S. cryptandrus along dirt roads and neighbouring habitats and tested if the species’ presence is related to soil characteristics and specific habitats. At the local scale, we assessed the species frequency, and tested if its occurrence is related to distance from roads and the total cover of resident grassland in a heavily infested site of Pannonic sand steppes, a priority habitat type of the EU Habitats Directive (Natura 2000 code: 6260*;
In Hungary, S. cryptandrus was discovered in 2016, near the city of Kiskunhalas (Danube–Tisza Interfluve, Central Hungary) and in the city of Debrecen (eastern Hungary;
The study was conducted in the southern part of the Danube–Tisza Interfluve, Central Hungary, where the largest number of S. cryptandrus occurrence data was recorded prior to our systematic mapping (Suppl. material
Since we planned to map the current distribution of S. cryptandrus, we did not have an a priori defined study area. As a starting point, we used S. cryptandrus occurrence data from the database of the Kiskunság National Park and from the co-authors of this paper, in the vicinity of the first record of the species, south-east of the city of Kiskunhalas (
We implemented the mapping by driving slowly (max 15 km/h) on dirt roads and visually searching for S. cryptandrus individuals along the roads. We decided to sample along dirt roads because (a) large populations of S. cryptandrus were reported to occur along dirt roads (
Sporobolus cryptandrus in the study area, south-east of the city of Kiskunhalas. The species most frequently grows on and along dirt roads (upper left picture; 7 February, 2023), but it can also form monodominant patches away from roads (upper right picture; 7 February, 2023). It can establish and grow in natural grasslands (lower left picture; large green tussocks in the foreground are S. cryptandrus, whereas small yellowish bunches are native Festuca vaginata, 13 September, 2022) and take over dominance (lower right picture; 13 September, 2022). Pictures were taken by G. Kröel-Dulay and A. Rigó.
During mapping, a new S. cryptandrus occurrence was only recorded when we were at least 120 m (checked on GPS) from an already documented occurrence. We recorded both the presence along the road, defined as within 5 meters of the road edge, and the presence further away from the road, defined as more than 5 meters from the road edge. Based on our field experience, we think that we can, with high probability, spot average-sized S. cryptandrus individuals to at least 30 m in open grasslands, where the species occurs. Starting in the close vicinity of the first reported occurrence (recorded in 2016) and moving gradually away, we sampled all dirt roads unless they were closer than 1 km to an already sampled parallel road. We gradually expanded the search area based on newly detected occurrences, until no additional occurrences were found within a ca. 2 km radius from the outermost occurrences. In the case of already known or accidentally found sporadic occurrences, we only searched their neighbourhood. Mapping in some areas was constrained by the lack of dirt roads or inaccessibility (e.g. fenced areas). While driving, we recorded not only S. cryptandrus occurrences, but also the tracks we drove, in order to report the verified absence of the species.
For the visualisation of S. cryptandrus distribution in the landscape scale, we plotted our data using the EEA reference grid cells of the ETRS89-LAEA Europe coordinate system (EPSG 3035;
Distribution of Sporobolus cryptandrus within the study area at a 1-km resolution in April 2023 and the first documented record in 2016. Verified absence denotes cells where at least 500 m dirt roads were sampled, and no S. cryptandrus was found, while no data denotes cells that were not sampled, or less than 500 m dirt roads were sampled and no S. cryptandrus was found. The core area was defined as the convex hull around the outermost occupied cells that are not separated from the large areas of mostly contiguous distribution (occupied cells) by more than one empty cell. Sporobolus cryptandrus presence further away from roads means cells where the species was present more than 5 m away from roads (in almost all of these cases, it was found also along the roads).
We studied the local-scale distribution of S. cryptandrus in a heavily infested sand dune system close to the first record of the species within the study area (coordinates: 46.414, 19.550 EPSG 4326; Fig.
Statistical analyses were performed using R version 4.1.1 (
We modelled the effect of habitat types, and the percentage sand and SOM contents of the soil on the presence of S. cryptandrus with three separate binomial generalized linear models. We used separate models because sand and SOM contents were correlated, and habitat types were also related to soil properties. For the three models, we first used all sampling points but found significant spatial autocorrelation in the residuals based on Moran’s tests in the DHARMa R package (
In order to explain S. cryptandrus’ presence within a highly infested grassland site in the local survey, we included the distance from the nearest dirt road and the total cover of resident grassland as explanatory variables in two separate binomial generalized linear models. The dirt road map was obtained by digitizing frequently used roads from a satellite image, and we used the Near tool in ESRI ArcMap 10.8 to get a distance measure for each plot (n = 100). The cover of resident grassland was obtained by summing up the percentage cover of each species present in the given plot (without S. cryptandrus). Plots with S. cryptandrus cover above 5% (15 plots) were left out of this analysis, because a substantial S. cryptandrus cover may itself induce a decrease in the cover of the resident species. We tested for but found no spatial autocorrelation in the model residuals by using the DHARMa R package (
We sampled (drove) a total of 1326 km and detected 902 S. cryptandrus occurrences (with the constraint that neighbouring occurrences are at least 120 m from each other), spread over 266 1-km2 units of the EEA reference grid. After merging these with previous data collected between 2016 and 2022, S. cryptandrus occurred in 282 mapping units (red cells in Fig.
Across the core area of its present distribution in the study area, the probability of S. cryptandrus presence was negatively related to soil organic matter content (χ2 = 43.71, df = 1, p < 0.001; Fig.
The relationship between the probability of Sporobolus cryptandrus presence at the landscape scale and a soil organic matter (SOM) content b soil sand content, and c the presence or absence of major habitat types within a 20-m radius around the sampling points along dirt roads.
In the heavily infested Pannonic sand steppe site in the middle of S. cryptandrus distribution, the species was present in 39% of vegetation plots, and out of these, S. cryptandrus was the dominant species (i.e. species with the highest cover in a study plot) in nine vegetation plots. The probability of S. cryptandrus presence was not related to how far a plot was located from dirt roads (χ2 = 1.07, df = 1, p = 0.300; Fig.
The modelled relationship between the probability of Sporobolus cryptandrus presence in plots within a heavily infested open grassland site and a the distance from the nearest dirt road (n = 100) and b the total cover of the resident grassland community (n = 85). To study the relationship with the total cover, we left out plots with the S. cryptandrus cover over 5%, because we wanted to avoid S. cryptandrus substantially affecting the resident community in our samples.
Our systematic mapping showed that in April 2023, S. cryptandrus had a largely contiguous distribution in a 591 km2 area, occurring in about half of the mapping units within this area. This is striking because the species was first recorded in the region in 2016 (
The fast spreading of S. cryptandrus may be explained by its prolific seed production (a single panicle can produce ten thousand seeds;
We detected S. cryptandrus further away from roads in 26% of occurrences (in almost all of these cases, the species was found also along the roads), particularly in the centre of the core area, in the vicinity of the first record of the species, where dirt roads were highly infested with S. cryptandrus. This distribution pattern suggests that S. cryptandrus spreads primarily along dirt roads but has an ability to enter natural grasslands (Fig.
We found that the presence of S. cryptandrus is positively related to the sand content of the soil, which is in agreement with previous reports on the species. In fact, the common English name of S. cryptandrus is sand dropseed, which reflects its preference for sandy soils in its native range (
The relationships of S. cryptandrus occurrence with the presence of specific habitats were consistent with the correlations between the species’ presence and soil characteristics. The probability of S. cryptandrus presence along dirt roads was positively related to the presence of open grasslands and coniferous forests, habitat types that typically appear on soils with high sand and low soil organic matter content in the region (
Our local survey in a sand dune site showed that once established, S. cryptandrus can become very frequent (it was present in 39 out of 100 random vegetation plots) in natural sand grasslands within about six years. Furthermore, these occurrences were not correlated with the distance from dirt roads but were associated with low (< 50%) cover of the resident grassland. These results suggest that when S. cryptandrus enters natural grasslands, its further spread may become independent from dirt roads. Moreover, S. cryptandrus has become even dominant in several stands (in nine vegetation plots) since its first record in 2016. This implies that, as a dominant species, S. cryptandrus likely has a strong impact on community structure and ecosystem functions. Previous studies in the region demonstrated that the high cover of S. cryptandrus was associated with reduced species richness, increased vegetation height, and decreased abundance of certain plant functional groups (e.g. other perennials and insect-pollinated species;
Collectively, the results of our landscape- and local-scale study revealed that S. cryptandrus has an ability to spread very fast along dirt roads, and these roadside populations likely serve as sources for its establishment in neighbouring grasslands, where the invasion of this grass poses a serious threat primarily to open perennial grasslands on unproductive sandy soils. These grasslands, the Pannonic sand steppes, are a priority habitat type of the EU Habitats Directive (Natura 2000 code: 6260*;
The preference of S. cryptandrus to open grasslands at this stage of its invasion may be related to the fact that bare soils can provide a more favourable environment for seed germination of this grass. Previous studies reported that both diurnal temperature fluctuations, which are larger in bare soil than in soil beneath a closed grassland canopy (
The rapid spread of S. cryptandrus and its potential risk for natural sand grasslands call for developing effective management strategies against the species. Given the large population size and area covered by the species in our study area, full eradication is unrealistic. Therefore, preventing S. cryptandrus from further spread along dirt roads and particularly from entering natural grasslands should be the priorities for action. Our results highlight the need for regular monitoring of dirt roads in actual infested and high-risk areas, and for blocking roads in sites where S. cryptandrus are present in high densities. Our map on the current distribution of S. cryptandrus may help to identify roads that should be carefully monitored for new infestations within and around the study area, and newly established and small populations should be eradicated. Regarding the natural grassland sites where S. cryptandrus already reached relatively high abundance, further studies are needed to assess how various habitat management practices (e.g. grazing, mowing, prescribed burning or their combination) affect the competition between S. cryptandrus and resident native plant species. Finally, if the mass invasion of S. cryptandrus cannot be stopped or at least slowed down, representative stands of unique Pannonic sand steppes should be designated, which are intensively surveyed and actively kept free of the species.
We thank the reviewer Pavol Eliáš and the anonymous reviewer for their valuable comments to improve our manuscript.
The authors have declared that no competing interests exist.
No ethical statement was reported.
Support from the National Laboratory for Health Security programme (RRF-2.3.1-21-2022-00006) and from the Sustainable Development and Technologies National Programme of the Hungarian Academy of Sciences (FFT NP FTA) is highly appreciated.
György Kröel-Dulay: Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Validation, Writing – original draft, Writing – review & editing.
Attila Rigó: Conceptualization, Data curation, Investigation, Methodology, Validation, Writing – review & editing.
Eszter Tanács: Conceptualization, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – review & editing.
Katalin Szitár: Formal analysis, Visualization, Writing – review & editing.
Gábor Ónodi: Conceptualization, Investigation, Validation, Writing – review & editing.
Eszter Aradi: Data curation, Investigation, Validation, Writing – review & editing.
Zsolt Bakró-Nagy: Data curation, Investigation, Validation, Writing – review & editing.
Marianna Biró: Methodology, Writing – review & editing.
Zoltán Botta-Dukát: Conceptualization, Formal analysis, Writing – review & editing.
Tibor Kalapos: Conceptualization, Writing – review & editing.
András Kelemen: Data curation, Investigation, Validation, Writing – review & editing.
Annamária Laborczi: Methodology, Writing – review & editing.
László Pásztor: Methodology, Writing – review & editing.
Quinter Akinyi Rabuogi: Data curation, Investigation, Writing – review & editing.
Andrea Mojzes: Conceptualization, Investigation, Validation, Writing – original draft, Writing – review & editing.
Attila Rigó https://orcid.org/0009-0005-8401-8922
Eszter Tanács https://orcid.org/0000-0003-1953-9340
Katalin Szitár https://orcid.org/0000-0002-8810-540X
Tibor Kalapos https://orcid.org/0000-0002-1393-5580
László Pásztor https://orcid.org/0000-0002-1605-4412
Andrea Mojzes https://orcid.org/0000-0003-2171-403X
Data that support the findings of this study are available from the authors upon reasonable request. R scripts used for data analysis are available in Suppl. material
Global distribution of sand dropseed (Sporobolus cryptandrus (Torr.) A. Gray) as an alien plant species based on literature data
Data type: xlsx
R scripts and the results of binomial generalized linear models used to test the effect of explanatory variables on the presence of Sporobolus cryptandrus
Data type: pdf