First spectral reflectance dataset of a highly invasive and aggressive alien plant species, Equisetum hyemale (snakegrass), in South Africa: Towards early detection and effective eradication

Abstract

This paper presents the first spectral reflectance dataset of snakegrass (Equisetum hyemale L.) invasive alien species recorded in South Africa. A total of 338 plant specimens were collected in Howick, KwaZulu-Natal province, and carefully stored in a cooler box for less than 24 hours of collection to retain the structural and biochemical state of the specimens and their overall characteristics. Then, spectral reflectance measurements were collected under laboratory conditions using the PSR-300 Spectral Evolution full-range spectrometer, equipped with a bifurbucated cable, a leaf clip and an artificial lighting system. Next, spectral preprocessing was performed in R statistical software to remove noisy spectra and regions and perform averaging per sample. The dataset is critical for early detection of the species and spatial distribution mapping using remotely piloted systems and earth observation satellites, providing essential information for aiding containment and eradication efforts.

Specifications Table

Subject	Earth & Environmental Sciences
Specific subject area	Remote sensing, field spectroscopy, invasion biology, geospatial science
Type of data	Raw (.SED format), Table (.csv format)
Data collection	Samples of the plant species were collected on the 16th of April 2024. The samples were harvested using with a sterilised blade and sealed in a plastic bag. The sealed samples were then placed in a cooler box filled with ice to preserve their quality during transportation to the laboratory.
Data source location	The plant samples were collected in Howick, KwaZulu-Natal, with geographic coordinates (29.48115°S; 30.21735°E). The spectral reflectance was collected in the lab at the University of Johannesburg.
Data accessibility	Repository name: Zenodo Data identification number: 10.5281/zenodo.13363924 Direct URL to data: https://zenodo.org/records/13363925
Related research article

1. Value of the Data

• •Equisetum hyemale has been of interest in the past decades to biologists and paleontologists. Therefore, the spectral signature of the species can open fields to remote sensing research and improve an understanding of the distribution of the species, especially in non-native ranges.
• •Researchers can use the data as inputs to machine learning algorithms to detect and discriminate the species from other similar and co-occurring species. Moreover, the data contains a full spectral coverage, i.e., 350 nm – 2500 nm; hence, it can be resampled to aerial and upcoming and existing satellite sensor configurations towards operational mapping and monitoring of the distribution patterns.
• •Currently, there is no existing spectral library of this species, to the best of our knowledge. Therefore, the dataset is a critical contribution to the characterization of E. hyemale species that is currently listed as a “target for nationwide eradication”, i.e., category 1a, by the National Environmental Management Biodiversity Act of 2004. The spectral dataset provided here can aid in the discrimination of the species from its native counterpart, i.e., Equisetum ramosissimum, within the same genus. Scientists can develop robust models for E. hyemale detection and monitoring.

2. Background

Invasion by Alien Invasive Species (AIS) are considered one of the major global causes of biodiversity loss and ecosystem functioning [16]. Increased human movement and international trade played a significant role in the introduction of AIS [18]. Alien invasive Plants (AIPs) have been introduced in different regions for various purposes including food, shelter, ornamental and medicine for many years [8]. Among them, ornamental plants have a major contribution to the escape, naturalization and invasiveness of (AIPs) [14,18].

AIPs can directly or indirectly affect the native plants by reducing or introducing new pollinators and pests, alter soil nutrients, alter fire regimes and water quantity to the extent that native plants cannot acclimatize to the changing environment [3]. In addition to the natural impacts, AIPs can also affect food security by spreading into the agricultural fields and becoming toxic to livestock [17]. A notable example of a fire-altering invasive plant in western North America is Bromus tectorum, which has altered the fire regimes so significantly that native grasses cannot recover, impacting herbivores dependent on them [3].

Further impacts of AIPs in Eastern Africa, many AIPs have spread throughout the rangelands and thus reducing the capacity of such rangelands to support livestock. Additionally, AIPs have an impact on the livelihood of rural communities that depend on natural resources [19]. South Africa is no exception to the impact of AIPs; a notable example of a widespread AIPs is lantana camara that is been reported to have reduced the grazing potential by almost 80% of invaded areas [17]; reported to have reduced the invertebrate density and can be toxic to livestock in consumed [13]. Managing and controlling AIPs can help achieve the United Nations Sustainable Development Goals (SDGs). For example, protecting native biodiversity can assist in achieving SDG 15 and ultimately, improve ecosystem services, food and water security, human health and contribute to climate change alleviation [12].

A typical example is the Equisetum hyemale (snakegrass) in South Africa. It is an ornamental evergreen rhizomous joined perennial herb native to temperate Asia, America and Europe [2;6]. The species is a popular ornamental plant and can be an aggressive invader once planted. It reproduces mainly through the rhizomes and can also produce spores [15]. Additionally, a broken piece of the plant or rhizomes can regenerate into a new plant. Equisetum hyemale contains silicone that makes it resistant to herbicides. Uprooting is shown not to be effective as the broken underground rhizomes can regenerate [9]. Although there is no scientific evidence of a naturalized population in South Africa, the species is believed to be sold in nurseries and planted in home gardens. People usually confuse the alien E. hyemale with the native E. ramosissimum because they have morphological similarities. Consequently, because of high silica content, E. hyemale in large quantities can be toxic to livestock if they consume contaminated hay. Furthermore, E. hyemale can impact crop yield and native plants by releasing substances that can suppress the growth of neighboring plants [4].

Unlike the traditional methods of fieldwork that require manpower, costs of fieldwork and is time-consuming, hyperspectral remote sensing has been used to discriminate species from co-occurring species because of the narrow and continuous bands it provides [16]. Early detection of alien species using remote sensing can assist with the management and control of alien species at an early stage of their invasion [7]. Additionally, early detection of aggressive invaders can increase the chances of eradication other than attempting to eradicate a matured population. Spectral reflectance of the species is one step to discriminating that species from other species for distribution map using classification [16].

3. Data Description

In this report, we present the spectral library of Equisetum hyemale L. (snakegrass) collected between 16 April and 17 April 2024 (Figs. 1-4). Fig. 1 shows a raw spectral library representing averaged spectra of 22 individuals of E. hyemale specimens before any processing.

Fig. 1.

Averaged raw (unprocessed) spectral reflectance of Equisetum hyemale (snakegrass). Each spectral reflectance curve represents averaged (9-12 measurements) per E. hyemale specimen.

Fig. 4.

The minimum, mean and maximum reflectance of Equisetum hyemale.

Fig. 2 is a pre-processed spectral library by filtering noisy regions known to be insignificant for vegetation analysis [7]. These regions include 350 – 399 nm, 1350 – 1465 nm, and 1790 – 1960 nm. In our analysis the region between 2350 and 2500 nm was also noisy and removed.

Fig. 2.

Mean of all 22 samples without the noisy regions, i.e., 350 – 399 nm, 1350 – 1465 nm, 1790 – 1960 nm and 2350 – 2500 nm.

Fig. 3 shows the reflectance of each of the 22 specimens plotted individually.

Fig. 3.

Averaged and filtered spectral library of E. hyemale, plotted individually.

Fig. 4 shows the summary statistics of spectral signatures for all 22 specimens (i.e., Minimum, Maximum and the Mean). Generally, the full-range spectral signature can be categorized into regions: Blue (350 – 449 nm), Green (450 – 549 nm), Red (550 – 649 nm), Red-edge (650 – 749 nm), Near-infrared (NIR, 750 – 1299 nm) and Shortwave infrared (SWIR, 1300 – 2500 nm). The region from 400 – 700 nm (Visible region) is sensitive to the chlorophyll content, where the peaks can be seen in the Green region, with a Minimum reflectance of <20% and a maximum of approximately 35%. As can be expected, there is a pronounced chlorophyll absorption of in the Red region, i.e., <20%, and the peak reflectance of >60% in the NIR regions of the electromagnetic spectrum. It can be noticed that the spectral reflectance range in the NIR is the greatest across the spectral range, indicative of the complex cell structure of the various components of E. hyemale measured. The transitional region known as the Red-edge region, is known for its sensitivity to chlorophyll amount changes, e.g., a decline because of biotic and abiotic stress. The SWIR is characterized by absorption due to leaf-water-content [10].

4. Experimental Design, Materials and Methods

We collected a total of 22 specimens, and each was sealed as several stocks in a separate plastic bag (see Fig. 5d). The collected samples were brought to the lab for spectral measurements. We used the Spectral Evolution PSR-3500 full-range spectrometer to collect the spectral reflectance measurements of the samples collected [5;7]. This spectrometer encompasses a spectral range spanning from 350 nm to 2500 nm, with spectral resolutions of 3.5 nm, 10 nm, and 7 nm at 350 nm – 1000 nm, 1500 nm, and 2100 nm, respectively. The spectral bands in the 350 nm–1000 nm, 1500 nm, and 2100 nm regions are characterized by nominal spectral sampling intervals of 1.5 nm, 3.8 nm, and 2.5 nm, respectively [5]. The spectrometer’s power supply produces an artificial light that runs through the bifurcated cable that is connected to the leaf clip (Fig. 5a-c). All 22 specimens collected were preserved in a cooler box and spectral measurements were taken within 24 hours of collection (Fig. 5d). The leaf clip was used because the spectral reflectance was collected in the laboratory. We turned off the laboratory lights to ensure that the only light that was available was from the power supply and to ensure that there was no additional light that could alter the results. This light is useful in a laboratory to account for the absence of sunlight in the field. Prior to the acquisition of the spectral measurements, we calibrated the spectrometer by placing the white reflectance on the leaf clip. This process was performed after every measurement of each specimen (i.e., up to 15 measurements per specimen).

Fig. 5.

Experimental setup using PSR-3500 Spectral Evolution spectrometer (a), equipped with a handheld Personal Digital Assistant (PDA) (b). An image of Equisetum hyemale (snakegrass) aerial shoot (c) and Cooler box to temporally keep the specimen condition prior to laboratory measurements (d).

The measurements were collected from various parts of the plant specimen such as the tips (Fig. 6a), areas near the nodes (Fig. 6b), where the lateral branches emerged (Fig. 6d, close to the rhizomes (Fig. 6e), and in the middle of the stock (Fig. 6c). This was to ensure complete characterization and that each sample spectra were representative of the entire plant. Whenever anomalous measurements were detected, they were flagged and removed. This process resulted in a total of 338 spectral measurements. Next, we averaged the remaining spectral measurements by plant specimen (or sample), with each averaged consisting of about 9 to 12 measurements. After the removal of all the anomalies, we further analyzed the data in R statistical software [11] to filter all noisy regions that are not important for vegetation analysis, i.e., 350 – 399 nm, 1350 – 1465 nm, 1790 – 1960 nm and 2350 – 2500 nm [7]. The R script to visualize the spectral measurement using R statistical software is provided on GitHub (https://github.com/mkganyago/SpectralEvolutionFileReader), and it is linked to this data’s repository [1].

Fig. 6.

Illustration of the plant parts where the spectral measurements were taken. [a] At the tip of plant, [b] near the node, [c] internodes, [d] lateral branches and [e] near the rhizomes.

Limitations

The data was collected under lab conditions and may differ from those collected under field conditions where a myriad of materials such as litter and soil background exist and affects the output reflectance signature of the species.

Ethics statement

The authors confirm that, after reading the ethical requirements for publication in Data in Brief, the current work does not involve human subjects, animal experiments, or any data collected from social media platforms.

Credit author statement

Lesibana Sedibana: Conceptualization, Writing – original draft, Writing – review & editing, Methodology, Investigation, Visualization; Yessoufou Kowiyou: Writing – review & editing, Supervision, Conceptualization; Mahlatse Kganyago: Methodology, Visualization, Validation, Conceptualization, Software, Formal analysis, Writing – review & editing, supervision, Data curation; Thulisile Jaca: Writing – review & editing, Conceptualization, Funding acquisition.

Data Availability

• ZenodoFirst spectral Reflectance Dataset of Equisetum hyemale (Snake grass) Invasive Alien Plant (Original data).