Plant invasion science in protected areas: progress and priorities

Abstract

Invasive alien species are a major problem for managers of protected areas (PAs) worldwide. Until the 1980s biological invasions were widely considered to be largely confined to anthropogenically disturbed sites and the widespread disruption of ecosystems in PAs by invasive species was not globally perceived as a major threat. A working group of the SCOPE program on biological invasions in the 1980s showed that PAs are not spared from major disruptive effects of invasions. Early research focused on descriptive studies of the extent to which PAs were invaded. More recent research explored drivers of invasion, and in the last decade much work has focused on understanding the impacts of invasions. We review the current understanding of alien plant invasions in PAs, focusing on four themes: (1) the status and macroecological patterns of alien plant invasions; (2) the threats that invasive alien plants (IAPs) pose and the impacts detected to date; (3) the current focus of invasion science in PAs; and (4) research priorities for advancing science-based management and policy. Of a sample of 59 widespread IAP species from a representative sample of 135 PAs globally, trees make up the largest proportion (32%), followed by perennial herbs (17%) and shrubs (15%). About 1857 papers have been published on alien species in PAs; 45% have focused on alien plants. Some textbook examples of impacts by IAPs originate from PAs, illustrating the severe threat to the core function of PAs. Impacts have been quantified at the species and community levels through the displacement and alteration of habitats. In some cases, native species abundance, diversity and estimated species richness have been altered, but reversed following control. At an ecosystem level, invasive plants have radically altered fire regimes in several PAs, in some cases causing regime shifts and transforming woodlands or savannas to grasslands. Invasions have also had a major impact on nutrient cycles. Protected areas are performing an increasingly important part of the global response to stem the rate of environmental change. Despite this, integrated efforts involving science, management and policy that are sufficiently resourced to generate insights on the status and dynamics of IAPs in PAs are insufficient or even lacking. Such efforts are needed to pave the way for monitoring trends, revising legislation and policies, and improving management interventions to reduce the extent and magnitude of impacts of invasive plants in PAs. While policy instruments to support management of non-native species date back to the 1930s, there has been a substantial increase in legislative support and general awareness since the early 2000s. Still, opportunities to improve research for PAs need to be created. Towards this goal, the establishment of a global PA research network could provide a unique vehicle to explore questions across species or functional groups and systems, at a scale currently beyond existing abilities. Developing an integrated global database with standardized, quantitative information could form part of such a networks function.

Appendices

Appendix 1: Search terms used to extract literature records from the SCOPUS database (12/05/2016)

((((((TITLE-ABS-KEY(invasive) OR TITLE-ABS-KEY(alien) OR TITLE-ABS-KEY(non-native) OR TITLE-ABS-KEY(weed))) AND TITLE-ABS-KEY (plant)) AND TITLE-ABS-KEY(“national park”) OR TITLE-ABS-KEY(“nature reserve”) OR TITLE-ABS-KEY(“protected area”) OR TITLE-ABS-KEY(wilderness)))) AND (EXCLUDE(DOCTYPE,“er”)) = 830.

Appendix 2: Text analysis used in Figs. 4 and 5

Text analysis was carried out in R version 3.2.5 (R Core Team 2013) with RStudio version 0.99.896 (RStudio 2013) and quanteda (Benoit and Nulty 2016), an R package for quantitative analysis of textual data. Year of publication, title and abstracts were extracted for 830 alien plant papers in protected areas from Scopus for the period 1969–2015. Titles and abstracts were combined in a.csv file and years grouped by 20 eras. The.csv file was imported as a corpus {quanteda} object into R with publication year describing the article attributes as docvars {quanteda} and titles + abstracts representing the text to be analyzed. We created a document-feature matrix using dfm {quanteda}, which counts available words by article and attributes. While doing so we removed all English stopwords (very common words or adjectives that can hinder extracting keywords), numbers, punctuation marks, separators, symbols and selected words of limited value (for example, introduction, study). We also stemmed words to ensure better comparison across similar words (for example, management and manage) and removed sparse words to reduce the size of the matrix. The resulting matrix contained 2000 words across 20 eras. Using R’s corrplot package (Wei and Simko 2016) we computed a correlation of the relative frequency of words across the different eras and conducted a Pearson’s product moment correlation test at a 95% confidence level. These results were visualized using the corrplot {corrplot} command (Fig. 4).