Introduction

During biological invasions, lag times occur when there is a delay between the time of a species' introduction to a new region and the time when it begins to spread or invade. Lag times could be caused by ecological barriers, such as delayed arrival of a required mutualist [1]. Evolutionary factors, such as time required for recombination and adaptation, have been suggested as potential explanation for lag times [2]. Demographic factors such as Allee effects may also be important [3]. Time lags of more than 50 years between first introduction and subsequent invasion are often assumed to be common in plants [4]. In a statistical analysis estimating lag times for woody plants introduced to Europe, the average time between introduction and first spread (escape from cultivation) was estimated at 170 years for trees and 131 years for shrubs [5]. These lag times were inferred indirectly, based on the time difference between when the plants appeared on planting lists and when they appeared in a published flora or sometimes in unpublished floristic data. The use of heterogeneous historical information presents some limitations: actual dates of introductions are sometimes not known and correspondence between recorded dates of naturalization and actual dates of naturalization or start of spread are typically not known. Furthermore, estimates of lag times are coarse-scale when several decades or more pass in the publication of subsequent floras, and floras vary in comprehensiveness and criteria for including species. An analysis of lag times in Australian cultivated plants using planting lists to indicate date of arrival and herbarium specimens as evidence of naturalization estimated a mode time of 149 years between introduction and naturalization [6]. Analyses of herbarium records [7] can provide valuable information about lag times, but they suffer from limitations because many herbarium specimens are collected opportunistically, potentially long after first naturalization, and the collected plant's status (i.e. planted or naturalized) is not always clear.

The common presumption of long lag times for invasions [8] requires testing through direct and frequent observations of the behavior of many introduced species over time. Wangen and Webster [9] point out that the invasion process can include multiple lag phases. For example, a patchy landscape and a long time to invader maturity can cause sporadic rates of spread long after an invasion has started. This study focuses on the lag time between introduction and the start of spread when plantings have been deliberately made in the immediate vicinity of a semi-natural environment. I used historic planting records from the Harold L. Lyon Arboretum (Honolulu, Hawai'i) and documentation of spread from these plantings to directly determine lag times between planting and the start of spread among 23 plants that subsequently became invasive pests in the Hawaiian Islands.

Results

Among hundreds of introduced species that were deliberately planted at the Harold L. Lyon Arboretum, 23 species have become recognized as invasive pests in Hawai'i (Table 1). For these invaders, Lyon's historical records allowed determination of the lag time between planting (Figure 1) and first evidence of spread (Figure 2). Most of the invasive plants (91%) were woody species, with an average lag time of 14 years (range 2–22 years). Among the two herbaceous invaders, the average time lag was 5 years (range 4–6 years). No invaders started to spread between 23 and 89 years after planting (Figure 2). This retrospective analysis does not allow detection of lag times greater than 89 years, but the lack of any invasive species that started to spread in a 66-year period (Figure 2) suggests that most invasions can be expected to start within a decade or two of first planting in disturbed habitats that are in the vicinity of appropriate natural or semi-natural vegetation. Based on field observations (Table 1), the woody invaders require 2–10 years to reach reproductive maturity in Hawai'i, while the herbaceous invaders require 1–3 years; therefore, all of the invaders began to spread within the first few years after reaching maturity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Cecropia obtusifolia (A) and Falcataria moluccana (B) are examples of early plantings in the Manoa Valley that became invasive.

Natural or semi-natural vegetation is visible in the background of both photos, less than 500 m from these plantings. Photos: E. Caum, 31 December 1922.

https://doi.org/10.1371/journal.pone.0004462.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Frequency distribution for time lags between first planting and start of spread in Manoa Valley (Honolulu, Hawai'i) for plant species that became invasive.

Records extend for 89 years.

https://doi.org/10.1371/journal.pone.0004462.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Time lag between first planting and start of invasion in Manoa Valley (Honolulu, HI) for plants that eventually became recognized as invasive pests.

https://doi.org/10.1371/journal.pone.0004462.t001

Discussion

Previously reported lag times [5], [6] were an order of magnitude greater than those reported here. Findings reported here are from a tropical environment, while most prior estimates of time lags have come from temperate environments. It is possible that lag times are shorter in tropical environments where plants can grow year round, and where narrower temperature extremes impose fewer restrictions on establishment. Kowarik [5] noted that many of the species with long lag times in Europe were adapted to warmer climates and their recent spread may have been associated with warming of parts of Europe. In contrast, temperature is unlikely to restrict spread of most tropical invaders. Differences in methodology and circumstances could also explain the shorter lag times reported here. In this study, the initial planting occurred within disturbed but naturally recovering vegetation, in the immediate vicinity of natural or semi-natural environments. In contrast, when date of first introduction is recorded in gardens or urban areas, these locations may be isolated from appropriate natural habitats, hindering spread into natural areas and resulting in long lag times.

A third consideration is that previous estimates of lag times may have been inflated because they were not determined by direct or systematic observations. Naturalization dates determined from herbarium specimens often over-estimate time to naturalization because herbarium specimens are usually not collected in a systematic or regular fashion. Dates of introduction are subject to similar errors, but in some cases they are known more precisely. Analyses of planting lists, floras and herbarium records suggest that the likelihood of detection outside of cultivation, naturalization, and/or area of spread of introduced species increase with residence time [10] and planting frequency [11]. This general pattern is true by necessity, since it is impossible for naturalization or spread to occur instantaneously due to the time required to reach reproductive maturity. But such a pattern is also to be expected due to the nature of herbarium collections: species that have been in residence for a longer period of time, or that have been planted at a greater frequency (more points of naturalization) are more likely to be collected as herbarium specimens. These facts complicate estimates of time lags based on herbarium specimens. It is rare to find studies recording direct field observations of invaders during the early naturalization process, but when such observations have been made, lag times between introductions and the start of spread of an invader have often been short [12]–[14]. Similarly, in the present study, deliberate and frequent searches were made to document spread, and much shorter lag times for plant invasions are evident.

The lag time at the start on an invasion is sometimes defined in mathematical terms as the time required for an invader to reach exponential growth or spread after the first planting or spontaneous occurrence in an area [15]. To determine the lag time using this approach requires analysis of records over many time intervals [15], [16], whereas the present study focused simply on the start of spread for species that became invasive pests. Defining lag time as the transition time to exponential growth is theoretically appealing; however, this lag time is difficult to measure because the exponential growth curve is expected to be exceedingly shallow at the start of exponential growth or spread [17, p. 178], and conclusions are likely to be sensitive to statistical methodologies and assumptions about observation intensity over time. In one of the only experimental studies to examine lag times in this way, Memmott et al. [18] released insect populations of different sizes and deliberately monitored their growth and spread over time. They found that although the smaller populations took longer to reach a given size (an apparent “lag”), the growth of all populations was entirely consistent with exponential growth. There were no unexplained lags. Additional detailed studies of this type are certainly needed. In plants, which commonly require several years to reach reproductive age, heterogeneous rates of exponential expansion are expected at least during the first few generations after arrival [9], so it is important to start with an appropriate null model for exponential growth before seeking alternate explanations for apparent lags.

Although in some cases long lag times are possible due to delayed introduction of a required mutualist [19], change in climate [20], or imposition of a new disturbance regime [21], the results reported here indicate that in all cases, species that became invasive pests began spreading within a few years after the initial plantings reached maturity in disturbed but naturally recovering vegetation. This finding, although from a single tropical region and involving a limited set of species, suggests that carefully controlled field trials [22]–[24] or “in ground” evaluation [25] could become a practical element of risk assessment for plant invaders. In Hawai'i, many invaders could have been identified prior to becoming pests, through their spread within a few years after maturation. A field trial approach is not a substitute for pre-entry risk assessment [26], [27]; rather, it could be most useful and feasible for potentially high-value plant introductions when conventional pre-entry risk assessments are indecisive.

Botanic gardens of all types have great potential to contribute to our understanding of plant invasions, as recently shown by Dawson et al. [28]. Botanical garden records provide a valuable resource documenting planting dates and locations, while observations of spread and plant limitations can be easily made for large numbers of plant species. Closer collaboration between ecologists and botanical garden staff would likely result in new insights into invasion ecology while also yielding new approaches and successes in plant conservation.

Materials and Methods

Acknowledgments

I thank Ray Baker for introducing me to historical records relating to plantings at the Harold L. Lyon Arboretum and for discussions of plant behavior at the Arboretum. Ray Baker, Petr Pyšek, Marcel Rejmánek and Dennis Hansen made helpful comments on an earlier version of this manuscript.