I found one adult E. paralias plant led to the localised establishment of 213 plants at NZ’s southern-most incursion (Scotts Beach) and 484 plants at the northern-most (Karekare) incursion (Table 1). The plant populations were extremely healthy with 96% of plants at both sites having < 5% vegetation damage. The majority of plants were seedlings < 4cm tall (Fig. 1). At Scotts beach seedlings tended to be closer to the adult plant, whereas at Karekare seedlings were distributed along the dispersal gradient (Fig. 2). Although the date of establishment is unknown, the results indicate that these populations are relatively new, healthy, and without control would rapidly expand at both sites. These findings also indicate E. paralias would become a major pest along the majority of NZ’s coastlines, and strongly support it’s Unwanted Organism status and the requirement for its continued eradication. For other countries with coastal areas that have similar climatic envelopes as NZ and South-Eastern Australia, eradication of this plant should be considered if populations are discovered.
Plants established on average further from the adult plant at Karekare than at Scotts Beach (Table 1, t = -9.445, df = 288, p-value = < 0.001). The percentiles of effective dispersal distance were also all further at Karekare than Scotts Beach (Table 1). Plant distributions around the adult plant also varied between sites (Fig. 3). Approximately three quarters of established plants occurred inland from the adult at Scotts beach, while at Karekare the distribution was more uniform around the adult (Fig. 3). I expect that the arrangement of the habitat at each site is the main driver of this result. At Scotts Beach the offspring of the adult plant were restricted in the habitat they could establish, due to the adult establishing in a relatively narrow surge zone. The bush edge, where shade would inhibit establishment, was approximately 8.5m inland from the adult plant, while the high tide mark on the seaward side was 13m away. Seed may have been washed away on the seaward side of the adult plant at Scotts Beach, but a survey 15km either side of the Scotts Beach infestation found no other populations. Yang et al. (2012), in a study of seed dispersal of coastal species found that 90% of their buoyant twig segments (used to mimic seeds) returned to the same section of coast where they were released, with the majority of those settling tens of metres from the release point. At Scotts Beach seed may have been washed inland by sea surges rather than washed out to sea, resulting in the observed fan shaped distribution (Fig. 3). At Karekare, the more uniform distribution of seeds may be due to the adult establishing in the back dune of a large dune system away from the storm surge zone, and with potential habitat for > 80m in all directions (Fig. 2).
The maximum effective dispersal distance of E. paralias (15.8m; Table 1) was 3.6 times further than the maximum dispersal distance predicted by the trait based dispeRsal model (Tamme et al. 2014) for ballochory for herbaceous species (4.4m) but approximately half the predicted distance if E. paralias is considered a shrub (27.8m). The 50th and 95th percentiles for both sites (Fig. 2; Table 1) were greater than those predicted in Bullock et al. (2017) for ballistic herbaceous species (< 1m in height and < 10mg seed size; 50th = 1.063m, 95th = 4.926) or wind dispersed herbaceous species (0.1-0.8m in height; 50th = 0.129m, 95th = 0.679m). The greater dispersal distance I observed compared to those predicted by the majority of dispersal models is probably due to the effective dispersal distances being impacted by secondary dispersal of seeds by wind. At Karekare, I observed seeds easily rolling along the top of the sand with a wind gust, dispersing seeds further than they would have initially been dispersed ballistically. The results support recent calls for developing dispersal kernels that consider the total dispersal kernel where dispersal by multiple vectors are considered (Rogers et al. 2019). This is especially important when managing invasive species where underestimation of the dispersal kernel by looking at a singular vector could lead to underestimation of spread and missed individuals or populations during surveillance monitoring.
A strong surveillance programme and management of dispersal and establishment pathways is needed to keep coastlines free from E. paralias. For Australia and New Zealand, targeted control of seed source populations where seed has the highest likelihood of reaching NZ or Australia’s more northern coastlines will reduce the probability of long-distance dispersal events occurring. Identification of seed source populations and high likelihood establishment sites is a critical avenue for future research. Such research will also help with the location and timing for a national surveillance programme in NZ. Development of multiple tools such as remote sensing and weed detection dogs to increase the probability of detecting plants and to delimit incursions would be valuable (Cherry et al. 2016). Future surveillance monitoring for E. paralias should consider the wind patterns at sites when searching for new individuals or populations.
To minimise secondary dispersal by wind or ocean currents removal of the seedbank could be considered at incursion sites. These results provide quick, realistic boundaries for delimiting the potential seedbank, e.g. 90% of presumed offspring were within 5.5m of the adult plant at both sites (Table 1). More in-depth analysis of the seedbank through soil sampling may provide better results but are more labour intensive. Development of low cost, effective methods for removal or treatment of the seedbank would be valuable for future eradication efforts.
In conclusion, E. paralias is a highly invasive and transformative weed that without control can become a serious threat for temperate coastal ecosystems. The establishment of a single adult can quickly lead to the localised establishment of > 450 healthy plants. The land-based maximum dispersal distance for E. paralias is larger than would be predicted by trait-based models alone, however the land-based maximum dispersal distance still appears to be relatively small (15.8m). Early detection of E. paralias populations through a comprehensive surveillance program and removal of any new populations (seed and establishing plants) will be important for any region or country attempting zero-density or eradication of E. paralias.