1. Globally, invasive non-native plants are an increasing threat to indigenous biodiversity and ecosystems, but management can be compromised by poor efficacy of control methods, harmful non-target effects or secondary invasions by other non-native plant species.

2. A 5-year field trial compared two stakeholder-selected control methods for heather, a European plant invading native ecosystems in and adjoining Tongariro National Park in New Zealand. The control methods were a selective herbicide (Pasture Kleen®; 2,4-D ester) and biocontrol with an introduced beetle Lochmaea suturalis (Coleoptera: Chrysomelidae).

3. Biocontrol reduced mean heather cover by 97%, slightly more than herbicide at 87%, compared with a 20% increase in heather under no management.

4. Cover of native dicots, the most species-rich plant group, increased following biocontrol. In contrast, herbicide application had major non-target effects on native dicots, reducing their percentage cover and species richness. Native monocot cover and species richness increased following both herbicide and biocontrol treatments.

5. A similar 8-fold increase in non-native monocots occurred following both biocontrol and herbicide treatments. Overall, secondary invasion was greatest with biocontrol because non-native dicot cover also increased, whereas herbicide almost eliminated non-native dicots. 6. Synthesis and applications. Biocontrol and herbicide treatments both controlled heather but herbicide application was associated with severe non-target impacts on native dicots. Benefits to the native flora were consequently greatest in the biocontrol treatment, despite greater secondary invasion. Control strategies for management of widespread non-native plants to optimize ecosystem outcomes should include more consideration of biocontrol.

This dataset if from 5-year field trial compared two stakeholder-selected control methods for heather Calluna vugarus, a European plant invading native ecosystems in and adjoining Tongariro National Park in New Zealand. The control methods were a selective herbicide (Pasture Kleen®; 2,4-D ester) and biocontrol with an introduced beetle Lochmaea suturalis (Coleoptera: Chrysomelidae). Data include percentage cover of heather and competing plant species and species richness within the study plots

Twenty-four 5 × 5 m plots (six blocks of four plots) were located around the periphery of a developing heather beetle outbreak at 39° 21’ 57.3” S, 175° 42’ 55.1” E during November 2007. In each block one plot was randomly assigned to one of four treatments; 1. Control (insecticide spray to protect vegetation from beetle feeding), 2. Biocontrol (to expose heather to beetle feeding only), 3. Herbicide (herbicide + insecticide to protect vegetation from beetle feeding but expose it to herbicide), and 4. Biocontrol + herbicide (to expose vegetation to beetle feeding and herbicide).

The herbicide used was Pasture Kleen® @ 6.5ml/L applied in December 2007 and 2008 to match the method being employed by the NZ Defence Force (NZDF) within the WMTA. The insecticide used was a synthetic pyrethroid Karate Zeon @ 1ml/15L + 0.3ml/L Vapor Guard, which was found to successfully eliminate beetles from treated plots. A concurrent insecticide check experiment at a locality where heather beetle was absent indicated that insecticide did not directly affect plant cover or species richness (Appendix 1 SI). Insecticide applications were made in 2007 (November), 2008 (January, February, March, September, October (twice), December), 2009 (January, February, September), 2010 (October), and 2011 (October). The frequency was reduced over time because, after the feeding front of heather beetles had moved well past the plots, there was less re-invasion of the plots by beetles. Herbicide and insecticide were applied separately, mostly on different dates, to a 7 × 7 m area to avoid edge effects, with a minimum buffer of 3 m to the edge of the next 7 × 7 m treatment area. On the one occasion when herbicide and insecticide were applied on the same date, the first application was allowed to dry before the second application. Both herbicide and insecticide were applied to run-off. Plots not treated with herbicide or insecticide were sprayed with water to run-off on the same dates the chemicals were applied. The insecticide treatment was found to be highly effective at reducing defoliation of heather (Appendix 2 SI).

Absolute percentage of cover all vascular and non-vascular (bryophytes, clubmosses and lichens) plant species was assessed visually in four 50 × 50 cm subplots, one in each plot corner, between the 1^st and 12^th February in each of 2008, 2009, 2010, and 2012.

We conducted repeated measures ANOVAs on the data. Percentage cover ("Raw cover data" worksheet) was converted to proportions and analysed for heather, clubmosses, lichens, and bryophytes, and cover and species richness were analysed for summed groups of native dicots, native monocots, non-native dicots (excluding heather) and non-native monocots. To avoid pseudoreplication, the average cover across the four quadrats (se, ne, nw, sw)  per plot in the "Raw cover data" was analysed and the total number of plant species in all four quadrats was summed in the species richness analysis.