This thesis developed various methods to quantify and better understand paropsine (Coleoptera: Chrysomelidae) defoliation within Eucalyptus, with the overarching goal of reducing herbivore damage in plantation forests. To date, paropsine preferences and impacts on Eucalyptus in New Zealand remain poorly understood with only a few published studies (Lin et al., 2017a; Radics et al., 2018; T. Withers et al., 2017). Furthermore, defoliation was previously measured with visual ground-based assessment; as a consequence, quantifying defoliation is semi-subjective. Research needs to be undertaken to fill these knowledge gaps. This thesis expands on the work of Lin (2017) and includes E. bosistoana families and other Eucalyptus species identified as priority species by the New Zeland Dryland Forest Initiative (NZDFI). This thesis is composed of five chapters, which seek to improve our knowledge of Eucalyptus tolerance and resistance to paropsine defoliation. The first chapter is a review of previous literature, followed by four separate, but interconnected chapters detailing experimental results, which are summarised in a final concluding chapter.

In chapter one, my literature review identifies knowledge gaps, especially with regard to Eucalyptus tolerance and resistance (Lin (2017a) already reviewed differences in Eucalyptus species resistance in the North Island) to paropsine beetles in the South Island, what drives paropsine beetle preferences, and the way that defoliation is currently assessed in the field. These gaps led me to develop four main research questions, each of which was explored by one of the four subsequent chapters. My research questions were:

1. Are there Eucalyptus species, Eucalyptus bosistoana families, clones or even provenances that are more resistant or tolerant to paropsine defoliation?
2. Are there Eucalyptus species more tolerant to controlled artificial defoliation at different water deficit?
3. Is there a relationship between Eucalyptus foliar chemistry and paropsine defoliation?
4. Can remote sensing be used to reliably quantify Eucalyptus paropsine defoliation?

In chapter two, seven Eucalyptus species from two subgenera (Symphyomyrtus subg.: E. bosistoana, E. camaldulensis, E. tricarpa, E. quadrangulata, and E. cladocalyx; Eucalyptus subg.: E. globoidea and E. macrorhyncha), 74 Eucalyptus bosistoana families and 132 Eucalyptus bosistoana genotypes (clones) were measured in the Marlborough region between December 2019 and March 2021 to investigate variation in their resistance and tolerance to paropsine defoliation. Resistance was measured with the Crown Damage Index (CDI) and tolerance was measured with growth (height gain, DBH gain and new stem length) over time. Compared to the Symphyomyrtus subgenus, the Eucalyptus subgenus was generally more resistant to paropsine defoliation, with the exception of E. cladocalyx. Eucalyptus cladocalyx, E. macrorhyncha and E. globoidea were the most resistant to paropsine defoliation. Eucalyptus bosistoana’s most resistant family and genotype were family 805 and genotype 839a. Eucalyptus quadrangulata, E. bosistoana, E. tricarpa and E. camaldulensis were all potentially tolerant regarding the new stem growth, whereas E. cladocalyx, E. macrorhyncha and E. globoidea were the most tolerant regarding the height and DBH growth. The E. bosistoana family 835 and genotype 24a were potentially tolerant.

In chapter three, two Eucalyptus species (E. bosistoana and E. globoidea) were placed in a greenhouse for 101 days to test defoliation and water tolerance in a more controlled environment. Three artificial defoliation severity levels, two water levels and two defoliation frequency levels were applied. Tree growth (height and diameter gain), dry biomass (total, leaf, stem, root) as well as leaf carbon and nitrogen content were measured. Both species were negatively affected by water stress, meaning that planting these species in drylands will likely exacerbate the negative impact of paropsines. Nevertheless, E. globoidea was more resistant to water deficit and defoliation than E. bosistoana. Low levels of defoliation stimulated E. bosistoana biomass (overcompensation), but high levels of defoliation negatively impacted this species. E. globoidea may generate taller trees as a mechanism to tolerate defoliation, whereas E. bosistoana may produce trees with denser overall biomass (stem in particular) and narrower trunk. Compared to trees that had only been defoliated once, trees that had been defoliated twice grew less. This demonstrated that both species exhibit some elements of tolerance to defoliation, but E. globoidea is more tolerant than E. bosistoana.

In chapter four, leaves from seven Eucalyptus species from two subgenera (Symphyomyrtus subg.: E. bosistoana, E. camaldulensis, E. tricarpa, E. quadrangulata, and E. cladocalyx; Eucalyptus subg.: E. globoidea and E. macrorhyncha) were collected in a trial from the Marlborough region to study the potential relationship between foliar chemistry and paropsine defoliation. The leaves were then oven-dried, ground, extracted with ethanol and analysed with a High Performance Liquid Chromatography (HPLC). The Eucalyptus subgenus Symphyomyrtus and Eucalyptus foliar compound diversity were distinct. These preliminary experimental results suggest a relationship between foliar chemicals and paropsine defoliation. Defoliated species had higher foliar compound concentration and richness than non-defoliated species, except for E. camaldulensis, which displayed low foliar compound concentration and richness despite being significantly defoliated. Two compounds specific to E. cladocalyx may be paropsine repellent.

In chapter five, six Eucalyptus species from two subgenera (Symphyomyrtus subg.: E. bosistoana, E. camaldulensis, E. tricarpa, E. quadrangulata, and E. cladocalyx; Eucalyptus subg.: E. globoidea) were measured in a trial from the Canterbury region with three LiDAR sensors to assess the potential of this technology to accurately detect defoliation. LiDAR data were compared to Crown Damage Index (CDI) field measurements to assess its prediction accuracy. The 5 % accuracy difference among the three LiDAR sensors under evaluation indicated that each could be useful for predicting paropsine defoliation and show good promise for further experiments. The most effective LiDAR metrics for predicting paropsine defoliation on Eucalyptus trees were itot, zimean, imax, imean, and isd, with zkurt, zpcum2 to zpcum9, p1th to p3th, pzabovemean, pzabove2, and ipcumzq10 to 70 being less useful. All of these metrics are related to either tree height or foliage density (canopy cover).

My results contributed to a better understanding of tolerance and resistance to paropsine beetles within Eucalyptus. This contributes to limiting the effect of paropsines on the Eucalyptus forest industry. These findings are a valuable starting point to deeper exploration and guiding future Eucalyptus breeding in terms of paropsine herbivory.