Summary: We overview an array of common garden genetics trials established across multiple sites and embedded in restoration plantings in the harsh, modified Midlands region of the island of Tasmania, Australia. The steps in establishing the common gardens, the diversity of experimental designs employed and their objectives are described. These trials provide a long‐term framework to test species choices and provenancing strategies, as well as enhance our understanding of adaptive genetic variation within key eucalypt species used for ecological restoration. Early results signal genetic‐based differences in early‐age performance, and we showcase the multiple benefits of integrating research infrastructure into restoration plantings.

Summary: There is an ongoing search for cost‐effective, practical and successful methods for landscape‐scale restoration required to meet the challenges posed by the United Nations Decade on Restoration 2021–2013. Mechanised cultivation techniques (rip mounding and Wilco spot cultivation), a range of revegetation strategies (dense reafforestation, riparian corridors, vegetation islets and scattered tree plantings) and best horticultural practice were employed to successfully establish extensive areas of local native trees and shrubs in highly altered agricultural landscapes within the Midlands of Tasmania, Australia. This region has been intensely farmed for the last 200 years and is characterised by past failures in native tree plantings. Between 2009 and 2018, 1800 ha were revegetated with a suite of hardy local native species to enhance, buffer and connect native vegetation remnants. These techniques were applied at a landscape scale to create biodiversity corridors that crossed the Midlands. Riparian corridors 100–200 m wide were created along 21 km of riverbank at an average density of 440 stems/ha. Fenced vegetation islets and individually caged scattered trees were established within and surrounding native vegetation remnants, at an average density of 143 stems/ha. Whole paddocks were densely reafforested at up to 800 stems/ha. Despite this, re‐establishing structurally complex and species diverse native vegetation, which can provide habitat for local native animals, proved challenging. In areas highly altered by a long history of farming, repeated interventions and long‐term relationships with landholders will be needed to achieve restoration objectives.

Summary: At the dawn of the Anthropocene, with the imminent threat of climate change delivering 3–4°C rise in temperature by the end of the century and biodiversity loss across the globe, restoration projects need to focus on re‐establishing connectivity in vegetation structure at a landscape scale to facilitate the movement of wildlife. To achieve this requires long‐term commitment, robust partnerships and planning and excellent planting technology underpinned by research. In this Special Issue, consisting of 15 papers, we present a multidisciplinary, multi‐institutional, science‐based approach to environmental restoration, focused on a single geographic region, the Midlands of Tasmania. This introductory paper describes the breadth of the material covered in the series and sets the scene for following papers by describing the Midlands, its geography, climate and history, its extraordinary natural values as a biodiversity hotspot, the degree of degradation that has resulted from 200 years of intensive farming and the objectives of the Midlands restoration project. The Midlands also offers opportunities as a model system for landscape scale restoration given it is a circumscribed region, heterogeneous in land forms and land uses. Furthermore, in land management, there is a high level of cooperation between land owners, government departments, environmental agencies and university researchers. We describe how the contributions from a wide range of disciplines can be focused to meet the challenges of ecological restoration in highly altered agricultural landscapes.

Research highlights: We present evidence indicating that covariation of functional traits among populations of a forest tree is not due to genetic constraints, but rather selective covariance arising from local adaptation to different facets of the climate, namely rainfall and temperature. Background and Aims: Traits frequently covary among natural populations. Such covariation can be caused by pleiotropy and/or linkage disequilibrium, but also may arise when the traits are genetically independent as a direct consequence of natural selection, drift, mutation and/or gene flow. Of particular interest are cases of selective covariance, where natural selection directly generates among-population covariance in a set of genetically independent traits. We here studied the causes of population-level covariation in two key traits in the Australian tree Eucalyptus pauciflora. Materials and Methods: We studied covariation in seedling lignotuber size and vegetative juvenility using 37 populations sampled from throughout the geographic and ecological ranges of E. pauciflora on the island of Tasmania. We integrated evidence from multiple sources: (i) comparison of patterns of trait covariation within and among populations; (ii) climate-trait modelling using machine-learning algorithms; and (iii) selection analysis linking trait variation to field growth in an arid environment. Results: We showed strong covariation among populations compared with the weak genetic correlation within populations for the focal traits. Population differentiation in these genetically independent traits was correlated with different home-site climate variables (lignotuber size with temperature; vegetative juvenility with rainfall), which spatially covaried. The role of selection in shaping the population differentiation in lignotuber size was supported by its relationship with fitness measured in the field. Conclusions: Our study highlights the multi-trait nature of adaptation likely to occur as tree species respond to spatial and temporal changes in climate.

Habitat fragmentation is a key factor causing variation in important mating system parameters in plants, but its effect is variable. We studied mating system variation among 276 native trees from 37 populations of Eucalyptus pauciflora from Tasmania. We assayed 10 microsatellite loci from 1359 open-pollinated progeny from these trees. Across Tasmania the species’ mating system was characterised by a high outcrossing rate (tm = 0.90) but moderate bi-parental inbreeding (tm–ts = 0.16) and moderate correlated paternity (rP = 0.20) in comparison to other eucalypt species. Despite significant differences in outcrossing rate and correlated paternity among populations, this variation was not correlated with fragmentation. Nevertheless, fragmentation was inversely correlated with the number of germinants per gram of seed capsule content. Outcrossing rate had been reported previously to decrease with increasing altitude in mainland populations of E. pauciflora, but this was not the case in Tasmania. However, a small but significant decrease in correlated paternity occurred with increasing altitude and a decrease in bi-parental inbreeding with increasing altitude was evident in fragmented populations only. It is argued that strong, but incomplete self-incompatibility mechanisms may buffer the mating system from changes in population density and pollinators. While seed yields from highly fragmented populations were reduced, in most cases the seed obtained is unlikely to be more inbred than that from non-fragmented populations and, thus, is likely to be as suitable for use in local forest restoration.