INTRODUCTION

Tropical rainforests represent a disproportionate and dynamic component of the world’s carbon budget (Mitchard 2018, Friedlingstein et al. 2022), contain a majority of Earth’s terrestrial biodiversity (Barlow et al. 2007, Gibson et al. 2011), and provide a diverse array of ecosystem services (Watson et al. 2018). Despite their significance, these ecosystems are under severe anthropogenic pressure, with an estimated 17% reduction in moist tropical forest extent from 1990 to 2019 (Vancutsem et al. 2021). At the same time, active restoration of terrestrial ecosystems is essential for addressing global challenges such as climate change, biodiversity decline, and land degradation (Girardin et al. 2021). The latest IPCC report (Riahi et al. 2022), highlights afforestation, reforestation, and revegetation perhaps in addition to improved forest management, agroforestry, and soil carbon sequestration as the primary CO2 removal methods currently in use. The United Nations (UN) declared 2021-2030 the Decade of Ecosystem Restoration (United Nations General Assembly 2019). However, the increasing frequency of hot, dry periods is exacerbating global tree mortality (Hammond et al. 2022) and is predicted to increase the frequency of lethal canopy temperatures being reached in tropical forests (Doughty et al. 2023), posing a significant threat to the success of restoration efforts.
In response to climate change, plants can migrate, adapt, or become extinct (Corlett & Westcott 2013, Merila & Hendry 2014). However, the rapid rate of anthropogenic climate change and the shallow latitudinal temperature gradient in the tropics mean that migration is largely restricted to upward shifts in elevation (Colwell & Feeley 2024). Additionally, the long generation times of forest tree species limit their ability to adapt quickly, making intraspecific phenotypic variation – both genetic variation accumulated over generations and phenotypic plasticity – crucial for adaptation (Merila & Hendry 2014). When variation exists for divergent selection to act on, local adaptation can result in local populations with a fitness advantage over nonlocal populations in their home site (Brancalion et al. 2018, Muehleisen et al. 2020). Understanding and leveraging intraspecific variation is therefore essential for assessing and enhancing the success of restoration plantings in the face of rapid climate change.
One strategy to ensure that restoration plantings are successful in establishing resilient forests is the appropriate choice of seed sources. Restoration guidelines provide advice on how to maintain the genetic diversity and adaptive capacity of restored forest populations through the consideration of the geographic origin of seed, i.e. provenance (Offord & Meagher 2009, McDonald et al. 2016, Malavasi et al. 2018). Although a strategy of selecting local provenances has traditionally been preferred by practitioners (Cooperet al. 2018), this can reduce fitness in cases when forest fragmentation results in inbreeding (Schlaepfer et al. 2018) or when rapid climate change causes a mismatch between the current or future climate with the historic conditions a population is adapted (Aitken & Whitlock 2013, Gellie et al. 2016). These issues have led to the recommendation of alternative provenancing strategies incorporating non-local seed sources (Breed et al. 2013, Jordanet al. 2024). Given the strong role climate can have in driving selection (Steane et al. 2014, Cordero et al. 2021, Ravnet al. 2024), the matching of seed sources to future climates (known as predictive or climate-adjusted provenancing) aims to introduce to an area genetic material that is better adapted and therefore also more resilient to climate change (Breed et al. 2013).
Despite the known advantages that strategies such as climate-matching provenance can have for restoration outcomes, the current industry default in rainforest restoration plantings is to source seed locally (Cooper et al. 2018). A key challenge in adopting alternative provenance selection in tropical rainforests is the extensive genetic information required to recommend a suitable strategy. Evidence for local adaptation requires direct evidence of local populations having fitness advantages over non-local populations, with full life-cycle analysis under reciprocal common garden conditions being considered the gold standard for determining true local adaptation (Kawecki & Ebert 2004, Meek et al. 2022, Schwinning et al. 2022). Synthesis studies to date are heavily biased toward temperate species of commercial importance with very few, if any, tropical trees included (Hereford 2009, Matesanz & Ramirez-Valiente 2019). Their findings also reveal how the presence and strength of local adaptation are species and context-dependent. This limits our understanding of the prevalence and strength of local adaptation in tropical rainforest trees, which typically have lower population densities than temperate species, but high reproductive efficiency over large distances maintaining similar levels of genetic diversity (Degen & Sebbenn 2015). However, provenance trials to test for local adaptation are highly resource-intensive (Sorket al. 2013) and typically focus on a single species. This has limited the broad applicability of climate-matching provenance strategies in tropical systems where forests are extremely speciose and understanding the genetic background of each species is not feasible.
While a ‘one-size fits all’ approach to provenancing is generally not recommended or ideal, the combination of extremely high species diversity in tropical rainforests, and the limited resources available for restoration activities necessitates a simplified, easy-to-implement provenancing strategy. One possibility could be grouping the strategy according to species plant functional traits, such as wood density, that describe the slow-fast life-history continuum (Reich 2014). Faster-growing species such as pioneers have shorter lifespans and earlier reproductive capacity than slower-growing species such as mature-phase species. As such, early successional stage species may show higher rates of molecular evolution and therefore be more likely to have developed local adaptation than slower-growing species (Smith & Donoghue 2008, Smith & Beaulieu 2009, Müller & Albach 2010). However, with studies of local adaptation severely lacking for tropical rainforest tree species, there is currently insufficient experimental evidence to support this. Additionally, while meta-analyses are useful to discern the patterns and drivers of local adaptation in studies across multiple species, these studies often vary in methodology, site maintenance, and conditions that are not easily accounted for. There is therefore a need for multi-species studies of local adaptation in tropical trees that are key to rainforest restoration.
We investigated the prevalence of local adaptation in tropical rainforest tree species commonly used in restoration plantings in the Australian Wet Tropics. To do so we established three field-based provenance trials, including an upland and lowland site that differ in climate but have similar high-nutrient soils, and an additional lowland site on a low-nutrient soil. Saplings of 16 species were sourced from lowland and upland populations and reciprocally planted at each site, where initial growth and survival were monitored over 1.5 years. Specifically, we tested the following questions:
  1. How do provenance, site, and provenance × site interactions affect sapling growth and survival?
  2. If provenance × site interactions exist, do local provenances outperform non-local provenances at their local site?
  3. If there is variation across species in their provenance differentiation, is this more pronounced in faster-growing species?