Introduction

Marine ecosystems worldwide are experiencing dramatic shifts in environmental conditions due to climate change, the most evident of which is a steady increase in sea surface temperature (SST) (Cheunget al. 2013). These changes can affect marine organisms in different ways, such as by altering the structure of trophic webs (e.g., Hyndes et al. 2016), biasing sex ratios in species with temperature-dependent sex determination (e.g., Miyoshi et al.2020), and redefining the geographical ranges of species (e.g., Pinskyet al. 2020). In order to guide natural resource management under this changing marine landscape, it is crucial to make future predictions of suitable habitat for target species as accurately as possible.
Species distribution models (SDMs), which estimate relationships between species’ presence data and environmental predictors, have been used extensively to predict potential changes in species’ distributions under climate change scenarios (Guisan et al. 2017). The majority of SDMs are constructed at the species-level or even higher taxonomic levels, and this is particularly true for applications to marine species (Robinsonet al. 2011; Robinson et al. 2017; Chefaoui et al.2018; Jayathilake & Costello 2018; Melo-Merino et al. 2020). One fundamental and critical assumption underlying species-level SDMs is niche conservatism, which assumes that all populations of a species have analogous environmental requirements and respond in a similar way to a changing environment (Guisan et al. 2017; Smith et al.2019). But this assumption ignores intra-specific variation, in particular local adaptation and phenotypic plasticity, which are frequently observed especially in broadly distributed taxa (e.g., Marín-Guirao et al. 2016; Duarte et al. 2018; King et al.2018; Benito Garzón et al. 2019; Peterson et al. 2019; Zhang et al. 2020b).
SDMs constructed with data for lineages below the species level can account for possible local adaptations and therefore can provide more reliable niche estimations and habitat suitability projections for species with intraspecific variation. For instance, a species-level SDM for the threatened Japanese crayfish Cambaroides japonicus (De Haan 1841) predicted that this species might lose a large proportion of its suitable habitat in the future, whereas lineage-level SDMs for the same species predicted a weaker impact of climate change overall (Zhanget al. 2021). The importance of taxonomic units (i.e., above and below the species level) in distribution modelling has recently been recognized (Benito Garzón et al. 2019; Peterson et al.2019; Smith et al. 2019; Collart et al. 2021), which has resulted in more SDM applications for terrestrial and freshwater species that consider intra-specific variation (e.g., Ikeda et al. 2017; Razgour et al. 2019; Zhang et al. 2021). Conversely, relatively few SDM studies have investigated this issue in the marine realm (but see Assis et al. 2018a; Cacciapaglia & van Woesik 2018; Lowen et al. 2019).
Seagrasses are one of the most critical habitat engineers (along with seaweeds, mangroves, and coral reefs) of tropical coastal marine environments. They not only harbor rich marine biodiversity in seagrass meadows, but also provide a number of ecosystem services, such as primary productivity, habitat restoration, resources for marine life, and human recreation, among others (Unsworth et al. 2018). Maintaining these services is key to achieving conservation and economic goals under global change. Yet, seagrass ecosystems are declining worldwide at an annual rate of 7% due to multiple natural and human-mediated disturbances (Orth et al. 2006; Waycott et al. 2009). It is noteworthy that climate change has received considerable attention as a major factor for the increasing loss of seagrass meadows (Jordà et al. 2012; Thomson et al. 2015; Repolho et al. 2017; Duarte et al. 2018; Smale et al. 2019). This is particularly true for the tropical Indo-Pacific bioregion, which supports the most seagrass diversity and a high diversity of associated flora and fauna (Short et al. 2007) but has suffered from striking degradation of seagrass coverage (Coleset al. 2011; Rasheed & Unsworth 2011; Grech et al. 2012; Chefaoui et al. 2018; Olsen et al. 2018; Brodie et al. 2020). Given the global ecological roles of seagrasses, it is crucial to make accurate forecasts of their distribution patterns in the face of climate change, but seagrasses are “among the least-studied groups” (Melo-Merino et al. 2020) with respect to range shift projections. The majority (if not all) of SDM studies on seagrasses have been at the species level and therefore did not incorporate potential intraspecific variation.
The seagrass Thalassia hemprichii (Ehrenberg) Ascherson (Hydrocharitaceae) is a perennial climax species that is widely distributed in the tropical Indo-Pacific bioregion (Green & Short 2003), extending from Australia, the peripheral limit of its eastern range (Hernawan et al. 2017), to East Africa in the West Indian Ocean (Jahnke et al. 2019a). It reproduces sexually via seeds and asexually via vegetative growth of rhizomes. Uprooted adult plants can potentially float for months and hence colonize distant areas (Wuet al. 2016). In addition, this seagrass forms buoyant seeds that remain afloat for long enough to disperse a few hundreds of kilometers (Lacap et al. 2002). A recent survey revealed that seedlings can also disperse for over a month due to the accumulation of oxygen in the body tissue (Wu et al. 2016). Thus, T. hemprichii has excellent long-distance dispersal potential that may play a significant role in shaping population genetic structure (Lowe & Allendorf 2010). This species may be particularly vulnerable to climate change because it exhibits spatial separation of the sexes (dioecious), reinforced by physiological and morphological differentiation of each sex to variable microhabitats (Hultine et al. 2016). Recent genetic studies ofT. hemprichii detected genetic lineage divisions in the East and West Indo-Pacific Ocean (Hernawan et al. 2017; Jahnke et al. 2019a), but we still do not have a clear understanding of the distribution of lineages across the entire tropical Indo-Pacific region, or whether these diverged lineages are expected to respond differentially to climate change.
In the present study, we used T. hemprichii as a model to: (i) examine range-wide divergence of genetic lineages in the tropical Indo-Pacific Ocean; (ii) test if phylogeographical lineages exist, and if so, quantify niche differentiation between distinct lineages; (iii) predict climate change impacts on the species’ range with species-level and lineage-level SDMs. By incorporating potential intra-specific variation, our SDMs can provide more realistic predictions on how climate change will shift future distributions of a habitat-forming seagrass, thus generating valuable knowledge for guiding the long-term management of this species in the tropical Indo-Pacific coast.