Strong scale-dependent relationships between fine-root function and soil properties uncovered with spatially coupled sampling

Abstract

Introduction

Trait-based ecology often seeks to define organisms' responses to broad gradients, assuming that coarse-scale biotic and abiotic conditions are the dominant drivers of their distribution and functioning (Shipley et al., 2016). Fine-root trait research has historically operated under this assumption that coarse gradients drive function, leading to a frequent spatial mismatch between fine-root and environmental sampling. However, this practice may substantially hinder the identification of certain influences on fine roots, the smallest and most distal modular units of the root system (by one definition < 2 mm diameter (D); Freschet et al., 2021). Fine roots must balance multiple functions: localizing and acquiring water and all essential plant nutrients, competing with microbes and other plants, defending against pathogens and interfacing with belowground symbionts. This multifunctionality is overlaid on a highly heterogeneous soil environment. A single edaphic variable, whether it is a resource or physical characteristic, could be relatively homogeneously or heterogeneously distributed, and this heterogeneity can also occur in various patterns – gradients, patches or combinations of both – that manifest at various spatial scales (Fig. 1; Ettema & Wardle, 2002; Yavitt et al., 2009). Given the small size and modular nature of fine roots, and the immense heterogeneity of the soil environment they navigate, it is highly probable that fine roots plastically adjust their functioning in response to the soil environment they directly encounter (de Kroon et al., 2005). Aligning the spatial scales of fine-root and soil samples could thus expose the clearest fine-root functional responses to edaphic drivers, particularly those that occur at fine spatial scales.

Fig. 1

Open in figure viewerPowerPoint

Conceptual representation of heterogeneity in a soil property that extensive root systems could encounter. Soil properties can be relatively homogeneous, or heterogeneity can be patterned as gradients, patches or combinations of both (shown here) at various spatial scales. Colors represent the relative values of a single soil resource or physical characteristic from low (light) to high (dark).

Individual edaphic variables often have independent degrees and spatial patterns of heterogeneity, forming a mosaic across the landscape (Fig. 1; Ettema & Wardle, 2002; Yavitt et al., 2009). Overlaid, multiple variables thus form a multidimensional soil mosaic. For soil nutrients, the source of inputs to ecosystems is likely to structure their spatial distributions. Beyond the spatial structuring that may result from the source, distributions could be restructured by the effects of microtopography, plant nutrient cycling, decomposition or animal activity (Roy & Singh, 1994; Waring et al., 2015; Osborne et al., 2017). Phosphorus (P) and exchangeable base cations, including magnesium (Mg), potassium (K) and calcium (Ca), primarily enter ecosystems from the weathering of parent materials. The type and age of parent materials and the climatic conditions that influence weathering rates thus tend to drive the quantities and patterns of these resources (Walker & Syers, 1976; Kaspari & Powers, 2016; Waring et al., 2021). For example, as soils age, total P and plant-available inorganic orthophosphate (PO₄) decline as rocks are heavily weathered and P is transformed into less-accessible forms or leached (Walker & Syers, 1976). As soils age, total nitrogen (N) increases (Walker & Syers, 1976). Nitrogen can also be weathered from parent materials, but the primary natural input into ecosystems is biological N fixation, which tends to structure heterogeneity at finer spatial scales resulting from plant activity and litter and microbial ‘hotspots’ of activity (Vitousek et al., 2013; Waring et al., 2015, 2016, 2021; Akana et al., 2023).

Due to the broad relationships between ecosystem age and total N and P, the tropics are expected to be primarily limited by P (Vitousek et al., 2010; Dalling et al., 2016; Cunha et al., 2022). However, readily available plant inorganic N – extractable nitrate (NO₃) and ammonium (NH₄) – could also shape plant responses; N can be highly variable in space and has been shown to influence tropical plant fine-root biomass, morphology and arbuscular mycorrhizal colonization (Wurzburger & Wright, 2015; Waring et al., 2016). In a broader sense, due to their multifunctionality, it is probable that fine roots respond to multiple soil variables at once. This is perhaps especially likely in the tropics, where nutrient colimitation may be more common due to the long-term weathering of all rock-derived nutrients (Kaspari & Powers, 2016).

Despite the complexity of the soil environment, fine-root studies have tended to sample roots at finer spatial resolutions than soils. For example, a study along a 1000-m elevational gradient in the French Alps sampled fine roots in five 400 m² plots per elevation separated by 300–2000 m of distance, but soil cores collected from plots were pooled to each elevation; this spatial mismatch could have obscured or blurred some responses, especially considering that most intraspecific fine-root trait variation was seen within elevations rather than among them (Weemstra et al., 2021). Another study in British Columbia, Canada, found the majority of intraspecific fine-root trait variation occurred within individual trees, rather than across a c. 220-km latitudinal gradient (Defrenne et al., 2019). It is thus clear that fine roots can vary significantly at fine spatial scales, but the causes of this variation are not.

There is also evidence that certain soil properties can vary the most at fine spatial scales, while others predominantly vary at coarse ones. A N cycle study conducted in seasonally dry neotropical forests found ≥ 50% of variation for 13 of the 16 soil variables occurred within 300 m² subplots rather than among sites separated by 60 km, even though sites had functionally distinct plant communities (Waring et al., 2016). In a temperate forest, soil NO₃ and NH₄ varied substantially at spatial scales < 1 m, considerably smaller than the root system of juvenile trees (Akana et al., 2023). These trends of high soil heterogeneity at fine spatial scales coupled with high fine-root trait variation at fine biological (i.e. within individuals or within species) or spatial scales could be directly related. Hence, to accurately and consistently describe fine-root functional responses to edaphic gradients, it is likely that soil sampling resolution needs to closely match fine-root sampling resolution.

In addition to being multifunctional, fine-root traits are also multidimensional (Kramer-Walter et al., 2016; Bergmann et al., 2020; Carmona et al., 2021; Weigelt et al., 2021). Across species globally, some traits have been shown to express multiple trade-offs and coordinations, such that high specific root length (SRL) trades off with large D and high arbuscular mycorrhizal colonization intensity (AM%), and independently, high root tissue density (RTD) trades off with high N concentration (N%); these trade-offs are interpreted as a do-it-yourself vs outsourcing trade-off with respect to symbiotic dependence, and a conservation trade-off between fine roots with high longevity vs high metabolic activity (Bergmann et al., 2020). This multidimensionality appears to stem from fine roots' cylindrical shape, cortex allometry and the flexibility of cell wall thickness and number, which may permit immense trait diversity and plasticity (Zhang et al., 2024). Since these intrinsic relationships among fine-root traits could constrain their responses to the environment, they should be explicitly considered in research design.

Besides constraining potential fine-root responses, strong relationships among fine-root traits could allow for measurement proxies for harder-to-measure traits and be leveraged to statistically impute data for less commonly measured fine-root traits based on a wealth of existing morphological data (Iversen et al., 2021). Although physiological fine-root traits like root respiration rate (RESP) and potential acid phosphomonoesterase activity rate (PME) capture direct plant impacts on biogeochemical cycles of carbon, N and P, and directly reflect plant energetic costs and potential nutrient acquisition, measuring them in the field is complicated and time sensitive. As a result, fine-root data are biased toward traits such as D, SRL, RTD and N%. Measuring suites of fine-root traits on each fine-root sample could help to identify common traits as proxies for physiological, symbiotic or other traits more directly related to functioning.

Prior works have sought to understand fine-root trait trade-offs and coordinations in isolation to explain variation generally, but a more specific approach is to relate fine-root variation to environmental conditions. Recent works are striving to relate individual traits and the multidimensional fine-root trait space to edaphic variation to advance theory by contextualizing these trade-offs ecologically (Laliberté, 2017; Guilbeault-Mayers & Laliberté, 2024). Fine-root responses to edaphic variables are often complex, with some evidence that acquisition is downregulated as edaphic resources increase and other evidence showing the opposite. In fact, extremely nutrient-limited ecosystems can host the greatest diversity of nutrient-acquisition strategies (Zemunik et al., 2015; Dallstream et al., 2023). For example, PME has been found to be downregulated with increasing soil P or P addition in tropical forests, indicating acquisitiveness was increased under limitation (Ushio et al., 2015; Guilbeault-Mayers et al., 2020; Cabugao et al., 2021; Lugli et al., 2021). However, along a soil chronosequence in New Zealand, SRL was similar between high- and low-P sites, but in the low-P sites, RTD increased and root N and P decreased, indicating that some acquisitiveness was decreased under limitation (Holdaway et al., 2011). For this reason, defining relationships between individual fine-root traits and edaphic properties is important, but the interpretation of plant acquisitiveness requires consideration of many traits in concert.

The main goal of the present study was to identify the dominant spatial scales of variation for fine-root traits and soil variables. A secondary goal was to identify intrinsic drivers of fine-root traits (i.e. fine-root trait coordinations and trade-offs) including fine-root trait proxies for physiological functions and other root–root relationships. Finally, we aimed to identify the edaphic drivers of fine-root function (i.e. root–soil relationships) and to examine how spatial scales of variation affected these relationships. Such questions about the intrinsic and edaphic drivers of fine-root trait variation are especially relevant for tropical forests, which disproportionately contribute to global carbon sequestration and will regulate feedbacks to climate change (Wieder et al., 2015; Allen et al., 2020). The tropics are also considered unique in terms of the abiotic factors and mineral nutrients expected to limit plant productivity, but are relatively under-sampled and underrepresented in global and vegetation models (Cusack et al., 2024).

We employed a nested, high-resolution sampling scheme that coupled fine-root samples to their adjacent soils across an edaphic gradient in the seasonally dry tropical forests of northwestern Costa Rica. Sampling scale spanned from within trees (> 1 m), to among trees within a site (2.5–60 m), to among sites (15–60 km). We sampled a single tree species, Handroanthus ochraceus, that is common and widely distributed in the region and thus perhaps more likely to express substantial trait variation (Powers et al., 2009). To parsimoniously evaluate root–root and root–soil relationships, we emphasized multivariate analyses given the multidimensional nature of both. Multiple P acquisition mechanisms were quantified due to their putative importance in shaping tropical fine-root responses: SRL and AM% increase the volume of soil accessed for a given carbon construction cost, which helps plants acquire low-mobility nutrients, such as PO₄, whereas PME mineralizes PO₄ from organic substrates (i.e. phosphomonoesters).

We hypothesized the following. (1) Substantial fine-root trait variation would occur within the finest of the three sampling scales (i.e. greater than one-third of total variance within trees), and substantial soil variation would occur at the finest of two sampling scales (i.e. greater than one-half of total variance within sites). (2) As observed for interspecific trait relationships, intraspecific fine-root trait relationships would also show that D and AM% coordinate with each other and trade off with SRL and PME. Independently, RTD would trade off with N% and RESP. (3) Fine-root traits related to P acquisition would be downregulated as soil PO₄ increased. For example, we expected that PME, AM% and/or SRL would decrease as plant-available soil PO₄ increased. Generally, root morphological traits would become more conservative as soil PO₄, NH₄ and NO₃ increased; for example, SRL would decrease and RTD would increase. (4) If the spatial scales of soil and fine-root trait variation are similar and properly captured, the root–soil relationships observed should be stronger.