The role of terrain-mediated hydroclimate in vegetation recovery after wildfire

Throughout communities and ecosystems both within and downstream of mountain forests, there is an increasing risk of wildfire. After a wildfire, stakeholder management will vary depending on the rate and spatial heterogeneity of forest re-establishment. However, forest re-establishment and recovery after a wildfire is closely linked to interactions between the temporal evolution of plant-available water (PAW) and spatial patterns in available energy. Therefore, we propose a conceptual model that describes spatial heterogeneity in long-term watershed recovery rate as a function of topographically-mediated interactions between available energy and the movement of water in the subsurface (i.e. subsurface hydrologic redistribution). As vegetation becomes re-established across a burned landscape in response to topographic and subsurface controls on water and energy, canopies shade the ground surface and reduce wind speed creating positive feedbacks that increase PAW. Furthermore, slope aspect differentially impacts the spatial patterns in regrowth and re-establishment. South aspect slopes receive high solar radiation, and consequently are warmer and drier, with lower standing biomass and greater drought stress and mortality compared to north aspect slopes. To date, most assessments of these impacts have taken a bulk approach, or an implicitly one-dimensional conceptual approach that does not include spatial heterogeneity in hydroclimate influenced by topography and vegetation. The presented conceptual model sets a starting point to further our understanding of the spatio-temporal evolution of PAW storage, energy availability, and vegetation re-establishment and survival in forested catchments after a wildfire. The model also provides a template for collaboration with diverse stakeholders to aid the co-production of next generation management tools to mitigate the negative impacts of future wildfires.


Introduction
Wildfire impacts on water resources are an increasing risk for communities and ecosystems in and downstream of mountain forests. Climate change has increased the frequency and burned area of wildfire in the western US over recent decades, altering the land surface energy balance resulting in surprising impacts on mountain snowpacks, ecosystem water availability, and streamflow generation (Kampf et al 2022). While the immediate impacts of fire on air quality, carbon storage, peak streamflow, and sediment delivery are well quantified, the long-term impacts are poorly understood (Wine et al 2018). In large part, this long-term uncertainty arises from spatial and temporal heterogeneity with respect to if and how vegetation recovers from disturbance in a warmer and likely drier climate where water limitation becomes more widespread, especially during late summer. Because vegetation both responds to and alters local hydroclimate, predicting recovery requires more complex and nuanced representations of the water cycle in complex terrain. Here, we review what is known and present a new conceptual model about terrain-mediated interactions between water, energy, and vegetation that can be expected to result in spatial heterogeneity in forest recovery and water resources in the western US.
As climate change increases the frequency, duration, and area under water stress, forest reestablishment and recovery after a wildfire will be closely linked to interactions between the temporal evolution of plant-available water (PAW) and spatial patterns in available energy (Brooks et al 2015, Fan et al 2019. High intensity wildfires remove surface vegetation and soil organic matter, leaving landscapes barren. The removal of organic matter and development of hydrophobic compounds alters the structure of the soil reducing infiltration and decreasing surface water storage. Consequently, immediately following a wildfire the landscape is prone to: (a) high spatial variability in infiltration (Onda et al 2008), (b) increased runoff response to input (Moody et al 2008), and the resulting (c) high rates of erosion which rapidly changes the physiography of the landscape (Kampf et al 2016). The altered geomorphology creates spatial and temporal patterns of hydrologic, sediment, and nutrient connectivity different from that of the pre-burn landscape. While these short-term responses have been studied at the plot (Benavides-Solorio and MacDonald 2001), hillslope (Schmeer et al 2018), and watershed scales (Saxe et al 2018), few studies have considered how these changes influence PAW storage across complex terrain at longer-term timescales.
We propose a conceptual model that describes spatial heterogeneity in long-term watershed recovery rate as a function of topographically-mediated interactions between available energy and the movement and storage of water in the shallow subsurface (i.e. subsurface hydrologic redistribution). Subsurface hydrologic redistribution acts as the connection between short-and long-term recovery timescales because it controls the distribution of PAW across complex terrain (McNamara et al 2005, Hwang et al 2012, McGuire et al 2018. The distribution of PAW then controls the spatial patterns of vegetation establishment and survival after a wildfire (e.g. Rodriguez-Iturbe 2000, Litaor et al 2008, Zapata-Rios et al 2015, Boisramé et al 2018. Spatially variable and temporally dynamic subsurface hydrologic redistribution has only recently been incorporated into predictive ecohydrological models. We argue that expansion of work in this area is required for models to get the 'right answers for the right reasons' (Kirchner 2006). In particular, understanding the temporally and spatially variable evolution of PAW after a wildfire will improve predictive capabilities on (1) how and where seedlings establish in the short term, and (2) the likelihood of seedling successful re-establishment and survival across dry years which are likely to become more frequent in a warming climate.

Hydrologic redistribution and PAW heterogeneity
The processes that control the post-wildfire spatial heterogeneity of water availability in complex terrain determines the patterns of long-term biomass accumulation (Swetnam et al 2017, Boisramé et al 2018, annual gross primary production (Tai et al 2021), and vegetation resilience to drought and insect attack (Tai et al 2021). These patterns are widespread throughout western US forests and are driven by subsurface hydrological redistribution from ridgeline to valley and at locations where local slope decreases (Rodriguez-Iturbe 2000, Litaor et al 2008, Zapata-Rios et al 2015, Swetnam et al 2017, Boisramé et al 2018. Similarly, antecedent soil moisture influences the partitioning of incoming precipitation amongst runoff, groundwater recharge, and evapotranspiration (e.g. Hwang et al 2012). The interactions between spatially variable and temporally dynamic subsurface hydrologic redistribution and topographically-driven patterns in energy availability have only recently begun to be incorporated into ecohydrologic models. Thus, the influence of these processes has yet to be fully explored in relation to vegetation reestablishment after a wildfire.
Many ecohydrologic studies have investigated post-wildfire recovery as bulk responses across scales (i.e. plot to watershed). This implicit onedimensional conceptual model of a system likely lacks transferability to other locations due to differences in complex physiographic characteristics that influence hydroclimate experienced by vegetation during recovery. Topography complicates the post-wildfire recovery processes through the redistribution of water both above and below the ground surface. Mountain snowpack is often redistributed from exposed areas and deposited on leeward slopes or where physical wind breaks occur such as surviving vegetation. During spring snowmelt, increased infiltration occurs where snow accumulates on leeward slopes, solar shaded areas, and the leeward sides of forest stands (Musselman et al 2012, Broxton et al 2015, Harpold et al 2015, Webb 2017, Webb et al 2020. Following spring snowmelt, rain events create an additional distinct infiltration pattern from runoff and run-on around vegetation driven largely by gravity flow and upslope contributing areas (Hwang et al 2012, Rossi et al 2018. However, the seasonally variable patterns of water input to the system also has seasonally varying output associated with the increased demand during the growing season. These complex processes of spatially and temporally variable patterns of water input, subsurface hydrologic redistribution, and water output result in spatially heterogeneous distribution of PAW and energy availability across a landscape. As vegetation re-establishes across a burned landscape in response to topographic and subsurface controls on PAW, inputs of organic matter and increases in surface roughness promote infiltration of upslope runoff, limit erosion (Boisramé et al 2018), and increase the rate of soil development (Brooks et al 2015) creating positive feedbacks that increase PAW storage. Vegetation regrowth in these favored locations alter boundary layer characteristics that influence evaporation from the ground surface, specifically reducing wind speed and shading the ground surface (Gouttevin et al 2015, Rossi et al 2018. During winter, shading and sheltering of the snowpack by regrowth reduces sublimation and evaporation, increasing the snow volume in these locations (Rinehart et al 2008, Musselman et al 2012. This shading also delays the onset of snowmelt and slows the rate of melt, thus shading alters the volume and timing of soil water input (Moeser et al 2020). During summer, shading and wind sheltering of the ground surface limits evaporation further, indirectly increasing PAW and decreasing the ratio of evaporation to transpiration. Consequently, there is the potential for positive feedback where initial vegetation re-establishment in topographically favored micro-hydroclimates increases the likelihood of longer-term survival.
Furthermore, slope aspect should differentially impact the spatial patterns in regrowth and reestablishment via topographically driven hydrologic redistribution. In the northern Hemisphere, south and west aspect slopes receive high solar radiation, and consequently are warmer and drier, with lower standing biomass ( Thus, topographically driven hydrologic subsidy should result in greater spatial variability in vegetation recovery in higher energy south and west aspect environments than on lower energy north and east aspect slopes or slopes shaded by adjacent topography. This is particularly important in predicting rates and trajectory of recovery in a watershed from wildfire in drought conditions compared to the climate when pre-burn vegetation patterns were established.

Conceptual model and testable hypothesis
The inferential evidence for hydrologic redistribution in complex terrain driving ecosystem structure and function is strong and widespread. These observations present both a notable challenge and opportunity to predict vegetation response and recovery after wildfire in complex terrain. To address this challenge, we propose a conceptual model that includes interactions between terrain-mediated subsurface hydrologic redistribution, energy availability, and vegetation structure (e.g. Rinehart et al 2008, Thompson et al 2011, Swetnam et al 2017 that differentially influences both short-term responses and long-term forest re-establishment after a wildfire (figure 1). Thus, we present figure 1 as a new conceptual model of topographically driven hydrologically redistributed PAW and vegetation recovery after a wildfire for generic hillslopes. Just as productivity, species composition, and drought response vary in undisturbed forests, we expect that the vegetation recovery rate and spatial heterogeneity will vary predictably between a southerly aspect (figure 1(a)) and northerly aspect (figure 1(b)) due to differences in energy availability. It is important to note that the influence of complex terrain and resulting distribution of PAW also alters the timeline of vegetation re-establishment and survival depending on physiographic parameters yet to be fully constrained. Therefore, we also present a timeline of four stages for post-wildfire vegetation recovery and associated ecohydrologic impacts in the form of a testable vegetation recovery hypothesis (table 1).
This testable hypothesis focuses on the temporal evolution of vegetation recovery driven by interactions between spatial patterns in water availability as a function of slope position, spatial patterns in available energy related to aspect, and vegetation influences on local microclimate. Specifically, the proposed four stage vegetation recovery focuses on the ecohydrologic response of evaporation, transpiration, canopy interception, infiltration, soil moisture, and energy demand reaching the ground surface. As vegetation establishes and survives, tree canopies will increase interception of incoming precipitation. However, the shading of radiation and increase in surface roughness overcomes these interception losses, resulting in an increase in PAW as previously described (Musselman et al 2008, Veatch et al 2009, O'Donnell et al 2021. Initial ground cover recovery (Stage 2) will be shallow rooted vegetation that utilizes shallow soil moisture, allowing deeper soil moisture retention (Tromp-van Meerveld and McDonnell 2006). As saplings establish (Stage 3), vegetation will access PAW from both shallow and deep soil moisture. We hypothesize that vegetation adolescence (Stage 4) will take approximately 5-10 years and the long-term spatial patterns of vegetation will emerge during this stage. However, long-term recovery beyond 10 years is an outstanding question due to uncertainty with respect to how vegetation recovers in a warmer, drier climate.
One example of our conceptual model and timeline is the recovery from the Thompson Ridge fire which occurred during the summer of 2013 in the Jemez mountains, New Mexico USA. Vegetation recovery from this fire has occurred on the north aspect but remains lacking on south aspect slopes However, there remains a need to systematically evaluate these processes at multiple sites in a variety of conditions. Our presented conceptual figure (figure 1) and testable vegetation recovery hypothesis (table 1) are posed as a starting point to conduct such a systematic evaluations in future investigations that incorporate multiple eco-regions.
Post-wildfire recovery will vary between ecoregions due to differences in vegetation species, phenology, latitude, geology, and climate patterns. However, further work is necessary to constrain the physiographic and environmental factors that constrain PAW and the associated vegetation reestablishment and survival. Once vegetation is reestablished and survives (figure 1(iii)), the spatial patterns of infiltration, depth to groundwater, and evapotranspiration will vary in a predictable seasonal manner that may be projected into the future (e.g. Boisramé et al 2018). These time-scales of recovery are likely to vary based on seasonality and form (rain vs snow) of precipitation relative to the growing season, different topographic thresholds such as hillslope angle, and geological impacts on subsurface water storage and transmissivity. Additionally, the distance to seed source and distribution mechanism (e.g. Kiel and Turner 2022) as well as hillslope angle influencing geomorphic changes (e.g. Kampf et al 2016) are likely important drivers in recovery after a  wildfire. Thus, further understanding of the influence of complex topography across multiple eco-regions on the subsurface hydrologic redistribution of PAW and association with vegetation establishment and survival will improve models for long-term predictive capabilities.
Long-term predictive capabilities in vegetation recovery may also vary across scales. While this paper focuses on the headwater catchment and hillslopes, larger scale processes are important to consider. At the mountain range scale, windward sides of mountains may receive higher amounts of precipitation due to orographic lifting, increasing PAW differently than at the hillslope scale as described in this paper. This may result in recovery patterns that vary by scale, though further investigations are necessary.
Considering the transferability of the presented conceptual model, the physical processes described in this paper will also drive recovery elsewhere. However, differences in climate and vegetation type, as previously mentioned, may alter the timeline. For example, wildfire in the mountains of south Africa have been linked to similar PAW storage processes as described herein (van

Concluding remarks
Burned areas experience both short-and long-term alterations to runoff and water quality that can be costly for municipalities and landowners, with large uncertainties associated with planning and recovery efforts. One of the largest uncertainties, particularly in seasonally water-limited forests characteristic of much of western North America, is understanding the mechanisms driving both spatial and temporal variability in vegetation re-establishment and survival after a wildfire and the ecohydrological impacts of this variability. Constraining these uncertainties in the face of climate change and more extreme wildfires will serve as an important link between science and resource management for ecosystems and communities that rely on these water supplies.
To date, most assessments of these impacts have taken a bulk approach, or an implicitly onedimensional conceptual approach that does not include spatial heterogeneity in hydroclimate influenced by topography and vegetation. Understanding of the natural heterogeneity of these processes across space and time in complex terrain is a major gap in current understanding. The presented conceptual model that includes influences of complex terrain in addition to the testable vegetation recovery hypothesis associated with time of recovery will address this knowledge gap. This sets a starting point to further our understanding of the spatio-temporal evolution of PAW storage and vegetation re-establishment and survival in forested catchments after a wildfire. The presented conceptual model also provides a template for discussions with diverse stakeholders to help integrate multiple interests and assist in the coproduction of knowledge to develop next generation management tools to mitigate the negative impacts of future wildfires.

Data availability statement
No new data were created or analyzed in this study.