Dry soils also inhibit reproduction, survival, dispersal, and development of symptoms in host plants . Soil moisture conditions experienced by the pathogen themselves arise from interactions among the precipitation regime, soil depth, drainage, and atmospheric evaporative demand, and thus reflect the interplay of edaphic and climatic conditions. Finally, Pc disease is also often suppressed in rich soils where organic matter content exceeds 5% , probably because of predation by other soil organisms in the diverse microfaunal communities sustained in these soils . Projections of potential future risk therefore require techniques to assess the impact of multiple environmental changes and their interactions on pathogen range and epidemiology. Here, we apply a mechanistic modeling approach to explore how climate change could impact pathogen range and activity. We explore how simultaneous changes in temperature, precipitation, snow-pack extent, and evaporative demand might impact the range of a well-characterized pathogen under different climate scenarios. To do this we couple two existing models that describe controls on the range of the generalist root pathogen Pc in the state of California and surrounding regions in the states of Oregon, Nevada, and Arizona in the southwest USA. Pc occurs in this region but its range is poorly delineated. In other warm climates such as southern Australia and Hawaii, Pc has had a devastating effect on timber production, natural forests and agriculture € .
Modeling the climatic and edaphic limits on its potential range in the US southwest will help determine the risks posed by this pathogen, procona florida container particularly since there is not as yet a detailed understanding of the susceptibility of native species to Pc infection.To address the research questions, we link two existing models that describe: Pc winter survival probabilities and Pc disease severity during the spring growing season . We first model winter soil temperatures. Second, we use the modeled soil temperatures to estimate Pc survival with an existing, validated survival model . Next, we expand a probabilistic soil moisture model to account for snowpack contributions. We use the projected soil moisture fluctuations to drive a stochastic pathogen risk model, previously employed to describe disease range, relative risk on different soil types, and response to different irrigation regimes at sites in Western Australia and Oregon . The output of this model is a metric of the likelihood of disease expansion throughout a hosts’ root system, lying between 0 and 1 . Fourth, we identify regions with high soil organic content from the national State Soil Geographic Database and Soil Survey Geographic Database data sets and exclude these regions from the potential Pc range . Finally, we compute a relative Pc risk as the product of the risk of winter survival, absence of suppressive soil conditions, and moisture-controlled host colonization. We apply this modeling framework to baseline and future climate scenarios across the southwest USA. We repeat the analyses using high resolution downscaled climate and soil data for the San Francisco Bay Area , allowing us to explore the effects of microtopography, orography, and fine-grained changes in soil properties, typical of the California coastline, on the spatial patterns of Pc disease risk and its projected climate sensitivity.We use three different climate datasets: monthly regional historical climate observations interpolated to a 140-km2 grid for the 1950–2000 period ; bias-corrected and spatially downscaled National Center for Atmospheric Research Community Climate System Model 3.0 monthly simulations run for the IPCC A2 and B1 climate scenarios on the same 140 9 140 km grid ; and 30-year average monthly projections from the Geophysical Fluid Dynamics Laboratory climate models for the same scenarios, downscaled to a high resolution grid over the San Francisco Bay area for the 1970–2000 and 2035–2065 periods.
To obtain representative climate characteristics, we average the monthly regional data over 10 year intervals . The climate components obtained for each dataset are summarized in Table 1. We estimate average daily temperature for the Bay Area climate surfaces as a simple mean of monthly maximum and minimum temperature; and estimate the potential evaporation at the southwestern US scale using the Priestly–Taylor equation. One of the limitations of this approach lies in the considerable uncertainty surrounding the use of downscaled climate datasets. Our goal in using these high resolution data sets was to: develop a reasonable baseline drawn from interpolated observations and compare this against plausible climate futures to elucidate the range of potential Pc responses and how they could manifest themselves at regional scales. We do not interpret these future cases as predictions or forecasts, but illustrative scenarios to offer insight into process interactions. Regional soil data, specifically percentage clay, organic matter content and soil depth data are derived from STATSGO . Both data sets derive from soil maps with irregular mapping units ranging from <1 ha in size to >20 000 ha . The soil data were thus mapped to the climate grid scales used for both the Bay Area and the regional scale. We use tabulated hydraulic parameters to estimate the properties of the water retention curve for each soil type on the basis of the percentage clay content, classifying <10% clay soils as sands, 10–15% clay content as loamy sands, 15–20% clay contents as sandy loams, 20–40% clay content as loams and >40% clay contents as clays. Georeferenced locations where Pc has been isolated from natural or agricultural soils in California were obtained from the Phytophthora Database , Forest Phytophthoras of the World , and recent literature sources .The southwestern USA contains a climatically, geologically, and biologically diverse set of landscapes . Climatically, the region encompasses cool, humid, oceanic climates near the coast, arid interior valleys and deserts with continental climates, and montane and nivial uplands in the Sierra Nevada mountain ranges. Precipitation tends to be highly seasonal with the majority falling during a 4-month period from December to March. The diversity of climatic types is reflected in the diversity of terrestrial vegetation, with important vegetation communities including coastal rainforests, dry chapparal, Mediterranean oak savannas, montane forests, and subalpine regions . Within the study area, the California Floristic Province is recognized as a biodiversity hotspot , and the state is home to over 7600 plant species . The San Francisco Bay Area, which is used as a higher resolution case study maintains much of the same complexity of climate, geology, and biology, due to strong ocean-inland gradients in temperature and annual precipitation , and the activity of various tectonic faults in the region . The diversity of climate and soil types in the study area can be directly parameterized within the stochastic soil moisture and pathogen models. Note that we have allowed spatial variations in evaporative demand to be driven by radiation, temperature, and humidity , and have not incorporated an explicit treatment of varying vegetation cover on Emax. The rationale for this lies in the fact that surface conductance loses sensitivity to LAI for low humidity and LAI > 2, suggesting that the first order controls on Emax variation lie in energy limitation rather than vegetation characteristics . The diversity of the vegetation also presents a challenge for parameterization of the pathogen dynamics in the model. It is clearly not feasible to parameterize specific host resistance terms for Pc for all 7600 vascular plant species in the study area; furthermore assessments of susceptibility of the native vegetation to Pc are strikingly incomplete, and remain insufficiently quantitative to allow a direct comparison to be made. The problem is similar to applying this model to Pc in Western Australia, where far more work on Pc susceptibility has been performed. In that comparably biodiverse landscape, 3084 of the 5710 native plant species were found to be susceptible or highly susceptible to Pc infection . This determination, however, required intensive surveys of native vegetation in representative habitats, as well as comparative inoculation greenhouse trials , and still provides only qualitative information about the relative vulnerability of host plants. No comparable investigations have as yet taken place in the US south west, procona London container although numerous common native species are known to be vulnerable to Pc, including manzanita species , chapparal species , MonTherey pine, sycamore, western sword fern, coast live oak , bay laurel and madrone .
In the absence of detailed observations of host–pathogen interactions for the diversity of species in Western Australia, we explored climate limitations on range using a combination of: laboratory estimates of Pc growth rates in sterile media and host resistance parameters taken from a moderately susceptible host and estimated to be 0.1 day 1 . We have repeated this approach for the US southwest with the following rationalizations: moderately susceptible host dynamics ensure that climate effects are clear. For highly susceptible hosts, climatic drivers are relatively unimportant, since any serendipitous infection is likely to lead to mortality. For unsusceptible hosts, climate is equally unimportant. Thus to explore the impacts of climate change, moderately susceptible host assumptions are sensible; Evidence from infections of common tree species suggests that the hosts are indeed moderately susceptible to Pc, with infections most common in wet areas and mortality commonly associated with concurrent pest or drought stress ; for the sake of identifying climate sensitivity specifically, holding host vulnerability constant across the study area provides a useful control, and in the absence of more detailed information, is the most parsimonious modeling approach. As a check on the impacts of this assumption, we have also evaluated the sensitivity of the model results to a doubling or halving of host resistance.The effects of a warming climate on Pc risk vary depending on the risk factors and specific climate scenario being assessed. Warming changes winter survival in a straightforward fashion: under the A2 scenario survival increases dramatically so that the region in which more than half of a Pc population would survive the winter increases from 43% to 72% of the study area. More modest increases in winter-survival probabilities arise under the B1 climate scenario, in which Pc winter survival becomes probable over 65% of the study area . The effects of climate change on soil moisture and Pc spring activity are more complex. At the regional scale, climate change reduces the risk posed by Pc across the majority of the study area. However, the changes are spatially variable. Pc risk declines markedly in the Central Valley. It is largely unchanged in coastal northern California and Oregon, where rainfall levels are projected to remain high. Its range is also unchanged in the south-eastern part of the region, which is significantly water limited under contemporary scenarios and projected to remain so. Pc risk increases in the north-eastern extent of the study area. The increase in Pc risk in this area is greatest in the high emissions scenario. In the lower emissions scenario, comparable increases in Pc extent occur inland in the southern extent of the range. In both cases, these increases indicate an interaction of warmer temperatures with unchanged or slightly enhanced rainfall. The potential complexity of the interactions between changing water and temperature in one of these southern locations is illustrated in Fig. 5 for a site in the southwestern part of the region. In this location, Pc risk increases under the B1 scenario but decreases under the A2 scenario. Figure 5 shows a decomposition of the projected changes into those due to temperature and those due to changes in soil moisture. As shown, increasing temperatures increase Pc risk from the baseline case for both A2 and B1 scenarios, but in the A2 scenario, a decrease in soil moisture more than offsets the effect of warmer spring conditions. Conversely, under the B1 conditions, the slight increase in soil moisture increases pathogen risk at this location, but only when both temperature and soil moisture increase together does the large predicted increase in pathogen risk occur. While these threshold-dynamics are not general across the study range, they illustrate the potential for highly nonlinear pathogen responses to interactions in changing temperature and moisture conditions, and highlight the importance of considering the impacts of synchronous changes in climate on pathogen dynamics. As summarized in Table 2 and illustrated in Fig. 2, water limitation reduces Pc risk over a range of 340 000 km2 across the region for the high emissions scenario, with an average decrease in the Pc risk of 0.28. Pc risk is reduced over 40 000 km2 for the low emissions scenario, by 0.26 on average.