**4. Discussion**

#### *4.1. Vulnerability and Adaptative Vegetation Management*

Here we have demonstrated an analytical framework to document relative climate-change vulnerabilities among major upland vegetation types that dominate desert and montane forests of the interior Western USA. While many of these types are well represented in protected and managed public lands [46] that does not shield them from effects of climate change. By integrating available

information, we identified types and places where signals of climate change stress are emerging, and where they can be foreseen over upcoming decades. This e ffort drew inspiration from many similar e fforts on the overall structure, measurements used, and available data, but here our analysis is based on synecology and facilitates practical application of results to vegetation at scales commonly addressed in biodiversity conservation and natural resource management. By applying a systematic framework to climate change vulnerability assessment, it generates actionable information targeted to both policy-makers and land managers in support of natural resource conservation decisions.

While traditional natural resource managemen<sup>t</sup> has tended to be 'retrospective'—utilizing knowledge of past and current conditions to inform today's managemen<sup>t</sup> actions—conservation professionals are increasingly required to more rigorously forecast future conditions. This forecasting strives to determine the nature and magnitude of change likely to occur, and then translate that knowledge to current decision-making timeframes. It is no longer su fficient to assess "how are we doing?" and then decide what actions should be prioritized for the upcoming 15-year managemen<sup>t</sup> plan. One must now ask "how is it changing, and by when?" and then translate that knowledge back into actions to take within one or more planning horizons.

Climate change adaptation includes actions that enable ecosystems and people to better cope with or adjust to changing conditions. Some have categorized major strategies into three areas, including resistance, resilience, and facilitated transformation [47–49]. Where vulnerability assessments indicate low vulnerability over upcoming decades, managemen<sup>t</sup> can concentrate on resistance-based strategies; aiming to prevent ecosystem degradation. Where moderate to high vulnerability is indicated, strategies focused on restoring resilience are the priority. Where vulnerability is indicated as being very high over upcoming decades, options for facilitated transformation need to be identified.

Results of this analysis sugges<sup>t</sup> adaptation strategies that suit the character of the vegetation type. For example, as described in the results, warm desert shrublands and semi-desert grassland types already score into the high vulnerability range. It would be prudent for planners and managers to evaluate current landscape patterns and identify zones where they can anticipate plant invasions from neighboring vegetation [50]. Where degraded from prior land uses, restoration of native herb diversity and nitrogen fixing taxa are also needed. Monitoring for pollinator population trends, invasive plant expansion, and shrub regeneration, are also increasingly urgent.

Further north and upslope, pinyon-juniper woodlands currently tend to score in the low-moderate range of vulnerability, but my mid-century, they score in the moderate to high range of vulnerability. Actions to maintain or restore resilience in these forests are needed [51]. These could include protection of remaining "old growth" stands while restoring natural wildfire regimes and tree canopy densities in the surroundings. Over upcoming decades, as temperature and precipitation patterns change, models of wildfire regimes will need to be updated and customized to local conditions. Monitoring for invasive plant expansion, such as from cheatgrass (*Bromus tectorum*), e ffects of drought stress, and tree regeneration will all increase in urgency.

Looking out to the upcoming decades towards the mid-21st century, nearly all types assessed here would benefit from a set of resilience-based strategies, so these investments in the near-term may limit needs for more extreme measures later in the century.

#### *4.2. Methodological Issues*

The HCCVI could be considered an Indicator-Based Vulnerability Assessment (IBVA), where a series of indicators for exposure or resilience are measured and then combined to approximate relative climate change vulnerability. Tonmoy et al. [29] completed a meta-analysis of vulnerability assessments and provide useful insights into strengths and weaknesses in methods for IBVA design and indicator aggregation. Our HCCVI could be categorized as a hybrid IBVA that combines simulation modeling (e.g., for project climate conditions) with the present value of other indicators (e.g., for sensitivity and adaptive capacity measures) to arrive at a vulnerability score. We normalized each indicator and used an arithmetic mean for their combination. This approach to aggregation requires a high level of independence among component indicators and considerable knowledge of the relationship between each indicator and overall vulnerability. While these vegetation types are well understood and documented, and indicators used here reflect high-quality data, we still cannot presume all assumptions have been fully met with regards to aggregation rules.

In contrast to some of the relatively few other assessment methods for natural community types, we have limited our scope to well-described types characterized by native vegetation. Where others have chosen to assess socio-ecological systems [17,18], the level of uncertainty increases considerably. While we acknowledge the relative utility of formally integrating human dimensions into ecosystem concepts, we feel that the introduced uncertainty is considerable. Our approach aims to limit this uncertainty by first assessing the natural community type, and then in subsequent steps, these outputs can be brought together with human dimensions a ffecting a given landscape of interest.

Also, in contrast to many other assessments, here we have emphasized measuring climate exposure along a trend line from the mid-20th century baseline through recent conditions, and into the future time frame of the upcoming decades. This trend-based approach helps to address the considerable uncertainty associated with climate projections by first grounding "current" measures from already observed climate change, and then by focusing on the upcoming decades though the mid-21st century.

One of the most significant methodological challenges to our framework is the application of the climate exposure measure. We anticipate much additional e ffort to better tie climate trend data more directly to driving ecological processes (e.g., biomass productivity, hydrologic regime, fire regime) to provide more robust predictions. Similarly, some factors a ffecting resilience will change over upcoming decades, and so the ability to create reliable forecasts of changing conditions, such as those resulting from future development patterns or invasive species spread, will add precision to overall resilience forecasts. Other challenges identified with this framework included the treatment of functional species groups and "keystone" species. While both concepts for vulnerability measures are desirable and likely provide important contributions, limits to current knowledge become apparent when one attempts to identify and assess species for each category. In addition, locating data sets for component measures that span the range of a given type will remain a challenge. A systematic and regional approach such as this that we have taken can highlight needs for investments in data sets critical to addressing climate change.
