Repurposing long-term ecological studies for climate change

Ecologists have long known that static systems can be complex, with population, community, and ecosystem processes interacting across timescales in a spatially heterogeneous environment. Understanding the direct and indirect consequences of climate change is even more daunting, given its dynamic nature, wherein multiple time-variant physical factors influence biological processes across a broad range of temporal and spatial scales. A particularly critical need is for longitudinal records of populations, communities, and environmental variables with sufficient duration to span significant climatic change. Few such records specifically designed to assess climate-change impacts span more than one or two decades. In this issue, Wang et al. report a 35-y record of ecological change in an Arizona desert-stream ecosystem that has experienced a substantial increase in temperature and decrease in discharge (1).

In the first 15 y of the study, Sycamore Creek experienced seasonal and annual fluctuations in discharge, including major floods and periods of no surface flow. In the face of these variations, composition and biomass of macroinvertebrate communities were relatively stable. The most recent 10 y of the observation period had higher temperatures, fewer and smaller floods, and more zero-flow days, and were characterized by reduced diversity, higher community sensitivity to hydrological fluctuation, and persistent compositional shifts. These findings reveal stabilizing mechanisms underlying aquatic communities and destabilizing effects of climate change, as well as feedback effects (e.g., replacement of algal mats by emergent plants, further increasing water temperatures, and dampening streamflow). Wang et al. (1) provide a glimpse of the future of Sycamore Creek and other arid streams as increasing air temperatures lead to lower discharge and higher water temperature. The Sycamore Creek ecosystem is undergoing a state transition—one that is probably incomplete as well as transient. As the authors note, continuing climate change will lead to further fundamental changes in flow regimes, community composition, and ecosystem properties.

Ecosystem transformations are underway across the globe, and resource managers are scrambling to understand how to manage transitions, minimize biodiversity loss, and avoid disruption of ecosystem services (2, 3). Under continued climate change in the coming decades, many places will undergo serial transitions, with transitional states disrupted or redirected by additional extreme events or simple accrual of change in biologically important climate-related factors. Anticipating these changes, and making management decisions under them, will benefit from a foundation of longitudinal research.

In PNAS, Wang et al. report a 35-y record of ecological change in an Arizona desert-stream ecosystem that has experienced substantial increase in temperature and decrease in discharge.

The ecological research community was largely caught flat-footed when climate change emerged as an imminent environmental concern in the 1980s. Fortunately, the 1970s and 1980s saw widespread emplacement of long-term ecological studies and monitoring programs. In the United States of America, many were initiated in the 1980s under the National Science Foundation’s Long-Term Ecological Research and related programs. Those programs were intended to support research on topical questions of the day, recognizing that studies spanning multiple years or decades were necessary to smooth out high-frequency fluctuations, understand cyclic disturbances, examine slow processes, and detect trends (4, 5). Although some scientists argued at the time that climate change was a dynamic determinant of ecological states (6, 7), climate was seldom incorporated as a major research question or design element.

Despite these design limitations, long-term ecological studies initiated in the late 20th Century are paying off, revealing impacts of four decades of accumulated climate change on a variety of ecosystems (8). Of particular note are projects initially designed around other questions, but later remodeled to examine climate change (Fig. 1). Wang et al. (1) is in this category: initially developed as a set of short-term studies, it was modified and extended as a longer-term study of interannual hydrological fluctuation. Continuation over the past two decades has put the authors in position to examine consequences of recently rising temperatures. By its long focus on ecological effects of hydrological variability, their study was preadapted for climate-change application. In hindsight, more precise inferences could have been made had the study incorporated water-temperature monitoring and capacity to examine species-level shifts within families and functional groups (difficult or impossible in the field, but possible now with DNA analysis of archived specimens). Repurposed studies inevitably suffer from limitations imposed by historical decisions on what to measure, record, and archive.

Fig. 1.

Varieties of long-term ecological studies of climate change. Prospective studies collect real-time data as climate change progresses. Retrospective studies examine past changes. The horizontal arrow between these classes indicates that, as time proceeds, the results of prospective studies will be used in retrospective studies. Retrospective studies are active or passive, depending on the role of humans (or their instruments and technologies) in past observations. Passive studies, usually from geohistorical archives, can cover historic periods or extend deeper in time. Active studies include intentional studies of climate-change impacts, long-term ecological studies remodeled to address climate-change questions, and opportunistic studies in which human-recorded ecological information from any archival source is used to examine temporal change.

Because of the dearth of intentional climate-change studies with time depth beyond a decade or two, for the foreseeable future, ecologists will rely heavily on remodeled long-term studies (e.g., ref. 1) and on opportunistic longitudinal studies, in which data or observations collected for entirely different reasons are repurposed for climate change. The latter category includes a vast range of information sources, including archived ecological data (e.g., permanent plots or site-specific sampling records), floristic and faunistic inventories, fisheries records, place-based traditional knowledge, witness-tree and land-survey records, and anecdotal accounts or observations (e.g., field notes, ship’s logs, trade records, diaries, and photographs). All these classes of information, and many more, are being compared with recent observations to assess change. All bear nontrivial uncertainties (incompleteness, errors, biases, loss, or deterioration of information). Yet, all are useful, providing irreplaceable information on ecological states at specific times, places, and climatic states. Deficiencies can be identified and often quantified or even offset. These information sources are similar in that they involve active, real-time observations and recording by human agents. Traditional knowledge is also an active source, involving individual or cultural observations, differing only by its propagation through time by means of oral communication and collective memory.

All these studies are retrospective; they compare information about past ecological and environmental states with later states, often the present. These active retrospective studies have much in common with retrospective research involving passive records, in which past ecological and environmental states have been recorded by nonhuman agents and processes. Passive records, generally studied under the rubric of paleoecology, are diverse in nature, including tree-ring demographic records, rodent-midden hoards of plant materials, predator-collected bone deposits, and sedimentary accumulations of sundry materials in lakes, wetlands, estuaries, and oceans. Passive records extend observational capacities over longer timespans than active records, encompassing a greater range of climate change and ecological responses than possible otherwise (9). They are also useful in more recent settings, including the past century, by providing records of variables that weren’t recorded by scientific observers (10).

Retrospective studies, whether active or passive, require careful consideration of the context in which information was accumulated. Even in the most ideal setting, analysis and inference are constrained by decisions made by the scientists who designed the study, and those who maintained the study and its records over the years. Constraints increase with remodeled studies, and even more with opportunistic studies. In all cases, the intellectual context, methods, and practical circumstances of the observers need to be recognized, acknowledged, and accounted for in the study. The recent controversy over application of Humboldt’s 1802 botanical observations to assess Andean climate-change effects is a case in point (11). Paleoecological records have their own pitfalls. As with active studies, attribution of drivers and mechanisms can be challenging. Paleoecology has its own subdiscipline, taphonomy, concerned with understanding the fate of ecological information as it is incorporated into a passive archive, extracted into a dataset, and used to infer an ecological state that no longer exists (12). Along the way, information is retained but often degraded or distorted, much like ecological information actively incorporated into written or digital archives.

Despite these difficulties, retrospective records remain the best available sources of empirical information about climate-change impacts. Prospective studies, aimed specifically at recording ecological response to ongoing climate change, will become increasingly important in the coming decades. As these studies are developed, we should remember that today’s prospective study will become tomorrow’s retrospective study. Future scientists analyzing data from the best-designed prospective studies will grapple with the same issues we face now in retrospective studies. What can we be measuring now that our counterparts in 2048, 2073, or 2123 will thank us for? This is not an easy question. Novel technical capacities, environmental stresses, and ecological ideas will arise in the interim. But experience of the past can inform the preparation for the future. Both active and passive retrospective studies have already identified fundamental phenomena, including nonstationary variation, climate-paced recruitment pulses, abrupt transformations, novel environments and communities, and climate-driven species translocations. We can look comprehensively at existing long-term ecological records and ask what we wish our predecessors had recorded, measured, collected, and archived in 1998, 1973, or 1923. Studies like Wang et al. (1) are useful not only for their ecological insights but also for indicating what might be missing from today’s study design and protocols.

Anticipating the future is an exercise in imagination. In pondering what we wish scientists of the past had measured, we must put ourselves in their place and consider what they knew and didn’t know. In that context, how might they have anticipated current needs and capacities, even if unknowingly? Although the genetic role of DNA was unknown in 1923, dried plant and animal specimens were routinely archived by taxonomists, if not most ecologists. Application of DNA, isotope, and other technologies to specimen archives of earlier generations are yielding significant rewards (13). Spatially precise sampling of species distributions along elevational gradients was prescient in supporting comparisons a century later (14). Pivoting to the future, we can ask what scientists may find valuable, however mundane now. Gene Likens noted in 1983 that long-term studies should deliver benefits far beyond the lives and careers of the initiators (15). Climate change will preoccupy ecologists and resource managers for generations, and current efforts should keep our successors’ needs in mind.