The Plant Phenology Monitoring Design for the National Ecological Observatory Network Sarah Elmendorf

Home , Phenology

University of Montana ScholarWorks at University of Montana Numerical Terradynamic Simulation Group Numerical Terradynamic Simulation Group Publications

4-2016 The plant phenology monitoring design for The National Ecological Observatory Network Sarah Elmendorf

Katherine D. Jones

Benjamin I. Cook

Jeffrey M. Diez

Carolyn A. F. Enquist

See next page for additional authors

Let us know how access to this document benefits ouy . Follow this and additional works at: https://scholarworks.umt.edu/ntsg_pubs

Recommended Citation Elmendorf, S. C., K. D. Jones, B. I. Cook, J. M. Diez, C. A. F. Enquist, R. A. Hufft, M. O. Jones, S. J. Mazer, A. J. Miller-Rushing, D. J. P. Moore, M. D. Schwartz, and J. F. Weltzin. 2016. The lp ant phenology monitoring design for The aN tional Ecological Observatory Network. Ecosphere 7(4):e01303. 10.1002/ecs2.1303

This Article is brought to you for free and open access by the Numerical Terradynamic Simulation Group at ScholarWorks at University of Montana. It has been accepted for inclusion in Numerical Terradynamic Simulation Group Publications by an authorized administrator of ScholarWorks at University of Montana. For more information, please contact [email protected]. Authors Sarah Elmendorf, Katherine D. Jones, Benjamin I. Cook, Jeffrey M. Diez, Carolyn A. F. Enquist, Rebecca A. Hufft, Matthew O. Jones, Susan J. Mazer, Abraham J. Miller-Rushing, David J. P. Moore, Mark D. Schwartz, and Jake F. Weltzin

This article is available at ScholarWorks at University of Montana: https://scholarworks.umt.edu/ntsg_pubs/405 SPECIAL FEATURE: NEON DESIGN

The plant phenology monitoring design for The National Ecological Observatory Network Sarah C. Elmendorf,1,2,†Katherine D. Jones,1 Benjamin I. Cook,3 Jeffrey M. Diez,4 Carolyn A. F. Enquist,5,6 Rebecca A. Hufft,7 Matthew O. Jones,8 Susan J. Mazer,9 Abraham J. Miller-Rushing,10 David J. P. Moore,11 Mark D. Schwartz,12 and Jake F. Weltzin13

1The National Ecological Observatory Network, 1685 38th St., Boulder, Colorado 80301 USA 2Department of Ecology and Evolutionary Biology, University of Colorado, Boulder, Colorado 80309 USA 3NASA Goddard Institute for Space Studies, 2880 Broadway, New York, New York 10025 USA 4Department of Botany and Plant Sciences, University of California, Riverside, California 92521 USA 5USA National Phenology Network, National Coordinating Office, 1955 E. 6th Street, Tucson, Arizona 85719 USA 6DOI Southwest Climate Science Center, U.S. Geological Survey, 1064 E. Lowell Street, Tucson, Arizona 85721 USA 7Denver Botanic Gardens, 909 York Street, Denver, Colorado 80206 USA 8Department of Forest Ecosystems and Society, Oregon State University, Corvallis, Oregon 97331 USA 9Department of Ecology, Evolution and Marine Biology, University of California, Santa Barbara, California 93106 USA 10National Park Service, Acadia National Park and Schoodic Education and Research Center, Bar Harbor, Maine 04660 USA 11School of Natural Resources and the Environment, University of Arizona, 1064 East Lowell Street, Tucson, Arizona 85721 USA 12Department of Geography, University of Wisconsin-Milwaukee, PO Box 413, Milwaukee, Wisconsin 53201 USA 13US Geological Survey, 1955 East 6th St., Tucson, Arizona 85721 USA

Citation: Elmendorf, S. C., K. D. Jones, B. I. Cook, J. M. Diez, C. A. F. Enquist, R. A. Hufft, M. O. Jones, S. J. Mazer, A. J. Miller-Rushing, D. J. P. Moore, M. D. Schwartz, and J. F. Weltzin. 2016. The plant phenology monitoring design for The National Ecological Observatory Network. Ecosphere 7(4):e01303. 10.1002/ecs2.1303

Abstract. Phenology is an integrative science that comprises the study of recurring biological activities or events. In an era of rapidly changing climate, the relationship between the timing of those events and environmental cues such as temperature, snowmelt, water availability, or day length are of particular interest. This article provides an overview of the observer-based plant phenology sampling conducted by the U.S. National Ecological Observatory Network (NEON), the resulting data, and the rationale behind the design. Trained technicians will conduct regular in situ observations of plant phenology at all terrestrial NEON sites for the 30-yr life of the observatory. Standardized and coordinated data across the network of sites can be used to quantify the direction and magnitude of the relationships between phenology and environmental forcings, as well as the degree to which these relationships vary among sites, among species, among phenophases, and through time. Vegetation at NEON sites will also be monitored with tower-based cameras, satellite remote sensing, and annual high-resolution airborne remote sensing. Ground-based measurements can be used to calibrate and improve satellite-derived phenometrics. NEON’s phenology monitoring design is complementary to existing phenology research efforts and citizen science initiatives throughout the world and will produce interoperable data. By collocating plant phenology observations with a suite of additional meteorological, biophysical, and ecological measurements (e.g., climate, carbon flux, plant productivity, population dynamics of consumers) at 47 terrestrial sites, the NEON design will enable continental-scale inference about the status, trends, causes, and ecological consequences of phenological change.

Key words: long-term monitoring; NEON; open-source data; plant phenology; sample design; Special Feature: NEON Design.

Received 30 July 2015; revised 6 November 2015; accepted 12 November 2015. Corresponding Editor: E.-L. Hinckley. Copyright: © 2016 Elmendorf et al. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. † E-mail: [email protected]

v www.esajournals.org 1 April 2016 v Volume 7(4) v Article e01303 SPECIAL FEATURE: NEON DESIGN ELMENDORF ET AL.

Introduction within a growing season, and the southern limit by insufficient chilling to break bud dormancy. The overarching mission of NEON is to enable Phenological plasticity may be a beneficial trait. understanding and forecasting of the impacts of Species whose activity patterns closely track in- climate change, land use change, and the intro- terannual climate variability tend to have im- duction of invasive species on ecosystem struc- proved growth, productivity, or reproductive ture and function (see Thorpe et al., unpublished success than those that do not (Cleland et al. manuscript). Tracking the timing of seasonally 2012). In other cases, however, early greenup or recurring life cycle events (phenology) is thus a floral bud development in response to anoma- natural focal area of study for the Observatory. lously early arrival of spring can be detrimental. Plant phenological transitions may be triggered Phenological advancement in response to warm by a variety of cues, including chilling, spring spring temperatures followed by a late frost can temperature, growing degree days, and daylight have catastrophic effects on fruit and seed pro- (Chuine 2000); many of these factors are likely to duction, and canopy development (Inouye 2008, shift significantly over the next 30 yr (IPCC 2013). Hufkens et al. 2012). Changes in phenology have been observed for Climate-induced changes in phenology can many taxa across the earth (Parmesan and Yohe create feedbacks that alter biogeochemical cy- 2003). The onset of spring phenological events cling and species interactions (Melillo et al. 2014). advanced at an estimated mean rate of 1.2 d per Changes in the timing of leaf budburst and se- decade from 1955 to 2002, across the Northern nescence affect surface radiation, near surface Hemisphere, likely caused by recent climate temperature, hydrology, and carbon cycling warming (Schwartz et al. 2006). Observational (Churkina et al. 2005, Bonan 2008, Richardson and experimental studies indicate that plants et al. 2010, Jeong et al. 2012, 2013). An analysis flower on average ~5 d earlier per 1 °C increase in of more than a dozen models included in the spring temperature (Wolkovich et al. 2012) and North American Carbon Program (NACP) Inter- current projections indicate that spring phenolo- im Synthesis indicated across all models, sites, gy could advance by between 1 and 10 d over the and years of data, for each forest type; errors of planned 30-yr lifespan of the NEON observatory up to 25 d in predictions of “spring onset” were (IPCC 2013). Many species, however, delay flow- common, and errors of up to 50 d were observed ering in response to increases in winter or spring (Richardson et al. 2012). From the general posi- temperatures (Mazer et al. 2013), and there is tive relationship between carbon uptake and sea- still much to learn about the causes of variation son length derived from a synthesis of a range among species and higher taxa in the direction of eddy covariance sites, the largest phenologi- and magnitude of their phenological responses cal errors in current models would translate into to both temperature and rainfall (Mazer et al. between ~150 and ~450 g/m2 of carbon annually 2013, 2015). (Churkina et al. 2005). Differential responses to Beyond providing an indicator of climate phenological cues between plants, consumers, change, the timing of phenological transitions and/or pollinators can disrupt the overlap in ac- is also a potentially important driver of demo- tivity periods among interacting organisms, po- graphic trajectories and biogeographic distri- tentially resulting in changes in species fecundity butions of individual taxa, and of ecological and cascading effects on the food chain (Strode processes including species interactions and rates 2003, McKinney et al. 2012) or local extinction of biogeochemical cycling (Morisette et al. 2008). of consumer populations (Singer and Parmesan Phenological traits may physiologically constrain 2010). broad-scale distribution patterns of species; phe- Plant phenology has been studied at a range nology is consistently an important predictor of geographic and temporal scales, and by em- in process-based species distributions models ploying a variety of tools, including: recording (Chuine 2010 and references therein). Models of in situ observations, experimental manipulation temperate deciduous trees, for example, indicate of abiotic factors, modeling, remote sensing, that the northern range limits are constrained and digital photography (Cleland et al. 2007). by the ability to complete reproductive cycles Understanding and reconciling the information

v www.esajournals.org 2 April 2016 v Volume 7(4) v Article e01303 SPECIAL FEATURE: NEON DESIGN ELMENDORF ET AL.

contributed at each scale is challenging (Mo- environmental drivers that affect plant phenol- risette et al. 2008) and observations at multiple ogy, as well as the functional consequences of scales are rare (but see Liang et al. 2011). This ar- changing phenology for a range of ecosystem ticle provides an overview of the plant phenolo- types and processes. The resulting scientific gy sampling that will occur within NEON sites, knowledge can inform decision-making process- including observation protocols, the spatial and es related to natural resource conservation and temporal frequency of monitoring, the taxa tar- management, control of invasive species and geted for observations, and the rationale for the infectious disease, and efforts related to societal sampling regime that was selected (Box 1). The climate change adaptation (Enquist et al. 2014). science design, developed by a technical working group comprised of phenology experts from Sampling Design academic institutions, government and nonprofit agencies, reflects current best practices in mon- Measurements itoring terrestrial plant phenology. The aims of Plant phenology is typically quantified by the plant phenology monitoring dovetail with observing the date of onset and the duration those of the NEON project more generally: to im- of particular phenophases, which may include prove the understanding and forecasting of eco- both vegetative and reproductive events. Specific logical change at continental scales (Schimel et al. phenophase definitions have not been univer- 2011). From its earliest inception, the design of sally adopted across monitoring networks. the NEON project as a whole has focused on gen- Without common units, data interoperability erating core measurements that address the data becomes a limiting factor in data integration. needs common to the broadest possible commu- Consistent with NEON’s commitment to use nity of data users (AIBS 2000). This differs from existing scientifically accepted, vetted, and stan- many site-based or PI-driven projects in that the dardized protocols wherever possible, NEON data are intended to answer multiple questions, will employ USA National Phenology Network rather than tailored to a specific hypothesis test. (USA-NPN) phenophase definitions and proto- By providing integrated and multiscale suites cols (Denny et al. 2014). of measurements on the seasonal progression Advantages of USA-NPN protocols and the of a diversity of taxa and ecosystem process- reasons for selecting this standard for NEON in es at intensively measured sites, data collected situ phenology observations include: (1) status- by NEON will enable the scientific community based monitoring, or the practice of reporting the to develop mechanistic linkages between the phenological condition of an individual at any

Box 1 NEON’s Contribution NEON is poised to advance the field of phenology by: 1. Accumulating high quality, long-term, standardized measurements recorded by trained technicians across 20 major ecosystem types found within the United States. 2. Observing replicate individuals of select species to quantify intraspecific variation in the timing of phenological events within and across years, facilitating precise population-level estimates of phenology. 3. Observing multiple species to characterize the range of phenological response patterns across species, functional groups and life history strategies. 4. Collocating plant phenological measurements with other terrestrial and atmospheric measurements data, which may be used to understand relationships between climate, phenology, ecosystem processes, and biodiversity. 5. Providing open-access, standardized data sets using common protocols and units, in order to facilitate synthetic analyses using data from NEON together with data from other large-scale monitoring networks.

v www.esajournals.org 3 April 2016 v Volume 7(4) v Article e01303 SPECIAL FEATURE: NEON DESIGN ELMENDORF ET AL. time that it is monitored; (2) repeated tracking of the continental United States. Mapping from marked and georeferenced replicate individual USA-NPN definitions to BBCH definitions is fea- perennials and patches of annual/clonal herbs, sible for many phenophases (Denny et al. 2014). and (3) incorporation of both status and “intensi- The phenology protocol includes repeated ty” definitions for phenophases (Kao et al. 2012, assessment of phenophase status and intensity on Denny et al. 2014). Using status-based rather each individual (see section Temporal distribution than first-event monitoring is a departure from of sampling, below, for more details), as well as an many historical phenological monitoring pro- annual assessment of individual-level covariates tocols, but has the advantage that events (such that can affect phenology. Due to resource con- as leaf emergence in Mediterranean climates, or straints, only a subset of the USA-NPN-defined flowering in many desert species) that may occur phenophases (as described by Denny et al. 2014) multiple times during a single year can be cap- will be targeted in NEON phenology sampling tured. Status monitoring also allows the explic- protocols, with the greatest focus on leaf phenolo- it quantification of uncertainties in phenophase gy. The focus on canopy development was select- transition dates (which occur in continuous time) ed, based on recommendations of a NSF Research that are introduced by monitoring in discrete Coordination Network (USA-NPN National Co- temporal bouts, as well quantifying the dura- ordinating Office 2012), to facilitate linkages with tion of phenophases rather than just their date NEON’s measurements of ecosystem processes of onset. Monitoring marked individuals/small such as landscape phenology and carbon cycling. patches ensures that the recorded dates of phe- To connect phenological measurements to plant nological events, or their duration, are decoupled health, productivity, and canopy position, NEON from population size (Miller-Rushing et al. 2008). will measure the size (stem diameter, % cover, Status monitoring overcomes weaknesses of height, and canopy dimensions), disease status, event monitoring, and when coupled with a reg- health condition, and structure of each individual ular sampling frequency, enables more accurate plant or patch once per year. These annual mea- phenophase change estimates. Repeated track- surements will be consistent with those taken on ing of the same georeferenced individual allows other plants in the network as part of NEON’s the NEON phenology measurements to be used vegetation structure and productivity protocol as phenoclimate monitors (like cloned lilacs; see (see Meier and Jones 2015 for details). Schwartz et al. 2012) rather than conflating variation within a population with climate effects. The Phased sampling design protocols employed include intensity metrics Two priorities were identified for NEON’s (e.g., percentage of the canopy that is full with plant phenology observations: Phenology of dom- leaves) along with phenophase status (e.g., one inants, which includes estimating the mean and or more live, unfolded leaves visible). These data intraspecific variance of phenological timing in can be used to estimate mean population onset dominant species within each site (see Phase I, and end dates for each phenophase, as well as below), and Community phenology, focused on track the seasonal progression of development capturing a range of species-specific phenologies throughout the active period. Together, these that represent the plant community at each NEON data should provide better linkages to ecosystem site (Phase II). Dominants are targeted specifically function and remotely sensed phenological data to facilitate linkages to ecosystem function based than existing “first event” phenological data sets, on the assumption that species contribute to which typically quantify the phenological sta- ecosystem properties roughly in proportion to tus of only the most extreme individuals within their relative abundances (Grime 1998). Sampling a population of unknown size (Miller-Rushing of dominant species’ phenology will enable link- et al. 2008). Although other phenophase defini- ing phenological events and patterns observed tions exist [e.g., the BBCH scale, commonly used to aboveground processes captured at other scales in agricultural systems, as well as across Europe by other NEON measurement systems (including (Meier 2001, Koch et al. 2007)], the USA-NPN ecosystem productivity and respiration, and car- scales were selected for interoperability with bon, water, and nutrient cycling), and to the large-scale distributed monitoring data sets in ground-based land-surface phenology signal

v www.esajournals.org 4 April 2016 v Volume 7(4) v Article e01303 SPECIAL FEATURE: NEON DESIGN ELMENDORF ET AL. observed via remote sensing methods. It will in the 4th year of operational sampling and will also provide critical information on intraspecific continue for the remainder of NEON operations variation in phenology patterns, which are poorly at each site. Species to be monitored in Phase II captured when monitoring efforts are limited to will include dominant species (the three species a census of one to several individuals per site. studied as part of Phase I at each site) and up Sampling of community-level phenology will to 17 additional species per site that collectively inform questions regarding interspecific variation represent a range of functional groups and life in the timing and duration of phenological phases, history strategies. Phase II will inform both the and their sensitivity to climate. The resulting range of phenological patterns occurring at a site, data set will enable assessment of the degree as well as predictive models of the sensitivities of to which phenological timing and climate sen- particular species based on their traits (Buckley sitivity vary based on functional groups or growth and Kingsolver 2012). forms (e.g., natives/exotics, overstory/understory, perennial/annual, deciduous/evergreen, herba- Spatial distribution of sampling ceous/woody, early and late-season). These pat- A common critique of much of the existing terns can enable generalizations regarding the ground-phenology observation data is that ob- likely phenological responses and sensitivities of servations are limited in space and are reported species beyond those targeted for regular as points, whereas remote sensing data pixels observation. from commonly used satellite products used NEON will implement phenological monitor- to model phenology range from 30 m to >1 km ing in two phases in order to accomplish both (Schwartz and Hanes 2010). While some studies inter- and intra-specific sampling goals. During have found little spatial autocorrelation in a Phase I (phenology of dominants), implemented single plant species’ phenological response given during the first three full (i.e., all sites operation- uniform temperature over small areas (Schwartz al) years of sampling, phenological observations et al. 2014), dispersion of monitored individuals will concentrate on intensive monitoring of three throughout a larger area is important to en- dominant species at each of the 47 terrestrial sites. compass variation in plant phenology within The NSF Research Coordination Network (RCN) the sampling area caused by microenvironmen- report (USA‐NPN National Coordinating Office tal variation, genetic variation, or both. To fa- 2012) recommends a minimum of 5–10 replicate cilitate repeatable observation of multiple individuals sampled for vegetative phenology individuals over a relatively large area, while per site per species, with an ideal sampling in- keeping travel time to a minimum, marked tensity of 20–30 individuals. In the absence of individuals will be situated along a fixed, 800-m existing data sufficient to statistically determine square “loop” transect (200 m on a side), with smaller minimum sample sizes for particular the four edges oriented in the four cardinal species and sites, NEON will target the higher directions. This size is comparable to the ~250 m end of this range in order to quantify intraspecif- MODIS pixel size, which is commonly used in ic variation in phenological timing for the three satellite-based phenology assessments. most dominant species at each site (see section This loop will be situated within or near Temporal distribution of sampling, below, for de- NEON’s flux tower airshed. The distance of the tails of monitoring frequency). transect from the tower will be site specific based Phase II (community phenology), will follow on identified exclusion areas around tower in- Phase I and consist of more limited sampling strumentation, and will be placed to facilitate than Phase I in terms of frequency and the num- inclusion of individuals located within sampling ber of replicate individuals per species (mini- plots used for NEON’s biomass and productivity mum of five individuals per species per site), measurement (see Meier and Jones 2014) (Fig. 1). but will have an increased number of species. Collocation of the phenology transect with the The focal shift will alter which individuals are instrument tower will allow meteorological and monitored, but keep the total number of plants biophysical data collected by tower-mounted monitored per site at ~90–100 due to budgetary sensors to be used directly in analysis of phe- limitations. Phase II monitoring will commence nological data (e.g., how local climate affects

v www.esajournals.org 5 April 2016 v Volume 7(4) v Article e01303 SPECIAL FEATURE: NEON DESIGN ELMENDORF ET AL.

Fig. 1. Layout of phenology transect (teal square) with respect to the NEON Tower (cross shape), the airshed (wedge shapes), and the Tower Plant Productivity plots (yellow squares) (figure credit: Rachel Krauss, 2015).

phenology) and vice versa (e.g., how leaf status additional insight into the realized environmen- affects daily carbon flux). NEON’s tower loca- tal heterogeneity of the various sites. tions are positioned such that the tower airshed is situated in a spatially and structurally homog- Temporal distribution of sampling enous area with the goal of a minimum of 80% A standard sampling frequency for phenology contribution from the representative ecosystem, has not been prescribed by the ecological com- ensuring that plants selected for phenological munity. Typically, sampling frequency varies by monitoring are also located within a regionally species, environment, sampling objectives, and representative habitat type. The assumption is budgetary and logistical constraints. Accuracy that the intraspecific variation in phenological re- of measurements can be improved by either sponses will, in general, be from individuals sub- increased precision of measurements (i.e., more ject to similar environmental conditions. Even frequent sampling or more extensive training) so, microtopographic features may still affect or by increased sample size (Nguyen et al. 2009). variation in observed phenological response. Ad- Phenophase status assessment using USA-NPN ditional information such as slope, aspect, com- definitions shows good interobserver agreement, munity composition, above-ground biomass, with volunteers classifying plants into the same and canopy chemistry as derived from NEONs phenophase as professional botanists 91% of the airborne observation system may provide time (Fuccillo et al. 2014), a number which

v www.esajournals.org 6 April 2016 v Volume 7(4) v Article e01303 SPECIAL FEATURE: NEON DESIGN ELMENDORF ET AL. should increase for trained NEON technicians. izing phenological forcing models is compro- Variability in phenology is seen across a wide mised. During Phase II, the frequency of pheno- range of scales, including differences between logical observations will be reduced to two times ecosystems, sites, plots, species, and individuals a week during transitional phases in order to (Diez et al. 2012, 2014). Therefore, the ideal accommodate sampling of a greater number of frequency of sampling depends on analysis goals species. (e.g., fitting a thermal forcing model vs. long- Phenologically active periods will vary among term trend detection vs. quantifying intraspecific species both spatially across the continent, and variation in phenology), as well as the degree interannually at each site. In order to catch the of intraspecific and interannual variation in full growing season for all selected species, phenology. Mazer et al. (2015) found that twice- NEON will aim to commence weekly sampling weekly sampling over a 3-yr period was suf- 3 weeks prior to the earliest anticipated onset of ficient to detect statistically significant the first phenophase (based on the earliest date associations between winter monthly rainfall observed in recent records for the species). This and/or mean temperature (and their interactions), date will be determined using local information, and the onset dates of vegetative growth, flow- where available (such as at LTER sites where his- ering, and fruiting in four species monitored torical phenological data exist, or indicator plants in California across broad environmental con- at a nearby, lower elevation site), or from histori- ditions. An NSF Research Coordination Network cal MODIS data, in sites where local information (RCN) report on phenology (2012) suggests a is not available to guide sampling. Start of season sampling interval of 2–4 times per week. Miller- metrics based on remote sensing data are typical- Rushing et al. (2008) recommend sampling every ly biased towards early dates (White et al. 2009, second day to ensure a 97% chance of detecting Ganguly et al. 2010), so this should provide an a significant change in flowering date over 10 yr “earliest” outer bound on start of season. of sampling, based on existing long-term flow- Once budbreak or initial growth is observed, ering data collected in Massachusetts and the observation frequency will increase from Colorado. These recommendations assumed re- once a week to either three times (Phase I) or two alistic anticipated rates of climate warming and times (Phase II) a week. The intensive sampling interannual variability in temperature, in addi- stage ends once full-sized leaves have emerged/ tion to a sensitivity of flowering date to tem- full canopy has formed, and sampling frequen- perature of 1 d/°C. A more recent synthesis of cy is reduced to once a week or once every other long-term phenology data sets worldwide week to survey for open flowers. Three weeks (Wolkovich et al. 2012), however, suggests that before the anticipated first date of senescence, flowering phenology will, on average, shift at based on local and/or MODIS data, sampling a rate of 5–6 d/°C. Therefore, less frequent sam- frequency will increase again to weekly (if pre- pling may be adequate for many species for viously reduced to every other week). At the simple trend detection. first sign of leaf senescence (i.e., fall color), ob- Following the RCN recommendations, during servation frequency will, once more, increase to the first 3 yr of sampling, the phenological status two times a week sampling until <5% of leaves of dominant species (Phase I) will be observed remain or until three consecutive censuses of no three times a week during key transition periods change have been observed. (i.e., leaf emergence and senescence, Table 1). Re- sulting data will be used to inform the sampling Species selection intensity necessary to characterize the mean Species selection is guided by several over- (± 3 d SE) for leaf phenology transition dates for arching principles: (1) The focus of NEON the three dominant species at the site in subse- sampling is to characterize the ecology of the quent years. This target is based on an analysis site. Therefore, an effort is made to sample by Jeong et al. (2012), who concluded that when species that are representative of the plant observational error in estimating population community at each site. (2) High level require- mean transition days for key phenological events ments of the NEON project focus on invasive (e.g., budbreak) is greater than ± 3 d, parameter- species as a driver of ecological change and

v www.esajournals.org 7 April 2016 v Volume 7(4) v Article e01303 SPECIAL FEATURE: NEON DESIGN ELMENDORF ET AL.

Table 1. Proposed rule sets for specific growth forms for phenology sampling at sites withdefined awell- growing season†.

Monitor Sample Sample Sample indicator 3×/week‡ 1×/week 2×/week indi- until all until all until all Sample vidual plants plants plants 1×/week Growth form for show show Then§ Then show until Then Cactus Breaking NA No more End sampling NA NA NA NA flower fresh season buds flowers Deciduous Breaking >50% of ≥95% of Commence Monitor One or <5% of End broadleaf leaf or canopy canopy every other indicator more canopy sam- flower full with full with week indi- colored full with pling buds leaves or leaves monitoring viduals leaves green or season three for open for one colored consecu- flowers or more leaves tive colored bouts of leaves no change Deciduous Breaking >50% of ≥95% or Commence Monitor One or <5% of End conifer needle canopy more of every other indicator more canopy sam- buds full with canopy week indi- colored full with pling needles full with monitoring viduals needles green or season or three needles for open for one colored consecu- pollen or more needles tive cones colored bouts of needles no change Drought Breaking Young No more Commence Monitor One or <5% of End deciduous leaf leaves young every other indicator more canopy sam- broadleaf buds leaves week indi- colored full with pling monitoring viduals leaves green or season for open for one colored flowers or more leaves colored leaves§ Evergreen Breaking Young No more Commence End NA NA NA Broadleaf leaf leaves young every other sampling buds leaves week season monitoring when no for open more flowers fresh flowers are present Evergreen Breaking Young No more Commence End NA NA NA conifer needle needles young every other sampling buds needles week season monitoring when no for open more pollen fresh cones pollen cones are present Evergreen Breaking Young No more Commence End NA NA NA forb leaf leaves young every other sampling buds leaves week season monitoring when no for open more flowers fresh flowers are present

v www.esajournals.org 8 April 2016 v Volume 7(4) v Article e01303 SPECIAL FEATURE: NEON DESIGN ELMENDORF ET AL.

Table 1 (Continued) Monitor Sample Sample Sample indicator 3×/week‡ 1×/week 2×/week indi- until all until all until all Sample vidual plants plants plants 1×/week Growth form for show show Then§ Then show until Then Forb Initial One or NA Commence Monitor NA No more End growth more every other indicator full-sized sam- fully week indi- leaves pling unfolded monitoring viduals are season leaves for evidence present flowering of phenology senescence Graminoid Initial >50% of ≥95% of Commence Monitor <95% green <5% of End growth plant is plant is every other indicator leaves plant is sam- green or green week indi- green pling three monitoring viduals season consecu- for for >5% tive flowering leaf bouts of phenology senes- no cence change Pine Emerging Young No young Commence End NA NA NA needles needles leaves every other sampling or week season pollen monitoring when no cone for open more develop- pollen cone fresh ment pollen cones visible Semi- Breaking Young No more Commence Monitor One or <5% of End evergreen leaf or leaves young every indica- more canopy sam- broadleaf¶ flower OR leaves other tor colored full pling buds >50% of OR 95% week indi- leaves with season canopy or more monitor- viduals green full of ing for for one or with canopy open or colored leaves is full flowers more leaves OR with colored three leaves leaves# consecu- tive bouts of no change † This is generally applicable to temperate or boreal systems; sites lacking a distinct growing season where growth occurs year round or is episodic such that a growing season cannot be defined will be monitored on a weekly basis. ‡ Three times a week in Phase I sampling, two times a week in Phase II. § If flowering phenology precedes leaf/needle budbreak skip the steps outlined in this column and decrease monitoring to watching indicator individuals for fall senescence or end monitoring for the season as specified in the following column. ¶ Semievergreen broadleaf growth form may be used for species in which life history varies with latitude. Monitoring strategy should be driven by phenophase observations. # Seasonal monitoring may end at this point if senescence does not occur. data integration with other large-scale moni- along an established route to sample reliably toring projects. These goals dictate a particular without extensive trail-building. To meet these focus on invasive species and taxa that are the goals, the taxa selected for plant phenology focus of more widely distributed phenological monitoring include (Phase I) three dominant monitoring. (3) The long-term nature of NEON species from each site, plus (Phase II) up to monitoring, and a commitment to minimal site 17 additional taxa. Phase II species selections disturbance, requires that any taxa selected for first targets noxious weeds and species that monitoring be present in sufficient numbers are the focus of other national phenological

v www.esajournals.org 9 April 2016 v Volume 7(4) v Article e01303 SPECIAL FEATURE: NEON DESIGN ELMENDORF ET AL. monitoring programs, with the remaining spe- stood for predicting phenology. It will also serve cies selected based on rank abundance. to concentrate monitoring efforts on species that Prior to commencing phenology observations are relatively common locally, complimenting at a given site, NEON will conduct quantita- targeted selection of campaign species which tive vegetation surveys within 20–30 randomly have large geographic ranges and concentrate placed plots within the tower airshed to assess monitoring efforts on taxa that cover multiple site-specific species abundances. These baseline sites. vegetation surveys are collected according to NEON’s standard protocols for plant diversity Site-specific modifications and vegetation structure (Barnett 2015, Meier Modifications will be made for sites with and Jones 2015), and are used to inform imple- growing seasons or species with life histories mentation details for both the phenology and that differ from the typical temperate deciduous vegetation structure measurements at NEON model. For example, sampling may begin earlier sites. Within each surveyed plot, abundance of than described above to capture flowering phe- overstory species is quantified via basal area per nophases for plants that flower prior to leaf species, and the abundance of understory species production. Additionally, sampling frequency is quantified by percent cover. Three dominant will need to be modified at sites without a species will be identified at each site for Phase clear seasonal greening pattern (e.g., tropical I phenology monitoring. The dominant species ecosystems or Mediterranean climates where selected will include the two most abundant species may leaf out or flower multiple times canopy species plus the single most abundant per year in response to episodic rainfall); in understory species for sites with >50% canopy these cases, year-round sampling with longer closure, and the two most abundant understory intercensus intervals will be necessary to capture species plus the single most dominant oversto- phenological trends. Modifications will also need ry species for sites with <50% canopy closure. to be made for cropped (agricultural) sites. At At sites with no defined woody overstory, e.g., these sites, NEON will monitor the cultivated grasslands, all three species will be selected from species; in most cases, the selected species will the herbaceous community. Understory and can- vary by year to track crop rotations and will opy species frequently occupy discrete temporal likely not have the diversity to support Phase niches, with the understory species, or in some II sampling. Details of monitoring, including cases understory individuals, showing advanced frequency and replication, may be adjusted phenology relative to that of canopy-forming in- based on the initial data collected at each site dividuals (Richardson and O’Keefe 2009). and budgetary constraints. All site specific de- Additional species to be sampled for Phase II tails including site-specific modifications, species include up to two invasive species, and up to 5 selection, and targeted sampling windows will USA-NPN “campaign taxa” and/or Project Bud- be captured, tracked, and made available to burst (PBB) “10 most wanted” species, with the end users as part of the NEON phenology remaining 10 species at each site picked based sampling protocol (available through the NEON on rank abundance. These exceptions to the rank web portal; www.neonscience.org). abundance selection process are made to inten- tionally target species that either contribute to Applications of phenology data NEON’s ability to address its invasive species NEON plant phenology data will provide grand challenge questions or contribute to NE- foundational information about the variability ON’s ability to align data collection with existing in plant phenology across populations, com- national citizen science data collection efforts munities, and landscapes, which can be used (USA-NPN and PBB taxa). The large number of to validate remotely sensed land-surface phe- species selected should ensure that a diversity of nology products, and parameterize land-surface plant growth forms, invasives, and natives are models. Accurate representation of intra- and selected at sites where they are present, without inter-annual variability in vegetation phenology requiring any a priori definition of “functional is critical for correctly predicting net CO2 up- group”, a concept which is not yet well under- take (Desai 2010). Estimates of the vegetation

v www.esajournals.org 10 April 2016 v Volume 7(4) v Article e01303 SPECIAL FEATURE: NEON DESIGN ELMENDORF ET AL. start of season and end of season, key param- those predictions (Gelman and Hill 2007). Mul- eters in most land-surface models, are typically tisite, multispecies data sets, as well as extensive derived from remote sensing estimates or phys- local-scale climatological data are required for iological models based on chilling and forcing these types of models. NEON will expand the units (e.g., degree days). However, most satellite- taxonomic representation of phenological data, derived phenology estimates have not been measuring as many as 20 plant species at each validated using ground data (Fisher and of 47 sites outfitted with sensors that measure Mustard 2007), and realistic parameterization biophysical parameters. These data can form the of physiologically based phenological models basis of an expanded phenological modeling for wild species is limited to the very few framework across taxa and ecosystems. species for which relevant data are available A second avenue for upscaling phenological (Jeong et al. 2012). An evaluation of vegetation measurements at NEON sites is using in situ phenology in 14 terrestrial biosphere models measurements to validate or calibrate phenologi- found that for deciduous forests an early start cal measurements taken at broader spatial scales of season bias of 2 weeks or more was typical (e.g., phenocam- or satellite remote sensing across all models which resulted in a 13% over based-phenometrics). Successful scaling from estimate of gross ecosystem productivity ground observations to larger spatial scales typi- (Richardson et al. 2012). Such misrepresentation cally involves weighting species-specific phenol- of phenology has consequences beyond ecosys- ogy measurements by their coverage on the land- tem productivity estimates. When terrestrial and scape (see Liang et al. 2011, 2014, Melaas et al. atmospheric models are not properly coupled, 2016). Colocated plot-based measurements of reductions in temperature associated with the vegetation cover and structure, as well as vege- onset of leaf emergence and associated increases tation maps that can be built from NEON’s high- in transpiration are often misrepresented (Levis resolution hyperspectral and LiDAR remote and Bonan 2004). This insufficient coupling sensing data sets make NEON sites particularly during critical phenological stages can lead to well-suited to refining this type of scaling and errors in modeled microclimate and weather developing similar routines that can be applied patterns, and thus present cascading effects on in a diversity of ecosystem types. In addition other model components. High quality, long- to the human-based observations detailed here, term, standardized phenological measurements NEON will collect landscape images multiple across major ecosystem types will be critical times per day using stationary cameras (pheno- components for improving model development cams) mounted on each flux tower. These data and accuracy. give a digital record of the seasonality of green- Quantifying the range of phenological re- ing and browning over larger scales. For maxi- sponses across a diversity of species and sites mal interoperability, NEON phenocam installa- will aid in upscaling phenology measurements tion and programming follows the PhenoCam from the level of individual plants to communi- Network protocols (Richardson and Klosterman ties and ecosystems. One approach to upscaling 2015). Additional information on the timing of phenological data is through the development plant phenology can be informed by NEON’s bi- of more accurate phenological forcing models, weekly collection of leaf area index (LAI) digital as well as quantifying the uncertainty in phe- hemispherical photos within the tower airshed nology estimated from such models for sites and carbon flux estimates processed at half-hour and locations where direct measurements are intervals from the instrumented tower. These not available. Bayesian hierarchical models are data streams, augmented with annual submeter a promising avenue forward in community phe- hyperspectral and LiDAR remote sensing data nology forecasting (see Ibáñez et al. 2010, Diez will be valuable in determining statistical and et al. 2012, for examples, applied to individual mechanistic associations between aboveground, sites with multiple taxa, or single taxa measured belowground, and landscape scale seasonal dy- across multiple sites). Hierarchical models can namics. be leveraged to generate predictions for new An ultimate goal includes not only upscaling species or locations, as well as uncertainties on of ground-based measurements but also using

v www.esajournals.org 11 April 2016 v Volume 7(4) v Article e01303 SPECIAL FEATURE: NEON DESIGN ELMENDORF ET AL. both ground- and larger scale measurements to phenological asynchrony between interacting down-scale from larger scale greening indices to species and potential consequences to shifts in guide local-scale decision-making. Phenological overlapping activity periods throughout the data are used in a number of natural resource duration of the observatory. management activities (Enquist et al. 2014). Ac- The development of integrated, interoperable curate phenological forecasts or real-time pheno- data sets will enhance the utility of data collect- logical tracking can aid land managers in timing ed by NEON and other programs. A number of controlled burns, mechanical harvesting, pesti- other programs (e.g., USA National Phenology cide, and/or herbicide applications for maximum Network (https://www.usanpn.org/), Project efficiency in controlling invasive species. Data on Budburst (budburst.org), Long Term Ecological seasonal growth and senescence patterns can in- Research (LTER) Network sites (http://www.lter- form wildfire predictions. Similarly, information net.edu/), National Parks (http://science.nature. on peak flowering and leaf color change dates nps.gov/im/monitor/), the Pan European Phenol- can help promote and plan for seasonal tourism ogy Project (PEP725; http://www.pep725.eu/)), as coincident with wildflower or fall foliage view- well as multiple long-term, PI-directed research ing. Last, recent studies theorize that a species’ projects also take phenology measurements. ability to make appropriate phenological adjust- NEON data will augment and compliment these ments to a changing climate may be predictive efforts, providing replication and longevity of of its future success in a changing climate (Wil- measurements that are difficult to achieve with- lis et al. 2010, Pau et al. 2011). This suggests that out a centralized source of funding. Because of an improved understanding of species-specific NEON’s planned infrastructure, its potential to phenological sensitivities could be used to link ground-based measurements, landscape identify particularly vulnerable native taxa for green-up and brown-down metrics, and ecosys- protection, or prioritize invasive species for tem processes is unique (Keller et al. 2008). removal. One limitation of the NEON design for The dominant species in all plant communi- phenology is that the financial and logistical ties generally represent key resources for ani- commitment required to measure phenology mals that depend on them for food or shelter. alongside a large suite of other parameters Consequently, phenological shifts in the onset, (see Lunch (2014) for the full list of NEON duration, and abundance of vegetative and re- data products) constrains the total number of productive resources detected by phenological NEON sites. As a result, NEON sites are spa- monitoring program can alert resource man- tially sparse compared to continent-wide cit- agers of changes that may affect the commu- izen science observation efforts, such as the nity composition, population dynamics, and USA-NPN, Project BudBurst and affiliated persistence of insects, pollinators, birds, and national and regional monitoring networks. mammals at site or regional scales. This goal Because NEON uses nationally standardized requires monitoring of the animals that interact protocols, however, data from the intensively with the focal plant species at NEON sites. In studied NEON sites can be readily combined addition to the plant phenology observations with existing and ongoing efforts to facilitate described here, terrestrial protocols that con- continental-scale analysis and forecasting. To tribute to phenological monitoring at NEON further this effort, an international group of sites include trapping of (1) mosquitoes and (2) phenology researchers and computer scientists small mammals throughout the active grow- is developing an ontology for plant ontology, ing season, These data may be used to quanti- with the aim of annotating diverse datasets to fy phenology of mosquito emergence and an- facilitate data discovery and integration. By nual population dynamics and small mammal combining ground-based observations with reproductive periods, respectively (Thibault other North American plant phenological 2014, Hoekman et al. in press). Integration of monitoring programs, existing data sets (e.g., NEON phenology data with surveillance data Wolkovich et al. 2012), the PhenoCam network on other taxa, conducted either by NEON or (http://phenocam.sr.unh.edu/webcam/), satel- by PIs working at NEON sites, can help track lite imagery (e.g., MODIS land cover dynam-

v www.esajournals.org 12 April 2016 v Volume 7(4) v Article e01303 SPECIAL FEATURE: NEON DESIGN ELMENDORF ET AL. ics http://modis.gsfc.nasa.gov/data/dataprod/), is based upon work supported by the National Science and/or models, in situ phenology observations Foundation under Cooperative Service Agreement made by NEON can contribute important in- EF-1029808. puts to an annual “green wave” (Schwartz 1998, Ault et al. 2015) projection over the con- Literature Cited tinent. On a more local scale, phenology field observations, phenocams, remote sensing, and AIBS. 2000. Report to the National Science Foundation temperature and precipitation data can be used from the second workshop on the development together to understand the drivers of phenolo- of a National Ecological Observatory Network gy of regionally important plant species to im- (NEON). San Diego Supercomputer Center, La prove range management practices (Browning Jolla, California, USA, March 9-13 2000. Available online at http://ibrcs.aibs.org/reports/pdf/NEON2_ et al. 2015). Mar2000.pdf Changes in plant phenology are widely re- Ault, T. R., A. K. Macalady, G. T. Pederson, J. L. Betan- garded as “fingerprints of climate change” or court, and M. D. Schwartz. 2011. Northern hemi- “climate change indicators” (e.g., U.S. Environ- sphere modes of variability and the timing of mental Protection Agency 2014); indeed, plant spring in western North America. Journal of Cli- phenology is an exemplary essential species trait mate 24:4003–4014. in the ongoing development of Essential Biodi- Ault, T. R., M. D. Schwartz, R. Zurita-Milla, J. F. Welt- versity Variables (EBV’s) targeted for internation- zin, and J. L. Betancourt. 2015. Trends and natural al monitoring (Pereira et al. 2013). Many of the variability of spring onset in the coterminous Unit- meteorological and atmospheric measurements ed States as evaluated by a new gridded dataset of at NEON sites are Essential Climate Variables spring indices. Journal of Climate 28:8363–8378. Barnett, D. 2015. TOS protocol and procedure: plant (Bojinski et al. 2014) and could facilitate em- diversity sampling. NEON document # NEON. pirical understanding of ecological responses DOC.014042. Available online at: http://data.neon- to change. Ongoing efforts both nationally and inc.org/documents. internationally (e.g., PEP725), will continue to Bojinski, S., M. Verstraete, T. C. Peterson, C. Richter, document patterns of plant phenology over large A. Simmons, and M. Zemp. 2014. The concept of spatial extents. Leveraging data from NEON will essential climate variables in support of climate enable the extrapolation not only of patterns of research, applications, and policy. Bulletin of the plant phenological shifts across the continent American Meteorological Society 95:1431–1443. (e.g., Ault et al. 2011, Jeong et al. 2013), but po- Bonan, G. B. 2008. Forests and climate change: forc- tentially also of the functional consequences of ings, feedbacks, and the climate benefits of forests. these shifts. Collocated measurements conduct- Science 320:1444–1449. Browning, D. M., A. Rango, J. W. Karl, C. M. Laney, E. ed by NEON will elucidate the degree to which R. Vivoni, and C. E. Tweedie. 2015. Emerging tech- plant phenological status is broadly indicative of nological and cultural shifts advancing drylands related ecosystem processes for which continent- research and management. Frontiers in Ecology wide data are sparse, such as below-ground and the Environment 13:52–60. phenology, carbon flux, seasonal biomass accu- Buckley, L. B., and J. G. Kingsolver. 2012. Functional mulation. In turn, the analysis, synthesis, and and phylogenetic approaches to forecasting spe- application of phenological information will fa- cies’ responses to climate change. Annual Review cilitate decision-making related to critical ecolog- of Ecology, Evolution, and Systematics 43:205–226. ical issues that affect societal well-being now and Chuine, I. 2000. A unified model for budburst of trees. into the future. Journal of Theoretical Biology 207:337–347. Chuine, I. 2010. Why does phenology drive species distribution? Philosophical Transactions of the cknowledgments A Royal Society B: Biological Sciences 365:3149–3160. Churkina, G., D. Schimel, B. H. Braswell, and X. Xiao. We thank Shirley Papuga, Yuri Springer, and Lee 2005. Spatial analysis of growing season length Stanish for helpful comments on the manuscript. control over net ecosystem exchange. Global Any use of trade, product, or firm names is for Change Biology 11:1777–1787. descriptive purposes only and does not imply en- Cleland, E. E., I. Chuine, A. Menzel, H. A. Mooney, and dorsement by the U.S. Government. This material M. D. Schwartz. 2007. Shifting plant phenology in

v www.esajournals.org 13 April 2016 v Volume 7(4) v Article e01303 SPECIAL FEATURE: NEON DESIGN ELMENDORF ET AL.

response to global change. Trends in Ecology & Hufkens, K., M. A. Friedl, T. F. Keenan, O. Sonnentag, Evolution 22:357–365. A. Bailey, J. O’Keefe, and A. D. Richardson. 2012. Cleland, E. E., J. M. Allen, T. M. Crimmins, J. A. Ecological impacts of a widespread frost event fol- Dunne, S. Pau, S. E. Travers, E. S. Zavaleta, and E. lowing early spring leaf-out. Global Change Biolo- M. Wolkovich. 2012. Phenological tracking enables gy 18:2365–2377. positive species responses to climate change. Ecol- Ibáñez, I., R. B. Primack, A. J. Miller-Rushing, E. Ell- ogy 93:1765–1771. wood, H. Higuchi, S. D. Lee, H. Kobori, and J. A. Si- Denny, E. G., K. L. Gerst, A. J. Miller-Rushing, G. L. lander. 2010. Forecasting phenology under global Tierney, T. M. Crimmins, C. A. Enquist, P. Guer- warming. Philosophical Transactions of the Royal tin, A. H. Rosemartin, M. D. Schwartz, and K. A. Society B: Biological Sciences 365:3247–3260. Thomas. 2014. Standardized phenology monitor- Inouye, D. W. 2008. Effects of climate change on phe- ing methods to track plant and animal activity for nology, frost damage, and floral abundance of science and resource management applications. In- montane wildflowers. Ecology 89:353–362. ternational Journal of Biometeorology 58:591–601. IPCC. 2013. Summary for Policymakers. Pages 3–29 in Desai, A. R. 2010. Climatic and phenological controls T. F. Stocker, D. Quin, G. K. Plattner, M. Tignor, S. on coherent regional interannual variability of car- K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex, and bon dioxide flux in a heterogeneous landscape. P. M. Midgley, editors. Climate Change 2013: The Journal of Geophysical Research 115:G00J02. Physical Science Basis. Contribution of Working Diez, J. M., I. Ibáñez, A. J. Miller-Rushing, S. J. Mazer, Group I to the Fifth Assessment Report of the In- T. M. Crimmins, M. A. Crimmins, C. D. Bertelsen, tergovernmental Panel on Climate Change. Cam- and D. W. Inouye. 2012. Forecasting phenology: bridge University Press, Cambridge, UK. from species variability to community patterns. Jeong, S.-J., D. Medvigy, E. Shevliakova, and S. Ma- Ecology Letters 15:545–553. lyshev. 2012. Uncertainties in terrestrial carbon Diez, J. M., I. Ibáñez, J. A. Jr Silander, R. Primack, H. budgets related to spring phenology. Journal of Higuchi, H. Kobori, A. Sen, and T. Y. James. 2014. Geophysical Research 117:G01030. Beyond seasonal climate: statistical estimation of Jeong, S.-J., D. Medvigy, E. Shevliakova, and S. Ma- phenological responses to weather. Ecological Ap- lyshev. 2013. Predicting changes in temperate plications 24:1793–1802. forest budburst using continental-scale observa- Enquist, C. A., J. L. Kellermann, K. L. Gerst, and A. J. tions and models. Geophysical Research Letters Miller-Rushing. 2014. Phenology research for nat- 40:619–632. ural resource management in the United States. Jolly, W. M., R. Nemani, and S. W. Running. 2005. A International Journal of Biometeorology 58:579– generalized, bioclimatic index to predict foliar phe- 589. nology in response to climate. Global Change Biol- Fisher, J. I., and J. F. Mustard. 2007. Cross-scalar ogy 11:619–632. satellite phenology from ground, Landsat, and Kao, R. H., C. M. Gibson, R. E. Gallery, C. L. Meier, D. T. MODIS data. Remote Sensing of Environment Barnett, K. M. Docherty, K. K. Blevins, P. D. Travers, 109:261–273. E. Azuaje, and Y. P. Springer. 2012. NEON terres- Fuccillo, K. K., T. M. Crimmins, C. E. de Rivera, and trial field observations: designing continental-scale, T. S. Elder. 2014. Assessing accuracy in citizen standardized sampling. Ecosphere 3:115. science-based plant phenology monitoring. Inter- Keller, M., D. S. Schimel, W. W. Hargrove, and F. M. national Journal of Biometeorology 59:917–926. Hoffman. 2008. A continental strategy for the Na- Ganguly, S., M. A. Friedl, B. Tan, X. Zhang, and M. Ver- tional Ecological Observatory Network. Frontiers ma. 2010. Land surface phenology from MODIS: in Ecology and the Environment 6:282–284. characterization of the Collection 5 global land Koch, E., E. Bruns, F. M. Chmielewski, C. Defila, W. cover dynamics product. Remote Sensing of Envi- Lipa, and A. Menzel. 2007. Guidelines for plant ronment 114:1805–1816. phenological observations. World Climate Data Gelman, A., and J. Hill. 2007. Data analysis using regres- and Monitoring Programme, Geneva, Switzerland. sion and multilevel/hierarchical models. Cambridge Levis, S., and G. B. Bonan. 2004. Simulating spring- University Press, New York, New York, USA. time temperature patterns in the community at- Grime, J. P. 1998. Benefits of plant diversity to ecosys- mosphere model coupled to the community land tems: immediate, filter and founder effects. Journal model using prognostic leaf area. Journal of Cli- of Ecology 86:902–910. mate 17:4531–4540. Hoekman, D., et al. in press. Design for mosquito abun- Liang, L. A., M. D. Schwartz, and S. L. Fei. 2011. Val- dance, diversity, and phenology sampling within idating satellite phenology through intensive the National Ecological Observatory Network. ground observation and landscape scaling in a

v www.esajournals.org 14 April 2016 v Volume 7(4) v Article e01303 SPECIAL FEATURE: NEON DESIGN ELMENDORF ET AL.

mixed seasonal forest. Remote Sensing of Environ- Morisette, J. T., A. D. Richardson, A. K. Knapp, J.I. ment 115:143–157. Fisher, E. A. Graham, J. Abatzoglou, B. E. Wilson, Liang, L., M. D. Schwartz, Z. Wang, F. Gao, C. B. D. D. Breshears, G. M. Henebry, and J. M. Hanes. Schaaf, B. Tan, J. T. Morisette, and X. Zhang. 2014. 2008. Tracking the rhythm of the seasons in the face A cross comparison of spatiotemporally enhanced of global change: phenological research in the 21st springtime phenological measurements from sat- century. Frontiers in Ecology and the Environment ellites and ground in a northern U.S. mixed for- 7:253–260. est. IEEE Transactions on Geoscience and Remote Nguyen, H. T., O. Kosheleva, V. Kreinvich, and S. Sensing 52:7513–7526. Ferson. 2009. Trade-off between sample size and Lunch, C. 2014. NEON Level 1, Level 2 and Level 3 accuracy: case of measurements under interval Data Products Catalog. NEON document # NEON. uncertainty. International Journal of Approximate DOC.002652. Available online at: http://www.neo- Reasoning 50:1164–1176. ninc.org/ Parmesan, C., and G. Yohe. 2003. A globally coherent Mazer, S. J., S. E. Travers, B. I. Cook, T. J. Davies, K. fingerprint of climate change impacts across natu- Bolmgren, N. J. B. Kraft, N. Salamin, and D. W. In- ral systems. Nature 421:37–42. ouye. 2013. Flowering date of taxonomic families Pau, S., E. M. Wolkovich, B. I. Cook, T. J. Davies, N. J. predicts phenological sensitivity to temperature: B. Kraft, K. Bolmgren, J. L. Betancourt, and E. E. implications for forecasting the effects of climate Cleland. 2011. Predicting phenology by integrat- change on unstudied taxa. American Journal of ing ecology, evolution and climate science. Global Botany 100:1–17. Change Biology 17:3633–3643. Mazer, S. J., K. L. Gerst, E. R. Matthews, and A. Even- Pereira, H. M., et al. 2013. Essential biodiversity vari- den. 2015. Species-specific phenological responses ables. Science 339:277–278. to winter temperature and precipitation in a water- Richardson, A., and S. Klosterman. 2015. Phenocam limited ecosystem. Ecosphere 6:98. installation instructions. Available online at: http:// McKinney, A. M., P. J. CaraDonna, D. W. Inouye, phenocam.sr.unh.edu/pdf/PhenoCam_Install_In- B. Barr, C. D. Bertelsen, and N. M. Waser. 2012. structions.pdf Asynchronous changes in phenology of migrat- Richardson, A., and J. O’Keefe. 2009. Phenological ing Broad-tailed Hummingbirds and their early- differences between understory and overstory: a season nectar resources. Ecology 93:1987–1993. case study using the Long-Term Harvard Forest Meier, U. 2001. Growth stages of mono-and dicotyle- Records. Pages 88–117 in A. Noormets, editor. Phe- donous plants. BBCH Monograph. Federal Biolog- nology of ecosystem processes. Springer-Verlag, ical Research Centre for Agriculture and Forestry, New York, New York, USA. Germany. Richardson, A. D., et al. 2010. Influence of spring Meier, C., and K. Jones. 2014. TOS science design for and autumn phenological transitions on forest plant biomass, productivity, and leaf area index. ecosystem productivity. Philosophical Transac- NEON document # NEON.DOC.000914. Available tions of the Royal Society B: Biological Sciences online at: http://data.neoninc.org/documents. 365:3227–3246. Meier, C., and K. Jones. 2015. TOS protocol and proce- Richardson, A. D., et al. 2012. Terrestrial biosphere dure: measurement of vegetation structure. NEON models need better representation of vegetation document # NEON.DOC.000987. Available online phenology: results from the North American Car- at: http://data.neoninc.org/documents. bon Program site synthesis. Global Change Biolo- Melaas, E. K., M. A. Friedl, and A. D. Richardson. 2016. gy 18:566–584. Multi-scale modeling of spring phenology across Schimel, D. S., et al. 2011. The National Ecological Deciduous Forests in the Eastern United States. Network 2011 Science Strategy: enabling continen- Global Change Biology 22:792–805. tal-scale ecological forecasting. Melillo, J. M., T. Richmond, and G. W. Yohe, editors. Schwartz, M. D. 1998. Green-wave phenology. Nature 2014. Climate change impacts in the United States: 394:839–840. the third national climate assessment. U.S. Global Schwartz, M. D., and J. M. Hanes. 2010. Intercompar- Change Research Program, Washington, D.C., USA. ing multiple measures of the onset of spring in Miller-Rushing, A. J., D. W. Inouye, and R. B. Primack. eastern North America. International Journal of 2008. How well do first flowering dates measure Climatology 30:1614–1626. plant responses to climate change? The effects of Schwartz, M. D., R. Ahas, and A. Aasa. 2006. Onset of population size and sampling frequency. Journal of spring starting earlier across the Northern Hemi- Ecology 96:1289–1296. sphere. Global Change Biology 12:343–351.

v www.esajournals.org 15 April 2016 v Volume 7(4) v Article e01303 SPECIAL FEATURE: NEON DESIGN ELMENDORF ET AL.

Schwartz, M. D., J. L. Betancourt, and J. F. Weltzin. USA-NPN National Coordinating Office. 2012. Identi- 2012. From Caprio’s lilacs to the USA National Phe- fying and prioritizing data products and tools for nology Network. Frontiers in Ecology and the En- use in science, management, and decision making. vironment 10:324–327. Research Coordination Network workshop report. Schwartz, M. D., J. M. Hanes, and L. Liang. 2014. Sep- USA-NPN Programmatic Series 2012-002. http:// arating temperature from other factors in pheno- www.usanpn.org. logical measurements. International Journal of White, M. A., D. Beurs, M. Kirsten, K. Didan, D. W. Biometeorology 58:1699–1704. Inouye, A. Richardson, O. P. Jensen, J. O’Keefe, Singer, M. C., and C. Parmesan. 2010. Phenological G. Zhang, and R. R. Nemani. 2009. Intercompari- asynchrony between herbivorous insects and their son, interpretation, and assessment of spring phe- hosts: signal of climate change or pre-existing adap- nology in North America estimated from remote tive strategy? Philosophical Transactions of the sensing for 1982–2006. Global Change Biology Royal Society B: Biological Sciences 365:3161–3176. 15:2335–2359. Strode, P. K. 2003. Implications of climate change for Willis, C. G., B. R. Ruhfel, R. B. Primack, A. J. Mill- North American wood warblers (Parulidae). Glob- er-Rushing, J. B. Losos, and C. C. Davis. 2010. al Change Biology 9:1137–1144. Favorable climate change response explains non- Thibault, K. M. 2014. TOS science design for small native species’ success in Thoreau’s woods. PLoS mammal abundance and diversity. NEON docu- ONE 5:e8878. ment # NEON.DOC.000915. Available online at: Wolkovich, E. M., B. I. Cook, J. M. Allen, T. M. http://data.neoninc.org/documents. Crimmins, J. L. Betancourt, S. E. Travers, S. Pau, U.S. Environmental Protection Agency. 2014. Climate J. Regetz, T. J. Davies, and N. J. B. Kraft. 2012. change indicators in the United States, 2014. Third Warming experiments underpredict plant phe- edition. EPA 430-R-14-004. www.epa.gov/climat- nological responses to climate change. Nature echange/indicators. 485:494–497.

v www.esajournals.org 16 April 2016 v Volume 7(4) v Article e01303