Focus on recent, present and future Arctic and boreal productivity and biomass changes

The reduction of cold temperature constraints on photosynthesis in recent decades has led to extended growing seasons and increased plant productivity (greening) in significant parts of Polar, Arctic and Boreal regions, here called northern lands. However, most territories within these regions display stable productivity in recent years. Smaller portions of Arctic and Boreal regions show reduced productivity (browning). Summer drought and wildfires are the best documented drivers causing browning of continental areas. Yet factors like winter warming events dampening the greening effect of more maritime regions have remained elusive, least monitored and least understood. A Norway-US network project called ArcticBiomass was launched in 2013 to further reveal both positive and negative effects of climate change on biomass in Arctic and Boreal regions. This focus collection named Focus on Recent, Present and Future Arctic and Boreal Productivity and Biomass Changes includes 24 articles and is an important outcome of this work and addresses recent changes in phenology, biomass and productivity and the mechanisms. These mechanisms include former human interactions (legacies) and drivers that control such changes (both greening and browning), along with consequences for local, regional and global scale processes. We complete our synthesis by stressing remaining challenges and knowledge gaps, and provide an outlook on future needs and research questions in the study of climate and human driven interactions in terrestrial Arctic and Boreal ecosystems.


Introduction
Ecosystem responses to Arctic warming have the potential to feedback either positively or negatively to the Earth's climate system depending on latitude, changes in disturbance regime, vegetation distribution and productivity (McGuire et al 2009). The lower albedo of shrub and forest vegetation compared with tundra, for example, results in a positive feedback on temperature (Bala et al 2007). Global warming presented as a warming rate, e.g. 0.2°C/decade (ACIA 2004), does not reflect the fact that cold seasons are warming faster than the warm seasons, especially in the Arctic due to positive feedbacks (e.g. albedotemperature feedback (Meredith et al 2019)). The initiation, termination and performance of many biological processes, e.g. plant growth, are tied to threshold temperatures. The trend in timing of these thresholds, and cumulative air temperatures driving them, may have the effect of enhancing vegetation productivity.
Conversely, increased productivity of Arctic vegetation resulting from warmer temperatures tends to result in increased carbon dioxide (CO 2 ) uptake by net photosynthesis, providing a negative feedback to rising temperatures (Field et al 2007, Speed et al 2010. As a result of this trend, between 1982 and 2011, Arctic tundra vegetation increased both in terms of peak productivity, greening and growing season length and this finding is supported by a wide range of field site mea- . These insights are not obvious from measurements of warming rates alone, and the relative importance of these competing feedbacks. Thus, the cumulative effect of changing Arctic vegetation on the climate system and CO 2 fluxes, is still not very well known, particularly in Svalbard (Speed et al 2010), and Fennoscandia (Väisänen et al 2014). Drought can modify these feedback effects, decoupling warming and productivity as well as the balance of gross photosynthesis and plant respiration, which varies substantially across plant functional types (Chapin et al 1996).
The only effective way to map biomass and plant productivity in such large and remote areas as Alaska and Svalbard, as well as in other territories of the Arctic, is using remote sensing together with necessary in situ measurements and observations (Karlsen et al 2009).
The aim of our synthesis is to provide an overview of the studies in this focus issue and to place their findings within the broader context of ecosystem-climate dynamics. Our overview is organized into four sections that focus on studies that were primarily relevant to: (1) Changes in the physical environment over high latitude regions and associated ecological and phenological changes in Arctic/Boreal vegetation, including vegetation-mediated responses and climate feedbacks; (2) Actual and potential biomass change influenced by (local) climate, natural disturbances, human impacts (e.g. resource extraction and legacies) and impacts on humans; (3) Transformation of open tundra vegetation to tall shrub tundra or forests, due to warming and other processes, influencing local and global climate, albedo and climate feedback mechanisms; and (4) Integration of in situ observations and manipulation experiments including remote sensing and other data sources to advance methodological approaches for measuring and monitoring.
We then discuss the key collective advances made in each of these areas by the studies within this focus issue, as well as opportunities for future research.

Northern lands (45 0 N)
The synthesis comprise studies conducted north of 45°N comprising both the Boreal and Arctic zones and in figure 1 we can see the areas that were covered by the different studies. In order to define both regions, the latest version of the MODIS International Geosphere-Biosphere Programme (IGBP) land cover map  (270). In spring, the correlation between the first snow-free day mapped by MODIS data and snow data from 36 of 40 meteorological stations was highly significant (p<0.05), with a of bias of less than 10 d for 34 of the stations. In autumn, 31 of 40 stations show highly significant (p<0.05) correlation with MODIS data, and the bias was less than 10 d for 27 of the stations.
Park et al (2016) analyzed satellite observations during 33 years in order to assess changes in growing season metrics (onset: SOS, end: EOS and length: LOS) and seasonal total gross primary productivity using the growing season summed NDVI (GSSNDVI). They found that LOS had lengthened by 2.60 d dec −1 (p<0.05) due to an earlier onset of SOS (−1.61 d dec −1 , p<0.05) and a delayed EOS (0.67 d dec −1 , p<0.1) the past three decades. The changes in growing season metrics were stronger in Eurasia (EA) and in boreal regions compared to North America (NA) and the arctic regions.
Reports on tundra greening are numerous and varied, and include increases in summer, spring, and autumn temperatures, as well as growing season length  (2016) made use of 30 years of a 1 km spatial resolution dataset (AVHRR) from 1986 to 2015 to calculate annual maximum NDVI over parts of Svalbard, and they found positive trends in maximum NDVI (+29%) and average summer temperature (+59%), which were significantly positively correlated with each other. On the same high arctic island, Karlsen et al (2018) found a a non-significant increase of 15% for the integrated MODIS NDVI defined as OP NDVI (onset to peak NDVI) for the period 2000-2014. At both local and regional scales, the latter NDVI-measure was found to predict biomass.
However, an increasing number of northern regions currently show declining productivity (browning). Such trends are evident both in North America and Eurasia, and factors assumed to contribute to this decline include recent reductions in summer moisture (see Verbyla 2015). Browning has also been attributed to mire and pond formation from increasing precipitation in some regions te Beest et al (2016) found that when reindeer reduce shrub abundance/cover and height, summer albedo increases in both Betula nana-dominated heath vegetation and Salix glauca-dominated shrub tundra. Model results reveal associated lower net radiation, together with latent and sensible heat fluxes in heavilygrazed areas in all shrub-dominated vegetation types. Hence reindeer have a potential cooling effect on climate by increasing summer albedo and decreasing net radiation. Herbivory may have long-term consequences in warmed and ambient high Arctic tundra according to the study by Little et al (2017). They found that significantly more dead vascular plant material was found within warmed open-top chambers during a 12 year period compared to ambient plots, regardless of grazing history, but in contrast to many short-term experiments no difference in the amount of living material was found.
Belowground plant biomass allocation in tundra ecosystems and its relationship with temperature is of importance for modelling and analysis of climate change (Wang et al 2016). They found that plant community biomass-temperature relationships were significantly different between above and belowground biomass. Tundra ecosystems through altered litter input and distribution in the soil, as well as possible changes in root turnover.
The timing and duration of different pheno-phases within a plant's life cycle are critical for plant performance and growth. In the High Arctic, the start of many of these phenological phases is determined by the start of snowmelt, which can change in a changing climate (Semenchuk et al 2016). In order to assess any change in these phases, Semenchuk et al (2016) tested if snowmelt data control the timing and duration of phenological periods in Svalbard using a timing gradient from natural to experimentally altered snow depths. All pheno-phases followed irrespective of timing of occurrence, vegetative or reproductive nature, and three of the four phenological periods were fixed for most species such as the indigenous, hence the last was aperiodic. Semenchuk et al (2016) concluded that periodic species like the dwarf shrubs Dryas octopetala and Cassiope tetragona and the grass species Luzula arcuata are likely to be limited in their ability to adapt to changing snowmelt dates, and thus may be disadvantaged compared to some of the invading species on Svalbard such as the herb Rumex longifolius the grass Deschampsia cespitosa. Gillespie et al (2016) studied the plant-pollinator interactions in High Arctic (Svalbard) and in their study they altered the timing of flowering phenology, using snow fences and opentop chambers. As expected, deep snow plots delay snow melt timing and this in turn delay the first and peak flowering dates of the plants, hence shortened the prefloration period overall. The OTCs, however, counteracted the delay in first and peak flowering to some extent. There was no effect of treatment on length of flowering season, although for all variables there were species-specific responses. The insect flower-visitor community was species poor, and although evidence of disruption to phenological overlaps was not found, reduced insect-flower visitation rates to flowers in plots with deep snow may have occurred, due to limited observation methods.
3.2. Actual and potential biomass change and productivity influenced by (local) climate, natural disturbances, human impacts (e.g. resource extraction and legacies) and impacts on humans Bjerke et al (2015) detail the effects of above ground ice accumulation on meadow productivity by using ground observation and remotely sensed data (GIMMS NDVI3g data). Five contrasting snow season types were identified; snow-rich season with no soil frost or no ground-ice through low snow and considerable soil frost and ground-ice. Conditions of shallow snow depth and shallow soil frost or aboveground ice formation which may be a result of more frequent warming events, are rare at present but are predicted to become the dominant snow season type in Low Arctic and the Boreal lowlands. Agricultural productivity was lowest after winters with high accumulation of plant-damaging, hermetic above-ground ice formation Deep soil frost by itself did not reduce primary productivity. Loranty et al (2016) quantified the distribution of vegetation productivity trends, wildfire, and near-surface soil carbon, according to vegetation type, across the continuous and discontinuous permafrost. Zones. They observed positive trends in vegetation productivity in areas of continuous permafrost, whereas areas underlain by discontinuous permafrost have proportionally less positive productivity trends.

Transformation of open tundra vegetation to
shrub tundra or forests, due to warming and other processes, influencing local and global climate, albedo and climate feedback mechanisms. growth and expansion. Evidence for a suite of 23 proximal controls that operate directly on shrub growth and expansion was found, while the evidence base was only focused on just four controls like air temperature, snow dynamics, soil moisture and herbivory. In particular, 65% of the evidence was generated within the warmest tundra climes (i.e. Low Arctic), while 24% of the evidence was from only one of 28 floristic sectors, indicating huge gaps (Martin et al 2017) in the available evidence (figure 1) in, so not comprehensive enough to answer key questions concerning Pan-Arctic shrub change.
Significant expansion of shrubs has been observed within Arctic and Boreal regions during recent decades and in eastern sub-arctic Canada, where dendrochronological studies have demonstrated that the majority of shrub stands sampled were young, since the dominant stems were developed after 1990 (Paradis et al 2016) which seem to be in accordance with Park et al (2016). Stratified sampling of shrubs revealed that woody biomass was maximal within the lower canopy stratum, whereas foliar biomass tracked the development of the respective stands' vertical structure. Shrub height as a parameter explained snow depth, winter ground level temperature and summed freezing-degree days, while woody biomass best explained summer ground level temperature. Hence, shrub canopy structure will exert significant control on the abiotic environment in subarctic ecosystems (Paradis et al 2016).
Shrub expansion is more likely to occur in areas with high soil moisture and nutrient availability, conditions typically found in sub-surface water channels called water tracks. Curasi et al (2016) quantified the distribution of water tracks and their contribution CO 2 dynamics during the growing season for a Siberian tundra landscape using field measurements, satellite observations and meteorological data. They found that water tracks occupied 7.4% of the study area, and account for a slightly larger proportion of growing season carbon uptake relative to surrounding tundra and also larger relative to graminoids within the same water tracks. Water tracks are an important component of this landscape and they will influence ecosystem structural and functional responses to climate, and is therefore of importance for modeling.
Increased wetness confounded Landsat-derived NDVI trends central Alaska North Slope region (Raynolds and Walker 2016) during 1985-2011. Regional trends showed decreases in NDVI for most vegetation types, but increases in tasseled-cap greenness, greatest for shrub dominated vegetation, and tasseled-cap wetness. This was consistent with thawing of polygon ice wedges. Increasing cover of water may be masking increases in vegetation when summarized using the more water-sensitive NDVI. This is also consistent with reduced NDVI due to melting of tundra as well as mire and pond formation from increasing precipitation in some regions of Northwest Siberia (Miles and Esau 2016), which show that care should be taken when relying solely on NDVI data (Raynolds and Walker 2016).
Disturbances can have particularly large effects on Arctic ecosystems when ecosystem structure and function are controlled by strong feedbacks between soil conditions, vegetation moisture, and sub-surface thermal regime (Cameron and Lantz 2016). An example of such disturbance includes e.g. road construction and maintenance on vegetation structure and biomass along the Dempster Highway. Using field data and very high-resolution remotely sensed data, Cameron and Lantz (2016) found that increased shrub proliferation adjacent to the road was caused by greater soil moisture.
3.4. Integration of in situ observations and manipulation experiments including remote sensing and other data sources to advance methodological approaches for measuring and monitoring. Buchhorn et al (2016) showed that satellites provide the only practical source of data for estimating biomass of large and remote areas of the Arctic. Researchers have found that the normalized difference vegetation index (NDVI) correlates well with biomass sampled on the ground. However, errors in NDVI and biomass estimates due to bidirectional reflectance distribution function (BRDF) effects are not well reported. Finally, they found that studies that only sampling a narrow range of biomass and NDVI may produce equations that are more difficult to correct for BRDF effects.
Brazhnik and Shugart (2015) applied the new spatially-explicit gap-dynamics model SIBBORK towards a better understanding of how transition zones, namely treelines or forest lines, which are under-sampled and difficult to model, may change in the near future. They found that a 2°C change in annual average air temperature will significantly alter the structure, composition, and productivity of boreal forest stands both at the northern treeline by 2040, and at the southern treeline by 2050.
Juutinen et al (2017) assessed the spatial variation and seasonal dynamics of leaf-area index (LAI) linking ground observations and very-high-spatial resolution multispectral satellite images (e.g. Worldview-2). They illustrated how the short growing season, rapid development of the LAI, yearly climatic variation, and timing of satellite data should be accounted for in matching imagery and field verification data in the Arctic. Among the main plant functional types, graminoid LAI displayed the largest seasonal amplitudes and was the main cause of varying NDVI spatial patterning.
The utilization of earth observation data in vegetation monitoring is highly dependent on a long heritage of ground-based observations in the Arctic (Walker et al 2016). Several products of the Conservation of Arctic Flora and Fauna are key to our current understanding (Shuchman et al 2015, Christensen et al 2013). They have concluded that there is an urgent need for more consistent standards of plot-based observations and recommend improvements regarding the linkage between plot-based observations biodiversity studies and satellite-based observations of Arctic vegetation.

New tools
Bratsch et al (2017) assessed the ability of hyperspectral remote sensing data to estimate low arctic tundra biomass in Alaska. The main result was that the ability to identify unique biomass-spectra relationships within respective vegetation types using hyperspectral sensors was decreased during the peak of the growing season, since shrubs obscure lower-statured, bryophyte-dominated vegetation types. Hence, this study supports previous studies that shrubs control the spectral reflectance in Low Arctic communities (see also Forbes et al 2010).

Further work and concluding remarks
Winter warming induced damage (arctic browning), particularly on dwarf-shrub vegetation (e.g. Cassiope tetragona) should be followed up via field monitoring and remote sensing (including UAV) in the future. Other vegetation changes in Arctic tundra and boreal regions, including the warming and thawing of permafrost (see Loranty et al 2016, 2018), should be followed up. UAV based sensors and satellite sensors with very high spatial resolution (30-50 cm) may be efficient tools for detection of coastal erosion, human impacts (tourism), changes in tundra vegetation cover/stature, as well as permafrost thaw, and should be followed annually in the future. UAV can also be used as an upscaling tool. Another problem is that there exists huge gaps (Martin et al 2017) in the Arctic and Boreal regions concerning long-term monitoring sites, including large areas that are not monitored at all (figure 1). These gaps currently afford us insufficient evidence of what is transpiring ecologically within these regions. However, a recent paper published by Virkkala et al (2019) in ERL provide detailed maps of potential new sampling locations in Arctic and the northern boreal region (the land north of the Arctic Circle (66,5 o N)), hence help prioritize future research efforts concerning environmental change in the north.