4. Discussion
This study first revealed the new knowledge about the development mode of ecosystem-level productivity over the NH in the first decade of this century, e.g., the productivity of the LEP- and HEP-kind ecosystems behaving with weaker tendencies of increasing than that of the whole NH ecosystem. Actually, it is hard to acquire such properties of the LEP and HEP ecosystems in the traditional ecology. In addition to such brief rules of EP trends, more EP characteristics were also derived in details. For the NA, its middle-latitudes performed with comparable EP development modes between its LEP (0.15 year-1), AEP (0.15 year-1), and HEP (0.14 year-1) ecosystems, but its high-latitudes exhibited a distinctive mode – LEP (0.05 year-1) weaker than AEP (0.16 year-1), far weaker than HEP (0.20 year-1). This meant that the components of the middle-latitude ecosystems in the NA had relatively similar modes in productivity increasing, but the ecosystem components at the high-latitudes behaved in a different mode –in productivity increasing the LEP ecosystems were weak whereas the HEP ecosystems were strong. For the NE, its middle- and high-latitudes demonstrated the comparative EP development modes but slightly weaker and stronger amplitudes between the LEP, AEP, and HEP ecosystems (0.28 vs. 0.36 year-1; 0.30 vs. 0.39 year-1; 0.27 vs. 0.35 year-1, respectively). North Europe and North Asia performed with discrepant development modes for the EP of the LEP, AEP, and HEP ecosystems (0.25 vs. 0.34 year-1; 0.26 vs. 0.37 year-1; 0.25 vs. 0.34 year-1), a little like the opposites of the middle- and high-latitudes of NE. In a whole sense, in EP development the NE region was not only more intense but also more consistent than the NA.
We found that the development of the Arctic circumpolar EP can be statistically attributed to the inter-annual variations of SP, and this gave a confirming answer to the first question. Our results pointed that the INDVI-characterized EP was more closely associated to snow-onset than -end SP. Three potential mechanisms, in terms of light (temperature), nutrient, and water as the three kinds of fundamental resources required by all plants, might account for these results. First, compared to snow-end SP, snow-onset SP is more indicative to its insulation capacity, which may decide the temperatures experienced by soils and plants during winter (Sturm et al., 1997) and further influences ecosystem productivity in the growing season (Billings & Bliss 1959), as evidenced by that altered snow density may affect plant growths (Rixen et al., 2008). Second, compared to snow-end SP, snow-onset SP is more related with its thickness and cover. In some cases, increased snow covers can be anticipated to add plant nitrogen concentrations (Aerts et al., 2009), which proved to be able to determine vegetation productivity eventually (Walsh et al.,1997). Third, in contrast to snow-end SP, snow-onset SP tends to be able to better project the variable of snow water equivalent, which serves as an essentially dependent factor of supporting ecosystem growth and productivity (Mark et al., 2015). The underlying hydrological process can also refer to the disentangled mechanisms of winter SP shifting plant activity in northern ecosystems (Wang et al.,2018). The finding as reviewed above can help to establish the theoretical basis for developing the EP-SP ecology.
We answered the second question that the Arctic circumpolar EP, indeed, was more sensitive to the SP around, and further, we noticed that this sensitivity occurred more to north for the NA but more to south for the NE. In effect, the findings have basically validated our proposed conceptual framework – periconnection – for revealing more special traits in ecology. But in contrast to its initial proposal based on a simple analogy just from one scheme-similar case, the discussion of periconnection here needed to cover the potential causes as many as possible for its theoretical establishment eventually in the future studies. Substantially, as a derivative from “teleconnection”, “periconnection” shall feature the same theoretical bases, involving solar radiative transfers, land surface-atmosphere interactions, and atmospheric circulations (Mikolajewicz et al., 1997; Allen & Luptowitz 2017; Mamalakis et al., 2018). The principles in macroecology can somehow explain the “heterogeneous-response”-causing eco-effects such as the spatial-heterogeneity effect (Chuine 2000) and the force-accumulation effect (Walsh et al., 1997). Further, the possible principles in specific reasoning here may involve the common drivers making simultaneous effects to EP and SP, e.g., temperature decreasing driven by wind reaches the thresholds of triggering the chilling effect (He et al., 2018), and the characterized directional eco-effect can just verify this hypothesis to some extent. Totally, the scientific essence of the neighborhood eco-effect is characterization of the time-lagging eco-effect (Walsh et al.,1997; Chuine 2000; Wu et al., 2015; Lin & Jiang 2017), alternatively in a spatial way here. As regard to the directional eco-effect, the phenomena that the NE and NA had contrasting ecosystem responses to recent climate changes and fire disturbances at the northern high latitudes were observed, based on remote sensing data and atmospheric transport model simulations (Goetz et al., 2007). In other words, atmospheric transports may be a key kind of external triggers of such directional eco-effects, and hence, taking atmospheric circulations into account may somehow explain why the EP-SP sensitivities occurred more to north for the NA but more to south for the NE. In addition, earlier experiments showed that the circulation response to maximum Eurasian snowfall is focused downstream in early winter, whereas upstream response is more evident from the North American experiments (Henderson et al., 2013). Such mechanisms may lay a more solid theoretical foundation for periconnection. However, all of these explanations are still hypothetical inferences, and more following endeavors shall be made for determining the real causes.
In summary, compared to the endeavors in attempting to attribute the interannual EP anomalies to the interannual SP variations based on direct correlations between the same geo-grids (Vaganov et al.,1999; Peng et al., 2013; Wang et al., 2018), this study taking the SP around into account as well is relatively rare. The new findings can help people to make a novel step toward more comprehensively understanding how climate changes affect the developments of ecosystems. Our inferences can increase some new macroecological information for improving the performance of ecosystem-climate-interaction modules in various dynamic global vegetation models (McCarthy et al., 2012; Murray 2014; Yanget al., 2015; IPCC 2007). However, a lot of endeavors further need to be conducted to truly set up the new methodological system of “periconnection”. The statistical analyses operated in this study need more explorations for reliably exposing the mechanisms of their performance, and more exploratory analyses (Brown & Robinson 2011) need to be carried out to explore other kinds of eco-effects for enriching the theoretical framework of the proposed concept. Moreover, whether the SP in some particular surrounding geo-grids really alludes the EP anomalies is still an open question. After all, the impacts of snow-onset and -end SP on EP have been observed to be nonlinear (Piaoet al., 2016), which further adds the difficulty in just using a pursuit of statistical relations between the current SP and EP indices to predict the EP in future climate scenarios. Therefore, well-designed manipulation experiments are also needed to improve our understanding of the ecological interactions between EP and SP, and, in a broader sense, of global carbon balance and ecosystem feedbacks to the ongoing climate changes.