Corresponding author: Cecilia T. Satta ( ctsatta@uniss.it ) Academic editor: Lucilla Capotondi
© 2019 Ingrid Kröncke, Hermann Neumann, Joachim W. Dippner, Sally Holbrook, Thomas Lamy, Robert Miller, Bachisio Mario Padedda, Silvia Pulina, Daniel C. Reed, Marko Reinikainen, Cecilia T. Satta, Nicola Sechi, Thomas Soltwedel, Sanna Suikkanen, Antonella Lugliè.
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation:
Kröncke I, Neumann H, Dippner JW, Holbrook S, Lamy T, Miller R, Padedda BM, Pulina S, Reed DC, Reinikainen M, Satta CT, Sechi N, Soltwedel T, Suikkanen S, Lugliè A (2019) Comparison of biological and ecological long-term trends related to northern hemisphere climate in different marine ecosystems. In: Mazzocchi MG, Capotondi L, Freppaz M, Lugliè A, Campanaro A (Eds) Italian Long-Term Ecological Research for understanding ecosystem diversity and functioning. Case studies from aquatic, terrestrial and transitional domains. Nature Conservation 34: 311-341. https://doi.org/10.3897/natureconservation.34.30209
|
Data from five sites of the International Long Term Ecological Research (ILTER) network in the North-Eastern Pacific, Western Arctic Ocean, Northern Baltic Sea, South-Eastern North Sea and in the Western Mediterranean Sea were analyzed by dynamic factor analysis (DFA) to trace common multi-year trends in abundance and composition of phytoplankton, benthic fauna and temperate reef fish. Multiannual trends were related to climate and environmental variables to study interactions. Two common trends in biological responses were detected, with temperature and climate indices as explanatory variables in four of the five LTER sites considered. Only one trend was observed at the fifth site, the Northern Baltic Sea, where no explanatory variables were identified. Our findings revealed quasi-synchronous biological shifts in the different marine ecosystems coincident with the 2000 climatic regime shift and provided evidence on a possible further biological shift around 2010. The observed biological modifications were coupled with abrupt or continuous increase in sea water and air temperature confirming the key-role of temperature in structuring marine communities.
ILTER network, marine communities, multiannual trends, biological shifts, climate changes, climate indices, northern hemisphere
Long-term ecological research aiming at understanding sources of natural variability in species composition, dominant species and functional diversity of marine communities is essential for disentangling the effects of natural environmental drivers and climate variation from the direct effects of local human activities (
An analysis of changes across different biological components and geographic regions in relation to climate indices has the potential to provide a novel insight into the role of climate in influencing trends in marine biodiversity (
The trends in biodiversity described above were strongly correlated with increasing sea surface temperatures (SST) and regional climate variability (
Combining different climate modes with vertical propagating Rossby waves to and from the stratosphere (
CRS are often considered as changes in the trend of global mean air temperature.
Comparative ecosystem analysis has been effectively used to improve our understanding of the processes controlling the biodiversity, productivity, and resilience of marine ecosystems (
The International Long-term Ecological Research Network (ILTER), founded in 1993, comprises 44 active member LTER networks representing 700 LTER Sites and ~80 LTSER Platforms across all continents, active in the fields of ecosystem, critical zone and socio-ecological research (
The Santa Barbara Coastal LTER (https://deims.org/dbd399ed-9c26-4621-b479-7ab505c8aa35) in southern California (USA, Fig.
The arctic LTER Observatory HAUSGARTEN (https://deims.org/f6d9ed12-6bc1-47fb-8e81-ef24e9579596) in the Fram Strait between NE Greenland and the Svalbard archipelago (Fig.
The Baltic Sea LTER site Seili (https://deims.org/9d4222a2-c50f-4fac-8b1d-3b685072b34d) is located in the Archipelago Sea in Finland (60°15'33"N; 21°57'39"E) (Fig.
The LTER North Sea Benthos site (https://deims.org/50946250-c0fa-41b0-a917-17d2a3992eee) constitutes a network of seven sampling areas in the North Sea. Box A, the area used for this study, is situated in the German Bight about 25 nautical miles northwest of the Island of Helgoland, in close proximity to the 30 m depth contour near the former glacial valley of the River Elbe (Fig.
The Mediterranean Sea LTER site Gulf of Olbia (IT14-002-M) (https://deims.org/3178d0fb-0789-4992-9c51-1ddb50b7e871) has belonged to the LTER-Italy network since 2006. It is situated on the eastern coast of Sardinia (Italy, Fig.
The collection of LTER data started in 1992 for phytoplankton. Details in sampling and methods are given by
Table
Location, depth, sampling periods and designs, climate and biological data sets used for the DFA at the five LTER sites.
Santa Barbara Coastal LTER | LTER Observatory HAUSGARTEN | Baltic Sea LTER Seili | LTER North Sea Benthos Observatory | Western Mediterranean Sea LTER Gulf of Olbia | |
---|---|---|---|---|---|
Location | USA, Santa Barbara Channel | Germany, Arctic Ocean Fram Strait | Finland, Archipelago Sea | Germany, North Sea, German Bight | Italy, Tyrrhenian Sea (Sardinia), Gulf of Olbia |
Geographical coordinates | ~34°25'N, ~119°57'W | ~79°N, ~4°20'E | ~60°15'N, ~21°58'E | ~54°22'N, ~7°10'E | ~40°55'N, ~9°33'E |
Sampling depth | 7-11 m | 250-5500 m | Surface to twice the Secchi depth | 36-43 m | Surface water layer (-30 cm) |
Sampling period | July to August | Autonomously year round sampling; field study June | July to September | July to August | July to August |
Sampling design | 7 transects | Satellite remote sensing, annual sampling in summer months with water samplers, plankton nets, multicorers | 1-2 times per month sampling using a water sampler | Nine random replicates per year with a 2 m beam trawl | Fortnightly or monthly samplings at one station collecting water samples with a Niskin bottle |
Climate data | PDO | IL | NAO | NSE | WeMO |
Summer bottom temperature | Water temperature (0–100 m) | SST | SST | Air temperature | |
Period | 2002-2016 | 2000-2016 | 1991-2014 | 1998–2017 | 1996-2014 |
Biological data | Abundance of 12 fish species | Chlorophyll a, phytoplankton composition (6 groups), ash-free dry weight, biogenic sediment compounds, bacterial numbers and biomasses, meiofauna abundance | Phytoplankton biomass (9 groups) | Abundance of 16 epifauna species | Phytoplankton cell density as percentage composition on the base of contribution of 9 major groups |
DFA is a smoothing and dimension reduction technique to identify common trends in multivariate time-series and to determine the effect of explanatory variables. The “common trends” represent the underlying dynamic pattern over time in the considered system (
The time series were modeled as a function of a linear combination of common trends, a constant level parameter, two or more explanatory variables, and noise (
DFA was used to analyze time series of phytoplankton (Baltic Sea, Western Mediterranean, Arctic Ocean), benthic fauna (North Sea, Arctic Ocean), benthic bacteria (Arctic Ocean) and temperate reef fish (Santa Barbara Channel). Water or air temperature and various climate indices (PDO, IL, NAO, NSE, WeMO) were used as explanatory variables. Several DFA models were tested for each time-series, ranging from the simplest (one common trend plus noise) to the most complex (two common trends, up to three explanatory variables plus noise). Models were fitted with both a diagonal covariance matrix and a symmetric non-diagonal covariance matrix and compared using the Akaike’s information criterion (AIC) (
The transformed and standardized time-series of the different biological components of marine communities and environmental parameters are presented in Fig.
Transformed and standardized time-series at the various LTER sites. AFDW = organic matter in the sediments as ash-free dry weight, CPE = sediment-bound chloroplastic pigments, PROT = particulate proteins in the sediments.
Common trends (left) and corresponding factor loadings (right) for the five LTER sites obtained by means of DFA. Only factor loadings above the cut-off of 0.2 in absolute value are shown. Common trends and factor loadings are untitled. Dashed line in graphs indicates the confidence interval of the DFA model. Bac = Bacillariophyceae, Chl = Chlorophyceae/Chlorophyta, Coc = Coccolithophyceae, Cry = Cryptophyceae, Cya = Cyanophyceae, Din = Dinophyceae, nano = nanoplankton/nanoflagellates, Ppou = Phaecystis pouchetii, CHLa = chlorophyll a, AFDW = organic matter in the sediments as ash-free dry weight, CPE = sediment-bound chloroplastic pigments, PROT = particulate proteins in the sediments, Meio = meiofauna.
Model selection based on values of Akaike’s information criterion (AIC). The optimal Dynamic Factor Analysis model with one or two trends is given in bold. TEMP = summer bottom temperature, PDO = Pacific Decadal Oscillation index, SST = sea surface temperature, PRES = pressure, LAT = latitude, LONG = longitude, NAO = North Atlantic Oscillation index, NSE = North Sea Environmental index, AirTemp = air temperature, WeMO = Western Mediterranean Oscillation index.
Diagonal | Non diagonal | |||
---|---|---|---|---|
Number of trends | 1 | 2 | 1 | 2 |
SANTA BARBARA | ||||
M common trend + noise | 530 | 518 | 477 | 480 |
M common trend + noise + TEMP | 525 | 515 | 457 | 462 |
M common trend + noise + PDO | 525 | 521 | 436 | 444 |
M common trend + noise + TEMP + PDO | 517 | 506 | 416 | 415 |
HAUSGARTEN | ||||
M common trend + noise | 500 | 489 | x | x |
M common trend + noise + SST | 506 | 479 | x | x |
M common trend + noise + PRES | 510 | 493 | x | x |
M common trend + noise + LAT | 515 | 501 | x | x |
M common trend + noise + LONG | 518 | 475 | x | x |
M common trend + noise + SST + PRES | 503 | 478 | x | x |
M common trend + noise + SST + LAT | 520 | 485 | x | x |
M common trend + noise + SST + LONG | 515 | 467 | x | x |
M common trend + noise + PRES + LAT | 518 | 468 | x | x |
M common trend + noise + PRES + LONG | 473 | 480 | x | x |
M common trend + noise + LAT + LONG | 521 | 460 | x | x |
M common trend + noise + SST + PRES + LAT | 499 | 424 | x | x |
M common trend + noise + SST + PRES + LONG | 491 | 473 | x | x |
M common trend + noise + SST + LAT + LONG | 508 | 428 | x | x |
M common trend + noise + PRES + LAT + LONG | 478 | 439 | x | x |
M common trend + noise + PRES + LAT + LONG + SST | 506 | 479 | x | x |
SEILI | ||||
M common trend + noise | 549 | 554 | 569 | 582 |
M common trend + noise + SST | 555 | 557 | 574 | 583 |
M common trend + noise + NAO | 552 | 555 | 569 | 582 |
M common trend + noise + SST + NAO | 559 | 560 | 572 | 581 |
NORTH SEA | ||||
M common trend + noise | 874 | 855 | 790 | 784 |
M common trend + noise + SST | 871 | 843 | 768 | 761 |
M common trend + noise + NSE | 883 | 857 | 728 | 721 |
M common trend + noise + SST + NSE | 876 | 845 | 626 | 620 |
GULF OF OLBIA | ||||
M common trend + noise | 379 | 369 | 317 | 329 |
M common trend + noise + AirTemp | 381 | 374 | 314 | 325 |
M common trend + noise + WeMO | 387 | 376 | 313 | 313 |
M common trend + noise + AirTemp + WeMO | 394 | 379 | 309 | 308 |
LTER Santa Barbara Channel: Sixteen dynamic factor models were calculated to estimate the underlying common trend of the fish time-series differing in the covariance matrix employed (diagonal versus non-diagonal), the number of trends (one or two) and the included explanatory variables (mean summer bottom temperature, PDO index or both). The AIC values indicated that the best model fit was obtained for a non-diagonal matrix with two common trends explained by bottom temperature and the PDO index (AIC = 415) (Table
Both trends showed overall increases from 2001–2017 (Fig.
Measures of fit (diagonal elements of error covariance matrix) for the different time-series. Relatively low diagonal elements of the error covariance matrix (< 0.50) indicate good fit. AFDW = organic matter in the sediments as ash-free dry weight, CPE = sediment-bound chloroplastic pigments, PROT = particulate proteins in the sediments.
SANTA BARBARA | HAUSGARTEN | SEILI | NORTH SEA | GULF OF OLBIA | |||||
---|---|---|---|---|---|---|---|---|---|
B. frenatus | 0.68 | Chlorophyll a | 0.47 | Cyanophyceae | 0.51 | O. albida | 0.18 | Dinophyceae | 0.64 |
C. punctipinnis | 0.18 | nanoflagellates | 0.18 | Cryptophyceae | 0.76 | P. minutus | 0.77 | Bacillariophyceae | 0.34 |
E. jacksoni | 0.58 | Dinophyceae | 0.27 | Dinophyceae | 0.85 | A. rubens | 0.65 | Cryptophyceae | 0.49 |
H. caryi | 0.64 | P. pouchetii | 0.20 | Coccolithophyceae | 0.89 | C. allmanni | 0.40 | Chrysophyceae | 0.41 |
O. californica | 0.23 | Coccolithophyceae | 0.20 | Chrysophyceae | 0.83 | P. bispinosus | 0.49 | Euglenophyceae | 0.68 |
O. pictus | 0.42 | Bacillariophyceae | 0.14 | Bacillariophyceae | 0.80 | L. holsatus | 0.32 | Prasinophyceae | 0.76 |
P. clathratus | 0.35 | AFDW | 0.49 | Euglenophyceae | 0.95 | B. luteum | 0.69 | Chlorophyceae | 0.52 |
P. furcatus | 0.58 | CPE | 0.12 | Chlorophyta | 0.41 | C. crangon | 0.43 | others | 0.63 |
R. vacca | 0.42 | PROT | 0.13 | Mesodinium | 0.79 | O. ophiura | 0.51 | nanoplankton | 0.25 |
R. nicholsii | 0.43 | LIPIDS | 0.65 | A. irregularis | 0.82 | ||||
S. mystinus | 0.29 | Bact_Numb | 0.67 | A. laterna | 0.61 | ||||
S. pulcher | 0.57 | Bact_Vol | 0.58 | C. cassivelaunus | 0.78 | ||||
Bact_Biom | 0.38 | E. nitida | 0.65 | ||||||
Meio | 0.35 | P. bernhardus | 0.62 | ||||||
T. communis | 0.50 |
The estimated t-values for the explanatory variables bottom temperature and PDO are given in Table
Estimated t-values for the explanatory variables. Only values above 0.2 are shown. TEMP = summer bottom temperature, PDO = Pacific Decadal Oscillation index, SST = sea surface temperature, PRES = pressure, LAT = latitude, NSE = North Sea Environmental index, AirTemp = air temperature, WeMO = Western Mediterranean Oscillation index, PROT = particulate proteins in the sediments.
t- values | |||
---|---|---|---|
SANTA BARBARA | TEMP | PDO | |
O. californica | -2.48 | ||
O. pictus | -3.36 | ||
P. clathratus | 2.44 | ||
B. frenatus | -2.27 | ||
C. punctipinnis | 6.14 | ||
E. jacksoni | -2.90 | ||
HAUSGARTEN | PRES | LAT | SST |
Bacillariophyceae | 4.08 | -4.47 | -4.64 |
Coccolithophyceae | -3.51 | 4.79 | |
PROT | 4.24 | ||
P. pouchetii | 3.65 | ||
NORTH SEA | SST | NSE | |
C. allmanni | -2.37 | -3.22 | |
E. nitida | -2.48 | ||
P. bernhardus | -2.71 | ||
C. crangon | -2.03 | ||
GULF OF OLBIA | AirTemp | WeMO | |
Chrysophyceae | -3.45 | ||
Dinophyceae | 2.38 | ||
Chlorophyceae | -2.87 |
LTER Observatory HAUSGARTEN: Thirty-two dynamic factor models, using SST as well as the intensity and positions of the IL as explanatory variables, revealed two trends in the planktonic community and underlying sediments. The AIC values (Table
t-values indicated a significant influence of SST on the proportion of Bacillariophyceae (-4.64) and Phaeocystis pouchetii (3.65) (Table
Meiofauna density at HAUSGARTEN observatory was the only faunal parameter included in the DFA. Unfortunately, meiofauna data had to be restricted to results provided by
LTER Baltic Sea Seili: Sixteen dynamic factor models were calculated to estimate the underlying common trend of the phytoplankton time-series differing in the covariance matrix employed (diagonal vs. non-diagonal), the number of trends (one or two) and the included explanatory variables, i.e. SST and NAO (Table
LTER North Sea Benthos Observatory: Sixteen dynamic factor models were calculated to estimate the underlying common trend of the epibenthic time-series differing in the used covariance matrix (diagonal and non-diagonal), and the number of trends (one or two) and the included explanatory variable (SST or NSE or both). The AIC values indicated that the best model fit was obtained for a non-diagonal matrix with two common trends and SST and NSE as explanatory variables (AIC = 620) (Table
The first common trend showed a sharp decrease from 1998 to 2007, followed by an increase afterwards (Fig.
The estimated t-values for the explanatory variables SST and NSE are given in Table
Mediterranean Sea LTER Gulf of Olbia: The best fit of the sixteen model outputs for phytoplankton was one with a non-diagonal error covariance matrix, containing two common trends (Fig.
The first common trend showed higher and similar values at the beginning (1996–1997) and at the end of the considered time series (2013–2015), with a central part of lower values (1998–2012, minimum value in 2005) (Fig.
According to the largest factor loadings (Fig.
Considering the estimated t-values (Table
Data from the five LTER sites considered in this study revealed an overall increase in temperature (SST, bottom and air temperature; data not shown). SST increased 0.8–3 °C at the northeastern Pacific, North Sea and Baltic Sea sites between 1995 and 2016. Air temperature at the LTER site in the Western Mediterranean Sea increased by 0.8–1.1 °C, while at the Arctic LTER site HAUSGARTEN SST increased by 0.06 °C y-1 for the time-period 1997–2010 (
Figure
The comparison of climatic and long-term ecological data at the five LTER sites located in the north-eastern Pacific, the western Arctic Ocean, the northern Baltic Sea, the southern North Sea and the western Mediterranean Sea revealed a great similarity in common trends in marine systems in the northern hemisphere.
At four of the five LTER sites considered in this study, the DFA revealed the presence of two common trends with temperature and climate indices as explanatory variables. Despite the different biological components and marine ecosystems analyzed, multiannual common trends within a site were quasi-synchronous, but usually in opposite directions. Common trends first crossed near the end of 1990s/early 2000s and then again around 2010, which coincided with the 2000 CRS and probably a new, yet to be described CRS around 2010. We found one inversion of tendencies of biological components at the Arctic LTER HAUSGARTEN and two inversions at sites in the north-eastern Pacific Ocean (Santa Barbara), the Western Mediterranean Sea (Gulf of Olbia) and the North Sea. A similar signal was also observed at the Baltic Sea site (Seili), where the single common trend reached its maximum in 2005, for which no explanatory variable was identified, similar to other long-term plankton studies (e.g. in the North Sea,
The congruent trends in climate indices and CRS at the five LTER sites suggest that decadal fluctuations in atmospheric and ocean circulation are teleconnected between the Atlantic and Pacific Ocean regions as also found for previously reported CRS (
We found consistent biological modifications to abrupt or continuous increases in sea water and air temperature and associated climatic indices at all five of the LTER sites analyzed in this study.
Temperature modulates a multitude of processes at the cellular, organismic and ecosystem levels of organization, and it can act as a mediator between organisms and climate (
In our study, at the south-eastern North Sea LTER site, the first occurrence of the epifaunal angular crab Goneplax rhomboides coincided with the 2000 CSR. The angular crab extended its distribution range from the north-eastern Atlantic to the North Sea, which was facilitated by an increase in water temperature (
The biomass and diversity of reef fish at the Santa Barbara Coastal LTER site also shifted towards warm-temperate species with warm water affinities (
It is widely recognized that temperature has also a strong influence on the physiology of planktonic organisms because it controls basic metabolic processes (
Quantitative and qualitative changes in phytoplankton have unavoidable consequences at the other trophic levels. Several studies of benthic communities revealed that their diversity and function depended on the amount and quality of organic matter produced in the water column, even at extreme depths (
Time series data collected at the five widely distributed marine ILTER sites in the Northern Hemisphere revealed relatively synchronous changes in the abundance and species composition of marine biota across the study regions. The community changes coincided with the 2000 CRS and provide the first data on a possible additional CRS in 2010, highlighting the existence of teleconnections among the climate modes in the Northern Hemisphere. Although recent climate models predict further increases in global temperature, future CRS might cause unexpected changes in marine communities. The significant relationship described in this study between climate modes and marine communities mediated by sea water or air temperature in the Northern Hemisphere can aid in predicting future changes in these marine systems in response to climate variability and ocean warming. Such improvements in ecological forecasting should prove useful to environmental managers responsible for implementing measures aimed at mitigating adverse ecological effects of future changes in climate. Our findings highlight the importance of spatially distributed quantitative Long-Term Ecological Research in developing a predictive understanding of ecological responses to climate change in the world’s oceans and they underpin the continued need for long-term research within the global and the national LTER networks, as already suggested in similar studies (
The respective LTER sites are funded and/or coordinated by the National Science Foundation (USA), the Alfred-Wegener-Institute, Helmholtz-Center for Polar and Marine Research (Germany), the Finnish Environment Institute (Finland), SW Finland Centre for Economic Development, Transport and the Environment (Finland), University of Turku, Archipelago Research Institute (Finland), the Senckenberg Gesellschaft für Naturforschung (Germany), and the University of Sassari (Italy). The authors are indebted to Sultan Hameed for providing the IL time series. The authors declare that they have no conflicts of interest.