Thresholds of gross primary production for the metabolic balance of marine planktonic communities

The notion that less productive marine planktonic communities tend to be heterotrophic was tested by synthesizing reported estimates of the relationships between the net community production or community respiration and gross primary production (GPP), allowing calculation of the threshold GPP separating less productive, heterotrophic communities from more productive, autotrophic ones. A total of 35 estimates of the threshold GPP were assembled, derived from reports of comparative analyses of individual regions (Mediterranean Sea, Atlantic Ocean, Southern Ocean, Pacific Ocean, and Indian Ocean) and global comparative analyses for open-ocean and coastal environments, time-series analyses of changes in planktonic metabolism at individual locations, experimental manipulations in mesocosms, and a semi-empirical modeling exercise. Planktonic communities of the open ocean and continental shelf showed threshold GPP values ranging 30-fold, from 0.34 mmol O2 m23 d21 to 9.45 mmol O2 m23 d21, with those for estuarine and coastal locations reaching 50.60 mmol O2 m23 d21. Antarctic and ultra-oligotrophic ecosystems showed the lowest threshold GPP values (,2.2 mmol O2 m23 d21), with a general consistency across approaches for a given ecosystem. Plankton community respiration in the absence of or under low primary production is not negligible and is supported by semi-labile dissolved organic carbon. The analysis of GPP thresholds suggests that allochthonous organic inputs to the less productive regions of the ocean must be in the order of 5–6 mmol C m22 d21, consistent with recent estimates of allochthonous inputs of organic carbon to the ocean. The metabolic balance of marine communities, the balance between their rates of autotrophic production of organic matter (gross primary production, GPP) and their respiratory remineralization (community respiration, R), is a key property determining net community production (NCP 5 GPP 2 R) of the communities. NCP affects the function and role of marine communities in material fluxes (Odum 1956). Autotrophic communities (GPP . R, NCP . 0) act as sinks of CO2 and inorganic nutrients and sources of organic matter and O2. Conversely, heterotrophic communities (GPP , R, NCP , 0) act as sources of CO2 and inorganic nutrients and sinks of organic matter and O2. Early depictions of the metabolic balance of aquatic ecosystems assumed the ocean to be in close metabolic balance, being marginally autotrophic (Odum 1956) to support carbon burial and fisheries yield. This perception has dominated views of oceanographers for decades (Williams 1998), to become implicit in consensus depictions of the ocean carbon (Prentice et al. 2001). However, recent reports of net heterotrophic community metabolism (i.e., NCP , 0) of planktonic communities studied in various regions of the ocean, including the subtropical Atlantic (Duarte et al. 2001; González et al. 2001; Harrison et al. 2001), the subtropical N. Pacific (Williams et al. 2004), and the Southern Ocean (Agustı́ et al. 2004), have revealed a wider prevalence of heterotrophic communities in the ocean than previously considered, particularly in the least productive oceanic regions (Duarte and Prairie 2005). A comparative analysis of aquatic community metabolism conducted a decade ago (Duarte and Agustı́ 1998) concluded volumetric oceanic respiration rates to be scaled as the half-power of GPP (i.e., R ,GPP0.5), implying that highly productive communities tend to be autotrophic whereas the least productive ones tend to be heterotrophic. Duarte and Agustı́ (1998) inferred a threshold GPP of 1.09 mmol O2 m23 d21 separating less productive, heterotrophic oceanic planktonic communities from more productive, autotrophic planktonic communities. However, these inferences, apparently in conflict with parallel claims of a widespread close metabolic balance in marine communities and proportionate scaling of R to GPP (i.e., R ,GPP1; Williams 1998), were argued to be the result of artifacts resulting from the use of Model I regression analysis and volumetric, instead of areally integrated, rates (Williams and Bower 1999). The ensuing debate has not yet delivered a clear consensus on this issue (cf., del Giorgio and Duarte 2002; Karl et al. 2003; Duarte and Prairie 2005). However, it has stimulated much-needed research to expand the meager (Williams and del Giorgio 2005) empirical basis upon which these debates were originally based (Duarte et al. 1999; Williams and Bower 1999; del Giorgio and Duarte 2002). The number of analyses of the relationship between R and GPP in marine ecosystems, including regional surveys (Duarte et al. 2001; Serret et al. 2002; Agustı́ et al. 2004), time-series analyses (Duarte et al. 2004, Navarro et al. 2004, Williams and Duarte 2004), experimental tests (Duarte et al. 2004; Agustı́ and Duarte 2005; Olsen et al. 2006), and modeling analyses (López-Urrutia et al. 2006), and a global data base of published marine planktonic GPP and R estimates (Robinson and Williams 2005), has increased greatly in response to this debate. This enhanced empirical basis provides a new test of the existence, * Corresponding author: [email protected] Limnol. Oceanogr., 54(3), 2009, 1015–1022 E 2009, by the American Society of Limnology and Oceanography, Inc.