78
in the dark. GPP can be many times greater than
either 14C production or NPP, and NPP can in fact
be a negative value if phytoplankton spend too
large a proportion of their day below the compensation depth and total daily respiration by phytoplankton exceeds GPP (Cole et al. 1991, 1992; Howarth et al. 1996). In ocean systems, the depth
above which rates of GPP and phytoplankton respiration are equal when integrated over the water
column is called the "critical depth." When the
depth of the mixed layer is greater than this critical
depth, so that phytoplankton are mixed in such a
way that their total respiration rate integrated over
the water column exceeds their total rate of GPP,
NPP is negative. When the depth of mixing is shallow compared with the critical depth, GPP becomes
proportionally greater than phytoplankton respiration, and NPP increases. This concept has shown
great utility in explaining the seasonal limits to phytoplankton growth in the North Atlantic Ocean
(Sverdrup 1953).
Perhaps the greatest challenge associated with
the use of incubation techniques is the difference
between the natural ecosystem and the portion of
the system that is contained in the bottle. These socalled "bottle effects" have been the focus of large
amounts of study, most of which has been more
valuable for highlighting the problems than for providing an unambiguous set of solutions. A few of
the key issues are discussed below.
One area of concern that now seems to have
achieved consensus regards contamination with
trace metals. From the earliest incubation experiments, it was recognized that contamination of the
sample in the bottle would compromise the measurement. Rigorous cleaning regimes were defined
with most methods. However, in the 1980s, a revolution in the analytical ability to measure trace
metals in seawater completely changed the perspective of the word "clean" for the open ocean.
Trace metals that had been thought to exist at micromolar concentrations were found to exist at
nanomolar and even picomolar concentrations
(Bruland 1980; Boyle et al. 1981). Careful incubations that compared natural concentrations with
the previous levels of cleanliness showed large differences in production rates, with production usually higher at the natural levels of trace metals and
production partially suppressed due to toxic effects
from the higher levels of trace metals found usually
Robert W. Howarth and Anthony F. Michaels
traditional techniques (e.g., Fitzwater et al. 1982).
Conversely, small additions of trace metals above
background levels sometimes seem to enhance production after a few days, particularly in parts of the
ocean where macronutrients persisted through the
year (Martin et al. 1989). These experiments indicated that trace metal limitation for iron, zinc, and
perhaps other metals could be an important control
on these ecosystems.
In the past decade, a standard, trace-metal clean
protocol has evolved for oceanic ecosystems. This
protocol carefully redefines every step of the
14C protocol to remove even trace levels of contamination in the sample. All of the sampling gear and
the incubation bottles are rigorously cleaned with a
multistage acid-cleaning procedure. Samples are
collected with samplers that can pass through the
air-sea interface closed and open and close at depth
to gather the sample. General Oceanics Go-Flo bottles are most commonly used. These are deployed
on nonmetal cables (usually Kevlar) or on tracemetal clean rosettes. Samples are drawn into the
incubation bottles using the same precautions that
the trace-metal chemists use to prevent contamination. The 14C innoculum is prepared to either reduce the trace-element concentration or cleaned by
passing it through a chelation column. Using these
precautions, trace-metal concentrations in the incubation bottles are near background. Rates of production with "clean techniques" appear to be twoto fourfold higher than those from before these
techniques were adopted (Michaels and Knap
1996). In coastal waters and most freshwater systems, trace-element concentrations are naturally
higher and these extreme measures may not be
required.
Another general problem with both the light and
dark bottle method and the 14C technique is the inherent isolation of the phytoplankton from the natural system by placement in a bottle (Collos et al.
1993). Within the incubation bottle, turbulence is
reduced, which may reduce the rate of nutrient uptake due to increased boundary layers around cells,
particularly in high-turbulence ecosystems such as
rivers and estuaries (Nixon et al. 1979; Richey et
al. 1990; Howarth et al. 1992). The total nutrient
pools or pools of DIC and CO 2 in incubation bottles
may also become depleted to the extent the water
is isolated from pathways of remineralization and
mixing. Within 6 to 24 hours, there can be signifi-
in the dark. GPP can be many times greater than
either 14C production or NPP, and NPP can in fact
be a negative value if phytoplankton spend too
large a proportion of their day below the compensation depth and total daily respiration by phytoplankton exceeds GPP (Cole et al. 1991, 1992; Howarth et al. 1996). In ocean systems, the depth
above which rates of GPP and phytoplankton respiration are equal when integrated over the water
column is called the "critical depth." When the
depth of the mixed layer is greater than this critical
depth, so that phytoplankton are mixed in such a
way that their total respiration rate integrated over
the water column exceeds their total rate of GPP,
NPP is negative. When the depth of mixing is shallow compared with the critical depth, GPP becomes
proportionally greater than phytoplankton respiration, and NPP increases. This concept has shown
great utility in explaining the seasonal limits to phytoplankton growth in the North Atlantic Ocean
(Sverdrup 1953).
Perhaps the greatest challenge associated with
the use of incubation techniques is the difference
between the natural ecosystem and the portion of
the system that is contained in the bottle. These socalled "bottle effects" have been the focus of large
amounts of study, most of which has been more
valuable for highlighting the problems than for providing an unambiguous set of solutions. A few of
the key issues are discussed below.
One area of concern that now seems to have
achieved consensus regards contamination with
trace metals. From the earliest incubation experiments, it was recognized that contamination of the
sample in the bottle would compromise the measurement. Rigorous cleaning regimes were defined
with most methods. However, in the 1980s, a revolution in the analytical ability to measure trace
metals in seawater completely changed the perspective of the word "clean" for the open ocean.
Trace metals that had been thought to exist at micromolar concentrations were found to exist at
nanomolar and even picomolar concentrations
(Bruland 1980; Boyle et al. 1981). Careful incubations that compared natural concentrations with
the previous levels of cleanliness showed large differences in production rates, with production usually higher at the natural levels of trace metals and
production partially suppressed due to toxic effects
from the higher levels of trace metals found usually
Robert W. Howarth and Anthony F. Michaels
traditional techniques (e.g., Fitzwater et al. 1982).
Conversely, small additions of trace metals above
background levels sometimes seem to enhance production after a few days, particularly in parts of the
ocean where macronutrients persisted through the
year (Martin et al. 1989). These experiments indicated that trace metal limitation for iron, zinc, and
perhaps other metals could be an important control
on these ecosystems.
In the past decade, a standard, trace-metal clean
protocol has evolved for oceanic ecosystems. This
protocol carefully redefines every step of the
14C protocol to remove even trace levels of contamination in the sample. All of the sampling gear and
the incubation bottles are rigorously cleaned with a
multistage acid-cleaning procedure. Samples are
collected with samplers that can pass through the
air-sea interface closed and open and close at depth
to gather the sample. General Oceanics Go-Flo bottles are most commonly used. These are deployed
on nonmetal cables (usually Kevlar) or on tracemetal clean rosettes. Samples are drawn into the
incubation bottles using the same precautions that
the trace-metal chemists use to prevent contamination. The 14C innoculum is prepared to either reduce the trace-element concentration or cleaned by
passing it through a chelation column. Using these
precautions, trace-metal concentrations in the incubation bottles are near background. Rates of production with "clean techniques" appear to be twoto fourfold higher than those from before these
techniques were adopted (Michaels and Knap
1996). In coastal waters and most freshwater systems, trace-element concentrations are naturally
higher and these extreme measures may not be
required.
Another general problem with both the light and
dark bottle method and the 14C technique is the inherent isolation of the phytoplankton from the natural system by placement in a bottle (Collos et al.
1993). Within the incubation bottle, turbulence is
reduced, which may reduce the rate of nutrient uptake due to increased boundary layers around cells,
particularly in high-turbulence ecosystems such as
rivers and estuaries (Nixon et al. 1979; Richey et
al. 1990; Howarth et al. 1992). The total nutrient
pools or pools of DIC and CO 2 in incubation bottles
may also become depleted to the extent the water
is isolated from pathways of remineralization and
mixing. Within 6 to 24 hours, there can be signifi-