Measurements of CH4 concentrations in the bottom water during two discrete sampling periods in subsequent years above different cold seeps at the Pacific margin off Costa Rica indicate large-scale variations of CH4 release. CH4 is emitted from mud extrusions and a slide scar at 1000–2300 m water depth. Maximum CH4 concentrations were found to be lower above all investigated sites in autumn 2003 than in autumn 2002 although seep sites are up to 300 km apart. Tidal and current changes were observed but found to apply only to individual seep sites. Increased seismic activity connected to the moment magnitude (MW) 6.4 earthquake offshore Costa Rica in June 2002 could have had an impact on all seep sites and thereby caused an increase in CH4 emission. This is supported by the largest variations of CH4 concentration found above mud extrusions located above faults likely more strongly affected by tectonic movements. Even though our data indicate a relation between seismicity and CH4 seepage, the relation is not proven, and future work is needed to comprehensively test this hypothesis.
 Natural CH4 seeps of varying intensity are found along most convergent continental margins. Several of these sites have been investigated in detail, e.g., the Cascadian Margin [Suess et al., 1999], the area offshore Peru [Dia et al., 1993], and the Barbados accretionary prism [Henry et al., 1996]. Convergent margins are crucial regions in terms of element recycling [Moore and Vrolijk, 1992]. Fluids and volatiles are mobilized due to the compaction of sediments that accrete in or subduct beneath margin wedges. CH4 is one of the most frequent observed compounds migrating from various sediment depths toward the sediment-water-interface. Most of it becomes anaerobically oxidized in the near-seafloor sediments, leading to the precipitation of authigenic carbonates [Han et al., 2004; Kulm et al., 1986; Teichert et al., 2005] and providing energy for vent-specific biota [Sibuet and Olu, 1998]. Only a fraction of CH4 escapes into the water column, its extent depending on the fluid pathway, the efficiency of oxidation processes, and the rate of upward flow [Linke et al., 2005; Luff and Wallmann, 2003]. A summary of cold seep activities with emphasis on CH4 cycling is provided by Boetius and Suess .
 Numerous cold seeps were examined in detail along the Costa Rican subduction zone over the past years [Bohrmann et al., 2002; Han et al., 2004; Hensen et al., 2004; Linke et al., 2005; Mau et al., 2006]. Along the Costa Rican subduction zone, the Cocos Plate is subducted beneath the Caribbean Plate at a rate of nearly 88 mm/yr since late Oligocene/early Miocene offshore Costa Rica [Kimura et al., 1997]. The subduction mechanism is proposed to be of erosive nature, that is not only the incoming crust and overlying sediments are subducted [Saffer et al., 2000] but also material from the upper plate is removed [Ranero and von Huene, 2000]. The tectonic erosion leads to extension of the continental margin, as evidenced by observations of listric faults which lead to a landward tilt of blocks of the basement [Meschede et al., 1999]. Some of the listric faults may penetrate the whole overriding plate down to the decollement [von Huene et al., 2004]. All seep sites reported in this paper are located at this part of the continental margin in water depths ranging from 1000–2300 m.
 The seep sites are associated with mud extrusions and scarps (Figure 1). Mud extrusions are driven by buoyancy forces that arise from bulk density differences between undercompacted fluid-rich clayey sediments and denser overlying sediments [Brown, 1990]. All mud extrusions reported in this paper are associated with deep reaching fault zones. Off Middle America, submarine landslides can be triggered by seamount subduction, which leads to a temporary uplift of the continental wedge during passage of the seamount and causes landslides on the over-steepened seaward side of the uplift [Ranero and von Huene, 2000; von Huene et al., 2000]. Several circular uplifts associated with slope failure structures have been identified along the continental margin of Costa Rica [von Huene et al., 2000], and are here referred to as scarps. All of these different structures show the typical signs of active CH4 seepage including chemosynthetic seep organisms and authigenic carbonates.
 The amount of CH4 discharging from cold seeps is difficult to estimate because of the high variability in space and time [Linke et al., 1994, 2005; Tryon et al., 1999] in particular on longer timescales. We had the opportunity to measure CH4 concentrations at several seep sites offshore Costa Rica in 2002 and approximately 12 months later (Figure 1). The temporal variability observed could have had several causes: seasonality, ocean current changes, tidal effects or differing seismic activity. In order to identify the most possible cause, we combined CH4 concentration data with oceanographic and seismic data.
 CH4 concentrations were measured in water samples collected with standard CTD/rosette equipment at four different cold seep sites offshore Costa Rica (Figures 1 and 2). Sampling took place in August/September 2002 aboard RV METEOR (M54-2/3) and in September 2003 using RV SONNE (SO173-3/4). For CH4-analyses aboard a modification of the vacuum degassing method described by Lammers and Suess  was used [Rehder et al., 1999]. Replicate analysis of samples of a single hydrocast yield a precision of ±10% for samples with CH4 concentration <2 nmol/L and ±5% for CH4 concentration >2 nmol/L.
 Tidal data were obtained from http://www.costarica.com/Home/Weather/Tides documenting the sea level changes at the port of Puntarenas, Costa Rica. These data correspond with short-term records of pressure changes at Mound Culebra and Mound 12 (Figure 3). These variations in pressure were recorded by a MAVS 3-axis acoustic current meter (NOBSKA) at Mound Culebra and a CTD (SBE16plus) at Mound 12 connected to a lander device [Pfannkuche and Linke, 2003] in 2002 (M54-2/3). Records of hydrostatic pressure exist from 16–23 September 2002 at Mound Culebra and from 23–24 September 2002 at Mound 12.
 Current measurements were obtained by upward looking ADCPs (Acoustic Doppler Current Profiler; RD Instruments) attached to different lander devices (Figure 4). ADCPs were located at Mound Culebra at 10°18.00′N/86°18.32′W in 1543 m water depth in 2002 and at 10°17.16′N/86°17.89′W in 1610 m water depth in 2003. Data from 20 m above the seafloor in 2002 and 2003 were used for comparison. At Mound 12 ADCPs were deployed at 8°55.87′N/84°18.85′W in 1020 m water depth in 2002 and at 8°55.61′N/84°18.40′W in 1023 m water depth in 2003. Data from 5 m above the seafloor in 2002 (1200 kHz ADCP did not cover a depth range as great as 20 m) and from 20 m above the seafloor 2003 were compared.
 Earthquakes located in the area of Figure 1 were selected from data provided by the Red Sismológica Nacional, Costa Rica (RSN: ICE-UCR). Earthquakes located by less than 5 stations and with a traveltime error of >0.6 s (root mean squared residual) were excluded. The energy released by earthquakes was calculated using the Gutenberg-Richter formula [Gutenberg and Richter, 1954]: log E = 11.8 + 1.5 ML, where E is energy in TJ (Terra Joule) and ML is local magnitude. Only earthquakes with a ML ≥ 3 were included in the calculation because of the limited magnitude detection level of the seismological network, especially offshore where no stations are situated.
 Variations of CH4 concentration were observed at cold seep sites investigated along the continental slope of Costa Rica between the sampling campaigns in 2002 and 2003, i.e., 11 to 12 months later. At all sites, the maximum CH4 concentration was lower in September 2003 compared to the maximum CH4 concentrations found in August/September 2002 (Figure 2 and Table 1). The values above the NW flank of Mound Culebra dropped from 42 nmol/L in August 2002 to 2 nmol/L in September 2003 (i.e., close to the regional background CH4 concentration, Figure 2b). At the summit of Mound Culebra, the CH4 concentration decreased from 24 nmol/L to 4 nmol/L between the investigations in 2002 and 2003 (Figure 2a). The values above Mound 10 and Mound 12 indicate the same trend (Figures 2c, 2f, and 2g). CH4 concentrations in Jaco Scarp are more variable not only in time but also in space. The CH4 profiles of 2003 show partly decreasing and partly increasing CH4 concentrations compared to the profiles of 2002 depending on water depth. Still, maximum CH4 concentrations at the upper edge of the talus apron decreased from 178 nmol/L in August 2002 to 64 nmol/L in September 2003 in 1850 m water depth (Figure 2d). Measurements at the SE rim of Jaco Scarp indicate a drop by 30% from 2002 to 2003 in 1700 m water depth (Figure 2e). The investigations indicate a general change to lower CH4 concentrations that is more obvious above the mud extrusions than in the scarp within a time frame of approximately one year.
Table 1. CH4 Concentrations at the Different Seep Sites
20 Aug 2002
21 Sep 2003
19 Aug 2002
22 Sep 2003
15 Sep 2002
22 Sep 2003
Edge of slide
Edge of slide
26 Aug 2002
19 Sep 2003
28 Aug 2002
23 Sep 2003
3 Sep 2002
24 Sep 2003
19 Sep 2002
24 Sep 2003
 The large areal extent of the observed CH4 concentration decline suggests a regional mechanism responsible for that change within this 11–12 month period between sampling. One possible change could have been seismic activity. The seepage sites are situated above a convergent plate boundary, an area of pronounced and continuous seismic activity. Protti et al.  reported that the central Costa Rican margin, historically the most seismically active region on the margin, can generate earthquakes up to MS 7.0 over a short recurrence interval. The segment offshore Nicoya Peninsula has also the potential for large MS 7.7 earthquakes and a recurrence interval of 50 years [DeShon et al., 2003].
 Temporal changes in concentration of chemical components of groundwater on land have been reported before many large earthquakes [King, 1986] going back to the 1960s when scientific studies in this field started. Such geochemical anomalies are thought to be driven by stress-induced crustal deformation, which affect the permeability of rocks [Favara et al., 2001; Italiano et al., 2001] and, in turn, fluid flow [Montgomery and Manga, 2003; Trique et al., 1999]. A natural laboratory in the French Alps was used by Trique et al.  to investigate this relationship. They showed that high water levels in lakes induced accelerated loading of the ground (deformation of the ground) suggesting enhanced fluid transport and causing the associated recorded radon emanation. Furthermore, Montgomery and Manga  summarized reported observations of hydrological response to earthquakes and could even show that the maximum distance of changes in water levels in wells is related to earthquake magnitude. Recently, the connection between CH4 seepage and seismic activity was hypothesized in marine settings by Obzhirov et al.  and Shakirov et al.  on the basis of CH4 data and earthquake events in the Sea of Okhotsk. For parts of our research area at the continental margin off Nicoya Peninsula, Costa Rica, Brown et al.  suggest that observed transient fluid flow events are linked to small scale seismic signals resembling tremors. Overall, a relationship between seismic activity and fluid flow is supported by observations on-land as well as, to a smaller extent, in marine tectonic settings.
 To identify a possible relationship of CH4 seepage and seismic activity off Costa Rica, we used preliminary earthquake data of 2002 and 2003 provided by the Red Sismológica Nacional, Costa Rica. The data cover only earthquakes with local magnitude ≥3. Thus we could only investigate a possible effect of seismicity of higher intensity on CH4-seepage. Tremor-related variations could not be studied. The earthquake record shows that CH4 concentrations in 2002 have been measured 2–3 month after increased seismic activity (Figure 5). This increase may be the consequence of a MW 6.4 earthquake that occurred near the plate interface on 16 June 2002 [DeShon et al., 2003]. In contrast, CH4 concentrations obtained in 2003 have been sampled after a period of less frequent, low magnitude earthquakes (Figure 5). Hence different seismic activities before sampling could have affected CH4-emission over a regional scale.
 Most of the epicenters of earthquakes recorded by RSN are located SE of Nicoya Peninsula, 200–300 km away from Mound Culebra and Mound 10 (Figure 6). This raises the question, whether CH4-emissions at these mounds could have been affected by these earthquakes. Montgomery and Manga , who compiled reported observations of water well responses to earthquakes, found that an earthquake of magnitude 3 can have an impact on well levels in a radius of ∼20 km, an earthquake of magnitude 4 on wells in a radius of roughly 50 km. By extrapolation of their model the MW 6.4 earthquake in June 2002 could have had a possible effect on hydrology in a radius of up to 500 km. The areas potentially influenced by the earthquakes offshore Costa Rica monitored by RSN are shown in Figure 6, illustrating that only the major event in June 2002 could have affected all the investigated seep sites. Apart from Mound 10, the other seep sites could have also been influenced by other earthquakes lower in magnitude, but we suggest that the increasing stress accumulation that resulted in the final slip along the plates recognized as the MW 6.4 earthquake could have caused a general increase of the CH4 output into the water column.
 The decline in CH4 concentration is more pronounced at the mud extrusions than at Jaco Scarp. The maximum concentrations observed above the mud extrusions in September 2003 never reach 50% of the maxima in the year before and are in some cases lower by an order of magnitude, whereas the maxima at the scarp reach at least 55% of the values of the previous year (Figure 2). In contrast to the scarp, mud extrusions are situated above deep-seated faults. Hensen et al.  showed that one end-member of the fluids expelled from mud extrusions originates from 10–15 km depth, i.e., from the subducted sediments, migrating most likely upward along faults. Ascending fluids push the zone of anaerobic oxidation of CH4 into shallow sediment depth or even through the sediment-water interface. Thus higher fluid discharge results in enhanced CH4 seepage [Luff and Wallmann, 2003]. Active faults are weak parts of the crust and it is not surprising that gases and fluids escape along this zones of least resistance [Favara et al., 2001; King, 1986]. Tectonic strain may be greatly amplified at active faults [King, 1986], and so the influence of seismic activity on fluid pathways connected to these geological structures is expected to be high. In contrast, the influence of tectonic activity is less pronounced at Jaco Scarp, where gas and fluid escape is mainly a result of exposed deeper sedimentary layers, hosting reduced geochemical compounds at elevated pore pressures. Thus CH4 concentration vary less at seep sites related to the scarp.
 Our interpretation suggests an influence of major earthquakes on fluid flow/CH4 discharge. However, it is based on data from only two field campaigns during similar time slots of subsequent years. The study complements the work by Brown et al.  relating small scale seismic events (tremors) to transient fluid flow events. They found fluid flow anomalies near the trench whereas we investigated fluid seepage in the mid-slope of the continental wedge where mud extrusions are near deeply penetrating faults. Earthquakes affect hydrology in various ways, e.g., expulsion of fluids from the seismogenic zone, increased permeability due to shaking of surface deposits or bedrock fractures, decreased permeability resulting from consolidation of surficial deposits [Montgomery and Manga, 2003]. Possibly different mechanisms apply near the trench, in contrast to further up-slope. Most of the dewatering is believed to take place near the trench, where faults and stratigraphic layers have approximately equivalent permeabilities. Landward, stratigraphic conduits decrease in permeability, but faults maintain high permeabilities [Moore and Vrolijk, 1992]. The sensitivity of different dewatering structures to seismicity is not well constrained.
 The variations in water column CH4 concentration could also be caused by seasonal changes, tidal influence or changes in ocean currents, but we found these possible factors less satisfactory. Concerning the seasonality, we determined the CH4 concentration during the same season in each year (in August/September 2002 and September 2003). Also, the CH4 emitting sites have been thoroughly investigated and are caused by subsedimentary fluid flow rather than degradation of young organic matter [Hensen et al., 2004; Soeding et al., 2003].
 Tidal control on fluid seepage as a result of changing hydrostatic pressure has been shown to be a factor for fluid seepage sites at Coal Oil Point, California [Boles et al., 2001] and at Hydrate Ridge, Oregon [Torres et al., 2002]. Yet, four of the seven sites were sampled in the same part of the tidal cycle either during high or low water level (above and below the mean tide height, respectively), i.e., high or low hydrostatic pressure (Figure 3). At Mound Culebra and the eastern rim of Jaco Scarp a partial influence cannot be excluded. However, a general trend toward measuring intervals near low tide/decreasing water levels in 2002 and high tide/raising water levels in 2003 is not shown by our data.
 Changes in the oceanographic current regime are another potential mechanism which would produce lower concentrations even above constantly emitting seep sites. Current speed influences dilution of CH4 and thus CH4 concentration in the water column. Changes in directions could cause the same sample location to be upstream of the source in the one year and downstream in the next, thus causing concentration changes. Unfortunately, currents were not recorded during sampling and current data are only available for short time intervals at Mound Culebra and Mound 12. We approximated the current conditions during sampling at the mounds using the available current data at the corresponding phase of the tidal cycle. Comparison of the estimated currents indicates potential changes in current regime, but the inferred changes differ from site to site. For example at Mound Culebra top, direction turned from NE/SE to NW/NE (Figure 4a) whereas at Mound 12 currents changed from NW/NE to north/NE (Figure 4b). Current velocities in 2002 were higher than in 2003 at Mound Culebra NW flank (Figure 4a), thus lower CH4 concentrations should have been measured in 2002 rather than higher CH4 concentrations, as has been observed. Therefore, although changes in the oceanographic current conditions occurred at individual sites, a general change in current speed or direction, affecting all seep sites uniformly and causing the observed decline in CH4 concentration, cannot be inferred from the data. It would be a very unlikely coincidence if the observed changes were toward lower concentrations at all seven sites that are 300 km apart.
5. Summary and Conclusion
 Maximum CH4 concentration decreased significantly from autumn 2002 to autumn 2003 in the water column above mud extrusions and in the area of a scarp offshore Costa Rica, all known to be active sites of fluid emissions [Bohrmann et al., 2002; Mau et al., 2006]. Hydrostatic pressure changes due to tides and oceanographic variations could explain the changing CH4 content at individual sites, but systematic changes in hydrostatic pressure and/or oceanographic variations affecting all sites have not been observed. The MW 6.4 earthquake at the seismogenic zone of Central America in 2002 could have had an impact on all investigated seep sites, which could have lead to increased discharge of CH4-rich fluids into the water column. Even though a continuous data set would be needed to verify a relationship between seismic activity and CH4 seepage, the data support such a hypothesis. Knowledge of the influence of large to small magnitude earthquakes on the output of CH4 at seep sites is thus likely to be required for the calculation of CH4 emissions from seep sites over longer timescales. The observed changes of fluid flow patterns on land before an earthquake and potentially similar effects at the seafloor is surely of interest in the ongoing discussion of precursors of large seismic events. Long-term investigations of fluid seepage/CH4 discharge at different areas (e.g., near the trench and further away) in correlation to environmental parameters and earthquake data are required to comprehensively address these questions.
 Many thanks to scientists, masters, and crews of research cruises SO 163 and M 54 for their support, information, and discussion. Thanks also to the Red Sismológica Nacional (ICE-UCR) for providing the earthquake data. The comments and suggestions of the anonymous reviewers considerably helped to improve the manuscript. This publication is contribution 64 of the SFB 574 “Volatiles and Fluids in Subduction Zones” at the University of Kiel. Susan Mau was partly funded through the German Ministry of Science and Education grant BMBF Az 03G0177A.