The experimental site is located to the south of Mexico City in Xochimilco (19° 16' 27.05'' N, 99° 05' 33'' W) at an altitude of 2240 masl. The climate is temperate with precipitation 600-1000 mm/year mostly from June to October. Mean annual temperature is 16 ^oC. The soils of the chinampas are of anthropic origin.

Recently, the remaining chinampas are fertilized with low-grade sewage and many of the channels have become stagnant and contaminated with garbage and domestic waste runoff. Increasingly, insecticides and chemical fertilizers are being used to cultivate new and "improved" plant varieties (Chapin 1988).

Three plots (6.5 x 28 m) covered mostly with grasses were cultivated with maize, amaranth or left fallow to monitor GHG. A systematic sampling was performed similarly in each plot. Maize and amaranth were planted on beds 40 cm wide with a 60 cm spacing between the rows on July 8^th 2012 and then harvested in January 2013. The crops were unfertilized and no herbicides or pesticides were applied. Weeds were removed when required and during the dry season, from September 2012 to January 2013 (harvest), once a week, 1.2 L of water from the channel was used to irrigate each plant. The plots with grass were left undisturbed and served as control.

CO₂, CH₄ and N₂O fluxes were monitored simultaneously from February 1^st 2012 to January 28^th 2013. Three chambers (25 cm length x 20 cm, internal diameter) were placed in the three plots of each treatment. They were designed as reported by Parkin et al. (2003) with a coated top and a sampling port fitted with a butyl rubber stopper. The chambers were inserted 5 cm into the soil. Gas sampling was done between 10:00 and 12:00 h. The covers were placed on the chambers and sealed airtight with Teflon tape. A 15 cm³ air sample was collected from the PVC chamber at 0, 20, 40 and 60 min after it was closed. The gas in the headspace was mixed by lushing 5 times with the air inside the chamber followed by gas collection for analysis. The 15 cm³ air sample was injected into 15 cm³ evacuated vials closed with a butyl rubber stopper and sealed with an aluminium cap pending analysis.

The headspace of the vials was analyzed for CO₂, CH₄ and N₂O on two Agilent Technologies 4890D gas chromatographs (GC) according to Serrano-Silva et al. (2011).

Each plot used to measure GHG fluxes was sampled by drilling 20 times the 0-20 cm layer. The soil samples from each plot were pooled (n = 9), sieved separately and characterized. The features measured to the soils were: pH, electrolytic conductivity (EC), water holding capacity (WHC), total N, organic C and soil texture, as described by Serrano-Silva et al. (2011).

Additionally, at the onset (February 2012) and end (January 2013) of the GHG monitoring, soil samples were taken from the 0-20, 20-40 and 40-60 cm layers in each plot to determine the total carbon (C_tot) and bulk density. Calculation of the net GWP was based on Robertson et al. (2000) and Thelen et al. (2010), taking into account soil C sequestration (Δ soil C GWP), emissions of GHG from the soil (soil N₂O flux + soil CH₄ flux), emissions of GHG from the fuel used for farming operations (which in this case were not used) (operation GHG flux) and the production of fertilizer and seeds (input GHG flux, were not used). The net GWP was calculated as:

Net GWP = Δ soil C GWP + soil N₂O flux + soil CH₄ flux + operation GHG flux + input GHG flux.

The overall GWP of the gasses emitted was calculated considering the GWP of 298 and 25 CO₂-equivalents for N₂O and CH₄, respectively (IPCC 2007).

Emissions of CO₂, CH₄ and N₂O were regressed on elapsed time, i.e. 0, 20, 40 and 60 min, using a linear model forced to pass through the origin, but allowing different slopes (production rates). The sample at time 0 accounted for the atmospheric CO₂, CH₄ and N₂O, and was subtracted from the measured values.

The C content in the 0-20, 20-40, 40-60 and 0-60 cm layers were subjected to a two-way analysis of variance using Proc GLM (SAS 1989) to test for a significant effect from layer, treatment and their interaction. Significant differences between treatments for CO₂, CH₄ and N₂O emission rates were determined using Proc Mixed considering repeated measurements (SAS 1989).

The total CO₂, N₂O and CH₄ emissions over the one-year period were calculated by linear interpolation of data points between each successive sampling event (Ussiri et al. 2009) and numerical integration of underlying area using the trapezoid rule (Whittaker and Robinson 1967).

The pH of the sandy clay loam soil was alkaline and EC ranged from 2.79 to 6.64 dS/m (Table I). The WHC of the soil ranged from 1888 to 2190 g/kg and total N from 5.92 to 6.17 g/kg, while the C_tot was considered high and ranged from 45.8 to 48.6 g/kg soil. None of the soil characteristics was significantly different between treatments.

TABLE I

The CO₂ emission did not show a clear pattern, but was higher by the end of 2012, and in the beginning of 2013 it ranged from 0.0012 to 6.0306 kg/ha/d (Fig. 1a). The emission of N₂O was considered low and ranged from -0.0065 to 0.0118 kg/ha/d (Fig. 1b). Sometimes negative values were obtained, i.e. reduction of N₂O was larger than its production. The emission of N₂O did not show large changes over time. The emission of CH₄ was low without a clear pattern (Fig. 1c). The flux of CH₄ ranged from -0.0249 to 0.0259 kg/ha/d and was mostly positive, so production prevailed over oxidation. The CO₂, N₂O and CH₄ emission rate was not affected significantly by treatment (Table II).

Fig. 1

TABLE II

The GWP of N₂O and that of CH₄ were similar in the different treatments and varied between 258 and 395, and between 26 and 44 kg CO₂-equivalents/ha/y, respectively (Table III). Consequently the GWP of the GHG was similar in the different treatments.

TABLE III

The organic C content of the soils ranged from 21.7 t/ha in the 0-20 cm layer of soil cultivated with amaranth to 28.4 t/ha in the 20-40 cm layer of uncultivated soil (Table IV). Soil layer, treatment and their interaction had no significant effect on the soil C content.

TABLE IV

Adverse effects of salinity and alkalinity on plants have been reported (Carrion et al. 2012). The high EC and pH found in the chinampas soil might inhibit growth of certain crops. The chinampas soil has a high organic matter content compared to arable soils of the regions, e.g. 7.2 g C/kg found in soil of Otumba (State of Mexico). Soils with high organic matter content do generally have a good fertility, and crop yields are high (Ball et al. 2007).

The constant application of sediment buries the organic material in the deeper soil layers. Consequently, the soil profile was organic rich, but with no clear gradient as normally found in arable soils (Table IV). The values found for C_tot in the 0-60 cm layer ranged from 73.9 in the maize cultivated soil to 81.7 t C/ha in the uncultivated soil, similar values have been reported in agricultural soils in the region. In the 0-60 cm layer of a conventional tilled soil with wheat and maize crop rotation and removal of residue in the valley of Mexico, the carbon content was 69.7 t C/ha (Dendooven et al. 2012).

Emissions of CO₂ were generally low in the first half of the year, but tended to increase towards the end of the year (Fig. 1a). During the dry season, i.e. mostly from November to May, channel water is used to irrigate the crops. The channel water is organic rich (Chavarría et al. 2010) and the mineralization of the applied organic material will increase CO₂ emissions.

Mineralization of the organic matter will provide nutrients for the crops, but this will also favour N₂O emissions, especially when an excess of mineralized N is present (Towprayoon et al. 2005). Additionally, frequent application of channel water will increase emissions of N₂O as the moisture content increases and denitrification is stimulated (Stewart et al. 2012). Cultivation of crops is also known to increase the emission of N₂O, as root exudates mineralization might stimulate denitrification (Kettunen et al. 2007).

In this study, N₂O emissions were generally low and occasionally even negative (Fig. 1b). Stewart et al. (2012) suggested that N₂O uptake can occur at relatively low soil moisture and temperature, and limited soil N. These conditions might be present in the chinampa soil, especially during dry spells in the rainy season.

In the chinampa soil, the CH₄ flux was mostly positive so the production of CH₄ was often larger than its oxidation. The high organic matter content (which stimulates microbial activity and oxygen consumption) and the regular irrigation with channel water, facilitate the creation of anaerobic microsites, and in consequence, methanogenesis.

CO₂, N₂O and CH₄ fluxes were not affected by crop. Management practices, such as irrigation, tillage and cropping system, as well as characteristics of the soils were similar in the study, so their effect on GHG emissions would be the same in the three treatments. The only different factor between treatments was the cultivated crop. It has to be considered, however, that crops-vegetables-lowers are regularly rotated in chinampa so it is very unlikely that crop will have an effect on GHG emissions.

N₂O contributed 91 % to the GWP of the GHG and CH₄ 9 %. N₂O is often the most important GHG from agricultural systems (Wan et al. 2012). It is only in rice-cultivation that CH₄ emissions are often more important than N₂O emissions (Horwath 2011). Cultivation of maize or amaranth had no significant effect on the GWP of the GHG. From this study, it can be assumed that the crop will have little effect on the GWP of the GHG emissions.

The GWP of the GHG was approximately 400 kg CO₂-equivalents/ha/y in a conventional agricultural system (tillage, maize monoculture, residue removal) in the valley of Mexico City in the year 2008-2009 and 230 kg CO₂-equivalents/ha/y in 2009-2010 (Dendooven et al. 2012). The values reported in this study were similar to those found in the arable soil mentioned above.