Agroforestry Systems

, Volume 76, Issue 1, pp 81–93

Soil characteristics below Erythrina poeppigiana in organic and conventional Costa Rican coffee plantations


    • Department of AgricultureUniversidad Autónoma Metropolitana-Xochimilco
  • Davey L. Jones
    • School of Agricultural and Forest SciencesUniversity of Wales
  • John Beer
    • Department of Agriculture and AgroforestryCATIE
  • Jean-Michel Harmand
    • CIRAD – Département Persyst UR Fonctionnement et Pilotage des Ecosystèmes de Plantations S/C UR SeqBio – IRD (Sup Agro)

DOI: 10.1007/s10457-008-9201-y

Cite this article as:
Payán, F., Jones, D.L., Beer, J. et al. Agroforest Syst (2009) 76: 81. doi:10.1007/s10457-008-9201-y


The impact of Erythrina poeppigiana on soil characteristics, at three different positions relative to the shade tree and from three different soil depths, was evaluated in pairs of comparable Costa Rican coffee farms (organic and conventional) in 2000 and 2004. In the conventional system at 0–5 cm, higher C and N concentrations were found close to the shade tree versus the positions 2 m from the trunk (5.04 vs. 4.18%). This positive effect could influence only 20% of the farm area when high population of E. poeppigiana were used. This finding highlighted the importance of E. poeppigiana in maintaining SOM levels. In contrast, the organic system showed similar C and N concentrations for all positions probably due to an even distribution of pruning residues and to the use of organic amendments. A trend to higher total C and N concentrations for organic farms in comparison to conventional farms was found. No significant temporal changes in soil C or N concentrations were found between 2000 and 2004.


Coffea arabicaSoil organic carbonNitrogenShade treesSoil organic matter fractions


The effect of single trees on surrounding soil pH, nutrient concentrations and bulk density, via litter distribution, may be modified by topographic and climatic variables like soil creep and predominant wind direction (Zinke 1962; Boettcher and Kalisz 1990). On abandoned pasture land in northeastern Costa Rica, Fisher (1995) found that the long term presence of 11 tree species (age 25 years) significantly increased the base captions, available P and organic C content of degraded soils, and also caused a reduction in soil bulk density. In Costa Rica, coffee shade trees are used principally to reduce the impact of adverse climatic conditions, especially high temperatures, and to positively influence the size and the quality of coffee beans (Muschler 2001). These shade trees also have an important role in the uptake and recycling of nutrients from the soil (Young 1999). Leguminous shade trees may also fix N, which may later be recycled to the coffee via natural litter fall, pruning residues and root/nodule turnover; e.g., Babbar and Zak (1994) reported higher N mineralization rates under leguminous coffee shade trees in comparison to an unshaded coffee plantation. Beer (1988) reported that almost 90% of the nutrients stored in the above ground biomass of the common leguminous shade tree Erythrina poeppigiana are returned annually to the soil surface, principally because of pollarding (1–3 times annually); these trees can contribute 5,000–12,000 kg of organic material (up to 300 kg N) ha−1 year −1. Nevertheless, the information on the impact of shade trees on soil characteristics in tropical regions, and particularly in low input agricultural systems, is very limited (Beer et al. 1998).

Depressed international coffee prices from 1999 to 2003 lead to a search for coffee niche markets, offering greater economic premiums. One option, adopted by some Costa Rican farmers, was organic coffee production in agroforestry systems (Lyngbaeck et al. 2001; Boyce et al. 1994). Many temperate region studies have shown that organic agricultural systems, in comparison to conventional agronomic practices, can have positive impacts on the soil including higher organic matter concentrations, microbial biomass, mineralizable N, promotion of a more friable soil structure and enhanced soil biological activity (Lotter 2003; Fließbach and Mäder 2000). For example, in organic farms in New Zealand and Australia, cation exchange capacity (CEC) and total soil N (0–15 cm depth) were higher in organically farmed soils, but pH, available soil P and S were higher in conventionally farmed soils possibly due to chemical inputs (Reganold et al. 1993; Reganold 1995; Wells et al. 2000). The real advantages of organic farming systems need to be evaluated in a greater number of farms and soil types, particularly in tropical regions, to determine whether these systems truly represent an ecologically and economically sustainable management strategy.

The central objectives of the research reported in this article were to analyze the impact of the shade tree E. poeppigiana on chemical and biological soil variables in comparable organic and conventional coffee farms; principally on soil C and N concentrations, pH, electrical conductivity, CO2 production and soil organic matter size fraction distribution.

Materials and methods

Farm selection criteria

Following a preliminary soil study in July 2000, five pairs of comparable farms were selected (five organic and five conventional; paired farms separated by no more than 500 m) in the municipalities of Aserrí (Aserrí 1 and Aserrí 2), Turrialba (CATIE), and Paraíso (Paraíso and Pejivalle), Costa Rica. An initial criterion was that the organic farms had to have been managed organically for at least 4 years (longest was ten and average was 7 years). A neighboring conventional farm, using standard herbicide and synthetic fertilizer doses, was then selected to complete each pair. Soil characteristics and management regimes were as similar as possible for each pair (Table 1). Soil bulk density was one of the indicators of comparability between paired farms (no significant differences between paired farms were found). E. poeppigiana was the principal shade tree on all farms selected. Shade tree spacing (approximately 4 m between trees), coffee type (mostly the “Caturra” cultivar), soil type, altitude, slope and farm size (less than 10 ha) were also taken into account when selecting sites (Table 1).
Table 1

Soil and management characteristics of five pairs of organic and conventional coffee plantations in Central Costa Rica


Soil subgroupa

Slope (%)

Coffee age (year); previous use

Years managed organically (in 2004)b

Liming (kg ha−1)

Fertilizers (1995–2000); total amount and NPK (kg ha−1 year−1)

Fertilizers (2000–2004); total amount and NPK (kg ha−1 year−1)

Herbicides (l ha−1 year−1); manual weed control

Pruning regime, pollarding height and residue distribution (i)c

Yield (kg fresh cherries ha−1 year−1)d

Aserrí 1 organic

Andic Haplustoll


30; forest


500 in 1997

2,500 of chicken manuree (50N, 50P, 25K; April 1996 only)

2,000 of compost (coffee pulp, cow manure and CaCO3: 5:5:2 in volume; October 2003)

Manual chopping (2–3 times a year1)

Partially 1 year−1; 2.5 m; residues evenly distributed


Aserrí 1 conventional

Andic Haplustoll


30; forest



675 of formula 18:5:5 (120N, 34P, 100K; Twice a year1)

675 of 18:5:5 (120N, 34P, 100K; once a year)

Glyphosate (1.0); paraquat (1.0); manual chopping (3 times a year)

Total 2 year−1; 1.5 m; no distribution of residues


Aserrí 2 organic

Andic Haplustoll


30; forest


750 in 1999

4,000 of chicken manure (80N, 80P, 40K; once a year; May–June; 1999–2003)

2,000 of compost (coffee pulp + chicken manure; 1:1 in volume; May 2003) 200 of KMAG (22% K2O = 44K) + 100 of CaCO3

Manual chopping (2–3 times a year1)

Partially 1 year−1; 2–3 m; residues are distributed


Aserrí 2 conventional

Andic Dystrustept


7; pastures/fallow land


500 in 1999

400 of 18:5:15 (72N, 20P, 60K; once a year)

Glyphosate (1.0); manual chopping1 3 times a year

Partially 2 year−1; 2 m; no distribution of residues


Aserrí 2 conventional (substitute)

Andic Dystrustept


10; pastures/fallow land



400 of 18:5:15 (72N, 20P, 60K; once a year)

Glyphosate (1.0); manual chopping (twice a year)

No pruning since 2000


CATIE organicg

Andic Dystrudept/Typic Hapludand


30; sugar cane


1,600; April 2001; June 2003

226.5 of KMAG (22% K2O = 50K; once a year; May)

226.5 of KMAG (22% K2O = 50K; 1 year−1 May)

Manual chopping (2–3 times a year)

Partial (2 year−1); 2–3 m; residues distributed


CATIE conventional

Typic Hapludand


30; sugar cane


1,500; April 1996

855 of 18:5:15 (154N, 42.7P, 128.2K; 3 times a year January, April, August)

350 of 18:5:15 (63N, 17.5P, 52.5K; May 2003)

Glyphosate (1.5) Paraquat (1.0); 2,4D (0.25); 4 times a year

Total (2 year−1); 1.5 m; no distribution residues


Pejivalle organic

Typic Haplohumult


30; sugar cane


1,000; November 2000

1,800 of compost (chopped grass; 9N, 9P, 9K; once a year; May)

1,800 of compost; (chopped grass; 9N, 9P, 9K; May 2003)

Manual chopping (2–3 times a year)

Partial (2 year−1); 2–3 m; residues distributed


Pejivalle conventional

Typic Haplohumult


15; forest



540 of 18:5:15 (97N, 27P, 81K; twice a year)

350 of 18:5:15 (63N, 17.5P, 52.5K; May 2003 only one)

Glyphosate (1.0); Oxiflurofen (0.25) (2–3 times a year)

Total (2 year−1) 1–5 m; no distribution of residues


Paraíso organic

Andic Haplohumult


>30; forest


1,500; August 1999

171 of earthworm manure; 143 of KMAG (22% K2O; 3.5N, 3.5P, 26K)

171 of earthworm manure; 143 of KMAG (22% K2O; 3.5N, 3.5P, 26K); 1,000 kg of compost (coffee pulp, chicken manure, CaCO3 5:5:2; once a year; September)

Manual chopping (2–3 times a year1)

Partial (2 year−1); 2.5 m; residues distributed


Paraíso conventional

Andic Haplohumult


30; pastures



640 of 18:5:15 (116N, 32P, 96K; once a year1)

No chemical fertilizers applied since 2001h 1,200 of chicken manure in part of the farm (January/04)

Paraquat (1.0); Oxiflurofen (0.25) (2 times a year1); manual chopping (2–3 times a year)

Total (2 year−1); 1–5 m; no distribution of residues


aSoil Survey Staff (2003); b All organic farms were conventional coffee farms before conversion; c With partial pruning, 2–3 branches remain on the tree; d Derived from local volume units (Fanegas; 1F ≃ 254 kg) as reported by farmers; e Content equivalent to: 2% N; 2% P2O5, 1% K2O (Fishersworring and Roβkamp 2001); f No trustable record was found; g The two soil pits were dug in opposite sides of the organic field and two different subgroups were classified; h The farmer did not applied chemical fertilizers due to economic problems; i 1 year−1 = once a year 2 year−1 = twice a year 3 year−1 = three times a year

All ten farmers were interviewed to determine plantation history including: age of the plantation, intensity of herbicide and fertilizer use (conventional farms), shade tree pruning regimes, production levels, use of lime to modify pH, pest and disease problems, and their general observations about the management of their coffee. Eight out of the ten farms were private properties and two were part of CATIE’s commercial farms. In Costa Rican coffee plantations it is common to find different coffee cultivars and plant ages within the same plantation as older coffee bushes are gradually replaced (Somarriba et al. 2001).

Selection of study sites within farms and of sampling positions within sites

Within both organic and conventional plantations, a central plot area (30 × 50 m) was selected avoiding edge effects. Similarity with the corresponding plot of the pair (similar age of shade trees, soil type and coffee-tree row spacing) were criteria. Within this central plot, three square 4 × 4 m sample sites were selected randomly. Each sample site contained one E. poepiggiana tree on each corner and four coffee rows (Fig. 1). Within each sample site, soil samples were initially collected (2000) from three different positions with respect to E. poeppigiana: (1) in the alley equidistant from two coffee rows and more than 2 m from a shade tree (“alley”); (2) below a coffee plant and more than 2 m from a shade tree (“bc > 2”); (3) below a coffee plant and less than 1 m from a randomly selected shade tree (“bc < 1”).
Fig. 1

Sampling design to compare soils in organic and conventional coffee farms in central Costa Rica

Soil sampling

In the initial study (2000), a helicoid-iron auger was used to obtain soil cores at 0–5, 5–10 and 10–20 cm depth after removing surface litter. The choice for examining these layers layer was related to the author’s interest in studying, along with other parallel researches in CATIE, the interactions between the shade tree E. poeppigiana and coffee bushes, which have relatively shallow root systems. This choice was also based on previous studies in the region (Kass et al. 1993; Lyngbaeck et al. 2001; Beer 1988) as well as on the detailed soil profile descriptions using the USDA methodology (Soil Survey Staff 2003). We classified one or two soil profiles in each farm, and they showed similar chemical conditions in deep layers. Additionally, some authors considered that in deep soil layers fresh-C inputs by plants are very low.

At each position and depth, 4–5 cores (20 cm apart) were mixed in a bulk sample. Twenty-seven bulk samples per farm were obtained (3 sample sites × 3 positions × 3 depths). The samples from the same depth and position from each of the three sites per farm were later mixed thoroughly to obtain nine compound samples per farm. Samples were kept in cold boxes with ice, conserved at 4°C until analysis.

A detailed soil survey and new soil sampling were carried out between April and June 2004, in all but one of the farms that were studied in 2000. In the Aserri 2 conventional farm, shade trees had been removed and this farm was substituted by a nearby conventional plantation with the same soil profile (less than 300 m from the original farm). On farms that had irregular topography (both organic and conventional), the plantation was divided into zones with different slopes, and separate soil pits were dug in each slope class to ensure that comparable areas were sampled. The soil profiles were described and samples were taken from the soil horizons to determine total C and N, texture, pH, Fe and Al concentrations in oxalate (to determine Andic characteristics), Ca, Mg, K and P concentrations. All data were used to classify these soils following USDA methodology (Soil Survey Staff 2003) and to ensure that the soil types in each pair of conventional and organic farms were comparable.

In 2004, only two positions, with respect to the shade tree, were sampled: (1) in the alley equidistant from two coffee rows and more that 2 m from a shade tree (“alley”); and (2) below a coffee plant and less than 1 m from a shade tree (“bc < 1”). The position “bc > 2” (below a coffee plant and more than 2 m from a shade tree) was not sampled in 2004 because the 2000 results indicated no differences to the “alley” position. The soil was sampled using mini soil pits (40 × 40 × 40 cm) to verify that horizon “A” was always sampled. The least disturbed wall of the mini soil pits was sampled at 0–5, 5–10 and 10–20 cm depths; samples distributed in three comparable areas, (according to the soil profile descriptions) within each of the paired farms were taken (2 positions × 3 sites × 3 depths). As in 2000, samples from the same depth and position were mixed to obtain six compound samples per farm.

Analytical methods

In the 2000 study, soil samples were air-dried and passed through a 2 mm stainless steel sieve. For total C and N analysis, approximately 3 g of sieved soil was pulverized; between 0.1 and 0.2 g were wrapped in tinfoil to be analyzed by gasification in a CHN2000 (Leco Corporation St. Joseph, MI, USA). Microbial respiration was measured using fresh soil (20 g) incubated at 22°C in an infrared gas analyzer (CIRAS-SC, PP Systems Ltd, Hitchin, UK). Electrical conductivity and pH (in water at a 1:1 v/v soil:water ratio) were measured with a Jenway 4010 conductivity meter and an Orion 410A pH meter. Organic matter fractions were obtained by the granulometric fractionation method (Kouakoua 1998). Subsamples of air dried soil (10 g) were dispersed in 50 ml sodium hexametaphosphate solution (5 g l−1) by shaking at 120 rpm for 1 h. The samples were sieved through three stacked 2000, 200 and 53 μm stainless steel sieves. The samples were washed using tap water until all clay, silt and fine organic materials were separated and the water passing through the sieves became clear. Two macroorganic matter fractions (53–200 and 200–2,000 μm) was collected, dried at 40°C, and weighed for intersystem comparisons. Size-fractions were analyzed at the CIRAD soil laboratory in Montpellier, France. In the 2004 study, soil total C and N were analyzed by gasification as described above for the 2000 study. Values for pH and electrical conductivity were measured in water at a 1:1 v/v soil:water ratio.

Weed biomass measurements were only made in the Pejivalle paired farms (the great similarity between soil profiles in these paired farms was taken into account for choosing this site). A square meter framework was located randomly on the ground in each of the three sampling sites and all weeds on the surface were collected and dried at 40°C for 3 days. An average of the three measurements for the conventional and organic farms was calculated.

Statistical analysis

Data were analyzed with ANOVA as a randomized complete block (the farm pairs) split-plot (systems as the main treatment and positions as sub-treatment) design with five replications. When significant statistical interactions between system or position were found, a t-test was performed. Data were analyzed using SAS, version 8 (SAS Institute 1999). For comparisons between farming systems, an average value of the positions within each farm was considered. Differences in total soil C and N between 2000 and 2004 were analyzed using ANOVA as a randomized complete block (the farm pairs) split–split-plot (systems as the main treatment and positions and years as sub-treatments) design.

Results and discussion

Total C and N soil concentrations

In 2000, significant differences between the three sampling positions (“bc > 2”, “alley” and “bc < 1”), for the 0–5 and 5–10 cm layer (P < 0.05), but not for the deepest layer (10–20 cm), were found for total soil C and N concentrations (Figs. 2a, 3a). At a depth of 0–5 cm in the conventional farms, the position nearest the shade tree (“bc < 1”) had higher soil C and N concentrations than the positions located 2 m from the trees (5.04 vs. 4.18% and 4.16% for C and 0.43 vs. 0.37 and 0.36% for N). Soil carbon and N concentrations for the three positions (0–5 cm layer) within the organic farms were similar. In the position “bc < 1”, no significant differences in soil C and N content were found between organic and conventional farms. However, soil C and N concentrations were higher in the organic treatments for the two positions located 2 m from the tree. In the 5–10 cm layer, at the position “bc > 2”, higher total C and N concentrations were observed in organic farms when compared with conventionally managed farms (Figs. 2b, 3b). In conventional farms, C and N concentrations were higher for “bc < 1” compared to “bc > 2”.
Fig. 2

Total soil C concentration for three depths in organic (□) and conventional (■) coffee farms in central Costa Rica. Treatments: alley, equidistant from two coffee rows and more than 2 m from a shade tree; bc > 2, below a coffee plant and more than 2 m from a shade tree; bc < 1, below a coffee plant and less than 1 m from a shade tree. Columns with the same letter are not significantly different (Lsmean-test P < 0.05)
Fig. 3

Total soil N concentration for three depths in organic (□) and conventional (■) coffee farms in central Costa Rica. Treatments: alley, equidistant from two coffee rows and more than 2 m from a shade tree; bc > 2, below a coffee plant and more than 2 m from a shade tree; bc < 1, below a coffee plant and less than 1 m from a shade tree. Columns with the same letter are not significantly different (Lsmean-test P < 0.05)

The results of the 2004 study were similar to the 2000 study. At the 0–5 cm depth within conventional farms, the position nearest to the tree trunk “bc < 1” had higher soil C and N concentrations than in the “alley” (5.67 vs. 4.17%), but within organic farms these two positions had similar C and N concentrations (Figs. 2d, 3d). In organic farms, the “alley” positions had higher values than their counterparts in conventional farms. No significant differences were found between organic and conventional farms for the position “bc < 1”. At soil depths of 5–10, 10–20 cm, there were no significant differences in soil C and N concentrations between positions in either organic or conventional systems (Figs. 2e, f; 3e, f). It is necessary to emphasize that these shallow soil layers were chosen taking into account previous soil description studies in the area Additionally, some authors have stated that in deep soil layer, fresh-C plant inputs are very low (Fontaine et al. 2007). This is particularly true for shaded coffee plantations where tilling activities are realized using mainly manual instruments.

No clear tendencies for changes in total soil C and N concentrations over time between 2000 and 2004 were found; i.e., split–split plot analysis did not detect a significant time effect for any position. “Priming effects” are defined as strong short-term changes which cause an extra decomposition and release of organic C or N following the addition of easily-decomposable organic substances to the soil (Kuzyakova et al. 2000; Ohm et al. 2007). The analysis of “Priming effect” processes can be considered a useful approach for explaining the absence of differences in soil C concentrations between 2000 and 2004. Even though 5,000–12,000 kg of fresh organic material from pruning residues are added to the soil every year, soil C concentrations remained practically at the same levels. The study of Payan et al. (2007) about the dynamics of macrooganic matter size-density fractions of C in Costa Rican soil added with fresh pruning residues of E. poppigiana has showed that the amount of C contained in the LF, MF and HF fell, on average, by 50, 57 and 60%, respectively, despite the of input of 12 Mg ha−1 year−1 over a 330-day study period. Unfortunately, the biochemical analyses using 14C labeled substrates as proposed by Ohm et al. (2007) to determine the impact of priming effect the obtained processes on soil organic matter fractions were beyond the reach of our study. However, the obtained results can encourage other researchers to assess the role of mechanisms involved in this phenomenon. Particularly, the acceleration of microbial activity and turnover and CO2 flush after the addition of easily-decomposable pruning residues of E. poeppigiana reviewed by Kuzyakova et al. (2000) and Ohm et al. (2007) can be studied.

In the 2000 study, four out of the five organic farms had greater soil C and N concentrations (0–5 cm) than the conventionally managed farms for “bc > 2” and three out of five for “alley” (Fig. 4a, b) while in the position closest to the tree (“bc < 1”), no tendency was observed (Fig. 4c). Similar results were observed in the 2004 study with greater soil C in organic farms in alleys (>2 m from E. poeppigiana) but no difference at bc < 1 (Fig. 4e, f). In conventional farms only, values close to the tree were always higher than values distant from the shade tree. This could reflect lower soil organic matter inputs distant from the tree in the conventional compared to the organic system due to different management practices in conventional farms such as more frequent and more intensive shade tree pruning, producing lower quantities of pruning residues per year (Beer 1988; Chesney 2000) that tend to be concentrated around the shade tree trunks (Table 1). The homogeneous distribution of chopped residues from the trees is more common in organic farms (Table 1) because organic coffee farmers depend widely on the nutrients provided by pruning residues and tried to spread these residues although facing higher labour costs. Furthermore, a lower tree pollarding height in conventional farms (Table 1) may reduce the area under the crown receiving natural litter fall. In contrast in organic coffee farms shade tree tended to be higher because farmers seek for a denser shade and therefore the area receiving natural litter fall is larger thus contributing to a more homogeneous litter distribution in the farm. Additionally, in organic farms, biomass inputs such as organic amendments, green manure as well as weed and pruning residues were evenly distributed, which should reduce any spatial differences in soil C and N.
Fig. 4

Carbon concentrations in the topsoil (0–5 cm depth) for five pairs of coffee farms, [organic (□) and conventional (■)] in Costa Rica. Treatments: alley, equidistant from two coffee rows and more than 2 m from a shade tree; bc > 2, below a coffee plant and more than 2 m from a shade tree; bc < 1, below a coffee plant and less than 1 m from a shade tree

E. poeppigiana and other related specie pruning residue inputs can lead to increases in soil C concentrations (Beer 1988; Beer et al. 1998). Higher total soil N concentrations (Trofuldalf soil in Brazil) were found under the crown of E. glauca than outside the circle of influence (0.32 vs. 0.24%, respectively) (Santana and Cabala-Rosand 1982). After 5 years, plots on a Typic Humitropept soil in Turrialba, Costa Rica, that received 9.2 Mg DM ha−1 year−1 of E. poeppigiana pruning residues, had higher soil C (3.73 vs. 3.11% C in controls) (Ramírez and Bornemisza 1990). The application of 3 Mg DM ha−1 year−1 of Leucaena leucocephala mulch (a legume with a similar residue quality to E. poeppigiana), in coffee plantations on a clay soil in Kenya, led to a 15% increase in soil C concentrations after 3 years (Kimemia et al. 2001). These studies help explain the higher C concentration under shade tree crowns (“bc < 1” compared to “alley”) in conventional farms in the present study. The differences in soil C and N concentrations between conventional and organic farms also could occur because of herbicide use that decreases weed biomass in the conventional farms leaving the topsoil exposed to POM runoff and with less C inputs from the decomposition of weed residues (root and shoot turnover). The loss of topsoil by run off (surface erosion) in Turrialba (35% slope) in a 5 month period after applying 16 Mg ha−1 year−1 of E. poeppigiana mulch, was five fold lower than in fields without mulching (Garzón 1991).

In the case of the Pejivalle farms, where soil characteristics were most similar, weed biomass was much higher in the organic farm (4.44 vs. 1.27 Mg ha−1). In shaded coffee plantations where no chemical herbicides were used (Masatepe, Nicaragua, 500 m a.s.l., average rainfall of 2,012 mm year−1) 3.35 Mg ha−1 of weed biomass (DM) were added to the system in a 7 month period after seven hand cuttings. In contrast, systems using herbicides (two applications of Paraquat + Simazine + 2,4-D; 1.6 l ha−1) and two hand cuttings a year for 3 years, only received 0.79 Mg ha−1 weed biomass residues (Aguilar and Staver 1997).

Comparison between management systems

ANOVA showed no significant differences between organic and conventionally managed systems in both the 2000 and 2004 studies. In 2000, at 0–5 cm depth, total C and N concentrations (averaged over all positions and farms) were 5.12 versus 4.46% C and 0.44 versus 0.39% N for organic and conventional systems, respectively. In 2004, the average values were 5.5 versus 4.6% C and 0.47 versus 0.41% N for organic and conventional farms, respectively. In 2000 and 2004, when averages of all positions for each farm were compared, four out of five organic farms had higher C concentrations that their conventional counterpart (Fig. 4d, g).

When data for systems (averaged over all positions) were compared between 2000 and 2004, no clear trends were found for changes in total C and N concentrations. In two conventional farms, soil C concentrations (averaged over all positions) increased from 2000 to 2004 (1.19% in CATIE CON and 2.99% in Paraíso CON). Management changes may explain these increases: in CATIE CON shade levels increased (taller shade trees) and in Paraíso CON, banana plants (625 plants ha−1) and chicken manure applications were included. Soil variables for conventional farms may show greater response to management changes while organic farms are more stable.

The C-to-N ratio was similar for both systems (11.9 and 11.7 for organic and conventional systems, respectively) suggesting that this soil characteristic is not easily modified by management (Russell 1988). Plots in Turrialba that received E. poeppigiana mulch over a 5 year period had similar C–N ratios (11.7 or average) (Ramírez and Bornemisza 1990). Wells et al. (2000) did not find significant differences in the soil C-to-N ratio between organic and conventional farms after a 4 year experiment in a loam soil in Australia; likewise, Lockeretz et al. (1981) did not find changes after a 25 years on 30 paired farms growing cereals in the midwestern USA. Stockdale et al. (2001), who also reported nine studies which found higher soil C concentrations under organic systems in comparison to conventionally managed systems, suggested that such increases are a consequence of the incorporation of green manures, composts, and crop residues into the soil. In the same way, Wells et al. (2000) concluded that the only factor that could explain higher soil C concentrations in organic vegetable farms was the addition of organic amendments (40–60 Mg ha−1 year−1). In New Zealand, a comparison of 16 paired farms with a range of cropping systems that included fruit, citrus, vegetables, dairy and pastures, found consistently higher C concentrations for organic farms (Reganold et al. 1993).

In 2004, after examining the soil profiles, comparable areas for inter-pair comparisons, with less influence of soil creep and erosion, were identified. Although a general trend to higher C concentrations in organic farms was suggested above, the differences between paired farms may be due to differences in original soil conditions and the previous land use. Although most of the study farms were planted with coffee around 30 years ago, some locations were planted with sugar cane before coffee was introduced (e.g., both systems in CATIE) while in other cases, secondary forest was the previous land use (e.g., both systems in Aserrí 1). However, in Pejivalle the organic farm had been used previously for sugar cane but the conventional counterpart was covered by a secondary forest. García et al. (1990) found a lower soil organic matter and cation exchange capacity, leading to a loss of base cations at 0–30 cm depth, in coffee plantation soils in comparison with secondary forest on Andisols in Veracruz, Mexico. On the other hand, regular burning of sugar cane (pre-harvest) can negatively affect soil organic matter (Schroth et al. 2001).

Low coffee prices during the study period (2000–2004) resulted in a decreased use of chemical and labor inputs for weed control as well as for fertilization in conventional farms. This management implied more C inputs to the soil. In Paraíso CON, the introduction in 2000 of banana (Musa sp.) plants (625 plants ha−1) as shade plants has been an alternative for diversification. Banana Plants have a high growth rate and high biomass inputs to the soil. These factors may have diminished the differences between organic and conventional farms, but this could not be supported quantitatively. Additionally, in Paraíso CON chicken manure applications were found. These inferences were in line with the findings of a detailed study (six sampling sites per farm) of C and N changes (2000–2004) in the same paired farms of our study at Paraíso (Zuloaga 2004). The study concluded that soil C in the conventional farm was more sensitive to management changes than in its organic counterpart.

pH and electrical conductivity

In the current study, mean pH values in the 0–5 cm layer were 5.31 versus 4.90 and 5.74 versus 4.88 for organic and conventional farms in 2000 and 2004, respectively. However, no significant differences were detected for pH (0–5 cm) between positions or for inter-system comparisons in the two study years. In 2004, organic systems in CATIE and Paraíso had higher average soil pH than conventional farms but recent additions of lime or amendments seemed to be the main reason for the observed difference rather than the use or absence of chemical fertilizer. No changes in pH (0–5 cm depth) between 2000 and 2004 were observed either for organic or conventional farms, except for CATIE where liming led to the change. Ramírez and Bornemisza (1990) also reported very small changes on soil pH in alley cropping systems in Turrialba after 5 years of applications of 9.2 DM Mg ha−1 year−1 of pruning residues (pH = 4.5 with E. poeppigiana residue additions and 4.2 in controls with no additions).

Trends to higher pH in organic systems in annual crops in New Zealand were found by Reganold et al. (1993) and Lotter (2003). Those studies suggested that not using chemical fertilizers was one of the main reasons for higher pH in organic farms. In contrast, Lockeretz et al. (1981) did not find significant differences in pH in a comparison of 30 paired organic and conventional wheat, corn and soybean farms. Boettcher and Kalisz (1990) found higher pH (Typic Dystropepts) under the tree species Liriodendron tulipifera when a dense herbaceous cover (26 species) was associated with the trees (pH = 5.6) in comparison to areas under the trees where herbaceous cover was absent (pH = 4.7). Overall average conductivity for organic versus conventional systems was very similar: 3.09 versus 2.88 dS m−1 at 0–5 cm; 2.12 versus 2.26 dS m−1 at 5–10 cm; and 1.67 versus 1.77 dS m−1 at 10–20 cm, respectively (no significant differences (P < 0.05) because of high rainfall in the study areas).

Soil respiration rates

Higher soil respiration rates under organic systems in comparison to conventional were only found for 0–5 cm (Table 2, 1.68 vs. 1.21 mg CO2–C kg−1 h−1 dry soil; respectively). Reganold et al. (1993) and Lotter (2003) reported higher soil respiration rates in organic farms, associated with higher microbial activity and soil C concentrations under organic systems. The highly diverse microbial communities present in organic systems also have been associated with a more efficient metabolic quotient qCO2 (Fließbach and Mäder 2000). No differences were found between the three study positions at any depth for either system.
Table 2

CO2 production (mg CO2-C kg−1 h−1) from three soil depths in organic and conventional coffee farms in central Costa Rica

Positions depths (cm)







bc > 2

bc < 1

bc > 2

bc < 1


1.76 (0.23)a

1.66 (0.22)

1.61 (0.11)


1.16 (0.10)

1.14 (0.11)

1.33 (0.14)



1.06 (0.29)

0.99 (0.13)

0.75 (0.02)


0.74 (0.01)

0.70 (0.02)

0.88 (0.10)



1.05 (0.11)

1.03 (0.11)

0.93 (0.05)


1.04 (0.09)

0.93 (0.07)

1.02 (0.10)


Treatments: alley, equidistant from two coffee rows and more than 2 m from a shade tree; bc > 2, below a coffee plant and more than 2 m from a shade tree; bc < 1, below a coffee plant and less than 1 m from a shade tree

aStandard error (n = 5)

bAverage values with the same letter within a row are not significantly different (P < 0.05, Duncan)

Organic matter size fractions

The average amount of the organic matter fraction (200–2,000 μm) found in organic farm soils in 2000 was lower than in conventional farms at all depths (four out of five pairs) but was only significant for 10–20 cm (Table 3). High variability between locations was found particularly at 5–10 cm. This fraction could be influenced by recent management or recent movements of organic inputs in the shallowest soil layers; e.g., a recent pruning or run off of macroorganic matter. Although our results were not conclusive, they are in line with other studies which found that the amount of this fraction was a sensitive measure of differences in SOM related to system management (Barrios et al. 1996). The size fraction (53–200 μm) did not showed any significant differences between systems.
Table 3

Amount of macroorganic matter (>200 μm) at three soil depths in five pairs of organic (Org) and conventional (Con) coffee farms in central Costa Rica

Farms depth (cm)

Aserri1 Orga

Aserri1 Con

Aserri2 Org

Aserri2 Con



Pejivalle Org

Pejivalle Con

Paraiso Org

Paraiso Con

Average Org

Average Con








































ag 100 g−1 dry soil

bAverage values with the same letter within a row are not significantly different (P < 0.05, Duncan)


In the two study years (2000 and 2004), surface soil C and N concentrations (0–5 cm) were higher close to E. poeppigiana in conventional farming systems but no evidence of this effect was found in organic farming systems. These higher values can be attributed to the concentration of tree pruning residues that farmers leave near to the tree trunks in conventional farms. This positive effect of the shade trees, in raising soil C and N concentrations, could influence only 20% of the total plantation area in conventional farms when high populations of E. poeppigiana are used (625 trees ha−1; circle of influence, with radius of 1 m assumed, thus affecting 1,964 m2 in each hectare). In organic systems, a better spatial distribution of pruning residues and of other organic amendments was observed, diminishing the differences between positions respect proximity of shade trees.

Higher total soil C and N soil concentrations in organic systems were detected in 2000 and 2004 probably due to greater organic matter inputs from shade trees, weed biomass and organic amendments. The differences between positions and between systems were detected only at the 0–5 cm depth. This suggests that the management period studied was too short to affect deeper layers. No significant changes between 2000 and 2004 were found in C or N concentrations in either system at any depth; the organic farms seem to have reached a stable level of soil C and N concentrations. At 0–5 cm depth, higher respiration rates were detected in organic compared to conventional farms.

The natural variability in soil conditions of the studied coffee plantations affected comparisons between organic and conventional systems and impeded definitive conclusions. In this study, the effect of E. poeppigiana proximity could be observed even in fields with a 45% slope. However, in the inter-system comparison slope differences between the farms could have affected the comparison between farming systems (CON vs. ORG). Nevertheless, the frequency analysis and the results from a pair of farms that had almost identical soil properties (Pejivalle 2000, and 2004) supported the hypothesis of higher soil C and N concentrations in organic in comparison to conventional farms.

The use of the mini soil-pit method (2004) instead of augers (2000) for soil sampling helped in the detection of differences between treatments in sloping field conditions. Ideally a larger number of paired farms should be included in future studies of this kind to reduce the confounding effects due to inter and intra-site variability; however, comparisons of commercial organic and conventional systems are limited by the availability of comparable site conditions and management. When the main aim of the study is compare short terms effects of fresh C-inputs by plant pruning residues the soil sampling should be concentrated on the 0–5 cm layer, where management and shade tree proximity have the greatest effect in C concentrations. However, the important role of biological transporters such as native earthworms (e.g., Pontoxcolex corethrurus) in C dynamic at deeper layers has to be considered. This species can bury fresh-C in zones were microbial conditions are depressed and change the dynamics of soil organic carbon conserved there during long time periods. Recent studies have shown that input of fresh plant C into deeper layers can induce priming effects in SOC decomposition in despite of its recalcitrant chemical composition (Fontaine et al. 2007). Therefore priming effects could be studied in both shallow and deeper soil layers. Finally, a better distribution of pruning residues in conventional farms can extend beneficial effects of higher organic matter concentrations to larger areas.


This work was funded by CONACYT (National Science and Technology Council) of the Mexican Government and by UAM (Autonomic Metropolitan University). We would also like to thank Dr. Andrew Owen for his experimental support, Dr Hector Zelaya-Turcios (in memoriam) for his inspiring advice, Mr. Carlos Vazquez for his support during field stage and an independent reviewer who suggested some valuable corrections, clarifications and additions to the draft article.

Copyright information

© Springer Science+Business Media B.V. 2009