Agroforestry Systems 60: 291–304, 2004.

© 2004 Kluwer Academic Publishers. Printed in the Netherlands.

Microclimate in agroforestry systems in central Amazonia: does canopy
closure matter to soil organisms?

Christopher Martius1,*, Hubert Höfer2, Marcos V.B. Garcia3, Jörg Römbke4,
Bernhard Förster4 and Werner Hanagarth2,†
1Center for Development Research (ZEF Bonn), Walter-Flex-Str. 3, D-53113 Bonn, Germany; 2Staatliches
Museum für Naturkunde Karlsruhe, Karlsruhe, Germany; 3Embrapa-Amazônia Ocidental, Manaus/Amazonas,
Brazil; 4ECT Oekotoxikologie GmbH, Flörsheim, Germany; *Author for correspondence (phone: + 00 49 228
731838: Fax: + 00 49 228 731889; e-mail:

Received 10 september 2002; accepted in revised form 17 October 2003.

Key words: Biodiversity, Rain forest, Soil fauna, Vegetation structure


Microclimate was recorded and soil organisms were collected 1997-1999 in ecosystem stands of contrasting
structure in central Amazonia 共a primary forest, a 12-year secondary forest, two different agroforestry systems, a
rubber tree 共Hevea brasiliensis) plantation, and a peach palm 共Bactris gasipaes) monoculture with a densely
closed canopy兲. The aim was to look at the effects of canopy closure on microclimate and soil organisms. Monthly
maxima temperature, average air and soil temperatures, and saturation deficit were highest in September 1997,
and total annual rainfall in 1997 was 12-28% lower than in the other study years. The monthly average litter
temperatures were consistently 2-4 °C higher in the plantation sites than in the rainforest and the secondary for-
est, and temperatures on single days 共not the monthly averages兲 in the plantations were up to 10 °C higher than
in the primary forest. The highest average litter and soil temperatures and the highest temperature maxima were
recorded in the agroforestry plantations. Canopy closure strongly determined the litter temperatures in the sites.
Soil macrofauna biomass was also strongly correlated to canopy closure 共linear regression, P ⫽ 0.05兲. We con-
clude that a well developed canopy effectively protects the soil macrofauna from high temperature variation and
drought stress. Therefore, optimizing these agroforestry systems for canopy closure may contribute to a better
management of the beneficial soil decomposer community.

Introduction failed 共e.g., Anderson 1990; al 1995兲. A potential
remedy is seen in the establishment of perennial
Rainforests in the central part of Amazonia are multi-species cropping systems, supposing that diver-
threatened by large-scale clearings but also by the sity of agricultural systems matters for productivity as
traditional slash-and-burn practice of small farmers it enhances the biomass and nutrient stocks 共Jolliffe
共Lal 1995兲. The tropical rainforest ecosystem is 1997; Naeem et al. 1996; Nijs and Roy 2000兲. An-
known for its low resilience against disturbance; and other option is to design cropping systems that
generally, the attempts to transform it into conven- involve trees and mimic the stand structure of the
tional agroecosystems based on annual crops have successional rainforest 共Ewel 1999兲, thus reproducing
in particular the microclimate in these stands. Micro-

This paper is dedicated to the memory of our dear friend and col- climate is a factor that determines the environmental
league, Dr. Werner Hanagarth, who unexpectedly passed away on conditions not only directly for the crop but also for
September 2, 2003

Germany. how.. an air humidity of 90-100%兲. which is cycles and physical properties of the soil 共e. the rubber trees and the sec- ditions for tropical soil fauna 共Borchert 1994兲. applied knowledge is available on heat or drought Seringueira兲.. Table 1兲. Mo- supposed to be higher and more frequent.K. Material and Methods aged according to these principles. a species of low quality wood together with weather station records of local climate 共Schizolobium amazonicum Ducke – Paricá兲 and two which are used to assess the macroclimate and the in. With higher temperatures. Höfer et al. four different the effects of stand structure 共degree of canopy clo. 共For details cf.B. Amelung et al. tree species of commercial use had been originally sure兲 on microclimate and indicators of soil life in the planted. we report on microclimate recordings from POA and POC. Vismia is a typical pioneer species in rainforest gaps 共Whitmore 1998兲.兲.. In anthropogenic systems. We first link microclimate data phylla King – Mogno. 2001兲 can thus be zil 共3º8'S. The experiment consisted of Rican cloud forest兲. a higher soil moisture and especially a lower variation of these factors. 1997. a small tree whose population influ- enced especially the litter production in these sites. ecologically sustainable. located close to the city of Manaus. . Most ondary growth were cut and burned and the area has soil organisms in rainforests require high moisture since then been used as for experimental agroforestry 共e.g. year of 1997. native high quality wood species 共Swietenia macro- ter-annual variability. In both systems.兲 Muell. In one of the polyculture systems 共sites named Here. The area was divided a period of three years including the extreme climatic into 90 experimental plots of a size of 32 ⫻ 8m each. and Carapa guianensis Aubl.g. 1995 for a Costa 共Feldmann et al. Although it is well known that most soil organisms The investigations took place in several sites on a in temperate forests 共e.s. carried out in cooperation between the Em- even be considered to be aquatic 共Collado and brapa Amazônia Ocidental and the Institute of Schmelz 2000兲. a 10 m distance was kept between thus.292 the soil organisms. Ori. iroba spaced 7 m. or due to a fungus infection. the rubber tree (Hevea bra- data loggers in different stands in central Amazonia siliensis – Seringueira兲. University of Hamburg. and Mogno and And- mass and decomposition rates兲 to microclimate. Collembola. It was dominated by Vismia guianensis 共Aubl. then data from soil – Andiroba兲 had been planted in intermittent rows of biological assessments 共soil fauna and microbial bio. Study sites peratures and fewer extremes. at the same time. gan to develop.. 1100 species of imental agroforestry stands in central Amazonia over vascular plants in these plots. Bohlman et al. It is believed that in stands man. 59º52' W兲. pers. Corrêa. 2001兲. We investigated the microclimate in several exper. 19 ha area that was cleared from primary rain forest batei兲 show narrow and low temperature preferences in 1979/1980 in favor of a plantation of rubber trees and are highly sensitive to high temperatures.g. canopy closure to soil organisms. rubber and Paricá spaced 4 m. Bra- Lavelle et al. to canopy closure in the stands. In 1992. soil fauna and soil microbes will encounter lower average surface tem. stands. and. Applied Botany. which have a much simplified structure com. Preisinger et al. were thought to be economically viable in the region pared to the natural forest. doned monocultures with mixed cropping systems est ecosystems 共cf. the installation of different multi-crop systems that ever. an El Niño 共ENSO兲 year 共Wolter and Out of these. tions 共altitude 44 ⫺ 50 m a. This plantation was abandoned in 1981 stress in animals and plants in the humid tropics. nocultures served as control plots. Amazonas. and The study area belongs to the agroforestry research that the beneficial soil biota that regulate nutrient station Embrapa Amazônia Ocidental. 共1994兲 recorded a total of approx. comm. Soil moisture and soil with the objective to study the recultivation of aban- temperature seem to be quite constant in natural for. 1995兲. soil mois. and some can research. The area is flat without eleva- maintained. the neigboring rows. climatic extremes are and. two polyculture systems were chosen Timlin 1998兲.兲 Choisy.. Isopoda. little 共Hevea brasiliensis 共H. One particular interest was to look at for the present study. ture becomes more important. and secondary growth be- more specifically on the limiting microclimatic con.l. Arg. The secondary vegetation com- ing up rapidily between the planted rows was not re- moved.

differences between the loggers were found not to known in Brazil as ‘latossolo amarelo’. between which secondary growth including trees. between which only annual plants. and periods of microclimatic measurements. A largely undisturbed primary rain forest site Temperature and air humidity were recorded with 共FLO兲 and a secondary forest site 共SEC兲 near to this small data loggers 共“Stowaway XTI Internal/External experimental site were also studied. Rainfall data 共January A peach palm 共B. Amazonia. all loggers were The soil in the region is a Xanthic Ferralsol calibrated. gasipaes兲 monoculture 共site PUP兲 1987 to December 2000兲 were used for comparison. established in 1992. Swietenia mac- rophylla. fore being used in the field. Bertholletia excelsa. Site Code Description Measurement Periods FLO 共Primary forest兲 Primary rain forest Aug 1997 – Mar 1998 May 1998 – Nov 1998 Nov 1998 – Apr 1999 SEC 共Secondary forest兲 Secondary forest established in 1984 Aug 1997 – Mar 1998 May 1998 – Nov 1998 Nov 1998 – Apr 1999 POA. In both minute intervals and stored as two-hour average val- stands. Porasset/MA. ex Spreng.B. range ⫺ 39 to 122 °C. brasilien. “pupunha” in Brazil兲. We recorded secondary forest area was growing since 1984 on a soil temperature at a depth of 5 cm. 293 Table 1. with 7 m high palms and a rubber tree 共H. Be- established. . Bactris gasipaes H. Differences in exceed 0. The rain forest Temperature Logger”. USA兲. 共Preisinger et al. established in 1992 n/a PUP 共Palm stand兲 Monoculture of peach palm 共Bactris gasipaes. a popular fruit ing capacity was higher in the primary forest soil in Amazonia兲. In this system no high- growing secondary vegetation was admitted but Pu. Data were recorded at 10 nia and Bellucia spp. 12 data points were obtained per day.K. Bertholletia excelsa H. Manaus.B. 共Orleans tree. during the study in the litter layer. The saturation deficit was calculated from air sis兲 monoculture plantation 共SER兲 were chosen for temperature and relative air humidity according to the comparison with the polyculture sites. May 1998 – Nov 1998 broma grandiflorum. was allowed兲 May 1998 – Nov 1998 Nov 1998 – Apr 1999 POL 共Fruit tree culture兲 Polyculture culture system consisting of 4 native Amazonian fruit trees 共Theo. were admitted 共the logger was placed between two rows兲 SER 共Rubber stand兲 Monoculture of rubber tree 共Hevea brasiliensis兲.兲 Benth. 共FLO兲 than in the other plots 共Martius et al. established in 1992. i. 共Cupuaçu. Mico. established May 1998 – Nov 1998 in 1992 n/a⫽ not available The other polyculture system 共site POL兲 consisted grain size. pH. 共Peach palm兲. Bixa orellana兲 planted in rows. Theobroma grandiflorum study plots were not significant.e.K. The studied puter Corporation. and air humidty a few centimeters period it was dominated by Vismia guianensis. organic matter and total N between the of four fruit tree species. sample areas of 40 ⫻ 40 m each were ues. 共Kudzu兲 was daily recordings 共January 1996 to April 1998兲 planted and a low grass cover had developed between obtained from the standard climate station run by the the rows. litter temperature former abandoned rubber plantation. Brazil. “Magnus formula” 共Lechner 1992兲. Carapa guianensis兲 planted in rows. when according to the FAO/UNESCO 共1990兲 classification. 共1999兲 for a nearby forest reserve. Only the water hold- 共Willd. above the litter layer. 1994兲. Monthly average climate data were computed from eraria phaseoloides 共Roxb. Onset Com- al. “Stow- site conforms to the descriptions given by Ribeiro et away RH Relative Humidity Logger”. Study sites at the Embrapa Amazônia Ocidental.. 2004兲.5 °C 共see Table 1 for details兲. and checked after two years of use. POC 共Wood tree Polyculture system consisting of 4 commercial species: the rubber tree 共Hevea Aug 1997 – Mar 1998 culture at two sites兲 brasiliensis兲 and three wood species 共Schizolobium amazonicum. Embrapa Amazônia Ocidental. 共Brazil nut兲 and Bixa orellana L. a bush that is cultivated to Climate and microclimate measurements produce food colorants兲.兲 Schum. Bactris gasipaes.

8 times. calibrated with 共2001兲. but in 1997.002兲 between average annual rainfall in the first half of this period Soil respiration was measured using the release of 共1971-1985: 2408 ⫾ 188 mm兲 and the second half CO2 from soil samples in a 17-chamber Infrared-Gas. Macrofauna 共including termites and Isopoda兲 tures in the period 1996-1998 were all recorded in was sampled with soil cores of 21 cm diameter ex. September and October 1997. age annual rainfall was also lower 共2234 ⫾ 219 mm兲 in the period 1971-1985. aver- Kurzatkowski et al.D. Soil samples 0-10 increased during the later years. given our main in- terest in the canopy as an isolating layer between at. during four months rainfall and at the secondary and primary forest areas 共n⫽20/ remained below this threshold 共Figure 2兲. 共2004兲. 共Sony DSC-P5兲 mounted on a tripod. comm. from October 1997 to May 1998. Canopy closure was recorded with a digital camera Römbke et al. We then transformed the original color picture into a graph containing only black and white Litter decomposition was determined by enclosing pixels using standard commercial graphics software 7-10 g of air-dried Vismia leaves in litterbags made 共Paintshop Pro 7兲. between July 1997 and March 1998兲 mally in this region. followed by a Tukey test for pair. 1995. This mesh width al- black pixels in the picture. Nakamura Results pers. Linear regressions vals. Climate (weather station) mosphere and soil. The differences in canopy lows the whole decomposer community to participate cover between the sites were assessed with an in the decomposition of the leaf material 共Höfer et al. a water level to point exactly upwards. For details cf. The annual were used to determine the relationships between decomposition rate was computed from a negative canopy closure and climatic as well as pedobiologi. Höfer et al. exponential regression fitted to the data using the cal parameters. Mesofauna was determined from soil cores 共6 1997 was also a very hot year: the highest monthly cm diameter兲 separated into litter layer and topsoil maxima 共Tmax in Figure 3. see below兲. Despite long-lasting controversies over how to best measure canopy closure 共cf. In the period in question here 共1996-98 for which Soil fauna density and biomass in the study plots were microclimate data are available. not more than 1-2 months per soil samples were taken from randomized points at year have a rainfall below 100 mm 共Ribeiro and Adis the polyculture plantations POA and POC 共n⫽10/site兲 1984兲. top兲 and the highest av- 0-5 cm. Both rain- . 1999兲. transported to the lab. software Sigmaplot. ANOVA on ranks. The minimum air tem- tracted in a Berlese-type extractor. in the city of Manaus. and the weight loss assessed. 2001兲. Cook et al. Nearby. 10 shots were taken at random points Litter decomposition in each site.5% for. For details cf. The sta- Soil respiration tistically significant difference 共P⫽0. the chosen approach appropriately captures the cover rate of the stands. Earthworms were peratures 共Tmin兲 still were elevated in the subsequent extracted from soil by repeatedly pouring a 0. 共1986-2000: 2700 ⫾ 270 mm兲 suggests that rainfall Absorption-Spectrometer 共IRGA兲.兲. half after 1986. and extracted in a Kempson apparatus 共Adis erage monthly air 共Tmed兲 and soil 共Tsoil兲 tempera- 1987兲. and measured for 24 h. The bags were retrieved after repeated inter- wise comparison 共Sigmastat 2. period. 2001兲.03兲. data set 1971-2000 in which the annual rainfall ex- sieved through a 4 mm sieve to remove roots and ceeded the average ⫹ 1 S. The year site兲.294 Canopy closure mol solution to areas of 4 m2 共n⫽1 or 2 per date兲 and manually collecting the emerging earthworms 共cf. The six years of the cm were taken in the field. and higher 共2392 ⫾ 344 Soil fauna mm兲 during 1986-2000 共Hanagarth unpubl.Canopy closure is the percentage of of nylon mesh 共width 10 mm兲. At every sampling event 共every three annual rainfall was below the average ⫺ 1 S. we think that. Nor- months. the year assessed from soil and litter samples with different 1997 was one of the few extremely dry years in which methods. Average annual rainfall at Embrapa between 1971 and 2000 was 2554 ⫾ 273 mm 共Figure 1兲. and adjusted to wide angle.D. Ganey and Block 1994. Bunnell and Vales 1990. all were in the second large animals.

Manaus. The relative Timlin 1998兲. Microclimate at the sites (data loggers) ber 1997. Note the interrupted y-axis. Amazônia. Manaus. bottom兲. middle兲. Monthly rainfall 1996-1999 共y-axis: monthly sums兲 at the weather station of Embrapa Amazonia Ocidental. was marked by an El Niño 共ENSO兲 event 共Wolter and duced in this period 共Figure 3. 295 Figure 1. air humidity was extremely low. Annual rainfall 1971-2000 at the weather station of Embrapa Amazônia Ocidental. Straight horizontal line: average rainfall 共1971-2000兲. Amazonia 共data courtesy of Embrapa兲. whereas the evapo- transpiration 共Piche evaporation兲 peaked in Septem.Dev. The year 1997 The average monthly litter temperatures as recorded with loggers in the litter layer of FLO and POA re- . and the saturation deficit was extremely high in this month 共Figure 3. dotted lines: ⫾ 1 St . fall and the number of rainy days were markedly re. Figure 2.

SEC兲. top兲. 共middle兲 total monthly rainfall and number of rain days per month. the recorded minima represent the night- when the forests 共FLO. time temperature values and did not differ. average monthly evapotranspiration and calculated total monthly satura- tion deficit. data courtesy of Embrapa兲: 共top兲 average monthly air and soil temperatures.7 °C in POA and POL 共Table 2兲 . Amazônia 共monthly values are based on daily readings at the station.296 Figure 3. top兲. Manaus. mained below the air temperatures recorded at the but they were consistently higher in the plantations Embrapa climate station 共circles in Figure 4. top兲. Extreme and the peach palm monoculture 共PUP兲 are compared. POA 共on average about 2 °C above FLO兲 and POL whereas the litter temperatures in the plantation POL 共almost 4 °C兲 共Table 2. 共bottom兲 air humidity. The litter temperatures differed very little and POL. maxima of up to 55. Figure 4. The differences were much higher than at the climate station 共Figure are caused by the higher maxima recorded in POA 4. Climate data recorded 1996-1999 at the weather station of Embrapa Amazonia Ocidental. one polyculture 共POC兲.

were probably recorded when the loggers. Wood tree cultures at two sites 共POA. litter of the sites POA and POL were not only harsher however. the average microclimatic conditions in the in the litter layer outside the case at these moments. in the Fruit tree culture 共POL兲 and in the Palm stand 共PUP兲. Manaus. POC兲. Tmed⫽ mean temperature. these observations confirm that in these two 共average temperatures higher by 2-4 °C. and humid- . 共middle兲 soil temperatures. became exposed to direct face more often than under a closed canopy. in the primary forest 共FLO兲. 297 Figure 4. In other sunlight. 共bottom兲 air humidity in the litter layer. secondary forest 共SEC兲. Amazonia. Embrapa Amazônia Ocidental. and may not represent the true temperature words. enclosed in sites sunlight radiation reached the litter and soil sur- translucent plastic cases. Monthly average microclimate values 1997-1999: 共top兲 litter temperatures.

0 100. probably also in POL兲 than Figure兲.5 86.7 22.7 0.7 Maximum 27.8 1. time litter temperatures in FLO and SEC almost did termining parameter for insect growth and develop.9 27. Figure 4. As the soil param.6 96.3 23. absolute maximum and minimum values recorded in each of the three study periods 共see Table 1兲.4 25.2 25. the litter stocks 共as indicator of the thickness of an even. the soil temperatures almost equaled in 2-hr intervals 共from which the monthly averages the temperatures in the litter layer.3 1. for example.2 30.9 26.6 1.7 20. but much .2 2.7 0. suggest 共Figure 5兲. 0. This again much more variable and unpredicTable than in the confirms the extreme conditions of 1997.9 23. 0. the data are separated in three ubsets. In the other temperature difference still was about 5 °C.5 24.1 28.9 46.8 27.2 17.6 22.1 1998-99 FLO SEC POA POC FLO SEC POA POC Average 24.6 °C much higher in POA.9 25. On clouded The monthly average air humidity at all sites was days 共e. the secondary for. by deduction.5 99.5 27. the amount and frequency of in closed stands like FLO 共Figure 4兲. and one plantation site 共POC兲.0 35.0 Minimum 24. these deviations from 100% is higher in SEC.8 25.3 16. that the differences in microclimate between the sites eters were similar in all sites 共Martius et al.4 2.0 23.4 0. in the secondary forest SEC.6 1998-98 FLO SEC POA POC FLO SEC POA POC PUP POL FLO SEC POA POC Average 25. Embrapa Amazônia Ocidental.1 22.298 Table 2.4 28. average humidity was almost always about rainy days the differences was still in the range sug- 100% in the primary forest 共FLO兲..4 28.6 Minimum 22.5 0. Soil temperature 共°C兲 Litter temperature 共°C兲 Relative air humidity 共%r. 1. est 共SEC兲.6 25.Dev. PUP: Palm stand. SER: Rub- ber stand.6 1.9 5. Figure 4.6 15.5 24. 0.9 2.4 26. Amazonia.7 32. Even on months. Litter and soil temperatures and relative air humidity in the study sites.8 1.0 32.0 28.⫽ standard deviation. same time points to the much drier conditions in the The lowest soil temperatures were recorded in the plantation site POA.3 25.5 100.5 2.0 1.2 2. SEC: Secondary forest.2 n/a 72. the sec- the soil temperatures were considerably lower than ondary forest SEC. Manaus. POA/POC: Wood tree culture in two different sites.7 26. in contrast to the other three primary forest FLO.h.7 96.0 Minimum 23.Dev.3 92. Even small av.1 26. FLO/SEC values during daytime in POA. n/a ⫽ not available. about 2 °C. they were also lower in POA 共Table 2.0 21.0 100.Dev.0 Maximum 27.0 27. standard deviations.6 22. they raised more than 10 °C above the ment 共Begon et al.7 25.1 30.9 0.4 39. the 9th and 10th peaks in Figure 6兲.6 29. Due to technical reasons 共logger battery life duration兲.9 22. and at the other sites.1 24.9 34.9 24. whereas in POA. Like for litter are much higher.6 100.2 3.0 100.6 50.2 55. The data set that shows the original data measured middle兲.0 30.6 n/a 76.6 90.1 3. no correlation of average soil temperature and little around soil temperature in FLO and SEC.1 36. of this data set 共Figure 6兲 reveals that although day- siderably large differences in temperature-days. and does this mainly in between the sites indicate that the soil organisms face the extreme year 1997 共left half of Figure 5兲.3 Averages.2 22.1 8.5 7.9 99.8 20. 2004兲 are even more marked than the monthly averages they cannot have influenced the soil microclimate.0 Maximum 28.9 36.7 100.0 25.8 22.3 26. daytime temperature peaks in the litter layer of POA tually insulating litter layer兲 was found.7 32.8 25. not differ.6 0. but not much harsher conditions in open-structured sites like so during the first months of 1998 共right part of the POA 共and.3 Std. a de. POL: Fruit tree culture.8 23.2 22. Whereas air temperature fluctuates Also.2 25. Std. Dev.4 26. In contrast. and the plantation site POA shows the litter temperatures 共Table 2兲.9 25. Site codes: FLO: Primary forest. bottom兲. and erage temperature differences of. gested above by the average values.6 25.6 26.9 Std. 1996兲.3 23. ity lower by 5-10 percent points兲.7 24.2 27.7 0.3 Std. A closer look at a short interval between FLO and POA 共Table 2兲 will sum up to con. the lowest in September/October 1997. and sites. the differences in soil temperature from 100% only very little. in one plantation site 共POA兲 共Table 2.9 22.2 22.兲 1997-98 FLO SEC POA FLO SEC POA POC FLO SEC POA Average 26.2 22.8 43.g.1 26.1 0.4 2.9 22. The air humidity in FLO deviates temperature. In FLO. were compiled兲 of the primary forest FLO.6 29.

Amazônia. air humidity in the litter layer 共fine black line on top of each graph兲 in primary forest 共FLO. Microclimate data: temperature in litter 共grey dots兲 and soil layer 共thick black lines. middle兲. bottom兲. 299 Figure 5. secondary forest 共SEC兲. secondary forest 共SEC. at Embrapa Amazonia Ocidental. . Figure 6. Amazônia. and Wood tree culture 共POA. Manaus. Manaus. during nine months 1997 and 1998. 2-hr intervals. and Wood tree culture 共POA兲 共2-hr intervals兲 during three weeks in 1997. at Embrapa Amazonia Ocidental. Daily litter temperature oscillation 共Litter-T兲in primary forest 共FLO兲. top兲.

Canopy closure Table 3. variability of canopy cover is much higher in POL ⬍ 0.003 ⬍ 0.001 no test principally in the polyculture area POC and in the SEC no test no test 0.001 SEC and POA than in FLO. 2004兲.001 no test POA no test 0. Soil biology The soil and litter temperatures were both highly correlated with canopy closure. 2001兲 peaked only in POA. Manaus. 共SER兲 rubber stand. How. which no difference was found. Only the The software does not allow to test if the difference between two rubber tree plantation 共SER. but termite . Box plots are for 10 samples per site. Soil microbial biomass was highest in POA 共Table 4兲. and FLO no test 0. 53%兲 and the polycul. 共SEC兲 secondary forest. SER ⬍ 0. Figure 7 and Table 3兲. for the comparison of canopy closure data between the study sites. Amazônia. and circles 5/95% percentiles.001 ⬍ 0. In the top row an example of one typical canopy closure photograph is given for each site: 共FLO兲 primary forest. although the power of The soil biological data followed different trends the test was satisfactory only for litter temperature 共Table 4兲.001 ⬍ 0.001 no test percent point below that in the primary forest. The highest canopy closure 共88%兲 was recorded in the primary rain forest 共FLO兲. 共POA.300 Figure 7. whiskers are 10/90% percentiles. straight lines within box are medians. 共POL兲 fruit tree culture and 共PUP兲 palm stand. Embrapa Amazônia Ocidental. p ⬍ 0. Based on this regression. Dotted lines are averages. P-values of the ANOVA on ranks. Table 1. followed by Tukey test. POC兲 wood tree culture at two sites.001 no test secondary forest 共SEC兲 canopy cover was only one POC ⬍ 0. Manaus. Average canopy closure 共% black pixels兲 in all sites at Embrapa Amazonia Ocidental. 共Höfer et al. POC and PUP. Most of the other sites SEC POA POC SER POL PUP also had a rather high closure of above 80%. cantly from each other.697 no test ⬍ 0. canopy closure and lowest in the rubber tree plantation and in POL can therefore be assumed to be a good proxy of the 共Kurzatkowski et al.002 ⬍ 0. For explanation of site codes see sure than the other sites 共and also differed signifi.005.001 ever. Box limits are 25/75% percentiles. values is smaller than the difference between two other values for ture POL 共19%兲 had a significantly lower canopy clo.001 ⬍ 0. Amazonia. Mesofauna biomass temperature in the litter layer.

23 3.13 2001兲 had their peaks both in FLO and in POC. sites with an open canopy like the multi-tree plantation sites 共POA.800兲. Microbial biomass. Therefore. only a few mg m–2 2713 1391 1368 2332 data points are available for these regressions. The decomposition rate as determined from rofauna taxa##兲 the litterbag experiments. failed to yield a mg m–2 satisfactory regression. k yr–1 2. it was near the conventionally desired 3322 2070 2305 2987 power of 0. top兲. 301 Table 4. 654 305 109 304 We established the regressions between canopy and different soil biological data: microbial biomass. macrofauna. The highest decomposition rates 共Kurzatkowski et al. Röm- Decomposition rate bke et al.兲. meso. Un- # . efficient protection against inter-annual climatic vari- Site ation as well as against microclimatic extremes. al- macrofauna though the power of the tests were generally low 共in Macrofauna 兲 Four dominant mac. the 4-macrofauna-taxa group. higher litter and soil temperatures. and the data should not be over-interpreted. and decomposition rate from litter bag experiments in the study sites. Also.⫹ mg m–2 macrofauna. biomass of different soil fauna groups. Termites mg m–2 2004兲 were measured in SER and PUP. termites. Ama. Microbial biomass in 0-5 cm soil Discussion and Conclusion A closed stand canopy as in the primary rain forest ␮g Cmic g–1 soil site has a strong influence on microclimate in the stands and soil fauna biomass 共Table 5兲. and earthworm biomass 共Martius unpublished.99 0. and for meso. mg m–2 and the joint biomass of the four dominant macro- 1541 259 397 963 fauna taxa. 2001兲 peaked in FLO. me- Earthworms sofauna. 4. Fruit tree culture POL Secondary forest SEC lar irradiation. termites.and macrofauna together.Meso.40 1. Primary forest FLO lower air humidity in the litter.and macrofauna biomass combined as well as the four dominant macrofauna groups pooled 共earth- worms. The canopy is able to buffer extreme climatic conditions 共Figure 397 423 460 409 291 282 376 5. The best correlation coefficients were ob- tained for macrofauna.72 0. POL兲 and the Wood tree plantation POA Wood tree plantation POC ## rubber tree monoculture 共SER兲 suffer from direct so- Rubber tree culture SER 兲Without earthworms. the two mo- noculture plantations. In contrast. but we conclude that canopy closure as a proxy of litter temperature Biomass mg m–2 Meso- has the potential to be a good predictor for the bio- fauna 兲Earthworms. Soil microbial biomass and earth- worms also yielded high R coefficients but low power. This might be due to the fact 2702 705 980 2537 that only one litter type was used in the litterbag ex- periment 共Höfer et al. Diplopoda and Isopoda. Manaus. unpubl. Höfer et al. Embrapa Amazônia Ocidental.78 1.and macrofauna together 共Table 5兲. and macrofauna. diplopods and isopods. meso. a parameter that we expect to integrate soil organismic activity. 609 679 937 655 mass of the group of the four most dominant macro- fauna taxa.95 3. mimicking the rain forests’ canopy closure in agroecosystems would provide an zonia. and that these tests # therefore are not representative of the whole soil de- composer community in the stands. and lower soil fauna Palm stand PUP biomass.

lowest cover兲. and daytime air humidity also decreased. in general is much more similar to tion in POA and POC buffers the soil against climatic FLO and SEC than to POA. Std.7 0. coeff. Although the admitted secondary vegeta- at POC.972 213.2 0.01 2. were relatively more abundant in the soil fraction of The plantations differed not only in canopy closure the samples in July and September 1997.302 Table 5. The fact that microclimatic conditions in the This gives a hint that a mosaic-like landscape design peach palm monoculture are similar to those in the in which plantations are interspersed with rainforest primary forest further underlines the prevalence of patches can offer additional improvement in planta.1 30.3 70.4 0.9 0. diplopods and isopods # der lower canopy closure 共SER and POL.1 ⫺ 3105. The the tree diversity of these two sites.67 42. 2001兲 and the has detrimental effects on microclimate in these macrofauna 共Höfer et al. although in all systems four different quent.965 0. however.e. termites. We assume 共although this ary growth in these stands is not considered here. The diversity of second.6 ⫺ 8447. 2002兲 patches.2 0. paralleled by differences in decomposition rates tures.938 0.976 0.968 Microbial biomass 共␮g Cmic g–1 soil–1兲 2.4 0. Figure 5 bottom兲.1 0. % r. The a period of unfavorable microclimatic conditions in rubber tree and peach palm plantations 共SER. Microclimate the stands. Also. 0.0 0. the two plantation sites that have the same have any strong influence on the microclimate within plant composition and stand structure.405 Termites mg m–2 41.986 0.738 0. POC and POL at least four tree spe.064 Decompos. Linear regressions 共y ⫽ ax⫹b兲 fitted to microclimate and soil biological data as variables 共y兲 dependent on canopy closure 共x.931 187.1 0.694 Meso. Franklin et al.1 226. We assume that POC is extremes when compared to POL. in the polyculture stands. but differed completely with regard to canopy 共Höfer et al. This effect is strongest cies were planted together. The same is true for the cover 共Figure 7兲. the differences in soil fauna densities between the stud- canopy closure in these stands was not correlated to ied systems averaged over the whole study period are the tree diversity.634 Air humid.753 Biomass of 4 dominant macrofauna groups#兲 mg m–2 240.0 0. during but also in the diversity of the planted crop trees.4 ⫺ 18423.2 0. stands.and macrofauna biomass mg m–2 132.3 0.151 0. 2001兲. i. 2001.3 ⫺ 11319.990 0.0 0.0 0.7 0.544 122. Nevertheless. Amazonia.172 3.173 兲 Earthworms.725 0. canopy cover in POA and enchytraeids and other microdrilid worms in FLO and .852 0..5 0. units ⫽ canopy closure in %兲. the upcoming vegetation that was withshort-term vertical migrations to changes in the tolerated between the planted rows of trees added to moisture and temperature conditions of the litter.980 117. diversity alone does not and POC. whereas in the polyculture litter fraction during the rest of the study period systems POA.049 Mesofauna biomass mg m–2 25. Bierregaard et al. Although a careful choice of markedly 共Figure 4. for the study sites at Embrapa Amazônia Ocidental.879 249.0 0. Manaus. error estimate Power 2 y a b R R Soil temperature °C 0.818 0.1 0.1 0. large-scale clear-cuttings 共although a fragmentation Additional data show that both the mesofauna of the rainforest ecosystem into too small remnants 共Mites and Collembola.243 Litter temperature °C 0. site structure over site diversity as a decisive factor tion stand microclimate than would be the case on of microclimatic conditions in the stand. They were relatively more abundant in the were monocultures. Hanagarth et al.6 ⫺ 10074. SER and PUP were both monocul.6 0.3 0. often tree crops were planted. the structurally different trees in plantations may con- An interesting detail is the difference between POA tribute to canopy closure.7 2872.522 Earthworms mg m–2 129.769 0. rate k yr–1 0. 共March 1998 to March 1999兲.414 0.1 31. this effect seems much better protected from temperature extremes by not to be related to the diversity of the secondary its close neighborhood to a primary forest stand growth but to the increased canopy closure in these 共which provides shadow to POC in the afternoon兲. Figure 7兲 POC was much higher than in POL 共the site with the the temperature extremes became much more fre.023 1.6 0. Dependent variable Units of y Regression Correl. PUP兲 the litter.155 Macrofauna biomass mg m–2 158.h. but was not measured兲 that the soil fauna reacts in POA and POC. 2001兲.953 0.591 177.

and Santos E. of degraded. Hanagarth necticut. Bohlman S. Bowers C.. Harper J. Martius C.. Hanagarth W. Andrias 15: Acknowledgements 141–153.B. Moisture late with these results 共Martius et al. Over the whole guilds? Soil Biology and Biochemistry 34: 367–372. 1990. and we and temperature patterns of canopy humus and forest floor soil of a montane cloud forest. and Zech W. Wolters V. 2000. logistically supported the study. Matelson T.. 共Tubificidae兲. Gascon C. Bonn. Amelung W.. Nutrient Cycling in Agroecosystems. A comparison of two techniques We thank the German Research Ministry for measuring canopy closure. canopy closure. Morais J. Appl. For..A.. ecosystem types instead of large-scale clear-cutting兲 FAO/ Unesco 1990. Jr. Garcia M. Natural systems as models for the design of sus- 共closed canopy. Nunes J. macrofauna was found in the soil. Eur. 69: 111–118. Brenner K. Bra- Vlek P.. ..N. Western Journal of Applied For- 共Bundesministerium für Bildung und Forschung – estry 9: 21–23.. In press.E.W. Gasparotto L. viduals. Franklin BMBF兲.. 146 pp. Collado R. and Dhillion S. decisive for the density and distribution of the soil Bunnell F. Feldmann F. Stutzman T.L. Roger P. measurement. Martins G. Idczak E. Japan. populations and communities. well developed canopy as in the peach palm monoc. 3rd 共ed. J.兲. Density and biomass of Acari and Collembola in primary forest. Bot.. Manaus. selho Nacional de Desenvolvimento Cientifico e Jolliffe P.. et al. USA. Höfer H. study period. 20: 101–107.. Biol.O. Martius C. Soil function in a changing a previous draft of this paper. Rome. and Block W. 2002兲. 1994. USA.e.sp. Sustainable management of soil resources in the hu- providing helpful information about canopy closure mid tropics. 1068 pp. Lovejoy T. VA-mycorrhizal fungi. and Vales D.. and support. Italie. Recultivation fauna and microflora of the soil together with the im. Soil map of the world. Bierregaard R. 1994. Blackwell Sci- 共During dry periods in El Niño years. Pedonais crassifaucis n. 2004. 1999兲. Water status and development of tropical trees ter layer.sp. but this was not so 2001. Ineson P.. fallow lying areas in central Amazonia with equili- portant functions these organisms exert in the ecosys. Franklin E. and Irwin L. Comparison of methods for es- fauna. microbial biomass and ac- zil. 303 SEC that occurred preferentially in the litter layer at References all sampling dates except in 1997 共Römbke et al.L. 1995. that the mimicking of a natural forest structure Ewel J. Annelida兲 from Central Amazonia. Lignin characteristics and density fractions of termite treated to deeper layers also in the dry season of 1998 nests in an Amazonian rain forest – indicators of termite feeding 共i. Two anonymous reviewers are also Lavelle P. Litter decomposition. Costa Rica. and Römbke J. Soil 共project number 0339675兲. 1995. of polyculture sites is much more extreme than in Can. These results indicate forest canopy cover. 2001. CA.R. Res. Tokyo. 1999. 1995. mosaic landscape of intermittent tainable systems of land use.. POA again was the a fragmented Forest. Trees 8: 115–125. Embrapa Amazônia Ocidental. 2004兲. Wildlife Society Bulletin 23: 711–717.. 1995.W. United Nations University Press. revised legend.. World can be successfully used for the management of mi. Pre- croclimatic conditions that affect the decomposer isinger H. J. London. Martius C. and ultimately. Spherical densiometers produce biased estimates of and high temperature peaks. 2001. in the 12-year old secondary forest 共Clitellata. 共Naididae兲 and Bothrioneurum righii n. 37: 229–235.gen. Are mixed populations of plant species more Technologico 共project number 03396/3兲 for financial productive than pure stands? Oikos 80共3兲: 595–602. Agroforestry Systems 45: 1–21... Soil Resources Report 60. and Nadkarni N.V. Biotropica 27共1兲: 13–19.W.R.H.A.. 1996. secondary and primary forests in Amazonia.. Kurzatkowski D.M. Ganey J. Yale University Press. Con- most extreme site with regard to this ratio. and Mesquita R.G.F. and the Brazilian Con. 1997. for Lal R. thanked for their comments which helped improving Heal O. the ratio of litter fauna to soil fauna was Begon M. relatively more ence Ltd. Germany for funding of the project E... Förster B.. and Schmelz R. J. brated polycultures: response of useful plants to inoculation with tem. 119 pp.G..L. Lepage M. Amazoniana 16: 共SEC兲 or in the vicinity to a closed forest 共as in POC兲 223–235. Redding.M. New Haven. are factors that offer protection from high variation Cook J.. fornia Center for Forestry..J. therefore assume that it is the microclimate in the lit- Borchert R. UK.W. Höfer H. Garcia M.... Moraes V. Beck L. 1997. 2001.. but a n.M. Ecology: indi- 50:50 in FLO and SEC but 30:70 in POA and POC. two new tropical soildwelling species of “aquatic” oligochaetes ulture 共PUP兲. that are most during seasonal drought. secondary regrowth and polycultures in central Amazonia. Lessons from Amazonia: the ecology and conservation of expressed in the rainy season. but not in POA and POC where the worms re. The microclimate in the litter and soil layers timating forest overstory cover: differences among techniques. FAO. and Townsend C.A.J. Bandeira A. and Lieberei R.. Structure and function of the soil fauna SHIFT ENV 52 “Soil fauna and litter decomposition” in Amazonian anthropogenic and natural ecosystems. Sincere thanks tivity of soil organisms in three agroforestry sites in central also go to Gary Nakamura of the University of Cali. Beck L.. Bignell D. Litter quantity and quality do not corre. June and September 1998兲.J.L. Amazonia.

Silva C. M.L. ties in central Amazonian primary and secondary forests and a semblage of plant species. Souza M. 2004.A.E.D. Coelho L. letting you access and read them immediately.F. Håkansson K. 1992. Lawton J. How important are species richness. 320 pp.B.. UK.C. Wolter K. Nijs I. 4th. 1996.G. Martins L. and Roy J. and Adis J.. Manaus. and Procópio L. and Hanagarth firme na Amazonia Central..D. An introduction to tropical rain forest. Höfer H. . do S.. Oxford University Press Inc..J.. 768 pp. Pedobiologia 43: 518–522..G. Hopkins M. Preisinger H. Germany. and Garcia M.J. CAB Interna- Manaus. 1984. Eur J Soil Ribeiro J.L.S. Pereira E.G.C.R. 1999.C.. edition...A. Sothers C. Römbke J. Berlin.B. Brazil.P. Angew Botanik 68: 40–46. All in-text references underlined in blue are linked to publications on ResearchGate. Brito J.V. Amazonas.H.A.M. 1999. Analysis of growth form types and floristic composition – how does 1997/98 rank? Weather 53: 315–324. In press. Biodiversity and plant productivity in a model as. spe. INPA-DFID.. Oikos 76: 259–264. 1998. M. Whitmore T.. Amazonica 14共12兲: 159–174.. 2nd cies evenness and interspecific differences to productivity? A ed. 1998. USA. Nutrient tential bias for bioecological studies in the Central Amazon.. and Thomp. and Timlin M... indentificação das plantas vasculares de uma floresta de terra Martius C. W. Naeem S.V. Meller M. 799 pp. Earthworm densi- son L.G.. Flora da Reserva Ducke. Local rainfall variability a po- forest and agroforestry systems in central Amazonia.J. Measuring the strength of ENSO 1994. D’AnsLax Taschenbuch für Chemiker und L. Assunção P. and Lieberei R. New York.. Vicentini A. Garcia M.H. Band I Physikalischchemische Daten. Mesquita Physiker. Römbke J. Oikos 88: 57–66. 2000. Acta Cycling in Agroecosystems. Guia de SpringerVerlag. Agroforestry for soil management. Siqueira M. Litter fall.F. litter stocks and decomposition rates in rain Ribeiro M.S.. 251 pp.. mathematical model. polyculture forestry plantation.. tional.. New York. 304 world: the role of invertebrate ecosystem engineers.A. Wallingford. Crawley M. Costa Biol 33共4兲: 159–193.C. Lohmannn Lechner M. 1997.N. of the spontaneous vegetation in an agricultural test area near Young A.