Root foraging elicits niche complementarity-dependent yield advantage in the ancient 'three sisters' (maize/bean/squash) polyculture

Abstract

Background and aims: Since ancient times in the Americas, maize, bean and squash have been grown together in a polyculture known as the 'three sisters'. This polyculture and its maize/bean variant have greater yield than component monocultures on a land-equivalent basis. This study shows that below-ground niche complementarity may contribute to this yield advantage.

Methods: Monocultures and polycultures of maize, bean and squash were grown in two seasons in field plots differing in nitrogen (N) and phosphorus (P) availability. Root growth patterns of individual crops and entire polycultures were determined using a modified DNA-based technique to discriminate roots of different species.

Key results: The maize/bean/squash and maize/bean polycultures had greater yield and biomass production on a land-equivalent basis than the monocultures. Increased biomass production was largely caused by a complementarity effect rather than a selection effect. The differences in root crown architecture and vertical root distribution among the components of the 'three sisters' suggest that these species have different, possibly complementary, nutrient foraging strategies. Maize foraged relatively shallower, common bean explored the vertical soil profile more equally, while the root placement of squash depended on P availability. The density of lateral root branching was significantly greater for all species in the polycultures than in the monocultures.

Conclusions: It is concluded that species differences in root foraging strategies increase total soil exploration, with consequent positive effects on the growth and yield of these ancient polycultures.

Keywords: Cucurbita; Phaseolus vulgaris; Zea mays; bean; intercropping; maize; niche complementarity; nitrogen; phosphorus; plant nutrition; polyculture; root architecture; root foraging; squash; ‘Three sisters’.

Fig. 1.

Visualization of 40-day-old root systems of maize, common bean and squash as simulated by SimRoot, a root architectural model. The primary (tap) roots of maize and bean are obscured by nodal and basal roots, respectively. From Postma and Lynch (2012).

Fig. 2.

Field layout of (A) maize or bean monoculture plot, (B) squash monoculture plot, (C) maize/bean/bean (MBB) and (D) maize/bean/squash (MBS) polyculture. Plants were always planted in a triangular station, and plants from monoculture maize were replaced by bean and squash. Within the stations, plants were 10 cm apart, and the distance between the centres of adjacent stations was 60 cm.

Fig. 3.

Expected (expt.) values, which are the weighted mean of monoculture values, and observed (obs.) values of shoot biomass and root volume density (RVD) for the different species in the two polycultures and four nutrient treatments 80 d after emergence in 2007. The contributions of each species are stacked to show the total value of the whole cropping system. Data are means of four replicates. Summary ANOVA results are provided at the system (above) and individual (below) levels, with detailed information presented in Supplementary Data Tables S3 and S4. Culture (Cult) represents the difference between observed and expected growth in polyculture; system (Sys) represents the difference between MBB and MBS; cropping (Crp) represents the difference between each species among the three cropping systems, monoculture, MBB (maize intercropped with two bean plants) and MBS (maize intercropped with bean and squash plants). The value in the ANOVA table is the χ² value. We left out non-significant values in order to condense the table. NS, not significant, *P < 0·05, **P < 0·01, ***P < 0.001.

Fig. 4.

Shoot biomass complementarity and selection effect for the cropping systems MBB (maize intercropped with two bean plants) and MBS (maize intercropped with bean and squash plants) with different nutrient treatments 80 d after emergence in 2007 and 50 d after emergence in 2008. Data are shown as mean ± s.e. Since the nutrient effect did not significantly affect the difference in shoot biomass, the nutrient data were pooled to conduct t-tests if changes in shoot biomass differed from zero. The results showed that the complementarity effect was significantly greater than zero (P = 0·00238) but the selection effect was not significantly different from zero (P = 0·361).

Fig. 5.

Land equivalent ratio (LER) of yield across all treatments in the 2007 (A) and 2008 (B) field experiments. Values are shown as mean ± s.e. The horizontal dashed line is drawn at an LER of 1. The LERs of MBB (maize intercropped with two bean plants) and MBS (maize intercropped with bean and squash plants) were >1 but were not significantly affected by nutrient application or cropping system.

Fig. 6.

Nitrogen (A) and phosphorus (B) per unit area comparing observed and expected polyculture performance across different nutrient treatments 80 d after emergence in 2007. Expected values of each polyculture system are based on the weighted average of the monocultures. Values are shown as mean ± s.e. ANOVA results show that the polycultures absorbed more nutrients than expected. Culture (Cult) represents the difference between observed and expected growth in polyculture; system (Sys) represents the difference between MBB (maize intercropped with two bean plants) and MBS (maize intercropped with bean and squash plants). The value in the ANOVA table is the χ² value. Non-significant factors are omitted for brevity. NS, not significant, *P < 0·05, **P < 0·01, ***P < 0·001.

Fig. 7.

Root volume density with depth of each species within soil cores across all cultivation and nutrient treatments. Monoculture (A) maize, common bean and squash roots (1–3) were sampled 80 d after emergence in 2007, and plants were sampled in MBB (maize intercropped with two bean plants) (B) and MBS (maize intercropped with bean and squash plants, namely the ‘three sisters’) (C). Values are shown as mean ± s.e.

Fig. 8.

Root length density variance (RLDV) within root cores across different treatments 80 d after emergence in 2007. The expected values are based on the average of the monocultures. Nitrogen did not significantly alter RLDV, but the expected RLDV of polyculture was greater than the observed value in both HP and LP plots. This occurred in both the MBB (maize intercropped with two bean plants) and the MBS (maize intercropped with bean and squash plants) system. Data are mean ± s.e. Since phosphorus treatment in 2007 was not replicated, RLDV between observed and expected values was analysed with the t-test. Asterisks above the lines indicate significant difference between observed and expected values under LP or HP for MBB and MBS. *P < 0·05, **P < 0·01, ***P < 0·001.