Agriculture As Matchmaker of an Unexpected Mutualism: Great Bustard Disperses and Enhances

Agriculture as matchmaker of an unexpected mutualism: Great bustard disperses and enhances emergence of domestic olive seeds

Miguel Delibesa, Casimiro Corbachob, Gemma Calvoa, José María Fedriania,∗

a Estación Biológica de Don˜ ana (CSIC), Avda. Américo Vespucio s/n, Isla de la Cartuja, 41092 Sevilla, Spain

b Área de Zoología, Facultad de Ciencias, Universidad de Extremadura, Avda. de Elvas s/n. 06071 Badajoz, Spain

Abstract

By changing the habitats and altering plant traits, agriculture has severely disrupted many plant–animal mutualisms. Interest- ingly, however, the intensification of agricultural practices could also facilitate mutualistic relationships between species with naturally mismatching phenotypes. We illustrate the potential of the great bustard (Otis tarda), a large steppe bird, as disperser of domestic olive (Olea europaea) seeds, originally a forest species. In an area of southwestern Spain, 30% of bustard faeces included olive stones (from 1 to 13). Only 1.7% of the bustard-ingested olive seeds were broken. Moreover, using a sowing experiment, we show bustard ingestion enhanced seedling emergence, which reached 8.8%, 3.4% and 0.0% for bustard-ingested, hand-depulped, and control seeds, respectively. As expected for Mediterranean habitats, seedling mortality was very high in the first summer for all seed treatments. In 6 out of 19 non-plowed patches within our study area, we found olive saplings of different ages likely to emerge from bustard-dispersed seeds. Given the large size of domestic olive fruits, bustards are among the few local animals able to disperse their seeds and thus to assist in the forestation of field boundaries and abandoned lands. Paradoxically, because bustards are rather restricted to open habitats, their success in shaping the habitat (i.e., ‘planting’ olive trees) could lead to their own removal from the resulting forested landscape.

Zusammenfassung

Indem sie Lebensräume und die Eigenschaften von Pflanzen verändert, hat die Landwirtschaft nachhaltig zahlreiche Tier-Pflanze-Mutualismen gestört. Indessen kann die Intensivierung der Landwirtschaft auch mutualistische Beziehungen ermöglichen, und zwar zwischen Arten, deren Phänotypen von Natur aus nicht zueinander passen.

Wir demonstrieren das Potential der Großtrappe Otis tarda (ursprünglich ein Steppenvogel) als Samenverbreiter für die Kulturolive Olea europaea, die ursprünglich eine Waldart ist. In einem Untersuchungsgebiet im Südwesten Spaniens enthiel- ten 30% der Kotproben der Großtrappe Olivenkerne (1–13 Stück). Nur 1,7% der von Großtrappen aufgenommenen Kerne waren zerbrochen. In einem Aussaatversuch zeigten wir, dass die Aufnahme durch Großtrappen die Keimung begünstigte: die Keimungsraten betrugen 8,8% für von Großtrappen aufgenommene Kerne, 3,4% für Kerne, die von Hand entpulpt wor- den waren, und 0,0% für unbehandelte Kontrollen. Wie für mediterrane Habitate erwartet, war die Keimmortalität im ersten Sommer bei allen Behandlungen sehr hoch. Auf 6 von 19 ungepflügten Stellen fanden wir in unserem Untersuchungsgebiet Olivenschößlinge, die vermutlich aus von Großtrappen verbreiteten Samen hervorgegangen waren.

∗ Corresponding author. Tel.: +34 954466700; fax: +34 95462125.

E-mail address: (J.M. Fedriani).

Angesichts der Größe der Kulturoliven gehören Großtrappen zu den wenigen einheimischen Arten, die ihre Samen verbreiten und damit zur Bewaldung von Feldrainen und Brachflächen beitragen können. Da die Großtrappe eher in offenen Lebensräumen vorkommt, könnte ihre erfolgreiche Habitatgestaltung durch das “Pflanzen” von Olivenbäumen zu ihrem eigenen Verschwinden aus der resultierenden aufgeforsteten Landschaft führen.

Keywords: Artificial selection; Fruit size; Habitat shaping; Mediterranean habitat; Old-field recolonization; Olea europaea; Seed dispersal; Spain

Introduction

Intensive agriculture has severely modified not only the patterns of land use worldwide, but also the genotypes and phenotypes of cultivated plant species. As a consequence, agricultural practices have disrupted many plant–animal mutualisms, which are necessary for sustaining life on Earth (e.g., pollination, seed dispersal; Bond 1994; Ellstrand

2002; Tylianakis, Didham, Bascompte, Wardle 2008). For instance, habitat changes and insecticides have generated an important global deficit of pollinators, which in some areas even threatens some crop yields (e.g. Klein et al., 2007; Aizen, Garibaldi, Cunningham, Klein 2009). Similarly, habitat loss and hunting are alarmingly reducing the num- ber and diversity of animal seed dispersers, which seriously threatens forest renewal (e.g. Moran, Catterall, Kanowski

2009). A less considered intriguing effect of agriculture is the enabling of novel ecological interactions when, for example, crop species brought beyond their natural range are polli- nated by native species (e.g. Kremen, Williams, Thorp

2002). Even less explored is the potential of agriculture as a ‘matchmaker’ of mutualistic interactions between species with naturally mismatching phenotypes.

In the case of plants dispersed by vertebrates, agriculture most often has disrupted this mutualism by modifying the composition and structure of habitats and changing the traits of cultivated plants. For example, by artificially enhancing fruit size, relatively small-sized dispersers become unable to swallow them and subsequently deliver their seeds. As a result, many potential seed dispersers become pulp predators (Rey and Gutiérrez 1996) and occasionally even agricultural pests (e.g. De Grazio 1978).

The olive tree (Olea europaea) and its seed consumers and dispersers represent a good example of such mutualism disruptions, as recently stated by Rey (2011). In Southern Spain, wild olive fruits are consumed by many species of small birds, especially of the genera Sylvia and Turdus, but most of them are unable to swallow the big drupes and/or to live in cultivated monospecific olive orchards, because of the lack of a diverse understorey vegetation and of comple- mentary food (Rey 1993, 2011; Rey, Gutiérrez, Alcántara,

Valera 1997). As a consequence, domestic olive seeds are thought to be scarcely dispersed by wildlife, in spite of many drupes remaining on the trees after harvest.

As mentioned above, however, human habitat modification and artificial selection of fruit traits could result in a matching

of naturally mismatched phenotypes. To illustrate this over- looked possibility, we report about the role of a large steppe bird, the great bustard (Otis tarda), as a legitimate disperser of a cultivar of domestic olive seeds in southwest Spain. We show that: (i) bustards regularly eat olive fruits; (ii) bustards regularly defecate undamaged seeds in sites apparently favor- able for seedling emergence; and (iii) emergence is enhanced as a result of seed passage through the bird gut. Seed pas- sage through bustard guts includes two different treatments potentially altering the amount and speed of seedling emer- gence: the mechanical removal of the pulp and the mechanical and chemical scarification of the stony endocarp (Traveset, Robertson, Rodríguez-Pérez 2007). Thus, (iv) we distin- guish between the effects of these two treatments following Samuels and Levey (2005). We also evaluate difference in size between bustard-ingested and non-ingested seeds and suggest potential underlying mechanisms. Finally, we discuss the potential of agriculture as a ‘matchmaker’ of new inter- actions, as exemplified by the bustard-olive tree relationship in southern Spain, and speculate about its implications in a scenario of reforestation of set-aside lands and in the con- text of invasive species research (e.g. Spennemann Allen

2000a).

Study site and system

The study was carried out at Llanos de la Albuera-Valverde de Leganés, an area of about 10,000 hectares near Bada- joz city (Extremadura, SW Spain). Climate is Mediterranean subhumid, with hot and dry summers and wet and mild win- ters. Annual rain is about 400–600 mm and average annual temperature 14–16 ◦C. The area is a flat, mainly unirrigated and very heterogeneous agro-ecosystem with cereal (about

45%), vineyard (15%), olive (13%), sunflower and legu- minous (10%) cultivations, and some open “dehesas” and pastures.

The oleaster or wild olive tree Olea europaea var. sylvestris L. (Oleaceae) is a small tree domesticated long ago at several times in several places to become the more characteristic cul- tivated tree in the Mediterranean Basin (Loumou Giourga

2003). The global cultivated surface of domestic varieties exceeded 100,000 square km in 2008, with 28% of this in Spain and Portugal (http://www.fao.org/corp/statistics/en). World average yield of olive drupes is about 1.7 tonnes/ha, mainly devoted to the production of oil. The fruit pericarp

comprises a thin epicarp, a fleshy mesocarp, and a stony endocarp that encloses the embryo. Olive orchards are monospecific stands with trees regularly distributed. They are important feeding places for wintering small-sized birds in Mediterranean Europe (e.g. Rey 1993).

In our study site, most olive trees belonged to the “car- rasquen˜ a” cultivar and spacing among trees was usually between 8 and 16 m. This variety is characterized by the large size of the fruits, which are used preferentially for direct consumption; usually harvest is early, at the begin- ning of the autumn, but many drupes (5–10% of the crop; unpublished data) remain unharvested on the trees, where they continue to mature before falling to the ground during the winter. Average measures of 161 fallen drupes, collected at the beginning of February 2011 under 16 different trees in

three orchards, were: weight (g) = 4.32 ± 0.08 (mean ± SE), length (mm) = 22.49 ± 0.22, width (mm) = 17.95 ± 0.12.

The great bustard is probably the heaviest extant flying bird

in the world. In Spain, mean weight of females is 4–5 kg and that of males is 10–12 kg (Alonso Palacín 2009). It ranges across central and southern Europe (with a small population in northern Morocco), Western Russia and some temperate areas of central and eastern Asia to the Pacific, occupying open steppe grasslands and extensively cultivated fields (Del Hoyo, Elliot, Sargatal 1996). The status of the species in the 2011 IUCN Red List is Vulnerable (IUCN 2011). Great bustards are omnivorous, eating mainly green plant mate- rial and secondarily insects, with grains (wheat, barley, etc.) and some other seeds also being common foods (Alonso and Palacín 2009). In Spain they are partially migratory, with most males and about half of the females making seasonal movements, which can reach up to 260 km. Natal dispersal averages 18 km but reaches up to 180 km (Alonso and Palacín

2009). In addition, when disturbed at their resting and feed- ing grounds, bustards often run or fly several hundred meters to some kilometres away (Sastre, Ponce, Palacín, Martín, & Alonso 2009). Thus, the potential of the species to disperse seeds over long distances is evident, but as yet unexplored. In our study site, about 200–250 bustards are resident and breed at the zone, but about 1500 individuals, particularly females, stay there during the winter (November to Febru- ary; Corbacho et al. 2005), which coincides with the olive ripening season.

Methods

To assess the potential of bustards as dispersers of domestic olives, we collected their seeds during the end of two non- consecutive ripening seasons (i.e., from January to February) in 2007 and 2011, respectively. During the 2007 season, we collected olive drupes and bustard faeces in and around three orchards (separated by 1.5–6.0 km) and used those samples for our seed sowing experiment (see below). In the 2011 sea- son, we collected bustard faeces in two of the three orchards and these samples were used to estimate the following

metrics: (i) the proportion of faeces with olive stones, (ii) the number of seeds per faecal sample, and (iii) the per- centage of seeds damaged after gastro-intestinal passage. In addition, we measured and compared the size of bustard- defecated and hand-depulped seeds collected in two target orchards.

The effect of bustard-ingestion on seedling emergence was evaluated through a common garden seed sowing experiment. To separate the effect of pulp removal from that of seed scar- ification on the percentage and speed of seedling emergence and survival (Samuels Levey 2005), we compared the fate of bustard-ingested (n = 250), hand-depulped (n = 175), and control seeds (i.e., whole ripe drupes with the pulp attached; n = 175) collected in the same tree olive orchards. Ripe olive fruits were collected from the ground under a minimum of five trees per orchard. A fraction of such drupes was depulped to obtain clean seeds, while the remaining fraction was used as “control” seeds in the sowing experiment. In March 2007,

seeds were shallowly (∼5 mm depth) covered with in situ

soil within open-bottomed plastic pots (18 cm in diameter)

set in open ground. Pots (10, 7 and 7 for bustard-ingested, hand-depulped, and control seeds, respectively) were buried about 14 cm, with the rim remaining 2 cm above the surface. In each pot, 25 seeds of a particular treatment were sowed. To avoid removal by vertebrate seed predators (e.g., rodents; Alcántara, Rey, Sánchez-Lafuente, Valera 2000), we pro- tected the sowings with wire cages. We monitored monthly seedling emergence and survival from the sowing to February

2011.

To evaluate whether olive seed dispersal by bustards lead to some seedling establishment within suitable habitats in our study area (i.e., non-plowed patches), we rigorously surveyed (in September of 2011) 19 unplowed patches of variable size within the plowed matrix of cultivated lands. Most of them (16) were small (10–15 m2 ) and corresponded to the area beneath Quercus ilex and Pinus pinea adult trees scattered throughout the study site (see Appendix A).

The remaining surveyed patches are two hedgerows (∼40 and 50 m2 , respectively) and an area (∼4000 m2 ) temporally waterlogged during winter storms. In addition, we surveyed

four linear transects (211, 304, 621, and 716 m) along olive cultivation edges that appeared to be seldom plowed.

Data on seed size were analyzed fitting generalized lin- ear mixed models using Glimmix procedure (SAS 2008). Because a preliminary mixed model for the percentage of seedling emergence did not converge, such data were ana- lyzed by fitting a generalized model using Genmod procedure (SAS 2008), with pot as the experimental unit. For the response variables seed size (length and width) and pro- portion (seedling emergence), we specified in the models the appropriate error (normal and binomial, respectively) and the canonical link function (SAS 2008). In the case of seed size, the source orchard and the block (i.e., faecal sample and the olive tree for bustard-ingested and control seeds, respectively) nested within orchard were included in the mixed models as random factors and thus their potential