Seed predation by amazon fish

Our experiment revealed unexpected behavioral strategies involved in seed predation of Mabea fistulifera by fish (Brycon cephalus), including temporal storage of seeds in the stomach followed by regurgitation and reingestion of individual seeds. Larger fish were faster at removing the seed coat and exposing the endosperm, due to precise oral manipulation of seeds.

Methods
Our experiment was conducted at the Laboratorio de Ictiologia Tropical (LIT) of the Federal University of Mato Grosso, Brazil, for 12 days during July 2019. Brycon cephalus individuals (N = 8, 10.1-28.0 cm standard length; maximum length is 45 cm TL, www.FishBase.org) were retrieved from a nearby fish-farm. Individual fish were kept in 61-liter aquaria, equipped with a water recirculation system, and water quality was monitored daily (temperature: 28.5 ± 0.6 ºC, pH: 7.03 ± 0.04, and dissolved oxygen 5.78 ± 0.42 mg/L).

Aquaria were protected with a dark screen, containing a small opening for the observer, aiming not to cause an observer's effect on the fish's behavior (Sabino & Carvalho, 2008), and equipped with remotely operated Go Pro cameras positioned above and on the sides of the tanks. During the experiment, focal fish were fed daily at 0800 h with individual seeds (one by one) extracted from ripe fruits of Mabea fistulifera (seed width: 6.04 ± 0.77 mm, length: 8.48 ± 0.22 mm, weight: 0.5 ± 0.4 g) and later, at 1700 h with extruded commercial feed, aiming to complement the species' diet. Mabea fistulifera Mart. (Euphorbiaceae) is a pioneer tree, commonly found in riparian forests. The fruit is a dry, subglobous capsule (length: ≈ 15.5–19.0 mm), whose epimesocarp cracks in explosive dehiscence, expelling the seeds (Vieira, 1991). Seed ingestion occurred ad libitum (Altmann, 1974), where we considered as satiated a fish that did not show interest on a seed after 15 min of it being offered. At that point, we removed non-consumed seeds. Observations during a pilot trial revealed that B. cephalus regurgitate seeds. Thus, we searched the tanks for regurgitated seeds at 15-min intervals, during 4 hrs. after the last feeding event. After 4 hrs., visual surveys were extended to 30-min intervals until the last regurgitated seed was observed (i.e., the fish regurgitated all the ingested seeds). We repeated the experiment four times per fish, with a 24 h period between trials.

We assessed four metrics of seed predation including: the (1) number of seeds ingested (NSI); (2) time to regurgitation of the first seed (TRFS), defined as the period between the ingestion of the last seed, without the fish having any food in their mouth, until the beginning of the oral manipulation activity; (3) time to regurgitation of the last seed (TRLS), defined as the period between the ingestion of the last seed, without the fish having any food in the mouth, until the time when oral manipulation activity of the last regurgitated seed was observed; and (4) handling time (HT), defined as the average time to chew each regurgitated seed that was reingested. We analyzed whether fish size influences these metrics of seed predation via separated mixed effects regression models, where fish size was the predictor variable and fish identity was included as a random effect to account for the lack of independence between multiple trials (N = 4) per fish. Prior to analysis, we transformed NSI via square root to enhance homogeneity of the variance. Regression models were implemented in R (version 3.6.3), package lme4 (Bates et al., 2015); significance of the fixed factor was assessed with Wald’s test, package car (Fox & Weisberg, 2019); plots were created in package ggplot2 (Wickham, 2016).