Experimental site
Two separate release experiments were conducted in August 2019 in the Arbor orchard in the Oulmes region at an altitude of 1240 m (33° 26′ 19.6′′ N, 5° 58′ 35.7′′ W). The first experiment was carried out on the Skarlet apple plot, abbreviated 19B with approximately 4 m tree high, and the Jeromine rootstock. The second one was conducted in the Golden delicious plot, namely 6A, with 300 trees 4 m high. Both plots have histories of high P. ulmi infestations and were commonly managed by 3–5 applications of acaricides. All selected trees in both release experiments are of similar age (6 years) and were not treated with acaricides at least 40 days prior to the experiments.
The initial population of the predatory mite, T. setubali, was obtained from the Riyad fruit orchard in the Tiddas region at 500 m of altitude (33° 33′ 37.0′′ N, 6° 15′ 40.8′′ W), where a population had naturally maintained on the Anna apple cultivar a few years ago. Both plots are located on the northern talus of the Moroccan Middle Atlas within a range of 50 km (Fig. 1).
Experimental design
A single release of T. setubali at 50, 100 and 200 individuals per tree was carried out on the 2 apple varieties with different initial P. ulmi densities. Before the beginning of experiments, each plot was sampled using a binomial procedure to determine initial prey densities. The first experiment carried out on the Skarlet cultivar has moderate initial P. ulmi density (4–6 P. ulmi motile stages/leaf), whilst the second was performed on a highly infested Golden delicious plot, comparatively to the economic threshold of 5 motiles/leaf, proposed by Nyrop et al. (1994) as a critical density related to P. ulmi in July. For each experiment, a control plot of 5 apple trees was selected for similarity in the varietal system, age of plantation, and field management.
A factorial design combining all modalities of release rate and sampling date within paired groups was adopted because it improves the experiment and reduces residual errors. Each plot is designed as a randomized complete block (RCBD), in which treatment levels are assigned to similar experimental units.
Release of T. setubali
The release timing was the same on both plots. One release point per tree test was selected in each of the two experiments. The release rate tested was 0 (no release control), 50, 100, and 200 predators per single test tree. The efficacy of T. setubali was evaluated based on the surviving number of P. ulmi eggs and motile stages every week after the initial release.
To determine P. ulmi eggs and motile stages densities, 16 randomly selected leaves (4 leaves per orientation) were taken from each test tree at 0, 7, 14, 21, 28, 35 and 42 days after release. The collected leaves were kept in coolers and transferred to the laboratory, where they were examined under a binocular stereomicroscope. The number of survived individuals after each treatment was minutely counted, and the results were recorded as prey density and the occupancy rate by one or more individuals per leaf.
Predatory mites were released between 8 and 10 am when the temperature and relative humidity were between 24 and 26 °C and 60 and 70%, respectively. Each release rate experiment was replicated on 5 test trees, separated by 2 buffer trees serving as windbreaks to reduce a possible wind-favoured mite dispersal. The average ambient temperature and relative humidity for the experiments were 30 °C and 55–65%.
Statistical analysis
Statistical analysis was performed in the R program (R Core Team 2019). Two-way multivariate analysis of variance (MANOVA) was used to test whether eggs and mobile stages of P. ulmi are influenced by the difference in release rate and sampling date. When the null hypothesis H0 was rejected, the normality of residuals was checked and normal distribution was approximated through square root transformation. Wilk’s lambda test was used to test whether there are differences between the mean densities of P. ulmi eggs and motiles. Furthermore, if the test revealed significant effects of treatment, sampling date, and treatment/sampling date interaction, a two-way analysis of variance (ANOVA), followed by a Tukey/Kramer HSD comparison test, was fitted to determine differences among release rates at each sampling date.
At each treatment, the Student t/test or Mann–Whitney nonparametric test was used, according to the Shapiro–Wilk test and Levene test for normality and homogeneity of variances to separate both groups.