Material and methods

Data compilation. The studies on the spatial association between cacti and nurse plants were searched in the ISI Web of Knowledge from 1986 to 2021 using the terms “cact*” or “spatial distribution” or “nurse”. We chose this time period because we did not find studies on the spatial association between cacti and nurse plants that used area sampling methods before 1986, and our study was performed in 2021. We found 1,023 studies that included at least one of the search terms. Of these 1,023 studies, only 58 studies had titles directly related to the topic. The examination of the abstract of these 58 studies showed that 28 studies had data relevant to the topic. Lastly, the examination of the full text of these 28 studies showed that 16 studies provided useful data on the association between cacti and nurse plants (Figure 1, Suppl. Mat.). These studies provided data on 39 cactus species, 76 nurse plant species, and 316 associations between cacti and nurse plants that were analyzed with the relative risk ratio (see below). The studies were performed in Mexico (6), USA (2), Mexico and USA (2), Argentina (2), Venezuela (2), Bolivia (1), and Chile (1). Figure 2.

Figure 1 Flow diagram of the literature search. 

Figure 2 Distribution of the studies on the spatial association between cacti and nurse plants. Each point can represent more than one study. 

For each study, we recorded author and year; plant community (i.e., arid tropical shrubland, desert shrubland, Larrea shrubland, semiarid shrubland, thorn scrubland, tropical dry forest, xeric shrubland); subfamily, tribe, species, and growth form of cacti (i.e., shrub cactus, columnar cactus, cylindrical cactus, globose cactus, flattened globose cactus sensu ^{Vázquez-Sánchez et al. 2012}); family, species, and growth form of nurse plants (i.e., grass, herb, shrub, tree, rosette). When cacti acted as nurse plants, we used the growth forms previously mentioned for cacti. In addition to these data, we recorded the observed number of cacti under nurse plants and in open spaces, and the expected number of cacti under nurse plants and in open spaces. When the number of cacti under nurse plants and in open spaces was reported graphically, we used the program WebPlotDigitizer version 4.6 (^{Rohatgi 2022}) to extract the numerical values.

For those studies that evaluated soil properties under nurse plants and in open spaces, we recorded family, species, and growth form of nurse plants as previously described, and numerical values of the soil properties under nurse plants and in open spaces. The numerical values of the soil properties obtained from the studies were expressed in the same units of measurement to compare them.

Data analysis. Data on the spatial association between cacti and nurse plants were analyzed with the relative risk ratio because it allowed us to compare the probability of cacti growing under shrubs versus the probability of cacti growing in open spaces. The relative risk ratio was calculated as: Relative risk ratio = ln [(observed number of cacti under nurse plants / expected number of cacti under nurse plants) / (observed number of cacti in open spaces / expected number of cacti in open spaces)], where ln is the natural logarithm. If the relative risk ratio = 0 the probability of cacti growing under nurse plants and the probability of cacti growing in open spaces were similar, if the relative risk ratio > 0 the probability of cacti growing under nurse plants was higher than the probability of cacti growing in open spaces, and if the relative risk ratio < 0 the probability of cacti growing under nurse plants was lower than the probability of cacti growing in open spaces. This ratio was used to evaluate the effect of subfamily, tribe, and growth form of cacti, the effect of family and growth form of nurse plants, and the effect of plant community on the association between cacti and nurse plants.

Data on the soil properties were analyzed with the ratio of medians because it allowed us to compare the soil properties under shrubs versus the soil properties in open spaces. We calculated the ratio of medians because several soil properties such as moisture, organic matter, nitrogen, and calcium carbonate content were estimated in percentage. The ratio of medians was calculated as: Ratio of medians = ln (numerical value of the soil property under nurse plants / numerical value of the soil property in open spaces), where ln is the natural logarithm. If the ratio of medians = 0 the soil property was similar under nurse plants and in open spaces, if the ratio of medians > 0 the soil property was higher under nurse plants than in open spaces, and if the ratio of medians < 0 the soil property was lower under nurse plants than in open spaces.

The relative risk ratio and ratio of medians were analyzed with the Wilcoxon one-sample signed-rank test to evaluate whether they differ from zero. We used this non-parametric test because the number of cacti under nurse plants and in open spaces are discrete variables that do not have a normal distribution. Moreover, several studies did not report the dispersion measures of the soil properties, nor the sample size, thus we could not use parametric tests. The non-parametric tests have been used in quantitative reviews of ecological aspects such as ecological restoration with ants (^{Casimiro et al. 2019}, ^{Ríos-Casanova & Godínez-Alvarez 2023}), and biodiversity and ecosystem services (^{Rey-Benayas et al. 2009}). The statistical tests were performed in the software PAST version 4.11 (^{Hammer et al. 2001}).

Results

Cacti: subfamilies, tribes, and growth forms. We obtained data for the subfamilies Opuntioideae and Cactoideae. For the subfamily Opuntioideae, the probability of cacti growing under nurse plants was lower than the probability of cacti growing in open spaces (Z = 2.49, P = 0.013). In contrast, for the subfamily Cactoideae, the probability of cacti growing under nurse plants was higher than the probability of cacti growing in open spaces (Z = 7.88, P < 0.000; Figure 3A).

Figure 3 Relative risk ratio for A) subfamilies, B) tribes, and C) growth forms of cacti. The ratios marked with * significantly differed from zero. 

As for the tribes, we obtained data for the tribes Cacteae, Cereeae, Notocacteae, Echinocereeae , and Trichocereeae. For the tribes Notocacteae and Trichocereeae, the probability of cacti growing under nurse plants and the probability of cacti growing in open spaces were similar (Notocacteae: Z = 1.60, P = 0.109, Trichocereeae: Z = 0.28, P = 0.779). For the tribes Cacteae, Cereeae, and Echinocereeae , the probability of cacti growing under nurse plants was higher than the probability of cacti growing in open spaces (Cacteae: Z = 3.67, P < 0.000, Cereeae: Z = 3.90, P < 0.000, Echinocereeae : Z = 5.88, P < 0.000; Figure 3B).

As for the growth forms, we obtained data for shrub cacti, columnar cacti, cylindrical cacti, and globose cacti. The probability of cylindrical, columnar, and globose cacti growing under nurse plants was higher than the probability of growing in open spaces (cylindrical cactus: Z = 3.06, P = 0.002, columnar cactus: Z = 6.78, P < 0.000, globose cactus: Z = 3.20, P = 0.001). Conversely, the probability of shrub cacti growing under nurse plants was lower than the probability of growing in open spaces (Z = 2.49, P = 0.013; Figure 3C).

Nurse plants: families and growth forms. We obtained data for 22 plant families. However, the Anacardiaceae, Bignoniaceae, Capparaceae, Celastraceae, Oleaceae, Rubiaceae, and Turneraceae were not analyzed because they had only one datum each (Figure 4A).

Figure 4 Relative risk ratio for A) families and B) growth forms of nurse plants. The ratios marked with * significantly differed from zero. 

The probability of cacti growing under nurse plants belonging to Acanthaceae, Asparagaceae, Boraginaceae, Bromeliaceae, Cactaceae, Capparidaceae, Fouquieriaceae, Rhamnaceae, Simaroubaceae, Solanaceae, Verbenaceae, and Zygophyllaceae was similar to the probability of cacti growing in open spaces (Acanthaceae: Z = 1.34, P = 0.18, Asparagaceae: Z = 1.99, P = 0.06, Boraginaceae: Z = 2.02, P = 0.06, Bromeliaceae: Z = 1.60, P = 0.11, Cactaceae: Z = 1.21, P = 0.23, Capparidaceae: Z = 0.53, P = 0.59, Fouquieriaceae: Z = 0.53, P = 0.59, Rhamnaceae: Z = 0.40, P = 0.69, Simaroubaceae: Z = 1.83, P = 0.07, Solanaceae: Z = 1.83, P = 0.07, Verbenaceae: Z = 1.78, P = 0.08, Zygophyllaceae: Z = 0.63, P = 0.53; Figure 4A). In contrast, the probability of cacti growing under nurse plants belonging to Asteraceae, Euphorbiaceae, and Fabaceae was higher than the probability of cacti growing in open spaces (Asteraceae: Z = 2.52, P = 0.01, Euphorbiaceae: Z = 2.16, P = 0.03, Fabaceae: Z = 5.63, P < 0.00; Figure 4A).

As for the growth forms, we obtained data for the herbs, shrubs, trees, rosettes, and cylindrical cacti. The probability of cacti growing under nurse plants such as trees, shrubs, and rosettes was higher than the probability of growing in open spaces (tree: Z = 3.78, P < 0.000, shrub: Z = 4.97, P < 0.000, rosette: Z = 2.43, P = 0.015; Figure 4B). In contrast, the probability of cacti growing under nurse plants such as herbs and cylindrical cacti was similar to the probability of growing in open spaces (herbs: Z = 0.45, P = 0.655, cylindrical cactus: Z = 1.60, P = 0.109; Figure 4B).

Plant communities. We obtained data for 10 plant communities. However, eight plant communities were comparable to the desert shrubland, thus we only considered the desert shrubland and tropical dry forest. In both plant communities, the probability of cacti growing under nurse plants was higher than the probability of cacti growing in open spaces (desert shrubland: Z = 6.66, P < 0.000, tropical dry forest: Z = 2.76, P < 0.006; Figure 5A).

Figure 5 Relative risk ratio and ratio of medians for A) plant communities, and B) soil physical and chemical properties, respectively. The ratios marked with * significantly differed from zero. EC = soil electrical conductivity, OM = soil organic matter, T = soil surface temperature. 

Soil properties. We obtained data for 13 soil properties. Of these soil properties, the soil surface temperature was higher in open spaces than under nurse plants (Z = 2.20, P = 0.028; Figure 5B).

The soil electrical conductivity, soil moisture, nitrogen, phosphorus, calcium, and magnesium were higher under nurse plants than in open spaces (electrical conductivity: Z = 2.52, P = 0.012, moisture: Z = 2.37, P = 0.018, nitrogen: Z = 2.19, P = 0.028, phosphorus: Z = 2.37, P = 0.018, calcium: Z = 2.50, P = 0.013, magnesium: Z = 2.40, P = 0.017; Figure 5B).

The soil organic matter, pH, calcium carbonate, potassium, sodium, and sulfates were similar under nurse plants and in open spaces (organic matter: Z = 1.57, P = 0.12, pH: Z = 0.65, P = 0.515, calcium carbonate: Z = 0.45, P = 0.655, potassium: Z = 1.35, P = 0.176, sodium: Z = 1.69, P = 0.109, sulfates: Z = 0.32, P = 0.753; Figure 5B).

Discussion

The results showed that cacti were associated to nurse plants. However, this association did not occur in all subfamilies, tribes, and growth forms of cacti, nor in families and growth forms of nurse plants. Moreover, nurse plants only ameliorated some soil properties. These results suggest that the association between cacti and nurse plants could be due to the amelioration of soil properties such as the soil surface temperature under the canopy of nurse plants. However, our review only considered studies that evaluated the spatial pattern of cactus association with area sampling methods by calculating the observed and expected number of cacti under nurse plants and in open spaces. Studies on the spatial pattern of cactus association with distance sampling methods, the association of cacti with nurse objects, and the experimental studies on seed germination and seedling survival under nurse plants and in open spaces were not considered in the review. Despite this limitation, as far as we know, this is the first review that quantitatively analyzed the association between cacti and nurse plants.

The shrub cacti of the subfamily Opuntioideae had a relatively low probability of being associated to nurse plants whereas the globose, cylindrical, and columnar cacti of the tribes Cacteae, Cereeae, and Echinocereeae , belonging to the subfamily Cactoideae, had a relatively high probability of being associated to nurse plants. These results suggest that some genera of the subfamily Opuntioideae such as Opuntia and Cylindropuntia can establish in open spaces probably due to their ability to propagate vegetatively (^{Evans et al. 2004}). The areoles of cladodes or terminal stem segments that fall to the ground can produce new roots and stems that originate new individuals. In fact, some species such as Opuntia rastrera, Opuntia macrocentra, and Cylindropuntia imbricata can propagate vegetatively in stressful environments like open spaces (^{Allen et al. 1991}, ^{Mandujano et al. 1998}), although this ability to propagate is seemingly higher in Opuntia than in Cylindropuntia due to the larger volume of cladodes to storage water and carbohydrates (^{Flores-Torres & Montaña 2012}).

In addition to the vegetative propagation, some species of Opuntia such as Opuntia pilifera can grow in open spaces because its cladodes are facing east and west. This orientation increases the absorption of photosynthetically active radiation, and decreases cladode temperature, thus allowing photosynthesis (^{Cano Santana et al. 1992}).

The probability of cacti growing under perennials and the probability of cacti growing in open spaces were similar in the Acanthaceae, Asparagaceae, Boraginaceae, Bromeliaceae, Cactaceae, Capparidaceae, Fouqueriaceae, Rhamnaceae, Simaroubaceae, Solanaceae, Verbenaceae, and Zygophyllaceae. In contrast, the probability of cacti growing under nurse plants was higher than the probability of cacti growing in open spaces in the Asteraceae, Euphorbiaceae, and Fabaceae. These results suggests that these two groups of plant families differ in the amelioration of soil properties under their canopies and thus, in the provision of an adequate microclimate for seed germination and seedling growth of cacti. Other studies (^{Flores & Jurado 2003}, ^{Godínez-Alvarez et al. 2003}) also found that the Fabaceae and Asteraceae had high numbers of species acting as nurse plants.

The nurse plants of some families such as Asparagaceae (Agave), Bromeliaceae (Hecthia), and Cactaceae (Echinocactus) can be rosettes or cylindrical cacti that do not provide enough shade, nor produce enough soil organic matter to facilitate cactus establishment. Moreover, the nurse plants of other families such as Solanaceae (Capsicum) are shrubs with sparse crowns and relatively superficial root systems unable to decrease the direct solar radiation and transport water from the deep, moist soil layers to the superficial and dry layers of soil. In contrast, the nurse plants of some families such as Asteraceae (Ambrosia), Euphorbiaceae (Jatropha), and Fabaceae (Prosopis) are trees and shrubs with dense crowns and deep root systems, particularly nurse plants in the Fabaceae family, that can decrease the direct solar radiation and soil surface temperature, and increase the soil moisture through hydraulic lift (^{Muñoz et al. 2008}). Moreover, these trees and shrubs can produce enough leaf litter to increase the organic matter and nutrient content of the soil (^{Cares et al. 2013}). In fact, our results showed that nurse plants increased the electrical conductivity, moisture, nitrogen, phosphorus, calcium, and magnesium content of soil. However, nurse plants did not increase the organic matter, pH, calcium carbonate, potassium sodium, and sulfate content of soil. ^{Flores & Jurado (2003)} also found that soil nitrogen content and litter are higher under nurse plants than in open spaces.

The association between cacti and nurse plants occurred in the desert shrubland and tropical dry forest. These results suggest that both plant communities have harsh conditions of temperature and rainfall that can limit cactus establishment. For instance, the desert shrubland has low mean annual rainfall (< 50 mm to < 700 mm) that occurs mainly in the summer when the solar radiation, soil surface temperature, and evapotranspiration are high (^{Challenger 1998}). In contrast, the tropical dry forest has a mean annual temperature of 20-29 (C and a mean annual rainfall of 600-1,200 mm (^{Challenger 1998}). However, the rainfall is seasonal with a dry season that can last 5-8 months (^{Challenger 1998}). Because of these climatic conditions, the successful establishment of cacti in the desert shrubland and tropical dry forest can occur only under the canopy of nurse plants.

This quantitative review showed that several studies have evaluated the spatial association between cacti and nurse plants. However, this plant association does not include all the members of the cactus family because some subfamilies, tribes, and growth forms were not associated to nurse plants. Also, the nurse plants only modified some soil properties. Lastly, this plant association has been evaluated only in the desert shrubland and tropical dry forest. Therefore, further studies are needed on a greater number of subfamilies, tribes, and growth forms of cacti in different plant communities. Moreover, further studies are needed on the amelioration of soil properties under the canopy of nurse plants and their influence on cactus establishment. Finally, most studies on the association between cacti and nurse plants have been performed in Mexico and the USA, thus more research is needed in other countries with a high cactus diversity in Central and South America.

Supplementary material

Supplemental data for this article can be accessed here: https://doi.org/10.17129/botsci.3520.

Supplemental material.